Protein Homeostasis Diseases: Mechanisms and Novel Therapies [1 ed.] 0128191325, 9780128191323

Protein Homeostasis Diseases: Mechanisms and Novel Therapies offers an interdisciplinary examination of the fundamental

760 17 18MB

English Pages 450 [418] Year 2020

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Protein Homeostasis Diseases: Mechanisms and Novel Therapies [1 ed.]
 0128191325, 9780128191323

Table of contents :
Chapter 1 - Protein folding: how, why, and beyond
Outline
Introduction
Protein conformational landscapes
Mutational perturbations to probe folding mechanisms and function
Disordered proteins-regions and unfolded states
Folding, stability, and binding in vivo
“Real proteins” and beyond
Acknowledgments
References
Chapter 2 - Protein homeostasis and disease
Outline
Abbreviations
Protein folding in vitro and in vivo
Effects of intracellular milieu composition on protein folding, misfolding, and stability in vivo
The first steps of in vivo folding and misfolding in the ribosomes: cotranslational versus posttranslational processes
Protein homeostasis networks
Molecular chaperones
Protein degradation: proteasome versus autophagy
Human misfolding diseases
Loss-of-function diseases
Gain-of-toxic function diseases
Acknowledgments
Conflict of interest
Funding
References
Chapter 3 - Caenorhabditis elegans as a model organism for protein homeostasis diseases
Outline
Abbreviations
Caenorhabditis elegans as a model organism
The proteostasis network is conserved in Caenorhabditis elegans
The heat shock response
The unfolded protein response of the endoplasmic reticulum and the mitochondria
The insulin-like signaling pathway
Caenorhabditis elegans as a model for protein misfolding diseases
Alzheimer’s disease
Disease mechanism
Caenorhabditis elegans Amyloid-β models
Utility of Caenorhabditis elegans Alzheimer’s disease models for drug discovery and identification of genetic modifiers
Tauopathies
Disease mechanism
Caenorhabditis elegans models of tauopathy
Utility of Caenorhabditis elegans tauopathy models for drug discovery and identification of genetic modifiers
Parkinson’s disease
Disease mechanism
Caenorhabditis elegans Parkinson’s disease models
Utility of Caenorhabditis elegans Parkinson’s disease models for drug discovery and identification of genetic modifiers
Polyglutamine diseases
Disease mechanism
Caenorhabditis elegans models of Huntington’s disease
Utility of Caenorhabditis elegans HD models for drug discovery and identification of genetic modifiers
Caenorhabditis elegans models of spinocerebellar ataxia
Utility of Caenorhabditis elegans spinocerebellar ataxia models for drug discovery and identification of genetic modifiers
Amyotrophic lateral sclerosis
Disease mechanism
Caenorhabditis elegans models of amyotrophic lateral sclerosis
Transthyretin amyloidosis
Disease mechanism
Caenorhabditis elegans disease models of transthyretin amyloidosis
Type II diabetes mellitus
Disease mechanism
Caenorhabditis elegans model of diabetes mellitus
Dialysis-related amyloidosis
Disease mechanism
Immunoglobulin light chain amyloidosis
Disease mechanism
Caenorhabditis elegans models of immunoglobulin light chain amyloidosis
Prion diseases
Conclusion
References
Chapter 4 - Proteome-scale studies of protein stability
Outline
Abbreviations
Introduction
Protein stability and unfolding
Biophysical methods to measure protein stability in vitro
Protein stability in vivo
Biological readouts to measure protein stability
Proteome-scale analysis based on aggregation in vivo
Proteome-scale analysis based on degradation in vivo
Structural analyses of proteins and proteomes
Cross-linking-based mass spectrometry
Hydroxyl radical footprinting (HRF)
Limited proteolysis-based mass spectrometry
Proteome-scale methods involving experimental denaturation of proteins
Denaturation probed by proteolysis sensitivity
Thermal denaturation probed by aggregation
Denaturation probed by methionine oxidation
Contributions to basic and applied biomedical research
Conclusions
Acknowledgments
References
Chapter 5 - Classifying disease-associated variants using measures of protein activity and stability
Outline
Abbreviations
Introduction
Selection of PTEN variants
Comparing multiplexed assays and computational predictions to assess variant effects
Loss of stability is a major source for loss of PTEN function
Conclusions
Methods
Rosetta ∆∆G calculations
Evolutionary sequence energies (Ẽ)
Phosphatase-MAVE and VAMP-seq data
Determining thresholds from receiver operating characteristic curves
Analysis scripts
Acknowledgments
References
Chapter 6 - Protein destabilization and degradation as a mechanism for hereditary disease
Outline
Abbreviations
Introduction to protein quality control
Protein quality control in hereditary diseases
Protein folding and refolding
Protein quality control–mediated degradation via the ubiquitin-proteasome system
Protein quality control degrons
Local versus global unfolding
Potential therapeutic approaches to protein quality control–linked hereditary diseases
Acknowledgments
Conflict of interest
Funding
References
Chapter 7 - Detection of amyloid aggregation in living systems
Outline
Abbreviations
Introduction
Techniques for detection of amyloid aggregation in vivo
Förster resonance energy transfer detection and fluorescence lifetime imaging
Bioluminescence imaging
Optical fiber bundles and fluorescence imaging
Cranial window or thinned-skull imaging using multiphoton microscopy
In vivo microdialysis
Positron emission tomography imaging
Animal models to test in vivo amyloid formation
Caenorhabditis elegans
Zebrafish
Mouse models
Note on using animal models for in vivo protein aggregation studies
In vivo complexity and how do in vivo detection assays provide insight into peripheral aspects contributing to neurodegener...
Replicating the multicellular complexity of the brain
Interaction of the brain with the periphery system
Future outlook
Acknowledgments
References
Chapter 8 - Molecular mechanisms of amyloid aggregation in human proteinopathies
Outline
Introduction
Protein aggregates: from dynamic oligomers to amyloid fibrils
Amyloid fibrils
Oligomers
Protofibrils
Mechanisms of amyloid aggregation
Nucleation-Polymerization
Nucleated conformational conversion
More general mechanisms
Application to different disease-related proteins
Amyloid β peptides
α-Synuclein
hIAPP/amylin
Concluding remarks
Acknowledgments
References
Chapter 9 - Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases
Outline
Abbreviations
Metal ions and amyloid formation in neurodegeneration
Protein misfolding and metal ions in neurodegeneration
Trace metal import and homeostasis in the brain
Zinc and toxic protein aggregates in Alzheimer’s disease
Zinc dyshomeostasis in Alzheimer’s disease
Zinc binding and aggregation of Aβ
Zinc binding to Aβ
Influence of zinc on Aβ aggregation
Zinc binding and tau aggregation
Zinc binding to tau
Influence of zinc on tau aggregation
Metal chelation therapies
Acknowledgments
References
Chapter 10 - Vitamin B6-dependent enzymes and disease
Abbreviations
Clinical and genetic bases of misfolding diseases and natural ligand therapies
Vitamin B6 enzymes and disease
Primary hyperoxaluria type 1 due to deficiency of AGT
Clinical and genetic features
Molecular mechanisms leading to the AGT deficit in primary hyperoxaluria type 1
The role of natural and unnatural ligands as chaperones for AGT
Steps forward to new therapeutic approaches for primary hyperoxaluria type 1
Conclusions
Acknowledgments
References
Chapter 11 - Galactosemia: opportunities for novel therapies
Outline
Abbreviations
Introduction: four types of galactosemia
Causes of pathology and current treatment
Potential novel therapies
Pharmacological chaperones for GALT deficiency
Pharmacological chaperones for other types of galactosemia
Conclusions
Acknowledgments
References
Chapter 12 - Protein homeostasis and regulation of intracellular trafficking of G protein-coupled receptors
Outline
Introduction
Proteostasis and quality control systems
Proteostasis
Quality control systems
Regulation of anterograde G protein-coupled receptor traffic from the endoplasmic reticulum to the cell surface plasma membrane
Sequence motifs that promote/prevent upward trafficking of G protein-coupled receptors
Posttranslational modifications in G protein-coupled receptors and intracellular trafficking
Association of G protein-coupled receptors and intracellular trafficking
Intracellular G protein-coupled receptor trafficking from the cell surface plasma membrane and beyond
G protein-coupled receptor internalization via clathrin-coated pits and a central role of arrestins
Divergent sorting of internalized G protein-coupled receptors and impact on receptor activity
Sorting of G protein-coupled receptors to the regulated recycling pathway via multiple endosome types
Sorting of G protein-coupled receptors to the lysosome
Multilocation signaling of G protein-coupled receptors within the endocytic network
Targeting misfolded G protein-coupled receptors with pharmacological chaperones
Conclusions
Acknowledgments
References
Chapter 13 - Structure-guided discovery of pharmacological chaperones targeting protein conformational and misfolding diseases
Outline
Abbreviations
Introduction
The premise and promise of pharmacological chaperone therapy
Advances in protein structural biology
Structure-guided understanding of disease-causing variants
Variants leading to decreased flexibility
Variants leading to increased flexibility
Variants that alter cofactor or substrate binding
Variants that alter binding interface
Structural basis of pharmacological chaperoning
Active-site inhibitory pharmacological chaperones for lysosomal storage disorders
Starting points for allosteric pharmacological chaperones
Pharmacological chaperones for defects in channels, transporters, and receptors
Pharmacological chaperones for structural proteins
Structural methods used in primary compound screening
Fragment-based approach
Crystallography-based screening
Concluding remarks
Statements
Acknowledgments
References
Chapter 14 - Virtual screening in drug discovery: a precious tool for a still-demanding challenge
Outline
Abbreviations
Introduction: the drug discovery process
Accurate determination of binding affinity in a receptor–ligand complex
Speeding up the search process through approximate scoring functions
From binding properties to the identification of molecular descriptors
Organizing distinct simulation techniques into a screening protocol
References
Chapter 15 - Differential scanning fluorimetry in the screening and validation of pharmacological chaperones for soluble and membrane pr...
Outline
Abbreviations
Background
Initial high-throughput screening by differential scanning fluorimetry
Differential scanning fluorimetry–monitored screening for soluble proteins
Experimental setup
Data analysis
Filtering of hits
Validation of primary hits by concentration-dependent differential scanning fluorimetry
Differential scanning fluorimetry–monitored screening for membrane proteins
Experimental setup
Conclusion
Acknowledgments
References
Chapter 16 - Cellular high-throughput screening
Outline
Abbreviations:
Introduction
Cellular high-throughput screening
Assay types
Primary screen
Negative screen
Counter screen
Positive control compound
Pharmacoperone types
Compound toxicity
Genetic diseases potentially amenable to treatment with chemical or pharmacological chaperones: G protein-coupled receptors...
G protein-coupled receptors
Enzyme diseases
Ion channel diseases
Lysosomal storage disorders
Challenges of high-throughput screening “hits”
Conclusion
Acknowledgments
References
Chapter 17 - High-throughput screening for intrinsically disordered proteins by using biophysical methods
Outline
Abbreviations
Introduction
Drug discovery and biophysics
Biophysical techniques and high-throughput screening
Intrinsically disordered proteins
Fluorescence
Introduction to fluorescence
Fluorescence intensity
Fluorescence polarization or anisotropy
Fluorescence resonance energy transfer
Fluorescence temperature-related intensity change
Fluorescence-based high-throughput screening: thermal shift assay
Nuclear magnetic resonance
Introduction to nuclear magnetic resonance
Ligand-based nuclear magnetic resonance screening methods
Relaxation-based methods
Nuclear Overhauser effect-based methods
Saturation transfer difference (STD) between protein and ligand via nuclear Overhauser effects
Transfer of 1H-polarization from water to the ligand
Target-based nuclear magnetic resonance screening methods
Surface plasmon resonance
Applications of biophysical techniques to hit identification and validation for intrinsically disordered proteins
Funding
References
Chapter 18 - Natural and pharmacological chaperones against accelerated protein degradation: uroporphyrinogen III synthase and congenita...
Outline
Abbreviations
Introduction
Heme group biosynthesis
Uroporphyrinogen III synthase
Congenital erythropoietic porphyria
The UROIIIS “stability defect” analyzed in vitro
UROIIIS is a kinetically stable protein
Congenital erythropoietic porphyria–causing mutations accelerate protein degradation in vitro
The irreversible unfolding of UROIIIS and the structure of the aggregates
UROIIIS intracellular homeostasis and congenital erythropoietic porphyria
UROIIIS intracellular concentration is altered in C73R-UROIIIS
Quantitative roadmap of the pathogenic mutations that are affected by impaired homeostasis
In vitro kinetic stability measures correlate with UROIIIS intracellular steady-state concentration for the congenital eryt...
UROIIIS homeostasis and proteasomal degradation
UROIIIS proteostasis restoration by proteasomal inhibition in animal models
Ciclopirox as a pharmacological chaperone for congenital erythropoietic porphyria
Pharmacological chaperones
Ciclopirox and congenital erythropoietic porphyria
Toward a modulation of the heme biosynthetic pathway
Acknowledgments
References

Citation preview

PROTEIN HOMEOSTASIS DISEASES Mechanisms and Novel Therapies

Edited by

ANGEL L. PEY Department of Physical Chemistry, Faculty of Sciences, University of Granada, Granada, Spain

Academic Press is an imprint of Elsevier 125 London Wall, London EC2Y 5AS, United Kingdom 525 B Street, Suite 1650, San Diego, CA 92101, United States 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom Copyright © 2020 Elsevier Inc. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library ISBN: 978-0-12-819132-3 For information on all Academic Press publications visit our website at https://www.elsevier.com/books-and-journals

Publisher: Andre Gerhard Wolff Acquisitions Editor: Peter B. Linsley Editorial Project Manager: Kristi Anderson Production Project Manager: Swapna Srinivasan Designer: Matthew Limbert Typeset by Thomson Digital

Contributors

Olga Abian Institute of Biocomputation and Physics of Complex Systems (BIFI), Joint Units IQFRCSIC-BIFI, and GBsC-CSIC-BIFI, University of Zaragoza, Zaragoza; Aragon Institute for Health Research (IIS Aragon), Zaragoza; Biomedical Research Networking Centre for Liver and Digestive Diseases (CIBERehd), Madrid; Department of Biochemistry and Molecular and Cell Biology, Universidad de Zaragoza, Zaragoza; Aragon Health Science Institute (IACS), Zaragoza, Spain Ilaria Bellezza Department of Experimental Medicine, University of Perugia, Perugia, Italy Ganeko Bernardo-Seisdedos Protein Stability and Inherited Disease Laboratory, CIC bioGUNE, Bizkaia Technology Park, Derio, Spain Isabel Betancor-Fernandez Department of Pathology, ITB, CIBERER, Hospital Universitario Canarias, Universidad de La Laguna, Tenerife, Spain Jean-Marc Blouin University of Bordeaux, Biotherapy of Genetic Diseases, Inflammatory Diseases and Cancers, Bordeaux, France Kerensa Broersen Applied Stem Cell Technologies, Faculty of Science and Technology, University of Twente, Enschede, The Netherlands Giulia Calloni Buchmann Institute for Molecular Life Sciences; Institute of Biophysical Chemistry, Goethe University Frankfurt, Frankfurt am Main, Germany Barbara Cellini Department of Experimental Medicine, University of Perugia, Perugia, Italy Caspar E. Christensen Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark Francisco Conejero-Lara Department of Physical Chemistry and Institute of Biotechnology, Faculty of Sciences, University of Granada, Spain Joana S. Cristóvão Biosystems and Integrative Sciences Institute; Department of Chemistry and Biochemistry, Faculty of Sciences, University of Lisbon, Lisbon, Portugal Marte I. Flydal Department of Biomedicine, University of Bergen, Bergen, Norway

xv

xvi

Contributors

Nicole Fontana Department of Pathology, ITB, CIBERER, Hospital Universitario Canarias, Universidad de La Laguna, Tenerife, Spain Douglas M. Fowler Departments of Genome Sciences and Bioengineering, University of Washington, Seattle, WA, United States David Gil Electron microscopy platform, CIC bioGUNE, Bizkaia Technology Park, Derio, Bizkaia, Spain Cláudio M. Gomes Biosystems and Integrative Sciences Institute; Department of Chemistry and Biochemistry, Faculty of Sciences, University of Lisbon, Lisbon, Portugal Sarah Good School of Molecular and Cell Biology and Astbury Centre for Structural Molecular Biology, University of Leeds; Leeds, United Kingdom Andreas M. Grabrucker Cellular Neurobiology and Neuro-Nanotechnology Lab, Department of Biological Sciences; Bernal Institute; Health Research Institute (HRI), University of Limerick, Limerick, Ireland Fedora Grande Department of Pharmacy, Health and Nutritional Sciences, University of Calabria, Rende, Italy Silvia Grottelli Department of Experimental Medicine, University of Perugia, Perugia, Italy Aylin C. Hanyaloglu Institute of Reproductive and Developmental Biology, Imperial College London, London, United Kingdom Rasmus Hartmann-Petersen Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark Emil Hausvik Department of Biomedicine, University of Bergen, Bergen, Norway Jo Ann Janovick Department of Cell Biology & Biochemistry, Texas Tech University Health Sciences Center, Lubbock, TX, United States Michael Maglegaard Jepsen Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark Kresten Lindorff-Larsen Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark Aurora Martinez Department of Biomedicine, University of Bergen, Bergen, Norway

Contributors

Thomas J. McCorvie Section of Structural Biology, Department of Medicine, Imperial College, London; Structural Genomics Consortium, Nuffield Department of Clinical Medicine, University of Oxford, Oxford, United Kingdom Oscar Millet Protein Stability and Inherited Disease Laboratory, CIC bioGUNE, Bizkaia Technology Park, Derio, Spain Guilherme G. Moreira Biosystems and Integrative Sciences Institute; Department of Chemistry and Biochemistry, Faculty of Sciences, University of Lisbon, Lisbon, Portugal Bertrand Morel Department of Physical Chemistry and Institute of Biotechnology, Faculty of Sciences, University of Granada, Spain Athi N. Naganathan Department of Biotechnology, Bhupat & Jyoti Mehta School of Biosciences, Indian Institute of Technology Madras (IITM), Chennai, India Jose L. Neira Institute of Biocomputation and Physics of Complex Systems (BIFI), Joint Units IQFR-CSICBIFI, and GBsC-CSIC-BIFI, University of Zaragoza, Zaragoza; Institute of Molecular and Cell Biology, Miguel Hernández University, Elche (Alicante), Spain Sofie V. Nielsen Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark Angel L. Pey Department of Physical Chemistry, Faculty of Sciences, University of Granada, Granada, Spain Emmanuel Richard University of Bordeaux, Biotherapy of Genetic Diseases, Inflamatory Diseases and Cancers, Bordeaux, France Bruno Rizzuti CNR-NANOTEC, Licryl-UOSCosenza, Department of Physics, University of Calabria, Rende, Italy Eduardo Salido Department of Pathology, ITB, CIBERER, Hospital Universitario Canarias, Universidad de La Laguna, Tenerife, Spain Signe M. Schenstrøm Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark Svein I. Støve Department of Biomedicine, University of Bergen, Bergen, Norway Amelie Stein Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark

xvii

xviii

Contributors

David J. Timson School of Pharmacy and Biomolecular Sciences, University of Brighton, Brighton, United Kingdom Alfredo Ulloa-Aguirre Red de Apoyo a la Investigación (RAI), National University of Mexico (UNAM) and Instituto Nacional de Ciencias Médicas y Nutrición SZ, Mexico City, Mexico Jarl Underhaug Department of Chemistry, University of Bergen, Bergen, Norway R. Martin Vabulas Buchmann Institute for Molecular Life Sciences; Institute of Biophysical Chemistry, Goethe University Frankfurt, Frankfurt am Main, Germany Patricija van Oosten-Hawle School of Molecular and Cell Biology and Astbury Centre for Structural Molecular Biology, University of Leeds, Leeds, United Kingdom Sonia Vega Institute of Biocomputation and Physics of Complex Systems (BIFI), Joint Units IQFR-CSICBIFI, and GBsC-CSIC-BIFI, University of Zaragoza, Zaragoza, Spain Adrian Velazquez-Campoy Institute of Biocomputation and Physics of Complex Systems (BIFI), Joint Units IQFR-CSICBIFI, and GBsC-CSIC-BIFI, University of Zaragoza, Zaragoza; Aragon Institute for Health Research (IIS Aragon), Zaragoza; Biomedical Research Networking Centre for Liver and Digestive Diseases (CIBERehd), Madrid; Department of Biochemistry and Molecular and Cell Biology, Universidad de Zaragoza, Zaragoza; Fundacion ARAID, Government of Aragon, Zaragoza, Spain Wyatt W. Yue Structural Genomics Consortium, Nuffield Department of Clinical Medicine, University of Oxford, Oxford, United Kingdom Teresa Zariñán Red de Apoyo a la Investigación (RAI), National University of Mexico (UNAM) and Instituto Nacional de Ciencias Médicas y Nutrición SZ, Mexico City, Mexico

Preface

Proteins are the most versatile biological macromolecules. They carry out biochemical reactions essential for metabolic homeostasis with exquisite efficiency and regulation, and control a wide variety of cellular processes by establishing specific interactions with other proteins, nucleic acids, biological membranes, and metabolites. This functional specificity and versatility of proteins is often tightly connected with their ability to acquire one- (or several) three-dimensional structures, a process generally referred to as protein folding. However, proteins also adopt alternative (often non-functional) threedimensional conformations that may compromise cellular metabolism and homeostasis through a process collectively known as protein misfolding. Consequently, living organisms have developed complex regulatory machineries to ensure that proteins are produced in the ribosome with fidelity, achieve their native structures, arrive to the cellular locations in which their functions are required, and are eventually targeted for their destruction to prevent misfolding. Alterations in this protein homeostasis naturally occur through changes (often inherited) in the protein sequence, and due to environmental stress or ageing, leading to human disease. These protein homeostasis diseases have become a huge social, economic, and human burden, particularly because in most of cases, these limit the life span and quality of life, and no efficient ways to treat them are available. The aim of this book is to present a current view of different aspects for protein homeostasis, focusing on its roles in human physiology and pathology and novel approaches to understanding and treating protein homeostasis diseases. Section I (Chapters 1 and 2) presents a brief overview of the physicochemical principles underlying protein folding and misfolding, using an integrative perspective that must ultimately take into account the complexity of protein homeostasis in vivo: from the vectorial synthesis of proteins in the ribosomes to their final destination(s) linked to subcellular protein traffic and degradation pathways. Section II (Chapters 3–5) focuses on state-of-the-art approaches aimed to understanding the effects of disease-associated mutations on protein folding and misfolding at the proteomic and organismal levels. This section also discusses our current (and limited) ability to predict pathogenicity of single amino acid changes from gene-targeted or whole-genome sequencing studies, large-scale studies on the effects of missense variants on protein stability and function, and the use of model organisms to understand protein homeostasis in disease and its dependence on cellular and individual contexts. Section III is dedicated to present in more detail basic mechanisms by which genetic variation leads to protein homeostasis defects in disease. This section introduces fundamental mechanisms that cause loss-of-function and gain-of-toxic-function

xix

xx

Preface

diseases from physicochemical to cellular points of view (Chapters 6–8). Then, some representative examples of diseases associated with altered protein aggregation, stability, and intracellular trafficking are presented in Chapters 9–12, with a particular focus on the modulation of protein folding and misfolding exerted by natural ligands binding to natively folded proteins. Finally, Section IV is devoted to the description of integrated and multi-disciplinary approaches, aimed to develop novel therapeutic approaches to treat protein homeostasis diseases. Chapters 13 and 14 describe the use of structural information to gain insight into effects on disease-associated mutations as well as to identify potential binding sites for pharmacological molecules using structure-based virtual screening. Chapters 15–18 are dedicated to describe promising screening campaigns that are using tools from biochemistry, biophysics, cell, and structural biology, to identify hit molecules and improve them as potential treatments for protein homeostasis diseases.

CHAPTER 1

Protein folding: how, why, and beyond Athi N. Naganathan

Department of Biotechnology, Bhupat & Jyoti Mehta School of Biosciences, Indian Institute of Technology Madras (IITM), Chennai, India

Outline Introduction Protein conformational landscapes Mutational perturbations to probe folding mechanisms and function Disordered proteins-regions and unfolded states Folding, stability, and binding in vivo “Real proteins” and beyond References

3 6 8 11 13 14 16

Abbreviations H/D  hydrogen/deuterium IDP  intrinsically disordered protein IDR  intrinsically disordered region MD  molecular dynamics NMR  nuclear magnetic resonance PolyQ  polyglutamine WSME  Wako-Saitô-Muñoz-Eaton

Introduction A universal feature of biological polymers is their conformational complexity. This is more apparent in proteins, the molecular machines that perform the majority of functions within the cell. With an arsenal of 20 amino acids with very different chemical and structural properties, the unstructured heteropolymers synthesized by the ribosomes pack themselves into unique structures. The time taken for this self-organization and the shape eventually adopted depend, in most cases, almost entirely on the precise patterning of amino acids in the polymer sequence and the length of the sequence. In other words, if the patterning is different even in one or a few positions, it is highly likely that the polymer does not fold or that it populates several competing structures. Thus, the complexity of protein folding can be broadly classified as originating from two different phenomena: the large conformational entropy of the polymeric chain and the multitude of noncovalent interactions that hold the structure together. Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00001-4 Copyright © 2020 Elsevier Inc. All rights reserved.

3

4

Protein Homeostasis Diseases

The phase space (i.e., the array of conformations) accessible to a protein chain is astronomical, and this arises primarily from the backbone Ramachandran angles1 that impart tremendous flexibility limited only by steric effects of the main-chain and sidechain atoms. For example, it is known that the entropic penalty for fixing an alanine in a folded-like conformation is ∼17 J mol−1 K−1;2 this translates to eight times more conformations (or main-chain dihedral possibilities) in the unfolded state compared to the folded state. If one were to consider a sequence of five alanines this would translate to a collection of >28,000 possible conformations apart from the fully folded conformation. This entropic stabilization of the polymeric chain needs to be substantially outweighed by favorable energetics to promote folding.3-7 Such elementary considerations are also evidence that simple chemical models of folding would not do justice to the underlying conformational heterogeneity. Therefore, the probability of randomly arriving at a heteropolymer sequence that folds to a specific conformation is next to impossible. The difference between a random heteropolymer and the evolutionarily selected extant protein sequence thus lies in the ability of the latter to fold in a biologically relevant timescale to one or a specific collection of conformations. How is this possible? Energy landscape theory, the currently accepted paradigm for protein folding, states that natural selection has fine-tuned the sequence patterning such that proteins fold by avoiding traps (misfolded nonnative states) that could arise due to conflicting energetics (or “frustration”).3,8 This helps them to more likely form favorable interactions, thus simultaneously decreasing energy and entropy and speeding up folding. The resulting conformational landscape in three dimensions is conventionally represented as a funnel with a deep minimum or a collection of a few minima. Protein folding landscapes are considered to be “minimally frustrated,”3,9 as natural selection would have eliminated most of the nonnative interactions through sequence changes to enable rapid folding and optimal functionality. Simple lattice models of protein folding that use polar (P) and hydrophobic (HP) partitioning of amino acid residue types are able to reproduce many of the basic thermodynamic and kinetic features observed experimentally.10-14 In such simulations, it is clear that random placements of hydrophobic and polar amino acids result in multitudes of traps, and no specific organized structure is populated. In fact, one has to predesign such lattice sequences to ensure that they fold to specific conformations following which other folding-related features are studied. Folding mechanisms in lattices can also be distinctly tuned through design of sequences with different local and nonlocal content.13 These simulations suggest that even if we have the computational power to sample all possible conformations for a real protein, the problem of protein structure-stability prediction boils down to the precise estimation of noncovalent energetics that hold the protein together. In other words, which one or a selected few has the lowest (free) energy while also accounting for solvent-associated energies and entropies?

Protein folding: how, why, and beyond

The noncovalent interactions that stabilize proteins are varied and include van der Waals, charge-charge and charge-dipole, hydrogen bonds, and cation-π interactions apart from hydrophobic free-energetic factors arising from polymer and solvent entropies (bulk and first shell of solvent).15 While a majority of these terms are routinely included in simulations and structure-prediction algorithms, the precise magnitude or even the relative balance between the different terms is challenging to estimate. This is a critical component of any predictive modeling, as folded states of most proteins exhibit stabilities of ∼10-30 kJ mol−1 at, say, 298 K. In this regard, polymer chain entropic freeenergy components are of the order of ∼500 kJ mol−1 for a 100-residue polypeptide. Many of the noncovalent interactions are of the order of just thermal energy or RT (∼2.5 kJ mol−1) or weaker but are very many in number and eventually dominate over the destabilizing entropic terms, thus stabilizing the folded state (or rather, just about stabilizing it). These conflicting free-energetic terms also complicate the prediction of protein structures from sequence information, as not only should the conformational space be sampled extensively but the (free-) energy functions should be reasonably detailed and accurate to account for the diversity in noncovalent energetics. In this regard, it is pertinent to note that the success of the fragment-assembly method pioneered by the group of Baker and coworkers highlights the importance of local energetics and backbone geometry.16 However, large-scale mini-protein design from the same group highlights the importance of buried non-polar surface area as a prominent feature to enhance protein stability potentially through higher-order effects.17 Lessons from ancestral protein design and studies on mesophilic-thermophilic protein pairs reinforce the role of hydrophobicity and a highly interconnected network of charge-charge interactions in determining stability.18-20 These studies underpin the role of multiple energetic terms in determining stability, designability, and foldability of proteins. However, designed proteins generally exhibit more complex folding thermodynamics and kinetics21−23 compared to the natural proteins, suggesting that there are still rules to be learned that are hidden in the layers of compensating energetic-entropic effects. How does one go about probing the folding mechanism of a protein? Experimentally, this involves constructing unfolding curves from varied experimental probes (circular dichroism, fluorescence, infrared) with temperature or denaturant as a first step followed by a simple chemical two- or three-state analysis to estimate stabilities and populations. This is followed by kinetic studies (stopped-flow kinetics or ultrafast spectroscopy)24,25 to probe for kinetic evidence to the proposed mechanism from the number of relaxation rates, amplitudes, and their dependence on the perturbation employed. This is generally complemented by various studies, including differential scanning calorimetry26,27 (information on stabilities, thermodynamic barriers, order of conformational transitions), nuclear magnetic resonance28-31 (NMR; structure, dynamics, population of folded and partially structured states), hydrogen-deuterium (H/D) exchange coupled

5

6

Protein Homeostasis Diseases

with mass spectrometry32,33 (rates, local and global stability estimates for most residues in the protein), and a host of single-molecule methods34-38 (kinetics, distances between specific positions, populations, pathway heterogeneity, transition path times). Experimental constraints (e.g., solubility) preclude the possibility of doing all possible experiments on the same protein, but the overall theme is to identify a self-consistent mechanism of folding that explains the observed experimental signals. Site-specific mutations are then introduced to test the hypothesis on folding mechanisms and functional roles of specific regions (vide infra).

Protein conformational landscapes While it is common to employ two- or three-state models to fit the equilibrium unfolding curves, it is important to realize that they are, at best, approximations with little predictive power.39,40 As discussed earlier, a statistical description that accounts for the diversity of conformations and noncovalent interaction types is best suited for explaining the experimental data (e.g., see the original classical works on statistical treatments of folding).41-43 In fact, if the precise magnitude of each of the noncovalent interactions from the perspective of the local dielectric environment is known, it should be possible to predict protein stabilities and mechanisms, explore the role of mutations, and identify competing conformations. Some of these conformations could exhibit different catalytic rates or activities, substrate promiscuity, and misfolding tendencies, thus enabling a detailed view of protein sequence-structure-function-misfolding connection. However, it is impossible to sample all possible conformations of a protein chain, and this is where computational modeling comes to the fore. All-atom molecular dynamics (MD) simulations in which all atoms are explicitly represented and a majority of the energetics are accounted for are arguably the ideal way to sample protein conformational space starting from the folded state.44,45 Recent developments in sampling, Markov model-based approaches, hardware and better force fields, and water models have made all-atom MD an attractive avenue to explore conformational transitions with great temporal and structural detail.46-49 Coarse graining, wherein residues are represented as single or multiple beads, can enhance conformational sampling and can transcend time-scale issues that typically plague all-atom MD simulations.50-53 An alternative is to employ statistical mechanical models that treat residues as units and, based on specific assumptions, construct a predefined physical collection of possible conformational states, the relative probabilities of which are determined by the energetics.54-57 While researchers can choose from any of the earlier-mentioned methods to sample conformational states, it is important to underscore the fact that coarsegrained and statistical models are arguably more predictive and quantitative but make more assumptions with fewer parameters.52,58 Particularly, the latter two methods generally assume that the interactions observed in the native structure determine the folding mechanism (Gō-model)59 and hence the relative population of partially structured states.

Protein folding: how, why, and beyond

Long-time-scale MD simulations have provided strong evidence that native contacts determine folding mechanisms,60 giving a recent fillip to Gō-model based approaches. Once the states are sampled adequately or predicted unfolding curves calibrated against experiments, the conformations are generally projected onto one- or two-order parameters (that can also serve as reaction coordinates), and this constitutes the conformational landscape of a protein (Fig. 1.1). They provide a simple view of the complex hyperdimensional space associated with polymers, information on the relative population

Figure 1.1  Protein conformational landscapes (color online). All representations have been generated from the Wako-Saitô-Muñoz-Eaton (WSME) statistical mechanical model. A spectral color-coding is employed going from low free energy (FE in kJ mol−1; blue) to high (red). Unfolded and folded-like conformations are indicated as U and F, respectively. The cartoon on the top right of each panel depicts the native conformation. (A) Experimentally consistent conformational landscape of the repeat protein IkBα.75 The coordinates m and n represent the starting structured residue and the number of structured residue in a single stretch, respectively. Numerous partially structured states are observed apart from the folded state that represent different numbers of structured repeats arising from weak packing of interrepeat interfaces. (B) The folding conformational landscape of the globular protein L, highlighting its funneled nature. Two macroscopic folding paths to the folded state starting from the unfolded state can also be seen (dashed curves), in agreement with single-molecule experiments that point to intermediates.76 The coordinates are the number of residues structured in the N and C termini, respectively. (C) Same as in panel A, but for the disordered protein CytR highlighting a flat landscape with multiple competing minima or conformations.77 The golden-colored regions in the cartoons represent residues assuming a folded-like conformational status. The model predicts even a fully folded conformation to coexist in the disordered ensemble that was recently validated experimentally.105 Source: Adapted with permission from Ref. [77]. Copyright (2013) American Chemical Society.

7

8

Protein Homeostasis Diseases

of varied states (that can be intermediates, excited, or partially structured states) and the barriers between them (Fig. 1.1). It should be possible to perform diffusive calculations of even simple Monte Carlo simulations on these landscapes to probe for time scales of interconversion. If landscapes are generated using all-atom MD, then it should be possible to identify even misfolded states that can serve as starting points for aggregation. Thus, conformational landscapes, if properly constructed, carry critical information and not just folding pathways but also the functionalities and regions of protein that are more prone to early or late unfolding. It is important to note that a careful choice of order parameters is needed, but in many cases, simple metrics like RMSD (root mean squared deviation from the starting structure), Rg (radius of gyration), Q (fraction of native contacts), or number of structured residues, are sufficient to provide a detailed picture of conformational landscapes (Fig. 1.1). Conformational landscapes have been constructed for a large number of proteins and have shown to be extremely useful in understanding protein folding mechanisms (number of macrostates, possible routes, and fluxes), function, allostery, fold switching, posttranslational modification, misfolding, and environment-dependent modulation of populations (the reference list is not exhaustive!).61-67 One-dimensional free-energy profiles have also been extensively employed to predict folding rates, identify intermediates, and probe for excited states that could be functionally relevant.55,68-70 Successes of these methods are, in turn, a validation of the “minimum frustration principle” described in the energy landscape theory of protein folding. One aspect learned from many of these simulations with a strong theoretical underpinning is that frustration has not been or cannot be completely eliminated, as there are constraints on a protein to not only fold reasonably quickly but to also function and exhibit a certain kinetic and thermodynamic stability for efficient regulation. For example, DNA-binding domains need to necessarily display a collection of positive-charged residues close in space to bind the anionic DNA. Proteins that need to bind aromatic ligands would have a patch of hydrophobic residues on their surface, and this would evolve at the expense of a higher aggregation tendency. Enzyme active sites have also been predicted to be frustrated from structural analysis.71 Such functionally driven frustration, in turn, makes certain regions of the protein more prone to local unfolding that can appear as intermediates or excited states in conformational landscapes (Fig. 1.1B). Importantly, they could determine the flexibility and hence the dynamics that is required for optimal catalytic activity. On the other hand, such partially structured states could determine the aggregation propensity of a protein, as has been shown on several systems.72-74

Mutational perturbations to probe folding mechanisms and function To understand many of the features discussed earlier, it is necessary to perturb proteins and probe for how they react (thermodynamic and kinetic stability, function, organismal fitness) to such perturbations. This is generally carried out through mutational studies

Protein folding: how, why, and beyond

that have revealed a gold mine of information on protein folding, function, evolution, cooperativity, and allostery.78-84 Mutations in proteins drive the adaptability of organisms to different environmental stresses and to generate new functionality or antibiotic resistance. Importantly, a majority of the inherited diseases have their origins in missense mutations that not only modulate stability but also have a silent (not yet quantified) impact on associated cellular processes including abundance, protein-protein/DNA interactions, sensitivity to changes in cytoplasmic composition, efficiency of degradation, and posttranslational modifications. Numerous studies have established that proteins are marginally stable, and random mutational events constantly change the stability of proteins in cellular populations.85,86 However, in many cases, it is seen that there is a stability threshold beyond which the protein cannot tolerate mutations and they unfold. The extent to which they tolerate mutations depends on the degree or order of a disorder in proteins and the interplay between them.87 For example, while most random mutations in protein tend to destabilize proteins, certain mutations can enhance protein stability or activity in the presence of other mutations. Such high-order nonadditive mutational effects, termed epistasis, can be either beneficial (positive epistasis) or detrimental (negative epistasis) to the organismal fitness that can be quantified based on their functionality.88 Ancestral protein reconstruction methods, deep mutational scanning, and directed evolution point to tremendous variation in the extent to which catalytic activity can be modulated upon mutations, highlighting their epistatic effects.82,84,89 Remarkably, epistatic interactions alone can be employed to calculate three-dimensional structures of proteins through deep mutational scans, thus tremendously reshaping conventional applications of mutational studies.90,91 Importantly, it is now established that protein mutations in interior (i.e., buried residues) not only alter properties of the residues in their immediate neighborhood (first shell of residues) but also distant residues (second shell and beyond).78 Such a propagative effect (in an equilibrium sense) is a hallmark of the network of noncovalent interactions within proteins that can readjust their packing and dynamics depending on the nature of the perturbation. Studies indicate dramatic changes in the dynamics, hydrogen-exchange protection factors, and chemical shifts for a large number of residues in proteins (at distances even >15-20 Å; Fig. 1.2A,C,E) on mutational perturbations.92-94 Repacking of the protein interior on mutations affects the specificity-affinity ratio for ligands on the surface,95 highlighting the “connectedness” of the protein interior with the surface. It is possible to accurately capture a majority of these equilibrium propagatory features via a variety of models, including network analysis of protein structures, MD simulations, and even detailed quantitative statistical mechanical modeling.96 This tightrope walk of protein sequences in the stability-activity phase space provides a peek into how nature could have engineered proteins with different functionalities. At a fundamental level, such tunability and context dependence of missense variants has its origins in the pliable nature of the intraprotein interaction network and the degree

9

10

Protein Homeostasis Diseases

of partitioning of stabilizing versus destabilizing interactions in specific regions within proteins. When interactions within a protein are perturbed, different regions react differently, depending on the strength and malleability of the local interaction network. This in turn (de)stabilizes certain interactions and modulates the population of partially structured states apart from the unfolded state, thus affecting the conformational landscape (Fig. 1.2B,D,F). Single-point mutations thus reshape the landscape irrespective

Figure 1.2  Mutations in protein interior modulate the population of partially structured states. The left-most column (A, C, and E) depicts cartoons of proteins with the effect of specific mutations (i.e., chemical shifts) mapped onto the structure.93,97,98 The middle and right-most columns (B, D, and F) are the folding landscapes of the wild type (WT) and the representative mutants from the perspective of the Wako-Saitô-Muñoz-Eaton (WSME) model.96 Arrows point to the appearance or a relatively higher population of partially structured states compared to the WT. Source: Adapted with permission from Ref. [96]. Copyright (2017) American Chemical Society.

Protein folding: how, why, and beyond

of whether they are buried or exposed to the solvent (Figs. 1.2, 1.3A) and potentially contribute to epistatic effects in the presence of a secondary mutation, aspects that can be extracted only from an intimate and quantitative modeling of protein conformational behavior. The modulation of the population need not be restricted to mutations but can also be achieved through ion-binding events (Fig. 1.3B), temperature changes (Fig. 1.3C,D), pH variations (Fig. 1.3E,F), generic charge screening from ionic-strength changes, and posttranslational modifications (e.g., phosphorylation, Fig. 1.3G,H), and therefore even upon ligand binding. It is also likely that certain mutations at specific positions can predispose proteins to undergo misfolding and aggregation by populating alternate conformations or even through rerouting of folding flux.

Disordered proteins-regions and unfolded states If the driving force for compaction is small due to weak sequence hydrophobicity, spatially proximity of similarly charged residues, or excess of glycine/proline or polar residues, a protein or regions of proteins can be completely disordered, even in native conditions.106,107 Disordered proteins can therefore be seen as an extreme manifestation of frustration at not only the level of sequence but through degeneracy in interactions; in fact, this is expected of polymeric systems that can be stabilized by either entropy and specific interactions alone or through a combination of both. Such intrinsically disordered proteins (IDPs) or regions (IDRs) have redefined the structure-function paradigm due to their ubiquity in organisms at different levels. Many IDPs act as hubs in proteinprotein interaction networks and acquire structure only in the presence of their partners, thus regulating function in a conditional manner. Being disordered can help in easier regulation (through degradation) and can potentially promote promiscuity in binding, aspects that are currently of great interest to protein scientists.108-111 Much of the focus over the last decade has now shifted from working on ordered domains to IDPs or proteins with a large chunk of IDRs. Many IDRs and IDPs have larger mutational rates compared to their ordered counterparts, suggesting that they are the hotbeds of evolution.112 Experiments and simulations are now increasingly identifying that IDPs populate pockets of local structures (and even fully folded conformations) while at the same time exhibiting random coil dimensions.113-115 For example, α-synuclein, which is largely expressed in the nervous system and implicated in Parkinson's disease, is an IDP that can adopt helical conformation near membranes and beta-sheet-like structures in aggregates. It can also undergo extensive posttranslational modifications ranging from phosphorylation to glycosylation, despite its size of just 140 amino acids.116-120 Similarly, polyglutamine (polyQ) diseases can have complex origins depending on the length of the polyQ tract and the associated regions.121,122 Such a feature is not intrinsic to just α-synuclein or proteins rich in polyQ tracts but to a large number of disordered proteins arising from the diverse nature of conformations present in equilibrium. Thus, even simple sequence-based approaches can provide a detailed

11

12 Protein Homeostasis Diseases

Figure 1.3  Role of mutations-solvent-environment in determining folding thermodynamic behaviors. All simulations are done at a (semi)quantitative level with the Wako-Saitô-Muñoz-Eaton (WSME) model56 that employs the number of structured residues as the reaction coordinate (RC). (A) Wild-type (WT) Barstar exhibits multistate unfolding behavior99 that can be eliminated by a specific mutation that relieves electrostatic frustration.69 (B) Free-energy profiles, at the respective midpoint conditions for the apo- and holo-forms of bovine lactalbumin,100 with the ion binding mimicked by a D82N mutation that again relieves electrostatic frustration.56 (C, D) Free-energy profiles (panel C) and probability densities (panel D) of the thermosensory protein Cnu that displays population redistribution within the native well upon temperature rise from 278 to 310 K.101 (E, F) Cnu is also sensitive to pH changes in the range between 5 and 7 due to protonation of a partially buried histidine that promotes electrostatic frustration in its vicinity.102 (G, H) Experimentally observed phosphorylation-driven conformational switch between folded- and unfoldedlike conformations in the disordered protein 4E-BP2103 can be quantitatively captured by the WSME model.104 (I, J) Apo- and holo-forms of the bovine Acyl-CoA-binding protein (ACBP) exhibit different packing and electrostatic interaction energy distributions contributing to differences in the free-energy profiles (panel I) that are suggestive of a conformational selection mechanism of binding.70 Note the depopulation of a partially structured state on binding acyl-coA ligand (at RC value ∼67, panel J). (K, L) Disordered CytR populates an excited, folded-like conformation in its native ensemble whose population can be reduced by a specific mutation of P33A.105 Source: Part A, Adapted with permission from Ref. [69]. Copyright (2015), American Chemical Society; Part B, Adapted with permission from Ref. [56]. Copyright (2012), American Chemical Society; Part C, D, Adapted with permission from Ref. [101]. Copyright (2017), American Chemical Society; Part E, F, Adapted with permission from Ref. [102]. Copyright (2018), American Chemical Society; Part G, H, Adapted from Ref. [104] with permission from the PCCP Owner Societies; Part I, J, Adapted from Ref. [70] with permission from the PCCP Owner Societies.

Protein folding: how, why, and beyond

view of the phase space accessible to IDPs based purely on the partitioning of charged residues in sequence.123-125 On the other hand, recent experiments point to how disorder in some proteins is agnostic to sequence composition or the resulting microstructure, pointing to an entropic mechanism for driving allostery,126 as expected from purely thermodynamic considerations.127-129 The heterogeneous landscapes accessible to IDPs in turn translates to an array of binding mechanisms. IDPs can therefore bind their partners on folding or fold upon binding, but simulations hint that a combination of the two mechanisms is also possible.77,130,131 Therefore, reshaping of conformational landscapes on mutations need not be restricted to folded proteins but can have a distinct effect on disordered proteins and regions, affecting the population of partially structured states and dimensions in a nonintuitive manner. For example, a single P33A mutation in the disordered CytR DNA-binding domain reduces the population of the excited folded conformation while weakening its binding affinity to the cognate site (Fig. 1.3K,L).105 Similarly, disordered tails in DNA-binding domains determine their specificity-affinity ratio to varied DNA sequences132,133 and help in one-dimensional sliding-transfer between different DNA strands,134,135 while tuning their charge patterning can effectively determine transcriptional activity.136 Similar to IDPs, the unfolded states of folded proteins have received equivalent attention, as the presence of specific structural elements or interactions can predispose folding reactions and determine stabilities. Unfolded states are more compact under folding conditions compared to unfolding conditions, can have very different structural preferences depending on the nature of the denaturant, and thus can speed up or slow down folding through various nonintuitive effects, aspects that are still being explored in detail.114,115,137-142 Studies on unfolded states have thus challenged the limits of experiments, theoretical arguments, and simulations, promoting a fresh look into numerous aspects, ranging from the assumptions implicit in experimental analyses to the nature of force fields in simulations.141-144 They have had positive knock-on effects in understanding the phase space of IDPs and the approaches employed to characterize them.

Folding, stability, and binding in vivo The protein folding-stability-function measurements discussed earlier are usually carried out under controlled conditions in vitro, thus simplifying the ease of experimentation and interpretation. However, in reality, proteins fold and function in highly crowded nonequilibrium environments within a cell (see Chapter 2). The cytoplasmic composition of prokaryotic or eukaryotic cells are highly varied with numerous ions, solutes, cofactors, and sugars. It is important to realize that the basic interactions that define folding also define the degree of association with different solutes and ligands. Accordingly, the protein conformation is affected not only by steric (excluded-volume) effects but also confinement, specific interactions, and nonspecific binding (mostly transient), all

13

14

Protein Homeostasis Diseases

of which fall under the category of “quinary” interactions.145-153 Moreover, the cell frequently undergoes shape changes that modulate the cytoplasmic composition, concentration of solutes, and hence crowding driven by a variety of factors including osmotic environment, temperature, cell division status, or even when being motile. These features are further compounded by complex structures such as membranes, DNA, and membrane-less organelles that contribute to a radically different microenvironment around them arising from a combination of electrostatic effects, solvation, and enrichment of specific solutes/ions. In-cell effects on folding, stability, and binding are now increasingly being studied through a combination of live-cell NMR (1H-15N or 19F-labeled proteins) and fluorescence microscopy.145,154 As expected, the extent to which a protein is stabilized and destabilized within a cell compared to in vitro conditions is highly protein dependent and has its origins in the nature, composition, and partitioning of charged residues on the protein surface.155 Remarkably, protein stability and kinetics are also affected by phases of the cell cycle and the compartment in which a protein is localized.156,157 Proteins exhibit a larger spread in their folding relaxation times in vivo158 when compared to in vitro conditions, potentially arising from the highly variable environment that (de)stabilizes specific conformations along the folding pathway. These studies suggest that there is an additional selection pressure in the evolution of proteins governed by the quinary environment. This is more evident in recent experiments in which eukaryotic proteins were reported to feel a stickier environment (i.e., slower diffusion) in E. coli while an endogenous counterpart is able to move relatively faster.159 The situation is more complicated when extrapolating IDPs' behavior from in vitro experiments, as they can exhibit very different structures or dynamics in the variable cellular environment. In fact, recent experiments-simulations on disordered DNAbinding proteins propose that the protein is fully folded in the vicinity of DNA160 owing to the large negative electrostatic potential of nucleic acids161 that compensates for unfavorable electrostatic interactions within the protein. Such conditional order has potentially been selected for efficient regulation but highlight the problems one encounters when extrapolating function from in vitro experiments.

“Real proteins” and beyond A large chunk of protein folding studies have been on small single-domain proteins that are either single-gene products or independently folding domains (∼30-100 residues) from larger proteins. The functional units of most proteins are significantly larger, ranging between 200 and 400 residues in length, on average. While the basic physics of folding and binding hold true for larger proteins, there are additional considerations that complicate their studies. In larger proteins, the number of misfolding traps is expected to be proportionally higher with nonlocal effects dominating the landscape. They also exist in multiple quarternary structural states apart from their predisposition to bind

Protein folding: how, why, and beyond

various agonists and antagonists, thus confounding the structure-function relationships. Last, large proteins have independently folded domains that interact in a complex manner, and a simple spectroscopic measure (circular dichroism or fluorescence) will not do justice to the expected intricate unfolding process. True to this expectation, misfolding is common in large proteins with multiple domains, but the extent to which this happens diminishes with increasing sequence divergence of adjacent domains.162,163 These works provide strong evidence for an additional level of complexity that is positively selected for during protein evolution. A majority of experimental studies on large proteins have exploited the power of NMR, H/D exchange, mass spectrometry, and single-molecule fluorescence resonance energy transfer combined with mutational perturbations to study various processes including ligand binding, allostery, and structural transitions. Energy landscape theory and its associated structure-based models have also been successful in capturing numerous features associated with large enzymes and membrane and multidomain proteins.164,165 Intensive MD simulations are increasingly providing vivid insights into structural changes and allosteric transitions occurring in ion channels. However, evidence is accumulating for the role of membranes and their lipid composition in determining and fine-tuning signals, aspects that were earlier not explored in detail in simulations.166,167 Since the time scale of folding scales with protein length, larger proteins will fold and unfold significantly slower; in other words, the associated structural transitions could be in the order of milliseconds or slower, a time scale that is still out of reach for simulations. Finally, there are not many models or computational approaches that can consistently predict protein structures from sequences, sequences that fold to specific structures, their stabilities, and time scales of various local and global structural transitions. Similarly, many diseases have their origins in small changes in amino acid sequences that can have a dramatic effect on solubility, protein-ligand (partner protein, DNA, or small molecules) interactions, and misfolding tendencies. All of these are intimately related to the energetics of interactions within and between proteins, solvent effects, and their relative magnitudes that can be tuned by the environment. Protein folding-stability-function studies from experimental approaches and computational treatments of the same are thus at crossroads where the onus is on continually improving the resolution (spatial and temporal) at which structural events are observed, not just for small, singe-domain proteins but for large functional units or enzymes, and for a detailed quantitative characterization of folding-function landscapes. An engaging crosstalk between experiments and simulations will constantly challenge these approaches while pushing the frontiers of what is possible. Importantly, they can potentially provide elusive solutions to diseases associated with protein (mis)folding that have their origins in missense mutations or environmental conditions that manifest as surprisingly small changes in atomic-level interaction energy terms.

15

16

Protein Homeostasis Diseases

Acknowledgments This chapter is built on decades of work by students, postdoctoral researchers, colleagues, and researchers across the globe. I apologize if I have missed out on obvious references, given the vast literature on protein folding and dynamics spanning varied approaches. A.N.N is a Wellcome Trust/DBT India Alliance Intermediate fellow.

References 1. Ramachandran GN, Ramakrishnan C, Sasisekharan V. Stereochemistry of polypeptide chain configurations. J Mol Biol 1963;7:95–9. 2. Daquino JA, Gomez J, Hilser VJ, Lee KH, Amzel LM, Freire E. The magnitude of the backbone conformational entropy change in protein folding. Proteins 1996;25(2):143–56. 3. Bryngelson JD, Onuchic JN, Socci ND, Wolynes PG. Funnels, pathways, and the energy landscape of protein-folding - a synthesis. Proteins 1995;21(3):167–95. 4. Veitshans T, Klimov D, Thirumalai D. Protein folding kinetics: timescales, pathways and energy landscapes in terms of sequence-dependent properties. Fold Des 1997;2(1):1–22. 5. Sali A, Shakhnovich EI, Karplus M. How does a protein fold? Nature. 1994;369:248–51. 6. Finkelstein AV, Shakhnovich EI. Theory of cooperative transitions in protein molecules. II. Phase diagram for a protein molecule in solution. Biopolymers 1989;28:1681–9. 7. Akmal A, Muñoz V. The nature of the free energy barriers to two-state folding. Proteins 2004;57(1):142– 52. 8. Ferreiro DU, Komives EA, Wolynes PG. Frustration in biomolecules. Q Rev Biophys 2014;47:285–363. 9. Socci ND, Onuchic JN, Wolynes PG. Protein folding mechanisms and the multidimensional folding funnel. Proteins 1998;32(2):136–58. 10. Socci ND, Onuchic JN, Wolynes PG. Diffusive dynamics of the reaction coordinate for protein folding funnels. J Chem Phys 1996;104(15):5860–8. 11. Chan HS, Dill KA. Protein folding in the landscape perspective: Chevron plots and non-Arrhenius kinetics. Proteins 1998;30(1):2–33. 12. Chan HS, Shimizu S, Kaya H. Cooperativity principles in protein folding. Methods Enzymol 2004;380:350–79. 13. Abkevich VI, Gutin AM, Shakhnovich E. Impact of local and non-local interactions on thermodynamics and kinetics of protein folding. J Mol Biol 1995;252:460–71. 14. Klimov DK, Thirumalai D. Cooperativity in protein folding: from lattice models with sidechains to real proteins. Fold Des 1998;3(2):127–39. 15. Baldwin RL. Energetics of protein folding. J Mol Biol 2007;371(2):283–301. 16. Rohl CA, Strauss CEM, Misura KMS, Baker D. Protein structure prediction using rosetta. Methods Enzymol 2004;383:66–93. 17. Rocklin GJ, Chidyausiku TM, Goreshnik I, et al. Global analysis of protein folding using massively parallel design, synthesis, and testing. Science 2017;357(6347):168–75. 18. Szilagyi A, Zavodszky P. Structural differences between mesophilic, moderately thermophilic and extremely thermophilic protein subunits: results of a comprehensive survey. Structure 2000;8(5):493– 504. 19. Risso VA, Gavira JA, Mejia-Carmona DF, Gaucher EA, Sanchez-Ruiz JM. Hyperstability and substrate promiscuity in laboratory resurrections of Precambrian beta-lactamases. J Am Chem Soc 2013;135(8):2899–902. 20. Zou T, Risso VA, Gavira JA, Sanchez-Ruiz JM, Ozkan SB. Evolution of conformational dynamics determines the conversion of a promiscuous generalist into a specialist enzyme. Mol Biol Evol 2015;32(1):132–43. 21. Basak S, Nobrega RP, Tavella D, et al. Networks of electrostatic and hydrophobic interactions modulate the complex folding free energy surface of a designed betaalpha protein. Proc Natl Acad Sci USA 2019;116(14):6806–11.

Protein folding: how, why, and beyond

22. Walters AL, Deka P, Corrent C, et al. The highly cooperative folding of small naturally occurring proteins is likely the result of natural selection. Cell 2007;128:613–24. 23. Sadqi M, de Alba E, Perez-Jimenez R, Sanchez-Ruiz JM, Muñoz V. A designed protein as experimental model of primordial folding. Proc Natl Acad Sci USA 2009;106:4127–32. 24. Jones CM, Henry ER, Hu Y, et al. Fast events in protein-folding initiated by nanosecond laser photolysis. Proc Natl Acad Sci USA 1993;90(24):11860–4. 25. Munoz V, Cerminara M. When fast is better: protein folding fundamentals and mechanisms from ultrafast approaches. Biochem J 2016;473(17):2545–59. 26. Ibarra-Molero B, Naganathan AN, Sanchez-Ruiz JM, Muñoz V. Modern analysis of protein folding by differential scanning calorimetry. Methods Enzymol 2016;567:281–318. 27. Sanchez-Ruiz JM. Probing free-energy surfaces with differential scanning calorimetry. Ann Rev Phys Chem 2011;62:231–55. 28. Neudecker P, Lundstrom P, Kay LE. Relaxation dispersion NMR spectroscopy as a tool for detailed studies of protein folding. Biophys J 2009;18:2045–54. 29. Tzeng SR, Kalodimos CG. Protein dynamics and allostery: an NMR view. Curr Opin Struct Biol 2011;21(1):62–7. 30. Wand AJ. The dark energy of proteins comes to light: conformational entropy and its role in protein function revealed by NMR relaxation. Curr Opin Struct Biol 2013;23(1):75–81. 31. Igumenova TI, Frederick KK, Wand AJ. Characterization of the fast dynamics of protein amino acid side chains using NMR relaxation in solution. Chem Rev 2006;106(5):1672–99. 32. Englander SW, Mayne L, Krishna MMG. Protein folding and misfolding: mechanism and principles. Q Rev Biophys 2007;40(4):287–326. 33. Englander SW. Protein folding intermediates and pathways studied by hydrogen exchange. Ann Rev Biophys Biomol Struct 2000;29:213–38. 34. Schuler B, Hofmann H. Single-molecule spectroscopy of protein folding dynamics-expanding scope and timescales. Curr Opin Struct Biol 2013;23:36–47. 35. Chung HS, Eaton WA. Protein folding transition path times from single molecule FRET. Curr Opin Struct Biol 2018;48:30–9. 36. Zoldak G, Rief M. Force as a single molecule probe of multidimensional protein energy landscapes. Curr Opin Struct Biol 2013;23(1):48–57. 37. Alegre-Cebollada J, Perez-Jimenez R, Kosuri P, Fernandez JM. Single-molecule force spectroscopy approach to enzyme catalysis. J Biol Chem 2010;285(25):18961–6. 38. Takahashi S, Kamagata K, Oikawa H. Where the complex things are: single molecule and ensemble spectroscopic investigations of protein folding dynamics. Curr Opin Struct Biol 2016;36:1–9. 39. Lumry R, Biltonen RL, Brandts JF. Validity of the “two-state” hypothesis for conformational transitions of proteins. Biopolymers 1966;4:917–44. 40. Naganathan AN, Perez-Jimenez R, Sanchez-Ruiz JM, Muñoz V. Robustness of downhill folding: guidelines for the analysis of equilibrium folding experiments on small proteins. Biochemistry 2005;44(20):7435–49. 41. Wako H, Saito N. Statistical mechanical theory of protein conformation. 1. General considerations and application to homopolymers. J Phys Soc Japan 1978;44(6):1931–8. 42. Ikegami A. Statistical thermodynamics of proteins and protein denaturation. Adv Chem Phys 1981;46:363–413. 43. Abe H, Go N. Noninteracting local-structure model of folding and unfolding transition in globular proteins. II. Application to two-dimensional lattice proteins. Biopolymers 1981;20(5):1013–31. 44. Best RB. Atomistic molecular simulations of protein folding. Curr Opin Struct Biol 2012;22: 52–61. 45. Piana S, Klepeis JL, Shaw DE. Assessing the accuracy of physical models used in protein-folding simulations: quantitative evidence from long molecular dynamics simulations. Curr Opin Struct Biol 2014;24:98–105. 46. Leone V, Marinelli F, Carloni P, Parrinello M. Targeting biomolecular flexibility with metadynamics. Curr Opin Struct Biol 2010;20:148–54. 47. Doshi U, Hamelberg D. Towards fast, rigorous and efficient conformational sampling of biomolecules: advances in accelerated molecular dynamics. Biochim Biophys Acta 2015;1850:878–88.

17

18

Protein Homeostasis Diseases

48. Perez A, Morrone JA, Simmerling C, Dill KA. Advances in free-energy-based simulations of protein folding and ligand binding. Curr Opin Struct Biol 2016;36:25–31. 49. Chodera JD, Noe F. Markov state models of biomolecular conformational dynamics. Curr Opin Struct Biol 2014;25:135–44. 50. Hyeon C, Thirumalai D. Capturing the essence of folding and functions of biomolecules using coarsegrained models. Nat Commun 2011;2:487. 51. Chan HS, Zhang Z, Wallin S, Liu Z. Cooperativity, Local-Nonlocal Coupling, and Nonnative Interactions: Principles of Protein Folding from Coarse-Grained Models. In: Leone SR, Cremer PS, Groves JT, Johnson MA, eds. Ann. Rev. Phys. Chem. Vol 622011:301-326. 52. Naganathan AN. Coarse-grained models of protein folding as detailed tools to connect with experiments. WIREs Comput Mol Sci 2013;3:504–14. 53. Kmiecik S, Gront D, Kolinski M, Wieteska L, Dawid AE, Kolinski A. Coarse-grained protein models and their applications. Chem Rev 2016;116(14):7898–936. 54. Wako H, Saito N. Statistical mechanical theory of protein conformation. 2. Folding pathway for protein. J Phys Soc Japan 1978;44(6):1939–45. 55. Muñoz V, Eaton WA. A simple model for calculating the kinetics of protein folding from threedimensional structures. Proc Natl Acad Sci USA 1999;96(20):11311–6. 56. Naganathan AN. Predictions from an ising-like statistical mechanical model on the dynamic and thermodynamic effects of protein surface electrostatics. J Chem Theory Comput 2012;8(11):4646–56. 57. Inanami T, Terada TP, Sasai M. Folding pathway of a multidomain protein depends on its topology of domain connectivity. Proc Natl Acad Sci USA 2014;111(45):15969–74. 58. Naganathan AN. Predictive modeling of protein folding thermodynamics, mutational effects and freeenergy landscapes. Proc Indian Natl Sci Acad 2016;82:1211–28. 59. Taketomi H, Ueda Y, Go N. Studies on protein folding, unfolding and fluctuations by computersimulation. 1. Effect of specific amino-acid sequence represented by specific inter-unit interactions. Inter J Prot Pep Res 1975;7(6):445–59. 60. Best RB, Hummer G, Eaton WA. Native contacts determine protein folding mechanisms in atomistic simulations. Proc Natl Acad Sci USA 2013;110(44):17874–9. 61. Truong HH, Kim BL, Schafer NP, Wolynes PG. Predictive energy landscapes for folding membrane protein assemblies. J Chem Phys 2015;143(24):243101. 62. Li W, Wolynes PG, Takada S. Frustration, specific sequence dependence, and nonlinearity in largeamplitude fluctuations of allosteric proteins. Proc Natl Acad Sci USA 2011;108(9):3504–9. 63. Ferreiro DU, Walczak AM, Komives EA, Wolynes PG. The energy landscapes of repeat-containing proteins: topology, cooperativity, and the folding funnels of one-dimensional architectures. PLOS Comp Biol 2008;4(5):e1000070. 64. Oliveberg M, Wolynes PG. The experimental survey of protein-folding energy landscapes. Q Rev Biophys 2005;38(3):245–88. 65. Giri Rao VH, Gosavi S. Using the folding landscapes of proteins to understand protein function. Curr Opin Struct Biol 2016;36:67–74. 66. Levy Y, Onuchic JN. Mechanisms of protein assembly: lessons from minimalist models. Acc Chem Res 2006;39:135–42. 67. Whitford PC, Onuchic JN. What protein folding teaches us about biological function and molecular machines. Curr Opin Struct Biol 2015;30:57–62. 68. Henry ER, Eaton WA. Combinatorial modeling of protein folding kinetics: free energy profiles and rates. Chem Phys 2004;307:163–85. 69. Naganathan AN, Sanchez-Ruiz JM, Munshi S, Suresh S. Are protein folding intermediates the evolutionary consequence of functional constraints? J Phys Chem B 2015;119:1323–33. 70. Munshi S, Naganathan AN. Imprints of function on the folding landscape: functional role for an intermediate in a conserved eukaryotic binding protein. Phys Chem Chem Phys 2015;17(16):11042–52. 71. Freiberger MI, Guzovsky AB, Wolynes PG, Parra RG, Ferreiro DU. Local frustration around enzyme active sites. Proc Natl Acad Sci USA 2019;116(10):4037–43. 72. Jahn TR, Parker MJ, Homans SW, Radford SE. Amyloid formation under physiological conditions proceeds via a native-like folding intermediate. Nat Struct Mol Biol 2006;13(3):195–201. 73. De Simone A, Dhulesia A, Soldi G, et al. Experimental free energy surfaces reveal the mechanisms of maintenance of protein solubility. Proc Natl Acad Sci USA 2011;108(52):21057–62.

Protein folding: how, why, and beyond

74. Sengupta I, Udgaonkar J. Monitoring site-specific conformational changes in real-time reveals a misfolding mechanism of the prion protein. eLife 2019;8:e44698. 75. Sivanandan S, Naganathan AN. A disorder-induced domino-like destabilization mechanism governs the folding and functional dynamics of the repeat protein IkBα. PLOS Comput Biol 2013;9:e1003403. 76. Aviram HY, Pirchi M, Barak Y, Riven I, Haran G. Two states or not two states: single-molecule folding studies of protein L. J Chem Phys 2018;148(12):123303. 77. Naganathan AN, Orozco M. The conformational landscape of an intrinsically disordered DNAbinding domain of a transcription regulator. J Phys Chem B 2013;117:13842–50. 78. Naganathan AN. Modulation of allosteric coupling by mutations: from protein dynamics and packing to altered native ensembles and function. Curr Opin Struct Biol 2019;54:1–9. 79. Guarnera E, Berezovsky IN. On the perturbation nature of allostery: sites, mutations, and signal modulation. Curr Opin Struct Biol 2019;56:18–27. 80. Tokuriki N, Tawfik DS. Stability effects of mutations and protein evolvability. Curr Opin Struct Biol 2009;19(5):596–604. 81. Sikosek T, Chan HS. Biophysics of protein evolution and evolutionary protein biophysics. J R Soc Interface 2014;11(100):20140419. 82. Risso VA, Sanchez-Ruiz JM, Ozkan SB. Biotechnological and protein-engineering implications of ancestral protein resurrection. Curr Opin Struct Biol 2018;51:106–15. 83. Petrovic D, Risso VA, Kamerlin SCL, Sanchez-Ruiz JM. Conformational dynamics and enzyme evolution. J R Soc Interface 2018;15(144). 84. Gupta K, Varadarajan R. Insights into protein structure, stability and function from saturation mutagenesis. Curr Opin Struct Biol 2018;50:117–25. 85. Bershtein S, Serohijos AW, Shakhnovich EI. Bridging the physical scales in evolutionary biology: from protein sequence space to fitness of organisms and populations. Curr Opin Struct Biol 2017;42:31–40. 86. Serohijos AWR, Shakhnovich EI. Merging molecular mechanism and evolution: theory and computation at the interface of biophysics and evolutionary population genetics. Curr Opin Struct Biol 2014;26:84–91. 87. Tokuriki N, Oldfield CJ, Uversky VN, Berezovksy IN, Tawfik DS. Do viral proteins possess unique biophysical features? Trends Biochem Sci 2008;34:53–9. 88. Horovitz A, Fleisher RC, Mondal T. Double-mutant cycles: new directions and applications. Curr Opin Struct Biol 2019;58:10–7. 89. Goldsmith M, Tawfik DS. Directed enzyme evolution: beyond the low-hanging fruit. Curr Opin Struct Biol 2012;22(4):406–12. 90. Rollins NJ, Brock KP, Poelwijk FJ, et al. Inferring protein 3D structure from deep mutation scans. Nat Genet 2019;51(7):1170–6. 91. Schmiedel JM, Lehner B. Determining protein structures using deep mutagenesis. Nat Genet 2019;51(7):1177–86. 92. Rajasekaran N, Sekhar A, Naganathan AN. A universal pattern in the percolation and dissipation of protein structural perturbations. J Phys Chem Lett 2017;8(19):4779–84. 93. Roche J, Caro JA, Dellarole M, et al. Structural, energetic, and dynamic responses of the native state ensemble of staphylococcal nuclease to cavity-creating mutations. Proteins 2013;81(6):1069–80. 94. Whitley MJ, Zhang J, Lee AL. Hydrophobic core mutations in CI2 globally perturb fast side-chain dynamics similarly without regard to position. Biochemistry 2008;47(33):8566–76. 95. Ben-David M, Huang H, Sun MGF, et al. Allosteric modulation of binding specificity by alternative packing of protein cores. J Mol Biol 2019;431(2):336–50. 96. Rajasekaran N, Suresh S, Gopi S, Raman K, Naganathan AN. A general mechanism for the propagation of mutational effects in proteins. Biochemistry 2017;56:294–305. 97. Haririnia A, Verma R, Purohit N, et al. Mutations in the hydrophobic core of ubiquitin differentially affect its recognition by receptor proteins. J Mol Biol 2008;375(4):979–96. 98. Bouvignies G, Vallurupalli P, Hansen DF, et al. Solution structure of a minor and transiently formed state of a T4 lysozyme mutant. Nature 2011;477(7362):111–4. 99. Sarkar SS, Udgaonkar JB, Krishnamoorthy G. Unfolding of a small protein proceeds via dry and wet globules and a solvated transition state. Biophys J 2013;105:2392–402. 100. Halskau O, Perez-Jimenez R, Ibarra-Molero B, et al. Large-scale modulation of thermodynamic protein folding barriers linked to electrostatics. Proc Natl Acad Sci USA 2008;105(25):8625–30.

19

20

Protein Homeostasis Diseases

101. Narayan A, Campos LA, Bhatia S, Fushman D, Naganathan AN. Graded structural polymorphism in a bacterial thermosensor protein. J Am Chem Soc 2017;139:792–802. 102. Narayan A, Naganathan AN. Switching protein conformational substates by protonation and mutation. J Phys Chem B 2018;122:11039–47. 103. Bah A, Vernon RM, Siddiqui Z, et al. Folding of an intrinsically disordered protein by phosphorylation as a regulatory switch. Nature 2015;519(7541):106–U240. 104. Gopi S, Rajasekaran N, Singh A, Ranu S, Naganathan AN. Energetic and topological determinants of a phosphorylation-induced disorder-to-order protein conformational switch. Phys Chem Chem Phys 2015;17:27264–9. 105. Munshi S, Rajendran D, Naganathan AN. Entropic control of an excited folded-like conformation in a disordered protein ensemble. J Mol Biol 2018;430(17):2688–94. 106. Uversky VN. What does it mean to be natively unfolded? Eur J Biochem 2002;269(1):2–12. 107. Uversky VN, Gillespie JR, Fink AL. Why are “natively unfolded” proteins unstructured under physiologic conditions? Proteins: Struct Funct Genet 2000;41(3):415–27. 108. van der Lee R, Buljan M, Lang B, et al. Classification of intrinsically disordered regions and proteins. Chem Rev 2014;114:6589–631. 109. Babu MM, van der Lee R, de Groot NS, Gsponer J. Intrinsically disordered proteins: regulation and disease. Curr Opin Struct Biol 2011;21:432–40. 110. Dyson HJ, Wright PE. Intrinsically unstructured proteins and their functions. Nature Rev Mol Cell Biol 2005;6(3):197–208. 111. Bah A, Forman-Kay JD. Modulation of intrinsically disordered protein function by post-translational modifications. J Biol Chem 2016;291(13):6696–705. 112. Uversky VN. A decade and a half of protein intrinsic disorder: biology still waits for physics. Protein Sci 2013;22:693–724. 113. Peran I, Holehouse AS, Carrico IS, Pappu RV, Bilsel O, Raleigh DP. Unfolded states under folding conditions accommodate sequence-specific conformational preferences with random coil-like dimensions. Proc Natl Acad Sci USA 2019;116(25):12301–10. 114. Meng WL, Lyle N, Luan BW, Raleigh DP, Pappu RV. Experiments and simulations show how longrange contacts can form in expanded unfolded proteins with negligible secondary structure. Proc Natl Acad Sci USA 2013;110(6):2123–8. 115. Meng W, Luan B, Lyle N, Pappu RV, Raleigh DP. The denatured state ensemble contains significant local and long-range structure under native conditions: analysis of the N-terminal domain of ribosomal protein L9. Biochemistry 2013;52(15):2662–71. 116. Breydo L, Wu JW, Uversky VN. Alpha-synuclein misfolding and Parkinson’s disease. Biochimica et biophysica acta 2012;1822(2):261–85. 117. Uversky VN, Eliezer D. Biophysics of Parkinson’s disease: structure and aggregation of alpha-synuclein. Curr Protein Peptide Sci 2009;10(5):483–99. 118. Abedini A, Raleigh DP. A critical assessment of the role of helical intermediates in amyloid formation by natively unfolded proteins and polypeptides. Protein Eng Des Sel 2009;22(8):453–9. 119. Uversky VN, Li J, Fink AL. Evidence for a partially folded intermediate in alpha-synuclein fibril formation. J Biol Chem 2001;276(14):10737–44. 120. Li J, Uversky VN, Fink AL. Conformational behavior of human alpha-synuclein is modulated by familial Parkinson’s disease point mutations A30P and A53T. Neurotoxicology 2002;23(4-5):553–67. 121. Matos CA, Almeida LP, Nobrega C. Proteolytic cleavage of polyglutamine disease-causing proteins: revisiting the toxic fragment hypothesis. Curr Pharm Des 2017;23(5):753–75. 122. Hoffner G, Djian P. Polyglutamine aggregation in huntington disease: does structure determine toxicity? Mol Neurobiol 2015;52(3):1297–314. 123. Das RK, Ruff KM, Pappu RV. Relating sequence encoded information to form and function of intrinsically disordered proteins. Curr Opin Struct Biol 2015;32:102–12. 124. Das RK, Pappu RV. Conformations of intrinsically disordered proteins are influenced by linear sequence distributions of oppositely charged residues. Proc Natl Acad Sci USA 2013;110(33):13392–7. 125. Mao AH, Crick SL, Vitalis A, Chicoine CL, Pappu RV. Net charge per residue modulates conformational ensembles of intrinsically disordered proteins. Proc Natl Acad Sci USA 2010;107:8183–8.

Protein folding: how, why, and beyond

126. Keul ND, Oruganty K, Schaper Bergman ET, et al. The entropic force generated by intrinsically disordered segments tunes protein function. Nature 2018;563(7732):584–8. 127. Motlagh HN, Wrabl JO, Li J, Hilser VJ. The ensemble nature of allostery. Nature 2014;508(7496):331– 9. 128. Hilser VJ, Thompson EB. Intrinsic disorder as a mechanism to optimize allosteric coupling in proteins. Proc Natl Acad Sci USA 2007;104(20):8311–5. 129. Rajasekaran N, Gopi S, Narayan A, Naganathan AN. Quantifying protein disorder through measures of excess conformational entropy. J Phys Chem B 2016;120:4341–50. 130. Paul F, Noe F, Weikl TR. Identifying conformational-selection and induced-fit aspects in the binding-induced folding of PMI from Markov state modeling of atomistic simulations. J Phys Chem B 2018;122(21):5649–56. 131. Mollica L, Bessa LM, Hanoulle X, Jensen MR, Blackledge M, Schneider R. Binding mechanisms of intrinsically disordered proteins: theory, simulation, and experiment. Front Mol Biosci 2016;3:52. 132. Vuzman D, Levy Y. Intrinsically disordered regions as affinity tuners in protein-DNA interactions. Mol Biosystems 2012;8(1):47–57. 133. Marcovitz A, Levy Y. Frustration in protein-DNA binding influences conformational switching and target search kinetics. Proc Natl Acad Sci USA 2011;108(44):17957–62. 134. Itoh Y, Murata A, Takahashi S, Kamagata K. Intrinsically disordered domain of tumor suppressor p53 facilitates target search by ultrafast transfer between different DNA strands. Nucleic Acids Res 2018;46(14):7261–9. 135. Subekti DRG, Murata A, Itoh Y, et al. The Disordered Linker in p53 participates in nonspecific binding to and one-dimensional sliding along DNA revealed by single-molecule fluorescence measurements. Biochemistry 2017;56(32):4134–44. 136. Sherry KP, Das RK, Pappu RV, Barrick D. Control of transcriptional activity by design of charge patterning in the intrinsically disordered RAM region of the Notch receptor. Proc Natl Acad Sci USA 2017;114(44):E9243–52. 137. Stenzoski NE, Luan B, Holehouse AS, Raleigh DP. The unfolded state of the C-terminal domain of L9 expands at low but not at elevated temperatures. Biophys J 2018;115(4):655–63. 138. Schuler B, Soranno A, Hofmann H, Nettels D. Single-molecule FRET spectroscopy and the polymer physics of unfolded and intrinsically disordered proteins. Annu Rev Biophys 2016;45:207–31. 139. Borgia A, Zheng W, Buholzer K, et al. Consistent view of polypeptide chain expansion in chemical denaturants from multiple experimental methods. J Am Chem Soc 2016;138(36):11714–26. 140. Riback JA, Bowman MA, Zmyslowski AM, et al. Innovative scattering analysis shows that hydrophobic disordered proteins are expanded in water. Science 2017;358(6360):238–41. 141. Song J, Gomes GN, Shi T, Gradinaru CC, Chan HS. Conformational heterogeneity and FRET data interpretation for dimensions of unfolded proteins. Biophys J 2017;113(5):1012–24. 142. Thirumalai D, Samanta HS, Maity H, Reddy G. Universal nature of collapsibility in the context of protein folding and evolution. Trends Biochem Sci 2019;44(8):675–87. 143. Henriques J, Cragnell C, Skepo M. Molecular dynamics simulations of intrinsically disordered proteins: force field evaluation and comparison with experiment. J Chem Theory Comput 2015;11(7):3420–31. 144. Narayan A, Bhattacharjee K, Naganathan AN. Thermally versus chemically denatured protein states. Biochemistry 2019;58(21):2519–23. 145. Davis CM, Gruebele M, Sukenik S. How does solvation in the cell affect protein folding and binding? Curr Opin Struct Biol 2018;48:23–9. 146. Sukenik S, Salam M, Wang Y, Gruebele M. In-cell titration of small solutes controls protein stability and aggregation. J Am Chem Soc 2018;140(33):10497–503. 147. Feng R, Gruebele M, Davis CM. Quantifying protein dynamics and stability in a living organism. Nat Commun 2019;10(1):1179. 148. Guseman AJ, Perez Goncalves GM, Speer SL, Young GB, Pielak GJ. Protein shape modulates crowding effects. Proc Natl Acad Sci USA 2018;115(43):10965–70. 149. Monteith WB, Cohen RD, Smith AE, Guzman-Cisneros E, Pielak GJ. Quinary structure modulates protein stability in cells. Proc Natl Acad Sci USA 2015;112(6):1739–42. 150. Smith AE, Zhou LZ, Gorensek AH, Senske M, Pielak GJ. In-cell thermodynamics and a new role for protein surfaces. Proc Natl Acad Sci USA 2016;113(7):1725–30.

21

22

Protein Homeostasis Diseases

151. Sukenik S, Ren P, Gruebele M. Weak protein-protein interactions in live cells are quantified by cellvolume modulation. Proc Natl Acad Sci USA 2017;114(26):6776–81. 152. Cohen RD, Pielak GJ. A cell is more than the sum of its (dilute) parts: a brief history of quinary structure. Protein Sci 2017;26(3):403–13. 153. Cheng K, Wu Q, Zhang Z, Pielak GJ, Liu M, Li C. Crowding and confinement can oppositely affect protein stability. Chemphyschem 2018;19(24):3350–5. 154. Smith AE, Zhang Z, Pielak GJ, Li C. NMR studies of protein folding and binding in cells and cell-like environments. Curr Opin Struct Biol 2015;30:7–16. 155. Guseman AJ, Speer SL, Perez Goncalves GM, Pielak GJ. Surface charge modulates protein-protein interactions in physiologically relevant environments. Biochemistry 2018;57(11):1681–4. 156. Wirth AJ, Platkov M, Gruebele M. Temporal variation of a protein folding energy landscape in the cell. J Am Chem Soc 2013;135:19215–21. 157. Dhar A, Girdhar K, Singh D, Gelman H, Ebbinghaus S, Gruebele M. Protein stability and folding kinetics in the nucleus and endoplasmic reticulum of eucaryotic cells. Biophys J 2011;101(2):421–30. 158. Dhar A, Ebbinghaus S, Shen Z, Mishra T, Gruebele M. The diffusion coefficient for PGK folding in eukaryotic cells. Biophys J 2010;99(9):L69–71. 159. Mu X, Choi S, Lang L, et al. Physicochemical code for quinary protein interactions in Escherichia coli. Proc Natl Acad Sci USA 2017;114(23):E4556–e4563. 160. Munshi S, Gopi S, Asampille G, et al. Tunable order-disorder continuum in protein-DNA interactions. Nucleic Acids Res 2018;46(17):8700–9. 161. Allred BE, Gebala M, Herschlag D. Determination of ion atmosphere effects on the nucleic acid electrostatic potential and ligand association using AH(+).C wobble formation in double-stranded DNA. J Am Chem Soc 2017;139(22):7540–8. 162. Borgia A, Kemplen KR, Borgia MB, et al. Transient misfolding dominates multidomain protein folding. Nat Commun 2015;6:8861. 163. Borgia MB, Borgia A, Best RB, et al. Single-molecule fluorescence reveals sequence-specific misfolding in multidomain proteins. Nature 2011;474(7353):662–5. 164. Lu W, Schafer NP, Wolynes PG. Energy landscape underlying spontaneous insertion and folding of an alpha-helical transmembrane protein into a bilayer. Nat Commun 2018;9(1):4949. 165. Kim BL, Schafer NP, Wolynes PG. Predictive energy landscapes for folding alpha-helical transmembrane proteins. Proc Natl Acad Sci USA 2014;111(30):11031–6. 166. Chavent M, Duncan AL, Sansom MS. Molecular dynamics simulations of membrane proteins and their interactions: from nanoscale to mesoscale. Curr Opin Struct Biol 2016;40:8–16. 167. Pliotas C, Naismith JH. Spectator no more, the role of the membrane in regulating ion channel function. Curr Opin Struct Biol 2017;45:59–66.

CHAPTER 2

Protein homeostasis and disease Angel L. Pey

Department of Physical Chemistry, Faculty of Sciences, University of Granada, Granada, Spain

Outline Protein folding in vitro and in vivo Effects of intracellular milieu composition on protein folding, misfolding, and stability in vivo The first steps of in vivo folding and misfolding in the ribosomes: cotranslational versus posttranslational processes Protein homeostasis networks Human misfolding diseases Loss-of-function diseases Gain-of-toxic function diseases References

23 26 27 28 31 32 32 34

Abbreviations ALP  autophagy-lysosomal pathway ER  endoplasmic reticulum GOF  gain-of-function HSF1  heat-shock transcription factor 1 HSP  heat-shock protein HSR  heat-shock response LOF  loss-of-function PN  protein homeostasis network UPR  unfolded protein response UPS  ubiquitin-dependent proteasomal degradation system

Protein folding in vitro and in vivo Proteins carry out a tremendous variety of functions, and this multifunctionality is deeply rooted into the structure and energetics of their native state. As introduced in Chapter 1, protein folding relies on the acquisition of native-state interactions (particularly burial of hydrophobic surface from the solvent) strongly coupled to the loss of conformational entropy. In this journey to the native structure, and even once the protein is folded, the polypeptide chain can adopt multiple conformations, an issue quantitatively and conceptually depicted in the protein energy landscape (Fig. 2.1A). In this landscape, the native

Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00002-6 Copyright © 2020 Elsevier Inc. All rights reserved.

23

24

Protein Homeostasis Diseases

Figure 2.1  Links between protein energy landscape, function, and misfolding. (A) Energy landscape (free-energy surface) for the human protein NAD(P)H:quinone oxidoreductase 1 (NQO1). Two folding intermediates (I1 and I2) are shown in which different regions of the protein (N- and C-terminal parts) are folded. (B) Mutational effects on the kinetic and thermodynamic stability of the native state and partially folded intermediates potentially leading to misfolding or inactivation. Source: Reproduced with permission from Ref. [3]. Copyright (2019) from Elsevier.

Protein homeostasis and disease

state often lies as a free-energy minimum in equilibrium with nonnative states, including partially folded states, which can have significant populations (Fig. 2.1A).1,2 The transition rates between different conformational states in this landscape mainly depend on the energy barrier that must be crossed for a given transition to occur. In some cases, these barriers are small, and thus, states readily interconvert under native conditions; this conformational dynamics is critical for protein function, involving, for instance, equilibrium between catalytically competent and noncompetent states (the basis of allosteric regulation). In other cases, kinetic barriers are large and thus make certain of these states being kinetically stable. Concomitant to the complexity of the energy landscape of proteins, the thermodynamics and kinetics that determine their relative population of different conformations in vitro or in vivo are expected to be complex. Although the native state is marginally stable (usually a few kcal·mol−1 more stable than the unfolded state), the folding process is fast, occurring in time scales from milliseconds to minutes.1,4 Thus, the population of nonnative states is usually low compared to that of the native state due to their higher free energy, although these states can be critical to understanding the protein function and life cycle.5 For instance, kinetic traps populated during the folding process may slowly proceed to the native state, exposing a hydrophobic surface prone to self-association and forming protein aggregates.6 In general, any process that prevents the protein from reaching the native state can be referred to as protein misfolding. Thus, protein folding and misfolding are competing processes inherently imprinted in the complex conformational landscapes of natural proteins. The basic relationships between protein sequence, energetics, stability, and function make these abovementioned protein features very sensitive to changes in the protein sequence (e.g., a single amino acid change; Chapter 1), and thus, these changes may promote protein misfolding. In an evolutionary context, single amino acid changes are basic tools to fine-tune protein function and stability or to promote acquisition of new functions.7,8 In this chapter, we will particularly focus on how single amino changes may promote protein misfolding, leading to disease by affecting the ability of proteins to work properly or causing them to become toxic in a living organism.9 From a very simple perspective, mutations may alter the equilibrium and kinetic relationships between the native and partially folded states, thus making either the native state more sensitive to misfolding and inactivation (Fig. 2.1B). A quantitative understanding of the relationships between protein sequence (and the variations thereof), energetics (effects on the protein energy landscape), stability, and function represents a big challenge for research fields such as biology, chemistry, biotechnology, and biomedicine. Remarkably, protein folding, misfolding, stability, and function may differ conceptually and quantitatively in vitro and in vivo. First, proteins are synthesized in the ribosomes vectorially from their N termini: the nascent chain grows tethered to the ribosome as this moves along the mRNA. Consequently, folding and misfolding are important issues even before the full protein has been synthesized. Accordingly, we can

25

26

Protein Homeostasis Diseases

envision that folding and misfolding compete before or after the full protein has been synthesized (cotranslationally versus posttranslationally). In addition, we find very high concentrations of macromolecules as well as low-molecular-weight compounds in the crowded intracellular milieu that may interact specifically or nonspecifically with the polypeptide chain, thus affecting its propensity to fold or misfold. And last but not least, to reach appropriate levels of folded and functional proteins and to prevent accumulation of misfolded species that could be harmful for cellular homeostasis and viability, complex intracellular machineries have naturally evolved to preserve protein homeostasis, referred to as protein homeostasis networks (PNs). These PNs are capable of shaping the energy landscape of proteins, altering protein folding and misfolding thermodynamics and kinetics in many different manners.

Effects of intracellular milieu composition on protein folding, misfolding, and stability in vivo Water makes up 60%–70% of the cellular mass/volume, whereas the rest of the cellular weight is made up of cosolutes such as macromolecules (proteins, nucleic acids, lipids) and low-molecular-mass solutes.1,10 This complex intracellular composition is not homogeneous within a single cell; it differs between cell types and depends on environmental conditions.10 Although these cosolutes may not radically alter the fundamental physics of protein folding, they can effectively modulate the protein energy landscape and thus affect protein function, folding, misfolding, and stability.1 Low-molecular-weight compounds can bind specifically and tightly to the protein native state (acting, for instance, as substrates, cofactors, or inhibitors), thus preferentially stabilizing binding-competent states, with a concomitant and large effect on protein folding and stability in vivo.10–12 Cosolutes can affect the protein energy landscape as a result of nonspecific but preferential interactions with certain conformational states. Generally, macromolecules and low-molecular-weight compounds may affect the amount and extent of the interaction of intracellular water with proteins, with remarkable effects on the protein energy landscape due to changes in solvation, thus affecting folding and stability.1,4,10 Studies in vitro and in living cells combined with theoretical models have supported that different types of cosolutes (macromolecules versus low-molecular-weight compounds) may have fundamentally different, and sometimes opposing, effects on protein folding and stability.4,10 Macromolecules are the cause of macromolecular crowding effects by which those protein conformations that occupy less volume are stabilized, and thus, these often stabilize the native state of proteins1,4 and/or promote self-association of partially folded states into aggregates.13 Cosolutes, particularly those with low molecular mass, can also produce nonspecific electrostatic and hydrophobic interactions with protein backbones and side chains, preferentially stabilizing unfolded or folded conformations depending on the nature and strength of these interactions.1,4,10 A well-known example of low-molecularweight cosolutes that stabilize protein native states are naturally occurring osmolytes,

Protein homeostasis and disease

which are synthesized and kept at high cellular concentrations in organisms living in harsh environments to prevent protein misfolding.1,14 Another curious example is the binding of small inorganic anions to natural ligand binding pockets that may modulate the stability of the native protein by inhibiting binding of largely stabilizing ligands such as substrates and cofactors.15 Thus, a complex picture on the effects of intracellular milieu composition on protein folding, misfolding, and stability emerges in which the net effects of this composition depends on the protein under study, cell type, subcellular location, and the perturbation exerted on the conformational landscape.1,4,10,16,17

The first steps of in vivo folding and misfolding in the ribosomes: cotranslational versus posttranslational processes A first layer of regulation of protein folding and misfolding in vivo, in which the intracellular milieu composition and PNs begin to operate, appears in the ribosomal synthesis of proteins. In a simple manner, ribosomal protein synthesis consists in three steps: initiation, in which a stable complex at the initiation codon is formed, elongation of the polypeptide chain as the ribosome moves along the actively translated mRNA and the protein remains tethered to the ribosome, and termination of synthesis at the codon stop.18 During ribosomal synthesis, the conformational landscape that is explored by the polypeptide as it grows may differ from that of the protein when is fully synthesized, consequently affecting its folding and misfolding propensities as well as their modulation by the intracellular milieu and PNs.19 The local environment of the ribosome can also significantly modulate protein stability once it has been fully synthesized.21 Therefore, cotranslational and posttranslational processes not present in in vitro studies can be critical to determine protein folding and misfolding in vivo. We can refer to, as cotranslational, any process affecting the properties of the nascent polypeptide before its full synthesis and release from the ribosome.22 These include those that may have an impact on the folding landscape of the protein as it is being synthesized, such as the translation rate of the mRNA, the threading of the polypeptide through the ribosome exit tunnel, and the interaction of this with elements recruited to the macromolecular ribosomal machinery, such as molecular chaperones and protein-modifying enzymes (e.g., those leading to ubiquitination or glycosylation).22 As found for in vitro protein folding (see Chapter 1), cotranslational processes operate in a wide range of time scales, from milliseconds to minutes and hours.22 Cotranslational processes can have multiple effects on the folding efficiency, assembly, misfolding, and function of proteins23; some of these are seemingly counterintuitive. For instance, codon translation rates are slow in mammalian cells (about 5 amino acids·s−1), and these can vary by one order of magnitude between synonymous codons encoding for the same amino acid.18 These differences in translation rates can strongly influence protein folding, misfolding, assembly, and subcellular targeting of proteins,23 leading to nonequilibrium folding dynamics due to the widely different time scales for protein synthesis and folding inside cells. In this

27

28

Protein Homeostasis Diseases

sense, cotranslational processes acting on protein folding and misfolding in vivo prior to the relevant rate-limiting step(s) may strongly affect the amount of active and properly targeted protein inside the cell,23 including the acquisition of secondary and tertiary structure by the nascent chain inside the ribosomal tunnel and its interaction with molecular chaperones and macromolecular crowding agents.22 Thus, cotranslational processes may also influence protein folding pathways, particularly in multidomain proteins in which individual domains can fold independently,6 and preventing the population of aggregation-prone intermediates.19,20,24 Indeed, proteins naturally folding in a posttranslational fashion can switch to a cotranslationally folding scenario just by introducing slow-translating synonymous codons.25 Interestingly, many hereditary misfolding diseases are caused by single amino changes (missense mutations) that impair protein folding and promote protein misfolding, although little attention has been paid to the potential effect of synonymous mutations through changes in translation rates and downstream effects on protein folding and assembly.23 These effects have been rarely explored when synonymous mutations are detected in patients or in genome-wide sequencing studies.

Protein homeostasis networks The term protein homeostasis describes an adequate balance of the ∼20,000 different protein species contained in the mammalian proteome to preserve cellular functions.26–28 Cells contain complex networks comprising over a thousand proteins (including chaperones and cochaperones and degradation protein machineries) that operate in a coordinated fashion, maintaining protein homeostasis (referred to as the PN).26,29 One of the main functions of the PN is preventing the accumulation of protein aggregates formed upon self-association of protein-misfolded molecules (often partially folded states) in the crowded cellular milieu that may compromise cellular function and viability.27,29,30 These partially folded states are commonly populated in the folding of multidomain and oligomeric proteins that constitute a large fraction of the eukaryotic proteome, and thus, efficient protein folding often requires the assistance of elements of the PN such as molecular chaperones.6,26 In this section, some of the most important systems contained in the PN that assist protein folding, intracellular traffic, and degradation are outlined. Molecular chaperones Molecular chaperones are components of the PNs that operate in a highly regulated and coordinated fashion. These assist in protein folding (foldase activity), prevent or revert protein aggregation (holdase and disaggregase activities, respectively), and, eventually, commit misfolded proteins for their degradation.26,28,30 This wide set of structurally diverse proteins is also known as heat-shock or stress proteins (HSP) due to their upregulation under stress conditions 26. Molecular chaperones are usually classified according to their size in six groups: HSP40, HSP60, HSP70, HSP90, HSP100, and the small HSPs (sHSPs).

Protein homeostasis and disease

These groups can act at different stages of protein folding, assembly, and traffic inside the cell.6,27,30–32 Some chaperones with a holdase activity are the sHSPs, the trigger factor, and eukaryotic ribosome-associated complex. These act at very early stages of protein folding, starting at the ribosome, by shielding nascent proteins against aggregation upon binding to exposed hydrophobic patches,6,30 During co- and posttranslational protein folding, newly synthesized proteins can be transferred to HSP70 and HSP90 chaperones that assist folding of partially folded intermediates through cycles of binding and release orchestrated by adenosine triphosphate (ATP) binding and hydrolysis, and by the presence of cochaperones (such as HSP40 and nucleotide-exchange factors).6,30 These chaperones are thus important for cotranslational folding of multidomain and oligomeric proteins, very common in eukaryotic proteomes, particularly due to the slow eukaryotic translation rates.6,18 HSP70 chaperones also commit protein clients to different intracellular fates, acting with disaggregase activities in a coordinated action with Hsp104,29,30 interacting with HSP90 chaperones and the ubiquitin ligase C terminus of HSC70-interacting protein (CHIP)mediating ubiquitination and commitment to the proteasome for degradation,26,30,33 and transferring clients to the large ATP-driven folding chambers of chaperonins (HSP60s) to resume and/or complete folding and assembly.6 HSP90 chaperones can act as proteostasis hubs, usually operating downstream of HSP70 in the folding process, and interacting with a variety of regulators and cochaperones,26 also linking folding pathways with those dedicated to detect and remove potentially harmful protein species, particular the proteasomal and autophagic pathways.26 HSP60 chaperones also act downstream of HSP70 chaperones, assisting protein folding of nascent chains often beyond the capacity of HSP70 chaperones.6 HSP60 chaperones contain a large chamber in which nonnative protein species can fold upon encapsulation in a folding-permissive environment.32 Binding to and release from HSP60 chaperones of client proteins is controlled by slow and concerted conformational changes in the chaperone system regulated by ATP binding, hydrolysis, and adenosine diphosphate (ADP) release.32 Particularly relevant, the eukaryotic HSP60 chaperone TCP-1 ring complex/chaperonin containing TCP-1 (TRiC/CCT) has specialized to assist folding of multidomain and oligomeric proteins, which comprise a large fraction of the eukaryotic proteome.34,35 The role of these chaperone systems in protein folding and traffic may vary among subcellular organelles. They seem to globally act in a similar manner in bacterial and eukaryotic cytosol, with about 20% of newly synthesized proteins interacting with HSP70 chaperones and about 10% of proteins with the HSP60s.6 However, some differences between cytosolic protein folding in prokaryotes and eukaryotes are expected simply from the different size and structural complexities of the proteins constituting their proteomes as well as their different translation rates.27 In the endoplasmic reticulum (ER), folding of eukaryotic proteins seem to follow a similar organizational scheme, although with some specialized functions for secretory proteins including disulfide bond formation and attachment

29

30

Protein Homeostasis Diseases

of carbohydrate moieties.6 Chaperones from the cytosol and ER lumen cooperate in the biogenesis and proper transport of integral membrane proteins.6 Most mitochondrial proteins are encoded in the nucleus and must be synthesized in the cytosol and imported to mitochondria upon interaction of their mitochondrial targeting sequences with the outer membrane translocase of the outer membrane (TOM) receptors, a process often mediated by presentation of the client proteins to the import machinery by Hsp70 and Hsp90 chaperones.29,36 Upon threading through the translocation pore, mitochondrial chaperones at the matrix may subsequently fold mitochondrial proteins.36 Nuclear and peroxisomal proteins typically fold and assemble in the cytosol prior to engaging with specific import machineries, although oligomeric assembly may also occur upon import assisted by chaperones and cochaperones residing inside these organelles.29,30,37–39 To cope with the inherent metastability of the proteome, protein homeostasis capacity can respond to environmental and cellular stress.39–41 These stress-induced mechanisms allow coordinated responses between different cellular organelles and even between tissues and organs.39–42 In the cytosol and nucleus, the main response mechanism (heat-shock response, or HSR) is orchestrated by the heat-shock transcription 1 (HSF1) that forms an inactive complex with HSP70 and HSP90 chaperones.29,40,41,43 Stress-induced increases in partially folded or misfolded proteins competes with these chaperones bound to HSF1, and the released HSF1 forms active trimers, consequently triggering the expression of HSR genes, such as molecular chaperones and components of the proteasomal degradation machinery, and attenuating protein synthesis to cope with proteotoxic stress.29,40,41 In the ER and mitochondria, the unfolded protein response (UPR) operates in a similar manner, although the molecular details of UPR activation can be somewhat different. Increased levels of misfolded species in the ER are detected by different signaling proteins [e.g., Inositol-requiring enzyme 1 (IRE-1), protein kinase R (PKR)-like endoplasmic reticulum kinase (PERK), Activating transcription factor 6 (ATF6)] that activate transcription factors involved in the production of different proteostasis components and also attenuate protein synthesis.40–42 Importantly, a significant crosstalk between these stressresponse mechanisms exists in different subcellular locations (ER, cytosol, mitochondria) and tissues, thus allowing coordinated stress-response at the cellular and organismal levels.40–42,44 Of note, this implies that proteotoxic stress in certain cell types or tissues may lead to reshaping proteostasis capacity at the organismal level.41,42 Protein degradation: proteasome versus autophagy Two main pathways exist in eukaryotic cells that account for protein degradation as a part of protein quality control: the proteasomal and autophagy-lysosomal degradation pathways.45 These two pathways are highly regulated and intertwined through different mechanisms.45 The proteasomal pathway is recognized as the most relevant for degradation of misfolded and damaged proteins in the cytosol and the nucleus.45–47(see also Chapter 6).

Protein homeostasis and disease

This pathway is dependent on the presence of degradation signals (degrons) around modifications of lysine residues in the client protein with ubiquitin moieties (ubiquitindependent proteasomal degradation system, or UPS), a posttranslational modification that shares with autophagy.45,48 A common feature in different types of degrons is the presence of structural disorder and flexibility. For instance, UPS-mediated degradation of short-lived proteins often follows the N-end rule, in which the presence of certain destabilizing amino acids and posttranslational modifications in unstructured N-terminal regions containing a lysine residue allows its efficient ubiquitination and consequent degradation.49,50 More generally, degradation by the UPS seems to involve the presence of three functionally different degrons in the client protein: a primary degron, a linear sequence in which ubiquitin E3 ligases attach; a secondary degron, close to the primary degron where one or several lysine residues are ubiquitinated; and a tertiary degron that acts as initiation site for degradation.48,51 These degrons are preferentially found in dynamic or partially unstructured regions of the client protein, thus allowing control of proteasomal degradation rates through modulation of protein local stability and dynamics.12,52 In addition, the proteasomal machinery may also degrade certain proteins in an ubiquitin-independent manner, mostly through recognition and initiation of degradation of disordered regions in the client substrates.46,53,54 The autophagy-lysosomal pathway (ALP) primarily copes with the removal of large structures (such as very large aggregates) and organelles beyond the capacity of the UPS.45,55 Consequently, both the UPS and the ALP are involved in the removal of unwanted (and potentially toxic) protein materials as well as in the recycling of amino acids during periods of starvation.45,55,56 It seems that the ALP pathway is also unidirectionally activated when the capacity of the UPS has been surpassed.45 Both the ALP and UPS pathways share a common recognition signal in lysine ubiquitination, and indeed, a given ubiquitin E3 ligase can specifically target different protein clients to the ALP or UPS pathways.45 However, the molecular mechanisms triggering ALP activation seem to be complex, and the determinants for the specific recognition of ubiquitinated substrates by the ALP are less known than those determining recognition by the UPS.45,55 The roles of the ALP in the development of misfolding diseases are potentially multifold, including removal of large and potentially toxic aggregates, regulation of the function and dysfunction of mitochondria and peroxisomes, and the transcriptional control of the antioxidant response.45,55,57

Human misfolding diseases Due to the intrinsic metastability of folded proteins, and the competition between folding, misfolding, and aggregation pathways, proteins are continuously at risk of misfolding in vivo. Particularly important, misfolding can be promoted by cellular and environmental stresses, as well as by inherited and somatic missense mutations in human genes, and are thus associated with a long list of human pathologies.2,5,9,58,59 It is interesting to note

31

32

Protein Homeostasis Diseases

that, despite the marginal thermodynamic stability of natural proteins, random missense mutations are generally destabilizing, particularly when these occur in a buried location in the structure.60 According to the phenotypic consequences of disease-associated mutations, protein misfolding diseases are simply classified in two large groups: loss-offunction (LOF) and gain-of-function (GOF) diseases. In LOF diseases, mutations generally compromise the ability of the protein to carry out one or several of their functions5 and often lead to inherited metabolic diseases.31,59,61–68 (see Chapters 6, 10–12, 16, 18). In GOF diseases, natively folded or unfolded proteins undergo aggregation through complex reaction mechanisms to form protein species that show different cellular toxicity, and these are associated with a set of neurological diseases with a large impact in modern society 9,28,30,69(Chapters 3, 7–9). Of note, in GOF diseases, the pathological consequences of protein aggregation are associated with either wild-type and mutated (i.e., inherited) protein species9,26 (Chapters 3, 8).

Loss-of-function diseases LOF diseases are typically caused by inherited missense mutations that substantially affect protein energy landscape, reducing the activity and stability of the protein in vivo.5,59,70 From clinical and biochemical points of view, these diseases are often associated with defects in the synthesis of physiologically important metabolites or lead to accumulation of toxic metabolic intermediates or products, and thus, these diseases are referred to as inborn errors of metabolism. Mechanistically, disease-associated mutations may cause LOF by affecting enzyme catalysis and regulation, protein:protein interactions, decreasing conformational stability and folding efficiency (thus enhancing intracellular degradation or aggregation to form inactive species), and altering intracellular transport (see Chapters 6, 10–12, 16, 18). Interestingly, disease-associated mutations often cause LOF through more than one of these mechanisms, supporting that mutational effects on the protein energy landscape can be transmitted efficiently, and to a variable extent, to those conformations associated with LOF (i.e., inactive, aggregation- or degradation-prone conformations) according to the simple model depicted in Fig. 2.1, a phenomenon known as pleiotropy.5,71–73 A deep understanding on the consequences of missense mutations on the energy landscape, and how this translates into different LOF mechanisms in vivo, is essential to develop novel therapeutic strategies to treat these otherwise orphan diseases (see Chapters 15–18). In this translation, different elements of the PNs such as molecular chaperones and the UPS can play important role in the manifestation of the disease and thus in the development of novel therapeutic strategies (Chapters 6, 15–18).

Gain-of-toxic function diseases Natural proteins and peptides have an intrinsic tendency to convert from a soluble native (folded or natively unfolded) state into a highly stable amyloid state2,9 (Chapters 7–9). This amyloid state is characterized by a threadlike form, highly ordered and tightly

Protein homeostasis and disease

packed structure rich in β-sheet and a remarkably high thermodynamic and kinetic stability2,9 (Chapter 8). The process by which this amyloid state is formed may follow complex sequences of molecular steps (i.e., reaction mechanisms) involving structural rearrangements, association, dissociation, nucleation, growth, and fragmentation steps forming low-molecular-weight oligomers that ultimately assemble into highly ordered fibrils as well as into amorphous and native-like aggregates2; (the reader is referred to Chapter 8 for a detailed description of these mechanisms). Formation of oligomers and fibrils from the native state may involve large kinetic barriers,74 and thus, mutations reducing native state stability can readily promote amyloid formation and hereditary amyloidosis.2,9 In addition, intracellular levels of many proteins seem to be close or even exceed their solubility limits, and thus, aggregation can be easily triggered by mutations or posttranslational modifications due to small changes in protein solubility 2 From the complex chain of reactions involved in the formation of amyloid fibrils, two states are thought to particularly contribute to amyloid pathogenesis: amyloid deposits that affect organ integrity and function, and oligomers that cause cellular toxicity.2,9 In the case of neurological disorders involving amyloid formation, oligomers seem to play the major pathogenic role, and amyloid fibers may provide a continuous supply of such small oligomers.9 These oligomers can target different cellular components causing cellular dysfunction and death, affecting membrane structure and integrity, Ca2+ fluxes triggering oxidative stress, and altering the function of membrane receptors and subcellular organelles such as mitochondria and the ER.2,9,74,75 A common feature of these oligomers related to their toxic effects appears to be the solvent exposure of hydrophobic surface, which is a surface that is often buried in amyloid aggregates, thus contributing to the lower toxicity of the latter.2,75 Although natural selection may have acted on natural proteins, reducing their potential to form toxic aggregates, and protein sequences may have coevolved with the components and regulation of the PN, protein aggregation is unavoidable and eventually leads to dysfunction and collapse of the PN.9 Amyloid aggregates are known to sequester multiple proteins with essential cellular functions, including elements of the PN such as molecular chaperones and components of protein degradation, transcription, and translation machineries.28,76 Oligomers and fibrils strongly interact with components of the PN, particularly with molecular chaperones that act on specific steps of the aggregation mechanism and with the UPS, and these interactions may reduce the toxicity of these protein species.26,29,76–79 Thus, these interactions with the PN represent both disease mechanisms and potential therapeutic strategies for amyloid diseases. In the one hand, these interactions can monopolize some of the cellular proteostasis functions, thus exceeding its capacity to maintain proteome stability and causing cellular toxicity.30 On the other hand, pharmacological modulation of PN components may prevent the accumulation or toxicity of oligomers with therapeutic applications.2,9,26,28,29 In these regards, it must be noted that proteostasis capacity decreases with aging,28–30 and thus, aging is considered to be a major factor contributing to amyloid diseases.

33

34

Protein Homeostasis Diseases

Protein aggregation and decreased solubility thus emerge as a consequence of this agedependent decline in proteostasis, although accumulation of misfolding-prone species can also contribute to the collapse of the PNs.28,29 Aging also reduces the organismal capacity for stress-response through transcriptional dysregulation (e.g., HSR and UPR) as well as the capacity of protein degradation pathways to clear misfolded species.28,40

Acknowledgments I want to thank Prof. Eduardo Salido and Dr. Bertrand Morel for critical reading of the manuscript.

Conflict of interest No conflict of interest declared.

Funding My work is supported by the ERDF/Spanish Ministry of Science, Innovation and Universities—State Research Agency (RTI2018-096246-B-I00).

References 1. Gruebele M, Dave K, Sukenik S. Globular protein folding in vitro and in vivo. Annu Rev Biophys 2016;45:233–51. 2. Dobson CM, Knowles TPJ,Vendruscolo M. The amyloid phenomenon and its significance in biology and medicine. Cold Spring Harb Perspect Biol 2019 Apr 1;. 3. Beaver SK, Mesa-Torres N, Pey AL, Timson DJ. NQO1: a target for the treatment of cancer and neurological diseases, and a model to understand loss of function disease mechanisms. Biochim Biophys Acta Proteins Proteom 2019;1867:663–76. 4. Chen T, Dave K, Gruebele M. Pressure- and heat-induced protein unfolding in bacterial cells: crowding vs. sticking. FEBS Lett 2018 Apr;592(8):1357–65. 5. Medina-Carmona E, Betancor-Fernández I, Santos J, et al. Insight into the specificity and severity of pathogenic mechanisms associated with missense mutations through experimental and structural perturbation analyses. Human Molecular Genetics 2019;28(1):1–15. 6. Hartl FU, Hayer-Hartl M. Converging concepts of protein folding in vitro and in vivo. Nat Struct Mol Biol 2009;16(6):574–81. 7. Soskine M, Tawfik DS. Mutational effects and the evolution of new protein functions. Nat Rev Genet 2010;11(8):572–82. 8. Tokuriki N, Tawfik DS. Stability effects of mutations and protein evolvability. Curr Opin Struct Biol 2009;19(5):596–604. 9. Chiti F, Dobson CM. Protein misfolding, amyloid formation, and human disease: a summary of progress over the last decade. Annu Rev Biochem 2017;86:27–68. 10. Davis CM, Gruebele M, Sukenik S. How does solvation in the cell affect protein folding and binding? Curr Opin Struct Biol 2018;48:23–9. 11. Martinez-Limon A, Alriquet M, Lang WH, Calloni G, Wittig I,Vabulas RM. Recognition of enzymes lacking bound cofactor by protein quality control. Proc Natl Acad Sci USA 2016;113(43):12156–61. 12. Medina-Carmona E, Palomino-Morales RJ, Fuchs JE, et al. Conformational dynamics is key to understanding loss-of-function of NQO1 cancer-associated polymorphisms and its correction by pharmacological ligands. Scientific Reports 2016;6:20331.

Protein homeostasis and disease

13. Kuznetsova IM, Turoverov KK, Uversky VN. What macromolecular crowding can do to a protein. Int J Mol Sci 2014;15(12):23090–140. 14. Yancey PH. Organic osmolytes as compatible, metabolic and counteracting cytoprotectants in high osmolarity and other stresses. J Exp Biol 2005;208(Pt 15):2819–30. 15. Pey AL. Anion-specific interaction with human NQO1 inhibits flavin binding. Int J Biol Macromol 2019;126:1223–33. 16. Dhar A, Girdhar K, Singh D, Gelman H, Ebbinghaus S, Gruebele M. Protein stability and folding kinetics in the nucleus and endoplasmic reticulum of eucaryotic cells. Biophys J 2011;101(2):421–30. 17. Feng R, Gruebele M, Davis CM. Quantifying protein dynamics and stability in a living organism. Nat Commun 2019;10(1):1179. 18. Sharma AK, Sormanni P, Ahmed N, et al. A chemical kinetic basis for measuring translation initiation and elongation rates from ribosome profiling data. PLoS Comput Biol 2019;15(5):e1007070. 19. Samelson AJ, Bolin E, Costello SM, Sharma AK, O’Brien EP, Marqusee S. Kinetic and structural comparison of a protein’s cotranslational folding and refolding pathways. Sci Adv 2018;4(5):eaas9098. 20. Guinn EJ,Tian P, Shin M, Best RB, Marqusee S. A small single-domain protein folds through the same pathway on and off the ribosome. Proc Natl Acad Sci USA 2018;115(48):12206–11. 21. Samelson AJ, Jensen MK, Soto RA, Cate JH, Marqusee S. Quantitative determination of ribosome nascent chain stability. Proc Natl Acad Sci USA 2016;113(47):13402–7. 22. Trovato F, O’Brien EP. Insights into cotranslational nascent protein behavior from computer simulations. Annu Rev Biophys 2016;45:345–69. 23. Sharma AK, O’Brien EP. Non-equilibrium coupling of protein structure and function to translationelongation kinetics. Curr Opin Struct Biol 2018;49:94–103. 24. Alexander LM, Goldman DH,Wee LM, Bustamante C. Non-equilibrium dynamics of a nascent polypeptide during translation suppress its misfolding. Nat Commun 2019;10(1):2709. 25. Nissley DA, Sharma AK, Ahmed N, et al. Accurate prediction of cellular co-translational folding indicates proteins can switch from post- to co-translational folding. Nat Commun 2016;7:10341. 26. Hartl FU, Bracher A, Hayer-Hartl M. Molecular chaperones in protein folding and proteostasis. Nature 2011;475(7356):324–32. 27. Kim YE, Hipp MS, Bracher A, Hayer-Hartl M, Hartl FU. Molecular chaperone functions in protein folding and proteostasis. Annu Rev Biochem 2013;82:323–55. 28. Labbadia J, Morimoto RI. The biology of proteostasis in aging and disease. Annu Rev Biochem 2015;84:435–64. 29. Hipp MS, Kasturi P, Hartl FU.The proteostasis network and its decline in ageing. Nat Rev Mol Cell Biol 2019;20(7):421–35. 30. Hipp MS, Park SH, Hartl FU. Proteostasis impairment in protein-misfolding and -aggregation diseases. Trends Cell Biol 2014;24(9):506–14. 31. Fernandez-Higuero JA, Betancor-Fernandez I, Mesa-Torres N, Muga A, Salido E, Pey AL. Structural and functional insights on the roles of molecular chaperones in the mistargeting and aggregation phenotypes associated with primary hyperoxaluria type I. Adv Protein Chem Struct Biol 2019;114:119–52. 32. Hayer-Hartl M, Bracher A, Hartl FU.The GroEL-GroES chaperonin machine: a nano-cage for protein folding. Trends Biochem Sci 2016;41(1):62–76. 33. Quintana-Gallardo L, Martin-Benito J, Marcilla M, Espadas G, Sabido E,Valpuesta JM. The cochaperone CHIP marks Hsp70- and Hsp90-bound substrates for degradation through a very flexible mechanism. Sci Rep 2019;9(1):5102. 34. Russmann F, Stemp MJ, Monkemeyer L, Etchells SA, Bracher A, Hartl FU. Folding of large multidomain proteins by partial encapsulation in the chaperonin TRiC/CCT. Proc Natl Acad Sci USA 2012;109(52):21208–15. 35. Cuellar J, Ludlam WG, Tensmeyer NC, et al. Structural and functional analysis of the role of the chaperonin CCT in mTOR complex assembly. Nat Commun 2019;10(1):2865. 36. Backes S, Herrmann JM. Protein translocation into the intermembrane space and matrix of mitochondria: mechanisms and driving forces. Front Mol Biosci 2017;4:83. 37. Baker A, Lanyon-Hogg T, Warriner SL. Peroxisome protein import: a complex journey. Biochem Soc Trans 2016;44(3):783–9.

35

36

Protein Homeostasis Diseases

38. Mesa-Torres N, Tomic N, Albert A, Salido E, Pey AL. Molecular recognition of PTS-1 cargo protein by Pex5p: implications for protein mistargeting in primary hyperoxaluria. Biomolecules 2015;5: 121–41. 39. Shibata Y, Morimoto RI. How the nucleus copes with proteotoxic stress. Curr Biol 2014;24(10):R463– 74. 40. Klaips CL, Jayaraj GG, Hartl FU. Pathways of cellular proteostasis in aging and disease. J Cell Biol 2018;217(1):51–63. 41. Sala AJ, Bott LC, Morimoto RI. Shaping proteostasis at the cellular, tissue, and organismal level. J Cell Biol 2017;216(5):1231–41. 42. van Oosten-Hawle P, Morimoto RI. Organismal proteostasis: role of cell-nonautonomous regulation and transcellular chaperone signaling. Genes Dev 2014;28(14):1533–43. 43. Li J, Labbadia J, Morimoto RI. Rethinking HSF1 in stress, development, and organismal health. Trends Cell Biol 2017;27(12):895–905. 44. Labbadia J, Brielmann RM, Neto MF, Lin YF, Haynes CM, Morimoto RI. Mitochondrial stress restores the heat shock response and prevents proteostasis collapse during aging. Cell Rep 2017;21(6):1481–94. 45. Dikic I. Proteasomal and autophagy degradation systems. Annu Rev Biochem. 2017;86:193–224. 46. Inobe T, Matouschek A. Paradigms of protein degradation by the proteasome. Curr Opin Struct Biol 2014;24:156–64. 47. Collins GA, Goldberg AL. The logic of the 26S proteasome. Cell 2017;169(5):792–806. 48. Guharoy M, Bhowmick P, Sallam M, Tompa P. Tripartite degrons confer diversity and specificity on regulated protein degradation in the ubiquitin-proteasome system. Nat Commun 2016;7:10239. 49. Varshavsky A. The ubiquitin system, autophagy, and regulated protein degradation. Annu Rev Biochem 2017;86:123–8. 50. Tasaki T, Sriram SM, Park KS, Kwon YT.The N-end rule pathway. Annu Rev Biochem 2012;81:261–89. 51. Braten O, Livneh I, Ziv T, et al. Numerous proteins with unique characteristics are degraded by the 26S proteasome following monoubiquitination. Proc Natl Acad Sci USA 2016;113(32):E4639–47. 52. Takahashi K, Matouschek A, Inobe T. Regulation of proteasomal degradation by modulating proteasomal initiation regions. ACS Chem Biol 2015;10(11):2537–43. 53. Baugh JM, Viktorova EG, Pilipenko EV. Proteasomes can degrade a significant proportion of cellular proteins independent of ubiquitination. J Mol Biol 2009;386(3):814–27. 54. Ben-Nissan G, Sharon M. Regulating the 20S proteasome ubiquitin-independent degradation pathway. Biomolecules 2014;4(3):862–84. 55. Hurley JH,Young LN. Mechanisms of autophagy initiation. Annu Rev Biochem 2017;86:225–44. 56. Vabulas RM, Hartl FU. Protein synthesis upon acute nutrient restriction relies on proteasome function. Science. 2005;310(5756):1960–3. 57. Sanchez-Martin P, Komatsu M. Physiological stress response by selective autophagy. J Mol Biol 2019;doi: 10.1016/j.jmb.2019.06.013. 58. Hartl FU. Protein misfolding diseases. Annu Rev Biochem 2017;86:21–6. 59. Muntau AC, Leandro J, Staudigl M, Mayer F, Gersting SW. Innovative strategies to treat protein misfolding in inborn errors of metabolism: pharmacological chaperones and proteostasis regulators. J Inherit Metab Dis 2014;37(4):505–23. 60. Tokuriki N, Stricher F, Schymkowitz J, Serrano L, Tawfik DS. The stability effects of protein mutations appear to be universally distributed. J Mol Biol 2007;369(5):1318–32. 61. Medina-Carmona E, Fuchs JE, Gavira JA, et al. Enhanced vulnerability of human proteins towards disease-associated inactivation through divergent evolution. Human Molecular Genetics 2017;26:3531–44. 62. Gomes CM. Protein misfolding in disease and small molecule therapies. Curr Top Med Chem 2012;12(22):2460–9. 63. McCorvie TJ, Kopec J, Pey AL, et al. Molecular basis of classic galactosemia from the structure of human galactose 1-phosphate uridylyltransferase. Hum Mol Genet 2016;25:2234–44. 64. Pey AL, Mesa-Torres N, Chiarelli LR, Valentini G. Structural and energetic basis of protein kinetic destabilization in human phosphoglycerate kinase 1 deficiency. Biochemistry 2013;52(7):1160–70. 65. Martinez A, Calvo AC, Teigen K, Pey AL. Rescuing proteins of low kinetic stability by chaperones and natural ligands phenylketonuria, a case study. Prog Mol Biol Transl Sci 2008;83:89–134.

Protein homeostasis and disease

66. Marinko JT, Huang H, Penn WD, Capra JA, Schlebach JP, Sanders CR. Folding and misfolding of human membrane proteins in health and disease: from single molecules to cellular proteostasis. Chem Rev 2019;119(9):5537–606. 67. Blouin JM, Bernardo-Seisdedos G, Sasso E, et al. Missense UROS mutations causing congenital erythropoietic porphyria reduce UROS homeostasis that can be rescued by proteasome inhibition. Hum Mol Genet 2017;26:1565–76. 68. Scheller R, Stein A, Nielsen SV, et al. Towards mechanistic models for genotype-phenotype correlations in phenylketonuria using protein stability calculations. Hum Mutat 2019;40(4):444–57. 69. Cremades N, Dobson CM. The contribution of biophysical and structural studies of protein self-assembly to the design of therapeutic strategies for amyloid diseases. Neurobiol Dis 2018;109(Pt. B):178– 90. 70. Stein A, Fowler DM, Hartmann-Petersen R, Lindorff-Larsen K. Biophysical and mechanistic models for disease-causing protein variants. Trends Biochem Sci 2019;44(7):575–88. 71. Mesa-Torres N, Betancor-Fernández I, Oppici E, Cellini B, Salido E, Pey AL. Evolutionary divergent suppressor mutations in conformational diseases. Genes 2018;9:E352. 72. Pey AL. Biophysical and functional perturbation analyses at cancer-associated P187 and K240 sites of the multifunctional NADP(H):quinone oxidoreductase 1. Int J Biol Macromol 2018;118:1912–23. 73. Fuchs JE, Muñoz IG, Timson DJ, Pey AL. Experimental and computational evidence on conformational fluctuations as a source of catalytic defects in genetic diseases. RSC Adv. 2016;6:58604. 74. Cremades N, Cohen SI, Deas E, et al. Direct observation of the interconversion of normal and toxic forms of alpha-synuclein. Cell 2012;149(5):1048–59. 75. Fusco G, Chen SW,Williamson PTF, et al. Structural basis of membrane disruption and cellular toxicity by alpha-synuclein oligomers. Science 2017;358(6369):1440–3. 76. Olzscha H, Schermann SM,Woerner AC, et al. Amyloid-like aggregates sequester numerous metastable proteins with essential cellular functions. Cell 2011;144(1):67–78. 77. Cox D, Whiten DR, Brown JWP, et al. The small heat shock protein Hsp27 binds alpha-synuclein fibrils, preventing elongation and cytotoxicity. J Biol Chem 2018;293(12):4486–97. 78. Kundel F, De S, Flagmeier P, et al. Hsp70 inhibits the nucleation and elongation of tau and sequesters tau aggregates with high affinity. ACS Chem Biol 2018;13(3):636–46. 79. Ciechanover A, Kwon YT. Protein quality control by molecular chaperones in neurodegeneration. Front Neurosci 2017;11:185.

37

CHAPTER 3

Caenorhabditis elegans as a model organism for protein homeostasis diseases Sarah Good, Patricija van Oosten-Hawle

School of Molecular and Cell Biology and Astbury Centre for Structural Molecular Biology, University of Leeds, Leeds, United Kingdom

Outline Caenorhabditis elegans as a model organism The proteostasis network is conserved in Caenorhabditis elegans Alzheimer’s disease Tauopathies Parkinson’s disease Polyglutamine diseases Amyotrophic lateral sclerosis Transthyretin amyloidosis Type II diabetes mellitus Dialysis-related amyloidosis Immunoglobulin light chain amyloidosis Prion diseases Conclusion References

43 43 46 53 54 56 58 59 60 61 61 61 62 62

Abbreviations 5-HT  5-hydroxytryptamine AD  Alzheimer’s disease ALS  amyotrophic lateral sclerosis APP  amyloid precursor protein ASH  Amphid neurons, single ciliated endings ATF  Activating transcription factor atfs  Activating transcription factor associated with stress α-syn  α-synuclein Aβ  amyloid-β β2m  β2-microglobulin CCT/TRiC  chaperonin containing TCP-1/T-complex protein-1 ring complex CDC  Cell division cycle related C. elegans  Caenorhabditis elegans ClpP  caseinolytic mitochondrial matrix peptidase proteolytic Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00003-8 Copyright © 2020 Elsevier Inc. All rights reserved.

41

42

Protein Homeostasis Diseases

CRYAβ  αβ-crystallin DAF  abnormal dauer formation DNJ  DNaJ domain DJ  Protein deglycase DRA  dialysis-related amyloidosis ER  endoplasmic reticulum eIF2  Eukaryotic Initiation Factor 2 FAC  familial amyloid cardiomyopathy FAP  familial amyloid polyneuropathy FTDP  frontotemporal dementia with Parkinsonism FUS  fused-in sarcoma GABA  γ-aminobutyric acid GFP  green fluorescent protein HAF  HAlF transporter HD  Huntington’s disease HSF  heat shock factor HSP  heat shock protein HSR  heat shock response IAPP  islet amyloid polypeptide IGF  insulin growth factor IIS  insulin/insulin growth factor signaling IRE  inositol-requiring enzyme LCs  immunoglobulin light chains MAP  microtubule-associated protein MAPT  microtubule associated protein tau MCSR  mitochondrial-to-cytosolic stress response MHCI  major histocompatibility complex I MJD  Machado-Joseph disease PD  Parkinson’s disease PEK  human PERK kinase homolog PERK  protein kinase R-like endoplasmic reticulum kinase PINK  PTEN-induced kinase PFD  prefoldin PHFs  paired helical filaments PHP  pseudohyperphosphorylated PN  proteostasis network PolyQ  polyglutamine RNAi  RNA interference SBMA  spinal and bulbar muscular atrophy SCA  spinocerebellar SERF  small EDRK-rich factor sHSPs  small heat shock proteins smg  suppressor with morphological effect on genitalia SOD1  superoxidase dismutase 1 SSA  senile systemic amyloidosis STI  stress induced phosphoprotein SUP Suppressor TDP-43  TAR DNA binding protein 43 TPR  tetratricopeptide repeat

Caenorhabditis elegans as a model organism for protein homeostasis diseases

TSEs  transmissible spongiform encephalopathies TTR  transthyretin UPR  unfolded protein response UPS  ubiquitin proteasome system XBP  x-box binding protein YFP  yellow fluorescent protein

Caenorhabditis elegans as a model organism Caenorhabditis elegans is a free-living, nonparasitic nematode with a lifespan of 2–3 weeks and a rapid reproductive cycle.1 Since its selection as a simple metazoan model for developmental biology in the early 1960s by Sydney Brenner and due to its short lifespan and ease to cultivate, C. elegans has been adapted and used to investigate a range of biological processes and their associated diseases.1 The adult is ∼1 mm in length and feeds on bacteria such as Escherichia coli and therefore can be easily grown on agar plates or in liquid culture. The entire cell lineage of 959 somatic cells of the adult hermaphrodite has been traced.2 In 1998, the genome sequence of C. elegans was completed, and it contains about 100 million base pairs. It is estimated that roughly 38% of C. elegans genes have a human ortholog.3 In addition, genetic manipulations, including forward genetic screens, reverse genetic RNAi screens, rapid mutation mapping, and transgenic animal generation, are easily performed, providing a powerful genetic toolbox that C. elegans researchers benefit from.

The proteostasis network is conserved in Caenorhabditis elegans Like humans, C. elegans are constantly under threat from misfolded proteins generated from stresses such as heat shock, oxidative stress, nutrient starvation, and metabolic imbalances. To counteract these proteotoxic challenges and maintain protein homeostasis, C. elegans possess highly conserved components comprising the proteostasis network (PN), encompassing molecular chaperones, transcription factor regulators, and degradation machinery in order to fold, unfold, and degrade misfolded proteins.4 C. elegans have 219 chaperone genes which can be grouped into functional chaperone families including the heat shock protein 90 family (HSP-90 in C. elegans), and the Hsp70, Hsp60, Hsp40, prefoldin, small heat shock protein (sHSPs), and tetratricopeptide repeat-domain-containing chaperone families as well as endoplasmic reticulum–specific and mitochondria-specific chaperones.5 To upregulate molecular chaperones and other factors counteracting protein misfolding, C. elegans also possess conserved stress-response mechanisms including the heat shock response (HSR) and the unfolded protein response (UPR) of the endoplasmic reticulum (UPRER) and mitochondria (UPRmito).6

43

44

Protein Homeostasis Diseases

The heat shock response The HSR is an ancient thermal stress response that is conserved between humans and C. elegans.7 In eukaryotes, the master regulator of the HSR is the highly conserved heat shock transcription factor1, Hsf1.7,8 Under normal cellular conditions, Hsf1 is sequestered in the cytosol in a complex with, Hsp90, Hsp40 and Hsp70.9–11 During stress conditions that lead to protein misfolding and aggregation, Hsp90, Hsp70, and Hsp40, dissociate, and Hsf1 trimerizes and translocates to the nucleus where it binds to a consensus sequence in the promoter of HSPs known as the heat shock element.12 This leads to the transcription of chaperones including Hsp90, Hsp70, and small heat shock proteins, which help to refold any misfolded proteins and maintain cellular proteostasis.9 Once proteostasis has been restored, the excess of Hsp70 acts as negative feedback on Hsf1, returning it to its cytosolic complex. The unfolded protein response of the endoplasmic reticulum and the mitochondria Another stress response pathway is the UPRER, which monitors and maintains protein folding and degradation in the ER.13 Similarly to humans, the UPRER in C. elegans is regulated by three proteins that sense ER stress, inositol-requiring enzyme 1 (IRE-1), activating transcription factor 6 (ATF-6), and a homolog of the human protein kinase R-like endoplasmic reticulum kinase (PERK), PEK-1.13–15 Upon ER stress, IRE-1 splices xbp-1 to generate a transcription factor that induces expression of UPR genes, including chaperones and ER-associated degradation proteins.14 This increases the folding capacity of the ER and allows for the degradation of any misfolded proteins. PEK-1 is a serine/threonine kinase that can phosphorylate the eukaryotic initiation factor 2 (eIF2α), thereby inhibiting protein translation initiation; this reduces the amount of proteins translated in the ER, thus easing the folding load of chaperones.16 ATF-6 is a leucine-zipper transcription factor usually retained at the ER membrane; however, in response to ER stress, ATF-6 transits to the Golgi, where it is processed to form a transcription factor that induces expression of genes that increase the folding capacity of the ER.15 A conceptually similar pathway to the compartmental stress response in the ER is the mitochondrial UPR (UPRmito).17 In C. elegans, mitochondrial stress induces the transcription of nuclear-encoded mitochondrial chaperone genes, including an orthologue of mitochondrial Hsp70, HSP-60, which act to deal with an increase in unfolded or misfolded proteins present in the organelle.18 Detection of this response is dependent on the ATP-dependent protease ClpP, which acts as a sensor by recognizing and degrading unfolded proteins that have accumulated beyond the capacity of mitochondrial chaperones.19 The HAIF transporter, HAF-1, then acts as a transporter for the peptides generated by ClpP degradation, facilitating their translocation into the cytosol to promote downstream signaling for transcriptional activation of mitochondrial chaperones.20

Caenorhabditis elegans as a model organism for protein homeostasis diseases

The insulin-like signaling pathway Another notable conserved signaling pathway is the insulin/insulin-like growth factor (IGF) signaling pathway (IIS), which regulates lifespan by controlling activity of the Forkhead Box O transcription factor DAF-16 (abnormal DAuer Formation) through the insulin/IGF-1 receptor DAF-2. DAF-16 mediates expression of target genes involved in the regulation of longevity, including molecular chaperones21 and sHSPs, such as HSP16 and HSP-12 families.22–25 Caenorhabditis elegans as a model for protein misfolding diseases Due to its short lifespan, C. elegans is a particularly useful model for a number of human protein misfolding diseases that are exacerbated during aging. In 1995, the first C. elegans protein misfolding disease model was generated, expressing the human amyloid-β peptide, a component of amyloid plaques found in muscle cells in the brain of patients with Alzheimer’s disease (AD).26 Since then, numerous human protein misfolding disease models have been created using C. elegans as a model system. These include AD, Parkinson’s disease (PD), Huntington’s disease (HD), spinocerebellar ataxia (SCA), type II diabetes mellitus (T2DM), amyloid lateral sclerosis, transthyretin (TTR) amyloidosis, dialysis-related amyloidosis (DRA), and immunoglobulin light chain (AL) amyloidosis, all of which will be the topic of this chapter and are described in more detail in the further sections. Many of these disease proteins have amyloid characteristics and self-assemble into aggregates in an age-dependent manner.27 Due to the transparency of C. elegans as a model system, this can be tracked, visualized, and quantified by tagging the protein with fluorescent proteins. This allows detailed information on the aggregation kinetics of human disease proteins throughout aging in real time. Moreover, expression of disease proteins in various tissues causes cellular toxicity, leading to measurable behavioral defects such as reduced motility or chemosensory defects. These measurable behavioral phenotypes can be utilized as readouts for genetic screens to identify modifiers of aggregation and aid in the understanding of molecular pathways underlying toxicity caused by amyloid protein aggregation. In addition, by integrating aggregation kinetics and behavioral defects, the influence of the aggregated proteins can be coupled to its cellular toxicity. Although C. elegans models of protein misfolding diseases have yielded substantial insight into the mechanisms of aggregation of these diseases in vivo, it is important to appreciate its limitations. Leaky expression of transgenes, erroneous expression patterns, or mosaic expression of extrachromosomal arrays can also mislead data.28 Furthermore, although genetic screens have been used to identify potential modifiers of aggregation, follow-up research usually lags behind the wealth of data obtained by a genetic screen in order to confirm scored gene functions in the pathogenesis of the disease in vertebrate model systems or human samples (Fig. 3.1).

45

46

Protein Homeostasis Diseases

Figure 3.1  Caenorhabditis elegans as a transgenic model of protein misfolding disease. Summary of transgenic C. elegans protein disease models expressing aggregation-prone human disease proteins in different tissues. Tissues are indicated as colored boxes, showing expression of human disease proteins modeled so far in C. elegans (Table 3.1).

Alzheimer’s disease Disease mechanism AD is one of the most common neurodegenerative disease, leading to progressive memory loss, cognitive dysfunction, and neurodegeneration.29 AD is caused by the formation of two types of inclusions in the brain: amyloid plaques, composed of aggregates of amyloid-β (Aβ) and neurofibrillary tangles, which are composed of tau and paired helical filaments (PHFs).29 These tangles occur in a number of diseases called tauopathies and will be explored further in the “Tauopathies” section. Aberrant cleavage of the amyloid precursor protein (APP) by β- and γ- secretases leads to the formation of Aβ. γ-Secretase cleavage can vary in the APP protein and can yield both Aβ40 and Aβ42, as well as release of the intracellular domain of APP.30 Antibody staining has shown that diffuse (non-neuritic) deposits are largely composed of Aβ42, while Aβ40 peptide is more abundantly generated.31 Mutations in the APP gene, presenilin (PS)-1 and PS-2, alter APP processing and the production of Aβ peptides, and have been linked to early onset in autosomal dominant AD.32,33 Mutations in apolipoprotein (apo) E4 (APOE4) have been linked to late-onset (>60 years) familial and sporadic AD and reported to impact Aβ aggregation, Aβ clearance, inflammation, and cerebrovascular events.34 Caenorhabditis elegans Amyloid-β models Although C. elegans has an APP-related gene (apl-1), it lacks the Aβ peptide and the β-secretase gene.35 The first C. elegans AD model was generated to express Aβ1-42 in the

Caenorhabditis elegans as a model organism for protein homeostasis diseases

body wall muscle of the animal, leading to age-dependent formation of intracellular amyloid deposits, which are immunoreactive to anti-Aβ antibodies and amyloid-specific dyes.26,36 Expression of Aβ deposits in this model also correlated with an age-progressive paralysis phenotype,36–38 which occurs more severely when animals are grown at 25°C.38 However, this model26,36 was later shown to produce Aβ3-42, not the intended Aβ1-42, due to aberrant cleavage of the Aβ transgene.38 Therefore, an additional model was created that truly produced Aβ1-42 (Table 3.1) and exhibits a severe paralysis phenotype when shifted to the restrictive temperature at 25°C.38,39 In order to disentangle age-related effects on the paralysis phenotype, a temperatureinducible Aβ strain was generated using the well-characterized C. elegans RNA-surveillance system smg(suppressor of morphological effect on genitalia).40 This system uses an abnormally long 3’ UTR sequence resulting in unstable mRNA that is degraded by the smg RNA surveillance pathway.41 Temperature upshift to 23°C inactivates the smg-1 system and allows for the translation of the stabilized Aβ transgene, leading to paralysis of almost all animals within 27 hours, demonstrating the strong, toxic nature of this protein.40 In another C. elegans AD model, a signal peptide sequence was added to the Aβ gene to direct the expressed peptide to the secretory pathway.42 An L17P substitution in this transgene sequence blocked the aggregation of Aβ, while a conservative L17V substitution did not,42 consistent with previous studies in vitro showing that nonconservative substitutions in the hydrophobic 17–21 core sequence prevented aggregation (Table 3.1).43,44 In order to more closely mimic the toxic effects of Aβ in the brain, an AD model was generated to express Aβ in all C. elegans neurons using a pan-neuronal promoter (Table 3.1).45 This strain exhibited reduced chemotaxis to the attractant benzaldehyde, increased serotonin or 5-hydroxytryptamine (5-HT) sensitivity, decreased lifespan, reduced serotonin-stimulated egg laying, and impaired associative learning and experience-dependent learning.45,46 Since glutamatergic neurons are particularly sensitive to Aβ expression in AD, Aβ has been expressed in the nematode under control of the glutamatergic eat-4 promoter, showing age-dependent glutamatergic neuron loss and neurodegeneration (Table 3.1).47 These effects were suppressed when certain genes were coexpressed with Aβ, including genes involved in clathrin-mediated endocytosis.47 Utility of Caenorhabditis elegans Alzheimer’s disease models for drug discovery and identification of genetic modifiers As expression of Aβ in C. elegans has been shown to be toxic to cells and cause a number of identifiable proteotoxic phenotypes, C. elegans has been used extensively to study genes or compounds that can ameliorate these. The UPRmito pathway has been shown to have a strong correlation in expression data sets from patients with AD brain tissue.48 This was further explored in an AD C. elegans model in which the induction of

47

Table 3.1  Caenorhabditis elegans transgenic models of protein misfolding diseases. Disease Protein

AD



APP Tau

PD

α-syn

Tissue

Genotype

Bwm Bwm Bwm Bwm Glut neu Pan-neu Pan-neu

unc-54p::human Aβ1-42 unc-54p::human Aβ3-42 unc-54p::human L17P Aβ3-42 myo-3p::human Aβ3-42:: let-851 eat-4p::human Aβ snb-1p::human Aβ1-42 rab-3p::APL-1::GFP apl-1 mutant aex-3p::human IN4R-MAPT aex-3p:: human P201L MAPT

Phenotype

Accumulation of amyloid deposits, rapid paralysis Accumulation of amyloid deposits, progressive paralysis No amyloid deposition Paralysis induced after upshift to 23˚C Loss of glutamatergic neurons Reduced chemotaxis, hypersensitive to 5-HT Shortened lifespan Larval lethality Pan-neu Uncoordinated Pan-neu Severe uncoordinated phenotype and progressive neurodegeneration Pan-neu aex-3p:: human V337M MAPT Server uncoordinated phenotype and progressive neurodegeneration Sensory neu mec-7p::human tau Age-dependent loss of touch nose response Sensory neu mec-7p::P301L tau Severe and progressive loss of touch nose response Sensory neu mec-7p::R406W tau Severe and progressive loss of touch nose response Pan-neu F25B3.3p::human tau Age-dependent uncoordinated locomotion Pan-neu F25B3.3p::human PHP tau Age-dependent uncoordinated locomotion, defective motor neuron development Pan-neu snb-1p::human 2N4R tau Mild late-onset uncoordinated phenotype Pan-neu snb-1p::human 2N4R A152T tau Early-onset paralysis, reduction of lifespan, neuronal dysfunction Bwm Progressive accumulation of deposits, reduction in motility unc-54p::human α-syn::YFP Bwm Progressive accumulation of deposits unc-54p::human α-syn::GFP DA neu DA loss and dendritic breaks dat-1p::human α-syn::GFP DA neu DA loss and dendritic breaks dat-1p::A53T a-syn Pan-neu Reduced thrashing rate, DA loss and dendritic breaks aex-3p::human α-syn Pan-neu Reduced thrashing rate, DA loss, and dendritic breaks aex-3p::A53T α-syn Motor neu acr-2p::human α-syn Reduced thrashing rate Motor neu acr-2p::A53T α-syn Reduced thrashing rate DA neu Neurite defects dat-1p::A30P α-syn DA neu Neurite defects dat-1p::A53T α-syn DA neu Severe neurite defects dat-1p::A56P α-syn Pan-neu No obvious phenotype unc-51p::human α-syn Pan-neu unc-51p:: A53T α Pan-neu -synunc-51p:: A30P α-syn

References

[39] [26,38] [42] [40] [47] [45] [148,149] [68] [68] [68] [70] [70] [70] [66] [66] [69] [69] [85] [89] [81] [81] [81] [81] [81] [81] [82] [82] [82] [91] [91] [91]

Disease Protein

HD

Tissue

HttBwm PolyQ Sensory neu Sensory neu Sensory neu Sensory neu Sensory neu Bwm Bwm Bwm Bwm Bwm Bwm Bwm Bwm BwmPanneu Pan-neu Pan-neu Pan-neu

SCA

Pan-neu Pan-neu MJD1- Pan-neu PolyQ Pan-neu Pan-neu

Genotype

Phenotype

References

unc-54p::GFP-Htt-polyQ mec-3p::Htt-polyQ::GFO osm-10p::Htt-Q2 polyQ osm-10p::Htt-Q23 polyQ osm-10p::Htt-Q95 polyQ osm-10p::Htt-Q150 polyQ

Reduced thrashing rate Neurodegeneration No sensory neuron phenotype No sensory neuron phenotype Progressive defective sensory neuron endings Progressive sensory neuron defect, death, progressive aggregate accumulation Diffuse fluorescence distribution and no motility defects Diffuse fluorescence distribution and no motility defects Diffuse fluorescence distribution and no motility defects Age-dependent foci formation and motility defect Age-dependent foci formation and motility defect Age-dependent foci formation and severe motility defect Localized discrete fluorescent foci Localized discrete fluorescent foci Localized discrete fluorescent foci and severe motility defect Diffuse fluorescent distribution, no motility defect Diffuse fluorescent distribution, no motility defect Diffuse fluorescent distribution, motility phenotype Diffuse fluorescent distribution in the neurons, motility phenotype Localized discrete fluorescent foci, severe motility phenotype Localized discrete fluorescent foci, severe motility phenotype Diffuse fluorescence distribution Diffuse fluorescence distribution Age-dependent aggregation

[123] [150] [151] [151] [151] [151]

unc-54p::Q0-YFP unc-54p::Q19-YFP unc-54p::Q29-YFP unc-54p::Q33-YFP unc-54p::Q35-YFP unc-54p::Q40-YFP unc-54p::Q44-YFP unc-54p::Q64-YFP unc-54p::Q82-YFP F25B3.3p::Q0-CFP F25B3.3p::Q19-CFP F25B3.3p::Q35-CFP F25B3.3p::Q40-CFP F25B3.3p::Q67-CFP F25B3.3p::Q86-CFP unc-119p::MJD1-17Q-GFP unc-119p::MJD1-91Q-GFP unc-119p::MJD1-130Q-GFP

[100] [100] [100] [100] [100] [100] [100] [100] [100] [98] [98] [98] [98] [98] [98] [108] [108] [108]

(Continued )

Table 3.1  Caenorhabditis elegans transgenic models of protein misfolding diseases. (Cont.) Disease Protein

ALS

SOD 1

Tissue

Genotype

Ubiquitous Ubiquitous Ubiquitous Ubiquitous Bwm Bwm Bwm Pan-neu Pan-neu

hsp16.2p::human SOD1 hsp16.2p::A4V SOD1 hsp16.2p:: G37V SOD1 hsp16.2p::G93A SOD1 unc-54p::human SOD1::YFP unc-54p::G85R SOD1-YFP unc-54p::G93A SOD1-YFP snb-1p::human SOD1 snb-1p::G85R SOD1

Motor neu Motor neu Ubiquitous Ubiquitous Ubiquitous TDP-43 Pan-neu Pan-neu Pan-neu Pan-neu Pan-neu Pan-neu Pan-neu Pan-neu

Phenotype

No sensitivity to paraquat under heat shock conditions Increased sensitivity to paraquat under heat shock conditions Increased sensitivity to paraquat under heat shock conditions Increased sensitivity to paraquat under heat shock conditions Diffuse fluorescence, no motility defects Punctate fluorescent pattern, reduction in thrashing rate Punctate fluorescent pattern, reduction in thrashing rate No motility defect Severe motility defect, decreased number of neuromuscular junctions and synaptic vesicles unc-25p::human SOD1 Progressive locomotion defect and paralysis phenotype unc-25p::G93A SOD1 Progressive locomotion defect and paralysis phenotype MosSCI A4V SOD1 Increased SOD1 protein accumulation, oxidative stress–induced neurodegeneration MosSCI H71Y SOD1 Increased SOD1 protein accumulation, oxidative stress–induced neurodegeneration MosSCI G85R SOD1 Increased SOD1 protein accumulation, oxidative stress–induced neurodegeneration snb-1p::human TDP-43 Uncoordinated phenotype, abnormal motor neuron synapses snb-p1::human TDP-43 Moderate motor defect, shortened lifespan snb-1p::G290A TDP-43 Severe motor dysfunction, degeneration of GABAergic neurons, shortened lifespan snb-1p::A315T TDP-43 Severe motor dysfunction, degeneration of GABAergic neurons, shortened lifespan snb-1p::M337V TDP-43 Severe motor dysfunction, degeneration of GABAergic neurons, shortened lifespan snb-1p::human TDP-43::YFP Severe locomotion defects, formation of fluorescent insoluble foci snb-1p::Q331K TDP-43::YFP Severe locomotion defects, formation of fluorescent insoluble foci snb-1p::M337V TDP-43::YFP Severe locomotion defects, formation of fluorescent insoluble foci

References

[122] [122] [122] [122] [84] [84] [84] [123] [123] [124] [124] [121] [121] [121] [127] [128] [128] [128] [128] [30] [30] [30]

Disease Protein

Tissue

Genotype

Phenotype

References

SSA TTR and FAP

Bwm Bwm Bwm Bwm Bwm

unc-54p::TTR::EGFP unc-54p::TTR1-80::EGFP unc-54p::TTR49-127::EGFP unc-54p::TTR81-127::EGFP unc-54p::V30M TTR

[133] [133] [133] [133] [131]

Bwm Bwm Ubiquitous Pan-neu, Bwm, + pharynx

unc-54p:: D18G TTR unc-54p::β2m hsp-16.2p::hIAPP lev-11p::hIAPP::YFP; tnt-4:: hIAPP::YFP; aex-3p::hIAPP::YFP

Diffuse fluorescence, decreased lifespan Diffuse fluorescence Localized fluorescent foci, decreased lifespan Localized fluorescent foci, decreased lifespan Impaired nociception, defective dendritic morphology, impaired locomotion Impaired nociception, Impaired locomotion Increased paralysis, delayed larval growth Defects in chemotaxis Insoluble aggregate formation; growth retardation

DRA β2m Type II IAPP diabetes

[131] [140] [137] [136]

Abbreviations: α-syn, α-synuclein; AD, Alzheimer’s disease; ALS, Amyotrophic lateral sclerosis; Bwm, Body wall muscle; DA neu, dopaminergic neurons; DRA, dialysis related amyloidosis; FAP, familial amyloid polyneuropathy; Glut neu, glutamatergic neurons; HD, Huntington’s disease; Motor neu, motor neurons; Pan-neu, pan-neuronal; PD, Parkinson’s disease; SCA, spinocerebellar ataxia; Sensory neu, sensory neurons; SSA, senile systemic amyloidosis.

52

Protein Homeostasis Diseases

this mitochondrial stress response was confirmed and shown to be essential for the maintenance of mitochondrial proteostasis and health.48 Further, increasing the expression of the UPRmito transcription factor atfs-1 (activating transcription factor associated with stress) in both a C. elegans AD disease model and a transgenic mouse model of AD reduced the number of amyloid aggregates in the models.48 To gain insight into the molecules interacting with Aβ, mass spectrometry analysis of proteins that coimmunoprecipitate with Aβ was performed in the Aβ3-42 body wall muscle model and identified several chaperone proteins that interact with intracellular Aβ, including two members of the HSP-70 family and three αB-crystallin-related sHSPs.49 sHSPs have been further explored in the Aβ3-42 model: HSP-16 was shown to colocalize with Aβ intracellular deposits and, when overexpressed, partially suppresses Aβ toxicity, suggesting a role for this chaperone in modulating intracellular Aβ.49,50 Heat shock treatment of a neuronal Aβ-expressing C. elegans strain has been shown to alleviate Aβ toxicity and diminish Aβ oligomerization, providing a link between upregulation of chaperones and protection against proteotoxicity.51 Furthermore, DNJ-27 (DNaJ domain), an ER luminal protein, has a protective role against Aβ aggregation and, when overexpressed in C. elegans, is found to ameliorate the deleterious paralysis phenotype associated with Aβ expression.52 Gene expression analysis has been performed on temperature-inducible Aβexpressing strains (Table 3.1), identifying 67 upregulated and 240 downregulated genes.40 αB-crystallin (CRYAB) and tumor necrosis factor-induced protein 1 (TNFAIP1) were among the induced genes, and these were also confirmed to be upregulated in AD patients’ brains.40 Furthermore, an RNA interference (RNAi) screen comprising all C. elegans chaperone genes identified a core set of chaperones that act as suppressors of Aβ-associated toxicity, including hsp-1 (Hsc70 in humans), hsp-90, eight subunits of the CCT/TRiC complex (chaperonin containing TCP-1/T-complex protein-1 ring complex), hsp-40, cell division cycle related protein 27 (cdc-27), and stress induced phosphoprotein 1 (sti-1).5 By comparison, knockdown of a similar set of chaperone genes in human cells using RNAi led to a similar increase of protein aggregates,5 demonstrating the conservation of the system. The IIS pathway has been shown to both increase the lifespan and suppress protein misfolding in C. elegans. Depletion of the IGF-1 ortholog in C. elegans, DAF-2, resulted in reduced toxicity associated with Aβ expression and led to daf-2 dependent activation of HSF-1 and DAF-16.37 Independently of the IIS pathway, dietary restriction has also been shown to act through HSF-1 to slow age-associated Aβ toxicity in C. elegans.53 C. elegans models of AD have been used extensively in order to test the efficacy of small-molecule modifiers of Aβ toxicity. The use of Ginkgo biloba tree extract EGb 761—a small molecule previously found to inhibit oligomerization of Aβ in vitro—alleviated paralysis and reduced chemotaxis deficiencies in this model by inhibiting Aβ oligomerization and deposits in neurons.45 More recent studies on this molecule in

Caenorhabditis elegans as a model organism for protein homeostasis diseases

mice54 and human trials have demonstrated a significant cognitive improvement,55 demonstrating the potential of this disease model for drug discovery.

Tauopathies Disease mechanism Tauopathies include AD and frontotemporal dementia with Parkinsonism (FTDP), and make up the most common variety of neurodegenerative disease.These diseases are characterized by the intracellular accumulation of PHFs composed mainly of posttranslationally modified tau.56,57 Tau is a microtubule-associated protein (MAPT), which in humans is encoded by the MAPT gene.58 It is normally highly soluble and natively unfolded and plays a role in the regulation of cytoskeletal organization.59 Several mutations in the human MAPT gene have been identified that reduce tau’s affinity for microtubules and increase its tendency for aggregation. Mutations for frontotemporal spectrum disorder commonly lie in or near the repeat domain (∆K280, P301L,V337M, R406W), which is important for the affinity of tau to microtubules.60 Caenorhabditis elegans models of tauopathy C. elegans encodes a protein similar to tau, a member of the microtubule-associated protein (MAP) family, ptl-1.61,62 ptl-1 is involved in egg hatching, touch responsiveness, neuronal integrity, and lifespan.63–65 Due to the wide expression of tau in neurons, human tau variants have been expressed using pan-neuronal promoters in C. elegans to model tauopathies,66–69 while targeted expression to a neuronal subpopulation has been used to resolve specific toxic effects.64,70 The first tau model was based on the overexpression of human 1N4RMAPT pan-neuronally; three mutant lines were generated expressing either wild-type MAPT or its FTD mutant variants P201L and V227M, where mutant strains showed a stronger age-progressive uncoordinated phenotype compared to the wild-type strain (Table 3.1).68 Ultrastructural studies of the C. elegans neurons in this model revealed protein inclusions in the axoplasm and axonal degeneration, which is consistent with observations in human patient samples.68 As touch neurons express 15 protofilament microtubules and PTL-1, this neuronal subtype was used to express the FTDP-associated P201L and R406W tau mutant variants, which led to an age-dependent reduction of the touch nose response. Moreover, neurons exhibited morphological abnormalities underlying the observed toxic phenotypes.70 Further pan-neuronal tau models have been generated, expressing wild-type human tau, a pseudohyperphosphorylated (PHP) tau, harboring nine glutamate substitutions mimicking an AD-like tau, and a construct replacing these substitutions with nonphosphorylatable alanine residues (Ala10).66 All three of these transgenic lines exhibited progressive age-dependent uncoordinated phenotypes, although PHP tau expression also induced a defective pattern of motor neuron development.66 In another

53

54

Protein Homeostasis Diseases

study, the mutant 1N4R-tau BV337M was coexpressed in C. elegans pan-neuronally, with either an aggregation prone 4R-tau fragment (F3∆K280) or a version that cannot aggregate (F3∆K280PP).67.These fragments had been used previously to induce or prevent tau aggregation in a human cell model.71 The anti-aggregation fragment led to mild defects in C. elegans, while the aggregation-prone fragment resulted in the accumulation of tau and behavioral defects including paralysis, axonal degeneration of γ-aminobutyric acid (GABA)-ergic and cholinergic motor neurons, and synaptic defects.67 Utility of Caenorhabditis elegans tauopathy models for drug discovery and identification of genetic modifiers C. elegans models of tauopathies have also been used to identify compounds or genes that suppress neurotoxicity. For example, small-molecule drugs bb14 and BSc3094 were previously identified to inhibit tau aggregation as well as improve the uncoordinated phenotype.67 Azaperone, a dopamine D2 receptor antagonist, was also shown to suppress the lethargic crawling defect associated with tau expression.72 A reverse genetic screen conducted in a C. elegans model expressing the tau mutant V337M identified genes that increased tau-induced toxicity, including kinases, chaperones, proteases, and phosphatases.73 Conversely, a forward genetic screen identified sut-1 as a suppressor of the uncoordinated phenotype associated with V337M pan-neuronal expression, suggesting that cytoskeletal function could be involved in the disease mechanism.74

Parkinson’s disease Disease mechanism PD is a neurodegenerative disorder associated with impaired motor control that is the result of the loss of dopaminergic (DA) neurons.75 The disease is characterized by the deposition of insoluble, intracellular inclusions called Lewy bodies within the substantia nigra in patients’ brains.76 The main component of these inclusions is α-synuclein (α-syn), a presynaptic neuronal protein that can aggregate into oligomers and fibrils.76 Most cases of PD are sporadic; however, up to 10% of known cases have genetic factors.75 Mutations in either the α-syn encoding gene PARK1 (A53T, A30P, and E46K) or leucine-rich repeat kinase 2 encoding gene LRRK2 (also known as PARK8) confer the autosomal dominant form of PD.75 Other mutations that have been linked to PD include PARK2, PARK6, and PARK7.77–79 PARK2 encodes parkin, an ubiquitin E3 ligase that modifies a range of proteins and causes autosomal recessive juvenile Parkinsonism.79 PARK6 encodes the PTEN-induced kinase (PINK1) and causes an autosomal recessive form of PD.78 PARK7 encodes protein deglycase DJ-1, which is associated with autosomal recessive juvenile PD and may act as a neuroprotective sensor of oxidative stress.77

Caenorhabditis elegans as a model organism for protein homeostasis diseases

Caenorhabditis elegans Parkinson’s disease models As C. elegans do not have an ortholog of α-syn,80 C. elegans PD models utilize transgenic expression of the human wild-type or mutant α-syn proteins. Expression of wild-type α-syn pan-neuronally, in DA neurons or in motor neurons, resulted in accelerated neuronal and dendritic loss, while expression in the motor neurons additionally caused motor defects.81 Disease-associated aggregating variants of α-syn expressed in C. elegans reduced pharyngeal pumping, motility, and lifespan.82,83 Expression of α-syn tagged with yellow fluorescent protein (YFP) in the body wall muscle of nematodes also showed formation of age-dependent insoluble inclusions and led to an age-dependent and progressive decline in motility.84,85 Other proteins associated with PD have also been investigated in C. elegans, including Parkin, LRRK2, DJ-1, PINK1, and ATP13A2.86–88 Mutations in PINK1, a mitochondrial serine/threonine-protein kinase, are associated with an autosomal recessive and early-onset PD,78 and mutations in the C. elegans homolog of pink-1 result in oxidative stress sensitivity, suggesting an antagonist role of this gene.87 Loss of another C. elegans PD-associated gene ortholog, LRRK2, suppressed the pink-1 mutant phenotype, suggesting an antagonistic role of the two proteins.87 DJ-1 is a redox-sensitive chaperone and a sensor for oxidative stress,77 and knockdown of DJ-1 orthologs in C. elegans were found to protect the nematode from glyoxal-induced death.86 Furthermore, the presence of DJ-1 was shown to protect mouse embryonic fibroblasts from the treatment of glyoxals.86 Utility of Caenorhabditis elegans Parkinson’s disease models for drug discovery and identification of genetic modifiers C. elegans models overexpressing the human α-syn protein have been used to identify genetic suppressors of toxicity using RNAi screens and transcriptome expression analysis.85,89–93 Genome-wide expression analysis using C. elegans expressing either wild-type or A53T mutant α-syn in neurons93 identified UPS and mitochondrial genes to be significantly upregulated, supporting the role of these complexes in mediating α-syn neurotoxicity.93 Another genetic screen for modifiers using a pan-neuronal model of PD identified 10 genes as suppressors of severe motor abnormalities, including four genes related to the endocytic pathway.91 A different RNAi screen using C. elegans expressing α-syn tagged with green fluorescent protein (GFP) in the body wall muscle identified 20 modifiers of aggregation.89 Interestingly, although the screen was performed in animals expressing α-syn in muscle cells, the suppressors also protected against the neurodegeneration induced in a PD DA model.89 In a different study, 80 genes were identified as modulators of α-syn inclusion formation in an α-syn-YFP body wall muscle model (Table 3.1), many of which were shown to be involved in protein quality control and vesicle trafficking in the ER/ Golgi complex.85 One of them, tdo-2, is involved the degradation of L-tryptophan.

55

56

Protein Homeostasis Diseases

Supplementation of the animals with L-tryptophan ameliorated the motility defect and resulted in an increased lifespan.92 This suggests the involvement of the tryptophan metabolism in modulating proteotoxicity, which could be used to investigate therapeutics in the future.92

Polyglutamine diseases Disease mechanism Polyglutamine (PolyQ) diseases are neurodegenerative disorders that are caused by the expansion of the trinucleotide repeat CAG.94 An expansion of 35–40 glutamine repeats causes protein misfolding, oligomerization, and aggregation in vitro and the formation of nuclear inclusions in vivo.95 Expansion of polyQ can arise in different proteins, and there are nine known disorders including HD (Huntingtin), spinal and bulbar muscular atrophy (SBMA; androgen receptor), and SCA (Ataxin). However, the proteins in each of these PolyQ-associated diseases are unrelated in sequence, structure, and function.94 These disorders are inherited in an autosomal dominant manner, except for SBMA, which is X-linked.96 Furthermore, several studies have highlighted the role of the PN in PolyQ diseases, whereby the recruitment of ubiquitin, HSPs, and proteasomal subunits into PolyQ aggregates has been demonstrated.97 Caenorhabditis elegans models of Huntington’s disease The best-characterized PolyQ disease is HD, which causes uncontrolled limb movements, cognitive dysfunction, and mental disorder in humans.94 In C. elegans, HD models have utilized expression of the human huntingtin protein with varying numbers of CAG repeats.98–100 In order explore the effect of the expression of a protein with a PolyQ expansion on neurons, a huntingtin fragment with a 150 polyQ tract (Htt-Q150) was expressed in the ASH (amphid neurons, single ciliated endings) sensory neurons of C. elegans to recapitulate the conditions of the disease.99 This model led to progressive neurodegeneration, and age-dependent protein aggregates of Htt-Q150 were observed in ASH neurons.99 The threshold of PolyQ aggregation in C. elegans was evaluated by expressing different polyQ tract lengths fused to YFP in the body wall muscle.100 These models demonstrated an aggregation threshold of 35–40 glutamine repeats, which is also seen in vitro.95,100 Expression of Q35 fused to YFP in the C. elegans muscle resulted in agedependent aggregation of the protein, accompanied by an increase in proteotoxicity, which leads to loss of motility.100 A similar threshold of 35–40 glutamines was observed when the protein was expressed in C. elegans neurons.98 However, the same protein aggregated less in certain types of neurons. For example, while expression of Q35 and Q40 was soluble in a subset of lateral neurons, it aggregated in the motor neurons of the same animal.98 This suggests neuron-specific differences in the capacity of the PN between different neuronal subsets.

Caenorhabditis elegans as a model organism for protein homeostasis diseases

Utility of Caenorhabditis elegans HD models for drug discovery and identification of genetic modifiers Nematodes contain a disaggregase complex comprising HSP-110, HSP-70, and J-proteins.101 This complex was shown to act on muscle (Q35::YFP) or intestinal (Q44::YFP)specific aggregation of PolyQ repeats in a C. elegans model: when class B J-protein dnj-13 and either class A J-protein dnj-12 or dnj-19 were knocked down in complex by RNAi, aggregation of Q35 was promoted, indicating mixed-class J-protein cooperation affects polyQ-containing protein aggregation.102 Furthermore, RNAi-mediated knockdown of hsp-1, encoding for the C. elegans version of constitutive Hsc70, or hsp-110 in C. elegans expressing muscle-specific Q35::YFP, resulted in increased aggregation of the PolyQ protein.103 In HD patient-derived neural cells, the overexpression of a J-protein, DNAJB1, was sufficient to reduce HttExon1Q97 aggregation, indicating that it could be a target for future therapeutics for HD.103 In C. elegans, a mitochondrial-to-cytosolic stress response (MCSR) was uncovered by perturbing mitochondrial protein homeostasis through depletion of hsp-6, the C. elegans ortholog of Hsp60.This in turn caused upregulation of cytosolic chaperone genes104 and improved cytosolic protein homeostasis, as Q35::YFP animals displayed fewer foci and a slower progression of the characteristic motility defect.100,104 This was further confirmed in a human cell line demonstrating that induction of the MCSR plays a beneficial role in proteostasis and reduces proteotoxicity.104 In a mutagenesis screen performed using the C. elegans model of HD expressing Q40 in muscle cells, modifiers of aggregation (moag), including moag-4, an ortholog of human small EDRK-rich factors, SERF1A and SERF2, were identified.105 SERF1A was previously identified as a genetic modifier of spinal muscular atrophy.106 Knockout of moag-4 reduced the number of aggregates in this model as well as nematode models of AD and PD by 75%.105 Importantly, SERF1A, and SERF2 were also shown to drive mutant huntingtin aggregation in a human neuroblastoma cell line.105 Caenorhabditis elegans models of spinocerebellar ataxia SCA is a progressive and degenerative hereditary neurological disorder. There are many types of SCA, including SCA 3 or Machado-Joseph disease (MJD), which is characterized by the aggregation of Ataxin-3.107 Varying lengths of PolyQ tracts in the Ataxin-3 expressing MJD gene, MJD1, including MJD1-17Q, MJD1-91Q, and MJD-1-130Q, were expressed pan-neuronally fused to a C-terminal GFP tag in C. elegans.108 While a strain expressing MJD1-17Q::GFP exhibited no aggregated foci or behavioral deficits, expression of MJD1-130Q::GFP caused a loss of motility, increased foci formation, and interrupted synaptic transmission via aberrant branching.108 Interesting, a study expressing the Ataxin 3 C-terminal fragment with either 45 or 63 PolyQ repeats tagged with YFP did not exhibit any age-related increase in aggregation of the proteins.109 Instead, the AT3(Q63) model demonstrated severe aggregation throughout adulthood, which

57

58

Protein Homeostasis Diseases

correlated with reduced motility, while the AT3(Q45) model exhibited few foci in early adulthood that did not increase with age.109 Utility of Caenorhabditis elegans spinocerebellar ataxia models for drug discovery and identification of genetic modifiers Expression of AT3(Q14), AT3(Q75), or AT3(Q130) pan-neuronally in C. elegans also showed polyQ length-dependent aggregation and led to dysfunctional motor neurons.110 In the background of a daf-2 mutant, the AT3(Q130) model displayed a decrease in aggregate formation and improved the motor neuron dysfunction, highlighting the importance of the IIS pathway in the maintenance of proteostasis.110 The HSR was also shown to be important in this model:AT3(Q130) expressed in the background of an hsf-1 mutant caused deleterious effects in transgenic animals.110 Furthermore, the compounds tetracycline, methacycline, and Epigallocatechin gallate (EGCG) have been shown to alleviate the defective motility phenotype associated with AT3(Q130) expression.111,112

Amyotrophic lateral sclerosis Disease mechanism Amyotrophic lateral sclerosis (ALS) is a disease that results in the death of motor neurons that control voluntary muscles, and is characterized by muscle stiffness, muscle twitching, and muscle atrophy.113 The majority of cases of ALS are sporadic, but mutations in several proteins, including in the gene superoxidase dismutase 1 (SOD1),114 the DNA/RNA binding proteins TAR DNA binding protein 43 (TDP-43) and fused-in-sarcoma, have been linked to familial ALS.115,116 SOD1 catalyzes the conversion of O2− into O2 and H2O2, and over 160 causative mutations have been found in SOD1 since its initial discovery.117 SOD1 mutations, including A4V, G37R, G85R, G93A, and H17Y, have been hypothesized to cause toxicity through gain of function, starting with an initial misfolding of SOD1 as the primary step in pathogenesis. In ALS, ubiquitinated aggregates are present in the affected regions of the brain and spinal cord and contain a hyperphosphorylated form of TDP-43.118 However, the pathogenic effect of the mutant protein is not well understood and could be due to gain of function, loss of function, or both. It is important to note that mice with elevated expression of wild-type TDP-43 have the characteristics of TDP-43 mutant proteins119; therefore, the expression level should be carefully considered in generating disease models.120 Caenorhabditis elegans models of amyotrophic lateral sclerosis To generate C. elegans models of ALS, two ALS-related human disease proteins have been expressed, namely SOD1 and TDP-43. Although C. elegans possess a SOD1 homolog which has ~71% similarity to the human protein, C. elegans ALS studies have largely utilised the expression of human wild-type SOD1 or ALS-related mutations in this variant.121 Transgenic strains of C. elegans have been created to express human wild-type and

Caenorhabditis elegans as a model organism for protein homeostasis diseases

mutant (A4V, G37R and G93A) SOD1 in most tissues.122 C. elegans expressing mutant SOD1 were more vulnerable to oxidative stress and exhibited discrete aggregates in the adult stage.122 Expression of wild-type SOD1 and mutant variants G85R, G93A, and 127X in C. elegans muscle cells showed that only the mutant SOD1 variants were proteotoxic and led to increased aggregation formation.84 Another human SOD1 mutation, G85R, when expressed pan-neuronally in C. elegans, impaired locomotion and neurotransmitter signaling and resulted in the formation of aggregates in certain mechanosensory neurons.123 However, when the wild-type SOD1 protein was expressed pan-neuronally, no motility or neurodegeneration defect was observed.123 In contrast to this, when Li and colleagues expressed either wild-type SOD1 or G93A SOD1 in the motor neurons of the nematode, phenotypes in paralysis and aggregation were observed, even in the wild-type SOD1 model.124,125 These discrepancies could be due to differences in transgene copy number, promoter expression, and the higher susceptibility of the motor neurons to toxicity compared to other types of neurons.124 More recent studies on the effects of SOD1 in C. elegans have used the C. elegans ortholog of the protein instead of the human protein in modeling the disease.121,126 Deletion of the sod-1 gene in C. elegans causes an increase in cytosolic and mitochondrial superoxide radicals (O2−) levels.126 Single-copy, sod-1 knock-in models, including A4V, H71Y and G85R mutations, showed an increase in SOD1 aggregation and oxidative stress in cholinergic motor neurons. SOD1 mutants H71T and G85R also led to degeneration of glutamatergic neurons, which is consistent with patient data.121 Ubiquitinated aggregates are present in the majority of affected regions in ALS patient brains and spinal cords.118 A major component of these is TDP-43, a protein with numerous functions in transcriptional repression, alternative mRNA splicing, and mRNA stability.118 A C. elegans model expressing wild-type TDP-43 pan-neuronally exhibits an uncoordinated phenotype, abnormal motor neuron synapses, and motor dysfunction, while expression of mutants G290A, A315T, or M337V caused a more severe motor defect.127,128 Hyperphosphorylation of wild-type TDP-43 also correlated with increased toxicity similar to that observed in ALS patients.128

Transthyretin amyloidosis Disease mechanism TTR is a visceral protein, secreted from the liver, which facilitates the transport of thyroid hormones and vitamin A in the blood and cerebrospinal fluid.129 In TTR systemic amyloidosis, the unaffected liver secretes TTR into the bloodstream, where it misfolds and aggregates, affecting other organs such as the heart and nervous system in a cell-nonautonomous proteotoxic manner.130 Misfolding and subsequent aggregation is associated with numerous diseases, including senile systemic amyloidosis (SSA), familial amyloid polyneuropathy (FAP), and familial amyloid cardiomyopathy.129 Aggregation of

59

60

Protein Homeostasis Diseases

wild-type TTR over time is associated with SSA, a sporadic nongenetic condition in which TTR accumulates around the heart, causing shortness of breath and an intolerance to exercise.129 Moreover, over 115 mutations in human TTR have been characterized, which make the protein more likely to aggregate and cause human diseases.131 Of these, the V30M mutation is the most common and causes FAP, a neurodegenerative disorder of the pain- and thermosensing neurons.132 Caenorhabditis elegans disease models of transthyretin amyloidosis A TTR model created alongside the first Aβ model was used to demonstrate its protective effect on Aβ-associated paralysis.26 Expression in C. elegans neurons of the FAP-associated mutation V30M showed that this exhibited nociceptive defects and morphological impairments, whereas these effects were not observed in the nonaggregating mutant T119M or the poorly secreted TTR mutant D18G variant, thus suggesting cell-nonautonomous proteotoxicity similar to that seen in TTR amyloidosis in humans.131 Amyloid deposits in patients have been found to contain C-terminal fragments of TTR as well as full-length TTR.133 Several types of fragments fused to GFP have been expressed in C. elegans body wall muscle cells, where nematodes expressing the C-terminal 81-127 residues of TTR developed aggregates and reduced lifespan and defective motility compared to the other fragments.133 This model was used for evaluating a component of green tea, EGCG, which was subsequently found to significantly inhibit the formation of aggregates and defective motility in this C-terminal TTR fragment strain.133

Type II diabetes mellitus Disease mechanism T2DM is characterized by insulin resistance and the loss of pancreatic β-cells caused by toxic aggregates resulting from the misfolding of human islet amyloid polypeptide (h-IAPP) 134. β-Cell loss begins early in the prognosis of the disease, and by diagnosis, 40%–60% of β-cell volume is lost. h-IAPP is a 37-amino acid hormone that is cosecreted with insulin, playing a role in glycemic regulation and gastric emptying.135 Caenorhabditis elegans model of diabetes mellitus An initial C. elegans model was created to express h-IAPP and mouse IAPP (m-IAPP), a nonamyloidogenic variant of IAPP, in the body wall muscle, pharynx, and neurons. The h-IAPP model exhibited a growth retardation phenotype, while the m-IAPP model did not exhibit any phenotypes. Chemical induction of HSPA1A, significantly improved this phenotype.136 Another C. elegans model was generated to express h-IAPP under the control of a heat-inducible promoter where, upon temperature upshift, the model displayed a chemotaxis defect.137

Caenorhabditis elegans as a model organism for protein homeostasis diseases

Dialysis-related amyloidosis Disease mechanism DRA is characterized by the deposition of β2-microglobulin (β2m) amyloid deposition in osteoarticular tissues.138 β2m, a component of the major histocompatibility complex class I, normally dissociates from cells and is removed from the system via the kidneys.138 However, in patients undergoing chronic hemodialysis treatment, β2m is not removed from the system, and levels rise by 60-fold. The increased concentrations of β2m leads to accumulation and aggregation in the plasma membrane of osteoarticular tissues due to the presence of collagen and glycosaminoglycans, which have been shown in vitro to be potent promoters of β2m amyloidogenesis.139 These deposits result in bone cysts, joint arthropathy, and carpal tunnel syndrome.138 C. elegans expressing wild-type, ∆N6, and P32G variants in the body wall muscle result in developmental defects when expressing in the highly amyloidogenic ∆N6 and P32G variants, but not with the wild-type β2m protein.140 All three β2m variants are, however, proteotoxic, resulting in decreased motility. Treatment with tetracyclines— which have previously shown to prevent Aβ aggregation—suppressed this proteotoxic effect.140,141

Immunoglobulin light chain amyloidosis Disease mechanism AL amyloidosis is caused by the accumulation and aggregation of monoclonal immunoglobulin light chains (LCs) produced by a bone marrow plasma cell clone. The disease is heterogeneous, and therefore, target organs vary from patient to patient; however, involvement of cardiac tissue is present in up to 75% of patients and can lead to chronic heart failure or fatal arrhythmias.142 Caenorhabditis elegans models of immunoglobulin light chain amyloidosis In order to mimic the effects of cardiotoxic LCs, a C. elegans–based assay has been established to investigate the underlying mechanisms of toxicity: the pharynx of C. elegans is considered to be evolutionarily related to the vertebrate heart, and thus, the effect of different organ tropisms from AL patients on pharyngeal pumping has been evaluated.143 Amyloidogenic LCs that are cardiotoxic in patients caused an impaired pharyngeal pumping rate in C. elegans.144 This pharyngeal impairment was dependent on LC concentration and resulted in a reduction in the lifespan of the animal, which was shown to be rescued by treatment with antioxidant agents.144

Prion diseases A characteristic of several neurodegenerative diseases is the spread of toxic, soluble oligomers across cells and tissues, thereby invading cells to propagate their aggregation-prone conformation in a prion-like process.145 One such class of proteins are prions, which are

61

62

Protein Homeostasis Diseases

self-propagating aggregates that underlie the infectious nature of transmissible spongiform encephalopathies in mammals.146 A prion model was created in C. elegans expressing the cytosolic yeast prion protein Sup35 in muscle cells.147 Sup35 is a highly toxic aggregate species resulting in severe toxicity and cellular dysfunction, with aggregates targeted to the autophagy pathway.147 The fluorescently tagged prion proteins were also observed in nonexpressing tissues of the nematode, demonstrating cell-to-cell transmission of the prion protein.147 Thus, future studies into the prion-like nature of other neurodegenerative disease proteins, including Aβ and tau, could further aid our understanding of this phenomenon.

Conclusion Animal models of protein homeostasis diseases have been instrumental in aiding our understanding of these diseases and the search for preventative measures and cures. Although higher-order animals such as mice mimic the human body more closely, lower-order animals such as C. elegans have been vital in aiding our understanding of the underlying molecular mechanisms of protein homeostasis disease. Despite a large evolutionary distance between nematodes and humans, human disease proteins expressed in C. elegans aggregate in an age-dependent manner and cause comparable cellular toxicity and phenotypic characteristics as observed in human cells. Further, as most cellular processes and protein quality control pathways are conserved between humans and nematodes, C. elegans models aid in the development of therapeutic interventions for humans, whereby nematodes can be used to screen medications before testing on higher-order animals.The simple nature and availability of strong genetic and imaging tools makes this nematode an extremely valuable bridge between in vitro studies and humans that will continue to aid in our understanding of these diseases.

References 1. Brenner S. The genetics of Caenorhabditis elegans. Genetics 1974;77:71–94. 2. Sulston JE, Schierenberg E, White JG, Thomson JN. The embryonic cell lineage of the nematode Caenorhabditis elegans. Dev Biol 1983;100:64–119. 3. Shaye DD, Greenwald I. Ortholist: a compendium of C. elegans genes with human orthologs. PLoS One 2011;6:e20085. 4. Labbadia J, Morimoto RI. The biology of proteostasis in aging and disease. Annu Rev Biochem 2015;84:435–64. 5. Brehme M,Voisine C, Rolland T, Wachi S, Soper JH, Zhu Y, Orton K,Villella A, Garza D,Vidal M, Ge H, Morimoto RI. A chaperome subnetwork safeguards proteostasis in aging and neurodegenerative disease. Cell Rep 2014;9:1135–50. 6. Balchin D, Hayer-Hartl M, Hartl FU. In vivo aspects of protein folding and quality control. Science 2016;353:aac4354. 7. Lindquist S. The Heat Shock Reponse. Annu Rev Biochem 1986;1151–91. 8. Hajdu-Cronin YM, Chen WJ, Sternberg PW. The L-type cyclin CYL-1 and the heat-shock-factor HSF-1 are required for heat-shock-induced protein expression in Caenorhabditis elegans. Genetics 2004;168:1937–49.

Caenorhabditis elegans as a model organism for protein homeostasis diseases

9. Abravaya K, Myers MP, Murphy SP, Morimoto RI.The human heat shock protein hsp70 interacts with HSF, the transcription factor that regulates heat shock gene expression. Genes Dev 1992;6:1153–64. 10. Nadeau K, Das A,Walsh CT. Hsp90 chaperonins possess ATPase activity and bind heat shock transcription factors and peptidyl prolyl isomerases. J Biol Chem 1993;268:1479–87. 11. Shi Y, Mosser DD, Morimoto RI. Molecular chaperones as HSF1-specific transcriptional repressors. Genes Dev 1998;12:654–66. 12. Anckar J, Sistonen L. Regulation of HSF 1 Function in the Heat Stress Response: Implications in Aging and Disease. Annu Rev Biochem 2011;80:1089–115. 13. Safra M, Ben-Hamo S, Kenyon C, Henis-Korenblit S.The ire-1 ER stress-response pathway is required for normal secretory-protein metabolism in C. elegans. J Cell Sci 2013;126:4136–46. 14. Shen X, Ellis RE, Sakaki K, Kaisfman RJ. Genetic interactions due to constitutive inducible gene regulation mediated by the unfolded protein in C. PLoS Genet 2005;1:355–68. 15. Walter P, Ron D. The Unfolded Protein Response: From Stress Pathway to Homeostatic Regulation. Science 2011;334:1081–1087. 16. Pellegrino MW, Nargund AM, Haynes CM. Signaling the mitochondrial unfolded protein response. Biochim Biophys Acta—Mol Cell Res 2013;1833:410–6. 17. Neupert W, Herrmann JM. Translocation of proteins into mitochondria. Annu Rev Biochem 2007;76:723–49. 18. Haynes CM, Petrova K, Benedetti C, Yang Y, Ron D. ClpP Mediates Activation of a Mitochondrial Unfolded Protein Response in C. elegans. Dev Cell 2007;13:467–80. 19. Haynes CM, Yang Y, Blais SP, Neubert TA, Ron D. The Matrix Peptide Exporter HAF-1 Signals a Mitochondrial UPR by Activating the Transcription Factor ZC376.7 in C. elegans. Mol Cell 2010;37: 529–40. 20. Kenyon CJ. The genetics of ageing. Nature 2010;464:504–12. 21. Dong M-Q,Venable JD, Au N, Xu T, Park SK, Cociorva D, Johnson JR, Dillin A,Yates III JR. Quantitative Mass Spectrometry Identifies Insulin Signaling Targets in C. elegans. Science 2007;317:660–3. 22. Halaschek-Wiener J, Khattra JS, McKay S, Pouzyrev A, Stott JM,Yang GS, Holt RA, Jones SJM, Marra MA, Brooks-Wilson AR, Riddle DL. Analysis of long-lived C. elegans daf-2 mutants using serial analysis of gene expression. Genome Res 2005;15:603–15. 23. McElwee JJ, Schuster E, Blanc E, Thomas JH, Gems D. Shared transcriptional signature in Caenorhabditis elegans dauer larvae and long-lived daf-2 mutants implicates detoxification system in longevity assurance. J Biol Chem 2004;279:44533–43. 24. Murphy CT, McCarroll SA, Lieb JD, Bargmann CI, Kamath RS, Fraser A, Ahringer J, Li H, Kenyon CJ. Genes that act downstream of DAF-16 to influence C. elegans lifespan. Nature 2003;424:277–83. 25. Link CD. Expression of human B-amyloid peptide in transgenic Caenorhabditis elegans. Neurobiology 1995;92:9368–72. 26. Hipp MS, Park SH, Hartl UU. Proteostasis impairment in protein-misfolding and -aggregation diseases. Trends Cell Biol 2014;. 27. Dolphin CT, Hope IA. Caenorhabditis elegans reporter fusion genes generated by seamless modification of large genomic DNA clones. Nucleic Acids Res 2006;34. 28. Huang Y, Mucke L. Alzheimer mechanisms and therapeutic strategies. Cell 2012;148:1204–22. 29. Zhang Y,Thompson R, Zhang H, Xu H. APP processing in Alzheimer’s disease. Mol Brain 2011;4:1–13. 30. Iwatsubo T, Odaka A, Suzuki N, Mizusawa H, Nukina N, Ihara Y.Visualization of Aβ42(43) and Aβ40 in senile plaques with end-specific Aβ monoclonals: Evidence that an initially deposited species is Aβ42(43). Neuron 1994;13:45–53. 31. Rogeav EI, Sherrington R, Rogeava EA, Levesque G, Ikeda M, Liang Y, Chi H, Lin C, Holman K, Tsuda T, Mar L, Sorbi S, Nacmias B, Placentini S, Amaducci L, Chumakov L, Cohen D, Lannfelt L, Fraser PE, Rommens JM, St George-Hyslop PH. Familial AD in kindreds with missense mutation in a gene on Chr1 related to AD type 3 gene. Nature 1995;376:775–8. 32. Sherrington R, Rogaev EI, Liang Y, Rogaeva EA, Levesque G, Ikeda M, Chi H, Lin C, Li G, Holman K,Tsuda T, Mar L, Foncin J-F, Bruni AC, Montesi MP, Sorbi S, Rainero I, Pinessi L, Nee L, Chumakov I, Pollen D, Brookes A, Sanseau P, Polinsky RJ,Wasco W, Da Silva HAR, Haines JL, Pericak-Vance MA, Tanzi RE, Roses AD, Fraser PE, Rommens JM, St George-Hyslop PH. Cloning of a gene bearing missense mutations in early-onset familial Alzheimer’s disease. Nature 1995;375:754–60.

63

64

Protein Homeostasis Diseases

33. Bertram L, Lill CM, Tanzi RE. The genetics of alzheimer disease: Back to the future. Neuron 2010;68:270–81. 34. Daigle, I., Li, C., 1993. api-i, a Caenorhabditis elegans gene encoding a protein related to the human f8-amyloid protein precursor 90, 12045-12049. 35. Link CD, Johnson CJ, Fonte V, Paupard MC, Hall DH, Styren S, Mathis CA, Klunk WE.Visualization of fibrillar amyloid deposits in living, transgenic Caenorhabditis elegans animals using the sensitive amyloid dye, X-34. Neurobiol Aging 2001;22:217–26. 36. Cohen E, Bieschke J, Perciavalle RM, Kelly JW, Dillin A. Opposing activities protect against age-onset proteotoxicity. Science 2006;313:1604–10. 37. McColl G, Roberts BR, Gunn AP, Perez KA, Tew DJ, Masters CL, Barnham KJ, Cherny RA, Bush AI. The Caenorhabditis elegans AB1-42 model of Alzheimer disease predominantly Expresses AB3-42. J Biol Chem 2009;284:22697–702. 38. McColl G, Roberts BR, Pukala TL, Kenche VB, Roberts CM, Link CD, Ryan TM, Masters CL, Barnham KJ, Bush AI, Cherny Ra. Utility of an improved model of amyloid-beta (Aβ1-42) toxicity in Caenorhabditis elegans for drug screening for Alzheimer’s disease. Mol Neurodegener 2012;7:57. 39. Link CD, Taft A, Kapulkin V, Duke K, Kim S, Fei Q, Wood DE, Sahagan BG. Gene expression analysis in a transgenic Caenorhabditis elegans Alzheimer’s disease model. Neurobiol Aging 2003;24:397–413. 40. Mango SE. Stop making nonsense: The C. elegans smg genes. Trends Genet 2001;17:646–53. 41. Fay DS, Fluet a, Johnson CJ, Link CD. In vivo aggregation of beta-amyloid peptide variants. J Neurochem 1998;71:1616–25. 42. Hilbich, C., Kisters-Woike, B., 1992. Substitution of Hydrophobic Amino Acids Reduces Amyloidogenicity of A-beta peptide. 43. Soto C, Castano EM, Frangione B, Inestrosa NC. The a-Helical to B-Strand Transition in the Amino-terminal Fragment of the Amyloid B-Peptide Modulates Amyloid Formation. J Biol Chem 1995;270:3063–7. 44. Wu Y,Wu Z, Butko P, Christen Y, Lambert MP, Klein WL, Link CD, Luo Y. Amyloid-B-induced pathological behaviors are suppressed by ginkgo biloba extract EGb 761 and ginkgolides in transgenic Caenorhabditis elegans. J Neurosci 2006;26:13102–13. 45. Dosanjh LE, Brown MK, Rao G, Link CD, Luo Y. Behavioral phenotyping of a transgenic Caenorhabditis elegans expressing neuronal amyloid-β. J Alzheimer’s Dis 2010;19:681–90. 46. Treusch S, Hamamichi S, Goodman JL, Matlack KES, Chung CY, Baru V, Shulman JM, Parrado A, Bevis BJ,Valastyan JS, Han H, Lindhagen-persson M, Reiman EM, Evans DA, Bennett DA, Olofsson A, Dejager PL,Tanzi RE, Caldwell KA, Caldwell GA, Lindquist S. Functional Links Between AB Toxicity, Endocytic Trafficking, and Alzheimer’s Disease Risk Factors in Yeast. Science 2011;334:1241–5. 47. Sorrentino V, Romani M, Mouchiroud L, Beck JS, Zhang H, D’Amico D, Moullan N, Potenza F, Schmid AW, Rietsch S, Counts SE, Auwerx J. Enhancing mitochondrial proteostasis reduces amyloid-β proteotoxicity. Nature 2017;552:187–93. 48. Fonte V, Kapulkin WJ, Taft A, Fluet A, Friedman D, Link CD. Interaction of intracellular β amyloid peptide with chaperone proteins. Proc Natl Acad Sci 2002;99:9439–44. 49. Fonte V, Kipp DR,Yerg J, Merin D, Forrestal M, Wagner E, Roberts CM, Link CD. Suppression of in vivo B-amyloid peptide toxicity by overexpression of the HSP-16.2 small chaperone protein. J Biol Chem 2008;283:784–91. 50. Wu Y, Cao Z, Klein WL, Luo Y. Heat shock treatment reduces beta amyloid toxicity in vivo by diminishing oligomers. Neurobiol Aging 2010;31:1055–8. 51. Muñoz-Lobato F, Rodríguez-Palero MJ, Naranjo-Galindo FJ, Shephard F, Gaffney CJ, Szewczyk NJ, Hamamichi S, Caldwell KA, Caldwell GA, Link CD, Miranda-Vizuete A. Protective role of DNJ-27/ ERdj5 in Caenorhabditis elegans models of human neurodegenerative diseases. Antioxid Redox Signal 2014;20:217–35. 52. Steinkraus KA, Smith ED, Davis C, Carr D, Pendergrass WR, Sutphin GL, Kennedy BK, Kaeberlein M. Dietary restriction suppresses proteotoxicity and enhances longevity by an hsf-1-dependent mechanism in Caenorhabditis elegans. Aging Cell 2008;7:394–404. 53. Liu X, Hao W, Qin Y, Decker Y, Wang X, Burkart M, Schötz K, Menger MD, Fassbender K, Liu Y. Long-term treatment with Ginkgo biloba extract EGb 761 improves symptoms and pathology in a transgenic mouse model of Alzheimer’s disease. Brain Behav Immun 2015;46:121–31.

Caenorhabditis elegans as a model organism for protein homeostasis diseases

54. Tan MS,Yu JT, Tan CC, Wang HF, Meng XF, Wang C, Jiang T, Zhu XC, Tan L. Efficacy and adverse effects of Ginkgo Biloba for cognitive impairment and dementia: A systematic review and meta-analysis. J Alzheimer’s Dis 2015;43:589–603. 55. Grundke-iqbal I, Iqbal K, Tung Y, Quinlan M, Wisniewski HM, Bindert LI. Abnormal phosphorylation of the microtubule-associated protein tau (tau) in Alzheimer cytoskeletal pathology. PNAS 1986;83:4913–7. 56. Kidd M. Paired helical filaments in Alzheimer’s Disease. Nature 1963;197:192–3. 57. Neve RL, Harris P, Kosik KS, Kurnit DM, Donlon TA. Identification of cDNA clones for the human microtubule-associated protein tau and chromosomal localization of the genes for tau and microtubule-associated protein 2. Mol. Brain Res 1986;1:271–80. 58. Wille H. Dimers and Antiparallel Helical Filaments Paired Tau In Vitro Protein from Microtubuleassociated Formed. J Cell Biol 2011;118:573–84. 59. Barghorn S, Zheng-Fischhofer Q, Ackmann M, Biernat J, Von Bergen M, Mandelkow EM, Mandelkow E. Structure, microtubule interactions, and paired helical filament aggregation by tau mutants of frontotemporal dementias. Biochemistry 2000;39:11714–21. 60. Goedert M, Baur CP, Ahringer J, Jakes R, Hasegawa M, Spillantini MG, Smith MJ, Hill F. PTL-1, a microtubule-associated protein with tau-like repeats from the nematode Caenorhabditis elegans. J Cell Sci 1996;109(Pt 1):2661–72. 61. McDermott JB, Aamodt S, Aamodt E. ptl-1, a Caenorhabditis elegans gene whose products are homologous to the τ microtubule-associated proteins. Biochemistry 1996;35:9415–23. 62. Chew YL, Fan X, Götz J, Nicholas HR. Regulation of age-related structural integrity in neurons by protein with tau-like repeats (PTL-1) is cell autonomous. Sci Rep 2014;4:1–9. 63. Chew YL, Fan X, Gotz J, Nicholas HR. PTL-1 regulates neuronal integrity and lifespan in C. elegans. J Cell Sci 2013;126:2079–91. 64. Gordon P, Hingula L, Krasny ML, Swienckowski JL, Pokrywka NJ, Raley-Susman KM. The invertebrate microtubule-associated protein PTL-1 functions in mechanosensation and development in Caenorhabditis elegans. Dev Genes Evol 2008;218:541–51. 65. Brandt R, Gergou A,Wacker I, Fath T, Hutter H. A Caenorhabditis elegans model of tau hyperphosphorylation: Induction of developmental defects by transgenic overexpression of Alzheimer’s disease-like modified tau. Neurobiol Aging 2009;30:22–33. 66. Fatouros C, Pir GJ, Biernat J, Koushika SP, Mandelkow E, Mandelkow EM, Schmidt E, Baumeister R. Inhibition of Tau aggregation in a novel Caenorhabditis elegans model of tauopathy mitigates proteotoxicity. Hum Mol Genet 2012;21:3587–603. 67. Kraemer BC, Zhang B, Leverenz JB, Thomas JH, Trojanowski JQ, Schellenberg GD. Neurodegeneration and defective neurotransmission in a Caenorhabditis elegans model of tauopathy. Proc Natl Acad Sci 2003;100:9980–5. 68. Pir GJ, Choudhary B, Mandelkow E, Mandelkow EM. Tau mutant A152T, a risk factor for FTD/ PSP, induces neuronal dysfunction and reduced lifespan independently of aggregation in a C. elegans Tauopathy model. Mol Neurodegener 2016;11:1–21. 69. Miyasaka T, Ding Z, Gengyo-Ando K, Oue M,Yamaguchi H, Mitani S, Ihara Y. Progressive neurodegeneration in C. elegans model of tauopathy. Neurobiol Dis 2005;20:372–83. 70. Khlistunova I, Biernat J, Wang Y, Pickhardt M, Von Bergen M, Gazova Z, Mandelkow E, Mandelkow EM. Inducible expression of tau repeat domain in cell models of tauopathy: Aggregation is toxic to cells but can be reversed by inhibitor drugs. J Biol Chem 2006;281:1205–14. 71. McCormick AV,Wheeler JM, Guthrie CR, Liachko NF, Kraemer BC. Dopamine D2 receptor antagonism suppresses tau aggregation and neurotoxicity. Biol Psychiatry 2013;73:464–71. 72. Kraemer BC, Burgess JK, Chen JH, Thomas JH, Schellenberg GD. Molecular pathways that influence human tau-induced pathology in Caenorhabditis elegans. Hum Mol Genet 2006;15:1483–96. 73. Kraemer BC, Schellenberg GD. SUT-1 enables tau-induced neurotoxicity in C. elegans. Hum Mol Genet 2007;16:1959–71. 74. Tan E-K, Skipper LM. Pathogenic Mutations in Parkinson Disease. Hum Mutat 2007;28:641–53. 75. Takeda A, Mallory M, Sundsmo M, Honer W, Hansen L, Masliah E. Abnormal accumulation of NACP/ alpha-synuclein in neurodegenerative disorders. Am J Pathol 1998;152:367–72.

65

66

Protein Homeostasis Diseases

76. Bonifati V, Rizzu P, van Baren MJ, Schaap O, Breedveld GJ, Kriegar E, Dekker MCJ, Squitieri F, Ibanez P, Joosse M, van Dongen JW,Vanacore N, van Switen JC, Brice A, Meco G, van Duijn CM, Oostra BA, Heutink P. Mutations in the DJ-1 gene associated with autosomal recessive early-onset parkinsonism. Science 2003;299:256–9. 77. Valente EM, Caputo V, Muqit MMK, Harvey K, Gispert S, Ali Z, Del Turco D, Bentivoglio AR, Healy DG, Albanese A, Nussbaum R, Gonzalez-Maldonado R, Deller T, Sergio Salvi P, Abou-Sleiman PM. Hereditary Early-Onset Parkinson’s Disease Caused by Mutations in PINK1. Science 2004;304:4. 78. Zhang Y, Gao J, Chung KKK, Huang H, Dawson VL, Dawson TM. Parkin functions as an E2-dependent ubiquitin- protein ligase and promotes the degradation of the synaptic vesicle-associated protein CDCrel-1. Proc Natl Acad Sci 2000;97:13354–9. 79. Harrington AJ, Hamamichi S, Caldwell GA, Caldwell KA. C. elegans as a model organism to investigate molecular pathways involved with Parkinson’s disease. Dev Dyn 2010;239:1282–95. 80. Lakso M, Vartiainen S, Moilanen A-M, Sirviö J, Thomas JH, Nass R, Blakely RD, Wong G. Dopaminergic neuronal loss and motor deficits in Caenorhabditis elegans overexpressing human α-synuclein. J Neurochem 2003;86:165–72. 81. Karpinar DP, Balijia MBG, Kugler S, Opazo F, Rezaei-Ghaleh N, Wender N, Kim H-Y, Taschenberger BH, Heise H, Kumar A, Riedel D, Fichtner L,Voigt A, Braus GH, Giller K, Becker S, Herzig A, Baldus M, Jackle H, Eimer S, Schulz JB, Griesinger C, Zweckstetter M. Pre-fibrillar a-synuclein variants with impaired B-structure increase neurotoxicity in Parkinson’s disease models. EMBO J 2009;28:3256–68. 82. Kuwahara T, Koyama A, Gengyo-Ando K, Masuda M, Kowa H, Tsunoda M, Mitani S, Iwatsubo T. Familial Parkinson mutant a-synuclein causes dopamine neuron dysfunction in transgenic Caenorhabditis elegans. J Biol Chem 2006;281:334–40. 83. Gidalevitz T, Krupinski T, Garcia S, Morimoto RI. Destabilizing protein polymorphisms in the genetic background direct phenotypic expression of mutant SOD1 toxicity. PLoS Genet 2009;5. 84. Van Ham TJ, Thijssen KL, Breitling R, Hofstra RMW, Plasterk RHA, Nollen EAA. C. elegans model identifies genetic modifiers of α-synuclein inclusion formation during aging. PLoS Genet 2008;4. 85. Lee JY, Song J, Kwon K, Jang S, Kim C, Baek K, Kim J, Park C. Human DJ-1 and its homologs are novel glyoxalases. Hum Mol Genet 2012;21:3215–25. 86. Sämann J, Hegermann J, von Gromoff E, Eimer S, Baumeister R, Schmidt E. Caenorhabditits elegans LRK-1 and PINK-1 act antagonistically in stress response and neurite outgrowth. J Biol Chem 2009;284:16482–91. 87. Springer W, Hoppe T, Schmidt E, Baumeister R. A Caenorhabditis elegans Parkin mutant with altered solubility couples α-synuclein aggregation to proteotoxic stress. Hum Mol Genet 2005;14:3407–23. 88. Hamamichi S, Rivas RN, Knight AL, Cao S, Caldwell Ka, Caldwell Ga. Hypothesis-based RNAi screening identifies neuroprotective genes in a Parkinson’s disease model. Proc Natl Acad Sci USA 2008;105:728–33. 89. Jadiya P, Fatima S, Baghel T, Mir SS, Nazir A. A systematic RNAi screen of neuroprotective genes identifies novel modulators of alpha-synuclein-associated effects in transgenic Caenorhabditis elegans. Mol Neurobiol 2016;53:6288–300. 90. Kuwahara T, Koyama A, Koyama S, Yoshina S, Ren CH, Kato T, Mitani S, Iwatsubo T. A systematic RNAi screen reveals involvement of endocytic pathway in neuronal dysfunction in α-synuclein transgenic C. elegans. Hum Mol Genet 2008;17:2997–3009. 91. van der Goot AT, Zhu W,Vazquez-Manrique RP, Seinstra RI, Dettmer K, Michels H, Farina F, Krijnen J, Melki R, Buijsman RC, Silva MR, Thijssen KL, Kema IP, Neri C, Oefner PJ, Nollen EAA. Delaying ageing and the ageing-associated decline in protein homeostasis by inhibition of tryptophan degradation. PNAS 2012;109:14912–7. 92. Vartiainen S, Pehkonen P, Lakso M, Nass R, Wong G. Identification of gene expression changes in transgenic C. elegans overexpressing human α-synuclein. Neurobiol Dis 2006;22:477–86. 93. Lieberman AP, Shakkottai VG, Albin RL. Polyglutamine Repeats in Neurodegenerative Diseases. Annu Rev Pathol Mech Dis 2019;14:1–27. 94. Duyao M, Ambrose C, Myers R, Novelletto A, Persichetti F, Frontali M, Folstein S, Ross C, Franz M, Abbott M, Gray J, Conneally P,Young A, Penney J, Hollingsworth Z, Shoulson I, Lazzarini A, Falek A, Koroshetz W, Sax D, Bird E,Vonsattel J, Bonilla E, Alvir J, Bickham Conde J, Cha J-H, Dure L, Gomez F, Ramos M, Sanchez-Ramos J, Snodgrass S, de Young M, Wexler N, Moscowitz C, Penchaszadeh G,

Caenorhabditis elegans as a model organism for protein homeostasis diseases

MacFarlane H, Anderson M, Jenkins B, Srinidhi J, Barnes G, Gusella J, MacDonald M. Trinucleotide repeat length instability and age of onset in Huntington’s disease. Nat Genet 1993;4:387–92. 95. Orr HT, Zoghbi HY. Trinucleotide Repeat Disorders. Annu Rev Neurosci 2007;30:575–621. 96. Chai Y, Koppenhafer SL, Bonini NM, Paulson HL. Analysis of the role of heat shock protein (Hsp) molecular chaperones in polyglutamine disease. J Neurosci 1999;19:10338–47. 97. Brignull HR, Moore FE, Tang SJ, Morimoto RI. Polyglutamine proteins at the pathogenic threshold display neuron-specific aggregation in a pan-neuronal Caenorhabditis elegans model. J Neurosci 2006;26:7597–606. 98. Faber PW, Alter JR, MacDonald ME, Hart AC. Polyglutamine-mediated dysfunction and apoptotic death of a Caenorhabditis elegans sensory neuron. Proc Natl Acad Sci USA 1999;96:179–84. 99. Morley JF, Brignull HR, Weyers JJ, Morimoto RI. The threshold for polyglutamine-expansion protein aggregation and cellular toxicity is dynamic and influenced by aging in Caenorhabditis elegans. Proc Natl Acad Sci USA 2002;99:10417–22. 100. Nillegoda NB, Kirstein J, Szlachcic A, Berynskyy M, Stank A, Stengel F, Arnsburg K, Gao X, Scior A, Aebersold R, Guilbride DL, Wade RC, Morimoto RI, Mayer MP, Bukau B. Crucial HSP70 cochaperone complex unlocks metazoan protein disaggregation. Nature 2015;524:247–51. 101. Kirstein J, Arnsburg K, Scior A, Szlachcic A, Guilbride DL, Morimoto RI, Bukau B, Nillegoda NB. In vivo properties of the disaggregase function of J-proteins and Hsc70 in Caenorhabditis elegans stress and aging. Aging Cell 2017;16:1414–24. 102. Scior A, Buntru A, Arnsburg K, Ast A, Iburg M, Juenemann K, Pigazzini ML, Mlody B, Puchkov D, Priller J, Wanker EE, Prigione A, Kirstein J. Complete suppression of Htt fibrilization and disaggregation of Htt fibrils by a trimeric chaperone complex. EMBO J 2018;37:282–99. 103. Kim HE, Grant AR, Simic MS, Kohnz RA, Nomura DK, Durieux J, Riera CE, Sanchez M, Kapernick E, Wolff S, Dillin A. Lipid Biosynthesis Coordinates a Mitochondrial-to-Cytosolic Stress Response. Cell 2016;166 1539-1552.e16. 104. van Ham TJ, Holmberg MA, van der Goot AT, Teuling E, Garcia-Arencibia M, Kim HE, Du D, Thijssen KL, Wiersma M, Burggraaff R, van Bergeijk P, van Rheenen J, Jerre van Veluw G, Hofstra RMW, Rubinsztein DC, Nollen EAA. Identification of MOAG-4/SERF as a regulator of age-related proteotoxicity. Cell 2010;142:601–12. 105. Scharf JM, Endrizzi MG, Wetter A, Huang S, Thompson TG, Zerres K, Dietrich WF, Wirth B, Kunkel LM. Identification of a candidate modifying gene for spinal muscular atrophy by comparative genomics. Nat Genet 1998;20:83–6. 106. Kawaguchi Y, Okamoto T, Taniwaki M, Aizawa M, Inoue M, Katayama S, Kawakami H, Nakamura S, Nishimura M, Akiguchi I, Kimura J, Narumiya S, Kakizuka A. CAG expansions in a novel gene for Machado-Joseph disease at chromosome 14q32.1. Nat Genet 1994;8:221–8. 107. Khan LA, Bauer PO, Miyazaki H, Lindenberg KS, Landwehrmeyer BG, Nukina N. Expanded polyglutamines impair synaptic transmission and ubiquitin – proteasome system in Caenorhabditis elegans. J Neurochem 2006;98:576–87. 108. Christie NTM, Lee AL, Fay HG, Gray AA, Kikis EA. Novel Polyglutamine Model Uncouples Proteotoxicity from Aging 2014;9. 109. Teixeira-Castro A, Ailion M, Jalles A, Brignull HR,Vilaca JL, Dias N, Rodrigues P, Oliveira JF, Carvalho AN, Morimoto RI, Maciel P. Neuron-specific proteotoxicity of mutant ataxin-3 in C. elegans: rescue by the DAF-16 and HSF-1 pathways. Hum Mol Genet 2011;20:2996–3009. 110. Amigoni L, Airoldi C, Natalello A, Romeo M, Diomede L, Tortora P, Elena M. Methacycline displays a strong efficacy in reducing toxicity in a SCA3 Caenorhabditis elegans model. BBA—Gen Subj 2019;1863:279–90. 111. Bonanomi M, Natalello A, Visentin C, Pastori V, Cornelli G, Colombo G, Malabarba MG, Doglia SM, Relini A, Regonesi ME, Tortora P. Epigallocatechin-3-gallate and tetracycline differently affect ataxin-3 fibrillogenesis and reduce toxicity in spinocerebellar ataxia type 3 model. Hum Mol Genet 2014;24:6542–52. 112. Chou TT, Trojanowski JQ, Kwong LK, Masliah E, Grossman M, Lee VM-Y, Feldman H, McCluskey LF, Clark CM, Miller BL,Truax AC, Kretzschmar HA, Bruce J, Micsenyi MC, Mackenzie IR, Schuck T, Feiden W, Sampathu DM, Neumann M. Ubiquitinated TDP-43 in Frontotemporal Lobar Degeneration and Amyotrophic Lateral Sclerosis. Science 2006;314:130–3.

67

68

Protein Homeostasis Diseases

113. Rosen DR, Siddique T, Patterson D, Figlewicz DA, Sapp P, Hentati A, Donaldson D, Goto J, O’Regan JP, Deng H-X, Rahmani Z, Krizus A, McKenna-Yasek D, Cayabyab A, Gaston SM, Berger R, Tanzi RE, Halperin JJ, Herzfeldt B, Van den Bergh R, Hung W-Y, Bird T, Deng G, Mulder DW, Smyth C, Laing NG, Soriano E, Pericak–Vance MA, Haines J, Rouleau GA, Gusella JS, Horvitz HR, Brown RH, Rosen, et al. Mutations in Cu / Zn superoxide dismutase gene are associated. Nature 1993;362:59–62. 114. Kabashi E,Valdmanis PN, Dion P, Spiegelman D, McConkey BJ,Vande Velde C, Bouchard JP, Lacomblez L, Pochigaeva K, Salachas F, Pradat PF, Camu W, Meininger V, Dupre N, Rouleau GA.TARDBP mutations in individuals with sporadic and familial amyotrophic lateral sclerosis. Nat Genet 2008;40:572–4. 115. Sreedharan J, Blair IP, Tripathi VB, Hu X, Vance C, Rogelj B, Ackerley S, Durnall JC, Williams KL, Buratti E, Baralle F, de Belleroche J, Mitchell JD, Leigh PN, Al-Chalabi A, Miller CC, Nicholson G, Shaw CE. TDP-43 Mutations in Familial and Sporadic Amyotrophic Lateral Sclerosis. Science 2008;249:1468–73. 116. Al-Chalabi A, Jones A, Troakes C, King A, Al-Sarraj S, Van Den Berg LH. The genetics and neuropathology of amyotrophic lateral sclerosis. Acta Neuropathol 2012;124:339–52. 117. Neumann M, Sampathu DM, Kwong LK, Truax AC, Micsenyi MC, Chou TT, Bruce J, Schuck T, Grossman M, Clark CM, McCluskey LF, Miller BL, Masliah E, Mackenzie IR, Feldman H, Feiden W, Kretzschmar HA, Trojanowski JQ, Lee VM-Y. Ubiquitinated TDP-43 in Frontotemporal Lobar Degeneration and Amyotrophic Lateral Sclerosis. Science 2006;314:130–3. 118. Xu Y-F, Gendron TF, Zhang Y-J, Lin W-L, D’Alton S, Sheng H, Casey MC, Tong J, Knight J, Yu X, Rademakers R, Boylan K, Hutton M, McGowan E, Dickson DW, Lewis J, Petrucelli L. Wild-Type Human TDP-43 Expression Causes TDP-43 Phosphorylation, Mitochondrial Aggregation, Motor Deficits, and Early Mortality in Transgenic Mice. J Neurosci 2010;30:10851–9. 119. Therrien M, Parker JA. Worming forward: Amyotrophic lateral sclerosis toxicity mechanisms and genetic interactions in Caenorhabditis elegans. Front Genet 2014;5:1–13. 120. Baskoylu SN,Yersak J, O’Hern P, Grosser S, Simon J, Kim S, Schuch K, Dimitriadi M,Yanagi KS, Lins J, Hart AC. Single copy/knock-in models of ALS SOD1 in C. elegans suggest loss and gain of function have different contributions to cholinergic and glutamatergic neurodegeneration. PLoS Genet 2018;14:1–28. 121. Oeda T, Shimohama S, Kitagawa N, Kohno R, Imura T, Shibasaki H, Ishii N. Oxidative stress causes abnormal accumulation of familial amyotrophic lateral sclerosis-related mutant SOD1 in transgenic Caenorhabditis elegans. Hum Mol Genet 2001;10:2013–23. 122. Wang J, Farr GW, Hall DH, Li F, Furtak K, Dreier L, Horwich AL. An ALS-linked mutant SOD1 produces a locomotor defect associated with aggregation and synaptic dysfunction when expressed in neurons of Caenorhabditis elegans. PLoS Genet 2009;5. 123. Li J, Li T, Zhang X, Tang Y, Yang J, Le W. Human superoxide dismutase 1 overexpression in motor neurons of Caenorhabditis elegans causes axon guidance defect and neurodegeneration. Neurobiol Aging 2013;35:837–46. 124. Watson MR, Lagow RD, Xu K, Zhang B, Bonini NM. A Drosophila model for amyotrophic lateral sclerosis reveals motor neuron damage by human SOD1. J Biol Chem 2008;283:24972–81. 125. Yanase S, Onodera A, Tedesco P, Johnson TE, Ishii N. SOD-1 deletions in Caenorhabditis elegans alter the localization of intracellular reactive oxygen species and show molecular compensation. J Gerontol Ser A Biol Sci Med Sci 2009;64:530–9. 126. Ash PEA, Zhang YJ, Roberts CM, Saldi T, Hutter H, Buratti E, Petrucelli L, Link CD. Neurotoxic effects of TDP-43 overexpression in C. elegans. Hum Mol Genet 2010;19:3206–18. 127. Liachko NF, Guthrie CR, Kraemer BC. Phosphorylation promotes neurotoxicity in a Caenorhabditis elegans model of TDP-43 proteinopathy. J Neurosci 2010;30:16208–19. 128. Hardiman O, Al-Chalabi A, Chio A, Corr EM, Logroscino G, Robberecht W, Shaw PJ, Simmons Z, Berg LH, van der. Amyotrophic lateral sclerosis. Prog Med Chem 2019;58:63–117. 129. Eisele YS, Monteiro C, Fearns C, Encalada SE, Wiseman RL, Powers ET, Kelly JW. Targeting protein aggregation for the treatment of degenerative diseases. Nat Rev Drug Discov 2015;14:759–80. 130. Madhivanan K, Greiner ER, Alves-Ferreira M, Soriano-Castell D, Rouzbeh N, Aguirre CA, Paulsson JF, Chapman J, Jiang X, Ooi FK, Lemos C, Dillin A, Prahlad V, Kelly JW, Encalada SE. Cellular clearance of circulating transthyretin decreases cell-nonautonomous proteotoxicity in Caenorhabditis elegans. Proc Natl Acad Sci 2018; 201801117.

Caenorhabditis elegans as a model organism for protein homeostasis diseases

131. Henze A, Homann T, Rohn I, Aschner M, Link CD, Kleuser B, Schweigert FJ, Schwerdtle T, Bornhorst J. Caenorhabditis elegans as a model system to study post-translational modifications of human transthyretin. Sci Rep 2016;6:37346. 132. Tsuda Y,Yamanaka K, Toyoshima R, Ueda M, Masuda T, Misumi Y, Ogura T, Ando Y. Development of transgenic Caenorhabditis elegans expressing human transthyretin as a model for drug screening. Sci Rep 2018;8:17884. 133. Hull RL, Westermark GT, Westermark P, Kahn SE. Islet amyloid: A critical entity in the pathogenesis of type 2 diabetes. J Clin Endocrinol Metab 2004;89:3629–43. 134. Pillay K, Govender P. Amylin uncovered: A review on the polypeptide responsible for type II diabetes. Biomed Res Int 2013;. 135. Rosas PC, Nagaraja GM, Kaur P, Panossian A, Wickman G, Garcia LR, Al-Khamis FA, Asea AAA. Hsp72 (HSPA1A) prevents human islet amyloid polypeptide aggregation and Toxicity: A new approach for type 2 diabetes treatment. PLoS One 2016;11:1–21. 136. Aldras Y, Singh S, Bode K, Bhowmick DC, Jeremic A, O’Halloran DM. An inducible model of human amylin overexpression reveals diverse transcriptional changes. Neurosci Lett 2019;704:212–9. 137. Scarponi R, Ricardi M, Albertazzi V, De Amicis S, Rastelli F, Zerbini L. Dialysis-related amyloidosis: challenges and solutions. Int J Nephrol Renovasc Dis 2016;9:319–28. 138. Relini A, De Stefano S, Torrassa S, Cavalleri O, Rolandi R, Gliozzi A, Giorgetti S, Raimondi S, Marchese L, Verga L, Rossi A, Stoppini M, Bellotti V. Heparin strongly enhances the formation of β2microglobulin amyloid fibrils in the presence of type I collagen. J Biol Chem 2008;283:4912–20. 139. Diomede L, Soria C, Romeo M, Giorgetti S, Marchese L, Mangione PP, Porcari R, Zorzoli I, Salmona M, Bellotti V, Stoppini M. C. elegans expressing human β2-microglobulin: a novel model for studying the relationship between the molecular assembly and the toxic phenotype. PLoS One 2012;7:e52314. 140. Diomede L, Cassata G, Fiordaliso F, Salio M, Ami D, Natalello A, Doglia SM, De Luigi A, Salmona M. Tetracycline and its analogues protect Caenorhabditis elegans from B amyloid-induced toxicity by targeting oligomers. Neurobiol Dis 2010;40:424–31. 141. Merlini G, Seldin DC, Gertz MA. Amyloidosis: pathogenesis and new therapeutic options. J Clin Oncol 2011;29:1924–33. 142. Diomede L, Rognoni P, Lavatelli F, Romeo M, di Fonzo A, Foray C, Fiordaliso F, Palladini G,Valentini V, Perfetti V, Salmona M, Merlini G. Investigating heart-specific toxicity of amyloidogenic immunoglobulin light chains: a lesson from C. elegans. Worm 2014;3:e965590. 143. Diomede L, Rognoni P, Lavatelli F, Romeo M, del Favero E, Cantu L, Ghibaudi E, di Fonzo A, Corbelli A, Fiordaliso F, Palladini G,Valentini V, Perfetti V, Salmona M, Merlini G. A Caenorhabditis elegans–based assay recognizes immunoglobulin light chains causing heart amyloidosis. Blood 2014;123:3543–52. 144 Nussbaum-Krammer CI, Morimoto RI. Caenorhabditis elegans as a model system for studying non-cellautonomous mechanisms in protein-misfolding diseases. Dis Model Mech 2014;7:31–9. 145 Prusiner SB. Prions. PNAS 1998;95:13363–83. 146 Nussbaum-Krammer CI, Park KW, Li L, Melki R, Morimoto RI. Spreading of a prion domain from cell-to-cell by vesicular transport in Caenorhabditis elegans. PLoS Genet 2013;9:21–3. 147. Ewald CY, Marfil V, Li C. Alzheimer-related protein APL-1 modulates lifespan through heterochronic gene regulation in Caenorhabditis elegans. Aging Cell 2016;15:1051–62. 148. Hornsten A, Lieberthal J, Fadia S, Malins R, Ha L, Xu X, Daigle I, Markowitz M, O’Connor G, Plasterk R, Li C. APL-1,a Caenorhabditis elegans protein related to the human beta-amyloid precursor protein, is essential for viability. Proc Natl Acad Sci 2016;104:1971–6. 149 Wang H, Lim PJ, Yin C, Rieckher M, Vogel BE, Monteiro MJ. Suppression of polyglutamine-induced toxicity in cell and animal models of Huntington’s disease by ubiquilin. Hum Mol Genet 2006;15:1025–41. 150 Faber PW, Voisine C, King DC, Bates EA, Hart AC. Glutamine/proline-rich PQE-1 proteins protect Caenorhabditis elegans neurons from huntingtin polyglutamine neurotoxicity. Proc Natl Acad Sci USA 2002;99:17131–6.

69

CHAPTER 4

Proteome-scale studies of protein stability Giulia Callonia,b, R. Martin Vabulasa,b

Buchmann Institute for Molecular Life Sciences, Goethe University Frankfurt, Frankfurt am Main, Germany Institute of Biophysical Chemistry, Goethe University Frankfurt, Frankfurt am Main, Germany

a

b

Outline Introduction Protein stability and unfolding Biophysical methods to measure protein stability in vitro Protein stability in vivo Biological readouts to measure protein stability Proteome-scale analysis based on aggregation in vivo Proteome-scale analysis based on degradation in vivo Structural analyses of proteins and proteomes Cross-linking-based mass spectrometry Hydroxyl radical footprinting (HRF) Limited proteolysis-based mass spectrometry Proteome-scale methods involving experimental denaturation of proteins Denaturation probed by proteolysis sensitivity Thermal denaturation probed by aggregation Denaturation probed by methionine oxidation Contributions to basic and applied biomedical research Conclusions References

72 72 73 74 75 76 77 78 79 80 80 81 82 83 83 84 85 86

Abbreviations CETSA  cellular thermal shift assay DARTS  drug affinity responsive target stability HDX  hydrogen-deuterium exchange HRF  hydroxyl radical footprinting IDR  isothermal dose-response iTRAQ  isobaric tags for relative and absolute quantitation MS  mass spectrometry NMR  nuclear magnetic resonance PN  proteostasis network SILAC  stable isotope labeling by amino acids in cell culture SPROX  stability of proteins from rates of oxidation TMT  tandem mass tags TPP  thermal proteome profiling Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00004-X Copyright © 2020 Elsevier Inc. All rights reserved.

71

72

Protein Homeostasis Diseases

UPS  ubiquitin-proteasome system XL-MS  cross-linking-based mass spectrometry

Introduction Cellular protein homeostasis, the proteostasis, is essential for proper function of any cell. A tightly connected network of proteins in charge of assuring proteome stability has evolved.1,2 The proteostasis network (PN) regulates protein synthesis, folding, transport, and degradation to maintain proteome integrity and limit the accumulation of protein aggregates.3 Significant advances in understanding of the PN role and the importance of protein stability in vivo have been made recently. Insufficient function of the PN has been linked with impaired proteome stability, disease development, and aging.4 Recent developments in the proteomics field contributed to these advances.5 From one side, the mass spectrometry (MS) instrumentation has reached the maturity level where high resolution and high accuracy measurements can be performed on a routine basis, which, in turn, allows investigation of even very complex biological systems, such as entire cellular and organismal proteomes. From the other side, data analysis strategies and algorithms have been developed to match the demands of the generated data. All of this has ensured the increasing popularity of high-throughput quantitative MS in cell biology.6 A variety of experimental approaches have been implemented to analyze protein stability at the proteome scale. This overview should help to give a first impression regarding their potential and motivate their implementation into the experimental arsenal of researchers investigating fundamental and applied aspects of proteome stability.

Protein stability and unfolding To become functional, a polypeptide has to reach its native folded state. The thermodynamic hypothesis of protein folding postulates that the three-dimensional structure of a native protein in its normal physiological milieu is the one in which its Gibbs free energy is lowest.7 The notion implies that, in a given environment, the native conformation is determined by the amino acid sequence of the protein. Although dynamic, native proteins display grossly stable structures.To a larger extent, the stability is determined by the hydrophobic effect, which results from the water molecule network around a protein. Another part of native stabilization contribute hydrogen bonds. In contrast, the unfolded polypeptide chain populates not one, but many less or more unfolded conformations with higher free energy. These conformations are unstable and make large motions. The folded state is dynamic—a protein constantly unfolds and folds again.8 The Gibbs energy difference between folded and unfolded states constitutes typically only several kilocalories per mole. In the case when the native protein is 10 kcal/mol more stable than the unfolded form, 1 out of 22 million protein molecules would be unfolded at room temperature: a solution with 1 mg of a 30 kDa protein would contain nearly a billion unfolded chains.9 The existence of many random conformations drives the

Proteome-scale studies of protein stability

unfolding by constituting a favorable entropy contribution to the Gibbs energy. For a protein of ca. 30 kDa size, the entropy effect of the polypeptide chain backbone flexibility would contribute ca. 400 kcal/mol.9 Thus, native conformation relies on a marginal stabilizing balance between large forces driving folding and unfolding (see Chapter 1). The unfolding of a small, single-domain protein is usually rapid, cooperative, and reversible. Large multidomain proteins may unfold stepwise, one domain after another. Accordingly, the folding/unfolding of larger proteins is characterized by trajectories containing intermediate states.10,11 Obviously, kinetic aspects of unfolding become critical while trying to understand and describe the unfolding of multidomain proteins. Further complication in respect to the biophysical analysis is the aggregation of unfolded species, because protein aggregation is irreversible and thus often precludes the equilibrium between folded and unfolded states, which is required for thermodynamic analysis of protein stability. Additional factors must be considered as well and include, among others, spontaneous deamination of asparagine and spontaneous oxidation of methionine and cysteine.

Biophysical methods to measure protein stability in vitro The key biophysical parameter of protein stability is the equilibrium constant, which is determined from the ratio of folded and unfolded polypeptides at standard conditions. For most proteins, the detection of the unfolded fraction is experimentally challenging because of its low concentration. To circumvent this problem, protein unfolding is usually increased artificially by adding chemical denaturants, such as urea or guanidinium chloride, or by increasing temperature. The equilibrium constant determined at the destabilizing conditions is then extrapolated to nonperturbed conditions.12 Any biochemical or biophysical technique that can detect and quantify the ratio unfolded/folded protein as a function of chemical or physical denaturant can be used to study protein folding and stability in vitro. Fluorescence, circular dichroism, or nuclear magnetic resonance (NMR) are only a couple of possible techniques to this end13,14 (see also Chapter 17 for a more detailed description of these techniques). The sensitivity and specificity of the signal remain the critical aspects in choosing an appropriate readout. New and interesting possibilities have been brought about by the recent advances in single-molecule spectroscopy techniques.15,16 The single-molecule manipulation known as optical tweezers deserves a special note.17 In this case, conformational changes in the protein are brought about by mechanical force. Using optical tweezers, new quantitative insights have been gained into the dynamics of unfolding and refolding of proteins in vitro, including that of traditionally “difficult” multidomain proteins.18

Protein stability in vivo Compared to the conditions used to analyze polypeptide folding and stability in vitro, proteins in the cell experience a different environment (see Chapter 2). Most striking is the high intracellular concentration of macromolecules. Protein concentration in the

73

74

Protein Homeostasis Diseases

cytosol can reach 300–400 g/L.19 Together with another major cellular constituent, the RNA, the high biopolymer concentration creates a very special milieu defined as macromolecular crowding. Due to the excluded volume effect, the macromolecular crowding constrains protein movements and increases local concentrations of small molecules significantly. This, in turn, can change intra- and intermolecular associations of a protein and, as consequence, affect protein stability, associations, and biochemical reactions.20 In addition to the macromolecular crowding, the biomolecular composition of the cells is highly dynamic. Cells are constantly adapting to the changing environment and stress by modulating their transcriptional and translational programs. Even under identical conditions, there are noticeable differences in respect to the RNA and protein sets among cells of the same type.21 This is in a stark contrast to the standardized solution conditions when analyzing protein folding and stability in vitro. Not only are bystander molecules constantly changing, but also individual proteins are not homogenous in chemical sense. From one site, proteins undergo evolutionary established and functionally relevant posttranslational processing and modifications. Some of them are gross and thus easily observable, such as proteolytic splitting or partial hydrolysis. Others are less noticeable and usually require specialized and targeted analyses for their identification, such as amino acid–specific phosphorylation or methylation.22 However small, the modifications can change both the thermodynamic and kinetic stability of the protein, affect the propensity to self-associate or interact with other proteins, modify its stability and turnover. Furthermore, the molecular heterogeneity might arise due to chemical damage of the amino acids following exposure to, for example, oxidizing or alkylating molecular species. The natural and environmentally caused modifications result in the heterogeneity of the protein ensemble, which, in turn, confounds the analyses of the biophysical stability of a given protein. Finally, even identical proteins can experience different pressures toward destabilization, misfolding, and aggregation in vivo due to the changed expression levels in the otherwise similar intracellular environment. It has been noticed that expression levels of human genes anticorrelate with the aggregation rates of the corresponding proteins determined in vitro.23 This anticorrelation was explained as a consequence of evolution aiming to reduce protein destabilization and aggregation. At the same time, it was suggested that proteins are expressed close to the edge of their “safety zone.” Consequently, mutations or other perturbations leading to increased levels of proteins do result in misfolding and aggregation of the respective proteins. Thus, not surprisingly, a battery of specialized proteins have evolved to support protein stability and functionality under normal and stress conditions. This machinery is composed in humans of ca. 1400 components and is called the PN.2 PN is versatile and efficient in safeguarding the proteome stability; yet, the potential of PN is not unlimited or ontogenetically constant. For example, the same aggregation-prone mutant of Aβ is kept soluble in a young individual but

Proteome-scale studies of protein stability

starts aggregating with an increasing age, when PN becomes defective or insufficient24 (see also Chapter 2). When analyzing protein stability in vivo, the aspect discussed above should be considered. Because of the complexity of the cellular environment, it is hardly possible to define the intracellular protein stability in absolute terms. Nevertheless, it is the in vivo behavior of proteins that matters in order to understand the diverse functions of a cell and cellular dysfunction during disease.

Biological readouts to measure protein stability Damaged, stress-misfolded, and otherwise destabilized proteins can undergo different fates in mammalian cells: refolding, aggregation, and degradation. These consequences are amenable for biochemical, biophysical, and microscopy analysis.25,26 The spectrum of experimental tools to access the above biological readouts is especially broad if only one particular polypeptide is to be investigated. Historically, biochemical fractionation and protein isolation based on protein-specific antibodies dominated the in vivo analysis of protein stability. More recently, techniques involving fluorescent protein fusions with the proteins of interest have become more popular.27–29 The latter approach has a number of advantages. Most importantly, microscopy allows observations to be performed in intact cells, including those that are difficult or too rare to be isolated in biochemical amounts. Secondly, a combination of fluorescent proteins with different spectral properties allows sophisticated experimental setups to address complex biological questions. Refolding of a stress-misfolded protein is the preferred outcome after perturbation subsides or the cell adapts to the changed environment. Refolded proteins can reassume their function, which precludes the loss-of-function toxicity usually following protein misfolding. Of note, some types of misfolded species become cytotoxic, which is called the gain-of-function toxicity. Obviously, this type of toxicity is prevented by successful refolding. Steady-state measurements cannot identify refolded species because they are identical to the native polypeptides. In contrast to the refolded state, there are well-established experimental possibilities to detect protein aggregates, microscopically and biochemically (see Chapters 7 and 8). Recently, there have been exciting developments in understanding cellular deposition of aggregated species in yeasts and mammalian cells.30,31 It turned out that cells have evolved an elaborated machinery to sort and sequester aggregates, and these processes seem to be heterogenous in regard of the type of aggregate. For example, the amyloid of polyglutamine-extended huntingtin fragment is sequestered in perivacuolar compartment as “insoluble protein deposit,” while more dynamic aggregates can be found in the “juxtanuclear quality compartment.”27 Finally, irreversibly misfolded proteins might undergo degradation by the cytosolic ubiquitin-proteasome system (UPS) or by vacuolar hydrolysis32,33 (see Chapters 2 and 6).

75

76

Protein Homeostasis Diseases

The former is typically associated with the tagging of target proteins with chains of the small protein ubiquitin. Polyubiquitin then mediates the recruitment of the tagged protein to the cytosolic degradation machine called proteasome. Therefore, the conjugated ubiquitin can be used analytically as a sign of (commitment for) degradation. In this case, additional controls are needed, since ubiquitin has a number of additional cellular functions in addition to its role in degradation.Vacuolar hydrolysis includes lysosomal and autophagic processes, both being highly linked.34 The most practical means of identifying these types of the misfolded protein degradation is the use of the respective inhibitors. The same is true in case of proteasomal degradation. Because of the clinical relevance of the UPS, many specific and potent proteasome inhibitors have been developed and are available for an experimenter.35

Proteome-scale analysis based on aggregation in vivo Proteins aggregate in vivo because of different reasons. Misfolding can take place because of increased conformational stress, for example, under increased temperature, because of chemical damage, such as oxidation, or because of destabilizing mutations. Refolding back to the native state in vivo is kinetically complicated due to sticky neighbor proteins and high concentrations of biomolecules in the cytosol. This is one reason why protein aggregation is such a widespread phenomenon.36 From the experimental point of view, protein aggregation is easily detectable using microscopic and biochemical techniques and has become a common readout to indicate destabilization of individual proteins or the entire proteome. Mutations increasing the aggregation propensity of neurodegeneration-causing proteins, such as Alzheimer’s Aβ peptide, Parkinson’s α-synuclein, and Huntington’s huntingtin, contributed to the current understanding of the relationship between aggregation and cytotoxicity.37 The amyloid hypothesis proposed that the intracellular and extracellular amyloid formed during aggregation of those proteins is at the basis of the neurodegeneration in the respective diseases. A pioneering study indicated a proteomedestabilizing effect of mutant huntingtin aggregates in a Caenorhabditis elegans model of Huntington’s disease.38 High-throughput quantitative MS was used to provide the general mechanistic explanation.39 The authors used artificial amyloidogenic polypeptides of different toxicity potential and uncovered a correlation between the extent of amyloidogenesis, bystander protein coaggregation, and cytotoxicity in human cell cultures. Similar results were obtained by means of quantitative proteomics also in the whole-organism models.40,41 The importance of the proteome-scale tools in analyzing protein aggregation in diverse pathologies is highlighted by the fact that a dedicated working group has been established in the Human Proteome Project.42 Another situation of widespread proteome destabilization and aggregation is aging. There are different theories trying to explain this phenomenon, including, first, the accumulation of protein damage over time and, second, the decreasing capacity of the

Proteome-scale studies of protein stability

PN.43 Regardless of the reasons, the aggregation of a significant number of proteins in old organisms and cells is well established experimentally. In contrast to the coaggregation analyses in the presence of an amyloidosis-driving key protein, there is no need of protein-specific enrichment in this case. An isolation of the insoluble cellular fraction is essentially enough to proceed with the quantitative MS. As an example, one of the earlier studies used stable isotope labeling with isobaric tags for relative and absolute quantitation (iTRAQ) to compare the insoluble fraction in young and old C. elegans worms.44 Another impactful study combined the analyses of protein aggregation with the changes of the PN during aging.45 Interestingly, reversible aggregation at a proteome scale was recently proposed as an adaptive reaction to stress.46 The authors demonstrated that during heat stress recovery, the disassembly of stress granules and restoration of translation activity was linked to the disaggregation of damaged proteins. In another proteomics study, stable isotope labeling by amino acids in cell culture (SILAC)–based quantification revealed the reversible aggregation of more than 170 proteins during heat shock in budding yeast cells.47 Here, misfolding was not necessary for aggregation of at least some of the analyzed proteins. Accordingly, the enzymatic activity could be retained in an in vitro reconstitution of the heat-induced aggregation.47 On similar lines, it was recently shown that reversible protein aggregation is a protective mechanism to ensure the cell cycle restarts after stress.48

Proteome-scale analysis based on degradation in vivo Another biological readout of the structural destabilization of a protein under given conditions is the degradation of the protein by the protein quality control machinery. However, the changed levels of a protein in the cell can arise not only from its increased degradation but from its decreased synthesis as well.The entanglement of this duality necessitates special experimental approaches. Historically, a pulse-labeling of nascent proteins with radioactive amino acids and subsequent chase in the nonradioactive medium was employed to quantify the degradation of individual proteins or the bulk proteome.49 The capacities to analyze proteome synthesis and degradation have been spectacularly expanded by the development of stable isotope–based quantitative proteomics.50,51 SILAC has emerged as a simple and powerful approach where one or several essential amino acids in the cellular or organismal “light” proteome are metabolically exchanged with those enriched in the stable isotopes 2H, 13C, and 15N to generate “heavy” proteomes.52 High-resolution mass spectrometers can resolve differentially labeled isotope clusters, thus allowing determination of the relative abundances of the respective peptides in light and heavy conditions. In addition to metabolic labeling, isotopic labels can be introduced chemically during sample processing.53 Here, the broadly used approaches are iTRAQ and tandem mass tags (TMT). An early implementation of the proteome labeling with stable isotopes to globally assess the rates of protein synthesis and degradation succeeded in determining the turn-

77

78

Protein Homeostasis Diseases

over of almost 600 proteins in a human cell line.54 Current setups and analytic procedures are more complex but reach several times larger coverage. For example, the turnover of more than 9600 proteins from five nondividing mammalian cells was determined, spanning the range from several hours to several months.55 Often, multitagging strategies combining metabolic and chemical labels are used.56 A recent example is the technique called “multiplexed proteome dynamic profiling,” which combines pulse SILAC with TMT labeling to compare the nascent and preexistent proteome stabilities under different conditions.57 Finally, labeling with amino acid analogs, such as azidohomoalanine (AHA), designed to ensure the enrichment of labeled protein by biochemical procedures, was established.58 One recent study combined SILAC with the AHA pulse to conclude that around 10% of mammalian proteins are degraded faster immediately after synthesis as compared to later time points.59 MS comparison of steady-state levels of proteins in cells with active and inhibited degradation systems is a simple and robust alternative to the experimental schemes described earlier. Especially for the proteasome, there is a number of well-established specific and potent inhibitors.35 For example, the application of MG-132 allowed researchers to conclude that riboflavin depletion from the cell culture medium for as short as 1 day destabilizes a significant fraction of the flavoproteins in a melanoma cell line, yet to a different extent.60 If combined with protein synthesis inhibitors, this approach has potential to access the dynamic equilibrium between proteome biogenesis and degradation. Importantly, the experimental setups using the inhibitors of proteasomal degradation should comply with the concerns of acute stress responses possibly induced in cells upon their application.61 A completely different approach is the MS-based identification of ubiquitin conjugation to a protein preceding its degradation by proteasome.62 During processing for MS measurement, proteins are digested with trypsin, which cuts the polypeptide sequence after lysine and arginine. The C terminus of ubiquitin, the side of ubiquitin attached to targets, displays the sequence Arg-Gly-Gly. This is why ubiquitinated peptides remain with the diglycine adduct upon processing with trypsin, which later can be detected in their tandem MS (MS/MS) spectra. At least two issues are associated with this method if used to identify destabilized and degradation-committed proteins. First, the trypsin digestion of neural-precursor-cell-expressed developmentally down-regulated (Nedd8) and interferon-stimulated gene (ISG15) also generates a glycine dipeptide on the lysines.62 Second, ubiquitination is known to have many other functions in addition to being a proteasomal degradation tag.63,64

Structural analyses of proteins and proteomes Although not directly, structural data do have the potential of explaining the changes of protein stability in different circumstances. For example, the detection of acetylation of a lysine within a ubiquitylation motif would suggest the impairment of the degradation of

Proteome-scale studies of protein stability

the protein by the UPS; the exposure of a hydrophobic patch would mean the increased propensity of the protein to aggregate; the detection of less interactors in a multimolecular complex and simultaneous increase of associations with molecular chaperones would suggest its destabilization. In combination with biochemical and computational methods, structural analyses provide molecular and atomistic understanding of the structure–function and structure–stability relationships. Standard approaches to analyze protein structure are X-ray crystallography, NMR spectroscopy, and electron microscopy. All of these biophysical methods are work intense and therefore low throughput in regard of results, which makes it difficult to implement them for proteome-scale studies. Since the conventional NMR spectroscopy measures protein structures in solution, it directly addresses the dynamics of a protein and thus is closest to serve the purpose of understanding the protein’s stability. Often, NMR spectroscopy and X-ray crystallography complement each other if applied to the analysis of the same protein and its variants.65 MS offers additional possibilities to investigate protein structures. Two methods, hydrogen-deuterium exchange (HDX) MS and the native MS, are low throughput and thus less suitable to be used for proteome-scale studies, at least currently. At the same time, three MS-based approaches can be implemented in a high-throughput manner to analyze protein structure and constitute the experimental basis for structural proteomics. These methods are the measurement of chemically cross-linked samples (XL-MS), the mass spectrometric hydroxyl radical footprinting (HRF), and the limited proteolysisbased MS.

Cross-linking-based mass spectrometry XL-MS uses chemical agents to covalently link neighboring residues in a polypeptide or between polypeptides in a complex.66,67 From the chemical point of view, the cross-linkers can be homobifunctional and heterobifunctional. The former have the same reactive groups; thus, they link the same functional groups in proteins. The latter have different reactive groups and therefore link different functional groups. There is also a group of trifunctional cross-linkers.This additional group can be designed to facilitate the affinity purification of cross-linked species.The length of the linker between the reactive groups determines the maximum distance between cross-linked groups in a protein and thus provides the key spacial information.The proteins are digested, and cross-linked residues are identified by MS. This information is used then as constrains in modeling the structure of the polypeptide or the complex under investigation. XL-MS was instrumental to elucidate structures of a number of large and dynamic multimolecular complexes.68–70 To identify cross-links, all possibilities have to be included in the search database for bottom-up analysis. This usually results in a vast increase of the search space, which requires an appropriate computational power possibly becoming a bottleneck for proteome-scale in vivo applications of the XL-MS. One possibility to tackle this issue are

79

80

Protein Homeostasis Diseases

MS-cleavable cross-linkers with a preferential cleavage site, which produces signature fragment ions upon MS/MS fragmentation. The combination of the MS-cleavable cross-linkers, different types of fragmentation of each MS precursor, and the cross-linking-tailored search algorithms ensure efficient proteome-wide applications of the XLMS. Using this approach, 326 protein-protein interactions in vivo were identified.71 In a more recent example, 3322 unique residue-to-residue contacts involving half of the mitochondrial proteome were detected by XL-MS.72

Hydroxyl radical footprinting (HRF) For HRF, proteins are covalently labeled by the hydroxyl radical. The labeling is then mapped using MS, which allows identification of the solvent-accessible surface of the proteins.73,74 This way, conformational changes due to ligand binding or protein destabilization upon stress can be identified. HRF is similar to another widely used MS-based footprinting method, the HDX. In contrast to the latter, HRF targets accessible side chains of the polypeptide, not the accessible backbone, and creates nonreversible modifications.74 Because of the label irreversibility, a broader range of processing conditions can be tested for optimization purposes. A hydroxyl radical is generated using Fenton’s reagent by the radiolysis of water or photolysis of hydrogen peroxide. The small size of the radical allows obtaining high-resolution mapping. From the other side, the reaction might cause protein unfolding, leading to artifactual conclusions regarding protein structure changes. Secondly, radical-modified side chains might affect protease recognition and cleavage during sample preparation for the MS measurements. Currently, HRF of nucleic acids in native cellular environment is better established than protein footprinting.73 Fenton chemistry and laser flash photolysis have been successfully used to label proteins in vivo with hydroxyl radical.75,76 To increase the throughput of HRF for intracellular studies, several obstacles remain to be overcome. One problem relates to the biochemical isolation of protein complexes before MS. Hydroxyl radical modification of amino acids can change the reactivity of the sequences to the affinity reagents, such as antibodies. To this end, multidimensional chromatography might offer an alternative for protein enrichment during HRF. There is also a need for specifically tailored data analysis tools to efficiently extract the information after HRF of complex samples.

Limited proteolysis-based mass spectrometry Here, cell lysates from different biological conditions are treated in vitro with an unspecific protease, such as proteinase K. Limitation of hydrolysis is achieved by short incubation time or by the addition of an irreversible protease inhibitor.The differential cleavage pattern is then identified by means of MS. The differences are finally related to the biological conditions, for example, the availability versus lack of an enzyme cofactor in the cell culture medium.77 The key difference between the conventional and limited prote-

Proteome-scale studies of protein stability

olysis is that the latter aims to keep the sample in its native conformation. Because of the folded domain architecture of the majority of protein in any proteome, cleavage usually happens in interdomain linkers or in loops between secondary structure elements of the domains. Due to their high proteolytic susceptibility, intrinsically disordered domains and proteins78 represent a more challenging target to analyze by the limited proteolysis. An early implementation of the protease sensitivity-based MS analysis of protein stability was the drug affinity responsive target stability (DARTS) method.79 As its name suggests, DARTS was motivated by the need to identify drug targets in complex cellular lysates. Since protein stability usually increases upon association with small-molecule ligands, the method relies on the stabilization of the ligand-bound proteins against hydrolysis.80 One recent example is the proteome-wide study to identify new targets for the immunosuppressive drug FK506.81 DARTS is a protein-centric method, which means that the changes of peptide pattern after digestion is not considered. In addition to the protein-centric DARTS, the peptide-centric analysis of limited proteolysis was successfully implemented as well.82 To ensure the reproducibility of hydrolysis patterns, the authors exploited the sensitivity and background subtraction potential of selected reaction monitoring, which allowed them to assess structural features of more than 1000 yeast proteins depending on the nutrient availability. The same approach was used to probe structural changes in response to l-arginine in human primary T cells.83 The resolution achieved by the method is not very high, being in the range of ca. 10 amino acids.77 However, since the thermodynamic protein stability is known to depend on the association with their ligands, the method did enable identification of yeast proteins binding fructose-1,6-bisphosphate and human proteins binding l-arginine.82,83 In this respect, large-scale identification of protein-small molecule interactions as exemplified by the abovementioned studies highlights the capacities of limited proteolysis-based MS to measure proteome stability.

Proteome-scale methods involving experimental denaturation of proteins Several MS-based methods have been developed for the direct measurements of cellular protein stability in a high-throughput manner. The key approach behind these methods is the exposure of proteomes to an increasing conformational stress and then following the unfolding of proteins by their increased sensitivity to unspecific proteolysis, the enhanced aggregation, or the higher availability of hydroxyl-reactive methionine residues. After probing, standard bottom-up proteomics is performed to quantify the respective fractions of proteins: intact/hydrolyzed, soluble/aggregated, or native/modified, accordingly. Quantification strategies can be either label free or involve labeling with stable isotopes. The majority of these methods are protein-centric and thus face the current challenges of the conventional bottom-up proteomics, such as the sensitivity and, especially, speed of the MS measurements to reach sufficient sequence coverage.

81

82

Protein Homeostasis Diseases

What is essentially different, though, is the structure of the quantitative data extracted from the MS-based proteome stability measurements. In a conventional proteomics experiment, the expected changes of protein concentration or protein modification in different biological conditions are fully open. In contrast, during the MS-based proteome stability measurements, the gradual increase of conformation stress results in incremental accumulation of unfolded species, typically in a sigmoidal dose-response relationship. The expected structure of the data facilitates the identification of the outliers and other experimental artifacts. An additional possibility to increase the reliability of the quantification is given by the possibility to normalize individual data points by globally measured values, for example, by the amounts of nonmodified peptides.84 The sigmoidal denaturation curves reflect the cooperative folding/misfolding behavior of polypeptides and allow determination of transition midpoints. The transition midpoints are usually used to compare the stability of proteins under different biological conditions, for example, the stability of an enzyme in the presence and absence of a cofactor. Based on the MS data, biophysical characterization of protein folding and ligand binding can be achieved in cases when the experimental setups ensure the reversibility of the unfolding reaction, for example, when using chemical denaturants.84–86

Denaturation probed by proteolysis sensitivity For this type of proteome stability determination, increasing the concentration of chemical denaturant, such as urea, is one possibility to unfold proteins in the cellular lysate. A nonspecific protease, such as thermolysin, is added then to hydrolyze the unfolded fractions of proteins, and the remaining folded fractions are quantified by MS. In these experiments, it is critical to separate the hydrolyzed products form intact polypeptides before MS processing with trypsin because intact proteins are quantified by bottom-up approaches. Initially, two-dimensional gel electrophoresis was used to separate hydrolyzed from nonhydrolyzed proteins before quantification.87 Later studies took advantage of metabolic and chemical labeling to facilitate quantification.88,89 These studies are often motivated by the aim to identify small-molecule binders in the cell. For example, thermolysin proteolysis of bacterial lysate in the presence of the nicotinamide adenine dinucleotide (NAD) identified 78 NAD-binding proteins in the Escherichia coli proteome.89 An impressive study analyzing thermal protein stability in vivo was published recently.90 The authors used 14 temperature points to increasingly unfold proteins in intact cells. The lysates then were treated with unspecific proteinase K to hydrolyze the unfolded fractions, and the remaining intact proteins were quantified using label-free MS. From the sigmoidal unfolding curves, the melting temperature was determined for more than 8000 proteins from E. coli bacteria, Saccharomyces cerevisiae yeasts, archaea Thermus thermophilus, and human cells. Since it is not sure whether thermodynamic equilibrium between folded and unfolded forms can be reached during heating in vivo, the inflection points of curves represent apparent melting temperatures of the respective proteins.90

Proteome-scale studies of protein stability

Notwithstanding this biophysical shortcoming, the bioinformatics analysis of the generated data immensely enriched our understanding of proteome buildup and exemplified the potential of proteome-scale analysis of protein stability.

Thermal denaturation probed by aggregation Similar to the above study by Leuenberger et al., the exposure of intact cells to the increasing temperature is used to generate intracellular protein melting curves in the cellular thermal shift assay (CETSA).91 However, CETSA does not involve unspecific protease to hydrolyze unfolded proteins but instead uses the biochemical separation of soluble from aggregated fractions to quantify protein melting. CETSA is based on the assumption that heated proteins in vivo unfold in a manner similar to the unfolding in vitro and that the unfolding is followed by rapid aggregation.92 Soon after its invention, CETSA was combined with high-throughput MS, evolving into a powerful tool to investigate proteome-wide stability of proteins, especially in regard to their stability changes induced by drugs and other small molecules.93,94 To increase the robustness and reliability of the generated data, several modifications have been introduced into the original protocol. For example, in a modified version heating was performed only at one temperature but in the presence of different amounts of a ligand, which was named isothermal dose-response CETSA (IDR-CETSA).95 IDR-CETSA results in reduced rates of false hits. Another group combined different temperatures with different concentrations of ligands, calling their method two-dimensional thermal proteome profiling (2D-TPP).96 One technical reason of improved reliability of these second-generation assays is the fact that the same TMT label set was used for quantification of proteomes within one temperature value.92 As mentioned before, thermal denaturation of proteins in vivo is not readily reversible, so the thermally unfolded proteins are not necessarily at equilibrium, which complicates the biophysical evaluation of ligand-binding affinities using cellular thermal shifts. Nevertheless, the CETSA/TPP methods allow generation of a wealth of powerful data helping to identify new drug targets97 and to uncover complex relationships between different cellular states and their proteome stability.98,99

Denaturation probed by methionine oxidation Stability of proteins from rates of oxidation (SPROX) utilizes the hydrogen peroxide– mediated oxidation of methionine residues to report protein unfolding.100 Methionine is a rare amino acid in protein sequences and is often buried in the native folded state of a polypeptide. SPROX uses chemical denaturation to shift the equilibrium between folded and unfolded fractions of proteins. Then hydrogen peroxide is supplied at a concentration and, for an incubation period, adjusted such that the exposed methionines are oxidized to sulfoxides. The protein oxidation reaction is stopped by adding excess methionine or catalase, and the oxidized peptides are mapped using MS.84,101 SPROX is in

83

84

Protein Homeostasis Diseases

its principle similar to hydroxyl radical fingerprinting; however, the latter requires much harsher reaction conditions. Although thermal denaturation can be used for SPROX as well,102 it seems that chemical denaturation is better suited for the analyses of complex protein mixtures. The rarity of methionines in proteins might cause potential coverage problems when applying SPROX to analyze entire proteomes. Nevertheless, this method was shown to be capable of proteome-scale studies of protein stability to discover small-molecule binders. For example, it was used to identify 10 protein targets of cyclosporin A in yeast.86 More recently, SPROX was used to identify 139 proteins in yeast lysate that can bind to the nonhydrolyzable adenosine triphosphate (ATP) analog adenylyl-imidodiphosphate (AMP-PNP).103 The quantification in the first study relied on metabolic labeling by means of SILAC, while the second study used chemical labeling with isobaric mass tags. To increase the coverage, methionine-containing peptides can be enriched before MS analysis.104 The potential of SPROX was impressively demonstrated in a recent study that quantified some 10,000 unique regions within ca. 3000 human proteins.105

Contributions to basic and applied biomedical research Together with the progress in genome biology, the technical developments of MS instrumentation and the advances in algorithms of MS data analysis have brought about new possibilities for deeper understanding of the human proteome complexity and function.5 Structural proteomics have emerged as a discrete and dynamic field of the high-throughput quantitative proteomics. Already, it fruitfully complements and extends classical structural biology approaches. For example, hydroxyl radical footprinting and limited proteolysis probing provides insights into the dynamic aspects of protein structure directly related to protein function. XL-MS is becoming an indispensable component while refining the large seminative cellular structures by means of cryoelectron microscopy.106 The high-throughput XL-MS allows characterization of even the highly complex connectivity between proteins in the cell or its different compartments.68–70 The MS-based determination of protein stability in vivo and ex vivo is an important application of the structural proteomics in biomedical research. The protein-centric view implies the importance of protein stability and turnover in all essential processes in the cell. Consequently, proteome stability has recently emerged as a novel framework to understand biological states, for example, young versus old, normal versus oncogenically transformed, and the pathogenesis of a number of diseases, for example, the neurodegeneration-linked pathologies. To this end, proteome-wide studies using indirect, biological readouts of protein stability, such as protein aggregation or degradation, are being increasingly performed.40,45 On a more quantitative side, the direct estimation of proteome stability has become possible as well. Best established currently are the approaches involving the chemical labeling of exposed amino84,107 and the pulse proteolysis

Proteome-scale studies of protein stability

of detergent-unfolded polypeptides.88 Both approaches were used to compare protein conformation and stability in a panel of breast cancer cell lines.66,88,108 Interestingly, there was little overlap between the sets of conformationally altered and thermodynamically destabilized proteins. In another study, SPROX was used to investigate age-related differences in the brain protein stability in mouse.109 Stability profiles were generated for 809 proteins.Their analysis supported the notion of the proteome destabilization during aging. The goal of therapeutic target identification have fueled the development of MSbased proteome-scale analyses of protein stability.79,86,94 It is safe to assume that this will remain the core application area of proteome stability measurements also in the future. Importantly, the variety of experimental approaches described in this chapter provides a possibility to compare the results obtained by different methods. This should be considered as a critical direction in the field on the way to its maturation. The standardization and extensive benchmarking of individual methods will facilitate the abovementioned objective. Faced with the urging need of fast identification of new and optimization of available therapeutics, the MS-based in vivo and ex vivo analyses of drug-induced changes of protein stability bear an exciting potential to develop to an essential toolbox in basic and translation biomedical research.

Conclusions The protein-centric view in modern cell biochemistry posits the importance of proteome stability to ensure proper cellular structure and function. Historically, the analyses of thermodynamic and kinetic stability of purified polypeptides in vitro contributed essentially to the current understanding of the structure–function relationship of biochemical processes. The advent of genomic biology combined with the progress in the MS instrumentation and data analysis algorithms have substantiated the ambition to understand the stability of proteins in their cellular entirety and complexity, that is, the stability of the cellular proteome. This goal is supported by the notion that the disturbances of the proteome stability is at the core of clinically relevant abnormal cellular states, such as tumorigenic transformation, neurodegeneration, metabolic diseases, and aging (see section, Biophysical methods to measure protein stability in vitro, for detailed description of representative examples). The current technical possibilities in the field of quantitative MS of proteins—as implicated in its bottom-up approaches—allow sensitive, accurate, and high-throughput measurements to cover significant parts of the human proteome, even from moderately sized samples. This has allowed studying the proteome-wide aggregation and degradation of proteins under different stress and pathologic conditions. In addition to the biological readouts of protein destabilization, a number of methods have been developed to test conformational changes of proteins on the proteome scale, such as measurement

85

86

Protein Homeostasis Diseases

of chemically cross-linked samples, the mass spectrometric footprinting, and the limited proteolysis-based MS. Finally, direct analysis of protein stability and biophysical characterization of protein folding and ligand binding can be achieved on the proteome scale in cases when experimental setups ensure the reversibility of the unfolding reaction.The respective methods involve the use of chemical denaturants and probe the unfolding of proteins by pulse proteolysis or exposed methionine oxidation. The proteome-scale studies of protein stability have significant novelty potential in basic and applied biomedical research. On one side, they provide experimental evidence to explore the involvement of the proteome stability in different physiological and pathological states. On the other side, they allow to identify and characterize protein targets of drugs and drug candidate molecules in the relevant in vivo environment in a high-throughput manner.These important areas of application have ensured the increasing interest and significant advances in the MS-based analyses of proteome stability in recent years.

Acknowledgments We thank the ERC (StG-311522) and DFG (EXC115) for funding of our work.

References 1. Balch WE, Morimoto RI, Dillin A, Kelly JW. Adapting proteostasis for disease intervention. Science 2008;319(5865):916–9. 2. Balchin D, Hayer-Hartl M, Hartl FU. In vivo aspects of protein folding and quality control. Science 2016;353(6294):aac4354. 3. Sala AJ, Bott LC, Morimoto RI. Shaping proteostasis at the cellular, tissue, and organismal level. J Cell Biol 2017;216(5):1231–41. 4. Klaips CL, Jayaraj GG, Hartl FU. Pathways of cellular proteostasis in aging and disease. J Cell Biol 2018;217(1):51–63. 5. Aebersold R, Mann M. Mass-spectrometric exploration of proteome structure and function. Nature 2016;537(7620):347–55. 6. Walther TC, Mann M. Mass spectrometry-based proteomics in cell biology. J Cell Biol 2010;190(4):491– 500. 7. Anfinsen CB. Principles that govern the folding of protein chains. Science 1973;181(4096):223–30. 8. Sekhar A, Kay LE. An NMR view of protein dynamics in health and disease. Annu Rev Biophys 2019;48:297–319. 9. Kazlauskas R. Engineering more stable proteins. Chem Soc Rev 2018;47(24):9026–45. 10. Fersht AR, Daggett V. Protein folding and unfolding at atomic resolution. Cell 2002;108(4):573–82. 11. Finkelstein AV, Badretdin AJ, Galzitskaya OV, Ivankov DN, Bogatyreva NS, Garbuzynskiy SO. There and back again: two views on the protein folding puzzle. Phys Life Rev 2017;21:56–71. 12. Shaw KL, Scholtz JM, Pace CN, Grimsley GR. Determining the conformational stability of a protein using urea denaturation curves. Methods Mol Biol 2009;490:41–55. 13. Bartlett AI, Radford SE. An expanding arsenal of experimental methods yields an explosion of insights into protein folding mechanisms. Nat Struct Mol Biol 2009;16(6):582–8. 14. Dobson CM. Experimental investigation of protein folding and misfolding. Methods 2004;34(1):4–14.

Proteome-scale studies of protein stability

15. Hofmann H. Single-molecule spectroscopy of unfolded proteins and chaperonin action. Biol Chem 2014;395(7–8):689–98. 16. Schuler B, Soranno A, Hofmann H, Nettels D. Single-molecule FRET spectroscopy and the polymer physics of unfolded and intrinsically disordered proteins. Annu Rev Biophys 2016;45:207–31. 17. Ritchie DB, Woodside MT. Probing the structural dynamics of proteins and nucleic acids with optical tweezers. Curr Opin Struct Biol 2015;34:43–51. 18. Bustamante CJ, Kaiser CM, Maillard RA, Goldman DH, Wilson CAM. Mechanisms of cellular proteostasis: insights from single-molecule approaches. Annu Rev Biophys 2014;43:119–40. 19. Zimmerman SB, Trach SO. Estimation of macromolecule concentrations and excluded volume effects for the cytoplasm of Escherichia coli. J Mol Biol 1991;222(3):599–620. 20. Minton AP. How can biochemical reactions within cells differ from those in test tubes? J Cell Sci 2006;119(Pt 14):2863–9. 21. Barteneva NS,Vorobjev IA. Heterogeneity of metazoan cells and beyond: to integrative analysis of cellular populations at single-cell level. Methods Mol Biol 2018;1745:3–23. 22. Qamar S, Wang G, Randle SJ, et al. FUS phase separation is modulated by a molecular chaperone and methylation of arginine cation-π interactions. Cell 2018;173(3):720–34 e15. 23. Tartaglia GG, Pechmann S, Dobson CM,Vendruscolo M. Life on the edge: a link between gene expression levels and aggregation rates of human proteins. Trends Biochem Sci 2007;32(5):204–6. 24. Cohen E, Bieschke J, Perciavalle RM, Kelly JW, Dillin A. Opposing activities protect against age-onset proteotoxicity. Science 2006;313(5793):1604–10. 25. Eisenberg DS, Sawaya MR. Structural studies of amyloid proteins at the molecular level. Annu Rev Biochem 2017;86:69–95. 26. Iadanza MG, Jackson MP, Hewitt EW, Ranson NA, Radford SE. A new era for understanding amyloid structures and disease. Nat Rev Mol Cell Biol 2018;19(12):755–73. 27. Kaganovich D, Kopito R, Frydman J. Misfolded proteins partition between two distinct quality control compartments. Nature 2008;454(7208):1088–95. 28. Waldo GS, Standish BM, Berendzen J,Terwilliger TC. Rapid protein-folding assay using green fluorescent protein. Nat Biotechnol 1999;17(7):691–5. 29. Wurth C, Guimard NK, Hecht MH. Mutations that reduce aggregation of the Alzheimer’s Abeta42 peptide: an unbiased search for the sequence determinants of Abeta amyloidogenesis. J Mol Biol 2002;319(5):1279–90. 30. Miller SBM, Mogk A, Bukau B. Spatially organized aggregation of misfolded proteins as cellular stress defense strategy. J Mol Biol 2015;427(7):1564–74. 31. Sontag EM, Samant RS, Frydman J. Mechanisms and functions of spatial protein quality control. Annu Rev Biochem 2017;86:97–122. 32. Ciechanover A, Kwon YT. Degradation of misfolded proteins in neurodegenerative diseases: therapeutic targets and strategies. Exp Mol Med 2015;47:e147. 33. Dikic I. Proteasomal and autophagic degradation systems. Annu Rev Biochem 2017;86:193–224. 34. Galluzzi L, Baehrecke EH, Ballabio A, et al. Molecular definitions of autophagy and related processes. EMBO J 2017;36(13):1811–36. 35. Huber EM, Groll M. Inhibitors for the immuno- and constitutive proteasome: current and future trends in drug development. Angew Chem Int Ed Engl 2012;51(35):8708–20. 36. Chiti F, Dobson CM. Protein misfolding, amyloid formation, and human disease: a summary of progress over the last decade. Annu Rev Biochem 2017;86:27–68. 37. Dobson CM. The amyloid phenomenon and its links with human disease. Cold Spring Harb Perspect Biol 2017;9(6). 38. Gidalevitz T, Ben-Zvi A, Ho KH, Brignull HR, Morimoto RI. Progressive disruption of cellular protein folding in models of polyglutamine diseases. Science 2006;311(5766):1471–4. 39. Olzscha H, Schermann SM,Woerner AC, et al. Amyloid-like aggregates sequester numerous metastable proteins with essential cellular functions. Cell 2011;144(1):67–78. 40. Hosp F, Gutiérrez-Ángel S, Schaefer MH, et al. Spatiotemporal proteomic profiling of Huntington’s disease inclusions reveals widespread loss of protein function. Cell Rep 2017;21(8):2291–303. 41. Kim YE, Hosp F, Frottin F, et al. Soluble oligomers of polyq-expanded huntingtin target a multiplicity of key cellular factors. Mol Cell 2016;63(6):951–64.

87

88

Protein Homeostasis Diseases

42. Boersema PJ, Melnik A, Hazenberg BPC, et al. Biology/disease-driven initiative on protein-aggregation diseases of the human proteome project: goals and progress to date. J Proteome Res 2018;17(12):4072– 84. 43. Taylor RC, Dillin A. Aging as an event of proteostasis collapse. Cold Spring Harb Perspect Biol 2011;3(5). 44. David DC, Ollikainen N, Trinidad JC, Cary MP, Burlingame AL, Kenyon C. Widespread protein aggregation as an inherent part of aging in C. elegans. PLoS Biol 2010;8(8):e1000450. 45. Walther DM, Kasturi P, Zheng M, et al.Widespread proteome remodeling and aggregation in aging C. elegans. Cell 2015;161(4):919–32. 46. Cherkasov V, Hofmann S, Druffel-Augustin S, et al. Coordination of translational control and protein homeostasis during severe heat stress. Curr Biol 2013;23(24):2452–62. 47. Wallace EWJ, Kear-Scott JL, Pilipenko EV, et al. Reversible, specific, active aggregates of endogenous proteins assemble upon heat stress. Cell 2015;162(6):1286–98. 48. Saad S, Cereghetti G, Feng Y, Picotti P, Peter M, Dechant R. Reversible protein aggregation is a protective mechanism to ensure cell cycle restart after stress. Nat Cell Biol 2017;19(10):1202–13. 49. Vabulas RM, Hartl FU. Protein synthesis upon acute nutrient restriction relies on proteasome function. Science 2005;310(5756):1960–3. 50. Cox J, Mann M. Quantitative, high-resolution proteomics for data-driven systems biology. Annu Rev Biochem 2011;80:273–99. 51. Pappireddi N, Martin L, Wühr M. A review on quantitative multiplexed proteomics. Chembiochem 2019;20(10):1210–24. 52. Mann M. Functional and quantitative proteomics using SILAC. Nat Rev Mol Cell Biol 2006;7(12):952– 8. 53. Käll L, Vitek O. Computational mass spectrometry-based proteomics. PLoS Comput Biol 2011;7(12):e1002277. 54. Doherty MK, Hammond DE, Clague MJ, Gaskell SJ, Beynon RJ. Turnover of the human proteome: determination of protein intracellular stability by dynamic SILAC. J Proteome Res 2009;8(1):104–12. 55. Mathieson T, Franken H, Kosinski J, et al. Systematic analysis of protein turnover in primary cells. Nat Commun 2018;9(1):689. 56. Jayapal KP, Sui S, Philp RJ, et al. Multitagging proteomic strategy to estimate protein turnover rates in dynamic systems. J Proteome Res 2010;9(5):2087–97. 57. Savitski MM, Zinn N, Faelth-Savitski M, et al. Multiplexed proteome dynamics profiling reveals mechanisms controlling protein homeostasis. Cell 2018;173(1):260–74 e25. 58. Dieterich DC, Link AJ, Graumann J, Tirrell DA, Schuman EM. Selective identification of newly synthesized proteins in mammalian cells using bioorthogonal noncanonical amino acid tagging (BONCAT). Proc Natl Acad Sci USA 2006;103(25):9482–7. 59. McShane E, Sin C, Zauber H, et al. Kinetic analysis of protein stability reveals age-dependent degradation. Cell 2016;167(3):803–15 e21. 60. Martínez-Limón A, Alriquet M, Lang W-H, Calloni G, Wittig I,Vabulas RM. Recognition of enzymes lacking bound cofactor by protein quality control. Proc Natl Acad Sci USA 2016;113(43):12156–61. 61. Mazroui R, Di Marco S, Kaufman RJ, Gallouzi I-E. Inhibition of the ubiquitin-proteasome system induces stress granule formation. Mol Biol Cell 2007;18(7):2603–18. 62. Beaudette P, Popp O, Dittmar G. Proteomic techniques to probe the ubiquitin landscape. Proteomics 2016;16(2):273–87. 63. Wagner SA, Beli P, Weinert BT, et al. A proteome-wide, quantitative survey of in vivo ubiquitylation sites reveals widespread regulatory roles. Mol Cell Proteomics 2011;10(10) M111.013284. 64. Yau R, Rape M. The increasing complexity of the ubiquitin code. Nat Cell Biol 2016;18(6):579–86. 65. Lienhart W-D, Gudipati V, Uhl MK, et al. Collapse of the native structure caused by a single amino acid exchange in human NAD(P)H:quinone oxidoreductase (1). FEBS J 2014;281(20):4691–704. 66. Liu F, Meng H, Fitzgerald MC. Large-scale analysis of breast cancer-related conformational changes in proteins using SILAC-SPROX. J Proteome Res 2017;16(9):3277–86. 67. Sinz A. Cross-linking/mass spectrometry for studying protein structures and protein-protein interactions: where are we now and where should we go from here? Angew Chem Int Ed Engl 2018;57(22):6390–6.

Proteome-scale studies of protein stability

68. Bui KH, von Appen A, DiGuilio AL, et al. Integrated structural analysis of the human nuclear pore complex scaffold. Cell 2013;155(6):1233–43. 69. Wang X, Cimermancic P,Yu C, et al. Molecular details underlying dynamic structures and regulation of the human 26S proteasome. Mol Cell Proteomics 2017;16(5):840–54. 70. Xu Y, Bernecky C, Lee C-T, et al. Architecture of the RNA polymerase II-Paf1C-TFIIS transcription elongation complex. Nat Commun 2017;8:15741. 71. Liu F, Rijkers DTS, Post H, Heck AJR. Proteome-wide profiling of protein assemblies by cross-linking mass spectrometry. Nat Methods 2015;12(12):1179–84. 72. Liu F, Lössl P, Rabbitts BM, Balaban RS, Heck AJR. The interactome of intact mitochondria by cross-linking mass spectrometry provides evidence for coexisting respiratory supercomplexes. Mol Cell Proteomics 2018;17(2):216–32. 73. Chea EE, Jones LM. Analyzing the structure of macromolecules in their native cellular environment using hydroxyl radical footprinting. Analyst 2018;143(4):798–807. 74. Wang L, Chance MR. Protein footprinting comes of age: mass spectrometry for biophysical structure assessment. Mol Cell Proteomics 2017;16(5):706–16. 75. Zhu Y, Serra A, Guo T, Park JE, Zhong Q, Sze SK. Application of nanosecond laser photolysis protein footprinting to study EGFR activation by EGF in cells. J Proteome Res 2017;16(6):2282–93. 76. Zhu Y, Guo T, Park JE, et al. Elucidating in vivo structural dynamics in integral membrane protein by hydroxyl radical footprinting. Mol Cell Proteomics 2009;8(8):1999–2010. 77. Schopper S, Kahraman A, Leuenberger P, et al. Measuring protein structural changes on a proteomewide scale using limited proteolysis-coupled mass spectrometry. Nat Protoc 2017;12(11):2391–410. 78. Pauwels K, Lebrun P, Tompa P. To be disordered or not to be disordered: is that still a question for proteins in the cell? Cell Mol Life Sci 2017;74(17):3185–204. 79. Lomenick B, Hao R, Jonai N, et al. Target identification using drug affinity responsive target stability (DARTS). Proc Natl Acad Sci USA 2009;106(51):21984–9. 80. Lomenick B, Olsen RW, Huang J. Identification of direct protein targets of small molecules. ACS Chem Biol 2011;6(1):34–46. 81. Kim D, Hwang H-Y, Kim JY, et al. FK506, an immunosuppressive drug, induces autophagy by binding to the V-ATPase catalytic subunit a in neuronal cells. J Proteome Res 2017;16(1):55–64. 82. Feng Y, De Franceschi G, Kahraman A, et al. Global analysis of protein structural changes in complex proteomes. Nat Biotechnol 2014;32(10):1036–44. 83. Geiger R, Rieckmann JC, Wolf T, et al. L-arginine modulates T cell metabolism and enhances survival and anti-tumor activity. Cell 2016;167(3):829–42 e13. 84. Strickland EC, Geer MA, Tran DT, et al. Thermodynamic analysis of protein-ligand binding interactions in complex biological mixtures using the stability of proteins from rates of oxidation. Nat Protoc 2013;8(1):148–61. 85. Park C, Marqusee S. Pulse proteolysis: a simple method for quantitative determination of protein stability and ligand binding. Nat Methods 2005;2(3):207–12. 86. West GM, Tucker CL, Xu T, et al. Quantitative proteomics approach for identifying protein-drug interactions in complex mixtures using protein stability measurements. Proc Natl Acad Sci USA 2010;107(20):9078–82. 87. Liu P-F, Kihara D, Park C. Energetics-based discovery of protein-ligand interactions on a proteomic scale. J Mol Biol 2011;408(1):147–62. 88. Adhikari J, Fitzgerald MC. SILAC-pulse proteolysis: a mass spectrometry-based method for discovery and cross-validation in proteome-wide studies of ligand binding. J Am Soc Mass Spectrom 2014;25(12):2073–83. 89. Zeng L, Shin W-H, Zhu X, et al. Discovery of nicotinamide adenine dinucleotide binding proteins in the Escherichia coli proteome using a combined energetic- and structural-bioinformatics-based approach. J Proteome Res 2017;16(2):470–80. 90. Leuenberger P, Ganscha S, Kahraman A, et al. Cell-wide analysis of protein thermal unfolding reveals determinants of thermostability. Science 2017;355(6327). 91. Martinez Molina D, Jafari R, Ignatushchenko M, et al. Monitoring drug target engagement in cells and tissues using the cellular thermal shift assay. Science 2013;341(6141):84–7.

89

90

Protein Homeostasis Diseases

92. Dai L, Prabhu N,Yu LY, Bacanu S, Ramos AD, Nordlund P. Horizontal cell biology: monitoring global changes of protein interaction states with the proteome-wide cellular thermal shift assay (CETSA). Annu Rev Biochem 2019;88:383–408. 93. Franken H, Mathieson T, Childs D, et al. Thermal proteome profiling for unbiased identification of direct and indirect drug targets using multiplexed quantitative mass spectrometry. Nat Protoc 2015;10(10):1567–93. 94. Savitski MM, Reinhard FBM, Franken H, et al.Tracking cancer drugs in living cells by thermal profiling of the proteome. Science 2014;346(6205):1255784. 95. Almqvist H, Axelsson H, Jafari R, et al. CETSA screening identifies known and novel thymidylate synthase inhibitors and slow intracellular activation of 5-fluorouracil. Nat Commun 2016;7:11040. 96. Becher I, Werner T, Doce C, et al. Thermal profiling reveals phenylalanine hydroxylase as an off-target of panobinostat. Nat Chem Biol 2016;12(11):908–10. 97. Martinez Molina D, Nordlund P. The cellular thermal shift assay: a novel biophysical assay for in situ drug target engagement and mechanistic biomarker studies. Annu Rev Pharmacol Toxicol 2016;56: 141–61. 98. Becher I, Andrés-Pons A, Romanov N, et al. Pervasive protein thermal stability variation during the cell cycle. Cell 2018;173(6):1495–507 e18. 99. Dai L, Zhao T, Bisteau X, et al. Modulation of protein-interaction states through the cell cycle. Cell 2018;173(6):1481–94 e13. 100. Geer MA, Fitzgerald MC. Energetics-based methods for protein folding and stability measurements. Annu Rev Anal Chem (Palo Alto Calif) 2014;7:209–28. 101. West GM, Tang L, Fitzgerald MC. Thermodynamic analysis of protein stability and ligand binding using a chemical modification- and mass spectrometry-based strategy. Anal Chem 2008;80(11):4175–85. 102. West GM,Thompson JW, Soderblom EJ, et al. Mass spectrometry-based thermal shift assay for proteinligand binding analysis. Anal Chem 2010;82(13):5573–81. 103. Tran DT, Adhikari J, Fitzgerald MC. Stable isotope labeling with amino acids in cell culture (SILAC)based strategy for proteome-wide thermodynamic analysis of protein-ligand binding interactions. Mol Cell Proteomics 2014;13(7):1800–13. 104. Dearmond PD, Xu Y, Strickland EC, Daniels KG, Fitzgerald MC. Thermodynamic analysis of proteinligand interactions in complex biological mixtures using a shotgun proteomics approach. J Proteome Res 2011;10(11):4948–58. 105. Walker EJ, Bettinger JQ, Welle KA, Hryhorenko JR, Ghaemmaghami S. Global analysis of methionine oxidation provides a census of folding stabilities for the human proteome. Proc Natl Acad Sci USA 2019;116(13):6081–90. 106. Faini M, Stengel F, Aebersold R. The evolving contribution of mass spectrometry to integrative structural biology. J Am Soc Mass Spectrom 2016;27(6):966–74. 107. Xu Y, Strickland EC, Fitzgerald MC. Thermodynamic analysis of protein folding and stability using a tryptophan modification protocol. Anal Chem 2014;86(14):7041–8. 108. Liu F, Fitzgerald MC. Large-scale analysis of breast cancer-related conformational changes in proteins using limited proteolysis. J Proteome Res 2016;15(12):4666–74. 109. Roberts JH, Liu F, Karnuta JM, Fitzgerald MC. Discovery of age-related protein folding stability differences in the mouse brain proteome. J Proteome Res 2016;15(12):4731–41.

CHAPTER 5

Classifying disease-associated variants using measures of protein activity and stability Michael Maglegaard Jepsena, Douglas M. Fowlerb, Rasmus Hartmann-Petersena, Amelie Steina, Kresten Lindorff-Larsena Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark Departments of Genome Sciences and Bioengineering, University of Washington, Seattle, WA, United States

a

b

Outline Introduction Selection of PTEN variants Comparing multiplexed assays and computational predictions to assess variant effects Loss of stability is a major source for loss of PTEN function Conclusions Methods Rosetta ∆∆G calculations Evolutionary sequence energies (Ẽ) Phosphatase-MAVE and VAMP-seq data Determining thresholds from receiver operating characteristic curves Analysis scripts References

92 96 97 100 102 104 104 104 105 105 105 105

Abbreviations ASD  autism spectrum disorder AUC  area under the curve FPR  false-positive rate gnomAD  genome aggregation database GREMLIN  generative regularized models of proteins MAVE  multiplexed assay of variant effects PHTS  PTEN hamartoma tumor syndrome PTEN  phosphatase and tensin homolog ROC  receiver operating characteristic TPR  true-positive rate VAMP-seq  variant abundance by massively parallel sequencing

Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00005-1 Copyright © 2020 Elsevier Inc. All rights reserved.

91

92

Protein Homeostasis Diseases

Introduction Technological advances have made human genome sequencing feasible in routine clinical contexts, revealing around 10,000 changes in the protein-coding regions of each individual,1 of which the majority will be very rare.2 Most of these changes will be missense variants in which a single amino acid is exchanged, contrasting, for example, around 200 protein-truncating changes.The resulting data avalanche highlights the challenge of interpreting the functional consequences of the many possible variants.3,4 While protein-truncating variants often lead to complete loss of function and can therefore generally be assumed to be detrimental, interpretation of functional consequences of missense mutations, in which a single amino acid is replaced by another, is more difficult. Here, we review and use state-of-the-art high-throughput methods for assessment of protein variants and discuss their performance in discriminating pathogenic from nondetrimental changes. PTEN (phosphatase and tensin homolog) is a dual-specificity phosphatase that acts on both protein and phosphatidylinositol 3,4,5-trisphosphate (PIP3) lipid substrates. Its lipid phosphatase function is important for PTEN’s role as a tumor suppressor where it counteracts phosphatidylinositol 3-kinase function.5 The structure of PTEN consists of two folded domains (Fig. 5.1A,B): an N-terminal phosphatase domain and a C-terminal C2 domain that acts to recruit PTEN to the membrane. Germline mutations in the PTEN gene are associated with various diseases including autism spectrum disorder (ASD) and different tumor-risk syndromes collectively known as PTEN hamartoma tumor syndrome (PHTS),5,6 and a number of variants of PTEN are known to be associated with PHTS and ASD (Fig. 5.1B). As discussed further below, we use PTEN as an example to compare and contrast different experimental and computational approaches to predict the disease outcome associated with PTEN variants. The same technological advances that enable high-throughput genome sequencing also allow for the generation and sequencing of large protein variant libraries. In combination with assays specific to the function of the protein of interest, these libraries enable highly parallel assessments of the functionality of each individual variant.12 These assays and the resulting datasets are referred to as MAVEs (multiplexed assays of variant effects).13 Briefly explained, a MAVE involves creating a large library of variants and subsequently selecting for a property of interest. Examples of selection methods include coupling protein function to cell growth, a fluorescent reporter, a transcriptional reporter, or selecting for ligand binding using phage or yeast display.14-17 As a result of the selection, variants in the library change in frequency depending on how well they are able to perform in the specific assay used. Finally, the frequency of each variant before and after selection is read out using next-generation DNA sequencing. In this way, MAVE experiments make it possible to calculate a score that quantifies the effect of the variant on the property and conditions selected.

Classifying disease-associated variants using measures of protein activity and stability

Figure 5.1  Overview of PTEN (phosphatase and tensin homolog) and the available single-variant data (color online). (A) We show the three-dimensional structure of PTEN (Protein Data Bank [PDB] ID: 1D5R7) with the phosphatase domain in yellow (top) and the C2 domain in pink (bottom). (B) PTEN domain architecture (top) and labels for variants that are seen at least 5 times in gnomAD (bottom; for details, see main text). Dark purple bars indicate positions where pathogenic variants have been described. (CF) Experimentally and computationally determined scores for single site variants. Green indicates wild type-like fitness/stability, magenta indicates low fitness/stability, and gray indicates missing data. (C) Phosphatase-MAVE,8 (D) VAMP-seq,9 (E) evolutionary sequence energies,10 (F) Rosetta cartesian ∆∆G,11 and (G) Venn diagram illustrating the availability of scores from each method and their intersections.

93

94

Protein Homeostasis Diseases

MAVEs have been applied to a substantial number of proteins and domains, overall indicating a surprising mutational tolerance, as many missense variants retain wild typelike function.18,19 At a general and somewhat simplified level, the selection system used in a MAVE may either (1) couple the function(s) of the protein, for example, its enzymatic activity, to the growth or another selectable phenotypic trait of the cell expressing the variant; or (2) be focused on one or a few specific properties of interest, for example, the stability of the variant or ability to bind a specific molecule. In practice, these two approaches are not clearly distinct because growth assays may often only depend on a subset of biological activities and because the specific molecular functions can play varying roles for the overall biological function of the protein. For practical reasons, MAVEs are typically implemented in single-cell model organisms or cultured mammalian cell lines. For biophysical properties like stability, this likely is not a major concern; however, with respect to interpretation of pathogenicity, especially for proteins embedded in complex cellular networks, differences between the model system and the human tissue may pose limitations. In the case of PTEN, two different MAVEs have been performed using different selection systems. As detailed further below, one experiment, which we term PhosphataseMAVE, involved coupling one of the intrinsic enzymatic functions of PTEN to the growth rate of the yeast Saccharomyces cerevisiae. The second experiment, termed VAMPseq (variant abundance by massively parallel sequencing), focused on how variants affect the cellular abundance of PTEN variants in a human cell line.The results from these two experiments provide us with the opportunity to use PTEN as a test case to compare the capability of both Phosphatase-MAVE and VAMP-seq in identifying pathogenic variants and to provide insight into the molecular mechanism(s) underlying pathogenicity. To determine residues important for PTEN function and clinically relevant genotype-phenotype relationships, Mighell et al. evaluated the effects of PTEN variants on lipid phosphatase activity in a yeast system.8 Specifically, the assay involved expressing a large library of PTEN variants in a yeast system in which the growth rate depends on the ability of PTEN to suppress the growth-inhibiting effect of the ectopically expressed catalytic subunit of human phosphatidylinositol 3-kinase. As described above, we term this experiment Phosphatase-MAVE because it integrates the consequences of the many different effects a missense variant can have, including intrinsic (lipid) phosphatase function, ability to localize to the cell membrane, and the overall stability and abundance of the protein. The resulting data (Fig. 5.1C) was shown to be able to discriminate likely pathogenic from benign alleles.8 Out of the 7638 possible single amino acid changes in PTEN, 6561 have accurate Phosphatase-MAVE measurements. While Phosphatase-MAVE data of disease-associated proteins are useful to clinical geneticists working to classify patient variants,20 it is more difficult to obtain direct insight into the molecular mechanism(s) underlying pathogenicity from these experiments.

Classifying disease-associated variants using measures of protein activity and stability

As described above, the outcome of the Phosphatase-MAVE experiment includes the combined effects of variants on several different molecular properties. Dissecting the mechanism(s) by which a variant impacts function will, however, be crucial for a deeper understanding of disease etiology and development of therapies. We and others have shown that loss of protein stability and subsequent degradation by the cellular protein quality control machinery is an important factor that underlies pathogenicity of variants in diverse proteins,21 including the cystic fibrosis transmembrane receptor,22 the metabolic enzyme phenylalanine hydroxylase,23,24 and the DNA-repair-associated proteins MutL homolog 1 (MLH1) and MSH2, in which hereditary variants are associated with increased cancer risk.25,26 These studies were based on low- or medium-throughput assessment of protein abundance, stability, and degradation. As a means to study protein stability and cellular abundance at greater throughput, Matreyek et al. recently introduced VAMP-seq as a MAVE of variant abundance.9 In VAMP-seq, a library of variants of the protein of interest is fused to green fluorescent protein (GFP). The library is expressed in cultured mammalian cells such that each cell expresses a single variant, and thus, each cell’s GFP fluorescence reports on the abundance of the protein variant. By sorting cells based on their fluorescence, and using nextgeneration sequencing to determine the frequency of every variant, a single VAMP-seq experiment provides quantitative abundance data for thousands of variants. Application of VAMP-seq to PTEN showed that many pathogenic variants were of low cellular abundance and that the resulting data (Fig. 5.1D) showed separation between pathogenic variants and common variants in the human population. Out of the 7638 possible single amino acid changes in PTEN, 4111 have accurate Phosphatase-MAVE measurements. In the absence of experimental data, computational methods can be used to predict the consequences of missense variants. We here complement and contrast the experimental Phosphatase-MAVE and VAMP-seq experiments with two computational methods, one that assesses sequence conservation through the calculation of an “evolutionary sequence energy” (Ẽ) and one that quantifies changes in protein stability (∆∆G). These methods have previously been applied to the identification and classification of pathogenic variants.4,27,28 By capturing how an amino acid sequence is constrained during evolution, changes in the evolutionary sequence energy may capture a wide range of variant effects beyond changes in protein stability but, for the same reason, do not impart direct mechanistic insight. In contrast, protein stability calculations are based on structural and mechanistic models for the effects of the individual variants. Loosely defined, Ẽ-values are conceptually similar to the outcome of the Phosphatase-MAVE experiment in that they capture a wide range of effects that a variant may have. In the same way, ∆∆G-values are related to the outcome of the VAMP-seq experiment by focusing more specifically on the effects of variants on protein stability. We used the software Gremlin10 to calculate evolutionary sequence energies (Fig. 5.1E). Gremlin takes as input a multiple sequence alignment of the protein and creates a

95

96

Protein Homeostasis Diseases

statistical model that accounts for both site conservation and covariation between sites. In this way, the changes in Ẽ capture more complex patterns of conservation, but calculating these values also requires a greater number of homologous sequences compared to simple site conservation alone. We used the Rosetta software to calculate changes in protein stability11 (∆∆G-values); these calculations do not require any information about the conservation of the protein but are instead based on a biophysical model of protein structure and folding together with a high-resolution, three-dimensional structure of the protein. As a consequence, ∆∆Gs are not available for loops and other residues missing from the experimentally determined structure (Fig. 5.1F). Out of the 7638 possible single amino acids, 2819 could be studied by all four methods (Fig. 5.1G); below, however, we focus only on a subset of 71 variants consisting of the 42 variants that are known to be pathogenic or the 29 that are found in the gnomAD database.29 We analyzed how well MAVE data and computational analyses perform in classifying disease-causing PTEN variants. We found that ∼60% of all disease-causing variants in PTEN appear to cause loss of function via loss of stability and thereby PTEN degradation. This result explains two observations regarding our ability to classify protein variants. First, methods that assess variant impact on a wider range of functional properties (Phosphatase-MAVE and Ẽ) perform slightly better in classifying disease variants than methods that focus more directly on protein stability and abundance (VAMP-seq and ∆∆G). Second, actually measuring these properties through MAVE experiments (Phosphatase-MAVE and VAMP-seq) gives rise to more accurate classifications compared to computational predictions (Ẽ and ∆∆G). We show that there is a large overlap between nonfunctional (Phosphatase-MAVE) and unstable (VAMP-seq) variants, indicating that loss of function often is due to loss of protein stability. Our results suggest that computational methods that combine structural modeling and evolutionary analyses could be useful in predicting variant effects when experimental data is not available.

Selection of PTEN variants As a source for pathogenic PTEN variants, we extracted 87 pathogenic missense variants from ClinVar,30 a public database containing information about phenotypic consequences of human genetic variation including curated assessments of both pathogenic and benign missense variants. Specifically, we only included variants that were marked as pathogenic or likely pathogenic, with review criteria provided and no conflicting annotations (accessed August 2018). We note that these variants may correspond to different disease outcomes ranging from ASD to PHTS, but we here collectively term them as pathogenic. As there are no PTEN variants classified as benign in ClinVar, we turned to gnomAD29 to examine whether there are variants that are sufficiently common in the population to make it likely that they do not cause disease.8,9,29 gnomAD is a database that

Classifying disease-associated variants using measures of protein activity and stability

aggregates results from genome and exome sequencing of >140,000 individuals from various disease-specific and population studies. Thus, while some of these individuals may have pathogenic PTEN variants, we expect that most variants-in particular, the more common ones-are likely not pathogenic. Of the 80 PTEN variants found in gnomAD (accessed April 2019), two (R173H and K289E) were classified as pathogenic in ClinVar, 32 variants were absent from the VAMP-seq data, 28 variants were located in unresolved parts of the crystal structure, and 3 variants were absent from the Phosphatase-MAVE data. As previously described,9 seven variants (A79T, M205V, D268E, S294R, Q298E, P354Q, and Y377F) were found at least five times in gnomAD and at an allele frequency greater than expected for a disease-causing PHTS variant with dominant inheritance. Of these seven variants, D268E was absent from the VAMP-seq data, S294R was absent from Phosphatase-MAVE, and four variants (S294R, Q298E, P354Q,Y377F) were found in unresolved parts of the PTEN crystal structure, and thus, we were not able to estimate ∆∆Gs for these variants. Thus, we used as likely benign variants A79T and M205V, which are found at allele frequencies of 1 × 10−4 and 2 × 10−5, respectively, in gnomAD. We supplemented these two variants with the remaining 27 variants that were found in gnomAD and were not annotated as pathogenic in ClinVar and where all four methods provided variant effect data. For convenience, we termed this aggregate data set as “gnomAD,” and remind the reader that although we have removed known pathogenic variants, we cannot exclude that some of the remaining variants in gnomAD are associated with ASD or PHTS. Our results below, however, suggest that most of them are likely not highly pathogenic.

Comparing multiplexed assays and computational predictions to assess variant effects We analyzed the distribution of scores for the gnomAD and pathogenic variants as obtained by the four different methods (Fig. 5.2). We note that the different methods have different scales and also differ in whether high or low numbers refer to a wild type-like readout. Specifically, for both experimental datasets (Phosphatase-MAVE, VAMP-seq), higher numbers correspond to the most functional/stable variants, whereas for the computational methods (Gremlin, Rosetta), the reverse is true. We opted to keep the scores as obtained from the different methods, but have reverted the orientation of the axes for the two computational data sets (Fig. 5.2, bottom). We found that all four methods show a clear difference in averages and distribution of values across the gnomAD and pathogenic variants, in agreement with previous analyses of the Phosphatase-MAVE and VAMP-seq data sets. Despite the clear differences in the distribution of scores for the two classes of variants, we also find that for each method, there is some overlap between the two (Fig. 5.2). Thus, for each of the four methods, some pathogenic variants would be scored errone-

97

98

Protein Homeostasis Diseases

Figure 5.2  Distributions for pathogenic and gnomAD variants (color online). Distributions of scores from each of the four methods for ClinVar pathogenic (mauve) and gnomAD (green) variants, shown as Raincloud plots.31 Boxes illustrate the 25th to 75th percentiles, with the median indicated by a horizontal line. Whiskers extend to minimum and maximum values, though they are at most 1.5 times the distance between the 25th and 75th percentiles (interquartile range) away from the box boundaries. As described in the main text, pathogenic variants were removed from the gnomAD set. A79T and M205V are specifically highlighted (crosses), as these are the most common PTEN variants in gnomAD and, thus, the most likely to be benign. Only variants for which all of the four methods provide data are included (see also Fig. 5.1G). Axes are oriented such that values near the bottom correspond to detrimental variants, while values near the top are wild type like.

ously as benign, and some gnomAD variants would be scored as pathogenic. In order to quantify the ability of each method to separate the pathogenic from the gnomAD variants, we performed a receiver operating characteristic (ROC) curve analysis of each of the four methods. In this analysis, we varied a cutoff value over the experimentally determined or calculated scores, and for each value, we quantified the number of true and false positive and negative classifications (Fig. 5.3).The resulting curves show the balance between sensitivity (the fraction of the pathogenic variants that are actually classified as pathogenic) and specificity (so that only a few gnomAD variants are misclassified as pathogenic). In line with expectations from the distribution of scores (Fig. 5.2), the resulting ROC curves clearly show that all methods have good predictive value for distinguishing pathogenic from gnomAD PTEN variants.The area under the curve (AUC)

Classifying disease-associated variants using measures of protein activity and stability

Figure 5.3  Receiver operating characteristic (ROC) curves for each method. ROC curves describe how well each method separates the ClinVar pathogenic variants from the gnomAD variants. As expected, methods assessing overall PTEN (phosphatase and tensin homolog) function perform better, with Phosphatase-MAVE experimental data (blue, AUC: 0.94) exceeding evolutionary sequence energies (green, AUC: 0.86) in overall sensitivity and specificity. Assessment of stability alone correctly identifies most pathogenic and gnomAD variants as well, again with experimental data (VAMP-seq, orange, AUC: 0.81) exceeding performance of Rosetta ∆∆G calculations (red, AUC: 0.73). The insert shows the AUC values and changes in AUC when moving between experimental and computational methods, and between methods that capture activity broadly or methods focused on stability.

provides a convenient measure to score how well a classifier may balance sensitivity and specificity across different thresholds: a ROC curve along the diagonal corresponds to a random prediction method and has AUC = 0.5, whereas a perfect classifier can separate pathogenic and benign variants with no overlap, and would have AUC = 1.When assessing the four methods in this way, we found that each of the two experimental approaches performed better than the corresponding purely computational method (Fig. 5.3).These

99

100

Protein Homeostasis Diseases

results are in line with the expectation that it is more informative to have experimental data available rather than a purely computational prediction of the corresponding effect. As also expected, we found that the two methods that probe a wider range of functionally relevant properties (Phosphatase-MAVE and Gremlin) are able to capture variant effects more accurately than those that focus more directly on protein stability (VAMPseq and Rosetta). Specifically, we find AUCs of 0.94 for Phosphatase-MAVE, 0.81 for VAMP-seq, 0.86 for Gremlin, and 0.73 for Rosetta. We also calculated precision-recall curves and determined F1 scores at the optimal cutoffs for each approach (see section, Methods), and observed the same trends. Specifically, the F1 scores are 0.90 for Phosphatase-MAVE, 0.76 for VAMP-seq, 0.82 for Gremlin and 0.71 for Rosetta.

Loss of stability is a major source for loss of PTEN function The analyses described above highlight that missense variants might cause disease through a number of different mechanisms, including loss (or gain) of intrinsic enzymatic function or ability to interact with substrates or effector molecules, or simply by loss of stability.4 The high AUCs for the methods that probe only stability but not other aspects contributing to protein function (VAMP-seq and Rosetta) suggest that loss of stability and resulting decrease in cellular abundance is a key driver for PTEN-associated diseases. This observation is in line with previous computational analyses of a range of proteins4,32-34 as well as experimental studies of disease-causing variants in selected proteins including PTEN.23-26,35-37 As described above, we found that Phosphatase-MAVE and Gremlin are better able to classify variants than VAMP-seq and Rosetta, suggesting that some variants cause loss of function and disease through other mechanisms than loss of stability. We have recently introduced two-dimensional plots of the change in stability, ∆∆G, as predicted by Rosetta and the evolutionary sequence energy estimated using Gremlin to analyze the interplay between loss of stability and other mechanisms for disease.24,26 Here, we supplement our analysis by a corresponding comparison of multiplexed experiments by using gnomAD and pathogenic variants to compare VAMP-seq and PhosphataseMAVE (Fig. 5.4A) and the computational analyses (Fig. 5.4B). As expected from the ROC curve analysis, loss of stability plays an important role for loss of function so that most gnomAD variants are stable and functional, and disease-causing variants are unstable and nonfunctional. We note that this observation supports our hypothesis that most of the PTEN variants found in gnomAD are not pathogenic. In both the experimental and computational analysis, we found, however, a number of variants that were stable (either by VAMP-seq or Rosetta calculations) but are likely not functional (by Phosphatase-MAVE or the evolutionary sequence energy). Such variants would be relevant candidates for analysis of biochemical function, ligand binding, or membrane interactions.

Classifying disease-associated variants using measures of protein activity and stability

Figure 5.4  Two-dimensional landscapes integrating changes in stability and fitness (color online). Combining scores from assessment of stability and function shows that most gnomAD variants (green) are wild type like, while most pathogenic variants (purple) are correctly identified by one, and often both, metrics. Comparing the landscape based on experimental data (left) to that based on predictions (right) shows that availability of experimental data leads to better separation of pathogenic variants, but also that the predictions will provide a good starting point at substantially lower cost. The dashed lines correspond to cutoff values derived from the receiver operating characteristic (ROC) curves, as those are values that give rise to the point on the ROC curve closest to the upper left corner (true-positive rate [TPR] = 1, false-positive rate [FPR] = 0).

101

102

Protein Homeostasis Diseases

We used the ROC-curve analyses to derive cutoffs between the gnomAD and pathogenic variants for all four sources of data to separate each of the two-dimensional plots into four quadrants. Through this analysis, we found that 27 of the 42 pathogenic variants fall in the unstable-and-nonfunctional quadrant in the experimental analysis (Fig. 5.4A), and 24 of the 42 variants fall in the same quadrant in the computational analysis (Fig. 5.4B).Thus, together, these analyses suggest that about 60% of disease-causing variants in PTEN cause disease via loss of stability and cellular abundance. The observation that most unstable variants are nonfunctional but that not all nonfunctional variants are unstable is also reflected in the distribution of scores (Fig. 5.2) and ROC analyses (Fig. 5.3).Thus, if we examine the ROC curves in the region of high specificity (10% false-positive rate), we reach a relatively high sensitivity (true-positive rate) for all methods (Phosphatase-MAVE, 0.90;VAMP-seq, 0.66; Gremlin,0.66; Rosetta, 0.53). Thus, one can find ∼60% of the pathogenic variants at a relatively low false discovery rate simply by examining loss of stability.

Conclusions Human exome and genome sequencing methods provide new opportunities to understand mechanistic origins of diseases and to use DNA sequencing for patient diagnosis. Large-scale sequencing efforts are revealing millions of missense variants, and we need improved and robust tools to assess potential pathogenicity of individual variants. As many variants are extremely rare, it is difficult to derive robust genotype-phenotype correlations for these even from large-scale patient studies. Experimental studies of variant effects can be extremely useful to assess functional effects and biological consequences of variants. Integrating insights from one or a few variants in retrospective analyses or in-depth mechanistic studies may sometimes also be used for assessing pathogenicity of newly discovered variants. More recently, however, multiplexed assays that use phenotypic selection and deep sequencing to study the effects of thousands of variants in a single experiment have enabled new approaches to classify variants by experiments. Such experiments can provide a quantitative assessment of variant effects for all possible missense variants that may, in turn, be used as a “look-up table” for any newly discovered variant. For example, in a recent study, the effect of variants of the BRCA1 gene were assayed using saturation genome editing, leading to very accurate predictions of clinical outcomes.20 Performing activity-based MAVE experiments such as Phosphatase-MAVE requires establishing a selectable system that relates protein function to, for example, cell growth, fluorescence, or binding. While extremely powerful to assess pathogenicity, such experiments result from the complex interplay of many different effects a variant may have, and thus, they less directly provide mechanistic insight into how the variants cause disease. On the other hand, VAMP-seq experiments provide a more general approach that can

Classifying disease-associated variants using measures of protein activity and stability

be used to study effects of variants across many different proteins and probe directly how missense variants affect function via perturbing cellular abundance. This generality, however, comes at a cost:VAMP-seq experiments miss some pathogenic variants that are detrimental for reasons other than loss of stability. On the other hand, the experiments directly suggest a mechanistic origin for the variant effects. In the foreseeable future, it will likely be difficult to establish and execute robust activity-based MAVE experiments for a very large number of proteins. Thus, computational methods for prediction of variant effects continues to be an important alternative to experiments. Such computational methods may be focused on either a single type of effect (e.g. stability, interactions, etc.) or combine multiple effects either directly or indirectly via analyses of evolutionary conservation. We have here taken advantage of the availability of both Phosphatase-MAVE and VAMP-seq data for the human protein PTEN to analyze how well these methods are able to classify known pathogenic variants, and we have compared the results to computational methods. Further, by contrasting analyses focused on protein stability with results from methods that include other aspects of function, we have been able to assess the importance of loss of stability as a mechanism for loss of PTEN function in disease. Our results show that all four methods that we analyzed are able to provide relatively accurate classifications of variant effects with AUCs ranging from 0.73 to 0.94 (Fig. 5.3). Here, we remind the reader that we in our selection of pathogenic variants have combined disease-causing variants for different diseases (ASD and PHTS), although it has been suggested that ASD results from variants with milder effects compared to those that give rise to PHTS.8 Similarly, we note that we have used also rare variants from gnomAD as a proxy for benign variant, and that some of these may indeed give rise to ASD or PHTS. Either choice may limit the accuracy in our analysis. With these caveats in mind, we highlight two observations regarding prediction accuracies. First, the experimental methods give rise to AUCs that are about 0.1 unit higher than the corresponding computational method (Phosphatase-MAVE [0.94] versus Gremlin [0.86] and VAMP-seq [0.81] versus Rosetta [0.73]). Second, the methods that probe function more generally give rise to AUCs that are about 0.1 unit higher compared to those that probe only protein stability (Phosphatase-MAVE [0.94] versus VAMP-seq [0.81] and Gremlin [0.86] versus Rosetta [0.73]). Similar trends were observed for precision-recall-curve-based F1 scores. Thus, we find that ∼60% of diseasecausing variants in PTEN appear to arise due to loss of protein stability, and these can be discovered relatively accurately using VAMP-seq and Rosetta ∆∆G calculations. Our results, however, reveal additional variants that are better captured by methods such as Phosphatase-MAVE that also include more direct measures of activity. One advantage of the approaches that focus on protein stability is that they provide a more direct mechanistic model for how the variants cause disease. Indeed, our two-dimensional analysis of stability and “function” reveals that many, but not all, dis-

103

104

Protein Homeostasis Diseases

ease-causing variants appear to cause loss of function because the variants cause loss of stability and decreased cellular abundance (Fig. 5.4). Thus, experiments or predictions that probe stability and other properties relevant for function can help guide follow-up studies, as protein homeostasis assays are likely insightful for variants that are unstable and nonfunctional, while, for example, biochemical characterization may be more appropriate for those that are nonfunctional yet stable. We conclude that at the current stage, experimental methods still perform better than the corresponding computational approaches and that many, but not all, pathogenic variants in PTEN appear to give rise to disease via loss of stability and cellular abundance. In the future, we aim to explore prediction methods that combine the evolutionary sequence energy, Ẽ, with ∆∆G predictions to improve accuracy further,26 and to provide a more automated assessment of disease mechanism. We also aim to extend these analyses using additional proteins and diseases and to examine the generality of these results and conclusions.

Methods Rosetta ∆∆G calculations We used the “cartesian_ddg”11 application in Rosetta with the “beta_nov_16” variant of the Rosetta energy function to perform the ∆∆G calculations. This approach relies on scoring the calculated conformations in cartesian space and include the “cart_bonded” scoreterm of the Rosetta energy function.11,38 As input for these calculations, we used the crystal structure of PTEN, including the bound ligand (Protein Data Bank [PDB] ID: 1D5R) in the stability calculations.We used the software Babel39 to parameterize the ligand. To accommodate for the missing loop in the crystal structure (residues 281 to 313), we added the flag “-missing_density_to_jump” during relaxation of the structure. When calculating ∆∆G, we used three iterations, which were subsequently averaged. For each mutation, we subtracted the average stability of the wild-type protein from that of the variant.The resulting difference in stability was finally divided by 2.9 to bring the ∆∆G values from Rosetta energy units onto a scale corresponding to kcal/mol (Frank DiMaio, University of Washington; personal correspondence). Note that this rescaling does not affect the result of our analysis of variant classification.

Evolutionary sequence energies (Ẽ) We used HHblits40 to create a multiple sequence alignment based on the UniProt sequence of PTEN (UniProt AC P60484). Next, we used Gremlin10 to build a sequence model capturing both conservation and pairwise residue covariation from this alignment, and then we calculated the log odds score between each single residue variant and the wild type.

Classifying disease-associated variants using measures of protein activity and stability

Phosphatase-MAVE and VAMP-seq data The VAMP-seq data can be accessed and downloaded at https://abundance. gs.washington.edu/shiny/stability/. The Phosphatase-MAVE data can be accessed in the supplementary material of the online article from Mighell et al.8

Determining thresholds from receiver operating characteristic curves Thresholds for separating likely pathogenic from likely benign variants were calculated for each metric by determining the point on the ROC curve that is closest to (0,1), the optimum where perfect specificity and sensitivity would be achieved. In this case, we gave comparable weight to detection of pathogenic and benign variants, but note here that, depending on the application, different choices of threshold may be most suitable.

Analysis scripts The scripts used in the analyses and for making the figures of this manuscript are available at https://github.com/KULL-Centre/papers/tree/master/2019/PTEN-variantsJepsen-et-al.

Acknowledgments We thank Dr. Sofie V. Nielsen, Mustapha C. Ahmed, and Prof. Fritz Roth for helpful discussions and comments. This work is supported by a Novo Nordisk Foundation Challenge Grant (PRISM) (D.M.F, R.H.-P., A.S., K.L.-L.), the Lundbeck Foundation (A.S.), and the National Institute of General Medical Sciences (1R01GM109110 and 1RM1HG010461 to D.M.F.). D.M.F. is a CIFAR Azrieli Global Scholar.

References 1. Kiryluk K, Goldstein DB, Rowe JW, Gharavi AG,Wapner R, Chung WK. Precision medicine in internal medicine. Ann Intern Med 2019;170(9):635. doi: 10.7326/M18-0425. 2. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature 2016;536(7616):285–91. doi: 10.1038/nature19057. 3. Manolio TA, Fowler DM, Starita LM, et al. Bedside back to bench: building bridges between basic and clinical genomic research. Cell 2017;169(1):6–12. doi: 10.1016/j.cell.2017.03.005. 4. Stein A, Fowler DM, Hartmann-Petersen R, Lindorff-Larsen K. Biophysical and mechanistic models for disease-causing protein variants. Trends Biochem Sci 2019;44(7):575–88. doi: 10.1016/j. tibs.2019.01.003. 5. Yehia L, Ngeow J, Eng C. PTEN-opathies: from biological insights to evidence-based precision medicine. J Clin Invest 2019;129(2):452–64. doi: 10.1172/JCI121277. 6. Leslie NR, Longy M. Inherited PTEN mutations and the prediction of phenotype. Semin Cell Dev Biol 2016;52:30–8. doi: 10.1016/j.semcdb.2016.01.030. 7. Lee J-O,Yang H, Georgescu M-M, et al. Crystal structure of the PTEN tumor suppressor: implications for its phosphoinositide phosphatase activity and membrane association. Cell 1999;99(3):323–34. doi: 10.1016/S0092-8674(00)81663-3. 8. Mighell TL, Evans-Dutson S, O’Roak BJ. A saturation mutagenesis approach to understanding PTEN lipid phosphatase activity and genotype-phenotype relationships. Am J Hum Genet 2018;102(5):943– 55. doi: 10.1016/j.ajhg.2018.03.018.

105

106

Protein Homeostasis Diseases

9. Matreyek KA, Starita LM, Stephany JJ, et al. Multiplex assessment of protein variant abundance by massively parallel sequencing. Nat Genet 2018;50(6):874. doi: 10.1038/s41588-018-0122-z. 10. Balakrishnan S, Kamisetty H, Carbonell JG, Lee S-I, Langmead CJ. Learning generative models for protein fold families. Proteins Struct Funct Bioinforma 2011;79(4):1061–78. doi: 10.1002/prot.22934. 11. Park H, Bradley P, Greisen P, et al. Simultaneous optimization of biomolecular energy functions on features from small molecules and macromolecules. J Chem Theory Comput 2016;12(12):6201–12. doi: 10.1021/acs.jctc.6b00819. 12. Fowler DM, Fields S. Deep mutational scanning: a new style of protein science. Nat Methods 2014;11(8):801–7. doi: 10.1038/nmeth.3027. 13. Gasperini M, Starita L, Shendure J. The power of multiplexed functional analysis of genetic variants. Nat Protoc 2016;11(10):1782–7. doi: 10.1038/nprot.2016.135. 14. Fowler DM, Araya CL, Fleishman SJ, et al. High-resolution mapping of protein sequence-function relationships. Nat Methods 2010;7(9):741–6. doi: 10.1038/nmeth.1492. 15. Starita LM,Young DL, Islam M, et al. Massively parallel functional analysis of BRCA1 RING domain variants. Genetics 2015;200(2):413–22. doi: 10.1534/genetics.115.175802. 16. Roscoe BP,Thayer KM, Zeldovich KB, Fushman D, Bolon DNA. Analyses of the effects of all ubiquitin point mutants on yeast growth rate. J Mol Biol 2013;425(8):1363–77. doi: 10.1016/j.jmb.2013.01.032. 17. Staller MV, Holehouse AS, Swain-Lenz D, Das RK, Pappu RV, Cohen BA. A high-throughput mutational scan of an intrinsically disordered acidic transcriptional activation domain. Cell Syst 2018;6(4):444–55. doi: 10.1016/j.cels.2018.01.015 e6. 18. Gray VE, Hause RJ, Fowler DM. Analysis of large-scale mutagenesis data to assess the impact of single amino acid substitutions. Genetics 2017;207(1):53–61. doi: 10.1534/genetics.117.300064. 19. Kinney JB, McCandlish DM. Massively Parallel Assays and Quantitative Sequence-Function Relationships. Annu Rev Genomics Hum Genet 2019;20:99–127. doi: 10.1146/annurev-genom-083118-014845. 20. Findlay GM, Daza RM, Martin B, et al. Accurate classification of BRCA1 variants with saturation genome editing. Nature 2018;562(7726):217. doi: 10.1038/s41586-018-0461-z. 21. Clausen L, Abildgaard AB, Gersing SK, Stein A, Lindorff-Larsen K, Hartmann-Petersen R. Protein stability and degradation in health and disease. In: Donev R editor. Advances in protein chemistry and structural biology.Vol. 114. Molecular Chaperones in Human Disorders. Cambridge, Massachusetts: Academic Press; 2019. p. pp. 61–83. doi: 10.1016/bs.apcsb.2018.09.002. 22. Callebaut I, Hoffmann B, Mornon J-P. The implications of CFTR structural studies for cystic fibrosis drug development. Curr Opin Pharmacol 2017;34:112–8. doi: 10.1016/j.coph.2017.09.006. 23. Pey AL, Stricher F, Serrano L, Martinez A. Predicted effects of missense mutations on native-state stability account for phenotypic outcome in phenylketonuria, a paradigm of misfolding diseases. Am J Hum Genet 2007;81(5):1006–24. doi: 10.1086/521879. 24. Scheller R, Stein A, Nielsen SV, et al.Toward mechanistic models for genotype-phenotype correlations in phenylketonuria using protein stability calculations. Hum Mutat 2019;40(4):444–57. doi: 10.1002/ humu.23707. 25. Nielsen SV, Stein A, Dinitzen AB, et al. Predicting the impact of Lynch syndrome-causing missense mutations from structural calculations. PLOS Genet 2017;13(4):e1006739. doi: 10.1371/journal. pgen.1006739. 26. Abildgaard AB, Stein A, Nielsen SV, et al. Computational and cellular studies reveal structural destabilization and degradation of MLH1 variants in Lynch syndrome. eLife 2019;8:e49138. doi: 10.7554/ eLife.49138. 27. Hopf TA, Ingraham JB, Poelwijk FJ, et al. Mutation effects predicted from sequence co-variation. Nat Biotechnol 2017;35(2):128–35. doi: 10.1038/nbt.3769. 28. Feinauer C, Weigt M. Context-aware prediction of pathogenicity of missense mutations involved in human disease. BioRxiv 2017;103051. doi: 10.1101/103051. 29. Karczewski KJ, Francioli LC, Tiao G, et al. Variation across 141,456 human exomes and genomes reveals the spectrum of loss-of-function intolerance across human protein-coding genes. BioRxiv 2019;531210. doi: 10.1101/531210. 30. Landrum MJ, Lee JM, Benson M, et al. ClinVar: improving access to variant interpretations and supporting evidence. Nucleic Acids Res 2018;46(D1):D1062–7. doi: 10.1093/nar/gkx1153.

Classifying disease-associated variants using measures of protein activity and stability

31. Allen M, Poggiali D, Whitaker K, Marshall TR, Kievit R. Raincloud plots: a multi-platform tool for robust data visualization [PeerJ Preprints]. https://peerj.com/preprints/27137/. 32. Casadio R,Vassura M, Tiwari S, Fariselli P, Martelli PL. Correlating disease-related mutations to their effect on protein stability: a large-scale analysis of the human proteome. Hum Mutat 2011;32(10):1161– 70. doi: 10.1002/humu.21555. 33. Pal LR, Moult J. Genetic basis of common human disease: insight into the role of missense SNPs from genome-wide association studies. J Mol Biol 2015;427(13):2271–89. doi: 10.1016/j.jmb.2015.04.014. 34. Redler RL, Das J, Diaz JR, Dokholyan NV. Protein destabilization as a common factor in diverse inherited disorders. J Mol Evol 2016;82(1):11–6. doi: 10.1007/s00239-015-9717-5. 35. Bullock AN, Henckel J, Fersht AR. Quantitative analysis of residual folding and DNA binding in mutant p53 core domain: definition of mutant states for rescue in cancer therapy. Oncogene 2000;19(10):1245. doi: 10.1038/sj.onc.1203434. 36. Johnston SB, Raines RT. Conformational stability and catalytic activity of PTEN variants linked to cancers and autism spectrum disorders. Biochemistry 2015;54(7):1576–82. doi: 10.1021/acs. biochem.5b00028. 37. Johnston SB, Raines RT. PTENpred: a designer protein impact predictor for PTEN-related disorders. Journal of Computational Biology 2016;23(12):969–75. doi: 10.1089/cmb.2016.0058. 38. Alford RF, Leaver-Fay A, Jeliazkov JR, et al. The rosetta all-atom energy function for macromolecular modeling and design. J Chem Theory Comput 2017;13(6):3031–48. doi: 10.1021/acs.jctc.7b00125. 39. O’Boyle NM, Banck M, James CA, Morley C,Vandermeersch T, Hutchison GR. Open babel: an open chemical toolbox. J Cheminformatics 2011;3(1):33. doi: 10.1186/1758-2946-3-33. 40. Zimmermann L, Stephens A, Nam S-Z, et al. A completely reimplemented MPI bioinformatics toolkit with a new HHpred server at its core. J Mol Biol 2018;430(15):2237–43. doi: 10.1016/j. jmb.2017.12.007.

107

CHAPTER 6

Protein destabilization and degradation as a mechanism for hereditary disease Sofie V. Nielsen, Signe M. Schenstrøm, Caspar E. Christensen, Amelie Stein, Kresten Lindorff-Larsen, Rasmus Hartmann-Petersen Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark

Outline Introduction to protein quality control Protein quality control in hereditary diseases Protein folding and refolding Protein quality control– mediated degradation via the ubiquitin-proteasome system Protein quality control degrons Local versus global unfolding Potential therapeutic approaches to protein quality control–linked hereditary diseases References

112 113 115 115 117 118 119 121

Abbreviations ALS  amyotrophic lateral sclerosis BAG  Bcl-2-associated athanogene BiP  binding-immunoglobulin protein CFTR  Cystic fibrosis transmembrane conductance regulator CHIP  carboxy terminus of Hsp70-interacting protein CIC  capicua DUBs  deubiquitylating enzymes ERAD  ER-associated degradation GOF  gain-of-function HDACis  histone deacetylase inhibitors LOF  loss-of-function MDM  mouse double minute MALT  mucosa-associated lymphoid tissue lymphoma translocation protein MSH  MutS protein homolog PDI  protein disulfide-isomerase PAH  phenylalanine hydroxylase PQC  protein quality control PROTACs  proteolysis-targeting chimeras San1  Sir antagonist 1 Ssb  stress-seventy subfamily B STUB  STIP1 homology and U box-containing protein UPS  ubiquitin-proteasome system Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00006-3 Copyright © 2020 Elsevier Inc. All rights reserved.

111

112

Protein Homeostasis Diseases

Introduction to protein quality control Mammalian cells typically express more than 10,000 different protein species that are highly versatile and structurally diverse. During synthesis on the ribosome, the nascent proteins are produced as linear chains of amino acids that, in most cases, must fold into the native structure for the protein to become biologically active. The traditional protein structure-function paradigm states that protein function depends on a well-defined, folded three-dimensional structure determined by the amino acid sequence.1 Although most proteins need to be folded to function, many proteins are found to be marginally stable. While the mechanisms underlying this observation are not fully known, likely reasons include the necessity for proteins to be flexible to function and be degraded. Further, since most amino acid changes causes destabilization of the native fold,2,3 random genetic variation is likely to push proteins towards decreased stability, whereas selection for function will keep proteins folded; this balance also leads to marginal stability.4 Consequently, even minor changes in the protein sequence may affect the protein folding process and the stability of the native structure, and accordingly, mutations in genes encoding folded proteins may trigger a structural destabilization or misfolding of the protein.5–7 Typically, destabilized, partially unfolded, or misfolded proteins are characterized by exposing, at least transiently, otherwise buried hydrophobic residues to the solvent, hence failing to bury nonpolar amino acids in the core of the protein. This feature may drive a protein chain towards collapse and lead to formation of aggregates.8,9 Accordingly, a misfolding event potentially jeopardizes the function, not only of the misfolded protein itself, but also of other proteins that may be caught up in nonnative interactions with the misfolded protein, and therefore represents a significant threat to the cell and the health of the organism. While the molecular origins of such aggregation-induced cell toxicity are not generally understood, all cells have evolved several protective measures to counter the accumulation of misfolded proteins. Collectively, these mechanisms are known as protein quality control (PQC) systems, which serve to either refold or degrade the misfolded proteins.10,11 As nascent proteins fold, they often undergo transitions through various folding intermediates before reaching the native state (Fig. 6.1).12–14 During this process, molecular chaperones catalyze the folding and protect the nascent proteins from aggregation, and in a similar manner, chaperones may also catalyze the refolding of proteins that become damaged after synthesis or even disentangle protein aggregates (Fig. 6.1). PQC-associated degradation, on the other hand, relies on proteases to eliminate nonnative proteins from the cell.10,11 Since degradation obviously eliminates the function of proteins, these PQC systems must be highly specific to ensure that only incorrectly folded or damaged proteins are targeted. At the same time, the PQC system must also be broadly inclusive to ensure that any of the >10,000 structurally diverse cellular proteins can be targeted, as well as those derived from infections by other species, and the variation of defects a

Protein destabilization and degradation as a mechanism for hereditary disease

Figure 6.1  Protein folding, misfolding, aggregation, and degradation. Protein folding begins during translation on the ribosome and typical occurs through various folding intermediates until the native conformation is attained (top). The ubiquitin-proteasome system (UPS) degrades both folded proteins—as part of regulatory processes—and destabilized/misfolded proteins (trashcan). Molecular chaperones (not shown) catalyze folding events, but also disaggregation of protein aggregates (bottom). Aggregated proteins are typically cleared by autophagy (Pac-Man).

single protein may undergo is also vast. Finally, the relatively low ratio of abnormal to normal proteins in the cell and the fact that many proteins, during synthesis, naturally and transiently form intermediately folded states further increases the requirement of the PQC system to discriminate between substrates and nonsubstrates.15 Accordingly, the substrate selection of the PQC system is the result of a balance between specificity and the ability to engage multiple structurally and chemically diverse targets. Hence, a too-strictly-tuned PQC system would result in the degradation of many native or nearnative and harmless functional proteins, while a too-lax PQC system would result in accumulation of nonnative, nonfunctional, and potentially toxic protein species.

Protein quality control in hereditary diseases Under normal conditions, our PQC system appears perfectly balanced to ensure that structurally destabilized proteins are folded or eliminated from the cell. However, in various disease states, and particularly in the case of several genetic diseases, the PQC system can appear unbalanced. In general, these so-called protein misfolding or conformational

113

114

Protein Homeostasis Diseases

diseases separate into two groups: toxic-gain-of-function disorders that typically follow a dominant inheritance pattern and loss-of-function disorders that are normally recessive. In the case of the gain-of-function group of protein misfolding disorders, genetic variations lead to misfolded proteins, but the PQC system, for reasons that are still unclear, fails to clear the misfolded proteins from the cell.16,17 In turn, this often leads to the formation of protein aggregates,18 such as those observed in Huntington’s disease19 and other polyglutamine expansion disorders,20 or amyloid fibrils typically associated with neurodegenerative pathologies such as Alzheimer’s and Parkinson’s disease.5,9,21–23 Indeed it has been suggested that the activity of the PQC system declines with age,24 which may explain why several of these disorders display an age-dependent occurrence. Diseases with loss-of-function phenotypes are generally caused by the cell lacking activity of the affected protein. Obviously, this can occur by a mutation in the active site of an enzyme, which will render the enzyme nonfunctional. However, since typically only a few residues in a protein constitute an active site or an essential binding motif, the lossof-function phenotype is statistically more likely to be the result of mutations rendering the protein structurally unstable,25 which in turn leads to degradation, insufficient cellular amounts of the protein, and a loss-of-function phenotype. Numerous examples of this have been reported (Table 6.1), including metabolic disorders, developmental disorders, lysosomal storage diseases, neurological diseases,26–30 and cancers.2,31–34 Cystic fibrosis constitutes another well described example of PQC-mediated clearance of the common cystic fibrosis transmembrane conductance regulator (CFTR) ∆F508 disease-causing variant.35 Table 6.1  Examples of protein destabilization/misfolding in hereditary diseases. Disease

Gene

Inheritancea

References

Lynch syndrome Phenylketonuria Cystic fibrosis Parkinson’s disease Parkinson’s disease Megaloblastic anemia Birt-Hogg-Dubé syndrome Lesch-Nyhan disease von Hippel-Lindau disease Gaucher disease Neurofibromatosis type 2 OTULIN-related autoinflammatory syndrome Machado-Joseph disease Amyotrophic lateral sclerosis

MSH2 PAH CFTR PARK7 PARK2 DHFR FLCN HPRT1 VHL GBA NF2 OTULIN

Dominant LOF Recessive LOF Recessive LOF Recessive LOF Recessive LOF Recessive LOF Dominant LOF X-linked LOF Dominant LOF Recessive LOF Recessive LOF Recessive LOF

[32,33] [26,27] [35] [28,116] [117] [118,119] [120] [121] [31] [122] [123] [124]

ATX3 SOD1

Dominant GOF Dominant GOF

[17] [125,126]

a

GOF, gain of function; LOF, loss of function.

Protein destabilization and degradation as a mechanism for hereditary disease

Protein folding and refolding The PQC system is constantly active and operates already as proteins are synthesized. During translation, a set of specific chaperones associates with the ribosome and facilitates de novo folding. This event occurs cotranslationally, thus enabling the nascent polypeptide chain to start folding of the N-terminal end while still associated with the ribosome.8,36,37 These ribosome-associated chaperones constitute one of two independent chaperone networks in eukaryotes.37 Specialized chaperones have been shown to be ribosome associated, such as the heat shock protein 70 (HSP70)–type chaperones (stress-seventy subfamily B, Ssb1 and Ssb2) in yeast that hinder misfolding and aggregation of nascent polypeptide chains.37,38 Downstream of translation, the other cytosolic chaperones continue to assist the newly synthesized proteins that require additional folding assistance. Furthermore, these chaperones also aid refolding of other proteins that happen to misfold.39,40 This role is predominantly performed by HSP70s, HSP90s, and certain chaperonins (e.g., HSP60s)5,8 (see also Chapter 2). In addition to their role in assisting proteins to fold and preventing misfolding, some chaperones also participate in targeting misfolded proteins for degradation.5,41–45 Thus, in cases where attempts to refold a misfolded protein fail, the chaperones instead redirect the protein for degradation. This process is controlled by various cochaperones, such as certain Bcl-2-associated athanogene (BAG) domain co-chaperones that directly establish interaction between the molecular chaperones and the cellular protein degradation systems.46–49

Protein quality control–mediated degradation via the ubiquitinproteasome system The eukaryotic cell manages intracellular proteolysis through two major systems: the UPS and the autophagy-lysosome pathway. While the UPS typically degrades soluble proteins, the autophagy system generally eliminates larger insoluble protein aggregates (Fig. 6.1) that may derive from proteins that have escaped the UPS.50 In general, the UPS is responsible for the majority of intracellular protein degradation, and UPS substrates count both native proteins that are degraded as part of regulatory events (such as the degradation of cyclins that ensures cell cycle progression) and abnormal/misfolded proteins that are degraded as part of the PQC system (Fig. 6.1).50 For recent advances regarding autophagy-mediated clearance of misfolded proteins, we refer to recent reviews on this topic.51,52 Degradation of proteins through the UPS involves the successive occurrence of two events: the first is the covalent attachment of ubiquitin molecules to the substrate, an event referred to as ubiquitylation. The second includes degradation of the ubiquitylated substrate by the 26S proteasome, which constitutes the proteolytic part of the UPS.53 With few exceptions, only proteins conjugated to ubiquitin are degraded by the proteasome.54

115

116

Protein Homeostasis Diseases

The conjugation of ubiquitin to a protein occurs in three steps and requires the consecutive action of three enzyme families: the ubiquitin-activating enzymes (E1), the ubiquitin-conjugating enzymes (E2), and the ubiquitin-protein ligases (E3).54–56 The first step of this enzymatic cascade is the activation of a ubiquitin moiety in an adenosine triphosphate (ATP)–consuming reaction catalyzed by an E1 enzyme. The E1 catalyzes the formation of a thioester linkage between a cysteine residue in the E1 and the ubiquitin C-terminal carboxyl group.56–59 Next, the ubiquitin-loaded E1 associates with an E2 and, in a trans-thiolation reaction, transfers the activated ubiquitin to the active site cysteine of the E2.57,58 The final step, attachment of the ubiquitin moiety to the target protein, involves an E3 ubiquitin-protein ligase that recognizes, directly or indirectly via adaptor proteins, both the target protein and the ubiquitin-loaded E2 and catalyzes the transfer of the ubiquitin moiety from the E2 to the target, creating an isopeptide bond between the C-terminal carboxyl group of ubiquitin and the ε-amino group in a lysine residue in the target protein.57,60 Sequential cycles of the ubiquitylation cascade add additional ubiquitin moieties to a lysine residue in ubiquitin, thus generating a polyubiquitin chain on the target protein. Polyubiquitylated proteins are recognized by the 26S proteasome, unfolded, and degraded while the ubiquitin chain is cleaved off by proteasome-associated deubiquitylating enzymes (DUBs) and thus recycled.60–63 The E3s play a key role for the substrate specificity of the UPS, while the E2 enzymes, in most cases, determine the length and the topology of the ubiquitin chain that is formed, and together, E2s and E3s make the UPS remarkable, diverse, and specific.59,64,65 The sheer number of E3s encoded in eukaryotic genomes emphasizes their importance for substrate specificity: the human genome encodes two E1s, approximately 40 E2s, but more than 600 E3s,58,66–68 while yeast contains one E1, approximately 15 E2s, and about 100 E3s.53,64,69 Obviously, most of the E3s play a role in the regulated turnover of folded and functional proteins, but several E3s have been connected with the PQC system. For instance, the mammalian E3, called carboxy terminus of Hsp70-interacting protein (CHIP) or STIP1 homology and U box-containing protein (STUB1), has been shown to bind directly to HSP70 and HSP90 and catalyze the ubiquitylation of chaperone-bound proteins.70 Certain transmembrane E3s associate with the ER membrane to target misfolded secretory proteins for proteasomal degradation via the so-called endoplasmic reticulum–associated degradation (ERAD) pathway.71,72 Similar to the situation with CHIP, the substrate recognition in this ERAD system also relies on chaperones such as the ER chaperones binding-immunoglobulin protein (BiP), protein disulfide-isomerase (PDI), and calreticulin.73–75 Not all PQC E3s, however, depend on chaperones for substrate recognition. In the yeast nucleus, the E3 San1 has been shown to operate in a chaperone-independent manner by directly associating with its substrate proteins.76 Although there is presently no known human San1 orthologue, multiple human E3s display San1-like characteristics

Protein destabilization and degradation as a mechanism for hereditary disease

and may, in a similar manner, directly associate with destabilized or misfolded targets.77 Finally, a number of E3s appear to have overlapping substrate specificity, resulting in a significant cross talk between different degradation pathways.78 Accordingly, a distinct combination of mutually redundant PQC E3s and molecular chaperones were recently found to target cytosolic and nuclear proteins in yeast cells.41 The observed redundancy among the different PQC components is likely more pronounced in mammals, and it demonstrates the importance of the system on both a cellular and organismal level. However, it presently creates challenges when attempting to map degradation pathways and elucidating the roles of individual PQC components. If a misfolded protein can be targeted for degradation by multiple E3s, inhibiting or knocking out a single one may not cause any observable effect on target protein half-life, as other E3s will just take over. Perhaps this could be solved by constructing E3 double knockout cells as done in yeast,41 but presently there is little evidence to exclude the possibility that a handful or even more E3s could target the same protein. Identifying E3s for a known substrate by mass spectrometry–based approaches is also not trivial. Substrate-E3 interactions are weak and transient, and in combination with the (assumed) redundancy, this makes them challenging to capture. Thus, the number of proteins known to be degraded via PQC substantially outnumbers those where we know which specific components (E3s, chaperones, etc.) are directly responsible for substrate recognition.

Protein quality control degrons For most proteins, E3s or their adaptor proteins recognize so-called degradation signals, degrons, that are local structural or sequence features in the target proteins.79 Most of the known degrons are short, linear peptide motifs that contain specific sequence or structural patterns recognizable to their cognate E3s (Fig. 6.2). In addition, a substrate lysine residue for ubiquitylation must be proximal to the E3 but does not have to be part of the degron sequence per se (Fig. 6.2).80,81 Although the PQC E3s are clearly able to distinguish a substrate protein from a nonsubstrate protein, the exact nature of the structural or sequence features that they recognize is, despite substantial recent efforts,76,79,81–83 still rather poorly defined. The PQC degrons are, however, likely hydrophobic regions81 that are normally buried within the core of native proteins but exposed upon a misfolding event.79,84 For example, in the case of the yeast Mat2α transcription factor, mapping studies have revealed a 19-residue degron segment that forms an amphipathic helix85 that is recognized by the PQC E3 Doa10, and since the effect of this degron is lost when hydrophobic residues are mutated, exposed hydrophobicity appears to play an essential role at least for this PQC degron.79,85 Similarly, Sir antagonist 1 (San1) interacts with its substrates via repeated short hydrophobic regions76 and primarily target hydrophobic proteins.81,86 Thus, PQC E3s and molecular chaperones,87,88 at least to some extent, display an overlapping preference for hydrophobic substrates.

117

118

Protein Homeostasis Diseases

Figure 6.2  Degrons and recognition of protein quality control (PQC) targets (color online). A stable and folded protein (substrate) may become destabilized or structurally perturbed during stress conditions or as a result of mutations. In turn, this may lead to the exposure of a degron (floppy tail), which is recognized by a PQC E3 (blue), which, along with its cognate E2 (red), catalyzes the ubiquitylation of a lysine residue (black stick) in the PQC substrate. Once ubiquitylated, the substrate protein is degraded by the 26S proteasome (trashcan).

Local versus global unfolding Since degrons determine how efficiently proteins are recognized by E3s, ultimately the protein turnover must depend on both the specific nature of the degron and the extent of its exposure. One model assumes that degrons are shielded in well-folded proteins but exposed and recognized upon unfolding. Accordingly, a key question is therefore how much structural destabilization is tolerated before degrons are sufficiently exposed. According to a two-state equilibrium model, a protein’s structural stability in vitro is defined by the free energy required to completely unfold the native state. However, although certain point mutations may cause a dramatic destabilization of the encoded protein, such global unfolding events are generally unlikely to occur in cells. In contrast, the possible spectrum of locally unfolded conformations may be large and involve smaller free energy differences between the native, folded state and the partially unfolded conformation (see Chapters 1 and 2). Thus, many disease-linked destabilized protein variants likely display an increased population of such partially unfolded conformations, and the PQC system must therefore be sufficiently sensitive to detect even transiently and partially unfolded structures. In agreement with this, we recently found that the degree of structural destabilization correlates with the turnover rate,26,33 so the more dramatic a structural destabilization is, the more efficiently the protein is degraded. However, for both the Lynch syndrome-linked MutS protein homolog (MSH2) protein and the phenylketonuria-linked PAH protein, a structural destabilization of as little as ∼3 kcal/ mol is sufficient to trigger accelerated degradation,26,33 and a comparable number was also previously found for disease-causing variants in TP53.34 Although this figure is likely to vary from protein to protein, depending on how tightly folded the wild-type protein is and the nature of the (presumably transiently) exposed degrons, a 3-kcal/mol destabilization is certainly not dramatic and unlikely to lead to global unfolding. It is, however,

Protein destabilization and degradation as a mechanism for hereditary disease

in agreement with a partial-unfolding model and also in line with genetic studies in yeast that have shown that the PQC system is highly diligent and inclined to target proteins that are only slightly structurally destabilized and remain functional.32,49,89,90 Finally, it supports that a large fraction of disease-linked missense variants operate by causing subtle structural destabilization, which, in turn, leads to PQC-mediated protein clearance and insufficient protein levels.

Potential therapeutic approaches to protein quality control–linked hereditary diseases Although genome editing techniques may eventually allow us to exchange a diseaselinked genetic variant with a nonpathogenic wild-type copy, this is presently not possible and also a path wreathed in a number of serious ethical concerns. However, the nature of the PQC-mediated clearance of structurally destabilized proteins potentially allows for other pharmaceutical approaches to mitigate the effects of those pathogenic variants in which the main cause is loss of stability. Importantly, since the PQC system relies on detecting changes in protein structure and stability rather than testing cellular proteins for function, some slightly destabilized and still functional protein variants are degraded. In genetics, examples of this have been known for years, where hypomorph alleles (gene variants displaying reduced function) can be suppressed, for example, by their overexpression or by blocking the degradation system. Indeed, genetic screens in yeast cells have utilized this to identify the E2s, E3s, chaperones, and proteases involved in the PQC system.32,49,89–92 Thus, in principle, a similar outcome could be achieved using small-molecule inhibitors of the degradative PQC system7,93 (see Chapter 18). The disadvantage of this approach is that such drugs may skew the finely balanced PQC system and might therefore cause toxicity. However, the advantage of this approach is that it could stabilize a number of different but collectively destabilized proteins and thus be broadly applicable for several unrelated hereditary diseases where missense variants cause loss of function through a subtle protein destabilization. The histone deacetylase inhibitors (HDACis) represent a promising class of drugs that operate in this manner. By inducing HSP90 acetylation, thereby preventing HSP90 substrate binding, the HDACis inhibit chaperone function and have been shown to display beneficial effects in correcting the cellular phenotypes of a number of conformational diseases.94–99 The heat shock– amplifying molecule arimoclomol works by increasing cellular amounts of HSP70, and has, in preclinical studies, demonstrated beneficial effects in various lysosomal storage diseases100 as well as in amyotrophic lateral sclerosis (ALS) models.101 Rather than blocking the degradation system or PQC network, an alternative approach could be to stabilize the misfolded protein variant, for example, through small molecules that bind the native form of the protein variant. In general, such small molecules could constitute cofactors, substrates, inhibitors, or more specialized compounds that associate with the protein. For instance, in the case of phenylketonuria, which is

119

120

Protein Homeostasis Diseases

linked to variants in the phenylalanine hydroxylase (PAH) enzyme, the PAH cosubstrate tetrahydrobiopterin has been shown to stabilize certain PAH variants26,102,103 and alleviate the symptoms in some affected individuals.104 In cystic fibrosis, it has been demonstrated that the common disease-linked CFTR F508∆ protein variant is not inherently inactive but is nonetheless targeted for proteasomal degradation.35 Consequently, the cells end up with insufficient amounts of the protein, and the disease can therefore be averted or alleviated by using small molecules to stabilize the CFTR F508∆ protein.105,106 More recently, allosteric inhibitors of the mucosa-associated lymphoid tissue lymphoma translocation protein (MALT1) paracaspase were shown to stabilize a disease-linked MALT1 variant and partly restore MALT1 substrate cleavage following a washout of the compound.107 Finally, cancer-linked genetic variants of p53 provide another example of this. Here, studies have identified small-molecule reactivators of p53 that function by structurally stabilizing certain p53 missense variants,108,109 while others block p53 ubiquitylation and degradation by inhibiting the E3 ubiquitin-protein ligase mouse double minute (MDM2).110,111 Similarly, the tumor suppressor capicua (CIC) was recently discovered to be unstable in sporadic glioblastomas, and the study demonstrates effects on survival when stabilizing CIC in mouse glioblastoma models.112 Here, the stabilization is accomplished by RNAi against the CIC targeting E3, and while this approach is not clinically applicable, it demonstrates the therapeutic potential of stabilizing a specific protein. Lastly, a potential approach for targeting the group of dominant gain-of-function disease proteins could include the so-called proteolysis-targeting chimeras (PROTACs), small molecules that can drive the interaction between a target protein and an E3 enzyme, leading to target ubiquitylation and degradation.113,114 Progress in this area has recently been reviewed.115

Acknowledgments The authors thank all members of the Linderstrøm-Lang Centre for Protein Science for helpful discussions and comments on the manuscript.

Conflict of interest No conflicting interests declared.

Funding Our work is financially supported by the Lundbeck Foundation (to A.S., K.L.-L. and R.H.-P.), the Danish Cancer Society (to R.H.-P.), the Novo Nordisk Foundation (to A.S., K.L.-L., and R.H.-P.), the A.P. Møller Foundation (to R.H.-P.), Aase and Ejnar Danielsens Foundation (to R.H.-P.), the Danish Rheumatism Association (to R.H.-P.),

Protein destabilization and degradation as a mechanism for hereditary disease

Foundation Jochum (to R.H.-P.), and the Danish Council for Independent Research (Natural Sciences) (to R.H.-P.).

References 1. Anfinsen CB. Principles that govern the folding of protein chains. Science 1973;181(4096):223–30. 2. Matreyek KA, Starita LM, Stephany JJ, et al. Multiplex assessment of protein variant abundance by massively parallel sequencing. Nat Genet 2018;50(6):874–82. 3. Tokuriki N, Stricher F, Schymkowitz J, Serrano L, Tawfik DS. The stability effects of protein mutations appear to be universally distributed. J Mol Biol 2007;369(5):1318–32. 4. Drummond DA. Protein evolution: innovative chaps. Curr Biol 2009;19(17):R740–2. 5. Hartl FU, Bracher A, Hayer-Hartl M. Molecular chaperones in protein folding and proteostasis. Nature 2011;475(7356):324–32. 6. Clausen L, Abildgaard AB, Gersing SK, Stein A, Lindorff-Larsen K, Hartmann-Petersen R. Protein stability and degradation in health and disease. Adv Protein Chem Struct Biol 2019;114:61–83. 7. Stein A, Fowler DM, Hartmann-Petersen R, Lindorff-Larsen K. Biophysical and mechanistic models for disease-causing protein variants. Trends Biochem Sci 2019;44(7):575–88. 8. Balchin D, Hayer-Hartl M, Hartl FU. In vivo aspects of protein folding and quality control. Science 2016;353(6294):aac4354. 9. Hartl FU. Protein misfolding diseases. Annu Rev Biochem 2017;86:21–6. 10. Enam C, Geffen Y, Ravid T, Gardner RG. Protein quality control degradation in the nucleus. Annu Rev Biochem 2018;87:725–49. 11. Pilla E, Schneider K, Bertolotti A. Coping with protein quality control failure. Annu Rev Cell Dev Biol 2017;33:439–65. 12. Kramer G, Shiber A, Bukau B. Mechanisms of cotranslational maturation of newly synthesized proteins. Annu Rev Biochem 2019;88:337–64. 13. Cabrita LD, Dobson CM, Christodoulou J. Protein folding on the ribosome. Curr Opin Struct Biol 2010;20(1):33–45. 14. Kaiser CM, Liu K. Folding up and moving on-nascent protein folding on the ribosome. J Mol Biol 2018;430(22):4580–91. 15. Shao S, Hegde RS. Target selection during protein quality control. Trends Biochem Sci 2016;41(2): 124–37. 16. Parakh S, Atkin JD. Protein folding alterations in amyotrophic lateral sclerosis. Brain Res 2016;1648 (Pt B):633–49. 17. Yang H, Li JJ, Liu S, et al. Aggregation of polyglutamine-expanded ataxin-3 sequesters its specific interacting partners into inclusions: implication in a loss-of-function pathology. Sci Rep 2014;4:6410. 18. Olzscha H, Schermann SM,Woerner AC, et al. Amyloid-like aggregates sequester numerous metastable proteins with essential cellular functions. Cell 2011;144(1):67–78. 19. Chen S, Ferrone FA,Wetzel R. Huntington’s disease age-of-onset linked to polyglutamine aggregation nucleation. Proc Natl Acad Sci USA 2002;99(18):11884–9. 20. Orr HT, Zoghbi HY. Trinucleotide repeat disorders. Annu Rev Neurosci 2007;30:575–621. 21. Chiti F, Dobson CM. Protein misfolding, functional amyloid, and human disease. Annu Rev Biochem 2006;75:333–66. 22. Hipp MS, Park SH, Hartl FU. Proteostasis impairment in protein-misfolding and -aggregation diseases. Trends Cell Biol 2014;24(9):506–14. 23. Powers ET, Morimoto RI, Dillin A, Kelly JW, Balch WE. Biological and chemical approaches to diseases of proteostasis deficiency. Annu Rev Biochem 2009;78:959–91. 24. Josefson R, Andersson R, Nystrom T. How and why do toxic conformers of aberrant proteins accumulate during ageing? Essays Biochem 2017;61(3):317–24. 25. Yue P, Li Z, Moult J. Loss of protein structure stability as a major causative factor in monogenic disease. J Mol Biol 2005;353(2):459–73. 26. Scheller R, Stein A, Nielsen SV, et al.Toward mechanistic models for genotype-phenotype correlations in phenylketonuria using protein stability calculations. Hum Mutat 2019;40(4):444–57.

121

122

Protein Homeostasis Diseases

27. Pey AL, Stricher F, Serrano L, Martinez A. Predicted effects of missense mutations on native-state stability account for phenotypic outcome in phenylketonuria, a paradigm of misfolding diseases. Am J Hum Genet 2007;81(5):1006–24. 28. Mathiassen SG, Larsen IB, Poulsen EG, et al. A two-step protein quality control pathway for a misfolded DJ-1 variant in fission yeast. J Biol Chem 2015;290(34):21141–53. 29. Henn IH, Gostner JM, Lackner P, Tatzelt J, Winklhofer KF. Pathogenic mutations inactivate parkin by distinct mechanisms. J Neurochem 2005;92(1):114–22. 30. Suri M, Evers JMG, Laskowski RA, et al. Protein structure and phenotypic analysis of pathogenic and population missense variants in STXBP1. Mol Genet Genomic Med 2017;5(5):495–507. 31. Knauth K, Cartwright E, Freund S, Bycroft M, Buchberger A. VHL mutations linked to type 2C von Hippel-Lindau disease cause extensive structural perturbations in pVHL. J Biol Chem 2009;284(16):10514–22. 32. Arlow T, Scott K, Wagenseller A, Gammie A. Proteasome inhibition rescues clinically significant unstable variants of the mismatch repair protein Msh2. Proc Natl Acad Sci USA 2013;110(1):246–51. 33. Nielsen SV, Stein A, Dinitzen AB, et al. Predicting the impact of Lynch syndrome-causing missense mutations from structural calculations. PLoS Genet 2017;13(4):e1006739. 34. Bullock AN, Henckel J, Fersht AR. Quantitative analysis of residual folding and DNA binding in mutant p53 core domain: definition of mutant states for rescue in cancer therapy. Oncogene 2000;19(10):1245–56. 35. Ahner A, Nakatsukasa K, Zhang H, Frizzell RA, Brodsky JL. Small heat-shock proteins select deltaF508-CFTR for endoplasmic reticulum-associated degradation. Mol Biol Cell 2007;18(3):806–14. 36. Sontag EM,Vonk WI, Frydman J. Sorting out the trash: the spatial nature of eukaryotic protein quality control. Curr Opin Cell Biol 2014;26C:139–46. 37. Pechmann S, Willmund F, Frydman J. The ribosome as a hub for protein quality control. Mol Cell 2013;49(3):411–21. 38. Lee J, Kim JH, Biter AB, Sielaff B, Lee S, Tsai FT. Heat shock protein (Hsp) 70 is an activator of the Hsp104 motor. Proc Natl Acad Sci U S A 2013;110(21):8513–8. 39. Ben-Zvi AP, Goloubinoff P. Review: mechanisms of disaggregation and refolding of stable protein aggregates by molecular chaperones. J Struct Biol 2001;135(2):84–93. 40. Voisine C, Pedersen JS, Morimoto RI. Chaperone networks: tipping the balance in protein folding diseases. Neurobiol Dis 2010;40(1):12–20. 41. Samant RS, Livingston CM, Sontag EM, Frydman J. Distinct proteostasis circuits cooperate in nuclear and cytoplasmic protein quality control. Nature 2018;563(7731):407–11. 42. Kandasamy G, Andreasson C. Hsp70-Hsp110 chaperones deliver ubiquitin-dependent and -independent substrates to the 26S proteasome for proteolysis in yeast. J Cell Sci 2018;131(6) pii: jcs210948. 43. Gowda NK, Kandasamy G, Froehlich MS, Dohmen RJ, Andreasson C. Hsp70 nucleotide exchange factor Fes1 is essential for ubiquitin-dependent degradation of misfolded cytosolic proteins. Proc Natl Acad Sci USA 2013;110(15):5975–80. 44. Kettern N, Dreiseidler M, Tawo R, Hohfeld J. Chaperone-assisted degradation: multiple paths to destruction. Biol Chem 2010;391(5):481–9. 45. Kriegenburg F, Ellgaard L, Hartmann-Petersen R. Molecular chaperones in targeting misfolded proteins for ubiquitin-dependent degradation. FEBS J 2012;279(4):532–42. 46. Demand J, Alberti S, Patterson C, Hohfeld J. Cooperation of a ubiquitin domain protein and an E3 ubiquitin ligase during chaperone/proteasome coupling. Curr Biol 2001;11(20):1569–77. 47. Alberti S, Demand J, Esser C, Emmerich N, Schild H, Hohfeld J. Ubiquitylation of BAG-1 suggests a novel regulatory mechanism during the sorting of chaperone substrates to the proteasome. J Biol Chem 2002;277(48):45920–7. 48. Arndt V, Daniel C, Nastainczyk W, Alberti S, Hohfeld J. BAG-2 acts as an inhibitor of the chaperoneassociated ubiquitin ligase CHIP. Mol Biol Cell 2005;16(12):5891–900. 49. Kriegenburg F, Jakopec V, Poulsen EG, et al. A chaperone-assisted degradation pathway targets kinetochore proteins to ensure genome stability. PLoS Genet 2014;10(1):e1004140. 50. Ciechanover A, Kwon YT. Protein quality control by molecular chaperones in neurodegeneration. Front Neurosci 2017;11:185. 51. Dikic I, Elazar Z. Mechanism and medical implications of mammalian autophagy. Nat Rev Mol Cell Biol 2018;19(6):349–64.

Protein destabilization and degradation as a mechanism for hereditary disease

52. Kriegenburg F, Ungermann C, Reggiori F. Coordination of autophagosome-lysosome fusion by Atg8 family members. Curr Biol 2018;28(8):R512–8. 53. Glickman MH, Ciechanover A. The ubiquitin-proteasome proteolytic pathway: destruction for the sake of construction. Physiol Rev 2002;82(2):373–428. 54. Hershko A, Ciechanover A. The ubiquitin system. Annu Rev Biochem 1998;67:425–79. 55. Kleiger G, Mayor T. Perilous journey: a tour of the ubiquitin-proteasome system. Trends Cell Biol 2014;24(6):352–9. 56. Pickart CM. Mechanisms underlying ubiquitination. Annu Rev Biochem 2001;70:503–33. 57. Dye BT, Schulman BA. Structural mechanisms underlying posttranslational modification by ubiquitinlike proteins. Annu Rev Biophys Biomol Struct 2007;36:131–50. 58. Komander D. The emerging complexity of protein ubiquitination. Biochem Soc Trans 2009;37 (Pt 5):937–53. 59. Pickart CM, Eddins MJ. Ubiquitin: structures, functions, mechanisms. Biochim Biophys Acta 2004;1695(1–3):55–72. 60. Pickart CM, Fushman D. Polyubiquitin chains: polymeric protein signals. Curr Opin Chem Biol 2004;8(6):610–6. 61. Lee MJ, Lee BH, Hanna J, King RW, Finley D.Trimming of ubiquitin chains by proteasome-associated deubiquitinating enzymes. Mol Cell Proteomics 2011;10(5):R110. 62. Reyes-Turcu FE,Ventii KH,Wilkinson KD. Regulation and cellular roles of ubiquitin-specific deubiquitinating enzymes. Annu Rev Biochem 2009;78:363–97. 63. Thrower JS, Hoffman L, Rechsteiner M, Pickart CM. Recognition of the polyubiquitin proteolytic signal. EMBO J 2000;19(1):94–102. 64. Ye Y, Rape M. Building ubiquitin chains: E2 enzymes at work. Nat Rev Mol Cell Biol 2009;10(11): 755–64. 65. Komander D, Rape M. The ubiquitin code. Annu Rev Biochem 2012;81:203–29. 66. Buetow L, Huang DT. Structural insights into the catalysis and regulation of E3 ubiquitin ligases. Nat Rev Mol Cell Biol 2016;17(10):626–42. 67. Semple CA. The comparative proteomics of ubiquitination in mouse. Genome Res 2003;13(6B): 1389–94. 68. Deshaies RJ, Joazeiro CA. RING domain E3 ubiquitin ligases. Annu Rev Biochem 2009;78:399–434. 69. Hanna J, Waterman D, Boselli M, Finley D. Spg5 protein regulates the proteasome in quiescence. J Biol Chem 2012;287(41):34400–9. 70. Arndt V, Rogon C, Hohfeld J. To be, or not to be--molecular chaperones in protein degradation. Cell Mol Life Sci 2007;64(19-20):2525–41. 71. Mehrtash AB, Hochstrasser M. Ubiquitin-dependent protein degradation at the endoplasmic reticulum and nuclear envelope. Semin Cell Dev Biol 2018; pii: S1084-9521(18)30067-3. 72. Wu X, Rapoport TA. Mechanistic insights into ER-associated protein degradation. Curr Opin Cell Biol 2018;53:22–8. 73. McCaffrey K, Braakman I. Protein quality control at the endoplasmic reticulum. Essays Biochem 2016;60(2):227–35. 74. Needham PG, Guerriero CJ, Brodsky JL. Chaperoning endoplasmic reticulum-associated degradation (ERAD) and protein conformational diseases. Cold Spring Harb Perspect Biol 2019;11(8) pii: a033928. 75. Pobre KFR, Poet GJ, Hendershot LM. The endoplasmic reticulum (ER) chaperone BiP is a master regulator of ER functions: getting by with a little help from ERdj friends. J Biol Chem 2019;294(6): 2098–108. 76. Rosenbaum JC, Fredrickson EK, Oeser ML, et al. Disorder targets misorder in nuclear quality control degradation: a disordered ubiquitin ligase directly recognizes its misfolded substrates. Mol Cell 2011;41(1):93–106. 77. Boomsma W, Nielsen SV, Lindorff-Larsen K, Hartmann-Petersen R, Ellgaard L. Bioinformatics analysis identifies several intrinsically disordered human E3 ubiquitin-protein ligases. PeerJ 2016;4:e1725. 78. Nillegoda NB, Theodoraki MA, Mandal AK, et al. Ubr1 and Ubr2 function in a quality control pathway for degradation of unfolded cytosolic proteins. Mol Biol Cell 2010;21(13):2102–16. 79. Ravid T, Hochstrasser M. Diversity of degradation signals in the ubiquitin-proteasome system. Nat Rev Mol Cell Biol 2008;9(9):679–90.

123

124

Protein Homeostasis Diseases

80. Guharoy M, Bhowmick P, Tompa P. Design principles involving protein disorder facilitate specific substrate selection and degradation by the ubiquitin-proteasome system. J Biol Chem 2016;291(13): 6723–31. 81. Fredrickson EK, Rosenbaum JC, Locke MN, Milac TI, Gardner RG. Exposed hydrophobicity is a key determinant of nuclear quality control degradation. Mol Biol Cell 2011;22:2384–95. 82. Maurer MJ, Spear ED, Yu AT, Lee EJ, Shahzad S, Michaelis S. Degradation signals for ubiquitin-proteasome dependent cytosolic protein quality control (CytoQC) in yeast. G3 (Bethesda ) 2016;6(7): 1853–66. 83. Geffen Y, Appleboim A, Gardner RG, Friedman N, Sadeh R, Ravid T. Mapping the landscape of a eukaryotic degronome. Mol Cell 2016;63(6):1055–65. 84. Ravid T, Kreft SG, Hochstrasser M. Membrane and soluble substrates of the Doa10 ubiquitin ligase are degraded by distinct pathways. EMBO J 2006;25(3):533–43. 85. Johnson PR, Swanson R, Rakhilina L, Hochstrasser M. Degradation signal masking by heterodimerization of MATalpha2 and MATa1 blocks their mutual destruction by the ubiquitin-proteasome pathway. Cell 1998;94(2):217–27. 86. Fredrickson EK, Clowes Candadai SV, Tam CH, Gardner RG. Means of self-preservation: how an intrinsically disordered ubiquitin-protein ligase averts self-destruction. Mol Biol Cell 2013;24(7): 1041–52. 87. Rudiger S, Germeroth L, Schneider-Mergener J, Bukau B. Substrate specificity of the DnaK chaperone determined by screening cellulose-bound peptide libraries. EMBO J 1997;16(7):1501–7. 88. Rudiger S, Buchberger A, Bukau B. Interaction of Hsp70 chaperones with substrates. Nat Struct Biol 1997;4(5):342–9. 89. Gardner RG, Nelson ZW, Gottschling DE. Degradation-mediated protein quality control in the nucleus. Cell 2005;120(6):803–15. 90. Kampmeyer C, Karakostova A, Schenstrom SM, et al. The exocyst subunit Sec3 is regulated by a protein quality control pathway. J Biol Chem 2017;292(37):15240–53. 91. Buschhorn BA, Kostova Z, Medicherla B,Wolf DH. A genome-wide screen identifies Yos9p as essential for ER-associated degradation of glycoproteins. FEBS Lett 2004;577(3):422–6. 92. Schwickart M, Huang X, Lill JR, et al. Deubiquitinase USP9X stabilizes MCL1 and promotes tumour cell survival. Nature 2010;463(7277):103–7. 93. Kampmeyer C, Nielsen SV, Clausen L, et al. Blocking protein quality control to counter hereditary cancers. Genes Chromosomes Cancer 2017;56(12):823–31. 94. Pankow S, Bamberger C, Calzolari D, et al. F508 CFTR interactome remodelling promotes rescue of cystic fibrosis. Nature 2015;528(7583):510–6. 95. Hutt DM, Herman D, Rodrigues AP, et al. Reduced histone deacetylase 7 activity restores function to misfolded CFTR in cystic fibrosis. Nat Chem Biol 2010;6(1):25–33. 96. Munkacsi AB, Chen FW, Brinkman MA, et al. An “exacerbate-reverse” strategy in yeast identifies histone deacetylase inhibition as a correction for cholesterol and sphingolipid transport defects in human Niemann-Pick type C disease. J Biol Chem 2011;286(27):23842–51. 97. Pipalia NH, Cosner CC, Huang A, et al. Histone deacetylase inhibitor treatment dramatically reduces cholesterol accumulation in Niemann-Pick type C1 mutant human fibroblasts. Proc Natl Acad Sci USA 2011;108(14):5620–5. 98. Yang C, Huntoon K, Ksendzovsky A, Zhuang Z, Lonser RR. Proteostasis modulators prolong missense VHL protein activity and halt tumor progression. Cell Rep 2013;3(1):52–9. 99. Coppede F.The potential of epigenetic therapies in neurodegenerative diseases. Front Genet 2014;5:220. 100. Kirkegaard T, Gray J, Priestman DA, et al. Heat shock protein-based therapy as a potential candidate for treating the sphingolipidoses. Sci Transl Med 2016;8(355) 355ra118. 101. Kalmar B, Lu CH, Greensmith L. The role of heat shock proteins in amyotrophic lateral sclerosis: the therapeutic potential of Arimoclomol. Pharmacol Ther 2014;141(1):40–54. 102. Erlandsen H, Pey AL, Gamez A, et al. Correction of kinetic and stability defects by tetrahydrobiopterin in phenylketonuria patients with certain phenylalanine hydroxylase mutations. Proc Natl Acad Sci USA 2004;101(48):16903–8.

Protein destabilization and degradation as a mechanism for hereditary disease

103. Muntau AC, Leandro J, Staudigl M, Mayer F, Gersting SW. Innovative strategies to treat protein misfolding in inborn errors of metabolism: pharmacological chaperones and proteostasis regulators. J Inherit Metab Dis 2014;37(4):505–23. 104. Blau N. Genetics of phenylketonuria: then and now. Hum Mutat 2016;37(6):508–15. 105. Arora K, Naren AP. Pharmacological correction of cystic fibrosis: molecular mechanisms at the plasma membrane to augment mutant CFTR function. Curr Drug Targets 2016;17(11):1275–81. 106. Van GF, Hadida S, Grootenhuis PD, et al. Correction of the F508del-CFTR protein processing defect in vitro by the investigational drug VX-809. Proc Natl Acad Sci U S A 2011;108(46):18843–8. 107. Quancard J, Klein T, Fung SY, et al. An allosteric MALT1 inhibitor is a molecular corrector rescuing function in an immunodeficient patient. Nat Chem Biol 2019;15(3):304–13. 108. Basse N, Kaar JL, Settanni G, Joerger AC, Rutherford TJ, Fersht AR. Toward the rational design of p53-stabilizing drugs: probing the surface of the oncogenic Y220C mutant. Chem Biol 2010;17(1): 46–56. 109. Liu X,Wilcken R, Joerger AC, et al. Small molecule induced reactivation of mutant p53 in cancer cells. Nucleic Acids Res 2013;41(12):6034–44. 110. Brown CJ, Lain S,Verma CS, Fersht AR, Lane DP. Awakening guardian angels: drugging the p53 pathway. Nat Rev Cancer 2009;9(12):862–73. 111. Zhao Y, Aguilar A, Bernard D, Wang S. Small-molecule inhibitors of the MDM2-p53 protein-protein interaction (MDM2 Inhibitors) in clinical trials for cancer treatment. J Med Chem 2015;58(3):1038–52. 112. Bunda S, Heir P, Metcalf J, et al. CIC protein instability contributes to tumorigenesis in glioblastoma. Nat Commun 2019;10(1):661. 113. Buckley DL,Van MI, Gareiss PC, et al.Targeting the von Hippel-Lindau E3 ubiquitin ligase using small molecules to disrupt the VHL/HIF-1alpha interaction. J Am Chem Soc 2012;134(10):4465–8. 114. Winter GE, Buckley DL, Paulk J, et al. Drug development. Phthalimide conjugation as a strategy for in vivo target protein degradation. Science 2015;348(6241):1376–81. 115. Lai AC, Crews CM. Induced protein degradation: an emerging drug discovery paradigm. Nat Rev Drug Discov 2017;16(2):101–14. 116. Olzmann JA, Brown K, Wilkinson KD, et al. Familial Parkinson’s disease-associated L166P mutation disrupts DJ-1 protein folding and function. J Biol Chem 2004;279(9):8506–15. 117. Fiesel FC, Caulfield TR, Moussaud-Lamodiere EL, et al. Structural and functional impact of parkinson disease-associated mutations in the E3 ubiquitin ligase parkin. Hum Mutat 2015;36(8):774–86. 118. Banka S, Blom HJ, Walter J, et al. Identification and characterization of an inborn error of metabolism caused by dihydrofolate reductase deficiency. Am J Hum Genet 2011;88(2):216–25. 119. Cario H, Smith DE, Blom H, et al. Dihydrofolate reductase deficiency due to a homozygous DHFR mutation causes megaloblastic anemia and cerebral folate deficiency leading to severe neurologic disease. Am J Hum Genet 2011;88(2):226–31. 120. Nahorski MS, Reiman A, Lim DH, et al. Birt Hogg-Dube syndrome-associated FLCN mutations disrupt protein stability. Hum Mutat 2011;32(8):921–9. 121. Fu R, Jinnah HA. Genotype-phenotype correlations in Lesch-Nyhan disease: moving beyond the gene. J Biol Chem 2012;287(5):2997–3008. 122. Ron I, Horowitz M. ER retention and degradation as the molecular basis underlying Gaucher disease heterogeneity. Hum Mol Genet 2005;14(16):2387–98. 123. Yang C, Asthagiri AR, Iyer RR, et al. Missense mutations in the NF2 gene result in the quantitative loss of merlin protein and minimally affect protein intrinsic function. Proc Natl Acad Sci USA 2011;108(12):4980–5. 124. Damgaard RB, Elliott PR, Swatek KN, et al. OTULIN deficiency in ORAS causes cell type-specific LUBAC degradation, dysregulated TNF signalling and cell death. EMBO Mol Med 2019;11(3) pii: e9324. 125. Volk AE, Weishaupt JH, Andersen PM, Ludolph AC, Kubisch C. Current knowledge and recent insights into the genetic basis of amyotrophic lateral sclerosis. Med Genet 2018;30(2):252–8. 126. Rosen DR, Siddique T, Patterson D, et al. Mutations in Cu/Zn superoxide dismutase gene are associated with familial amyotrophic lateral sclerosis. Nature 1993;362(6415):59–62.

125

CHAPTER 7

Detection of amyloid aggregation in living systems Kerensa Broersen Applied Stem Cell Technologies, Faculty of Science and Technology, University of Twente, Enschede, The Netherlands

Outline Introduction Techniques for detection of amyloid aggregation in vivo Förster resonance energy transfer detection and fluorescence lifetime imaging Bioluminescence imaging Optical fiber bundles and fluorescence imaging Cranial window or thinned-skull imaging using multiphoton microscopy In vivo microdialysis Positron emission tomography imaging Animal models to test in vivo amyloid formation Caenorhabditis elegans Zebrafish Mouse models Note on using animal models for in vivo protein aggregation studies In vivo complexity and how do in vivo detection assays provide insight into peripheral aspects contributing to neurodegeneration? Replicating the multicellular complexity of the brain Interaction of the brain with the periphery system Future outlook References

128 129 131 132 133 134 135 136 137 137 138 139 140 141 141 143 144 145

Abbreviations AAV  adeno-associated virus Aβ  amyloid-β ALS  amyotrophic lateral sclerosis APP  amyloid precursor protein CAA  cerebral amyloid angiopathy CFP  cyan fluorescent protein C. elegans  Caenorhabditis elegans ChR2  channelrhodopsin-2 CRISPR  clustered regularly interspaced short palindromic repeats DLB  dementia with Lewy bodies d.p.f.  days post fertilization FRET  Förster resonance energy transfer Gfap  glial fibrillary acidic protein Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00007-5 Copyright © 2020 Elsevier Inc. All rights reserved.

127

128

Protein Homeostasis Diseases

GFP  green fluorescent protein hESC  human embryonic stem cells IL-1β  interleukin-1β IPSC  induced pluripotent stem cells ISF  interstitial fluid KIN-19  casein kinase I isoform alpha LRRK2  leucine-rich repeat kinase 2 PET  positron emission tomography PiB  11C-Pittsburgh compound B PolyQ  polyglutamine PS  presenilin PTU  1-phenyl 2-thiourea RHO-1  Ras-like GTP-binding protein rhoA SOD  superoxide dismutase TEM  transmission electron microscopy TCSPC  time-correlated single photon counting UCH-L1  ubiquitin carboxy-terminal hydrolase L1 YFP  yellow fluorescent protein

Introduction Recent developments into in vivo imaging techniques provide direct insight into various biochemical processes in living systems. For example, it is nowadays possible to obtain high spatial and temporary resolution insight into brain processes such as glutamate neurotransmission1 or glucose utilization.2 Significant research effort has gone into in vitro detection of protein aggregates (Box 7.1). Nowadays, it is possible to obtain, with

BOX 7.1  Protein aggregation Protein aggregates are the result of aberrant self-association of protein molecules that, in the complex process of folding, failed to attain a three-dimensional fold that is dictated as the functional structural arrangement. These so-called misfolded proteins often have increased surface hydrophobic exposure leading to their hydrophobic interaction. Resulting aggregates may come in many forms and shapes but are generally regarded as dysfunctional and are sometimes toxic. Protein aggregation may be the result of increased generation of proteins as a result of mutations or decreased clearance pathways. Over the years, many diseases have been identified that contain a protein misfolding and aggregation component or their pathogenic pathways may even be primarily dictated by accumulation and deposition of protein aggregates. Examples of such diseases are Alzheimer’s disease, Parkinson’s disease, and Huntington’s disease. Also, various forms of cancer, diabetes, and some metabolic diseases have been identified to be associated with protein aggregation. How aggregation arises is a question that has tempted many researchers to delve deeper into the molecular mechanisms of the process. Even though great progress has been made, it is clear that the various diseases that are collectively characterized by a protein aggregation component are still highly heterogeneous.

Detection of amyloid aggregation in living systems

reasonable structural detail, insight into in vitro protein aggregation mechanisms by virtue of the wealth of methods available for detection of aggregated protein species. Nevertheless, in organisms, detection of this process proves to be more challenging, as most traditional techniques to study protein aggregation (Chapter 8) are not readily compatible with the optical properties and complexity of living tissue. At the same time, multicellular organism organization and cellular complexity may well add important clues on how environmental factors may affect the behavior of amyloidogenic proteins. Also, as many therapeutic strategies that show great promise in in vitro or in cellular assays fail in clinical trials, it is of utmost importance to understand better what in organism factors may contribute to such unpredictable outcomes. In this chapter, in vivo methods and the observations derived using them will be described to gain access to molecular mechanisms of protein aggregation within a living system. Animal models suitable for such studies will be discussed and placed into perspective of cellular complexity. We will address the question what we have learned by studying protein aggregation in vivo compared to the relative wealth of in vitro data available. How does aggregation occur in a multicellular intact organism? In vivo studies could shed light on initiation and pathological sequence of events, enable identification of previously unknown cellular factors, as well as pinpoint cell types implicated because disease progress can ideally be followed in individual animals throughout the entire process of disease.

Techniques for detection of amyloid aggregation in vivo A range of techniques has been developed over the years that allow direct visualization of in vivo aggregate formation in living animals or humans. Some of these assays require animals to be anesthetized, but others can accurately address protein aggregation related questions in awake animals. Some of these techniques, such as in vivo Förster resonance energy transfer (FRET) detection, require expression of a fluorescent probe to enable visualization of assembly processes, and, as such, these techniques are exclusively available for experimental transgene animal studies. Other methods, such as positron emission tomography (PET) scanning, rely on infusion of radiolabeled tracers that are later removed from the brain by means of the physiologically available clearance mechanisms. This technique therefore also allows for use in diagnostics in humans. Toxicity of some of these tracer molecules, however, limits possibilities to study, for example, aggregation kinetics and dynamics of the assembly process in high-resolution function of time and various parameters. This section will discuss the principles and possibilities of a number of methods described in literature to allow for in vivo visualization of protein aggregates in living whole animals. All these assays come with their own advantages, criteria, and limitations. Table 7.1 provides a summary of the various methods described in this paragraph.

129

130

Protein Homeostasis Diseases

Table 7.1  A summary of in vivo methodologies for detection of protein aggregation, animal models used and limitations of these assays.

Method

Förster resonance energy transfer (FRET) and fluorescence lifetime imaging Bioluminescence imaging

Optical fiber bundles and fluorescence imaging Cranial window imaging using multiphoton microscopy

Animal models reported in amyloid studies

Example references on use of technology for amyloid imaging

Limitations

Caenorhabditis elegans

[5,7]

Requires transparent animals. Signature is not easily distinguished from other autofluorescent species in the cell.

Transgenic mouse models

[18–20]

Resolution is limited by light scattering originating from the tissue. Gene transfer vectors used may instigate an immunological reaction (although alternative methods for gene transfer have been reported). Limited spatial resolution. Surgery required to insert cannula. Imaging requires light or full anesthetization. Requires major surgery, for example, cranial window installation or skull thinning. Both techniques can induce signs of chronic inflammation and changes of neuronal functioning. Radiolabel-specific limitations including retention, specificity, sensitivity, and blood–brain barrier permeability. Restraint of head movement. Radiolabeled markers are required. Invasive, blood–brain barrier permeability reversible impaired, the molecular-weight cutoff of the membrane used limits direct insight into deposition into aggregates.

No publications to date (Transgenic) mouse models

[31,33,36]

Positron emission tomography (PET)

Humans and transgenic mouse models

[59,62,63]

In vivo microdialysis

Mouse models

[38,46]

Detection of amyloid aggregation in living systems

In addition to the advantages and limitations of the imaging techniques that will be discussed, the heterogeneous makeup of the various tissues in the body give rise to their own associated challenges in imaging. Many of the now known neurodegenerative processes take place in the brain, a less-than-transparent, highly complex organ protected from the outside world by an, on average, 7-mm-thick bony structure called the cranium and a layer of skin. The optical characteristics of the different tissues that need to be penetrated to reach the target area may hamper imaging. Often, these challenges are circumvented by using transparent organisms, as will be described in this chapter as well.

Förster resonance energy transfer detection and fluorescence lifetime imaging The transparency of multicellular organisms, such as Caenorhabditis elegans (C. elegans) (see also Chapter 3) and early-development-stage zebrafish, has been exploited to study in vivo amyloid formation related to neurodegenerative disorders. For example, transgenic C. elegans models expressing yellow fluorescent protein (YFP)– or green fluorescent protein (GFP)–fused constructs of Parkinson’s disease–related α-synuclein have been generated to study in-organism aggregation processes.3,4 It was demonstrated that locomotion rate and life span of α-synuclein-expressing strains was considerably decreased. At the same time, FRET intensity increased with age, consistent with the appearance of granular inclusions indicative of aggregation.5 Exploiting the earlier observation that growing aggregates composed of α-synuclein exhibit some intrinsic fluorescence intensity,6 a sensor was developed to detect the assembly of α-synuclein oligomers within C. elegans.7 The sensor was constructed by generation of a fusion protein composed of YFP and α-synuclein, which shows sufficient spectral overlap with the intrinsic fluorescence characteristics of unlabeled α-synuclein aggregates,8 to give rise to a FRET signal. Detection of protein aggregates in this way relies on the fact that FRET can only occur when two fluorescent tags show sufficient spectral overlap and the fluorophores involved are located in close proximity, within approximately 70 Å, from each other,9 as in an aggregated form.The higher-order transparent nematode C. elegans was then used to show that the time-correlated single photon counting fluorescence lifetime imaging approach using YFP-tagged α-synuclein can provide insight into protein aggregation kinetics directly in a YFP-α-synuclein transgene intact organism by observation of a decrease in α-synuclein-YFP lifetime upon aggregation. Spectral characteristics could be assigned to specific species and aggregate sizes en route to amyloid fibril formation, including oligomeric and prefibrillar species as observed in ex vitro immune-TEM analysis. A later publication showed that a similar approach, coinjecting a C-terminal-labeled α-synuclein using cyan fluorescent protein and YFP, could also be used to quantify α-synuclein aggregation in C. elegans using FRET.5 Similar results were observed upon fusing a 40-residue polyglutamine extension, simulating the polyglutamine extension observed in Huntington’s disease, to YFP

131

132

Protein Homeostasis Diseases

and subsequent detection of fluorescence lifetime changes in C. elegans. An important observation was that aggregation, as observed by a lifetime decrease, was less prominent in a living C. elegans compared to in vitro observations,7 suggesting that perhaps in vitro assays overestimate aggregation rates, while in vivo aggregation-inhibiting or clearing factors may be active. More recently, C. elegans was used to demonstrate that aging is associated with the formation of aggregates from newly synthesized proteins that are toxic and that correlates with functional decline.10 This study used irreversible photoconversion of an mEOS2 label, a green-to-red convertible fluorescent protein. This was tagged to previously identified aging-related proteins casein kinase I isoform alpha (KIN-19), and Ras-like GTP-binding protein rhoA (RHO-1), two proteins previously observed to lose solubility upon aging of C. elegans,11 to distinguish newly synthesized proteins from their aggregates. Again, aggregation of fluorescently labeled proteins results in a reduction of fluorescence lifetime as a result of quenching, and using this approach to monitor KIN-19 and RHO-1 aggregation, it was observed that RHO-1 and KIN-19 colocalize forming mixed aggregates that are Congo red derivative X34, an amyloid stain, positive.10

Bioluminescence imaging Bioluminescence imaging is based on the conversion of luciferins into light by the enzyme luciferase from insects, such as the North American firefly, and sea species.12,13 Bioluminescence imaging can be used for in vivo detection of various processes in living animals, upon transgene generation expressing luciferase as a reporter gene, and injection of luciferin substrates as a result of the development of cameras that can detect light at high sensitivity.14,15 As excitation is not required, background signal in the animal is eliminated,16 while sensitivity depends on the wavelength of emitted light, light quantities generated, and overlapping tissue (reviewed by Ref. [17]). One application of interest in the field of protein aggregation is the use of split luciferase where catalytic activity of the enzyme is only regained upon close interaction of the domains (this approach is extensively discussed in Ref. [17]. Bioluminescence-based emission imaging has been used to noninvasively evaluate bioluminescence intensity emitted from the brains of a transgenic mouse model. A bigenic mouse model expressing mutant amyloid precursor protein (APP), both APP23 and CRND8a, and luciferase was generated to investigate aggregation of Alzheimer’s disease–related amyloid-β (Aβ) peptide using bioluminescence.18 The concomitant age-dependent increase in bioluminescence signal was interpreted as the deposition of Aβ into the brains of the transgenic animals. In the absence of clinical symptoms, changes in bioluminescence signal were found to correlate a Tg(PRNP-APPSweInd)8Dwst mouse line in which ‘Tg’ stands for transgene. This mouse model bears both the Swedish (K670N/M671L) and the Indiana (V717F) mutations in APP which stands for “amyloid precursor protein.”

Detection of amyloid aggregation in living systems

with the appearance of Aβ-containing amyloid plaques and glial fibrillary acidic protein (GFAP)–reactive astrocytes upon analysis of brain material derived from the biogenic mice. Using a similar approach, prion infection in transgenic mice expressing luciferase was effectively identified prior to symptomatic onset.19,20 Bioluminescence imaging has also been used to investigate a mutated form of superoxide dismutase (SOD) transgenic mouse model for amyotrophic lateral sclerosis (ALS). Upon generation of a doubletransgenic ALS (GFAP-luciferase-SODG93A) reporter mouse, living animals were studied for GFAP upregulation by means of bioluminescence. Interestingly, it was observed that GFAP was upregulated prior to the occurrence of clinical symptoms and that increased GFAP signals could be observed simultaneously both in the lumbar spinal cord projections as well as the periphery.21 This setup allowed also other observations that shed light on the progress of disease and deciphered specific pathological processes directly related with onset of the symptomatic stage, showing that GFAP is even more strongly induced in peripheral nerve Schwann cells.21

Optical fiber bundles and fluorescence imaging One of the major challenges in living whole-animal studies is the limited penetration of fluorescence excitation light as a result of tissue absorption.To increase detection sensitivity, a recent publication exploited the use of fiber optic–based imaging, effectively decreasing the distance between the source and the fluorescent target, to image bacteria in the lung of a mouse.22 In this way, the excitation light is delivered locally close to the tissue of interest inside the animal. Such a fiber optic–based technique is very suitable for in–living organism detection of fluorescently tagged proteins.To allow imaging, animals require light or full anesthetization. In this case, a recombinant Mycobacterium bovis strain expressing tdTomato, a relatively photostable dye with high quantum yield,23 was used to quantify the threshold of bacterial load to induce pulmonary infection in the lung of a mouse. An interesting in-brain application of fiber optics described the combination of a solid-state laser diode coupled to an optical fiber with optogenetics to study how optical stimulation could activate specific neurons in the brain.24 To this end, channelrhodopsin-2 (ChR2), a light-activated cation channel, was delivered by lentiviral methods to adult mice and rats by stereotactic injection to provide a light-driven stimulus selectively to ChR2-expressing neurons of the rodent vibrissal motor system.25 Following this work, a series of other studies have been performed using this technology to provide insight into all sorts of aspects of brain functioning, including the identification of specific neuronal networks that connect the basal forebrain and the cortex responsible for the regulation of cortical activity.26 One of the great questions in the field of amyloidosis is the specific sensitivity of subsets of neurons for a selective set of amyloidogenic proteins. Even though no publications to date report on the use of such fiber optic–based imaging technology to study protein aggregate formation in the brain, this technology could well hold promise for this application, especially in the light of the possibility to virally transfect specific subsets of cells in the brain.

133

134

Protein Homeostasis Diseases

Cranial window or thinned-skull imaging using multiphoton microscopy Two-photon fluorescence relies on the prediction that two photons can be absorbed simultaneously (within 0.1 fs) by a molecule during the same quantum event.27 The hence excited molecule decays to its normal state upon releasing an emission photon by a two-step process within a few nanoseconds. Two-photon laser scanning microscopy, which is based on the principle of two-photon fluorescence, allows fluorescently tagged molecules to be visualized up to a depth of more than 100 microns deep in a tissue. One of the benefits of two-photon microscopy is that different wavelengths can be selected for the two exciting photons. Access to the brain to allow for two-photon laser scanning microscopy is generally acquired either by skull thinning or cranial window techniques. Installation of a cranial window in mice brains followed by multiphoton imaging allows for nondestructive and direct three-dimensional visualization up to 900 µm deep into the cortex28–30 at micron resolution of various pathological processes taking place in a relatively large volume upon, for example, aging and/or neurodegeneration.31 More specifically, this method allows for longitudinal imaging of the same plaques and dendrites within an individual mouse over the course of a few weeks.31 Installation of a cranial window has been described in the past,32 and, based on this method, a number of seminal findings on deposition of proteins as well as their clearance and therapeutic removal by means of antibodies have been described in living animals. For example, to investigate the pathological involvement of various cell types and processes in the development of Alzheimer’s disease, longitudinal imaging by means of multiphoton microscopy has been employed using mice transgenic for APPswe/(presenilin)PS1d9xYFP.33 Using this strategy, it was observed that plaques appeared prior to microglial activation and recruitment to the plaques and that subsequently neurites become dysmorphic. These data provided important information on the sequence of events of Alzheimer’s disease pathology. To enable such imaging, an intracranial window was surgically introduced,31) and blood–brain barrier-permeable methoxy-XO4, a Congo red derivative, was intraperitoneally injected to stain amyloid plaques and cerebral amyloid angiopathy (CAA).34 Using Tg2576 mice, an Alzheimer’s disease model expressing human APP containing the Swedish mutation,35 enhanced GFP was intracortically expressed to visualize neurons. Subsequently, mice were administered with methoxy-XO4, and a cranial window was installed to allow visualization, using multiphoton imaging upon anesthetization of the mice, of neuronal processes in the vicinity of amyloid plaques in living animals. Using this approach, it was found that dystrophic neurons with reduced spine density were located in close proximity to, but did not penetrate, amyloid plaques in the cortex of Tg2576 mice. Similarly, using PSAPPb transgenic mice and transcranial two-photon imaging, it was demonstrated that dendrites in close proximity of amyloid deposits are b This mouse line carries mutations in APP KM670/671NL (Swedish), as well as PSEN1 (or PS which stands for presenilins) M146L (A>C).

Detection of amyloid aggregation in living systems

atrophic.36 These results were further confirmed using an immunofluorescent stain targeting amyloid plaques and in vivo multiphoton imaging of PDAPP mice treated with craniotomy followed by multiphoton microscopy performed under anesthesia.37 These findings are of seminal interest, as they showed that amyloid deposits could act as toxic mediators independent from oligomeric intermediates, which are primarily thought to induce neurotoxic action.

In vivo microdialysis In vivo microdialysis in amyloid research was developed to analyze soluble Aβ in the brain interstitial fluid (ISF) of a living, awake transgenic mouse,38 providing insight into the dynamic regulation of Aβ over time. This method is based on the implantation of a 38-kDa molecular-weight cutoff microdialysis probe into the hippocampus or striatum and quantification of Aβ in obtained ISF samples by enzyme-linked immunosorbent assay. One of the advantages of this technique is that experimental animals are awake, and data are not compromised by the use of anesthetics. Administration of anesthesia has been shown to be capable of inducing tau hyperphosphorylation, a modification that is known to be involved in the formation of intraneuronal neurofibrillary tangles of tau (reviewed by Ref. [39]), to promote oligomer formation by Aβ,40,41 and to increase Aβ production in cell cultures.42,43 The drawback of in vivo microdialysis is its invasive nature and that blood–brain barrier permeability was shown to be reversibly affected.44,45 Using this method, Aβ38 was identified as a major species in the ISF, and the half-life of Aβ in ISF was observed to be extended upon aging. This technique was further employed to determine brain ISF extracellular tau levels.46 This observation is of interest in light of other observations that showed that extracellular misfolded forms of tau may provide an intercellular means of transmitting misfolded conformations among the various neurons that make up specific brain regions.47,48 Within this context, also extracellular levels of Parkinson’s disease–related α-synuclein in response to neuronal activity were quantified using in vivo microdialysis.49 Attachment of a recording electrode to a microdialysis probe is capable of simultaneously measuring hippocampal electroencephalographic activity and ISF Aβ levels in freely moving Tg2576 mice.50 Induction of electrical seizures by continuous electrical stimulation upon infusion of a glutamate receptor 2/3 agonist via reverse microdialysis showed that increased neuronal activity co-occurred with performant pathway–mediated increased hippocampal ISF Aβ levels. Combining cranial window multiphoton microscopy with in vivo microdialysis appeared to be a highly powerful combination, elucidating that ISF Aβ concentration can affect the dynamic growth of amyloid plaques.51 In vivo microdialysis in mice also demonstrated a correlation between Aβ dynamics and the sleep-wake cycle, showing that sleep deprivation significantly increased ISF levels of Aβ.52 A significant drawback of in vivo microdialysis is the approximate 40-kDa cutoff of the membrane used for this technique, which does not allow direct insight into actual deposition into aggregates. However, it does provide

135

136

Protein Homeostasis Diseases

information on a potential pathogenic parameter by which concentration-dependent aggregation of Aβ may be regulated. Recent developments in the field of in vivo microdialysis resulted in the generation of a miniaturized probe, resolving issues with low temporal and spatial resolution in in vivo microdialysis techniques, using a channel array that can sample the neurotransmitter glutamate.53

Positron emission tomography imaging One of the ultimate examples of in vivo imaging of amyloid plaque deposition is the use of noninvasive PET imaging by means of infusion of a radioisotope against the suspected protein involved in a neuropathological process. The neutron-deficient radioisotope emits a positron that, upon interaction with an electron, generates two highenergy photons that can be detected by scintillator crystals coupled to photomultipliers outside of the body. Subsequently, images are reconstructed based on the within-tissue distribution of the radiotracer. In this way, PET is often used to complement the range of diagnostic measures to which suspected patients suffering from neurodegenerative disorders are exposed. The use of this technology as sole determinant in diagnostics has been questioned as a result of the observation that also nonpatients show a variety of deposits that are recognized by commonly used tracer molecules.54 Therefore, generally, diagnostics is based on the characterization of a number of the classical pathological features in conjunction with the reported clinical symptoms. The limited information on anatomical localization provided by PET imaging can be solved by combining PET readouts with X-ray computed tomography or magnetic resonance imaging.55,56 Amyloid protein–related PET tracers that have been generated to characterize amyloid burden in vivo include 18F-THK523,57 which recognizes phosphorylated tau but is significantly retained in the white matter58 and selectively binds to aggregates from Alzheimer’s disease pathology origin.59 The generation of 18F-AV-1451 (flortaucipir) against tau60 showed that this radioligand had improved affinity for tau and low white matter retention compared to previous-generation ligands, and the binding pattern of this ligand is directly associated with the extent of cognitive decline observed in progressive disease.61 The radioligand 11C-Pittsburgh compound B (PiB) is commonly used to detect fibrils of Aβ in the human brain.62 Binding studies of PiB to Aβ fibrils identified low and high binding sites to fibrils composed of Aβ1-42,63 but in this same publication, it was shown that PiB can identify the amount of Aβ deposits in human brains but not in the brains of transgenic PS1/APP mice. Moreover, even though this radioligand showed potential to detect disease progression in early stages, the PET signal obtained using this ligand remains stable upon the appearance of clinical symptoms.64,65 Moreover, familial forms of Alzheimer’s disease, characterized by diffuse plaque pathology and low levels of fibrillar forms of Aβ, showed negative PET scans.66 One of the reasons for this could be that clinical symptoms implicated in many neurodegenerative disorders nowadays have been observed to associate better with concentrations of prefibrillar species rather than

Detection of amyloid aggregation in living systems

mature fibrillar aggregates. In response to this, recently, an antibody-based blood–brainpermeable radioligand, di-scFv [124I]3D6-8D3, was reported that, apart from fibrillar forms of Aβ, also recognizes smaller soluble oligomers and protofibrils in transgenic mouse models.67 Positive PET scans using the radioligand 11C-PIB have also been observed for patients suffering from dementia with Lewy bodies, Parkinson’s patients with dementia,68 CAA,69,70 patients suffering from brain trauma,71 and older patients suffering from Down’s syndrome,72,73 demonstrating the wide but little differentiating power of this PET tracer. A review on amyloid PET imaging by means of 11C-PIB and 18Fflorbetapir, one of three fluorinated tracers that have been approved by the Food and Drug Administration (FDA) as diagnostic agents, in a range of amyloid diseases has been published before.74 The other two FDA-approved diagnostic fluorinated tracers are florbetaben75 and flutemetamol.76 PET images may be blurred and limit interpretation if the animal is allowed to move during acquisition, which is challenging, as some studies will take up to 20 minutes to complete.Therefore, subjects need to be restrained, or acquired images require frame-by-frame alignment or, preferably, motion tracking.77 Other limitations to PET are the high levels of noise and the limited spatial resolution.

Animal models to test in vivo amyloid formation All methods suitable for in vivo detection of amyloid formation, as discussed in the previous paragraph, with the exception of PET imaging, have been exploited only using animals. One of these assays based on FRET imaging requires transparent animals to allow in vivo visualization of amyloid processes. The animals most suitable for such studies are C. elegans, which are transparent at every stage of their life cycle, and early-age zebrafish. All other experimental methods can use nontransparent animals and often are based on mouse studies, although a procedure to generate transparent labeled nonliving mice has been described.78 In this section of the chapter, the suitability of various animal model systems to investigate in vivo amyloid aggregation will be discussed, and the progress made using the specific species under investigation involved will be addressed.

Caenorhabditis elegans C. elegans is a small, invertebrate transparent animal with simple, multicellular, 302 neuron-containing anatomy and rapid development (see also Chapter 3).79 The major anatomical features have been reviewed.80 The genetic makeup of C. elegans was originally described by Sydney Brenner, who used this nematode worm for his groundbreaking work on genetic regulation of organ development and programmed cell death. Reproduction of C. elegans occurs by hermaphrodite self-fertilization, allowing for maintenance of homozygous mutations without mating, and transgene lines are easily produced.81,82 The complete genome sequence was published in 1998,83 and it was reported that approximately 40% of its genes associated with human diseases have homologs in

137

138

Protein Homeostasis Diseases

the genome of C. elegans (reviewed by [84]). C. elegans is susceptible to genome-wide RNA interference by feeding.85 GFP-based visualization in living animals upon expression in C. elegans was reported as early as 1994,86 allowing for direct visualization of all sorts of processes by fluorescent labeling of proteins of interest. Apart from providing a very useful readout of studying in vivo genetic effects, a wide range of amyloid deposition experiments have been performed in C. elegans. For example, amyloid deposition of Aβ within the complex multicellular environment in vivo has been studied as a function of transthyretin coexpression using thioflavin S as a readout.87 Also, it has been observed that oxidative stress levels are increased prior to Aβ fibril formation, providing information on the sequence of pathologic events within the complex environment of the animal.88 In this work, by introducing a temperature-sensitive mutation in the mRNA surveillance system, a temperature-inducible Aβ-expressing C. elegans strain was generated, showing that the induction of Aβ expression co-occurred with significant oxidative stress as measured by protein carbonyl levels.88 C. elegans-expressing Aβ were also used to evaluate a newly developed sensitive amyloid stain, X-34, (1,4-bis[3-carboxy-4-hydroxyphenylethenyl]-benzene), a Congo red derivative, showing that this stain can sensitively detect amyloid fibrillar deposits composed from Aβ as well as transthyretin in living animals.89 At the same time, transgenes of C. elegans have been used to elucidate the involvement of autophagy in degradational pathways of Aβ.90 Consistent with this observation, a stress-sensitive mutant of C. elegans demonstrated a reduced survival upon RNA interference with mitophagy-relevant genes under conditions of stress and increased protein aggregation as well as increased oxidative stress levels, although nematodes were lysed prior to evaluation of aggregation by use of a commercially available protein aggregation assay.91

Zebrafish With a known and highly conserved genome that is closely related to the human genome,92 as well as reminiscent learning behavior,93,94 the vertebrate zebrafish, Danio rerio, highly qualifies as a model species to investigate protein aggregation–related neurodegeneration disorders. Zebrafish have a short reproductive cycle and generate a large number of progeny.Work with the zebrafish resulted in insights, using genetic mutations by means of microinjection of antisense morpholino oligonucleotides that can specifically bind their target mRNAs to block their translation or splicing, or transgenesis with a fluorescent reporter protein, into organization and vascularization of the brain and developmental processes.95–98 One of the first processes to be visualized in vivo in living embryo zebrafish was axonal outgrowth of spinal cord motor neurons.99 In addition to this, the small size of the zebrafish larvae enables well plate maintenance of these animals. For this reason, they have been exploited for large-scale screening of compounds that potentially qualify as therapeutic agents, an application that has been reviewed on various occasions,100–102 partly because blood–brain barrier regulation is fully functional at 10 days postfertilization, yielding crucial insight into blood–brain barrier permeability of

Detection of amyloid aggregation in living systems

compounds as well.103 One experimental work that explored reverse drug effects, using readouts for cardiac, intestinal, and visual function, reported that zebrafish models were partially capable of providing significant insight into adverse effects of small-molecule compounds.104 Further features that embrace the zebrafish as a model for studies in the field of neurodevelopment and neurodegeneration are that overall zebrafish brain structure105 and neurotransmitter pathways106 are reminiscent of those of humans, with some differences in terms of volume of cerebral hemispheres and forebrain structure (reviewed by Ref. [107]). The transparency of embryos from zebrafish provide unique in vivo insight into fluorescent-trackable processes such as aggregation of proteins. An individual neuron can be tracked over time and within an individual animal to allow spatial insight into the effects of specific disease-related mutations. Genetic studies on neurodegenerative disorders Alzheimer’s disease, Huntington’s disease, and Parkinson’s disease have been performed using zebrafish as a model system, and provided insight into pathological aspects including apoptotic response to polyglutamine (PolyQ) extension,108,109 γ-secretase-mediated APP processing,110 and genes implicated in Parkinson’s disease, such as DJ-1111,112 and ubiquitin carboxy-terminal hydrolase L1 (UCH-L1).113 A transgenic zebrafish model expressing a human tau-GFP fusion protein by microinjection was generated to investigate in vivo tau trafficking, cytoskeletal regulation, and posttranslational modification.114 Assembly of hyperphosphorylated tau into paired helical filaments could be observed following immunostaining. In vivo aggregation of ALS-related SOD1 G93A mutation was shown to develop axonopathy.115–117 An important limitation in the use of zebrafish for protein aggregation detection is the limited timeframe of their transparency, while protein aggregation is a time- and concentration-dependent process. As onset of pigmentation ensues at 24 hours after fertilization,118 the animal becomes progressively less transparent, which limits the time window during which in vivo processes can be effectively visualized, a time scale during which the blood–brain barrier lacks development. A number of efforts have been reported to prolong the transparency stages of living zebrafish upon development. These include the generation of albino strains119,120 or inhibition of pigmentation by using hydroquinone.121 A protocol that describes treatment of animals with 1-phenyl 2-thiourea (PTU) was reported to improve signal detection and optimized protocols, avoiding loss of embryo mortality, reduced hatching frequency, and teratogenesis to generate PTU-treated animals has been published.122 Administration of PTU well before first pigmentation was demonstrated to be required to block melanin pigmentation without affecting hatching frequency, zebrafish hatching enzyme messenger RNA levels, or survival.122 At the same time, the PTU-induced inhibition of pigmentation appeared to be reversible, as discontinuation of treatment with PTU led to pigmentation.

Mouse models As mice present a nontransparent rodent model, direct visualization of aggregation processes by means of fluorescent reporter–tagged proteins is not possible. On the other

139

140

Protein Homeostasis Diseases

hand, mice have been extensively used to gain insight into assembly of neurodegeneration-related aggregates via other means involving living animals, including in vivo microdialysis, bioluminescence imaging, or by means of multiphoton imaging after installing a cranial window. One exciting development in the field of live imaging of protein aggregation in a mouse model describes the expression of disease-generating proteins in the ocular lens, which allows noninvasive imaging in real time.123 Transgenic mice were generated to express GFP-labeled mutant huntingtin fragments or α-synuclein in the lens, resulting in opacity and supranuclear cataract formation detectable by 3 weeks of age observed in anesthetized mice by means of slit-lamp ophthalmoscopy and quasielastic laser light-scattering spectroscopy.

Note on using animal models for in vivo protein aggregation studies Living transgenic animal models combined with in vivo detection assays for protein aggregation continue to provide valuable information on neurodegenerative processes. This feature is largely due to the extensive battery of standardized experimental procedures available to generate transgenic animals and to test the impact of these genetic modifications for a wide range of pathological characteristics and processes. The connectome of C. elegans models has been fully mapped. This is advantageous when studying molecular mechanisms of changes in cellular and network response to pathological processes in this species. However, there are some clear limitations linked with the use of such animals. One of the limitations using transgenic animal models is that the agerelated characteristic of most sporadic forms of neurodegeneration cannot be replicated faithfully, as neurodegeneration as a function of aging in animals is a very rare event. Therefore, researchers turn to modifying genetic parameters that have been identified before to relate to familial forms of the pathologies investigated, often only partially reproducing the pathology as observed in aging humans. Lacking the human complexity and multifactorial character of most aging-related neurodegenerative disorders can be the result of comorbidities found in humans upon aging that complicate disease progress and hamper identification of a single contributing target. Posttranslational modifications occurring in proteins in the human brain are usually not replicated in animal models, and one striking example is the earlier-mentioned lack of PET tracer PiB to recognize amyloidogenic aggregates in mouse models, while this tracer shows a high level of specificity towards Alzheimer’s disease–implicated human plaques. Further, a variety of limitations has been reported related to specific neurotoxic and transgenic models employed for neurodegeneration-related protein aggregation studies. For example, high doses of paraquat can result in the development of pulmonary fibrosis,124 and the gene encoding for LRRK2 is sufficiently large to prevent the use of adenoassociated virus vectors.125

Detection of amyloid aggregation in living systems

In vivo complexity and how do in vivo detection assays provide insight into peripheral aspects contributing to neurodegeneration? It is of interest to evaluate how the current state of the art in terms of in vivo assays to study protein aggregation have affected our views on the role of protein aggregation in neurodegeneration when compared to the classical in vitro test tube– and cell culture– based assays.

Replicating the multicellular complexity of the brain Many of the classical assays investigating molecular mechanisms of neurodegeneration focus on the isolated neuron, or neuroblastoma cell, as the fundamental unit of the nervous system. Indeed, many valuable insights into neurodegenerative disorders have been derived from such simplified model systems. However, these models largely neglect the fact that the brain consists of a mixture of different cell types that show bidirectional interaction and impact each other’s behavior. One of the straightforward implications of this could be that the cellular complexity of the brain, consisting, apart from neuronal cells, also of astrocytes and microglia and their interactions, may impact on amyloid deposition. An interesting review126 highlighted how the various cell types present in the brain, including neurons, astrocytes, microglia, as well as the brain vasculature, respond to each other, affecting each other’s behavior. The authors plead for a move away from the single-cell biochemical assays that are performed to elucidate pathological processes at a molecular level toward studying more complex systems to understand the contribution of the various cell populations and their interactions in the brain toward a more systemic approach. The “tripartite synapse” hypothesis describes the observed tight bidirectional interaction that has been observed between astrocytes and neurons, and this theory describes well how synaptic transmission and plasticity are intimately regulated as a result of intercellular signaling pathways that act between these two cell types.127–129 For example, using fluorescence imaging, it was found that astrocytic intracellular Ca2+ concentrations rise in response to receptor-mediated binding of neurotransmitters127 released upon synaptic activity.128 Vice versa, synaptic transmission128 and cerebrovascular microcirculation130 are regulated in an astrocytic Ca2+-dependent manner by feedback signaling based on so-called astrocyte-secreted gliotransmitters. To address the complexity of interactions in such a multicellular system, the interaction between the various components of the brain has been modeled using computational approaches to show how environmental conditions affect the neuronal network as a function of astrocytic presence.131 Dissected postmortem brains derived from animal or human studies do provide some insight into the dynamics of, for example, Aβ generation, in response to various parameters, but preservation of DNA, which is sensitive to the various fixatives used for preservation of tissue,132–134 and the integrity of preserved RNA depends on factors including postmortem delay and the hypoxia state,135,136 while the preservation

141

142

Protein Homeostasis Diseases

of proteins depends on similar factors including postmortem delay between death and tissue processing, but vulnerability to degradation of proteins varies depending on the type of protein.137 All in all, even though investigations using postmortem brain tissue derived from humans or animals have resulted in many significant insights into pathophysiological processes, preparation of the brain for analysis may impact the outcome of such studies. For example, by means of in situ hybridization, the expression patterns of presenilins 1 and 2, one of the active components of γ-secretase responsible for cleaving APP into Aβ, in a rat brain was elucidated.138 These findings greatly contributed to the understanding of how familial mutations of these two proteins can contribute to familial forms of early-onset Alzheimer’s disease. The preservation limitations encountered demonstrate that studies of these processes in living animals provide the potential to significantly build on this knowledge. In response to this, a large number of researchers are now investigating the use of brain organoids, either derived from human embryonic stem cells (hESCs) or from induced pluripotent stem cells (IPSCs), to extend such observations in a living system.Various procedures to generate such organoids have been published, and, collectively, these methods describe the differentiation of either IPSCs or hESCs, by using a combination of growth factors, into a multilayered, threedimensional brain-like structure consisting of neurons and astrocytes that stain positive for various neuronal markers while showing electrophysiological activity and gene expression profiles, which are reasonably reminiscent to those of the human brain.139–141 (Box 7.2). One of the limitations of these organoids is their lack of vascularization, BOX 7.2  Organoids for neurodegeneration Organoids are currently being explored for their potential to recapitulate tissues and organs of the human body to describe a variety of physiological and pathological processes. The great advantage of using organoids, which are mostly derived from induced pluripotent stem cells (IPSCs) or human embryonic stem cells (hESCs), is that they resemble, in many ways, the three-dimensional organizational level of human tissues. Another interesting feature is that organoids can be reprogrammed by using various clustered regularly interspaced short palindromic repeats (CRISPR)–mediated genetic interventions pinpointing genetic factors in pathogenic processes. Nowadays, procedures have been described to redirect development of undifferentiated stem cells into a great number of different tissues and organs. These procedures vary somewhat in their approach, applying different growth factors at different time points and starting from different cell seed densities, but are generally well characterized by means of specific markers in the shape of transcription factors that should become activated or deactivated over time. A challenging aspect is to produce such organoids within a specific range of homogeneity to allow for statistical reproducibility of data, limited vascularization, and lack of maturation of organoids, which is often limited to early embryonic or fetal stages, not representing the aged stage that organs generally reside in when age-associated neurodegeneration kicks in. Various efforts have now been initiated to tackle these issues, including the application of microfluidic compartments, experimental validation of the maturation stage, and other technological advances to improve homogeneity of generated organoids.

Detection of amyloid aggregation in living systems

which generally leads to a growing core of necrotic cells deprived of nutrients available in the surrounding medium, as the organoid expands beyond a mathematically derived diameter of 1.4 mm.140,142 Such organoids are therefore sometimes grafted onto wellvascularized organs, and, in vivo two-photon imaging, combined with an optogenetics approach, was used to investigate neuronal activity and synaptic connectivity based on such a design.143 Even though aspects of protein deposition have been replicated in such brain organoids, in which IPSC-derived brain organoids from patients suffering from a familial form of Alzheimer’s disease or Down syndrome showed deposition of amyloid plaques and neurofibrillary tangles,144 no in vivo imaging to date has been performed to elucidate, for example, the kinetics or dynamics of amyloid deposition in such models. Clear limitations to the use of organoids to study physiological features at a molecular level include, apart from the abovementioned necrotic core formation induced by limited vascularization, the large variability obtained in the generated organoids145 and, most prominently, the general lack of maturation (reviewed by Ref. [146]), which, in light of the age-related association of most neurodegenerative disorders implicated in protein deposition, significantly impact application of the outcome for aging-related neurodegenerative disorders. Collectively, all model systems described in this paragraph have had their own significant contributions to the understanding of the workings of neurodegenerative disorders, but lacking the wealth of information that can be derived using in vivo detection of amyloid deposition in a living animal.

Interaction of the brain with the periphery system A factor that should be highlighted when studying brain-related neurodegenerative disorders involving amyloid deposition is the increasing awareness that the brain receives a wealth of information from its direct surrounding environment via the blood, humoral, and nervous system. This information is fed into the brain, and the brain responds to this information by modulating, for example, neuronal activity or glial activation, factors that are known to directly impact neurodegeneration-related protein aggregation processes. An interesting observation that illustrates the tight interaction between the brain and the peripheral system is circadian oscillation. Although the existence of peripheral clocks has been observed, circadian oscillation is primarily driven by the suprachiasmatic nuclei in the hypothalamus147,148 and is closely associated with the periphery synchronizing the body’s physiological functioning. Another example demonstrating an interaction between the brain and peripheral stimuli is the peripheral generation of inflammatory cytokines. Peripheral generation and secretion of the proinflammatory protein interleukin-1β (IL-1β) was found to be detected by the brain via activation of afferent fibers of the vagal nerve inducing various brain-driven effects, including modulation of brain norepinephrine levels.149 Also, it has been observed that immune-mediated signaling activated upon peripheral inflammation can induce behavioral changes, giving rise to hypotheses including the liver-brain inflammation axis.150 Further, recent findings on

143

144

Protein Homeostasis Diseases

the potential involvement of the gut microbiome in the pathogenic progress of Parkinson’s disease151 strongly suggest that organismal and even multiorganism complexity may well importantly contribute to disease process. Anatomical connections exist between the brain and the periphery, including the vagal nerve and the sympathetic nerve fibers, but also chemical interactions exist by means of small molecules, neuroendocrine peptides, cytokines, and growth factors. Many different communication pathways between peripheral organs and the brain have been identified, but their contributions toward neurodegenerative mechanisms have been largely neglected as a result of the use of isolated brain material. The means by which these organs interact is currently under investigation, but some outcomes highlight the interaction via neurotransmitters, humoral or monocyte-driven pathways, or signaling molecules that are trafficking via the blood or nervous system. These observations by themselves provide a crucial indication that addressing fundamental protein aggregation–related questions using simple in vitro model systems may provide very limited information on the contribution of complex intercellular and interorgan interactions of the living organism.

Future outlook Current important challenges in the field of amyloid deposition and related neurodegeneration are related to sequence-of-events questions as well as whole-systems biology. With the recognition that most amyloid deposition–related disorders are characterized by high complexity and their transient nature, while most of these features occur in the brain with input from a large number of peripheral organs and systems, in vivo wholeanimal imaging assays become an important tool to push this field of research forward. However, the technical possibilities to do this are somewhat hampered by limited transparency of higher-order multicellular organisms as well as by the penetration limit of most detection techniques. Also, parameters such as background fluorescence and resolution limits are in the way of gaining full appreciation of the dynamic nature of neurodegenerative processes. Currently, many exciting advances are being made to tackle these limitations by means of development of improved optics with better spatial and temporal resolution, data analysis protocols, and biomarker detection. For example, encapsulation of blood–brain barrier–impermeable molecular neuroimaging probes into nanoparticles composed of poly(n-butyl cyanoacrylate) dextran polymers enhanced blood–brain barrier permeability of the probe to enhance in vivo whole-brain imaging.152 Another more recent development is the generation of a probe that is able to map Aβ plaques in vivo in the brain by near-infrared detection. This hydrophilic probe, termed QM-FNSO3, binds to Aβ plaques, upon which the probe becomes fluorescent.153 Apart from these, many advances are currently being made in the field of optics, data acquisition, and analysis that will potentially provide the means in the future to obtain insight into amyloid deposition processes from a whole-body living organism perspective.

Detection of amyloid aggregation in living systems

Acknowledgments Research in the laboratory of K. Broersen is financially supported by ZonMw (The Netherlands Organisation for Health Research and Development), Alzheimer Nederland, and Stichting Fulbright Commission. Due to size limitations, not all of the relevant works to this chapter could be referred to.

References 1. Marvin JS, Borghuis BG, Tian L, Cichon J, Harnett MT, Akerboom J, Gordus A, Renninger SL, Chen T-W, Bargmann CI, Orger MB, Schreiter ER, Demb JB, Gan W-B, Hires SA, Looger LL. An optimized fluorescent probe for visualizing glutamate neurotransmission. Nature Meth 2013;10(2):162–5. 2. Keller JP, Marvin JS, Lacin H, Lemon WC, Shea J, Kim S, Lee RT, Koyama M, Keller PJ, Looger LL. In vivo glucose imaging in multiple model organisms with an engineered single-wavelength sensor. doi: https://doi.org/10.1101/571422. 3. van Ham T, Thijssen K, Breitling B, Hofstra R, Plasterk R, Nollen EA. C elegans model identifies genetic modifiers of α-synuclein inclusion formation during aging. PLoS Genet 2008;4(3):e100027. 4. van Ham T, Esposito A, Kumita J, Hsu ST, Kaminski Schierle G, Kaminski C, Dobson C, Nollen E, Bertoncini C. Towards multiparametric fluorescent imaging of amyloid formation studies of a YFP model of α-synuclein aggregation. J Mol Biol 2010;395(3):627–42. 5. Bodhicharla R, Nagarajan A, Winter J, Adenle A, Nazir A, Brady D, Vere K, Richens J, O’Shea P, Bell DR, de Pomerai D. Effects of α-synuclein overexpression in transgenic Caenorhabditis elegans strains. CNS Neurol Dis—Drug Targets 2012;11(8):965–75. 6. Del Mercato LL, Pompa PP, Maruccio G, Della Torre A, Sabella S, Tamburro AM, Cingolani R, Rinaldi R. Charge transport and intrinsic fluorescence in amyloid-like fibrils. Proc Natl Acad Sci USA 2007;104(46):18019–24. 7. Kaminski Schierle GS, Bertoncini CW, Chan FTS, van der Goot AT, Schwedler S, Skepper J, Schlachter S, van Ham T, Esposito A, Kumita JR, Nollen EAA, Dobson CM, Kaminski CF. A FRET sensor for non-invasive imaging of amyloid formation in vivo. Chemphyschem 2011;12(3):673–80 25. 8. Pinotsi D, Buell AK, Dobson CM, Kaminski Schierle GS, Kaminski CF. A label-free quantitative assay of amyloid fibril growth based on intrinsic fluorescence. ChemBioChem 2013;14(7):846–50. 9. Miyawaki A, Tsien RY. Monitoring protein conformations and interactions by fluorescence resonance energy transfer between mutants of Green Fluorescent protein. Appl Chim Genes Hybrid Prot (part B) 2000;327:472–500. 10. Huang C,Wagner-Valladolid S, Stephens AD, Jung R, Poudel C, Sinnige T, Lechler MC, Schlörit N, Lu M, Laine RF, Michel CH,Vendrusculo M, Kaminski CF, Kaminski Schierle GS, David DC. Intrinsically aggregation-prone proteins form amyloid-like aggregates and contribute to tissue aging in Caenorhabditis elegans. eLife 2019;8:e43059. 11. David DC, Ollikainen N, Trinidad JC, Cary MP, Burlingame AL, Kenyon C. Widespread protein aggregation as an inherent part of aging in C. elegans. PLoS Biol 2010;8:e1000450. 12. Thorne N, Inglese J, Auld DS. Illuminating insights into firefly luciferase and other bioluminescent reporters used in chemical biology. Chem Biol 2010;17(6):646–57. 13. Prescher JA, Contag CH. Guided by the light: visualizing biomolecular processes in living animals with bioluminescence. Curr Opin Chem Biol 2010;14(1):80–9. 14. Contag CH, Spilman SD, Contag PR, Oshiro M, Eames B, Dennery P, Stevenson DK, Benaron DA. Visualizing gene expression in living mammals using a bioluminescent reporter. Photochem Photobiol 1997;66(4):523–31. 15. Lyman GE, DeVincenzo JP. Determination of picogram amounts of ATP using the luciferin-luciferase enzyme system. Anal Biochem 1967;21(3):435–43. 16. Contag CH. Functional imaging using bioluminescent markers.Weissleder R, Ross BD, Rehemtulla A, Gambhir SS, editors. Molecular imaging: principles and practice. People’s Medical Publishing House; 2010. p. 118–138. 17. Keyaerts M, Caveliers V, Lahoutte T. Bioluminescence imaging: looking beyond the light. Trends Mol Med 2012;18(3):164–72.

145

146

Protein Homeostasis Diseases

18. Watts JC, Giles K, Grillo SK, Lemus A, DeArmond SJ, Prusiner SB. Bioluminescence imaging of Aβ deposition in bigenic mouse models of Alzheimer’s disease. Proc Natl Acad Sci USA 2011;108(6):2528– 33. 19. Zhu L, Ramboz S, Hewitt D, Boring L, Grass DS, Purchio AF. Non-invasive imaging of GFAP expression after neuronal damage in mice. Neurosci Lett 2004;367(2):210–2. 20. Tamgüney G, Francis KP, Giles K, Lemus A, DeArmond SJ, Prusiner SB. Measuring prions by bioluminescence imaging. Proc Natl Acad Sci USA 2009;106(35):15002–6. 21. Keller AF, Gravel M, Kriz J. Live imaging of amyotrophic lateral sclerosis pathogenesis: disease onset is characterized by marked induction of GFAP in Schwann cells. Glia 2009;57(10):1130–42. 22. Nooshabadi F,Yang H-J, Bixler JN, Kong Y, Cirillo JD, Maitland KC. Intravital fluorescence excitation in whole-animal optical imaging. PLoS One 2016;11(2):e0149932. 23. Goldgeier M, Fox CA, Zavislan JM, Harris D, Gonzalez S. Noninvasive imaging, treatment, and microscopic confirmation of clearance of basal cell carcinoma. Derma Surg 2003;29(3):205–10. 24. Aravanis AM, Wang L-P, Zhang F, Meltzer LA, Mogri MZ, Schneider MB, Deisseroth K. An optical neural interface: in vivo control of rodent motor cortex with integrated fiberoptic and optogenetic technology. J Neural Eng 2007;4(3):S143–56. 25. Zhang F, Wang LP, Boyden ES, Deisseroth K. Channelrhodopsin-2 and optical control of excitable cells. Nat Methods 2006;3(10):785–92. 26. Chaves-Coira I, Martín-Cortecero J, Nuñez A, Rodrigo-Angulo ML. Basal forebrain nuclei display distinct projecting pathways and functional circuits to sensory primary and prefrontal cortices in the rat. Front Neuroanat 2018;12:69. 27. Göppert-Mayer M. Über elementarakte mit zwei quantensprüngen. Ann Physik 1931;401(3):273–94. 28. Yang G, Pan F, Parkhurst CN, Grutzendler J, Gan WB. Thinned-skull cranial window technique for long-term imaging of the cortex in live mice. Nat Protoc 2010;5(2):201–8. 29. Shih AY, Driscoll JD, Drew PJ, Nishimura N, Schaffer CB, Kleinfeld D. Two-photon microscopy as a tool to study blood flow and neurovascular coupling in the rodent brain. J Cereb Blood Flow Metab 2012;32:1277–309. 30. Lathia JD, Gallagher J, Myers JT, Li M, Vasanji A, McLendon RE, Hjelmeland AB, Huang AY, Rich JN. Direct in vivo evidence for tumor propagation by glioblastoma cancer stem cells. PLoS One 2011;6:e24807. 31. Spires TL, Meyer-Luehmann M, Stern EA, McLean PJ, Skoch J, Nguyen PT, Bacskai BJ, Hyman BT. Dendritic spine abnormalities in amyloid precursor protein transgenic mice demonstrated by gene transfer and intravital multiphoton microscopy. J Neurosci 2005;25(31):7278–87. 32. Skoch J, Hickey GA, Kajdasz ST, Hyman BT, Bacskai BJ. In , editor. Amyloid proteins: methods and protocols. Humana Press; Totowa, 2004:349-364. 33. Meyer-Luehmann M, Spires-Jones TL, Prada C, Garcia-Alloza M, de Calignon A, Rozkalne A, Koeningsknecht-Talboo J, Holtzman DM, Bacskai BJ, Hyman BT. Rapid appearance and local toxicity of amyloid-β plaques in a mouse model of Alzheimer’s disease. Nature 2008;451(7179):720–4. 34. Klunk WE, Bacskai BJ, Mathis CA, Kajdasz ST, McLellan ME, Frosch MP, Debnath ML, Holt DP, Wang Y, Hyman BT. Imaging Aβ plaques in living transgenic mice with multiphoton microscopy and methoxy-X04, a systemically administered Congo red derivative. J Neuropathol Exp Neurol 2002;61(9):797–805. 35. Hsiao K, Chapman P, Nilsen S, Eckman C, Younkin S, Yang F, Cole G. Correlative memory deficits, abeta elevation, and amyloid plaques in transgenic mice. Science 1996;274(5284):99–102. 36. Tsai J, Grutzendler J, Duff K, Gan W-B. Fibrillar amyloid deposition leads to local synaptic abnormalities and breakage of neuronal branches. Nat Neurosci 2004;7(11):1181–3. 37. D’Amore JD, Kajdasz ST, McLellan ME, Bacskai BJ, Stern EA, Hyman BT. In vivo multiphoton imaging of a transgenic mouse model of Alzheimer disease reveals marked thioflavine-S-associated alterations in neurite trajectories. J Neuropathol Exp Neurol 2003;62(2):137–45. 38. Cirrito JR, May PC, O’Dell MA, Taylor JW, Parsadanian M, Cramer JW, Audia JE, Nissen JS, Bales KR, Paul SM, DeMattos RB, Holtzman DM. In vivo assessment of brain interstitial fluid with microdialysis reveals plaque-associated changes in amyloid-β metabolism and half-life. J Neurosci 2003;23(26):8844–53. 39. Whittington RA, Bretteville A, Dickler MF, Planel E. Anesthesia and tau pathology. Psychopharm Biol Psych 2013;47:147–55.

Detection of amyloid aggregation in living systems

40. Eckenhoff RG, Johansson JS, Wei H, Carnini A, Kang B, Wei W, Pidikiti R, Keller JM, Eckenhoff MF. Inhaled anesthetic enhancement of amyloid-beta oligomerization and cytotoxicity. Anesthesiology 2004;101(3):703–9. 41. Carnini A, Lear JD, Eckenhoff RG. Inhaled anesthetic modulation of amyloid beta(1-40) assembly and growth. Curr Alzheimer Res 2007;4(3):233–41. 42. Xie Z, Dong Y, Maeda U, Alfille P, Culley DJ, Crosby G, Tanzi RE. The common inhalation anesthetic isoflurane induces apoptosis and increases amyloid beta protein levels. Anesthesiology 2006;104(5):988– 94. 43. Zhang B, Dong Y, Zhang G, Moir RD, Xia W,Yue Y, Tian M, Culley DJ, Crosby G, Tanzi RE, Xie Z. The inhalation anesthetic desflurane induces caspase activation and increases amyloid beta-protein levels under hypoxic conditions. J Biol Chem 2008;283(18):11866–75. 44. Dykstra KH, Hsiao JK, Morrison PF, Bungay PM, Mefford IN, Scully MM, Dedrick RL. Quantitative examination of tissue concentration profiles associated with microdialysis. J Neurochem 1992;58(3):931– 40. 45. Morgan ME, Singhal D, Anderson BD. Quantitative assessment of blood-brain barrier damage during microdialysis. J Pharmacol Exp Ther 1996;277(2):1167–76. 46. Yamada K. In vivo microdialysis of brain interstitial fluid for the determination of extracellular tau levels. Methods Mol Biol 2017;1523:285–96. 47. Lasagna-Reeves CA, Castillo-Carranza DL, Sengupta U, Guerrero-Munoz MJ, Kiritoshi T, Neugebauer V, Jackson GR, Kayed R. Alzheimer brain-derived tau oligomers propagate pathology from endogenous tau. Sci Rep 2012;2:700. 48. Clavaguera F, Akatsu H, Fraser G, Crowther RA, Frank S, Hench J, Probst A, Winkler DT, Reichwald J, Staufenbiel M, Ghetti B, Goedert M, Tolnay M. Brain homogenates from human tauopathies induce tau inclusions in mouse brain. Proc Natl Acad Sci USA 2013;110(23):9535–40. 49. Yamada K, Iwatsubo T. Extracellular α-synuclein levels are regulated by neuronal activity. Mol Neurodegener 2018;13(1):9. 50. Cirrito JR, Yamada KA, Finn MB, Sloviter RS, Bales KR, May PC, Schoepp DD, Paul SM, Mennerick S, Holtzman DM. Synaptic activity regulates interstitial fluid amyloid-β levels in vivo. Neuron 2005;48(6):913–22. 51. Yan P, Bero AW, Cirrito JR, Xiao Q, Hu X, Wang Y, Gonzales E, Holtzman DM, Lee JM. Characterizing the appearance and growth of amyloid plaques in APP/PS1 mice. J Neurosci 2009;29(34):10706– 14. 52. Kang JE, Lim MM, Bateman RJ, Lee JJ, Smyth LP, Cirrito JR, Fujiki N, Nishino S, Holtzman DM. Amyloid-β dynamics are regulated by orexin and the sleep-wake cycle. Science 2009;326(5955):1005– 7. 53. Brink van den FTG, Phisonkunkasem T, Asthana A, Bomer JG, Maagdenburg van den AMJM, Tolner EA, Odijk M. A miniaturized push-pull-perfusion probe for few-second sampling of neurotransmitters in the mouse brain. Lab Chip 2019;19(8):1332–43. 54. Ossenkoppele R, Jansen WJ, Rabinovici GD, Knol DL, van der Flier WM, van Berckel BN, Scheltens P, Visser PJ, Amyloid PET, Study Group,Verfaillie SC, Zwan MD, Adriaanse SM, Lammertsma AA, Barkhof F, Jagust WJ, Miller BL, Rosen HJ, Landau SM,Villemagne VL, Rowe CC, Lee DY, Na DL, Seo SW, Sarazin M, Roe CM, Sabri O, Barthel H, Koglin N, Hodges J, Leyton CE,Vandenberghe R, van Laere K, Drzezga A, Forster S, Grimmer T, Sánchez-Juan P, Carril JM, Mok V, Camus V, Klunk WE, Cohen AD, Meyer PT, Hellwig S, Newberg A, Frederiksen KS, Fleisher AS, Mintun MA, Wolk DA, Nordberg A, Rinne JO, Chételat G, Lleo A, Blesa R, Fortea J, Madsen K, Rodrigue KM, Brooks DJ. Prevalence of amyloid PET positivity in dementia syndromes: a meta-analysis. JAMA 2015;313(19):1939–49. 55. Beyer T, Townsend DW, Brun T, Kinahan PE, Charron M, Roddy R, Jerin J,Young J, Byars L, Nutt R. A combined PET/CT scanner for clinical oncology. J Nucl Med 2000;41(8):1369–79. 56. Cohade C, Osman M, Pannu HK, Wahl RL. Uptake in supraclavicular area fat (“USA-Fat”): description on 18F-FDG PET/CT. J Nucl Med 2003;44(2):170–6. 57. Okamura N, Suemoto T, Furumoto S, Suzuki M, Shimadzu H, Akatsu H, Yamamoto T, Fujiwara H, Nemoto M, Maruyama M, Arai H, Yanai K, Sawada T, Kudo Y. Quinoline and benzimidazole derivatives: candidate probes for in vivo imaging of tau pathology in Alzheimer’s disease. J Neurosci 2005;25(47):10857–62.

147

148

Protein Homeostasis Diseases

58. Villemagne L, Furumoto S, Fodero-Tavoletti MT, Mulligan RS, Hodges J, Harada R,Yates P, Piguet O, Pejoska S, Doré V,Yanai K, Masters CL, Kudo Y, Rowe CC, Okamura N. In vivo evaluation of a novel tau imaging tracer for Alzheimer’s disease. Eur J Nucl Med Mol Imaging 2014;41(5):816–26. 59. Fodero-Tavoletti MT, Furumoto S, Taylor L, McLean CA, Mulligan RS, Birchall I, Harada R, Masters CL,Yanai K, Kudo Y, Rowe CC, Okamura N,Villemagne VL. Assessing THK523 selectivity for tau deposits in Alzheimer’s disease and non-Alzheimer’s disease tauopathies. Alzheimers Res Ther 2014;6(1):11. 60. Shcherbinin S, Schwarz AJ, Joshi A, Navitsky M, Flitter M, Shankle WR, Devous Sr MD, Mintun MA. Kinetics of the tau PET tracer 18F-AV-1451 (T807) in subjects with normal cognitive function, mild cognitive impairment, and Alzheimer disease. J Nucl Med 2016;57(10):1535–42. 61. Johnson KA, Schultz A, Betensky RA, Becker JA, Sepulcre J, Rentz D, Mormino E, Chhatwal J, Amariglio R, Papp K, Marshall G, Albers M, Mauro S, Pepin L, Alverio J, Judge K, Philiossaint M, Shoup T,Yokell D, Dickerson B, Gomez-Isla T, Hyman B,Vasdev N, Sperling R. Tau positron emission tomographic imaging in aging and early Alzheimer disease. Ann Neurol 2016;79(1):110–9. 62. Mathis CA, Wang Y, Holt DP, Huang G-F, Debnath ML, Klunk WE. Synthesis and evaluation of 11Clabeled 6-substituted 2-arylbenzothiazoles as amyloid imaging agents. J Med Chem 2003;46(13):2740– 54. 63. Klunk WE, Lopresti BJ, Ikonomovic MD, Lefterov IM, Koldamova RP, Abrahamson EE, Debnath ML, Holt DP, Huang G-f, Shao L, DeKosky ST, Price JC, Mathis CA. Binding of the positron emission tomography tracer Pittsburgh compound-B reflects the amount of amyloid-β in Alzheimer’s disease brain but not in transgenic mouse brain. J Neurosci 2005;25(46):10598–606. 64. Engler H, Forsberg A, Almkvist O, Blomquist G, Larsson E, Savitcheva I, Wall A, Ringheim A, Långström B, Nordberg A. Two-year follow-up of amyloid deposition in patients with Alzheimer’s disease. Brain 2006;129(Pt 11):2856–66. 65. Rodriguez-Vieitez E, Saint-Aubert L, Carter SF, Almkvist O, Farid K, Schöll M, Chiotis K, Thordardottir S, Graff C, Wall A, Långström B, Nordberg A. Diverging longitudinal changes in astrocytosis and amyloid PET in autosomal dominant Alzheimer’s disease. Brain 2016;139(Pt 3):922–36. 66. Schöll M, Wall A, Thordardottir S, Ferreira D, Bogdanovic N, Långström B, Almkvist O, Graff C, Nordberg A. Low PiB PET retention in presence of pathologic CSF biomarkers in Arctic APP mutation carriers. Neurology 2012;79(3):229–36. 67. Fang XT, Hultqvist G, Meier SR, Antoni G, Sehlin D, Syvänen S. High detection sensitivity with antibody-based PET radioligand for amyloid beta in the brain. Neuroimage 2019;184:881–8. 68. Edison P, Rowe CC, Rinne JO, Ng S, Ahmed I, Kemppainen N, Villemagne VL, O’Keefe G, Nagren K, Chaudhury KR, Masters CL, Brooks DJ. Amyloid load in Parkinson’s disease dementia and Lewy body dementia measured with [11C]PIB positron emission tomography. J Neurol Neurosurg Psychiatry 2008;79(12):1331–8. 69. Lockhart A, Lamb JR, Osredkar T, Sue LI, Joyce JN,Ye L, Libri V, Leppert D, Beach TG. PIB is a nonspecific imaging marker of amyloid-beta (Abeta) peptide-related cerebral amyloidosis. Brain J Neurol 2007;130(10):2607–15. 70. Ikonomovic MD, Klunk WE, Abrahamson EE, Mathis CA, Price JC, Tsopelas ND, Lopresti BJ, Ziolko S, Bi W, Paljug WR, Debnath ML, Hope CE, Isanski BA, Hamilton RL, DeKosky ST. Post-mortem correlates of in vivo PiB-PET amyloid imaging in a typical case of Alzheimer’s disease. Brain J Neurol 2008;131(Pt 6):1630–45. 71. Kawai N, Kawanishi M, Kudomi N, Maeda Y,Yamamoto Y, Nishiyama Y, Tamiya T. Detection of brain amyloid beta deposition in patients with neuropsychological impairment after traumatic brain injury: PET evaluation using Pittsburgh compound-B. Brain Inj 2013;27(9):1026–31. 72. Landt J, D’Abrera JC, Holland AJ, Aigbirhio FI, Fryer TD, Canales R, Hong YT, Menon DK, Baron JC, Zaman SH. Using positron emission tomography and Carbon 11-labeled Pittsburgh Compound B to image brain fibrillar β-amyloid in adults with down syndrome: safety, acceptability, and feasibility. Arch Neurol 2011;68(7):890–6. 73. Handen BL, Cohen AD, Channamalappa U, Bulova P, Cannon SA, Cohen WI, Mathis CA, Price JC, Klunk WE. Imaging brain amyloid in nondemented young adults with Down syndrome using Pittsburgh compound B. Alzheimers Dement 2012;8(6):496–501. 74. Catafau AM, Bullich S. Amyloid PET imaging: applications beyond Alzheimer’s disease. Clin Transl Imaging 2015;3(1):39–55.

Detection of amyloid aggregation in living systems

75. Ong KT, Villemagne VL, Bahar-Fuchs A, Lamb F, Langdon N, Catafau AM, Stephens AW, Seibyl J, Dinkelborg LM, Reininger CB, Putz B, Rohde B, Masters CL, Rowe CC. Aβ imaging with 18Fflorbetaben in prodromal Alzheimer’s disease: a prospective outcome study. J Neurol Neurosurg Psych 2015;86:431–6. 76. Curtis C, Gamez JE, Singh U, Sadowsky CH, Villena T, Sabbagh MN, Beach TG, Duara R, Fleisher AS, Frey KA, Walker Z, Hunjan A, Holmes C, Escovar YM, Vera CX, Agronin ME, Ross J, Bozoki A, Akinola M, Shi J, Vandenberghe R, Ikonomovic MD, Sherwin PF, Grachev ID, Farrar G, Smith APL, Buckley CJ, McLain R, Salloway S. Phase 3 trial of flutemetamol labeled with radioactive fluorine 18 imaging and neuritic plaque density. JAMA Neurol 2015;72(3):287–94. 77. Montgomery AJ, Thielemans K, Mehta MA, Turkheimer F, Mustafovic S, Grasby PM. Correction of head movement on PET studies: comparison of methods. J Nucl Med 2006;47(12):1936–44. 78. Yang B, Treweek JB, Kulkarni RP, Deverman BE, Chen C-K, Lubeck E, Shah S, Cai L, Gradinaru V. Single-cell phenotyping within transparent intact tissue through whole-body clearing. Cell 2014;158(4):945–58. 79. Brenner S. Genetics of Caenorhabditis elegans. Genetics 1974;77(1):71–94. 80. Corsi AK. A biochemist’s guide to C. elegans. Anal Biochem 2006;359(1):1–17. 81. Fire A. Integrative transformation of Caenorhabditis elegans. EMBO J 1986;5(10):2673–80. 82. Mello CC, Kramer JM, Stinchcomb D, Ambros V. Efficient gene transfer in C. elegans: extrachromosomal maintenance and integration of transforming sequences. EMBO J 1991;10(12):3959–70. 83. C. elegans Sequencing Consortium. Genome sequence of the nematode C. elegans: a platform for investigating biology. Science. 1998;282(5396):2012–2018. 84. Culetto E, Sattelle DB. A role for Caenorhabditis elegans in understanding the function and interactions of human disease genes. Hum Mol Genet 2000;9(6):869–77. 85. Kamath RS, Fraser A, Dong Y, Poulin G, Durbin R, Gotta M, Kanapin A, Le Bot N, Morena S, Sohrmann M,Welchman DP, Zipperlen P, Ahringer J. Systemic functional analysis of the Caenorhabditis elegans genome using RNAi. Nature 2003;421(6920):231–7. 86. Chalfie M, Tu Y, Euskirchen G, Ward WW, Prasher DC. Green fluorescent protein as a marker for gene expression. Science 1994;263(5148):802–5. 87. Link CD. Expression of human beta-amyloid peptide in transgenic Caenorhabditis elegans. Proc Natl Acad Sci USA 1995;92(20):9368–72. 88. Drake J, Link CD, Butterfield DA. Oxidative stress precedes fibrillar deposition of Alzheimer’s disease amyloid β-peptide (1-42) in a transgenic Caenorhabditis elegans model. Neurobiol Aging 2003;24(3):415–20. 89. Link CD, Johnson CJ, Fonte V, Paupard M-C, Hall DH, Styren S, Mathis CA, Klunk WE. Visualization of fibrillar amyloid deposits in living, transgenic Caenorhabditis elegans animals using the sensitive amyloid dye, X-34. Neurobiol Aging 2001;22(2):217–26. 90. Florez-McClure ML, Hohsfield LA, Fonte G, Bealor MT, Link C.D.. Decreased insulin-receptor signaling promotes the autophagic degradation of β-amyloid peptide in C. elegans. Autophagy 2007;3(6):569–80. 91. Civelek M, Mehrkens J-F, Carstens N-M, Fitzenberger E,Wenzel U. Inhibition of mitophagy decreases survival of Caenorhabditis elegans by increasing protein aggregation. Mol Cell Biochem 2019;452(1– 2):123–31. 92. Postlethwait JH, Woods IG, Ngo-Hazelett P, Yan YL, Kelly PD, Chu F, Huang H, Hill-Force A, Talbot WS. Zebrafish comparative genomics and the origins of vertebrate chromosomes. Genome Res 2000;10(12):1890–902. 93. Zhdanova IV, Wang SY, Leclair OU, Danilova NP. Melatonin promotes sleep-like state in zebrafish. Brain Res 2001;903(1–2):263–8. 94. Orger MB, Gahtan E, Muto A, Page-McCaw P, Smear MC, Baier H. Behavioral screening assays in zebrafish. Methods Cell Biol 2004;77:53–68. 95. Driever W, Solnica-Krezel L, Schier AF, Neuhauss SC, Malicki J, Stemple DL, Stainier DY, Zwartkruis F, Abdelilah S, Rangini Z, Belak J, Boggs C. A genetic screen for mutations affecting embryogenesis in zebrafish. Development 1996;123:37–46. 96. Haffter P, Granato M, Brand M, Mullins MC, Hammerschmidt M, Kane DA, Odenthal J, van Eeden FJ, Jiang YJ, Heisenberg CP, Kelsh RN, Furutani-Seiki M,Vogelsang E, Beuchle D, Schach U, Fabian C, Nüsslein-Volhard C. The identification of genes with unique and essential functions in the development of the zebrafish, Danio rerio. Development 1996;123:1–36.

149

150

Protein Homeostasis Diseases

97. Stainier DY. A glimpse into the molecular entrails of endoderm formation. Genes Dev 2002;16:893– 907. 98. Thisse C, Zon LI. Organogenesis — heart and blood formation from the zebrafish point of view. Science 2002;295(5554):457–62. 99. Eisen JS, Myers PZ, Westerfield M. Pathway selection by growth cones of identified motoneurones in live zebra fish embryos. Nature 1986;320(6059):269–71. 100. MacRae CA, Peterson RT. Zebrafish-based small molecule discovery. Chem Biol 2003;10(10):901–8. 101. Zon LI, Peterson RT. In vivo drug discovery in the zebrafish. Nat Rev Drug Discov 2005;4(1):35–44. 102. Murphey RD, Zon LI. Small molecule screening in the zebrafish. Methods 2006;39(3):255–61. 103. Goldsmith P, Fleming A. Screening methods employing zebrafish and the blood brain barrier. United States patent no. US 2006/0193776 A1; 2007. 104. Berghmans S, Butler P, Goldsmith P,Waldron G, Gardner I, Golder Z, Richards FM, Kimber G, Roach A, Alderton W, Fleming A. Zebrafish based assays for the assessment of cardiac, visual and gut function-potential safety screens for early drug discovery. J Pharmacol Toxicol Methods 2008;58(1):59–68. 105. Tropepe V, Sive HL. Can zebrafish be used as a model to study the neurodevelopmental causes of autism? Genes Brain Behav 2003;2(5):268–81. 106. Rink E,Wullimann MF. Connections of the ventral telencephalon (subpallium) in the zebrafish (Danio rerio). Brain Res 2004;1011(2):206–20. 107. Wullimann MF, Mueller T. Teleostean and mammalian forebrains contrasted: evidence from genes to behavior. J Comp Neurol 2004;475(2):143–62. 108. Schiffer NW, Broadley SA, Hirschberger T, Tavan P, Kretzschmar HA, Giese A, Haass C, Hartl FU, Schmid B. Identification of anti-prion compounds as efficient inhibitors of polyglutamine protein aggregation in a zebrafish model. J Biol Chem 2007;282(12):9195–203. 109. Miller VM, Nelson RF, Gouvion CM, Williams A, Rodriguez-Lebron E, Harper SQ, Davidson BL, Rebagliati MR, Paulson HL. CHIP suppresses polyglutamine aggregation and toxicity in vitro and in vivo. J Neurosci 2005;25(40):9152–61. 110. Musa A, Lehrach H, Russo VE. Distinct expression patterns of two zebrafish homologues of the human APP gene during embryonic development. Dev Genes Evol 2001;211(11):563–7. 111. Bretaud S, Allen C, Ingham PW, Bandmann O. p53-dependent neuronal cell death in a DJ-1-deficient zebrafish model of Parkinson’s disease. J Neurochem 2007;100(6):1626–35. 112. Chen L, Cagniard B, Mathews T, Jones S, Koh HC, Ding Y, Carvey PM, Ling Z, Kang UJ, Zhuang X. Age-dependent motor deficits and dopaminergic dysfunction in DJ-1 null mice. J Biol Chem 2005;280(22):21418–26. 113. Son OL, Kim HT, Ji MH,Yoo KW, Rhee M, Kim CH. Cloning and expression analysis of a Parkinson’s disease gene, uch-L1, and its promoter in zebrafish. Biochem Biophys Res Commun 2003;312(3):601–7. 114. Tomasiewicz HG, Flaherty DB, Soria JP, Wood JG. Transgenic zebrafish model of neurodegeneration. J Neurosci Res 2002;70(6):734–45. 115. Lemmens R,Van Hoecke A, Hersmus N, Geelen V, D’Hollander I,Thijs V,Van Den Bosch L, Carmeliet P, Robberecht W. Overexpression of mutant superoxide dismutase 1 causes a motor axonopathy in the zebrafish. Hum Mol Genet 2007;16(19):2359–65. 116. Laird AS, Robberecht W. Modeling neurodegenerative diseases in zebrafish embryos. Methods Mol Biol 2011;793:167–84. 117. Xi Y, Noble S, Ekker M. Modeling neurodegeneration in zebrafish. Curr Neurol Neurosci Rep 2011;11(3):274–82. 118. Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, Schilling TF. Stages of embryonic development of the zebrafish. Dev Dyn 1995;203(3):253–310. 119. Kelsh RN, Brand M, Jiang YJ, Heisenberg CP, Lin S, Haffter P, Odenthal J, Mullins MC, van Eeden FJ, Furutani-Seiki M, Granato M, Hammerschmidt M, Kane DA, Warga RM, Beuchle D,Vogelsang L, Nüsslein-Volhard C. Zebrafish pigmentation mutations and the process of neural crest development. Development 1996;123:369–89. 120. Kameyama K, Jiménes M, Muller J, Ishida Y, Hearing VJ. Regulation of mammalian melanogenesis by tyrosinase inhibition. Differentiation 1989;42(1):28–36. 121. Palumbo A, d’Ischia M, Misuraca G, Prota G. Mechanism of inhibition of melanogenesis by hydroquinone. Biochim Biophys Acta 1991;1073(1):85–90.

Detection of amyloid aggregation in living systems

122. Karlsson J, von Hofsten J, Olsson P-E. Generating transparent zebrafish: a refined method to improve detection of gene expression during embryonic development. Marine Biotech 2001;3(6):522–7. 123. Muchowski PJ, Ramsden R, Nguyen Q, Arnett EE, Greiling TM, Anderson SK, Clark JI. Noninvasive measurement of protein aggregation by mutant huntingtin fragments or alpha-synuclein in the lens. J Biol Chem 2008;283(10):6330–6. 124. Uversky VN. Neurotoxicant-induced animal models of Parkinson’s disease: understanding the role of rotenone, maneb and paraquat in neurodegeneration. Cell Tissue Res 2004;318(1):225–41. 125. Van der Perren A,Van den Haute C, Baekelandt V.Viral vector-based models of Parkinson’s disease. Curr Top Behav Neurosci 2015;22:271–301. Available from: https://doi.org/10.1007/7854_2014_310. 126. Strooper De B, Karran E. The cellular phase of Alzheimer’s disease. Cell 2016;164(4):603–15. 127. Porter JT, McCarthy KD. Astrocytic neurotransmitter receptors in situ and in vivo. Prog Neurobiol 1997;51(4):439–55. 128. Araque A, Martin ED, Perea G, Arellano JI, Buño W. Synaptically-released acetylcholine evokes Ca2+ elevations in astrocytes in hippocampal slices. J Neurosci 2002;22(7):2443–50. 129. Araque A, Sanzgiri RP, Parpura V, Haydon PG. Calcium elevation in astrocytes causes an NMDA receptor-dependent increase in the frequency of miniature synaptic currents in cultured hippocampal neurons. J Neurosci 1998;18(17):6822–9. 130. Mulligan SJ, MacVicar BA. Calcium transients in astrocyte end feet cause cerebrovascular constrictions. Nature 431;(7005):195–199. 131. Zareh M, Manshaei MH, Adibi M, Montazeri MA. Neurons and astrocytes interaction in neuronal network: a game-theoretic approach. J Theor Biol. 470:76–89. 132. Koppelstaetter C, Jennings P, Hochegger K, Perco P, Ischia R, Karkoszka H, Mayer G. Effect of tissue fixatives on telomere length determination by qualitative PCR. Mech Ageing Dev 2005;126(12):1331–3. 133. Kunkle RA, Miller JM, Alt DP, Cutlip RC, Cockett NE, Wang S, Richt JA, Thomsen BV, Hall SM. Determination of sheep prion gene polymorphisms from paraffin-embedded tissues. J Vet Diagn Invest 2006;18(5):443–7. 134. Miething F, Hering S, Hanschke B, Dressler J. Effect of fixation for the degradation of nuclear and mitochondrial DNA in different tissues. J Histochem Cytochem 2006;54(3):371–4. 135. Preece P, Cairns NJ. Quantifying mRNA in post-mortem human brain: influence of gender, age at death, post-mortem interval, brain pH, agonal state and inter-lobe mRNA variance. Brain Res Mol Brain Res 2003;118(1–2):60–71. 136. Tomita H, Vawter MP, Walsh DM, Evans SJ, Choudary PV, Li J, Overman KM, Atz ME, Myers RM, Jones EG, Watson SJ, Akil H, Bunney Jr WE. Effect of agonal and post-mortem factors on gene expression profile: quality control in microarray analyses of post-mortem human brain. Biol Psychiatry 2004;55(4):346–52. 137. Ferrer I, Santpere G, Arzberger T, Bell J, Blanco R, Boluda S, Budka H, Carmona M, Giaccone G, Krebs B, Limido L, Parchi P, Puig B, Strammiello R, Ströbel T, Kretzschmar H. Brain protein preservation largely depends on the postmortem storage temperature: implications for study of proteins in human neurologic diseases and management of brain banks: a BrainNet Europe Study. J Neuropathol Exp Neurol 2007;66(1):35–46. 138. Kovacs DM, Fausett HJ, Page KJ, Kim T-W, Moir RD, Merriam DE, Hollister RD, Hallmark OG, Mancini R, Felstenstein KM, Hyman BT, Tanzi RE, Wasco W. Alzheimer-associated presenilins 1 and 2: neuronal expression in brain and localization to intracellular membranes in mammalian cells. Nature Med 1996;2(2):224–9. 139. Yakoub AM. Cerebral organoids exhibit mature neurons and astrocytes and recapitulate electrophysiological activity of the human brain. Neural Regen Res 2019;14(5):757–61. 140. Lancaster MA, Renner M, Martin CA, Wenzel D, Bicknell LS, Hurles ME, Homfray T, Penninger JM, Jackson AP, Knoblich JA. Cerebral organoids model human brain development and microcephaly. Nature 2013;501(7467):373–9. 141. Sloan SA, Andersen J, Pas¸ca AM, Birey F, Pas¸ca SP. Generation and assembly of human brain regionspecific three-dimensional cultures. Nature Prot 2018;13(9):2062–85. 142. McMurtrey RJ. Analytic models of oxygen and nutrient diffusion, metabolism dynamics, and architecture optimization in 3D tissue constructs with applications and insights in cerebral organoids. Tissue Eng Part C Methods 2016;22(3):221–49.

151

152

Protein Homeostasis Diseases

143. Mansour AA, Gonçalves JT, Bloyd CW, Li H, Fernandes S, Quang D, Johnston S, Parylak SL, Jin X, Cage FH. An in vivo model of functional and vascularized human brain organoids. Nat Biotechnol 2018;36(5):432–41. 144. Gonzalez C, Armijo E, Bravo-Alegria J, Becerra-Calixto A, Mays CE, Soto C. Modeling amyloid beta and tau pathology in human cerebral organoids. Mol Psych 2018;23(12):2363–74. 145. Lancaster MA, Knoblich JA. Generation of cerebral organoids from human pluripotent stem cells. Nat Protoc 2014;9(10):2329–40. 146. Akkerman N, Defize LHK. Dawn of the organoid era. Bioessays 2017;39(4):1600244. 147. Stephan FK, Zucker I. Circadian rhythms in drinking behavior and locomotor activity of rats are eliminated by hypothalamic lesions. Proc Natl Acad Sci USA 1972;69:1583–6. 148. Moore RY, Eichler VB. Loss of a circadian adrenal corticosterone rhythm following suprachiasmatic lesions in the rat. Brain Res 1972;42:201–6. 149. Maier SF, Goehler LE, Fleshner M,Watkins LR.The role of the vagus nerve in cytokine-to-brain communication. Ann NY Acad Sci 1998;840:289–300. 150. DMello C, Swain MG. Liver–brain inflammation axis. Am J Physiol Gastrointest Liver Physiol 2011;301(5):G749–61. 151. Scheperjans F, Aho V, Pereira PA, Koskinen K, Paulin L, Pekkonen E, Haapaniemi E, Kaakkola S, Eerola-Rautio J, Pohja M, Kinnunen E, Murros K, Auvinen P. Gut microbiota are related to Parkinson’s disease and clinical phenotype. Mov Disord 2015;30(3):350–8. 152. Koffie RM, Farrar CT, Saidi L-J, William CM, Hyman BT, Spires-Jones TL. Nanoparticles enhance brain delivery of blood-brain barrier-impermeable probes for in vivo optical and magnetic resonance imaging. Proc Natl Acad Sci USA 2011;108(46):18837–42. 153. Fu W,Yan C, Guo Z, Zhang J, Zhang H, Tian H, Zhu W-H. Rational design of near-infrared aggregation-induced-emission-active probes: in situ mapping of amyloid-b plaques with ultrasensitivity and high-fidelity. J Am Chem Soc 2019;141(7):3171–7.

CHAPTER 8

Molecular mechanisms of amyloid aggregation in human proteinopathies Bertrand Morel, Francisco Conejero-Lara

Department of Physical Chemistry and Institute of Biotechnology, Faculty of Sciences, University of Granada, Spain

Outline Introduction Protein aggregates: from dynamic oligomers to amyloid fibrils Amyloid fibrils Oligomers Protofibrils Mechanisms of amyloid aggregation Nucleation-Polymerization Nucleated conformational conversion More general mechanisms Application to different disease-related proteins Amyloid β peptides α-Synuclein hIAPP/amylin Concluding remarks References

154 155 155 156 157 158 158 162 163 166 167 170 172 173 174

Abbreviations Aβ  amyloid β peptide AD  Alzheimer’s disease AFM  atomic force microscopy ANS  1-anilino-8-naphthalene sulfonate AS  α-synuclein CD  circular dichroism CMC  critical micellar concentration Cryo-EM  cryo-electron microscopy FRET  Förster resonance energy transfer FT-IR  Fourier-transform infrared spectroscopy hIAPP  human islet amyloid polypeptide LENP  Lumry-Eyring nucleated polymerization NCC  nucleated conformational conversion mechanism NMR  nuclear magnetic resonance NP  nucleation-polymerization mechanism

Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00008-7 Copyright © 2020 Elsevier Inc. All rights reserved.

153

154

Protein Homeostasis Diseases

PD  Parkinson’s disease SH3  Src-homology domain 3 TEM  transmission electron microscopy ThT  thioflavine T WT  wild -type

Introduction Protein misfolding and aggregation can cause a wide range of deleterious effects in living organisms. For this reason, they have evolved complex machineries with a variety of chaperones to assist correct folding or refold misfolded proteins1 or to direct misfolded proteins to clearance by degradation2,3 (see also Chapters 2 and 6). When this stringent control of aggregation fails, accumulation of protein deposits can lead to disease. An updated list of protein deposition disorders and the aggregating proteins involved has been reviewed recently.4 A major fraction of these diseases have been related with protein amyloid aggregates. Some of them are among the most prevalent and devastating neurodegenerative disorders, such as Alzheimer’s disease (AD), Parkinson’s disease (PD), and prion-related encephalopathies. Other protein deposition diseases can be systemic (light-chain amyloidosis) or affect specific tissues or organs (type II diabetes). The number of disorders identified as associated to protein aggregation is continuously increasing (see also Chapter 3). Over the last decades, considerable advances have been made by biophysical analyses in vitro using purified proteins dissolved in simple buffers. Although these experimental conditions are usually far from those existing in living cells, the highly detailed and accurate experimental data obtained from these controlled model systems have allowed researchers to derive theoretical mechanisms about the intricate aggregation processes aimed to be later extrapolated to in vivo conditions. Biophysical studies of amyloid aggregation are preferably carried out with proteins or polypeptides related to diseases. However, these proteins are often very difficult to handle because of their high aggregation propensity, making it difficult to obtain reproducible experimental data. The discovery that almost any protein can form amyloid fibrils under the appropriate conditions has led to intensive research using proteins unrelated to deposition diseases. In fact, a large fraction of the existing structural, mechanistic, and kinetic information about amyloid aggregation has been derived from these types of studies. In this chapter, we recapitulate the main structural and morphological features of protein amyloid aggregates. Then, we summarize the classic and more advanced techniques to follow experimentally amyloid aggregation in vitro and how the kinetic information obtained has been interpreted in the light of theoretical approaches to derive fine mechanistic details of the aggregation stages under various conditions. Finally, we focus on some specific amyloid-aggregating proteins related to several important diseases.

Molecular mechanisms of amyloid aggregation in human proteinopathies

Protein aggregates: from dynamic oligomers to amyloid fibrils Amyloid-related protein aggregates are extremely diverse. Aggregate types range from small-sized (few nanometers in diameter) and unstable oligomers formed by few, disordered polypeptide chains to the long (many microns) amyloid fibrils. Between these two extremes, there is a broad variety of aggregates with diverse sizes, shapes, and structures. Amyloid aggregation is a hierarchical process in which the highly ordered structure is progressively generated by successive conformational changes and assembly events.5 Structural elements produced in early aggregation steps can be preserved and propagated to further aggregation steps, giving rise to structural and morphological polymorphisms.6 It is therefore crucial to describe in detail the different types of aggregates in order to understand their relationships along the mechanism of assembly.

Amyloid fibrils Amyloid fibrils are possibly the most prominent and best characterized type of protein aggregates. They have typical structural and morphological characteristics. Under the transmission electron microscope or the atomic force microscope (AFM), they appear as long, unbranched fibers with a length of up to several microns and a diameter of 4 to 10 nanometers (Fig. 8.1A,B). They are made of two or more intertwined protofilaments that twist along the fiber axis in a ropelike fashion.7 Amyloid fibrils produce a

Figure 8.1  Morphology of different amyloid aggregate types. (A) Transmission electron microscope (TEM) image showing a mixture of mature fibrils and protofibrils of an SH3 domain; (B) mature fibrils of amyloid β (1–40) peptide visualized by atomic force microscopy (AFM); (C) and (D) AFM image of two distinct purified oligomers of amyloid β (1-42) peptide; and (E) TEM image of early oligomers and protofibrils of α-synuclein (AS) assembled in presence of detergent. Black segments indicate 200 nm in A and E, and 500 nm in B, C, and D. Source: Part A, Adapted with permission from Ref. [14].

155

156

Protein Homeostasis Diseases

characteristic X-ray diffraction pattern with a 4.7-Å meridional reflection and a 10-Å equatorial reflection,8 indicating a cross β-sheet ultrastructure in the protofilaments in which the polypeptide chains arrange perpendicularly to the fiber axis in parallel in-register β-sheets that extend along the filament. Increasingly accurate models of the internal amyloid structure have been obtained mainly from cryo-electron microscopy (cryo-EM)9 and solid-state nuclear magnetic resonance (NMR) spectroscopy.10 The high content in β-sheet structure in the amyloid fibrils is reflected in their characteristic circular dichroism (CD) and Fourier-transform infrared spectroscopy (FT-IR) spectra.11 In addition, certain dyes such as Congo Red and the fluorescent compound thioflavine T (ThT) bind strongly to the amyloid structure.12 Amyloid-bound Congo Red produces a characteristic green birefringence under polarized light. A strong increase in fluorescence quantum yield of ThT upon binding to the amyloid structure is a widely used and sensitive probe to monitor amyloid formation.13 Despite these general properties, amyloid fibrils show considerable polymorphism. The fibers can be assembled by different number of protofilaments and show distinct twist periodicities.7 In some cases, the filaments assemble in parallel-forming flat ribbons. Fiber morphology variations are the consequence of subtle changes in molecular ultrastructure in the fibril, as a result of the different protein sequences and/or of variations in the assembly conditions. Moreover, both morphology and internal structure self-replicate in new fibrils assembled from preformed seeds.6 The amyloid fibrils described are generally defined as “mature” fibrils because they are usually isolated from amyloid deposits from patients tissues or obtained in vitro as a final product after long incubations of the protein under amyloidogenic conditions.

Oligomers The term “oligomer” generally refers to protein aggregates of relatively small size (from dimers to hundreds of monomer chains) and variable shapes that behave as soluble entities in the sense that they do not tend to precipitate. These nonfibrillar protein aggregates have become the subject of increasing interest because of their implication in the mechanisms of cell toxicity associated with disease.15,16 Oligomeric aggregates have been described as extremely diverse in size, morphology, and internal structure.17 Due precisely to this diversity, there is considerable confusion in their identification and classification, as well as in their exact role in fibril assembly and in the mechanisms of toxicity.18 The main problem resides in that oligomers are usually highly dynamic, metastable, and very difficult to isolate and characterize. The smallest oligomers have usually a low level of stable structure and tend to convert to more stable species or dissociate upon dilution. For instance, freshly prepared, solubilized amyloid β (Aβ) peptides tend to rapidly self-associate above a certain critical concentrations in micelle-like oligomers in equilibrium with the monomers.19,20 In some studies, low-molecular-weight oligomers have been captured using chemical cross-linking.21 These studies indicate that oligomers are not single species but rather a distribution of species with different oligomerization degrees.

Molecular mechanisms of amyloid aggregation in human proteinopathies

More stable oligomers containing different degrees of β-sheet structure have been described for Aβ peptides, α-synuclein (AS), and other amyloidogenic proteins. Because of their higher stability, researchers have tried to isolate and characterize them with variable success. Oligomer preparations have been purified from protein deposits in tissues or prepared in vitro from soluble protein. Globular or spheroidal oligomers of Aβ (Fig. 8.1C,D) can usually be prepared in vitro by relatively short incubations of the soluble peptide in physiological pH buffers.20,22 Similar oligomers with globular shape can also be assembled from other soluble proteins, although each one requires specific conditions of incubation or addition of external compounds.23–26 In general, there is a considerable uncertainty in the aggregation order, size, and structural content, due to the fact that oligomer samples are heterogeneous and metastable. Moreover, the techniques used to characterize them differ in the type of information provided and its interpretation, and there is a significant influence of the source (in vitro or ex vivo), protocol, and/or experimental conditions of isolation or preparation. Oligomeric species are normally partially enriched in antiparallel β-sheet structure,20,27 in contrast to the parallel β-sheet of the mature amyloid fibrils. However, spherical Aβ oligomers that are intermediates in fibrillation have been shown to contain significant amount of parallel β-sheet structure, according to solid-state NMR analysis.28 The level of β-sheet structure has been correlated with oligomer size, although there is no wellestablished quantitative relationship for this correlation.15 Oligomers can also interact with dyes, such as Congo Red and ThT, although more weakly than amyloid fibrils. Due to a significant level of exposed hydrophobicity, the oligomers can bind the fluorescent probe 1-anilino-8-naphthalene sulfonate. Due to the difficulty in isolating them in pure and stable form, high-resolution structural information for oligomeric species is very scarce. Cryo-EM image reconstruction has allowed describing AS oligomers as hollow cylindrical structures.24 Another way to characterize structurally oligomers is by their immunological response.29 Conformation-specific antibodies or sera have been produced to recognize different amyloid-related structures. An antibody named A1130 is specific of a conformational epitope typically found in toxic prefibrillar oligomers from different proteins, whereas this antibody does not recognize the regular structure of the mature amyloid fibrils. In contrast, other antibodies, like the OC polyclonal antibody, can recognize the mature fibrils31 but do not react with A11-positive oligomers. Due to its specificity for toxic prefibrillar oligomers, the A11 antibody has been widely used to identify common structural features of oligomers from different sources.

Protofibrils In situ AFM imaging has shown that globular oligomers can associate laterally to form fibrillar structures.22 These fibrillar aggregates are often obtained under specific conditions in vitro at shorter times of incubation than mature fibrils (Fig. 8.1A,E).32,33 These fibrils, sometimes termed “protofibrils” or “wormlike” fibrils, are usually thinner

157

158

Protein Homeostasis Diseases

(3–6 nm), shorter, and curvilinear and lack the characteristic twist periodicity of mature fibrils. Another morphological type of protofibril is the annular protofibril,34 which raised interest because of its pore-like shape and has been related to a cytotoxic membrane permeabilization activity.35 Other types of protofibrillar aggregates such as rodlike protofibrils32 have also been described. Protofibrils have been described to contain a lower level of internal β-sheet structure, fewer residues involved in hydrogen bonds, and a lower ThT fluorescence response than the mature fibrils.36 These structural properties are shared with smaller oligomers from which they appear to be assembled, as described below. After prolonged incubation, the protofibrils frequently convert to mature fibrils. It is still unclear, however, whether the protofibrils are on-pathway intermediates in the formation of mature fibrils, as it was suggested initially, or they rather constitute offpathway aggregates.32,37 We will come back to this question later in this chapter. To sum up, proteins can form in a wide variety of aggregates with diverse sizes and shapes and different levels of structure. Given the aforementioned hierarchical nature of the amyloid aggregation process, it is highly relevant to determine the mechanisms by which these aggregates assemble and interconvert.

Mechanisms of amyloid aggregation During the last three decades, a vast amount of research on the kinetics of amyloid aggregation has been aimed to unveil its detailed mechanism, that is, how monomers associate into different oligomer types, how the different oligomeric and prefibrillar species are interrelated, and which of these species are able to assemble hierarchically into the different fibrillar aggregates. Particular focus has been set in identifying which oligomers and protofibrillar intermediates are on and off the pathway of fibrillation. High-quality kinetic information supported by structural characterization requires appropriate techniques that report about the time evolution of each species along the aggregation pathway. Unfortunately, most biophysical techniques are not fully specific of single aggregate species but probe molecular properties that could be attributable to different aggregate types or suffer from diverse interferences. Some techniques use intrinsic probes or physical properties, whereas others require addition of external probes, labeling with dyes, chemical cross-linking, or quenching of the aggregation and sample processing to different extents.12 Table 8.1 summarizes the main techniques used to obtain data about aggregation kinetics.

Nucleation-Polymerization Early studies of the kinetics of amyloid fibril formation measuring turbidity of the solution observed a sigmoidal shape with a lag phase of variable length, followed by a sharp growth phase, in which the amount of aggregates increases, rapidly reaching a final plateau.50,51 It was also observed that the length of the lag phase could be shortened or

Molecular mechanisms of amyloid aggregation in human proteinopathies

Table 8.1  Biophysical techniques used to follow amyloid aggregation kinetics. Technique

Properties

Advantages

Static/dynamic light scattering (SLS/DLS)

Particle molecular weight Hydrodynamic radius

Noninvasive, rapid, and sensitive

Thioflavine T (ThT) fluorescence

Cross-beta amyloid structure

ANS (1-anilino8-naphthalene sulfonate) fluorescence

Exposed surface hydrophobicity

Can be monitored in situ. Proportional to mass of amyloid aggregates Can be monitored in situ

Circular dichroism (CD)

Secondary and tertiary structure

Noninvasive. Very sensitive to structural changes

Fouriertransform infrared spectroscopy (FT-IR)

Secondary structure

Nuclear magnetic resonance (NMR)

Rich and varied structural and dynamic information

Noninvasive. Applicable to precipitates. Can differentiate parallel and antiparallel β-sheet Noninvasive. Each atom is a probe. High resolution. Specific for small, soluble species

Problems and interferences

References

Signal intensity depends on particle size. Traces of large aggregates could mask small oligomers. Needs addition of dye. Signal is not fully specific of amyloid aggregates.

[38]

Unspecific of aggregate type. Needs addition of dye. ANS can induce/enhance aggregation at high concentrations. Not discriminating between different types of β-sheets. Insensitive to disordered oligomers. Low sensitivity. Needs high protein concentrations.

[40]

Needs high concentration. High-resolution methods need isotope labeling of the protein. Blind to large aggregates.

[43]

[39]

[41]

[42]

(Continued)

159

160

Protein Homeostasis Diseases

Table 8.1  Biophysical techniques used to follow amyloid aggregation kinetics. (Cont.) Technique

Properties

Advantages

Transmission electron microscopy (TEM)

Aggregate morphology

Rich morphological information about the aggregates

Atomic force microscopy (AFM)

Aggregate morphology and size

Limited proteolysis

Conformational flexibility at regional resolution

Rich size and shape information. In wet mode, it allows time-lapse visualization of aggregate growth Applicable to any protein

Hydrogen/ Deuterium exchange– mass spectrometry/ NMR Single-molecule fluorescence

Conformational flexibility at regional/ residue resolution

Very rich, highresolution information

Oligomer size and interactions

Very sensitive to low-populated oligomers

Problems and interferences

References

Needs sample fixation and staining. Nonquantitative. Smaller oligomers are difficult to detect. Slow. Nonquantitative. In dry mode, aggregates may become altered.

[44]

Difficult processing of samples. Results can depend on sequence and protease specificity. Needs careful control of back exchange. Complex processing of samples.

[46]

Needs labeling of the protein with fluorescent dyes.

[49]

[45]

[47,48]

removed by addition of preformed fibril seeds, similar to that observed in crystallization.51 On the basis of the similarity between amyloid aggregation kinetics and the kinetics of other protein polymerization processes,52,53 these observations were interpreted in terms of a nucleation-polymerization (NP) mechanism (Fig. 8.2A), in which the formation of an aggregation nucleus sets a kinetic barrier for aggregate growth. In this model, the nucleus is considered thermodynamically unstable and is defined as the smallest aggregate species that can spontaneously grow by addition of protein monomers from the bulk solution. From a kinetic point of view, the overall rate of aggregation is limited by the initial accumulation of nuclei. From a structural point of view, the NP mechanism assumes that

Molecular mechanisms of amyloid aggregation in human proteinopathies

Figure 8.2  Basic mechanisms of amyloid fibril assembly.

the monomer acquires amyloid-like structure upon its association to form the nucleus (homogeneous nucleation) or upon its addition to the fibril ends. Equations to describe quantitatively the amyloid aggregation kinetics were subsequently derived for the NP mechanism with different degrees of approximation. Ferrone et al.54 considered the unfavorable nucleation event to be in preequilibrium with the soluble monomer, whereas the subsequent addition of monomers to the growing fibril was assumed irreversible. With these and other simplifications, an approximate integrated rate equation was obtained to describe the initial phase of amyloid aggregation curves, which were predicted to follow a parabolic function of time: (1) M (t ) = 1 2 k 2 K n c ( n *+2)t 2 In Eq. (1), M(t) is the mass concentration of fibrils, k+ is the rate constant of monomer addition to the fibril, Kn is the equilibrium constant for nucleation, c is the monomer concentration, and n* is the nucleus size in monomer units. According to Eq. (1), plotting the aggregate mass versus t2 should be approximately linear with a slope that scales with c(n*+2), since the monomer concentration does not decrease significantly during the early aggregation phase. A double log plot of the slopes versus the monomer concentration yields a slope of (n* + 2), from which the nucleus size could be derived.55 It was soon recognized that simple homogeneous nucleation could not account for many observations in the aggregation kinetics. For instance, human islet amyloid polypeptide (hIAPP) and insulin fibrillation showed much sharper sigmoidal aggregation curves than those predicted by the simple NP model.56,57 These and other studies indicated the presence of additional secondary mechanisms accelerating the formation of aggregation nuclei, including fibril-catalyzed secondary nucleation, in which new nuclei are developed on the surface of the fibrils, and fibril fragmentation, which increases the number of fibril ends (Fig. 8.2B,C).

161

162

Protein Homeostasis Diseases

A simple approach to explain this self-catalytic behavior in the fibrillation kinetics was proposed by Finke and Watzky,58 who used a minimal model for self-catalytic conversion: first, a protein monomer, A, slowly converts to an aggregation nucleus B: A → B; second, B catalyzes the additional conversion of A to B: A + B → 2B.This model successfully fitted a large set of aggregation kinetics but lacks any detail about the molecular nature of the different nucleation and growth events. Knowles and coworkers further clarified the influence of secondary nucleation processes on the shape of the fibrillation kinetics. They proposed detailed models and derived analytical equations to quantitatively describe extended fibrillation kinetics followed by ThT fluorescence, including primary nucleation, fibril fragmentation,59 and fibril-catalyzed nucleation.60 Global fitting analyses allowed them to derive rate constants for each individual step and explained correctly the experimental dependencies of parameters, such as the length of the lag phase or the half-times of aggregation, τ1/2, versus the monomer concentration, c.61 For many aggregating systems, τ1/2 is proportional to cγ, where the scaling exponent γ can be obtained from a double log plot of aggregation half-times versus monomer concentration.62 The scaling exponent γ is equal to −nc/2 for aggregation dominated by a nc-order primary nucleation, whereas γ is equal to −(n2 +1)/2 for a secondary nucleation of n2 order. This allowed derivation of information about the dominant nucleation mechanism. A scaling exponent γ = −0.5 indicates a nucleation independent from the monomer, as occurs when fibril fragmentation is the dominant process, for example, in aggregation under vigorous agitation. In contrast, under quiescent conditions, primary nucleation or fibril-catalyzed nucleation can become dominant, and |γ| ≥ 1.

Nucleated conformational conversion Despite a great success in the quantitative description of the fibrillation kinetics, the earlier-described NP models cannot fully explain some experimental observations in amyloid aggregation. As described earlier, many proteins can form a variety of oligomeric aggregates and protofibrillar species, either simultaneously or before the appearance of any amyloid fibrils. Since these oligomers are generally less structured than the mature amyloid fibrils, they can pass undetected in fibrillation kinetics experiments measured by ThT fluorescence, especially at low protein concentrations. However, in many cases, oligomers can accumulate during the lag phase of fibrillation, even to large extents, and there is no clear role assigned to them in the kinetics. Since oligomerization decreases the concentration of soluble monomers, it may compete with fibrillation and alter the apparent kinetics.63 This influence in the mechanisms of amyloid aggregation cannot be undervalued because oligomers are the most likely culprits of cell toxicity related to disease. Soon the question arose about whether these oligomers and protofibrils could act as possible on-pathway intermediates in the amyloid aggregation mechanism or if they

Molecular mechanisms of amyloid aggregation in human proteinopathies

are off-pathway species. By exploring the influence of monomer concentration on the rates of nucleation, it was found that micellar oligomers favor the formation of fibrillation nuclei above certain critical concentrations. This was observed, for instance, in fibrillation of amyloid β (1–40),64 the prion form of the yeast omnipotent suppresor protein 35 (Sup35),23 or hIAPP.65 Because these “structurally fluid” oligomers appeared to form rapidly and promote fibrillation, it was necessary to add this additional step to the aggregation mechanisms (two-step nucleation). This led to the nucleated conformational conversion (NCC) mechanism (Fig. 8.2D).23 According to the NCC model, the rate-limiting nucleation step is a conformational conversion of the protein within these dynamic oligomers, leading to oligomeric amyloid nuclei. In some cases, fibril growth was observed to occur by lateral association of the oligomers in a template-assisted conversion, according to AFM imaging.22,23,66 These studies indicated that some oligomeric species are on-pathway intermediates in the fibrillation process, at least under certain aggregation conditions.

More general mechanisms The influence of this preoligomerization process on the kinetics of fibrillation has been mathematically explored. Roberts and coworkers developed a Lumry-Eyring nucleated polymerization model67,68 in which they accounted for the influence of the kinetics and thermodynamics of different steps preceding and following the nucleation of the amyloid structure, including monomer unfolding, oligomerization, conformational conversion, fibril growth by monomer addition, and fibril growth by oligomer condensation. In their developments, conformational changes in the protein monomer, such as partial or global unfolding, as well as self-association into dynamic oligomers, were considered to be rapidly preequilibrated so that their influence in the overall kinetics depends on the thermodynamic magnitudes of these processes. On the other hand, the nucleation step, i.e., the conformational conversion of an oligomer of size x to an amyloid nucleus of the same size, was considered irreversible. By exploring different kinetic regimes, they concluded that the apparent aggregation kinetics parameters, such as the observed aggregation rates or half-times, are the result of the convolution of the different kinetic stages. It was also predicted that if nucleation is significantly faster than fibril growth, the apparent aggregation kinetics is of apparent x order and shows no lag phase. Due to the strong influence of the conformational and oligomerization processes preceding the rate-limiting nucleation step on the observed overall aggregation kinetics, it was of great interest to explore these processes in more detail. This was a great challenge because the aggregation milieu is a complex mixture of multiple interchanging states in very low concentrations.69 It was therefore necessary to find methods to detect and/or stabilize the precursor states in order to characterize them and identify their role in the primary nucleation event. One approach used was the stabilization of oligomeric species by chemical cross-linking. For instance, Bitan and coworkers used

163

164

Protein Homeostasis Diseases

a photo-induced cross-linking method to “freeze” the oligomer size distribution prior to amyloid assembly.15,70,71 Another interesting approach used single-molecule fluorescence methods with dye-labeled protein to characterize low-populated oligomers during the early steps of aggregation.49,72,73 Also, formation of spherical oligomers could be selectively monitored during the aggregation lag phase using a fluorescence probe that could only bind to oligomeric assemblies or fibrils but not monomers.66 All of these approaches supported a NCC mechanism for primary formation of amyloid nuclei, even at very low concentrations.This was also supported by molecular dynamics simulations using coarse-grained models.74 These simulations suggested that classical nucleation from soluble monomers (NP) is thermodynamically disfavored and can exceptionally occur at nonphysiologically high concentrations of peptides with low self-association tendencies. On the other hand, under typically physiological concentrations, nucleation must occur by a two-step (NCC) mechanism involving prefibrillar dynamic oligomers (even if they are rarely formed under some conditions). Intermolecular contacts within these oligomers help to lower the energy barrier to facilitate the conversion to amyloid structure. Significant details about the kinetics of amyloid formation, including the nucleation mechanism, fibril structure, and oligomer toxicity, have been achieved using well-behaved model protein systems, such as the SH3 domains.47,75,76 We and others have found that amyloid aggregation of SH3 domains can be finely modulated by environmental conditions (pH, temperature, presence of salt ions, etc.)33,77,78 or mutations.79 We applied a quantitative initial rate analysis to the aggregation of the SH3 domain of α-spectrin under mild acidic conditions, where nucleation is devoid of a lag phase.80 Under a wide range of conditions, aggregation follows n-order kinetics, with variable apparent order. Initial aggregation rates measured by ThT fluorescence and their dependence of the total protein concentration were interpreted with a modified NCC model80 (Fig. 8.3A). A key assumption was an unlimited size distribution of oligomers that can form by isodesmic self-association of a partially folded intermediate. Moreover, it was assumed that amyloid nuclei could form from oligomers of any size in a rate-limiting conformational conversion step. Mathematical elaboration led to a simple equation for the initial aggregation rates: k (Bx N )2 (2 − Bx N ) (2) r0 = F K A (1 − Bx N )2 In Eq. (2), xN is the fraction of native protein, kF is the first-order rate constant for conformational conversion to amyloid nuclei, and A and B are functions of the total protein concentration and the equilibrium constants of the unfolding and self-association processes preceding the nucleation.80 Eq. (2) allowed us to fit the dependence of the initial aggregation rates with the protein concentration observed experimentally (Fig. 8.3B). In a double logarithmic plot, this dependence has a sigmoidal shape with two limiting regimes. The inflexion point

Molecular mechanisms of amyloid aggregation in human proteinopathies

Figure 8.3  Extended nucleated conformational conversion (NCC) model of amyloid nucleation. (A) Conformational and oligomerization equilibrium preceding the rate-limiting conformational conversion step. (B) Experimental initial aggregation rates of an SH3 domain as a function of concentration under different conditions. The solid lines correspond to the fittings using Eq. (2). Source: Adapted from Ref. [80] under Creative Commons Attribution (CC BY) license.

corresponds to a critical concentration Ccrit  = (KA·KI)−1, which depends only on the thermodynamic stability of the aggregating intermediate (KI) and its propensity to selfassociate into oligomers (KA). At low concentrations (C > Ccrit), the self-association equilibrium shifts to high-order oligomers, which dominate the conformational conversion to amyloid nuclei so that the kinetics tends to become first order. This limit is reminiscent of the NCC mechanism. There is a region covering a wide range of concentrations around Ccrit where the apparent kinetic order derived from the slope of the double logarithmic plot is higher than 2. In fact, the amplitude of the inflexion and its slope at Ccrit varied with the conditions.This order depended on the relative magnitude of KI and KA. Low KI (unstable intermediate) and

165

166

Protein Homeostasis Diseases

high KA (intermediate very prone to self-association) give rise to a steep inflecion, whereas the opposite leads to a flatter transition. Accordingly, the shape of these curves contains thermodynamic information about the amyloidogenic intermediate and its oligomerization process.80 In addition, our model reconciles the NP and NCC mechanisms as limiting kinetic regimes of primary nucleation operating under different conditions. Using this approach, we further analyzed the influence of the presence of different salts in the aggregation rates and how they affect the thermodynamic magnitudes of the primary nucleation mechanism.82 We found that anions, but not cations, stabilize the amyloidogenic intermediate by direct preferential binding to this state relative to the native or fully unfolded states. This preferential stabilization and its consequent increase in the intermediate concentration decrease the energy barrier for the nucleation step, strongly accelerating aggregation. Subsequently, the influence of point mutations on the initial aggregation rates was analyzed with the same model.83 A series of mutations, distributed along the sequence, were designed to produce local stability changes on each structural element of the SH3 domain. By analyzing the initial aggregation rates of each mutant at different concentrations, we could derive the Gibbs energies of formation of the amyloidogenic intermediate and its self-association process. Using a type of analysis similar to the φ-value analysis of protein folding, information was derived about the structure of the partially unfolded amyloidogenic intermediate and how it interacts to form the early oligomeric species. To sum up, these kinetic analyses provided unprecedentedly detailed information about the molecular events that precede the rate-limiting step of amyloid nucleation. This information is very important because these processes strongly determine the overall aggregation rates and kinetic regime that can lead to different aggregate species and different cytotoxic effects, as described later. Despite these details about the early oligomeric species that participate in the primary nucleation process, the role of other oligomers and protofibrillar species in the nucleation and growth of mature amyloid fibrils seems more complex than anticipated by the NP and NCC mechanisms. Recent work with hen egg-white lysozyme and dimeric Aβ peptide have shown that increasing amounts of metastable oligomers and protofibrils formed above a critical concentration seem to act as off-pathway competitors of fibrillation and even to actively inhibit nucleation of mature fibrils.84 This inhibitory activity may, however, arise from a capacity of the early oligomers to retain and sequester amyloid nuclei that are formed within them, whereas other oligomeric species may be truly off the pathway of the fibrillation track.20

Application to different disease-related proteins Having established the main amyloid aggregation mechanisms, in this section we focus on three proteins whose aggregation is related to highly prominent diseases, i.e., amyloid β (Aβ) peptides related to AD, AS associated with PD, and the hIAPP, related to type II diabetes.

Molecular mechanisms of amyloid aggregation in human proteinopathies

Amyloid β peptides Because of their relevance in AD, the Aβ peptides have been the subject of a major frac­ tion of research about the aggregation kinetics in vitro. However, the details of the Aβ aggregation mechanisms, especially the earliest events of the nucleation process, are still not completely understood, and there remain many uncertainties about the identity and role of different oligomeric intermediates and how they are interrelated in the aggregation pathway. This is very important because many types of Aβ oligomers have been described, and it is still unclear which of them are involved in different cytotoxic activities related to disease.18,85 Aβ peptides are naturally produced by proteolytic cleavage of the transmembrane amyloid precursor protein. They comprise different lengths, from 27 to 49 amino acids. Peptides 1–40 (Aβ40) and 1–42 (Aβ42) are the most abundantly produced in about a 9:1 ratio in healthy individuals. Even though Aβ40 is more abundant, Aβ42 has a higher propensity to aggregate and is the main Aβ peptide in amyloid deposits of AD patients’ brains.86 Kinetic experiments usually start from fully soluble Aβ peptides, but carefully controlled sample preparation is crucial for reproducible results because the presence of uncontrolled traces of protofibrillar aggregates can strongly accelerate the process. Protocols to prepare soluble Aβ involve a step of disaggregation of the peptide (obtained from synthetic or recombinant sources) using a concentrated denaturant solution (urea or guanidinium-HCl), organic solvent (dimethyl sulfoxide or hexafluoroisopropanol) or high pH solution (0.1 M NaOH), followed by buffer exchange and filtration to remove remaining aggregates. In addition, a size-exclusion chromatographic step is very important to remove soluble high-molecular-weight oligomers from the low-molecularweight (LMW) fraction.87 This purified LMW fraction is considered free of prefibrillar aggregates, and, although it was originally assumed to represent the monomeric peptide, several studies showed that it actually corresponds to oligomeric species in equilibrium with the monomer.20,70 LMW-Aβ is essentially disordered according to CD spectra, and recent exhaustive solution NMR analysis has indicated that both soluble Aβ40 and Aβ42 mainly populate similar random coil conformations.88 Both Aβ peptides show, however, a significant tendency to form low-order oligomers, even at low subnanomolar concentrations.89 LMW-Aβ peptides manifest as a surfactant-like behavior, with a critical micellar concentration (CMC), above which micelle-like oligomers appear in the mixture, formed by hydrophobic interactions.90 These micellar oligomers accelerate Aβ aggregation.19,20 We have recently demonstrated that under identical conditions, Aβ42 has a lower CMC (25 µM) than Aβ40 (60 µM), consistent with a higher aggregation tendency.20 The CMC values vary considerably with the environmental conditions. Knowles and coworkers carried out detailed and rigorous analyses of Aβ40 and Aβ42 aggregation kinetics followed by ThT fluorescence in vitro, spanning concentration ranges from 0.5 µM to 6 µM for Aβ4260 and from 3.5 µM to 70 µM for Aβ40.81

167

168

Protein Homeostasis Diseases

Most of these concentration ranges fall below the reported CMC values of each peptide, except for Aβ40 at the highest concentrations studied. They used rigorous analytical equations to globally fit the sigmoidal ThT fluorescence kinetics and analyzed of the concentration dependence of the experimental half-times of fibrillation. As described earlier, the scaling exponent γ reports about the dominant nucleation mechanism. In the low-concentration range, both Aβ40 and Aβ42 showed scaling exponents of −1.2 and −1.3, respectively, suggesting a mixture of second-order primary nucleation and a fibrilcatalyzed secondary nucleation. While second-order nucleation was assumed to be a fixed parameter, it does not need to represent the actual size of the fibrillation nucleus.69 In contrast to other systems, fibril fragmentation was considered negligible in the generation of amyloid nuclei of Aβ. Primary nucleation and fibril elongation are much faster for Aβ42 than for Aβ40, whereas fibril-catalyzed secondary nucleation is only marginally dissimilar,81 suggesting that the two additional hydrophobic amino acids of Aβ42 enhance interactions in primary nucleation but are not so crucial for templated conformational conversion on the fibril surface. Interestingly, at higher concentrations above ca. 30 µM, a considerable decrease in the scaling component of Aβ40 from −1.2 to about −0.2 was interpreted as a saturation of the catalytic activity of the fibrils in a Michaelis-Menten-like manner, whereas this saturation effect was not observed for Aβ42 in the 0.5 to 6 µM range. An alternative interpretation to this effect proposed that the presence of metastable oligomers at high concentrations would compete with a monomer-dependent fibril nucleation pathway by decreasing the concentration of available monomers.63,84 These studies favored a NP model for fibrillation with competing off-pathway oligomerization. However, the fact that this tendency of the scaling exponent to saturate occurs near the CMC of Aβ40 suggests an influence of the micellar oligomers on the mechanism of formation of amyloid nuclei. Instead of an inhibitory effect, several studies have proposed that the micelles can actually accelerate the nucleation and the fibrillation process of Aβ by alternative mechanisms.19,20,66 We and others have shown that near and above the CMC both Aβ40 and Aβ42 micellar oligomers undergo a progressive conversion to larger oligomers enriched in β-sheet structure.20,22 The resulting oligomers have similar characteristics to the so-called “Aβ amylospheroids,” which have strong neurotoxicity.91 Aβ42 undergoes this conversion more rapidly than Aβ40, consistent with its lower CMC. Moreover, the oligomers can be fractionated by size-exclusion chromatography. While Aβ40 mainly forms only one type of globular oligomers, Aβ42 forms two morphologically and structurally distinct ones (Fig. 8.1C,D). Globular Aβ42 oligomers are similar to those formed by Aβ40, whereas rough and irregular Aβ42 oligomers are favored at higher concentrations and longer incubations and contain more β-sheet structure. Likewise, Ladiwala et al. 92 have reported that Aβ42 can form two different oligomers with different antibody reactivity. A11-positive oligomers form only at intermediate concentrations (around 20 to 25 µM, near the CMC of Aβ42). These oligomers

Molecular mechanisms of amyloid aggregation in human proteinopathies

are hydrophobic and relatively unstable and show high toxicity to mammalian cells. At longer incubation times, these oligomers convert to A11-negative, more stable oligomers that are less hydrophobic and less toxic. Interestingly, A11-positive Aβ oligomers have been associated with cognitive decline in AD modeling transgenic mice, whereas A11-negative but OC-positive oligomers accumulate around the amyloid plaques but do not impair cognition.93 These studies indicate that the structural arrangement of the polypeptide chain within these oligomers has significant influence in their neurotoxic activity. Interestingly, the purified, spherical Aβ40 oligomers can convert rapidly to mature fibrils upon further incubation at 37°C, indicating that these oligomers preserve in their composition the necessary “ingredients” for effective fibrillation. In contrast, neither of the two different oligomers of Aβ42 could fibrillate effectively in isolation and would appear to be off-pathway aggregates of amyloid fibrillation.20 Therefore, accumulation of different oligomeric species can favor alternative pathways of fibrillation, such as lateral oligomer association22 or oligomer-seeded fibrillation.20,94 Accordingly, there seem to be diverse competing pathways of oligomerization-fibrillation, but these apparently diverging processes do not need to be fully unconnected; they could also feed each other, providing aggregation seeds or new species to accelerate or delay the fibrillation process, depending on the conditions. This complex scenario is illustrated in Fig. 8.4.

Figure 8.4  Schematic illustration of the variety of nucleation, oligomerization, and fibrillation steps of Aβ. A, Homogeneous primary nucleation; B, fibril-catalyzed secondary nucleation; C, fibril fragmentation; D, micellization; E, conformational conversion; F, fibril elongation by monomer addition; G, micelle-catalyzed fibril elongation; H, fibrillation by lateral association.

169

170

Protein Homeostasis Diseases

α-Synuclein The main component of the intraneuronal inclusions known as Lewy bodies that characterize PD is AS. Moreover, several missense mutations in the gene encoding for AS produce early-onset PD, supporting a genetic link of AS with this disease.95 The most frequent of these mutations lead to the amino acid substitutions A56T, E46K, and A30P. AS is a 140-residue, intrinsically disordered protein highly expressed in brain cells. Its function has been related to neurotransmission and synaptic plasticity.96 As opposed to Aβ peptides, AS is highly soluble and does not aggregate within weeks in standard buffers in vitro. AS can, however, aggregate rapidly in presence of added aggregate seeds, under vigorous agitation, or in the presence of some type of hydrophobichydrophilic interface (liquid-air interface,97 lipid membranes or vesicles,98 or detergent micelles25). The difficulty in controlling the influence of these interfaces in experiments makes it very difficult to obtain reproducible kinetic data of AS aggregation for rigorous analysis. Early kinetic studies reported that AS fibrillation is nucleation dependent and can be seeded by preformed nuclei.99 Under continuous shaking or stirring, AS fibrillation occurs via a partially folded intermediate.100 Two different types of oligomers, one formed during the lag phase of fibrillation and a second smaller one that coexists with the fibrils,101 have been identified. According to AFM measurements, lag-phase oligomers are spherical, 2–6 nm in height, and can give rise to annular structures.102 Single-molecule fluorescence techniques with dye-labeled AS allowed a detailed characterization of the oligomeric fraction occurring during the lag phase of fibrillation, consisting in a wide range of sizes (from dimers to more than 100 molecules), but Förster resonance energy transfer (FRET) efficiencies indicated two types of oligomeric structures, i.e., small- to medium-sized oligomers (type-A oligomers) with low resistance to protease, and medium-to-large oligomers showing more stable structure (type B oligomers) and higher resistance to proteolysis.72 The type-A oligomers form without lag phase and convert to type-B oligomers that appear after a delay and lead then to fibril formation. These results were consistent with a NCC model for fibril nucleation, essentially identical to that represented in Fig. 8.3A. In a subsequent study, stable AS oligomers prepared at high protein concentrations were found not to seed fibrillation efficiently, but they, rather, increased the lag phases of fibrillation by inhibiting the primary nucleation process.103 These oligomers were actually composed of two subpopulations with different sizes and were structurally similar to the type-B oligomers, according to protease resistance, CD structure, and FRET efficiency. Moreover, these oligomers were found to be highly neurotoxic. Highly sensitive single-molecule FRET analysis during the time course of AS aggregation allowed detection of the formation of type-A and type-B oligomers over a very wide range of initial concentrations (from 0.05 to 140 µM), and a global analysis of the data permitted the determination of the rate constants of conformational conversion.104 Interestingly, the authors concluded that, in the case of AS, templated seeding

Molecular mechanisms of amyloid aggregation in human proteinopathies

is not relevant compared to primary generation of oligomers and that, as opposed to prions, it is cellular stress that is the mechanism that appears to favor generation of toxic oligomers, which, in turn, could migrate to other cells to induce more cellular stress and, therefore, oligomer spreading. The oligomerization behavior of AS is therefore reminiscent of that previously described for Aβ, where heterogeneous oligomerization processes could lead to fibril productive or unproductive pathways. The AS aggregation mechanism described so far corresponds to in vitro experiments under continuous shaking or stirring, which act by increasing the effective air-water interface and accelerating the nucleation process. AS can also nucleate amyloid fibrillation very effectively in presence of lipids and detergent.25,98 However, in its normal function, AS interacts with lipids and acquires diverse α-helical structures on the surface of membranes, vesicles, and micelles,105–107 and it has been described that lipid binding can also inhibit AS aggregation under some conditions.108,109 Whether lipids stimulate or prevent AS aggregation depends on the type of lipid and the lipid/AS ratio.110,111 Negatively charged membranes and micelles accelerate primary nucleation of amyloid oligomers by several orders of magnitude, but this effect occurs only at relatively low molar ratios.25,98 In contrast, high lipid-to-AS ratios decrease the formation of oligomeric nuclei. It appears that both lipid-bound AS and soluble, monomeric AS in rapid exchange are necessary for amyloid nucleation. Vesicle- or micelle-enhanced, primary nucleation of AS leads to formation of spheroidal oligomers that self-associate in curvilinear protofibrils (Fig. 8.1E), 25,37 which slowly convert to mature, straight fibrils. This two-step fibrillation mechanism is similar to that found for Aβ under micelle-favoring conditions (Fig. 8.4). However, while Aβ can self-catalyze nucleation by forming micelles itself, AS needs exogenous micellar or vesicular structures as nucleation sites. Whereas most research about AS fibrillation had been carried out with recombinantly produced AS, it was found that most cellular AS is N-terminally acetylated.112 N-acetylation has been found to protect AS from fibrillation both in presence of lipids and in buffers under shaking.113,114 It appears that N-acetylation enhances N-terminal helix propensity and favors protective interactions with membrane surfaces through a long-range conformational effect. Finally, considerable attention has been placed in understanding the effects of AS mutations associated to early-onset PD. The effect of the mutations on the aggregation propensity of the AS mutants has been found to be diverse and controversial. For instance, the A53T and A30P mutations accelerate oligomer formation, while A53T also favors fibrillation, A30P retardates it compared to wild-type (WT) AS.102 The E46K mutation has also been reported to accelerate AS fibrillation.115 However, these studies were carried out in bulk solution and with nonacetylated AS. A detailed study of the kinetics lipid-induced AS aggregation has found different effects116: membrane-

171

172

Protein Homeostasis Diseases

catalyzed protofibril formation and fibril-catalyzed secondary nucleation are strongly altered by the different mutations. A54T accelerates both processes, A30P accelerates only secondary nucleation but leaves unaffected lipid-induced fibril formation, and E46K accelerates secondary nucleation but reduces the rate of lipid-catalyzed fibrillation. Other mutations like H50Q and G51D strongly inhibit both processes. In contrast, seeded fibril elongation is only weakly affected by the mutations compared to WT AS. These mutational effects observed in lipid-catalyzed fibrillation are essentially coincident with those we found for nonacetylated A54T, A30P, and E46K AS mutants in a study of SDS-induced amyloid nucleation. However, while N-acetylation reduced strongly aggregation of WT AS, the mutations abolished this protective effect.25 Anyhow, these variable effects of the early-onset PD mutants do not show any correlation with the age of onset observed for familial forms of PD, suggesting that cellular factors may modulate differently aggregation propensity or the processes leading to neuronal toxicity.

hIAPP/amylin hIAPP, also known as amylin, is a 37-residue peptide whose aggregation is widely studied because of its involvement in type II diabetes. It is the primary component of aggregates that cause β-cell death in the pancreas, followed by the reduction in insulin production. Although it is still debated, the early steps of the fibrillation process seem to be critical in the dysfunction of pancreatic cells.117 As is the case of other diseases, such as the ones described earlier in this chapter, the detailed mechanism of hIAPP aggregation and the structure and properties of the early oligomers are still not fully characterized. Indeed, far fewer studies have been undertaken on the oligomers of hIAPP than on those related to AD and PD. Although most of the information obtained for small hIAPP oligomers has been derived from computational simulations, some studies have demonstrated experimentally that formation of extended structures in the low-order oligomers is the key in early hIAPP amyloid formation.118 These early species of the aggregation process imply different structural components such as helix-helix interactions119 and β-strands associations.120,121 Two decades ago, it was reported that the aggregation of hIAPP appeared to occur via a molten globule-like conformational state containing considerable secondary structure and a large number of solvent-exposed hydrophobic patches.122 Under specific conditions, helical intermediates have been observed in the early stages of the hIAPP aggregation. Such α-helical oligomers have been found to promote or enhance the rate of hIAPP amyloid formation.123 The N-terminal segment of hIAPP forms transient α-helices in solution, and intermolecular contacts between these helices could help to nucleate the oligomers and seed the development of amyloid fibrils. Furthermore, hIAPP α-helical species were also appreciated upon interaction with micelles or lipid membranes such as observed for AS25 accelerating the amyloid formation under these

Molecular mechanisms of amyloid aggregation in human proteinopathies

conditions. However, there is no clear evidence whether the α-helix oligomers of IAPP could be considered as on- or off-pathway oligomers. Then, electron paramagnetic resonance was employed to analyze the kinetics of IAPP aggregation, and these studies have shown that prenucleation intermediates induce the formation of intra-β-sheet structure and initiate the bonding network required for the possible nucleation mechanisms of fibril assembly.124 A further study using solution NMR, AFM, and mass spectrometry found evidence for rapid initial oligomerization of hIAPP into dimeric and trimeric species driven by interactions between Histidine 18 and Tyrosine 37. These low-order oligomers further evolve into micelle-like molten globule oligomers, composed of between 16 and 300 monomers, before going on to form fibrils.125 Similar micellar aggregates of hIAPP were reported more recently as key intermediates in fibrillation by Brender and coworkers.65 In addition, this study shows that relatively small variations in the experimental conditions, such as the peptide concentration, pH, and temperature, could affect strongly the CMC. From these experiments, it has been suggested that the hIAPP nucleation could occur following two different mechanisms: a less efficient process occurring within small oligomers and a more efficient process initiated within micelle-like oligomers involving more extensive contacts throughout the peptides. The combination of 2D-infrared spectroscopy and computational techniques has also identified an oligomeric species forming during the very early stages of the aggregation process and persisting until the sigmoidal rise of the fibrils.26,120 This oligomer has been found to be composed of parallel β-sheet structure in a specific region being further assigned as a partially disordered loop in the final amyloid fibrils structure. These characteristics in the hIAPP aggregation cascade are highly similar to those described earlier in this chapter for other disease-related proteins.

Concluding remarks In this chapter, we have summarized the most relevant mechanisms of amyloid aggregation in vitro. Although amyloid fibrillation is a highly hierarchical assembly process, the steps leading from the soluble protein to the highly ordered amyloid fibrillar structure are not necessarily arranged linearly, but there are diverging routes that are sometimes interconnected with the participation of a variety of oligomeric and prefibrillar interchanging species. A basic scheme involves a nucleation step mediated by dynamic intermolecular interactions between protein monomers, sometimes catalyzed by external surfaces, which help to lower the energy barrier for conversion to amyloid structure. Oligomeric nuclei can then evolve to fibrils by monomer addition and/or lateral associations. Variations between the possible aggregation routes, key species, and kinetic regimes are sometimes protein specific and can be modulated by the aggregation conditions leading, in many cases, to aggregate polymorphism.

173

174

Protein Homeostasis Diseases

Acknowledgments We acknowledge funding and support from the Spanish Ministry of Economy and Competitivity (grants: BIO2009-07317, BIO2013-40697-R, and BIO2016-76640-R) and from the European Regional Development Fund of the European Union.

References 1. Hartl FU, Bracher A, Hayer-Hartl M. Molecular chaperones in protein folding and proteostasis. Nature 2011;475(7356):324–32. 2. Goldberg AL. Protein degradation and protection against misfolded or damaged proteins. Nature 2003;426(6968):895–9. 3. Kirkin V, McEwan DG, Novak I, et al. A role for ubiquitin in selective autophagy. Mol Cell 2009;34(3):259–69. 4. Chiti F, Dobson CM. Protein misfolding, amyloid formation, and human disease: a summary of progress over the last decade. Annu Rev Biochem 2017;86:27–68. 5. Fitzpatrick AW, Debelouchina GT, Bayro MJ, et al. Atomic structure and hierarchical assembly of a cross-beta amyloid fibril. Proc Natl Acad Sci USA 2013;110(14):5468–73. 6. Petkova AT, Leapman RD, Guo Z, et al. Self-propagating, molecular-level polymorphism in Alzheimer’s beta-amyloid fibrils. Science 2005;307(5707):262–5. 7. Jimenez JL, Nettleton EJ, Bouchard M, et al.The protofilament structure of insulin amyloid fibrils. Proc Natl Acad Sci USA 2002;99(14):9196–201. 8. Sunde M, Blake C. The structure of amyloid fibrils by electron microscopy and X-ray diffraction. Adv Protein Chem 1997;50:123–59. 9. Gremer L, Scholzel D, Schenk C, et al. Fibril structure of amyloid-beta(1-42) by cryo-electron microscopy. Science 2017;358(6359):116–9. 10. Colvin MT, Silvers R, Ni QZ, et al. Atomic resolution structure of monomorphic Abeta42 amyloid fibrils. J Am Chem Soc 2016;138(30):9663–74. 11. Bouchard M, Zurdo J, Nettleton EJ, et al. Formation of insulin amyloid fibrils followed by FTIR simultaneously with CD and electron microscopy. Protein Sci 2000;9(10):1960–7. 12. Nilsson MR. Techniques to study amyloid fibril formation in vitro. Methods 2004;34(1):151–60. 13. LeVine 3rd H. Quantification of beta-sheet amyloid fibril structures with thioflavin T. Methods Enzymol 1999;309:274–84. 14. Morel B, Varela L, Conejero-Lara F. The thermodynamic stability of amyloid fibrils studied by differential scanning calorimetry. J Phys Chem B 2010;114(11):4010–9. 15. Ono K, Condron MM, Teplow DB. Structure-neurotoxicity relationships of amyloid beta-protein oligomers. Proc Natl Acad Sci USA 2009;106(35):14745–50. 16. Cline EN, Bicca MA, Viola KL, et al. The amyloid-beta oligomer hypothesis: beginning of the third decade. J Alzheimers Dis 2018;64(s1):S567–610. 17. Breydo L, Uversky VN. Structural, morphological, and functional diversity of amyloid oligomers. FEBS Lett 2015;589(19 Pt A):2640–8. 18. Benilova I, Karran E, De Strooper B. The toxic Abeta oligomer and Alzheimer’s disease: an emperor in need of clothes. Nat Neurosci 2012;15(3):349–57. 19. Sabate R, Estelrich J. Evidence of the existence of micelles in the fibrillogenesis of beta-amyloid peptide. J Phys Chem B 2005;109(21):11027–32. 20. Morel B, Carrasco MP, Jurado S, et al. Dynamic micellar oligomers of amyloid beta peptides play a crucial role in their aggregation mechanisms. Phys Chem Chem Phys 2018;20(31):20597–614. 21. Lopes DH, Sinha S, Rosensweig C, et al. Application of photochemical cross-linking to the study of oligomerization of amyloidogenic proteins. Methods Mol Biol 2012;849:11–21. 22. Fu Z, Aucoin D, Davis J, et al. Mechanism of nucleated conformational conversion of Abeta42. Biochemistry 2015;54(27):4197–207. 23. Serio TR, Cashikar AG, Kowal AS, et al. Nucleated conformational conversion and the replication of conformational information by a prion determinant. Science 2000;289(5483):1317–21.

Molecular mechanisms of amyloid aggregation in human proteinopathies

24. Chen SW, Drakulic S, Deas E, et al. Structural characterization of toxic oligomers that are kinetically trapped during alpha-synuclein fibril formation. Proc Natl Acad Sci USA 2015;112(16):E1994–2003. 25. Ruzafa D, Hernandez-Gomez YS, Bisello G, et al.The influence of N-terminal acetylation on micelleinduced conformational changes and aggregation of alpha-synuclein. PLoS One 2017;12(5):e0178576. 26. Serrano AL, Lomont JP,Tu LH, et al. A free energy barrier caused by the refolding of an oligomeric intermediate controls the lag time of amyloid formation by hIAPP. J Am Chem Soc 2017;139(46):16748– 58. 27. Stroud JC, Liu C, Teng PK, et al. Toxic fibrillar oligomers of amyloid-beta have cross-beta structure. Proc Natl Acad Sci USA 2012;109(20):7717–22. 28. Chimon S, Shaibat MA, Jones CR, et al. Evidence of fibril-like beta-sheet structures in a neurotoxic amyloid intermediate of Alzheimer’s beta-amyloid. Nat Struct Mol Biol 2007;14(12):1157–64. 29. Glabe CG. Structural classification of toxic amyloid oligomers. J Biol Chem 2008;283(44):29639–43. 30. Kayed R, Head E, Thompson JL, et al. Common structure of soluble amyloid oligomers implies common mechanism of pathogenesis. Science 2003;300(5618):486–9. 31. O’Nuallain B, Wetzel R. Conformational Abs recognizing a generic amyloid fibril epitope. Proc Natl Acad Sci USA 2002;99(3):1485–90. 32. Gosal WS, Morten IJ, Hewitt EW, et al. Competing pathways determine fibril morphology in the selfassembly of beta2-microglobulin into amyloid. J Mol Biol 2005;351(4):850–64. 33. Morel B,Varela L, Azuaga AI, et al. Environmental conditions affect the kinetics of nucleation of amyloid fibrils and determine their morphology. Biophys J 2010;99(11):3801–10. 34. Ding TT, Lee SJ, Rochet JC, et al. Annular alpha-synuclein protofibrils are produced when spherical protofibrils are incubated in solution or bound to brain-derived membranes. Biochemistry 2002;41(32):10209–17. 35. Lashuel HA, Hartley D, Petre BM, et al. Neurodegenerative disease: amyloid pores from pathogenic mutations. Nature 2002;418(6895):291. 36. Kodali R, Wetzel R. Polymorphism in the intermediates and products of amyloid assembly. Curr Opin Struct Biol 2007;17(1):48–57. 37. Brown JWP, Meisl G, Knowles TPJ, et al. Kinetic barriers to alpha-synuclein protofilament formation and conversion into mature fibrils. Chemical Communications 2018;54(56):7854–7. 38. Murphy RM, Pallitto MM. Probing the kinetics of beta-amyloid self-association. J Struct Biol 2000;130(2–3):109–22. 39. LeVine 3rd H. Thioflavine T interaction with synthetic Alzheimer’s disease beta-amyloid peptides: detection of amyloid aggregation in solution. Protein Sci 1993;2(3):404–10. 40. Bolognesi B, Kumita JR, Barros TP, et al. ANS binding reveals common features of cytotoxic amyloid species. ACS Chem Biol 2010;5(8):735–40. 41. Micsonai A, Wien F, Kernya L, et al. Accurate secondary structure prediction and fold recognition for circular dichroism spectroscopy. Proc Natl Acad Sci USA 2015;112(24):E3095–3103. 42. Sarroukh R, Goormaghtigh E, Ruysschaert JM, et al. ATR-FTIR: a “rejuvenated” tool to investigate amyloid proteins. Biochim Biophys Acta 2013;1828(10):2328–38. 43. Karamanos TK, Kalverda AP, Thompson GS, et al. Mechanisms of amyloid formation revealed by solution NMR. Prog Nucl Magn Reson Spectrosc 2015;88–89:86–104. 44. Vadukul DM, Al-Hilaly YK, Serpell LC. Methods for structural analysis of amyloid fibrils in misfolding diseases. Methods Mol Biol 2019;1873:109–22. 45. Blackley HK, Sanders GH, Davies MC, et al. In-situ atomic force microscopy study of beta-amyloid fibrillization. J Mol Biol 2000;298(5):833–40. 46. Polverino de Laureto P, Taddei N, Frare E, et al. Protein aggregation and amyloid fibril formation by an SH3 domain probed by limited proteolysis. J Mol Biol 2003;334(1):129–41. 47. Carulla N, Zhou M, Arimon M, et al. Experimental characterization of disordered and ordered aggregates populated during the process of amyloid fibril formation. Proc Natl Acad Sci USA 2009;106(19):7828–33. 48. Alexandrescu AT. Quenched hydrogen exchange NMR of amyloid fibrils. Methods Mol Biol 2016;1345:211–22. 49. Orte A, Birkett NR, Clarke RW, et al. Direct characterization of amyloidogenic oligomers by singlemolecule fluorescence. Proc Natl Acad Sci USA 2008;105(38):14424–9.

175

176

Protein Homeostasis Diseases

50. Jarrett JT, Lansbury Jr PT. Amyloid fibril formation requires a chemically discriminating nucleation event: studies of an amyloidogenic sequence from the bacterial protein OsmB. Biochemistry 1992;31(49):12345–52. 51. Come JH, Fraser PE, Lansbury Jr PT. A kinetic model for amyloid formation in the prion diseases: importance of seeding. Proc Natl Acad Sci USA 1993;90(13):5959–63. 52. Oosawa F, Kasai M. A theory of linear and helical aggregations of macromolecules. J Mol Biol 1962;4:10–21. 53. Ferrone FA, Hofrichter J, Eaton WA. Kinetics of sickle hemoglobin polymerization. II. A double nucleation mechanism. J Mol Biol 1985;183(4):611–31. 54. Ferrone F. Analysis of protein aggregation kinetics. Methods Enzymol 1999;309:256–74. 55. Chen S, Ferrone FA,Wetzel R. Huntington’s disease age-of-onset linked to polyglutamine aggregation nucleation. Proc Natl Acad Sci USA 2002;99(18):11884–9. 56. Padrick SB, Miranker AD. Islet amyloid: phase partitioning and secondary nucleation are central to the mechanism of fibrillogenesis. Biochemistry 2002;41(14):4694–703. 57. Librizzi F, Rischel C. The kinetic behavior of insulin fibrillation is determined by heterogeneous nucleation pathways. Protein Sci 2005;14(12):3129–34. 58. Morris AM, Watzky MA, Agar JN, et al. Fitting neurological protein aggregation kinetic data via a 2-step, minimal/“Ockham’s razor” model: the Finke-Watzky mechanism of nucleation followed by autocatalytic surface growth. Biochemistry 2008;47(8):2413–27. 59. Knowles TP, Waudby CA, Devlin GL, et al. An analytical solution to the kinetics of breakable filament assembly. Science 2009;326(5959):1533–7. 60. Cohen SI, Linse S, Luheshi LM, et al. Proliferation of amyloid-beta42 aggregates occurs through a secondary nucleation mechanism. Proc Natl Acad Sci USA 2013;110(24):9758–63. 61. Meisl G, Michaels TCT, Linse S, et al. Kinetic analysis of amyloid formation. Methods Mol Biol 2018;1779:181–96. 62. Cohen SI,Vendruscolo M, Dobson CM, et al. From macroscopic measurements to microscopic mechanisms of protein aggregation. J Mol Biol 2012;421(2–3):160–71. 63. Finkelstein AV, Dovidchenko NV, Galzitskaya OV. What is responsible for atypical dependence of the rate of amyloid formation on protein concentration: fibril-catalyzed initiation of new fibrils or competition with oligomers? J Phys Chem Lett 2018;9(5):1002–6. 64. Lomakin A, Teplow DB, Kirschner DA, et al. Kinetic theory of fibrillogenesis of amyloid beta-protein. Proc Natl Acad Sci USA 1997;94(15):7942–7. 65. Brender JR, Krishnamoorthy J, Sciacca MF, et al. Probing the sources of the apparent irreproducibility of amyloid formation: drastic changes in kinetics and a switch in mechanism due to micellelike oligomer formation at critical concentrations of IAPP. J Phys Chem B 2015;119(7):2886–96. 66. Lee J, Culyba EK, Powers ET, et al. Amyloid-beta forms fibrils by nucleated conformational conversion of oligomers. Nat Chem Biol 2011;7(9):602–9. 67. Li Y, Roberts CJ. Lumry-Eyring nucleated-polymerization model of protein aggregation kinetics. 2. Competing growth via condensation and chain polymerization. J Phys Chem B 2009;113(19):7020–32. 68. Andrews JM, Roberts CJ. A Lumry-Eyring nucleated polymerization model of protein aggregation kinetics: 1. Aggregation with pre-equilibrated unfolding. J Phys Chem B 2007;111(27):7897–913. 69. Arosio P, Knowles TP, Linse S. On the lag phase in amyloid fibril formation. Phys Chem Chem Phys 2015;17(12):7606–18. 70. Bitan G, Lomakin A, Teplow DB. Amyloid beta-protein oligomerization: prenucleation interactions revealed by photo-induced cross-linking of unmodified proteins. J Biol Chem 2001;276(37):35176–84. 71. Hayden EY, Conovaloff JL, Mason A, et al. Preparation of pure populations of amyloid beta-protein oligomers of defined size. Methods Mol Biol 2018;1779:3–12. 72. Cremades N, Cohen SI, Deas E, et al. Direct observation of the interconversion of normal and toxic forms of alpha-synuclein. Cell 2012;149(5):1048–59. 73. Yang J, Dear AJ, Michaels TCT, et al. Direct observation of oligomerization by single molecule fluorescence reveals a multistep aggregation mechanism for the yeast prion protein Ure2. J Am Chem Soc 2018;140(7):2493–503. 74. Saric A, Chebaro YC, Knowles TP, et al. Crucial role of nonspecific interactions in amyloid nucleation. Proc Natl Acad Sci USA 2014;111(50):17869–74.

Molecular mechanisms of amyloid aggregation in human proteinopathies

75. Jimenez JL, Guijarro JL, Orlova E, et al. Cryo-electron microscopy structure of an SH3 amyloid fibril and model of the molecular packing. EMBO Journal 1999;18(4):815–21. 76. Bucciantini M, Giannoni E, Chiti F, et al. Inherent toxicity of aggregates implies a common mechanism for protein misfolding diseases. Nature 2002;416(6880):507–11. 77. Zurdo J, Guijarro JI, Jimenez JL, et al. Dependence on solution conditions of aggregation and amyloid formation by an SH3 domain. J Mol Biol 2001;311(2):325–40. 78. Morel B, Casares S, Conejero-Lara F. A single mutation induces amyloid aggregation in the alphaspectrin SH3 domain: analysis of the early stages of fibril formation. J Mol Biol 2006;356(2):453–68. 79. Varela L, Morel B, Azuaga AI, et al. A single mutation in an SH3 domain increases amyloid aggregation by accelerating nucleation, but not by destabilizing thermodynamically the native state. FEBS Lett 2009;583(4):801–6. 80. Ruzafa D, Morel B, Varela L, et al. Characterization of oligomers of heterogeneous size as precursors of amyloid fibril nucleation of an SH3 domain: an experimental kinetics study. PLoS One 2012;7(11):e49690. 81. Meisl G, Yang X, Hellstrand E, et al. Differences in nucleation behavior underlie the contrasting aggregation kinetics of the Abeta40 and Abeta42 peptides. Proc Natl Acad Sci USA 2014;111(26):9384–9. 82. Ruzafa D, Conejero-Lara F, Morel B. Modulation of the stability of amyloidogenic precursors by anion binding strongly influences the rate of amyloid nucleation. Phys Chem Chem Phys 2013;15(37):15508– 17. 83. Ruzafa D,Varela L, Azuaga AI, et al. Mapping the structure of amyloid nucleation precursors by protein engineering kinetic analysis. Phys Chem Chem Phys 2014;16(7):2989–3000. 84. Hasecke F, Miti T, Perez C, et al. Origin of metastable oligomers and their effects on amyloid fibril self-assembly. Chem Sci 2018;9(27):5937–48. 85. De S, Wirthensohn DC, Flagmeier P, et al. Different soluble aggregates of Abeta42 can give rise to cellular toxicity through different mechanisms. Nat Commun 2019;10(1):1541. 86. Iwatsubo T, Odaka A, Suzuki N, et al. Visualization of A beta 42(43) and A beta 40 in senile plaques with end-specific A beta monoclonals: evidence that an initially deposited species is A beta 42(43). Neuron 1994;13(1):45–53. 87. Jan A, Hartley DM, Lashuel HA. Preparation and characterization of toxic Abeta aggregates for structural and functional studies in Alzheimer’s disease research. Nat Protoc 2010;5(6):1186–209. 88. Roche J, Shen Y, Lee JH, et al. Monomeric Abeta(1-40) and Abeta(1-42) peptides in solution adopt very similar ramachandran map distributions that closely resemble random coil. Biochemistry 2016;55(5):762–75. 89. Iljina M, Garcia GA, Dear AJ, et al. Quantitative analysis of co-oligomer formation by amyloid-beta peptide isoforms. Sci Rep 2016;6:28658. 90. Soreghan B, Kosmoski J, Glabe C. Surfactant properties of Alzheimer’s A beta peptides and the mechanism of amyloid aggregation. J Biol Chem 1994;269(46):28551–4. 91. Hoshi M, Sato M, Matsumoto S, et al. Spherical aggregates of β-amyloid (amylospheroid) show high neurotoxicity and activate tau protein kinase I/glycogen synthase kinase-3β. Proc Natl Acad Sci USA 2003;100(11):6370–5. 92. Ladiwala AR, Litt J, Kane RS, et al. Conformational differences between two amyloid beta oligomers of similar size and dissimilar toxicity. J Biol Chem 2012;287(29):24765–73. 93. Liu P, Reed MN, Kotilinek LA, et al. Quaternary structure defines a large class of amyloid-beta oligomers neutralized by sequestration. Cell Rep 2015;11(11):1760–71. 94. Economou NJ, Giammona MJ, Do TD, et al. Amyloid beta-protein assembly and Alzheimer’s disease: dodecamers of abeta42, but not of abeta40, seed fibril formation. J Am Chem Soc 2016;138(6):1772–5. 95. Nussbaum RL, Polymeropoulos MH. Genetics of Parkinson’s disease. Hum Mol Genet 1997;6(10):1687– 91. 96. Cheng F, Vivacqua G, Yu S. The role of alpha-synuclein in neurotransmission and synaptic plasticity. J Chem Neuroanat 2011;42(4):242–8. 97. Campioni S, Carret G, Jordens S, et al. The presence of an air-water interface affects formation and elongation of alpha-Synuclein fibrils. J Am Chem Soc 2014;136(7):2866–75. 98. Galvagnion C, Buell AK, Meisl G, et al. Lipid vesicles trigger alpha-synuclein aggregation by stimulating primary nucleation. Nat Chem Biol 2015;11(3):229–34.

177

178

Protein Homeostasis Diseases

99. Wood SJ, Wypych J, Steavenson S, et al. alpha-synuclein fibrillogenesis is nucleation-dependent, Implications for the pathogenesis of Parkinson’s disease. J Biol Chem 1999;274(28):19509–12. 100. Uversky VN, Li J, Fink AL. Evidence for a partially folded intermediate in alpha-synuclein fibril formation. J Biol Chem 2001;276(14):10737–44. 101. Dusa A, Kaylor J, Edridge S, et al. Characterization of oligomers during α-synuclein aggregation using intrinsic tryptophan fluorescence. Biochemistry 2006;45(8):2752–60. 102. Conway KA, Lee SJ, Rochet JC, et al. Acceleration of oligomerization, not fibrillization, is a shared property of both alpha-synuclein mutations linked to early-onset Parkinson’s disease: implications for pathogenesis and therapy. Proc Natl Acad Sci USA 2000;97(2):571–6. 103. Lorenzen N, Nielsen SB, Buell AK, et al.The role of stable alpha-synuclein oligomers in the molecular events underlying amyloid formation. J Am Chem Soc 2014;136(10):3859–68. 104. Iljina M, Garcia GA, Horrocks MH, et al. Kinetic model of the aggregation of alpha-synuclein provides insights into prion-like spreading. Proc Natl Acad Sci USA 2016;113(9):E1206–1215. 105. Georgieva ER, Ramlall TF, Borbat PP, et al. Membrane-bound alpha-synuclein forms an extended helix: long-distance pulsed ESR measurements using vesicles, bicelles, and rodlike micelles. J Am Chem Soc 2008;130(39):12856–7. 106. Broersen K, van den Brink D, Fraser G, et al. alpha-synuclein adopts an alpha-helical conformation in the presence of polyunsaturated fatty acids to hinder micelle formation. Biochemistry 2006;45(51):15610–6. 107. Bisaglia M, Tessari I, Pinato L, et al. A topological model of the interaction between alpha-synuclein and sodium dodecyl sulfate micelles. Biochemistry 2005;44(1):329–39. 108. Zhu M, Fink AL. Lipid binding inhibits alpha-synuclein fibril formation. J Biol Chem 2003;278(19):16873–7. 109. Martinez Z, Zhu M, Han SB, et al. GM1 specifically interacts with alpha-synuclein and inhibits fibrillation. Biochemistry 2007;46(7):1868–77. 110. Butterfield SM, Lashuel HA. Amyloidogenic protein membrane interactions: mechanistic insight from model systems. Angew Chem Int Edit 2010;49(33):5628–54. 111. O’Leary EI, Lee JC. Interplay between alpha-synuclein amyloid formation and membrane structure. Biochim Biophys Acta Proteins Proteom 2019;1867(5):483–91. 112. Anderson JP, Walker DE, Goldstein JM, et al. Phosphorylation of Ser-129 is the dominant pathological modification of alpha-synuclein in familial and sporadic Lewy body disease. J Biol Chem 2006;281(40):29739–52. 113. Bartels T, Kim NC, Luth ES, et al. N-alpha-acetylation of alpha-synuclein increases its helical folding propensity. GM1 binding specificity and resistance to aggregation. PLoS One 2014;9(7):e103727. 114. Kang L, Moriarty GM, Woods LA, et al. N-terminal acetylation of alpha-synuclein induces increased transient helical propensity and decreased aggregation rates in the intrinsically disordered monomer. Protein Sci 2012;21(7):911–7. 115. Fredenburg RA, Rospigliosi C, Meray RK, et al. The impact of the E46K mutation on the properties of alpha-synuclein in its monomeric and oligomeric states. Biochemistry 2007;46(24):7107–18. 116. Flagmeier P, Meisl G, Vendruscolo M, et al. Mutations associated with familial Parkinson’s disease alter the initiation and amplification steps of alpha-synuclein aggregation. Proc Natl Acad Sci USA 2016;113(37):10328–33. 117. Zraika S, Hull RL,Verchere CB, et al. Toxic oligomers and islet beta cell death: guilty by association or convicted by circumstantial evidence? Diabetologia 2010;53(6):1046–56. 118. Nagel-Steger L, Owen MC, Strodel B. An account of amyloid oligomers: facts and figures obtained from experiments and simulations. Chembiochem 2016;17(8):657–76. 119. Liu G, Prabhakar A, Aucoin D, et al. Mechanistic studies of peptide self-assembly: transient alfa-helices to stable beta-sheets. J Am Chem Soc 2010;132(51):18223–32. 120. Buchanan LE, Dunkelberger EB,Tran HQ, et al. Mechanism of IAPP amyloid fibril formation involves an intermediate with a transient beta-sheet. Proc Natl Acad Sci USA 2013;110(48):19285–90. 121. Dupuis NF, Wu C, Shea JE, et al. The amyloid formation mechanism in human IAPP: dimers have beta-strand monomer-monomer interfaces. J Am Chem Soc 2011;133(19):7240–3. 122. Kayed R, Bernhagen J, Greenfield N, et al. Conformational transitions of islet amyloid polypeptide (IAPP) in amyloid formation in vitro. J Mol Biol 1999;287(4):781–96.

Molecular mechanisms of amyloid aggregation in human proteinopathies

123. Vaiana SM, Best RB,Yau WM, et al. Evidence for a partially structured state of the amylin monomer. Biophys J 2009;97(11):2948–57. 124. Ruschak AM, Miranker AD. The role of prefibrillar structures in the assembly of a peptide amyloid. J Mol Biol 2009;393(1):214–26. 125. Wei L, Jiang P, Xu W, et al. The molecular basis of distinct aggregation pathways of islet amyloid polypeptide. J Biol Chem 2011;286(8):6291–300.

179

CHAPTER 9

Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases Joana S. Cristóvãoa,b, Guilherme G. Moreiraa,b, Andreas M. Grabruckerc,d,e, Cláudio M. Gomesa,b

Biosystems and Integrative Sciences Institute, Faculty of Sciences, University of Lisbon, Lisbon, Portugal Department of Chemistry and Biochemistry, Faculty of Sciences, University of Lisbon, Lisbon, Portugal c Cellular Neurobiology and Neuro-Nanotechnology Lab, Department of Biological Sciences, University of Limerick, Limerick, Ireland d Bernal Institute, University of Limerick, Limerick, Ireland e Health Research Institute (HRI), University of Limerick, Limerick, Ireland a

b

Outline Metal ions and amyloid formation in neurodegeneration Protein misfolding and metal ions in neurodegeneration Trace metal import and homeostasis in the brain Zinc and toxic protein aggregates in Alzheimer’s disease Zinc dyshomeostasis in Alzheimer’s disease Zinc binding and aggregation of Aβ Zinc binding and tau aggregation Metal chelation therapies References

182 182 182 185 185 187 188 190 191

Abbreviations Aβ  amyloid β peptide Aβ40  amyloid β peptide 1-40 Aβ42  amyloid β peptide 1-42 AD  Alzheimer’s disease APP/PS1  amyloid precursor protein/presenilin 1 BBB  blood–brain barrier CQ  clioquinol DMT  divalent metal transporter GI  gastrointestinal MMP  metalloproteinase MPAC  metal protein attenuating compounds MT  metallothionein PHF  paired helical filaments R1  tau 244-274 R2  tau 275-305 R3  tau 306-336 Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00009-9 Copyright © 2020 Elsevier Inc. All rights reserved.

181

182

Protein Homeostasis Diseases

R4  tau 337-368 ThT  thioflavin T ZIP  zrt/Irt-like protein ZnT  zinc transporter

Metal ions and amyloid formation in neurodegeneration Protein misfolding and metal ions in neurodegeneration Protein folding is an exquisitely regulated biological process through which a polypeptide is wrapped into a specific three-dimensional conformation that is critical for biological function.1 In the cell, this purely physics- and chemistry-driven process takes place in a highly complex biochemical environment whose interference is minimized by a tight proteostasis regulatory network that operates to maintain proteins in its adequate folding status. Throughout their lifetimes, proteins, which are highly dynamic entities, are, however, subject to physiological insults that may result in their conformational destabilization and potential misfolding. Protein quality control systems do not always efficiently cope with the accumulation of misfolded proteins, for instance, when the level of damaged proteins overloads the protective systems or as a result of decaying efficiency, as seen in aging. Like proteostasis networks, other biochemical processes become slightly impaired or deregulated under stress or aging, as is the case of metal ion homeostasis. Indeed, altered metallostasis, as proteostasis defects, is a common feature across age-related neurodegenerative conditions such as Alzheimer’s disease (AD). This imbalance in the levels of metal ions creates an opportunity for aberrant interactions between metal ions and the different polypeptides implicated in aggregation phenomena in neurodegeneration, such as amyloid β (Aβ), α-synuclein or tau, among many others. Interestingly, most of these aggregating peptides are intrinsically disordered proteins and, as such, expose metal-coordinating side chains (carboxylates, imidazole, thiols, etc.). Therefore, these metal–protein interactions modulate protein dynamics and aggregation pathways, influencing amyloid formation rates as well as the formation of both on- and off-pathway oligomers, aggregates, or precursors (for an extensive discussion on these aspects, including a systematic analysis of metal binding to amyloidogenic proteins, see Ref. [2–4]). As discussed in the following sections and illustrated with the effects of zinc binding to Aβ and tau in the context of AD, metals directly influence protein aggregation, modulating cytotoxicity of the formed species, as well as themselves becoming part of the protein inclusions identified in the patient brains, which are highly enriched in iron, copper, and, in particular, zinc.

Trace metal import and homeostasis in the brain Zinc, as well as copper and iron, is an essential trace metal for humans. It has numerous biological functions, and its availability, therefore, influences many physiological processes

Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases

in various organ systems. Zinc can participate as a free (aqueous) ion in cellular signaling pathways. Besides, being bound to proteins, zinc has a structural and regulatory role controlling protein folding and protein/enzyme function. Although the average human body contains approximately 2.3 g of zinc,5 the concentration of zinc varies between different organs. Interestingly, zinc is the second most prevalent trace metal found in the human brain and is only surpassed by iron. However, heme-bound iron in blood makes a significant contribution to the iron content of the brain. In contrast, zinc is concentrated in neural tissue. Within the brain, zinc is enriched in specific brain regions such as the hippocampus,6 where a high abundance of neurotransmitter vesicles containing zinc can be found at so-called “zincergic” synapses. Thus, also on the subcellular level, zinc is not equally distributed. To achieve the differences in the levels of zinc between organs, within organs, and within cells, and to maintain this intricate balance, zinc homeostasis is established and carefully controlled by many zinc transport and buffering proteins.These proteins mediate the uptake of zinc into enterocytes in the gastrointestinal system, the release of zinc from these cells into the blood circulation, the transport of zinc within the blood, and the crossing of the blood–brain barrier, as well as uptake into neural cells, and, finally, they control the subcellular distribution of zinc. Disruption of this gut–brain interaction at any of the abovementioned barriers may significantly affect zinc homeostasis within neural cells of the brain. Zinc ions, due to their charge, cannot freely pass cellular membranes. Therefore, uptake, release, and redistribution of zinc on tissue and subcellular levels are particularly dependent on membrane transport proteins (Fig. 9.1). Twenty-four transporters are known for zinc in humans.7 These proteins belong to two families, the SLC30A (ZnT) family of zinc transporters and the SLC39A (ZIP) family. While 10 SLC30A carriers mediate the export of zinc from cells to the extracellular fluid or from cellular organelles into the cytoplasm, 14 SLC39A transporters move zinc in the opposite direction. The mechanisms of both ZIP- and ZnT-mediated zinc transport are currently not well understood, although studies suggest that ZIP transport does neither require ATP nor K+ or Na+ gradients and could thus be a facilitated process driven by a concentration gradient.8 In addition, intracellular and extracellular zinc-binding proteins such as metallothioneins (MTs)9 and S100 proteins10 contribute to the transport and redistribution of zinc. Various physiological and also pathological processes challenge zinc homeostasis. For example, trace metals, and possibly ultra-trace metals, whether they have a biological function in our body or not, influence each other. This establishes the so-called metallome, which refers to the cellular and tissue distribution of free metal ions. An increase of copper, for example, due to competition for transporters and metal-binding proteins, will impact zinc levels.11 Therefore, if a chronic depletion or chronically elevated levels of a specific trace element occur that cannot be balanced, a new trace metal profile may be

183

184 Protein Homeostasis Diseases

Figure 9.1  Zinc homeostasis is established and carefully controlled by many processes involving zinc transport and buffering proteins in different tissues. (1) Absorption: Zinc is taken up from the diet in the gastrointestinal (GI) tract by ZIP (Zrt/Irt-like protein)-type zinc transporters such as ZIP4 and other ZIP family members (i.e., ZIP1 and ZIP2). Within cells, zinc can be buffered by zinc-binding proteins such as metallothioneins (MTs). ZnT type zinc transporters (ZnT1) release zinc into the bloodstream. (2) Transport: Zinc is transported in blood mostly bound to proteins such as albumin, α-2 macroglobulin, S100, and transferrin. (3) Blood–brain barrier (BBB) crossing: probably mediated by transporters such as ZIPs and DMT1 (divalent metal transporter 1). (4) Tissue distribution: Within the brain, zinc is again actively transported from the extracellular fluid into and out of neurons and glial cells. Within neurons, zinc is distributed to specific subcellular compartments, for example, imported into lysosomes (by ZnT2) endosomes (by ZnT4), or the Golgi apparatus (by ZnT5,6, and 7). Extracellularly as well as intracellularly, zinc is bound to proteins or found as “free” zinc. Within a neuron, zinc can be locally released from MT-3 after synaptic activity and binds to hundreds of proteins. However, zinc-binding proteins can also be found in extracellular (e.g., MTs, matrix metalloproteinases [MMPs], S100B). The presence of Aβ significantly disturbs metal homeostasis in the brain. By sequestering zinc into Aβ plaques, the equilibrium between free and protein-bound, as well as intra- and extracellular zinc, is affected leading to downstream effects in various biological pathways, as well as provoking changes in the expression levels of proteins involved in zinc homeostasis.

Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases

established.12 In the brain, this may have profound effects on the activation of immune responses, inflammatory processes, synaptic signaling, synapse formation and plasticity, oxidative stress, and protein aggregation. In AD, zinc imbalance in the brain may occur through several mechanisms, such as decreased gastrointestinal absorption, the accumulation of antagonistic trace metals, alterations in the levels and function of metal import and export proteins, and the abnormal accumulation and loss of metal-binding proteins.

Zinc and toxic protein aggregates in Alzheimer’s disease The extracellular aggregation of amyloid-β (Aβ) and the hyperphosphorylation of the intracellular protein tau are the two main pathological hallmarks of AD. While both Aβ and tau are metal-binding proteins, the abnormal accumulation of the zinc-binding Aβ may especially pose a significant challenge for zinc homeostasis in the brain. Consequently, abnormal zinc homeostasis and signaling in AD and their contribution to the AD pathology have been frequently reported.12

Zinc dyshomeostasis in Alzheimer’s disease Zinc dyshomeostasis in AD was reported to occur on a systemic level resulting from a zinc deficiency that, in most cases, was measured in plasma, serum, or cerebrospinal fluid of AD patients. However, detecting subclinical zinc deficiencies in patients is challenging, and reliable biomarkers are still missing. Nevertheless, despite a high heterogeneity among the studies investigating zinc levels in AD either in demographic terms or in methodological approaches, a recent meta-analysis showed that serum and plasma zinc is significantly decreased in AD patients compared with healthy controls.13 While these results suggest decreased systemic zinc levels, most likely induced by low dietary zinc levels in elderly and particularly in institutionalized subjects,14,15 increased brain zinc levels have been reported.16 The most likely reason for this is the ability of Aβ to bind zinc.17 Besides an increase in brain zinc, zinc becomes mislocalized and sequestered into plaques.This mislocalization has been hypothesized to be responsible for the majority of the biological effects of metal dyshomeostasis in AD.18,19 The chelation of zinc by amyloid plaques results in an abnormal distribution of the trace metal,20 leading to a deficiency of zinc in the vicinity of plaques. This deficiency may be further aggravated by the deregulation of zinc-binding proteins and zinc transporters in response to the plaque pathology. For example, expression of ZnTs is significantly increased, especially in Aβ plaques, compared with the surrounding tissue in the cortex of human AD brains21 and the amyloid precursor protein/presenilin 1 (APP/PS1) mouse model for AD.22 However, the mechanisms and role of this zinc transporter regulation in AD so far remain elusive. The MT family is formed by four proteins (MT-I to MT-IV) with low molecular weights and a high content of cysteines that mediate the binding of up to seven divalent

185

186

Protein Homeostasis Diseases

metal ions (mainly copper and zinc). While MT-I and MT-II are distributed intra- and extracellularly in the brain, MT-III, which is particularly expressed in the brain, is mainly found in neurons, especially zincergic neurons.23 There, MT-III may have a particular function such as participating in the activity and zinc-dependent strengthening of postsynaptic SH3 and Multiple Ankyrin Repeat Domains 3 (SHANK3) platforms during situations requiring synaptic plasticity.24 In AD, increased expression of MT-I and MT-II, mainly a result of the presence of free radicals and cytokines, may have a neuroprotective role. However, higher levels of MTs may lead to a profound shift in the pool of free and protein-bound zinc, lowering the availability of free zinc. The decreased availability of free trace metals may be a double-edged sword. On the one hand, copper binding might protect from oxidative stress and may act antiinflammatory. On the other hand, zinc binding may further affect cellular zinc-signaling dependent pathways, such as SHANK3-dependent synapse stabilization, that are already compromised by zinc-sequestration in amyloid plaques. Also, proteins of the S100 family may contribute to zinc homeostasis in the brain25 and may be affected by MT levels and zinc dyshomeostasis. S100 alarmins may not only regulate trace metal levels but also provide a link between metal and protein homeostasis. In AD, the expression of S100B and S100A9 proteins increases in response to inflammation.26 S100B is a zinc-binding protein that inhibits Aβ aggregation27 and is known to undergo metal binding–induced conformational changes that influence its ability to bind to Aβ. Similarly, matrix metalloproteinases (MMPs) are metal-binding proteins that provide a link between metallostasis and proteostasis. MMPs are zinc-dependent endopeptidases that degrade extracellular matrix proteins. Several MMPs also influence the processing of Aβ. MMPs are generally secreted as inactive proenzymes that become activated in a zinc-dependent manner. However, MMP activity is strictly controlled, as both increased and decreased activity may have pathological effects.28 In particular, levels of MMP-2 and MMP-9 are found to be elevated by the presence of Aβ.29 MMP-2 is directly linked to Aβ in the brain, and dysfunction in this enzyme may decrease the degradation of Aβ. Therefore, MMP-2 may have a protective role in AD. Despite increased levels, zinc dyshomeostasis, for example, through sequestration of zinc in amyloid plaques, may result in high levels of inactive MMP-2 promoting Aβ aggregation.Through S100 and MMPs, it is obvious that trace metal homeostasis and Aβ levels and aggregation are tightly interconnected. Although zinc is redox inert, zinc, through its complex protein interactions, plays a role in the regulation of redox-signaling pathways. Zinc, in contrast to copper and iron, is considered an antioxidant. Although these indirect antioxidant-like effects are present only under certain conditions,30 zinc deficiency in the vicinity of plaques may facilitate oxidative damage.31 In addition, the expression of some inflammatory cytokines is zinc dependent,32 and local zinc deficiency may facilitate proinflammatory signaling in the vicinity of plaques. Therefore, understanding zinc binding to Aβ and tau is essential for

Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases

unraveling the many pathomechanisms in AD and the development of targeted treatment strategies.

Zinc binding and aggregation of Aβ Amyloid-β peptide (Aβ) is a 39- to 43-amino-acid-long fragment of the APP that is found in the form of extracellular amyloid plaques in the brain of AD patients and is associated with several pathological features of the disease.33 The most abundant and toxic Aβ forms are Aβ42 and Aβ40. Aβ is an intrinsically unstructured monomer, highly flexible and prone to undertake several partly structured conformations, including N-terminal turns and β-sheet.34–37 The internal flexibility of the disordered N-terminal domain is essential for Aβ aggregation38,39 and the main binding site for metal interactions.40–43 Aβ aggregation proceeds through the formation of oligomers and protofibrils, formed through primary nucleated growth from soluble Aβ monomers; this process is frequently dominated by secondary nucleation processes where surface-catalyzed interactions of monomers and oligomers with existing fibrils nucleate the formation of new fibrils.44 Metal ions are among the most effective modulators of Aβ folding and aggregation, notably zinc. Zinc binding to Aβ Zinc binding to Aβ peptides is extensively reported and has been reported to induce fast conformational changes leading to aggregation-prone conformers.45–47 The minimal zinc-binding site on Aβ is 6HDSGYEVHH14, and the formation of a stable homodimer with two zinc-mediated interaction interfaces of Aβ occurs due to residue H6 and segment 11EVHH14.40,48–59 Additionally, it has also been reported that Aβ can bind one or two Zn2+ per peptide, depending on the number of available coordinating histidine residues46,50,52,53,57,59 with a wide binding affinity range (from 1 to 400 µM45,52,58,60–62). The formed Aβ:Zn mononuclear complex has a hairpin-like structure that seems to involve a salt bridge between residues L28-G22 and L16D23.49,57,59,63 However, zinc can bridge the imidazole rings of histidines from different peptides in a so-called “Aβ-Zn-Aβ” configuration.59,64 Dimerization and zinc crosslinking may be the seed for zinc-driven oligomerization leading to the formation of oligomers.49,55,56,59,64,65 However, other residues of the structure also are reported as participating in the peptide conformational changes after zinc binding.66–68 Regarding the conformational changes induced by zinc binding, reports suggest that Aβ forms a more compact conformation, adopting a β-sheet-enriched conformation with hydrophobic region exposure.45,47,57,68–70 Influence of zinc on Aβ aggregation Although significant evidence suggests that zinc binding to Aβ modulates its aggregation toward a nonfibrillar pathway, there are some discrepancies in the literature. These can,

187

188

Protein Homeostasis Diseases

however, be accounted for by factors related to the source of used Aβ, preparation of initial monomeric states, temperature, agitation, and concentrations of Aβ and zinc and even the materials of the consumables used to perform the experiments. The rapid binding of zinc to Aβ42 causes immediate conformational changes that promote its fast aggregation while preventing the formation of amyloid fibrillar species.45–47,68,71–78 Indeed, the aggregation profile of Aβ42 is dramatically changed even at extremely low molar ratios, as little as 0.01 mol equivalent of Zn2+.75,78 Posttranslational modifications, such as phosphorylation of serine 8 (pS8), induce the formation of a heterodimer between Aβ and pS8-Aβ79 and reduce zinc-induced aggregation of Aβ, as pS8-Aβ suppresses zinc-driven aggregation of nonmodified Aβ,80 due to interaction between the imidazole ring of H6 and the phosphate group of S8.52 Regarding Aβ42-zinc thioflavin T–reactive species, there are reports indicating that Aβ aggregation is inhibited in the presence of an equal molar concentration of zinc,68 while others have only observed a significant decrease in the lag phase and lower final fluorescence intensities.67,74,75 The time for the depletion of half of the Aβ42 monomer pool depends on Zn2+ concentration67,71,72 and is leveled out within the lag time of apo Aβ40, suggesting that secondary pathway Aβ aggregation processes dominate in the presence of Zn2+, but are slowed down.6 Seeding experiments revealed that Aβ oligomers formed by zinc do not affect68,75 nor delay metal-free Aβ aggregation.64,71 The effect of zinc over Aβ aggregation can indeed be reversed by ethylenediaminetetraacetic acid (EDTA),65,74 and some reports indicate that removal of Zn2+ by EDTA rapidly shifts the equilibrium back to the fibrillization pathway with faster kinetics.68 However, debate persists as again other studies report contradictory evidence.75 Recent work on the Aβ40 peptide showed that carrying out an aggregation reaction in the presence of zinc ions results in the formation of a homogenous population of small-sized stable oligomers with high cross-β sheet structure content and extended hydrophobic surface patches.81 Such oligomers retain toxicity features, thus suggesting that interactions of zinc with Aβ40 modulate the aggregation pathway, opening new routes for further explorations of the effect of zinc on Aβ aggregates.

Zinc binding and tau aggregation Tau is a microtubule-associated protein that binds tubulin and contributes to critical functions related to the regulation of the cytoskeleton. Tau is an intrinsically disordered protein whose hyperphosphorylation and release from microtubules results in its prompt aggregation into stacked paired helical filaments (PHFs) and which form inclusions called neurofibrillary tangles, which are associated with several neurodegenerative conditions, notably AD. There are six tau isoforms in the human central nervous system resulting from alternative splicing, and the expressed tau protein isoforms range from 352 to 441 amino acids. The larger human tau isoform is called full-length tau, abbreviated to hTau441 or hTau40.

Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases

Tau isoforms are composed of an amino-terminal acidic domain followed by two proline-rich domains and a carboxy-terminal microtubule-binding domain. The latter comprises a variable number of repeat regions, depending on the isoform. The four tau repeat regions (R) at the C-terminal domain consist of ∼30 amino acid long residues: R1 (tau 244–274), R2 (tau 275–305), R3 (tau 306–336), and R4 (tau 337–368). Half of the tau isoforms contain three repeat regions (R1, R3, and R4), while the other half contains four repeat regions (R1, R2, R3, and R4). The R3 repeat, which is present in all isoforms, comprises the 306VQIVYK311 hexapeptide, which is an established initiator of tau fibrillation. A similar hexapeptide, 375VQIINK280, is also present in the R2 repeat. These peptide segments are highly aggregation prone and are thus found to form the core of PHFs. In recent years, the potential role of zinc binding to tau as a contributor to aggregation and tauopathies has been gaining increasing interest, and several studies that provide mechanistic insights into the process are now emerging. Zinc binding to tau Given the large size of full-length tau, several studies use tau-derived fragments as models to investigate zinc binding. Studies on the tau 244–372 fragment (128 amino acids long, comprising the R1/R2/R3/R4 repeats) established that Cys-291 (at R2) and Cys-322 (at R3) are critical for Zn2+ binding, which occurs at moderate affinity (Kd ≈ 4 µM) and which is also likely coordinated by His-330 and His-362.82 This was confirmed by subsequent studies on the K19 fragment (99 amino acids long, comprising the R1/R3/R4 repeats)83 as well as on the full-length tau.84 Influence of zinc on tau aggregation Studies on the K19 fragment (99 amino acids long) suggest that enhancement of tau aggregation in the presence of zinc may result from intramolecular arrangements involving the zinc coordinating residues, thus generating an aggregation protein K19 conformer due to weakening of the inhibitory interaction between repeat regions R3 and R4.83 Binding of Zn2+ to the R3 repeat likely exposes the aggregation-prone hexapeptide, which would be otherwise shielded due to interaction with R4. However, zinc binding to the K19 fragment does not change the fibril morphology of the formed aggregates.83 Zinc binding to the tau fragment 244–372 yields, however, differences in the morphology of the formed aggregates in a zinc-dependent fashion. At low micromolar concentrations of Zn2+, the aggregation reaction is accelerated, and the fibril morphology is maintained. Higher concentrations of Zn2+ result in granular aggregates of tau, which can nevertheless be restored to fibrils upon metal chelation.82 Fibrils formed from full-length tau in the presence of zinc and inducers such as heparin or planar aromatic dyes (Congo Red, thiazin red, and thioflavin S) are cytotoxic,84 although studies on the Tau-R3 repeat fragment indicate an aggravation of toxicity effects caused by Zn2+. While both apo Tau-R3 and Zn2+-Tau-R3 aggregates are

189

190

Protein Homeostasis Diseases

taken up by cells in cell cultures, the latter are found intracellularly at higher levels. Whether this effect is related to the reported shorter fibrils and oligomers observed in Zn2+-Tau-R3 remains to be established.

Metal chelation therapies Given the tight interactions of zinc with Aβ and the resulting pathophysiological effects, different therapeutic strategies have been developed in the past and are currently under investigation that aimed at a targeted manipulation of metal homeostasis. It has been hypothesized that through restoring the metallome balance, a decrease in oxidative stress, tau phosphorylation, Aβ aggregation, synapse loss, and inflammation, can be achieved.85 This may be achieved by chelation and/or redistribution of cellular and tissue metal ion pools, and by dietary supplementation of specific metals. Initial treatment strategies were based on metal protein attenuating compounds (MPACs), of which Clioquinol (CQ) was among the first to enter clinical trials. CQ is a small molecule that can cross the blood–brain barrier. Due to its moderate affinity for zinc, it sequesters zinc from zincbinding sites with lower affinity. However, CQ releases zinc if the concentration gradient of zinc facilitates this or donates zinc to molecules/proteins with higher-affinity zincbinding sites. In initial studies in humans, oral CQ treatment for 36 weeks of severely affected AD patients significantly prevented cognitive deterioration. In follow-up studies, hydroxyquinoline (PBT2), a more soluble derivate of CQ that improved cognitive performance and reduced Aβ load in the APP/PS1 mouse model for AD, entered phase I and II clinical trials that enrolled patients with early AD. While PBT2 was reported as being safe, the effects of PBT2 were inconsistent, although executive dysfunction was significantly reduced in participants.86 Several other compounds targeting zinc homeostasis have also been tested in clinical studies. For instance, Zinc bis-DL-hydrogen aspartate resulted in improved memory, understanding, communication, and social interaction in 8 out of 10 patients. On the other hand, compounds such as zinc methionine, reaZin, zinc oxide + cupric oxide, and zinc gluconate85,87,88 resulted only in some cognitive improvements. Other compounds that target copper homeostasis, such as Cu-(II)-orotate-dihydrate and d-penicillamine, showed limited effects, while compounds targeting aluminum (and iron) levels such as desferrioxamine mesylate85 showed promising effects, slowing the rate of decline in daily living activities significantly. Interestingly, as abnormal trace metals levels have been reported in various neurodegenerative diseases, clinical trials with metal-targeting therapeutics for Huntington’s disease and Parkinson’s disease are planned or currently performed, in part using the same compounds. Unfortunately, metal homeostasis is very complex, and due to the diversity of biological pathways influenced by trace metals, selecting the right clinical readouts in trials is difficult. In addition, many of the interventional trials in this field have been

Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases

underpowered.85 Therefore, in the future, more rigorous and controlled studies need to be performed to confirm preclinical data, where targeting metal dyshomeostasis in AD is a promising approach. Another promising strategy is the use of nanocarriers for targeted manipulation of metal homeostasis in the brain. Nanocarriers such as inorganic or organic nanoparticles can be modified to facilitate blood–brain barrier crossing and release metals or MPACs in a controlled manner.89 While the targeted increase of zinc levels in the brain is hardly achievable through dietary supplementation due to the involved physiological barriers, polymeric nanoparticles have been used to significantly increase zinc within hours in the brain of mice.90 The same nanoparticles showed promising effects such as Aβ disaggregation, a reduction of inflammation, and synapse stabilization in the APP23 mouse model for AD.91 These and other preclinical studies show that the targeted manipulation of trace metal levels through MPACs and improved drug/metal delivery systems still are among the most promising therapeutic strategies for the prevention and treatment of AD.

Acknowledgments This work was supported by Fundação para a Ciência e a Tecnologia (Portugal) through grants UID/ Multi/04046/2013 (to BioISI), PTDC/NEU-NMC/2138/2014 (to C.M.G.), IF/01046/2014 (to C.G.), PhD fellowship SFRH/BD/101171/2014 (to J.C.), and by the Bial Foundation through grant PT/FB/BL2014-343 (to C.M.G.). The authors acknowledge networking support by the COST Action TD1304, the Network for the Biology of Zinc (Zinc-Net).

References 1. Gomes CM. Protein folding: an introduction. 1st ed. Springer; 2019. 2. Leal SS, Botelho HM, Gomes CM. Metal ions as modulators of protein conformation and misfolding in neurodegeneration. Coord Chem Rev 2012;256(19–20):2253–70. 3. Cristovao JS, Santos R, Gomes CM. Metals and neuronal metal binding proteins implicated in Alzheimer’s disease. Oxid Med Cell Longev 2016;2016:9812178. 4. Gomes CM, Wittung-Stafshede P. Metal ions, protein folding, and conformational states. In: Gomes CM, Wittung-Stafshede P editors. Protein folding and metal ions. CRC Press; 2010. p. 3–11. 5. Maret W. Metallomics: the science of biometals and biometalloids. In: Arruda MAZ, editor. Metallomics: the science of biometals, vol. 1055. Springer; 2018. pp. 1–20. 6. Grochowski C, Blicharska E, Krukow P, et al. Analysis of trace elements in human brain: its aim, methods, and concentration levels. Front Chem 2019;7:115. 7. Baltaci AK,Yuce K. Zinc transporter proteins. Neurochem Res 2018;43(3):517–30. 8. Lyubartseva G, Lovell MA. A potential role for zinc alterations in the pathogenesis of Alzheimer’s disease. Biofactors 2012;38(2):98–106. 9. Ziller A, Fraissinet-Tachet L. Metallothionein diversity and distribution in the tree of life: a multifunctional protein. Metallomics 2018;10(11):1549–59. 10. Gilston BA, Skaar EP, Chazin WJ. Binding of transition metals to S100 proteins. Sci China Life Sci 2016;59(8):792–801. 11. Sensi SL, Granzotto A, Siotto M, Squitti R. Copper and zinc dysregulation in Alzheimer’s disease. Trends Pharmacol Sci 2018;39(12):1049–63.

191

192

Protein Homeostasis Diseases

12. Pfaender S, Grabrucker AM. Characterization of biometal profiles in neurological disorders. Metallomics 2014;6(5):960–77. 13. Ventriglia M, Brewer GJ, Simonelli I, et al. Zinc in Alzheimer’s disease: a meta-analysis of serum, plasma, and cerebrospinal fluid studies. J Alzheimers Dis 2015;46(1):75–87. 14. Briefel RR, Bialostosky K, Kennedy-Stephenson J, McDowell MA, Ervin RB, Wright JD. Zinc intake of the U.S. population: findings from the third National Health and Nutrition Examination Survey, 1988-1994. J Nutr 2000;130(5S Suppl.):1367S–73S. 15. Abbasi AA, Rudman D. Undernutrition in the nursing home: prevalence, consequences, causes and prevention. Nutr Rev 1994;52(4):113–22. 16. Schrag M, Crofton A, Zabel M, et al. Effect of cerebral amyloid angiopathy on brain iron, copper, and zinc in Alzheimer’s disease. J Alzheimers Dis 2011;24(1):137–49. 17. Lovell MA, Robertson JD, Teesdale WJ, Campbell JL, Markesbery WR. Copper, iron and zinc in Alzheimer’s disease senile plaques. J Neurol Sci 1998;158(1):47–52. 18. Adlard PA, Parncutt JM, Finkelstein DI, Bush AI. Cognitive loss in zinc transporter-3 knock-out mice: a phenocopy for the synaptic and memory deficits of Alzheimer’s disease? J Neurosci 2010;30(5): 1631–6. 19. Grabrucker AM, Schmeisser MJ, Udvardi PT, et al. Amyloid beta protein-induced zinc sequestration leads to synaptic loss via dysregulation of the ProSAP2/Shank3 scaffold. Mol Neurodegener 2011;6:65. 20. Roberts BR, Ryan TM, Bush AI, Masters CL, Duce JA. The role of metallobiology and amyloid-beta peptides in Alzheimer’s disease. J Neurochem 2012;120(Suppl. 1):149–66. 21. Zhang LH, Wang X, Stoltenberg M, Danscher G, Huang L, Wang ZY. Abundant expression of zinc transporters in the amyloid plaques of Alzheimer’s disease brain. Brain Res Bull 2008;77(1):55–60. 22. Zhang LH,Wang X, Zheng ZH, et al. Altered expression and distribution of zinc transporters in APP/ PS1 transgenic mouse brain. Neurobiol Aging 2010;31(1):74–87. 23. Juarez-Rebollar D, Rios C, Nava-Ruiz C, Mendez-Armenta M. Metallothionein in brain disorders. Oxid Med Cell Longev 2017;2017:5828056. 24. Grabrucker S, Jannetti L, Eckert M, et al. Zinc deficiency dysregulates the synaptic ProSAP/Shank scaffold and might contribute to autism spectrum disorders. Brain 2014;137(Pt. 1):137–52. 25. Hagmeyer S, Cristovao JS, Mulvihill JJE, Boeckers TM, Gomes CM, Grabrucker AM. Zinc binding to S100B affords regulation of trace metal homeostasis and excitotoxicity in the brain. Front Mol Neurosci 2017;10:456. 26. Swindell WR, Johnston A, Xing X, et al. Robust shifts in S100a9 expression with aging: a novel mechanism for chronic inflammation. Sci Rep 2013;3:1215. 27. Cristovao JS, Morris VK, Cardoso I, et al. The neuronal S100B protein is a calcium-tuned suppressor of amyloid-beta aggregation. Sci Adv 2018;4(6) eaaq1702. 28. Brkic M, Balusu S, Libert C,Vandenbroucke RE. Friends or foes: matrix metalloproteinases and their multifaceted roles in neurodegenerative diseases. Mediators Inflamm 2015;2015:620581. 29. Wang XX, Tan MS, Yu JT, Tan L. Matrix metalloproteinases and their multiple roles in Alzheimer’s disease. Biomed Res Int 2014;2014:908636. 30. Maret W. The redox biology of redox-inert zinc ions. Free Radic Biol Med 2019;134:311–26. 31. Perry G, Cash AD, Smith MA. Alzheimer disease and oxidative stress. J Biomed Biotechnol 2002;2(3): 120–3. 32. Vasto S, Candore G, Listi F, et al. Inflammation, genes and zinc in Alzheimer’s disease. Brain Res Rev 2008;58(1):96–105. 33. Masters CL, Bateman R, Blennow K, Rowe CC, Sperling RA, Cummings JL. Alzheimer’s disease. Nat Rev Dis Primers 2015;1:15056. 34. Sgourakis NG, Yan Y, McCallum SA, Wang C, Garcia AE. The Alzheimer’s peptides Abeta40 and 42 adopt distinct conformations in water: a combined MD/NMR study. J Mol Biol 2007;368(5):1448–57. 35. Yang M, Teplow DB. Amyloid beta-protein monomer folding: free-energy surfaces reveal alloformspecific differences. J Mol Biol 2008;384(2):450–64. 36. Ono K, Condron MM,Teplow DB. Effects of the English (H6R) and Tottori (D7N) familial Alzheimer disease mutations on amyloid beta-protein assembly and toxicity. J Biol Chem 2010;285(30):23186–97. 37. Xu L, Chen Y, Wang X. Dual effects of familial Alzheimer’s disease mutations (D7H, D7N, and H6R) on amyloid beta peptide: correlation dynamics and zinc binding. Proteins 2014;82(12):3286–97.

Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases

38. Brannstrom K, Ohman A, Nilsson L, Pihl M, Sandblad L, Olofsson A. The N-terminal region of amyloid beta controls the aggregation rate and fibril stability at low pH through a gain of function mechanism. J Am Chem Soc 2014;136(31):10956–64. 39. Morris C, Cupples S, Kent TW, et al. N-terminal charged residues of amyloid-beta peptide modulate amyloidogenesis and interaction with lipid membrane. Chemistry 2018;24(38):9494–8. 40. Istrate AN, Kozin SA, Zhokhov SS, et al. Interplay of histidine residues of the Alzheimer’s disease Abeta peptide governs its Zn-induced oligomerization. Sci Rep 2016;6:21734. 41. Miller Y, Ma BY, Nussinov R. Metal binding sites in amyloid oligomers: complexes and mechanisms. Coordin Chem Rev 2012;256(19–20):2245–52. 42. Tougu V, Tiiman A, Palumaa P. Interactions of Zn(II) and Cu(II) ions with Alzheimer’s amyloid-beta peptide. Metal ion binding, contribution to fibrillization and toxicity. Metallomics 2011;3(3):250–61. 43. Miura T, Suzuki K, Kohata N, Takeuchi H. Metal binding modes of Alzheimer’s amyloid beta-peptide in insoluble aggregates and soluble complexes. Biochemistry 2000;39(23):7024–31. 44. Cohen SI, Linse S, Luheshi LM, et al. Proliferation of amyloid-beta42 aggregates occurs through a secondary nucleation mechanism. Proc Natl Acad Sci USA 2013;110(24):9758–63. 45. Chen WT, Liao YH, Yu HM, Cheng IH, Chen YR. Distinct effects of Zn2+, Cu2+, Fe3+, and Al3+ on amyloid-beta stability, oligomerization, and aggregation: amyloid-beta destabilization promotes annular protofibril formation. J Biol Chem 2011;286(11):9646–56. 46. Noy D, Solomonov I, Sinkevich O, Arad T, Kjaer K, Sagi I. Zinc-amyloid-beta interactions on a millisecond time-scale stabilize non-fibrillar Alzheimer-related species. J Am Chem Soc 2008;130(4): 1376–83. 47. Guo J, Yu L, Sun Y, Dong X. Kinetic insights into Zn(2+)-induced amyloid beta-protein aggregation revealed by stopped-flow fluorescence spectroscopy. J Phys Chem B 2017;121(16):3909–17. 48. Kozin SA, Kulikova AA, Istrate AN, et al. The English (H6R) familial Alzheimer’s disease mutation facilitates zinc-induced dimerization of the amyloid-beta metal-binding domain. Metallomics 2015;7(3):422–5. 49. Giannozzi P, Jansen K, La Penna G, et al. Zn induced structural aggregation patterns of beta-amyloid peptides by first-principle simulations and XAS measurements. Metallomics 2012;4(2):156–65. 50. Polshakov VI, Mantsyzov AB, Kozin SA, et al. A binuclear zinc interaction fold discovered in the homodimer of Alzheimer’s amyloid-beta fragment with taiwanese mutation D7H. Angewandte Chemie 2017;56(39):11734–9. 51. Furlan S, La Penna G. Modeling of the Zn2+ binding in the 1-16 region of the amyloid beta peptide involved in Alzheimer’s disease. Phys Chem Chem Phys 2009;11(30):6468–81. 52. Kulikova AA, Tsvetkov PO, Indeykina MI, et al. Phosphorylation of Ser8 promotes zinc-induced dimerization of the amyloid-beta metal-binding domain. Mol Biosyst 2014;10(10):2590–6. 53. Alies B, Conte-Daban A, Sayen S, et al. Zinc(II) binding site to the amyloid-beta peptide: insights from spectroscopic studies with a wide series of modified peptides. Inorg Chem 2016;55(20):10499–509. 54. Kozin SA, Mezentsev YV, Kulikova AA, et al. Zinc-induced dimerization of the amyloid-beta metalbinding domain 1-16 is mediated by residues 11-14. Mol Biosyst 2011;7(4):1053–5. 55. Minicozzi V, Stellato F, Comai M, et al. Identifying the minimal copper- and zinc-binding site sequence in amyloid-beta peptides. J Biol Chem 2008;283(16):10784–92. 56. Han D, Wang H, Yang P. Molecular modeling of zinc and copper binding with Alzheimer’s amyloid beta-peptide. Biometals 2008;21(2):189–96. 57. Pan L, Patterson JC. Molecular dynamics study of Zn(abeta) and Zn(abeta)2. PloS one 2013;8(9):e70681. 58. Tsvetkov PO, Kulikova AA, Golovin AV, et al. Minimal Zn(2+) binding site of amyloid-beta. Biophys J 2010;99(10):L84–6. 59. Pietropaolo A, Satriano C, Strano G, La Mendola D, Rizzarelli E. Different zinc(II) complex species and binding modes at Abeta N-terminus drive distinct long range cross-talks in the Abeta monomers. J Inorg Biochem 2015;153:367–76. 60. Noel S, Rodriguez SB, Sayen S, Guillon E, Faller P, Hureau C. Use of a new water-soluble Zn sensor to determine Zn affinity for the amyloid-beta peptide and relevant mutants. Metallomics 2014;6(7): 1220–2. 61. Talmard C, Bouzan A, Faller P. Zinc binding to amyloid-beta: isothermal titration calorimetry and Zn competition experiments with Zn sensors. Biochemistry 2007;46(47):13658–66.

193

194

Protein Homeostasis Diseases

62. Tougu V, Karafin A, Palumaa P. Binding of zinc(II) and copper(II) to the full-length Alzheimer’s amyloid-beta peptide. J Neurochem 2008;104(5):1249–59. 63. Xu L, Wang XJ, Wang XC. Effects of Zn2+ binding on the structural and dynamic properties of amyloid B peptide associated with Alzheimer’s disease: Asp(1) or Glu(11)? Acs Chem Neurosci 2013;4(11): 1458–68. 64. Stellato F, Menestrina G, Serra MD, et al. Metal binding in amyloid beta-peptides shows intra- and inter-peptide coordination modes. European Biophys J 2006;35(4):340–51. 65. Alies B, Solari PL, Hureau C, Faller P. Dynamics of Zn(II) binding as a key feature in the formation of amyloid fibrils by Abeta11-28. Inorg Chem 2012;51(1):701–8. 66. Au DF, Ostrovsky D, Fu R,Vugmeyster L. Solid-state NMR reveals a comprehensive view of the dynamics of the flexible, disordered N-terminal domain of amyloid-beta fibrils. J Biol Chem 2019;. 67. Abelein A, Graslund A, Danielsson J. Zinc as chaperone-mimicking agent for retardation of amyloid beta peptide fibril formation. Proc Natl Acad Sci USA 2015;112(17):5407–12. 68. Lee MC,Yu WC, Shih YH, et al. Zinc ion rapidly induces toxic, off-pathway amyloid-beta oligomers distinct from amyloid-beta derived diffusible ligands in Alzheimer’s disease. Sci Rep 2018;8(1):4772. 69. Shi H, Kang B, Lee JY. Zn(2+) effect on structure and residual hydrophobicity of amyloid beta-peptide monomers. J Phys Chem 2014;118(35):10355–61. 70. Rezaei-Ghaleh N, Giller K, Becker S, Zweckstetter M. Effect of zinc binding on beta-amyloid structure and dynamics: implications for Abeta aggregation. Biophys J 2011;101(5):1202–11. 71. Miller Y, Ma B, Nussinov R. Zinc ions promote Alzheimer Abeta aggregation via population shift of polymorphic states. Proc Natl Acad Sci USA 2010;107(21):9490–5. 72. Dong J, Shokes JE, Scott RA, Lynn DG. Modulating amyloid self-assembly and fibril morphology with Zn(II). J Am Chem Soc 2006;128(11):3540–2. 73. Morgan DM, Dong J, Jacob J, et al. Metal switch for amyloid formation: insight into the structure of the nucleus. J Am Chem Soc 2002;124(43):12644–5. 74. Zhang T, Pauly T, Nagel-Steger L. Stoichiometric Zn(2+) interferes with the self-association of Abeta42: insights from size distribution analysis. Int J Biol Macromol 2018;113:631–9. 75. Matheou CJ, Younan ND, Viles JH. The rapid exchange of zinc(2+) enables trace levels to profoundly influence amyloid-beta misfolding and dominates assembly outcomes in Cu(2+)/Zn(2+) mixtures. J Mol Biol 2016;428(14):2832–46. 76. Solomonov I, Korkotian E, Born B, et al. Zn2+-Abeta40 complexes form metastable quasi-spherical oligomers that are cytotoxic to cultured hippocampal neurons. J Biol Chem 2012;287(24):20555–64. 77. Bolognin S, Messori L, Drago D, Gabbiani C, Cendron L, Zatta P. Aluminum, copper, iron and zinc differentially alter amyloid-Abeta(1-42) aggregation and toxicity. Int J Biochem Cell Biol 2011;43(6): 877–85. 78. Innocenti M, Salvietti E, Guidotti M, et al. Trace copper(II) or zinc(II) ions drastically modify the aggregation behavior of amyloid-beta1-42: an AFM study. J Alzheimers Dis 2010;19(4):1323–9. 79. Mezentsev YV, Medvedev AE, Kechko OI, et al. Zinc-induced heterodimer formation between metalbinding domains of intact and naturally modified amyloid-beta species: implication to amyloid seeding in Alzheimer’s disease? J Biomol Struct Dyn 2016;34(11):2317–26. 80. Barykin EP, Petrushanko IY, Kozin SA, et al. Phosphorylation of the amyloid-beta peptide inhibits zinc-dependent aggregation, prevents Na, K-ATPase inhibition, and reduces cerebral plaque deposition. Front Mol Neurosci 2018;11:302. 81. Mannini B, Habchi J, Chia S, et al. Stabilization and characterization of cytotoxic abeta40 oligomers isolated from an aggregation reaction in the presence of zinc ions. ACS Chem Neurosci 2018;9(12): 2959–71. 82. Mo ZY, Zhu YZ, Zhu HL, Fan JB, Chen J, Liang Y. Low micromolar zinc accelerates the fibrillization of human tau via bridging of Cys-291 and Cys-322. J Biol Chem 2009;284(50):34648–57. 83. Jiji AC, Arshad A, Dhanya SR, Shabana PS, Mehjubin CK,Vijayan V. Zn(2+) interrupts R4-R3 association leading to accelerated aggregation of tau protein. Chemistry 2017;23(67):16976–9. 84. Hu JY, Zhang DL, Liu XL, et al. Pathological concentration of zinc dramatically accelerates abnormal aggregation of full-length human Tau and thereby significantly increases Tau toxicity in neuronal cells. Biochim Biophys Acta Mol Basis Dis 2017;1863(2):414–27.

Metals and amyloid gain-of-toxic mechanisms in neurodegenerative diseases

85. Adlard PA, Bush AI. Metals and Alzheimer’s disease: how far have we come in the clinic? J Alzheimers Dis 2018;62(3):1369–79. 86. Sampson EL, Jenagaratnam L, McShane R. Metal protein attenuating compounds for the treatment of Alzheimer’s dementia. Cochrane Database Syst Rev 2012;(5) CD005380. 87. Maylor EA, Simpson EE, Secker DL, et al. Effects of zinc supplementation on cognitive function in healthy middle-aged and older adults: the ZENITH study. Br J Nutr 2006;96(4):752–60. 88. Constantinidis J. Treatment of Alzheimers-disease by zinc-compounds. Drug Develop Res 1992;27(1): 1–14. 89. Chhabra R,Tosi G, Grabrucker AM. Emerging use of nanotechnology in the treatment of neurological disorders. Curr Pharm Des 2015;21(22):3111–30. 90. Chhabra R, Ruozi B, Vilella A, et al. Application of polymeric nanoparticles for CNS targeted zinc delivery in vivo. CNS Neurol Disord Drug Targets 2015;14(8):1041–53. 91. Vilella A, Belletti D, Sauer AK, et al. Reduced plaque size and inflammation in the APP23 mouse model for Alzheimer’s disease after chronic application of polymeric nanoparticles for CNS targeted zinc delivery. J Trace Elem Med Biol 2018;49:210–21.

195

CHAPTER 10

Vitamin B6-dependent enzymes and disease Barbara Cellinia, Isabel Betancor-Fernandezb, Silvia Grottellia, Nicole Fontanab, Ilaria Bellezzaa, Eduardo Salidob Department of Experimental Medicine, University of Perugia, Perugia, Italy Department of Pathology, ITB, CIBERER, Hospital Universitario Canarias, Universidad de La Laguna, Tenerife, Spain

a

b

Outline Clinical and genetic bases of misfolding diseases and natural ligand therapies Vitamin B6 enzymes and disease Primary hyperoxaluria type 1 due to deficiency of AGT Clinical and genetic features Molecular mechanisms leading to the AGT deficit in primary hyperoxaluria type 1 The role of natural and unnatural ligands as chaperones for AGT Steps forward to new therapeutic approaches for primary hyperoxaluria type 1 Conclusions References

198 198 203 203 206 208 213 214 214

Abbreviations AADC  aromatic amino acid decarboxylase AGT  alanine-glyoxylate aminotransferase ALAS2  δ-aminolevulinate synthase 2 AOA  aminooxyacetic acid CBS  cystathionine β-synthase CHO-GO  Chinese hamster ovary cells stably expressing glycolate oxidase ESRD  end-stage renal disease LDH  lactate dehydrogenase OAT ornithine δ-aminotransferase PCs  pharmacological chaperones PH  primary hyperoxaluria PH1  primary hyperoxaluria type 1 PL  pyridoxal PLP  pyridoxal 5’-phosphate PLP-Ez  PLP-dependent enzymes PM  pyridoxamine PN  pyridoxine PNP  pyridoxine 5’-phosphate

Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00010-5 Copyright © 2020 Elsevier Inc. All rights reserved.

197

198

Protein Homeostasis Diseases

Clinical and genetic bases of misfolding diseases and natural ligand therapies The term “misfolding disease” refers to a group of disorders caused by the inability of a protein to acquire its fully folded and active conformation. The molecular mechanism by which protein misfolding can generate a pathologic condition can be either gain of function, when the misfolded protein undergoes aggregation and/or becomes toxic for the cell, or loss of function, when the organism is damaged by the loss of the functional activity proper of the misfolded protein.1–3 The first situation is typical of neurodegenerative diseases like Alzheimer’s disease and Parkinson’s disease, characterized by the progressive accumulation of proteinaceous deposits eliciting neuronal cell death.4 On the other hand, a loss of function due to protein misfolding is becoming one of the most frequent mechanisms at the basis of inherited enzymatic deficits, such as lysosomal storage disorders and phenylketonuria.2 In the latter case, pathogenic missense mutations negatively interfere with the efficiency of the phenylalanine hydroxylase folding pathway and/or reduce the stability of the folded protein, thus strongly reducing the levels of functional enzyme without significantly affecting its intrinsic catalytic activity.2 It follows that any improvement of the folding efficiency would result in an increase of catalytic activity and the consequent rescue of the pathogenic effect. Based on this rationale, many efforts have been devoted in recent years to the development of smallmolecule therapeutics acting as pharmacological chaperones (PCs), that is, ligands that specifically bind a protein and shift the equilibrium toward the correct conformation.5,6 In this regard, the use of natural ligands, such as coenzymes or coenzyme analogs, is one of the preferred roads, currently employed with success in various misfolding diseases, including phenylketonuria.7

Vitamin B6 enzymes and disease Vitamin B6 (pyridoxine; PN) is one of the main water-soluble essential micronutrients, since it is involved in a wide range of physiological processes.8 Humans, like most animals, do not have the enzymes needed for its de novo synthesis and rely solely on external supplies. Vitamin B6 is made out of six vitamers (three phosphorylated and three nonphosphorylated at the 5’ carbon) that share a central pyridine and have different groups at the 4’ position of the pyridine ring: an aldehydic group in pyridoxal (PL), an amino methyl group in pyridoxamine (PM), and a hydroxyl methyl group in PN (Fig. 10.1). PL-5’ phosphate (PLP) is the active form of vitamin B6 and represents an important cofactor of a large number of enzymes (PLP-dependent enzymes, PLP-Ez) involved in more than a hundred reactions in the human metabolism. In all PLP-Ez, the coenzyme forms a covalent aldimine bond (internal Schiff base) with the ε-amino group of a lysine residue of the apoprotein. Upon binding to the active site, the α-amino group of the

Vitamin B6-dependent enzymes and disease

Figure 10.1  Molecular structures of the vitamin B6 forms.

amino acid substrate substitutes the lysine residue generating an external Schiff base. During catalysis, PLP works like an electron sink to facilitate a variety of reactions at the α, β, and γ carbons (C-2 to C-4) of amino acid substrates.9 At the α carbon, reactions can result in exchanges between L and D amino acids (racemizations), losses of one carbon (decarboxylations), and transfers of amino groups between two different substrates (transaminations). The basic mechanism involved in these reactions was described 60 years ago and then reviewed in the context of their structures.10 As a major example, PLP is the carrier of amino groups at the active site of aminotransferases, where it undergoes reversible transformations between its aldehyde form, PLP, which can accept an amino group, and its aminated form, PM phosphate (PMP), which can donate its amino group to an α-keto acid. Aminotransferases catalyze the so-called bimolecular ping-pong reactions, in which two half-reactions occur. In the first half-reaction, an amino donor substrate reacts with the enzyme active site generating the external aldimine. Then, a quinonoid intermediate is formed when the proton at the Cα of the external aldimine is extracted by an acid/base catalyst, a step facilitated by the electrophilic properties of PLP, which can act as an electron sink. The quinonoid is finally reprotonated at the C4’ of the coenzyme producing the α-keto acid and the enzyme in PMP form. In the second half-reaction, an incoming α-keto acid (amino acceptor) binds and, by following the same steps of the first half-reaction but in reverse order, is converted to the corresponding amino acid and leaves the active site regenerating the enzyme in the PLP form (Fig. 10.2) This electron sink effect of the pyridine portion of PLP has been considered central to the ping-pong mechanism, but it is not universally accepted.11 With the exception of PLP-dependent phosphorylases, which make use of the PLP phosphate group as acid/base catalysts, a large number of PLP-Ez generate a quinonoid intermediate, with variations depending on which group of the substrate Cα atom is cleaved off. Consequently, a common ancestor has been suggested for the majority of PLP-Ez.12 A grouping into α, β and γ classes, according to the carbon atom involved, has been also proposed, with sequence alignments supporting distant relation between α and γ classes and an independent evolution of the β group.13 More recently, taking into account structural and sequence information, PLP-Ez have been divided into four related classes in one family and the PLP-dependent phosphorylases in the

199

200

Protein Homeostasis Diseases

Figure 10.2  General catalytic mechanisms of aminotransferase reactions.

other.10 The five different structural folds characteristic of each of these groups have been reviewed.14,15 PLP plays a crucial role as prosthetic group for enzymes involved in numerous aspects of cellular metabolism. Its roles in important aspects of physiology, such as the biosynthesis of amino acids, tetrapyrroles, and neurotransmitters, as well as the degradation of cellular storage compounds, have been recently reviewed elsewhere.16 Nonetheless, PLP also plays a chaperone role because its binding influences the folding of many apoenzymes. Evidences supporting this notion have been provided at both protein and cellular levels. PLP is known to assist the folding of numerous PLP-Ez.17 For instance, it is well-known that cystalysin, a hemolytic protein produced by Treponema denticola, requires PLP for proper protein folding18 as well as for the stabilization of the dimeric structure.19 Among aminotransferases, it is also known that Escherichia coli aspartate aminotransferase requires PLP for proper stabilization in a native structure.20 On the other hand, bacterial serine hydroxymethyltransferase (SHMT), which takes part of the folate cycle in E. coli and Bacillus subtilis, does not require PLP for folding but only requires it for proper functioning of the protein.21 Nevertheless, PLP binding triggers a conformational change toward an ordered structure and favors the tetrameric arrangement in human mitochondrial SHMT.22,23 Very recently, cryo-electron microscopy analyses have demonstrated that PLP binding to human SHMT also affects the interaction of the protein with the deubiquitylating BRCA1/BRCA2-containing complex subunit 3 (BRCC36) isopeptidase complex, a newly discovered effect that regulates inflammatory responses.24 An important practical consequence of a better understanding of the role of PLP as a PC derives from the fact that several inborn errors of metabolism affect specific PLP-Ez, such as the deficits of δ-aminolevulinate synthase 2 (ALAS2), cystathionine β-synthase (CBS), ornithine δ-aminotransferase (OAT), aromatic amino acid decarboxylase

Vitamin B6-dependent enzymes and disease

(AADC), and alanine-glyoxylate aminotransferase (AGT) (reviewed by Refs. [17,25]). Some patients with inborn errors affecting PLP-Ez are PN responsive, even when they bear mutations that do not directly impair the binding of PLP to the enzyme. Thus, it was proposed that PLP does not only play a role in mediating catalytic activity but also in assisting protein folding and stability. For any particular pathogenic mutation affecting a PLP-Ez, more than one mechanism may be involved in explaining the molecular effects of PN, similar to what is known about the responsiveness of patients affected by phenylketonuria to tetrahydrobiopterin.26 Traditionally, empirical administration of pharmacological doses of vitamin B6 has shown highly variable response, from 90% in X-linked sideroblastic anemia (δ-aminolevulinate synthase deficiency) through 50% in homocystinuria (cystathionine β-synthase deficiency) to 5% in ornithinemia associated with gyrate atrophy (OAT deficiency).27 In some cases, such as infants with intractable epilepsy, pharmacological doses of vitamin B6 or PLP can be life-saving, but a better understanding of the structural mechanisms of disease is needed, since these medications can also have side effects such as peripheral neuropathy. In the following paragraphs, we review the role of vitamin B6 on several inborn errors of metabolism, with particular attention to primary hyperoxaluria type I (PH1), where we have concentrated most of our research efforts. X-linked sideroblastic anemia (Online Mendelian Inheritance in Man [OMIM] # 300751) is due to mutations in the ALAS2 gene, coding for a PLP-Ez that catalyzes the condensation of succinyl-CoA and glycine to produce δ-aminolevulinic acid, needed for heme synthesis. With variable age at onset, from birth to the ninth decade, but most commonly around the second decade of life, typically male patients present with hypochromic, microcytic anemia, a sideroblastic marrow, perinuclear mitochondrial iron deposits in erythrocyte precursors, and problems caused by iron overload. Most of the known mutations in the ALAS2 gene are missense substitutions and concentrate around the catalytic domain. Functional mutation analysis of the PN-responsive p.I471N mutation, in the same exon as the PLP-binding site, has shown that the apo-enzyme has reduced stability that could be improved by the cofactor.28 Other PN-responsive missense mutations include p.Y199H, p.R411C, p.R448Q, and p.R452C,29 while the change in p.D190V has been associated with PN resistance due to failure of mitochondrial import of the mutant enzyme.30 Homocystinuria due to CBS deficiency (OMIM # 236200) is an autosomal recessive disorder of sulfur metabolism, presenting typically in the first or second decade of life with mental retardation associated with Marfan-like eye, skeletal anomalies, and thromboembolic episodes. Biochemical features include increased urinary homocysteine and methionine. CBS catalyzes the condensation of serine with homocysteine. PLP is a cofactor for the CBS enzyme, and some patients respond to pharmacological doses of PN, resulting in a milder clinical phenotype31 with decreased plasma methionine levels and elimination of homocysteine from plasma and urine. With exceptions, a general correlation

201

202

Protein Homeostasis Diseases

between residual hepatic CBS activity and PN responsiveness has been reported.32,33 A chaperone mechanism of action for PLP on CBS stability has been suggested after observing that PN administration, in some patients, resulted in a significant increase in enzymatic activity in liver biopsies.32 Thus, the use of PLP precursor and other native state ligands for CBS is currently a promising therapeutic option for this disease (two more recent reviews: Refs. [34,35]). Gyrate atrophy of the choroid and retina due to deficiency of OAT (OMIM # 258870) presents with progressive chorioretinal degeneration and early cataract formation. Characteristic chorioretinal atrophy with progressive constriction of the visual fields leads to blindness before the sixth decade of life. The main sign of the disease is the accumulation of L-ornithine in plasma and, in some cases, also in urine, due to the deficit of OAT, which is involved in L-ornithine degradation in adults.36 Nonetheless, the molecular mechanisms underlying the disease, and in particular, the damage to retinal cells caused by ornithinemia, are not fully understood.37 OAT is a peculiar PLP-Ez in that it catalyzes a δ-aminotransferase reaction in which the δ-amino group of L-ornithine is transferred to α-ketoglutarate, generating glutamate semialdehyde and glutamate, respectively.38 The PLP cofactor is covalently bound to Lys292 and anchored at the active site through typical interactions with residues such as Asp263, which makes an ion pair with the pyridine nitrogen, Phe177, which generates a hydrophobic interaction with the pyridine ring, and Thr322, which is hydrogen bonded to the phosphate group.39 Recently, a deep investigation of the biochemical features of the human recombinant enzyme, along with site-directed mutagenesis approaches, have conclusively demonstrated that the functional unit of OAT is the dimer, but the protein assumes a tetrameric quaternary structure in solution. The overall structure and the quaternary assembly is strongly stabilized by PLP binding, thus suggesting that the coenzyme plays both a prosthetic and a chaperone role.40 Clinical data have shown that some cases of OAT deficiency are B6 responsive41–43 and that responsive patients display higher OAT activity in cell homogenates than unresponsive ones. One of the B6-responsive mutations is the p.A226V mutation, which probably interferes with cofactor binding due to increased steric hindrance. Patients bearing the p.V332M mutation are also responsive. OAT activity in fibroblast obtained from these patients increases when high concentrations of PLP are added to the culture media, and the authors have proposed a chaperoning mechanism for PLP to either enhance enzyme activity or decrease its degradation.44 The molecular and cellular effects of the p.V332M mutation in human OAT have been analyzed.45 The results indicate that the p.V332M variant does not display significant impairments in the overall transamination reaction, but during the catalytic cycle, it converts into the apo-form, which is prone to unfolding and aggregation.Vitamin B6 has been shown to play a chaperone role in this mutation, partly counteracting the folding defect, while it has no beneficial effects on mutations at the active site such as p.R180T.

Vitamin B6-dependent enzymes and disease

AADC deficiency (OMIM # 608643) is an autosomal recessive neurometabolic disorder that leads to severe hypotonia, movement disorders, developmental delay, and autonomic perturbations.46 The disease manifests in childhood, with a clinical phenotype ranging from mild to severe. More than 100 patients have been identified so far. The frequency of the disease is unknown, although higher incidence has been found in some regions of Taiwan and Japan due to founder effect.47 Patients are usually treated with monoamine oxidase inhibitors to improve the half-lives of the neurotransmitters dopamine and serotonin, or dopamine agonists to mimic their action.48,49 Some patients respond to the treatment with vitamin B6, but the relationship between responsiveness and patient genotype is far from being clarified.46,50 Interestingly, a phase one-half clinical trial has been recently concluded that indicates a good level of safety and tolerability of a gene therapy approach using adeno-associated viral vectors.51 AADC catalyzes the decarboxylation of 3,4-dihydroxyphenylalanine and 5-hydroxytryptophan to dopamine and serotonin, respectively.The enzyme represents the target of carbidopa, one of the drugs currently in use for the treatment of Parkinson’s disease.52 In recent years, human AADC has received great attention, and many efforts have been dedicated to the study of its biochemical properties as the base to unravel the effects of pathogenic mutations.48 In particular, analyses performed on the recombinant purified protein have allowed categorization of pathogenic mutations based on their effect on the structural and/or functional properties of the enzyme, as well as to suggest the most appropriate therapeutic strategy.53–55 It has been suggested that the PLP coenzyme could behave as PC for mutated forms of the enzyme that are prone to proteasomal degradation.54 The structural role of the coenzyme is also corroborated by crystallographic studies indicating that PLP binding induces a significant structural change in the protein, switching from an open to a closed conformation.56 This allows speculation that the apo-form could be more prone to be degraded inside the cell, in line with data obtained in cell model systems,57 and that responsive mutations could be those involving residues that play a key role during AADC folding.

Primary hyperoxaluria type 1 due to deficiency of AGT Clinical and genetic features PHs are a group of recessive inborn errors of glyoxylate metabolism characterized by overproduction of oxalate mainly in the liver and massive excretion of oxalate in urine.58 Oxalate, a dicarboxylic acid with high affinity for calcium, is an end product of metabolism without a known biological role in mammals. Although oxalate can be also absorbed in the gut from the diet, most of the oxalate produced by PH patients is generated by hepatocytes from glycolate, glyoxylate, hydroxyproline, and other precursors. Under normal circumstances, oxalate is efficiently removed by the kidneys and excreted in the urine without causing adverse effects. But the high oxalate excretion in

203

204

Protein Homeostasis Diseases

PH patients causes frequent calcium oxalate stone formation (urolithiasis) and deposits in the renal parenchyma (nephrocalcinosis), which result in kidney injury and eventual loss of renal function. At this point, oxalate levels increase dramatically in plasma, and widespread oxalate deposition (oxalosis) can be lethal unless aggressive measures are taken, such as double liver-kidney transplantation. To date, three types of PH have been characterized genetically and biochemically, but this review deals only with the most common form, PH type 1 (PH1), caused by mutations of the AGXT gene encoding liver-peroxisomal AGT (Fig. 10.3).58,59 Since glyoxylate cannot be converted to glycine, it is oxidized to oxalate by lactate dehydrogenase (LDH). Glycolate levels also increase and can be used as a secondary metabolite marker. More than 300 pathogenic (or likely pathogenic) mutations have been described, involving all exons (https://www.ncbi.nlm.nih.gov/clinvar).The majority are point mutations (missense, nonsense, and splice-site mutations), and approximately a quarter are minor deletions or insertions. The so-called “minor allele” haplotype has a number of polymorphic variants, including the p.P11L substitution (rs34116584, with allelic frequency

Figure 10.3  (A) Frequent mutated sites are represented in the three-dimensional (3D) structure of AGXT (Protein Data Bank ID: 2YOB). Left panel: using the mutagenesis tool by Pymol, we mutated proline in position 11 to leucine; glycine in position 170 to arginine; and isoleucine in position 244 to threonine. Right panel: AGXT 3D structure is shown. (B) AGXT exons and the principal PH1-associated mutations. Notice in red those presenting a higher allelic frequency: Gly170Arg, Ile244Thr, and Phe252Ile. (C) Classification of PH1-associated mutations based on their pathogenic mechanism (color online).

Vitamin B6-dependent enzymes and disease

Figure 10.3  (Continued)

ranging between 0.05 and 0.2 in the various populations studied). This minor AGT allele has a reduced enzymatic capacity and introduces a weak mitochondrial targeting signal that, in conjunction with other pathogenic amino acid substitutions (such as p.G170R, abbreviated as p.G170R-Mi), causes peroxisome-to-mitochondrion mistargeting of the AGT enzyme.60,61 No universal genotype-phenotype correlation has been found in PH1,62–64 except for patients homozygous for the p.G170R mutation, which appears to predict PN responsiveness. In addition, two studies have suggested a correlation between the type of mutation and the rate of progression to end-stage renal disease (ESRD), with best outcomes for p.G170R homozygous patients.63,65

205

206

Protein Homeostasis Diseases

Onset of PH1 symptoms is also delayed in p.G170R homozygotes compared to other genetic backgrounds. The benefit of p.G170R homozygosity may relate to lower oxalate excretion rates in this group in addition to marked reduction in urinary oxalate excretion when PN treatment is given, as recommended in current guidelines.66,67 PN is the only medication that is effective in lowering urinary oxalate in PH1, an effect that has been known for greater than 40 years. However, it is only efficacious in some patients, and the molecular basis for its effect is only partially understood. PN is a precursor of the AGT cofactor and can correct the localization of AGT in approximately one-third of PH1 patients.68,69 Responsiveness to the treatment, which has been defined by >30% reduction in urinary oxalate, is associated with specific mutations, as described earlier, and is often related to a favorable clinical outcome. PH1 patients may demonstrate a partial or a complete response.The latter is typically seen in patients homozygous for the p.G170R mutation of AGXT, who may demonstrate reduction in urine oxalate to near normal when receiving pharmacologic doses of PN. All suspected PH1 patients should be given PN pending genetic diagnosis. The doses range from 5 mg/kg/day up to doses of 20 mg/kg/day. Dose titration should be followed by assessment of effect on urinary oxalate.Very high doses may be associated with peripheral neuropathy. The trial treatment to assess responsiveness lasts for 3 months, and patients responding to PN are kept on this medication thereafter.66 However, responsiveness and sensitivity are difficult to assess in patients with chronic kidney disease/ESRD.

Molecular mechanisms leading to the AGT deficit in primary hyperoxaluria type 1 AGT is a dimeric PLP-dependent enzyme expressed in hepatocytes, where it catalyzes the almost irreversible transamination of L-alanine and glyoxylate to pyruvate and glycine, respectively.70 It belongs to the fold type I family of PLP-Ez, and each monomer is formed by a large N-terminal domain (residues 22 through 282) containing the active site and the dimerization interface, and a small domain (residues 283 through 392) (Fig. 10.4). The first 21 amino acids generate an extension connecting one monomer to the other.71 The protein is normally imported into the peroxisomal matrix, thanks to a C-terminal tripeptide and other ancillary signals. However, the “minor allele” (with p.P11L) has been proposed to favor a putative N-terminal mitochondrial-targeting sequence.59 Important interactions of the nascent protein with chaperones seem to be central to the mistargeting phenomenon.72,73 AGT is a very stable protein, although under certain conditions, it can undergo aggregation mediated by either hydrophobic effects, occurring upon partial unfolding and monomerization,74 or by electrostatic forces, occurring on a native-like dimeric form of the protein.75 The molecular and cellular bases of PH1 have been largely investigated in recent years by using various combinations of molecular, cellular and in silico approaches. The data obtained have been extensively reviewed elsewhere.6,72,73,76–80 The overall picture

Vitamin B6-dependent enzymes and disease

Figure 10.4  Possible prosthetic and chaperoning effects of B6 vitamers and aminotransferase inhibitor (AOA) on the alanine-glyoxylate aminotransferase (AGT) folding pathway. The unfolded chain is proposed to form a partly folded monomer (M*) that can either complete folding (M) and dimerize into a folded apodimer (D) before binding PLP (pyridoxal-5’ phosphate) (pathway to the left) or bind the cofactor to form a holo-monomer (MPLP) that then dimerizes (pathway to the right). It is also possible that a partially unfolded apodimer (D*) binds PLP to form the active fully folded holodimer (DPLP). The folded holodimer is further stabilized upon binding of its substrate (or the competitive inhibitor aminooxyacetic acid [AOA]) to achieve the three-dimensional structure described (Protein Data Bank ID: 1H0C). Most pathogenic mutations associated with PH1 cause folding defects that then lead to a variety of downstream effects including increased degradation or aggregation, dimer destabilization, or mistargeting. D, Folded apodimer; DPLP, folded holodimer; DPLP-AOA, AOA analogs-holodimer complex; D*, partly unfolded apodimer; D*PLP, partly unfolded holodimer; M, folded monomer; M*, partly unfolded monomer; U, Unfolded polypeptide chain.

coming out from the work of various groups is summarized in Fig. 10.5. Most mutations, including the most frequent in PH1 patients, do not influence the intrinsic catalytic activity of the protein, but rather decrease the efficiency of the folding process and/ or influence its stability. At the molecular level, the alterations can lead to an increased susceptibility of the dimer to aggregate or to the increased population of monomeric

207

208

Protein Homeostasis Diseases

Figure 10.5  Loss-of-function mechanisms and the corresponding action of the pharmacological agent. (A) Increased degradation activity. (B) Mistargeting into mitochondria. (C) Increased susceptibility to aggregate. (D) Catalytic activity is suppressed by a defect in the corresponding domain. (E) Protein partners interactions do not work properly.

intermediates prone to be imported into mitochondria in the presence of the minor allele polymorphism. In this regard, the presence of the minor allele usually exacerbates the effects of PH1-related mutations.61,81–83 It has been proposed that the minor allele could represent a threshold below which the level of instability of the protein leads to a pathogenic phenotype.73 Pathogenic mutations can affect the holo-form and/or the apo-form. However, apoAGT is, in general, more susceptible to the detrimental effect of mutations altering protein folding, while holoAGT displays an increased stability toward both chemical and thermal stress.72,73,84,85 At the cellular level, folding defects can result in either very low levels of expression, cytosolic or intraperoxisomal aggregation, increased binding of molecular chaperones, and/or the aberrant targeting of the protein to mitochondria.59,60,72,73,83,86–88 A further step in the understanding of the relation between the pathogenicity of mutations and the structural features of the mutation site has been recently published.77 The data indicate that the pathogenic effect of recurrent mutations not only depends on the local destabilizing effect but also on the ability to propagate on the entire protein structure remodeling its energy landscape. In addition, the interaction between pathogenic alleles, evolutionarily divergent gatekeeper residues, and posttranslational modifications should be also taken into account in the development of new therapeutic approaches for diseases as PH1.89

The role of natural and unnatural ligands as chaperones for AGT On the basis of the results reported above, PH1 has been unanimously defined as a misfolding disease, and efforts directed toward the identification of small molecules able to improve the AGT folding efficiency have been undertaken with the final aim of finding

Vitamin B6-dependent enzymes and disease

a noninvasive and specific treatment strategy to counteract the disease. Numerous studies point at the identification of PCs. PCs are drug-like molecules that improve the folding yield of misfolded variants by counteracting mutations that cause conformational defects.90 Compounds acting as PCs for diseases due to enzymatic deficits usually belong to three categories: (1) cofactors for enzymes acting through the binding of a coenzyme; (2) competitive inhibitors, which show high binding affinity and specificity; and (3) nonnatural molecules identified in high-throughput screening campaigns. As for the first group, the finding that PLP could behave as a chaperone for AGT has been known for some years, mainly because of data on purified proteins indicating that the loss of the coenzyme strongly decreases the stability of the protein against chemical and thermal stress.72,84,91 Subsequently, the chaperone action of the coenzyme has been extensively demonstrated in cellular systems expressing the most common variants associated with PH1 (i.e., p.G170R, p.F152I, p.I244T, and p.G41R associated with the minor allele)60 as well as less frequent variants (i.e., p.G47R on the minor allele and the variants showing mutations at p.Gly161).87,92 These variants display various combinations of defects including aberrant mitochondrial import, reduced dimer stability, and increased tendency to aggregation or intracellular degradation. In some cases, the effect of PLP is paired by the clinical response of PH1 patients to the therapy with PN.63,65,68 Overall, molecular and cellular analyses have highlighted that the chaperone role of the coenzyme is probably more complex than expected because PLP can play various roles during AGT folding, as outlined in Fig. 10.4. It has been shown to facilitate the correct folding of monomeric intermediates, but it could also promote dimerization of the folded apomonomer and even stabilize the dimer once folded.79 However, some data obtained in a cellular model of PH1 have shown that very high PN concentrations in the culture medium unexpectedly reduce the specific activity and the glyoxylate detoxification capability of cells.68,93 This effect is probably due to the finding that the supplementation with PN leads not only to the intracellular accumulation of PLP but also of PN 5’-phosphate (PNP), which competes with PLP for apoenzyme binding, giving rise to an inactive AGT-PNP complex. Thus, the global positive effect seen upon PN administration is probably a balance between the chaperone effect exerted by PLP and the inhibitory effect exerted by PNP. This could explain why most patients respond to low PN doses (< 10 mg/kg/day) and why urine oxalate levels do not correlate with serum B6 levels.68 Based on the inhibitory effects of PN administration, the effects of other B6 vitamers have been considered.93 PM and PL were found to be more effective than PN in rescuing two PN-responsive variants (p.F152I and p.G170R on the minor allele). Indeed, PM or PL supplementation causes the intracellular accumulation of PLP and PMP, the two active forms of the coenzyme exerting both a chaperone and a prosthetic role, without any accumulation of PNP.93 The great significance of this strategy is also supported by the finding that the administration of PM and PL to mice does not result in the subtle neurological effects seen with high PN doses and that PM is actually under phase II clinical trial for diabetes type I and type II (Pyridorin) and does not appear to be

209

210

Protein Homeostasis Diseases

toxic for humans, even at high doses.94,95 These results support the testing of PM and PL in animal disease models to validate their potential use as a novel coenzyme administration therapy for patients bearing PH1-causing mutations that affect protein folding. Another class of PCs is represented by compounds acting as competitive inhibitors of the enzyme involved. One of the molecules known as AGT inhibitors is aminooxyacetic acid (AOA).96 By kinetic analyses on purified proteins, AOA was found to act as a slow, tight-binding competitive inhibitor of AGT in the holo-form with a KI value in the low nanomolar range. The presence of AOA in the culture medium increases the amount of functional AGT of cells expressing conformational variants. In particular, it improves the folding of p.G41R-Ma, a variant mainly characterized by an increased susceptibility to aggregation and proteolytic degradation, and promotes the peroxisomal localization of p.G170R-Mi and p.I244T-Mi, two variants aberrantly targeted to mitochondria.97 However, the tight binding behavior and the low specificity of AOA make it unsuitable for use in therapy. Thus, based on the idea that AOA could represent the chemical scaffold for the development of more specific ligands, a preliminary screening campaign on several AOA derivates has been set up.97 It allowed identification of compounds active as AGT inhibitors with IC50 in the micromolar range (Table 10.1) and establishment Table 10.1  In vitro activity data of aminooxyacetic acid (AOA) analogs. Compound Structure

IC50(µM)

0.15

3

0.3

1.4

16.7

Vitamin B6-dependent enzymes and disease

of a preliminary structure-activity relationship around AOA. Another high-throughput screening has been also performed by using a phenotypic assay aimed at identifying molecules able to redirect AGT variants mistargeted to mitochondria back to their correct peroxisomal localization.98 All of these data represent a very important step toward the setup of a PC-based therapy, but rational design and testing of more effective and specific compounds should be performed in view of the development and the optimization of a lead compound. Nonetheless, studies carried out at the preclinical and/or clinical levels suggest that the treatment of rare diseases represents a challenge, mainly due to the complex pathogenesis of the disorders as well as to the occurrence of secondary events downstream to the primary defect. It follows that even when promising therapeutic strategies are in development, a complete correction of the disease alterations by a single treatment is rare. Rather, the targeting of two or more disease-pathogenic mechanisms by combination therapies has proven to be more effective. Combined treatments have shown both additive and synergic effects, and even therapies that are poorly effective when administered alone can greatly contribute to the overall efficacy when administered in combination with other approaches. Very representative examples of this strategy are lysosomal storage diseases, for which the use of combinations that couple protein-, cell- and gene-based therapies have been implemented in recent years.99 Other examples include Wilson’s disease,100 Crohn’s disease,101 and Alzheimer’s disease.102 Notably, a combined approach has been also used for PN-dependent epilepsy, whose symptoms can be more effectively reduced by a triple therapy that also includes the administration of PN.103 As far as PH1 is concerned, two classes of small molecules have been found to be effective in counteracting the misfolding caused by pathogenic mutations: B6 vitamers, which rescue conformational defects of the apo form of the variants, and AOA derivatives, which act as PCs and rescue conformational defects of the holo forms (Fig. 10.4). Thus, it can be presumed that the combined administration of B6 vitamers and AOA derivatives could be a promising option. In order to have preliminary insights into the possible efficacy of a combined treatment with B6 vitamers and compounds acting as PCs for AGT, we compared the effects of (1) B6 vitamers alone (PN, PM, or PL), (2) AOA alone, or (3) the combination of B6 vitamers and AOA on the specific activity and expression level of different AGT pathogenic variants. We chose AGT mutations affecting the folding of both the holo-and the apo-forms of the enzyme (i.e., p.G41R and p.F152I) or only the apoform (p.G170R).80 All of the analyzed variants were stably expressed in Chinese hamster ovary cells stably expressing glycolate oxidase CHO-GO and adapted for growing in a medium containing a PN concentration that mirrors the physiological B6 plasma levels.93 The data reported in Fig. 10.6 reveal that: (1) B6 vitamers increase the expression level of p.G41R-Ma (in the major haplotype), p.F152I-Mi, and p.G170R-Mi; (2)

211

212

Protein Homeostasis Diseases

Figure 10.6  Effect of combined B6 vitamers-aminooxyacetic acid (AOA) treatment on expression level and specific activity of folding-defective alanine-glyoxylate aminotransferase (AGT) variants. Chinese hamster ovary cells stably expressing glycolate oxidase (CHO-GO) cells stably expressing p.G41R-Ma, p.F152I-Mi, and p.G170R-Mi were grown for 3 weeks in low B6 medium in the absence or in the presence of B6 vitamers (10 µM) and/or AOA (50 µM), as reported in the figure key. (A) Immunoblot band volume. 2 g of soluble lysate were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), immunoblotted with rabbit anti-AGT (1:6000), and detected by chemiluminescence. Band volumes have been quantified by the QuantityOne software and expressed as relative to the corresponding untreated control. (B) Specific activity values. One hundred micrograms of soluble lysate were incubated with 0.5 mM L-alanine, 10 mM glyoxylate, and 200 uM pyridoxal-5’ phosphate (PLP) at 25°C in 100 mM K phosphate, pH 7.4. The catalytic activity was measured by a spectrophotometric assay coupled with lactate dehydrogenase. Bar graphs represent the mean + standard error of the mean (SEM).

AOA increases the expression level of p.F152I-Mi and p.G170R-Mi, while it does not significantly alter the expression level of p.G41R-Ma; and (3) as compared with the single B6 vitamer/AOA treatment, the combined treatment significantly increases the AGT expression level of p.G41R-Ma and p.F152I-Mi variants, while no significant variations are detected for the p.G170R-Mi variant. Similarly, the analyses of the transaminase activity (Fig. 10.6B) show that: (1) the treatment with PN, PM, or

Vitamin B6-dependent enzymes and disease

PL increases the specific activity of p.G41R-Ma, p.F152I-Mi, and p.G170R-Mi; (2) AOA increases the specific activity of p.F152I-Mi, while it does not significantly alter the specific activity of p.G41R-Ma and p.G170R-Mi; and (3) as compared with the single B6 vitamer/AOA treatment, the combined treatment significantly increases the specific activity of p.G41R-Ma and p.F152I-Mi, while no significant variations are detected for p.G170R-Mi. Overall, these data demonstrate that the combined B6 vitamers/AOA treatment is more efficient in counteracting the effects of the p.G41R and p.F152I mutations as compared with each single treatment, while no significant differences between the single and the combined treatment are observed for the p.G170R mutation. These data are in excellent agreement with the finding that while the folding defect of p.G170R-Mi is only related to the apo-form, both p.G41R-Ma and p.F152I-Mi have structural defects on both the apo- and the holo-forms. In fact, B6 vitamers act on the apo-form by exerting both a prosthetic and a chaperone role, while AOA interacts with PLP at the AGT active site, thus playing a role only in the stabilization of the holo-form.6 On these bases, we can envisage that variants affected by structural defects of both the holo- and the apo-forms could benefit from the combined treatment.

Steps forward to new therapeutic approaches for primary hyperoxaluria type 1 PH1 is a life-threatening disease that has recently attracted the interest of a growing number of academic groups and private companies. Thanks to the advent of innovative technologies in the field of biomedicine, the possibility to develop an effective therapeutic approach in the near future is increasing. Attempts toward therapies able to restore liver AGT expression by gene therapy have been carried out in mice models using either adeno-associated viral vectors or helper-dependent adenoviral vectors.104–106 Moreover, many efforts have been recently devoted to strategies able to prevent glyoxylate formation rather than restoring its detoxification, using small interfering RNA therapeutics targeting GO, the enzyme that converts glycolate to glyoxylate in liver peroxisomes,107 or LDH, which catalyzes glyoxylate oxidation generating oxalate.108 The promising results obtained have encouraged companies to start clinical trials that are currently ongoing (NCT03681184, NCT03392896). Another option validated in animal models and currently in clinical testing is based on the oral administration of oxalate-degrading bacteria109 or oxalate-degrading enzymes.110 Both are expected to reduce exogenous oxalate absorption, thus possibly preventing stone formation. Within this scenario, a place is also reserved for small-molecule therapies, which have the advantage of an easier administration and are usually associated with reduced side effects. Attempts toward a pharmacological inhibition of GO and LDH are currently ongoing, and they already allowed identification of hit compounds able to inhibit oxalate formation in primary hepatocytes from AGXT knockout mice.111,112

213

214

Protein Homeostasis Diseases

Conclusions For several decades, empirical vitamin B6 administration has been the only pharmacological treatment of PH trying to reduce the liver production of toxic oxalate in PH1 patients. In recent times, PH is diagnosed by genetic analysis at earlier stages of disease in more and more patients, raising the possibility of pharmacoprophylactic therapies aimed at delaying or preventing the loss of renal function, which is a tipping point in the natural history of the disease.With our current understanding of the molecular consequences of missense AGT mutations and the available cell and organism models, we are in a position to explore novel PCs that could, perhaps in combination with other orthogonal molecular therapies, achieve efficient prevention of renal damage in PH patients, avoiding the risks and long-term complications of liver (and kidney) transplantation and immunosuppression.

Acknowledgments The authors acknowledge the help from Dr. Sofía Salido (University of Jaen) in the preparation of Figs. 10.1 and 10.2. This work has been partially supported by grants from the Spanish Ministry of Science (SAF2015-69796 to E.S.) and by the Italian Ministry of University and Research (SIR project RBSI148BK3 to B.C.)

References 1. Ashaq I, Shajrul A, Akbar M, Rashid F. Protein misfolding diseases: in perspective of gain and loss of function. In: Proteostasis and chaperone surveillance. New Delhi: Springer India; 2015, p. 105–8. Avail­ able from: https://doi.org/10.1007/978-81-322-2467-9_6. 2. Valastyan JS, Lindquist S. Mechanisms of protein-folding diseases at a glance. Dis Model Mech 2014;7(1):9–14. doi: 10.1242/dmm.013474. 3. Winklhofer KF, Tatzelt J, Haass C. The two faces of protein misfolding: gain- and loss-of-function in neurodegenerative diseases. EMBO J 2008;27(2):336–49. doi: 10.1038/sj.emboj.7601930. 4. Chiti F, Dobson CM. Protein misfolding, amyloid formation, and human disease: a summa­ry of progress over the last decade. Ann Rev Biochem 2017;86:27–68. doi: 10.1146/annurev-biochem-061516-045115. 5. Aymami J, Barril X, Rodríguez-Pascau L, Martinell M. Pharmacological chaperones for enzyme enhancement therapy in genetic diseases. Pharm Pat Anal 2013;2(1):109–24. doi: 10.4155/ppa.12.74. 6. Oppici E, Montioli R, Dindo M, Cellini B. Natural and unnatural compounds rescue folding defects of human alanine:glyoxylate aminotransferase leading to primary hyperoxaluria type I. Curr Drug Targets 2016;17(13):1482–91 Available from: http://www.ncbi.nlm.nih.gov/pubmed/26931357. 7. Somaraju UR, Merrin M. Sapropterin dihydrochloride for phenylketonuria. Cochrane Database Syst Rev 2015;12(3):CD008005. doi: 10.1002/14651858.CD008005.pub4. 8. Hellmann H, Mooney S. Vitamin B6: a molecule for human health? Molecules 2010;15(1):442–59. doi: 10.3390/molecules15010442. 9. Percudani R, Peracchi A. The B6 database: a tool for the description and classification of vitamin B6-dependent enzymatic activities and of the corresponding protein families. BMC Bioinformatics 2009;10:273. doi: 10.1186/1471-2105-10-273. 10. Jansonius JN. Structure, evolution and action of vitamin B6-dependent enzymes. Curr Opin Struct Biol 1998;8(6):759–69 Available from: http://www.ncbi.nlm.nih.gov/pubmed/9914259. 11. Bach RD, Canepa C, Glukhovtsev MN. Influence of electrostatic effects on activation barriers in enzymatic reactions: pyridoxal 5‘-phosphate-dependent decarboxylation of α-Amino acids. J Am Chem Soc 1999;121(28):6542–55. doi: 10.1021/ja9907616.

Vitamin B6-dependent enzymes and disease

12. Dunathan HC, Voet JG. Stereochemical evidence for the evolution of pyridoxal-phosphate enzymes of various function from a common ancestor. Proc Natl Acad Sci USA 1974;71(10):3888–91. doi: 10.1073/pnas.71.10.3888. 13. Alexander FW, Sandmeier E, Mehta PK, Christen P. Evolutionary relationships among pyridoxal-5’-phosphate-dependent enzymes. Regio-specific alpha, beta and gamma families. Eur J Biochem 1994;219(3):953–60 Available from: http://www.ncbi.nlm.nih.gov/pubmed/8112347. 14. Catazaro J, Caprez A, Guru A, Swanson D, Powers R. Functional evolution of PLP-dependent enzymes based on active-site structural similarities. Proteins 2014;82(10):2597–608. doi: 10.1002/prot.24624. 15. Schneider G, Käck H, Lindqvist Y. The manifold of vitamin B6 dependent enzymes. Structure 2000;8(1):R1–6 Available from: http://www.ncbi.nlm.nih.gov/pubmed/10673430. 16. Parra M, Stahl S, Hellmann H. Vitamin B and its role in cell metabolism and physiology. Cells 2018;7(7)doi: 10.3390/cells7070084. 17. Cellini B, Montioli R, Oppici E, Astegno A, Voltattorni CB. The chaperone role of the pyridoxal 5’-phosphate and its implications for rare diseases involving B6-dependent enzymes. Clin Biochem 2014;47(3):158–65. doi: 10.1016/j.clinbiochem.2013.11.021. 18. Cellini B, Bertoldi M, Montioli R, Laurents DV, Paiardini A, Voltattorni CB. Dimerization, folding processes of Treponema denticola cystalysin, the role of pyridoxal 5’-phosphate. Biochemistry 2006;45(47):14140–54. doi: 10.1021/bi061496l. 19. Montioli R, Cellini B, Bertoldi M, Paiardini A, Voltattorni CB. An engineered folded PLP-bound monomer of Treponema denticola cystalysin reveals the effect of the dimeric structure on the catalytic properties of the enzyme. Proteins 2009;74(2):304–17. doi: 10.1002/prot.22160. 20. Deu E, Kirsch JF. Cofactor-directed reversible denaturation pathways: the cofactor-stabilized Escherichia coli aspartate aminotransferase homodimer unfolds through a pathway that differs from that of the apoenzyme. Biochemistry 2007;46(19):5819–29. doi: 10.1021/bi602632d. 21. Bhatt AN, Bhakuni V. Characterization of pyridoxal 5’-phosphate-binding domain and folding intermediate of Bacillus subtilis serine hydroxymethyltransferase: an autonomous folding domain. J Biochem 2008;144(3):295–303. doi: 10.1093/jb/mvn067. 22. Gandarias JM, Goiriena JJ, Ainz F, Rueda E, Fernández BM. Lipolysis by corticotropin in fat cells from hypothyroid rats. Effect of adenosine deaminase. Rev Esp Fisiol 1989;45(Suppl.):25–8 Available from: http://www.ncbi.nlm.nih.gov/pubmed/2561927. 23. Giardina G, Brunotti P, Fiascarelli A, Cicalini A, Costa MGS, Buckle AM, …, Cutruzzolà F. How pyridoxal 5’-phosphate differentially regulates human cytosolic and mitochondrial serine hydroxymethyltransferase oligomeric state. FEBS J 2015;282(7):1225–41. doi: 10.1111/febs.13211. 24. Walden M, Tian L, Ross RL, Sykora UM, Byrne DP, Hesketh EL, …, Zeqiraj E. Metabolic control of BRISC-SHMT2 assembly regulates immune signalling. Nature 2019;570(7760):194–9. doi: 10.1038/s41586-019-1232-1. 25. Wilson MP, Plecko B, Mills PB, Clayton PT. Disorders affecting vitamin B 6 metabolism. J Inherit Metab Dis 2019;doi: 10.1002/jimd.12060. 26. Erlandsen H, Pey AL, Gámez A, Pérez B, Desviat LR, Aguado C, …, Stevens RC. Correction of kinetic and stability defects by tetrahydrobiopterin in phenylketonuria patients with certain phenylalanine hydroxylase mutations. Proc Natl Acad Sci USA 2004;101(48):16903–8. doi: 10.1073/pnas.0407256101. 27. Clayton PT. B6-responsive disorders: a model of vitamin dependency. J Inherit Metab Dis 2006;29(2– 3):317–26. doi: 10.1007/s10545-005-0243-2. 28. Cotter PD, Baumann M, Bishop DF. Enzymatic defect in “X-linked” sideroblastic anemia: molecular evidence for erythroid delta-aminolevulinate synthase deficiency. Proc Natl Acad Sci USA 1992;89(9):4028–32. doi: 10.1073/pnas.89.9.4028. 29. Cotter PD, May A, Li L, Al-Sabah AI, Fitzsimons EJ, Cazzola M, Bishop DF. Four new mutations in the erythroid-specific 5-aminolevulinate synthase (ALAS2) gene causing X-linked sideroblastic anemia: increased pyridoxine responsiveness after removal of iron overload by phlebotomy and coinheritance of hereditary hemochromatosis. Blood 1999;93(5):1757–69 Available from: http://www. ncbi.nlm.nih.gov/pubmed/10029606. 30. Furuyama K, Fujita H, Nagai T, Yomogida K, Munakata H, Kondo M, …, Yamamoto M. Pyridoxine refractory X-linked sideroblastic anemia caused by a point mutation in the erythroid 5-aminolevulinate synthase gene. Blood 1997;90(2):822–30 Available from: http://www.ncbi.nlm.nih.gov/ pubmed/9226183.

215

216

Protein Homeostasis Diseases

31. Barber GW, Spaeth GL. The successful treatment of homocystinuria with pyridoxine. J Pediatr 1969;75(3):463–78 Available from: http://www.ncbi.nlm.nih.gov/pubmed/5804194. 32. Mudd SH, Levy HL, Kraus JP. Disorders of transsulfuration. In: Scriver CR, Beaudet AL, Sly WS, Valle D editors. The metabolic and molecular bases of inherited disease. 8th ed. New York: McGraw-Hill; 2001 p. 2007–56. 33. Schiff M, Blom HJ. Treatment of inherited homocystinurias. Neuropediatrics 2012;43(6):295–304. doi: 10.1055/s-0032-1329883. 34. Majtan T, Pey AL, Ereño-Orbea J, Martínez-Cruz LA, Kraus JP. Targeting cystathionine beta-synthase misfolding in homocystinuria by small ligands: state of the art and future directions. Curr Drug Targets 2016;17(13):1455–70 Available from: http://www.ncbi.nlm.nih.gov/pubmed/26931358. 35. Majtan T, Pey AL, Gimenez-Mascarell P, Martínez-Cruz LA, Szabo C, Kožich V, Kraus JP. Potential pharmacological chaperones for cystathionine beta-synthase-deficient homocystinuria. Handb Exp Pharmacol 2018;245:345–83. doi: 10.1007/164_2017_72. 36. Ginguay A, Cynober L, Curis E, Nicolis I. Ornithine aminotransferase, an important glutamatemetabolizing enzyme at the crossroads of multiple metabolic pathways. Biology 2017;6(4):18. doi: 10.3390/biology6010018. 37. Ginguay A, Cynober L, Curis E, Nicolis I. Ornithine aminotransferase, an important glutamatemetabolizing enzyme at the crossroads of multiple metabolic pathways. Biology 2017;6(1):E18. doi: 10.3390/biology6010018. 38. Seiler N. Ornithine aminotransferase, a potential target for the treatment of hyperammonemias. Curr Drug Targets 2000;1(2):119–53 Available from: http://www.ncbi.nlm.nih.gov/pubmed/11465067. 39. Shen BW, Hennig M, Hohenester E, Jansonius JN, Schirmer T. Crystal structure of human recombinant ornithine aminotransferase. J Mol Biol 1998;277(1):81–102. doi: 10.1006/jmbi.1997.1583. 40. Montioli R, Zamparelli C, Borri Voltattorni C, Cellini B. Oligomeric state and thermal stability of apo- and holo-human ornithine δ-aminotransferase. Protein J 2017;36(3):174–85. doi: 10.1007/ s10930-017-9710-5. 41. Berson EL, Shih VE, Sullivan PL. Ocular findings in patients with gyrate atrophy on pyridoxine and low-protein, low-arginine diets. Ophthalmology 1981;88(4):311–5 Available from: http://www.ncbi. nlm.nih.gov/pubmed/7254777. 42. Mashima YG, Weleber RG, Kennaway NG, Inana G. Genotype-phenotype correlation of a pyridoxine-responsive form of gyrate atrophy. Ophthalmic Genet 1999;20(4):219–24 Available from: http:// www.ncbi.nlm.nih.gov/pubmed/10617919. 43. Wirtz MK, Kennaway NG, Weleber RG. Heterogeneity and complementation analysis of fibroblasts from vitamin B6 responsive and nonresponsive patients with gyrate atrophy of the choroid and retina. J Inherit Metab Dis 1985;8(2):71–4 Available from: http://www.ncbi.nlm.nih.gov/ pubmed/3939534. 44. Kennaway NG, Stankova L, Wirtz MK, Weleber RG. Gyrate atrophy of the choroid and retina: characterization of mutant ornithine aminotransferase and mechanism of response to vitamin B6. Am J Human Genet 1989;44(3):344–52 Available from: http://www.ncbi.nlm.nih.gov/pubmed/2916580. 45. Montioli R, Desbats MA, Grottelli S, Doimo M, Bellezza I, Borri Voltattorni C, …, Cellini B. Molecular and cellular basis of ornithine δ-aminotransferase deficiency caused by the V332M mutation associated with gyrate atrophy of the choroid and retina. Biochim Biophys Acta 2018;1864(11):3629– 38. doi: 10.1016/j.bbadis.2018.08.032. 46. Wassenberg T, Molero-Luis M, Jeltsch K, Hoffmann GF, Assmann B, Blau N, …, Opladen T. Consensus guideline for the diagnosis and treatment of aromatic l-amino acid decarboxylase (AADC) deficiency. Orphanet J Rare Dis 2017;12(1):12. doi: 10.1186/s13023-016-0522-z. 47. Lee H-F, Tsai C-R, Chi C-S, Chang T-M, Lee H-J. Aromatic L-amino acid decarboxylase deficiency in Taiwan. Eur J Paediatr Neurol EJPN 2009;13(2):135–40. doi: 10.1016/j.ejpn.2008.03.008. 48. Cellini B, Montioli R, Oppici E, Voltattorni CB. Biochemical and computational approaches to improve the clinical treatment of dopa decarboxylase-related diseases: an overview. Open Biochem J 2012;6:131–8. doi: 10.2174/1874091X01206010131. 49. Himmelreich N, Montioli R, Bertoldi M, Carducci C, Leuzzi V, Gemperle C, …, Blau N. Aromatic amino acid decarboxylase deficiency: molecular and metabolic basis and therapeutic outlook. Mol Genet Metab 2019;127:12–22. doi: 10.1016/j.ymgme.2019.03.009.

Vitamin B6-dependent enzymes and disease

50. Pons R, Ford B, Chiriboga CA, Clayton PT, Hinton V, Hyland K, …, De Vivo DC. Aromatic L-amino acid decarboxylase deficiency: clinical features, treatment, and prognosis. Neurology 2004;62(7):1058– 65 Available from: http://www.ncbi.nlm.nih.gov/pubmed/15079002. 51. Chien Y-H, Lee N-C, Tseng S-H, Tai C-H, Muramatsu S, Byrne BJ, Hwu W-L. Efficacy and safety of AAV2 gene therapy in children with aromatic L-amino acid decarboxylase deficiency: an open-label, phase 1/2 trial. Lancet Child Adolescent Health 2017;1(4):265–73. doi: 10.1016/S23524642(17)30125-6. 52. Montioli R, Voltattorni CB, Bertoldi M. Parkinson’s disease: recent updates in the identification of human dopa decarboxylase inhibitors. Curr Drug Metab 2016;17(5):513–8 Available from: http:// www.ncbi.nlm.nih.gov/pubmed/27025882. 53. Montioli R, Dindo M, Giorgetti A, Piccoli S, Cellini B, Voltattorni CB. A comprehensive picture of the mutations associated with aromatic amino acid decarboxylase deficiency: from molecular mechanisms to therapy implications. Human Mol Genet 2014;23(20):5429–40. doi: 10.1093/hmg/ddu266. 54. Montioli R, Oppici E, Cellini B, Roncador A, Dindo M, Voltattorni CB. S250F variant associated with aromatic amino acid decarboxylase deficiency: molecular defects and intracellular rescue by pyridoxine. Human Mol Genet 2013;22(8):1615–24. doi: 10.1093/hmg/ddt011. 55. Montioli R, Paiardini A, Kurian MA, Dindo M, Rossignoli G, Heales SJR, …, Bertoldi M. The novel R347g pathogenic mutation of aromatic amino acid decarboxylase provides additional molecular insights into enzyme catalysis and deficiency. Biochim Biophys Acta 2016;1864(6):676–82. doi: 10.1016/j. bbapap.2016.03.011. 56. Giardina G, Montioli R, Gianni S, Cellini B, Paiardini A, Voltattorni CB, Cutruzzolà F. Open conformation of human DOPA decarboxylase reveals the mechanism of PLP addition to Group II decarboxylases. Proc Natl Acad Sci USA 2011;108(51):20514–9. doi: 10.1073/pnas.1111456108. 57. Matsuda N, Hayashi H, Miyatake S, Kuroiwa T, Kagamiyama H. Instability of the apo form of aromatic L-amino acid decarboxylase in vivo and in vitro: implications for the involvement of the flexible loop that covers the active site. J Biochem 2004;135(1):33–42 Available from: http://www.ncbi. nlm.nih.gov/pubmed/14999007. 58. Salido E, Pey AL, Rodriguez R, Lorenzo V. Primary hyperoxalurias: disorders of glyoxylate detoxification. Biochim Biophys Acta 2012;1822(9):1453–64. doi: 10.1016/j.bbadis.2012.03.004. 59. Purdue PE, Takada Y, Danpure CJ. Identification of mutations associated with peroxisome-to-mitochondrion mistargeting of alanine/glyoxylate aminotransferase in primary hyperoxaluria type 1. J Cell Biol 1990;111(6 Pt. 1):2341–51 Available from: http://www.ncbi.nlm.nih.gov/pubmed/1703535. 60. Fargue S, Lewin J, Rumsby G, Danpure CJ. Four of the most common mutations in primary hyperoxaluria type 1 unmask the cryptic mitochondrial targeting sequence of alanine:glyoxylate aminotransferase encoded by the polymorphic minor allele. J Biol Chem 2013;288(4):2475–84. doi: 10.1074/ jbc.M112.432617. 61. Lumb MJ, Danpure CJ. Functional synergism between the most common polymorphism in human alanine:glyoxylate aminotransferase and four of the most common disease-causing mutations. J Biol Chem 2000;275(46):36415–22. doi: 10.1074/jbc.M006693200. 62. Hopp K, Cogal AG, Bergstralh EJ, Seide BM, Olson JB, Meek AM, …. Rare Kidney Stone Consortium. Phenotype-genotype correlations and estimated carrier frequencies of primary hyperoxaluria. J Am Soc Nephrol 2015;26(10):2559–70. doi: 10.1681/ASN.2014070698. 63. Mandrile G, van Woerden CS, Berchialla P, Beck BB, Acquaviva Bourdain C, Hulton S-A, …. OxalEurope Consortium. Data from a large European study indicate that the outcome of primary hyperoxaluria type 1 correlates with the AGXT mutation type. Kidney Int 2014;86(6):1197–204. doi: 10.1038/ki.2014.222. 64. Milliner DS, Harris PC, Cogal AG, Lieske JC. Primary Hyperoxaluria Type 1. GeneReviews®. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20301460; 1993. 65. Zhao F, Bergstralh EJ, Mehta RA,Vaughan LE, Olson JB, Seide BM, …. Investigators of Rare Kidney Stone Consortium. Predictors of incident ESRD among patients with primary hyperoxaluria presenting prior to kidney failure. Clin J Am Soc Nephrol 2016;11(1):119–26. doi: 10.2215/CJN. 02810315. 66. Cochat P, Hulton S-A, Acquaviva C, Danpure CJ, Daudon M, De Marchi M, …, van Woerden CS. Primary hyperoxaluria type 1: indications for screening and guidance for diagnosis and treatment. Nephrol Dial Transplant 2012;27(5):1729–36. doi: 10.1093/ndt/gfs078.

217

218

Protein Homeostasis Diseases

67. Monico CG, Rossetti S, Olson JB, Milliner DS. Pyridoxine effect in type I primary hyperoxaluria is associated with the most common mutant allele. Kidney Int 2005;67(5):1704–9. doi: 10.1111/j.15231755.2005.00267.x. 68. Fargue S, Rumsby G, Danpure CJ. Multiple mechanisms of action of pyridoxine in primary hyperoxaluria type 1. Biochim Biophys Acta 2013;1832(10):1776–83. doi: 10.1016/j.bbadis.2013.04.010. 69. Hoppe B, Beck BB, Milliner DS. The primary hyperoxalurias. Kidney Int 2009;75(12):1264–71. doi: 10.1038/ki.2009.32. 70. Cellini B, Bertoldi M, Montioli R, Paiardini A, Borri Voltattorni C. Human wild-type alanine:glyoxylate aminotransferase and its naturally occurring G82E variant: functional properties and physiological implications. Biochem J 2007;408(1):39–50. doi: 10.1042/BJ20070637. 71. Zhang X, Roe SM, Hou Y, Bartlam M, Rao Z, Pearl LH, Danpure CJ. Crystal structure of alanine:glyoxylate aminotransferase and the relationship between genotype and enzymatic phenotype in primary hyperoxaluria type 1. J Mol Biol 2003;331(3):643–52 Available from: http://www.ncbi. nlm.nih.gov/pubmed/12899834. 72. Mesa-Torres N, Fabelo-Rosa I, Riverol D, Yunta C, Albert A, Salido E, Pey AL. The role of protein denaturation energetics and molecular chaperones in the aggregation and mistargeting of mutants causing primary hyperoxaluria type I. PloS One 2013;8(8):e71963. doi: 10.1371/journal.pone. 0071963. 73. Mesa-Torres N, Salido E, Pey AL. The lower limits for protein stability and foldability in primary hyperoxaluria type I. Biochim Biophys Acta 2014;1844(12):2355–65. doi: 10.1016/j.bbapap.2014.10.010. 74. Dindo M, Montioli R, Busato M, Giorgetti A, Cellini B, Borri Voltattorni C. Effects of interface mutations on the dimerization of alanine glyoxylate aminotransferase and implications in the mistargeting of the pathogenic variants F152I and I244T. Biochimie 2016;131:137–48. doi: 10.1016/j. biochi.2016.10.001. 75. Dindo M, Conter C, Cellini B. Electrostatic interactions drive native-like aggregation of human alanine:glyoxylate aminostransferase. FEBS J 2017;284(21):3739–64. doi: 10.1111/febs.14269. 76. Fernandez-Higuero JA, Betancor-Fernandez I, Mesa-Torres N, Muga A, Salido E, Pey AL. Structural and functional insights on the roles of molecular chaperones in the mistargeting and aggregation phenotypes associated with primary hyperoxaluria type I. Adv Protein Chem Struct Biol 2019;114:119– 52 Available from: http://www.ncbi.nlm.nih.gov/pubmed/30635080. 77. Medina-Carmona E, Betancor-Fernández I, Santos J, Mesa-Torres N, Grottelli S, Batlle C, …, Pey AL. Insight into the specificity and severity of pathogenic mechanisms associated with missense mutations through experimental and structural perturbation analyses. Human Mol Genet 2019;28(1):1–15. doi: 10.1093/hmg/ddy323. 78. Medina-Carmona E, Fuchs JE, Gavira JA, Mesa-Torres N, Neira JL, Salido E, …, Pey AL. Enhanced vulnerability of human proteins towards disease-associated inactivation through divergent evolution. Human Mol Genet 2017;26(18):3531–44. doi: 10.1093/hmg/ddx238. 79. Oppici E, Dindo M, Conter C, Borri Voltattorni C, Cellini B. Folding defects leading to primary hyperoxaluria. Handb Exp Pharmacol 2018;245:313–43. doi: 10.1007/164_2017_59. 80. Oppici E, Montioli R, Cellini B. Liver peroxisomal alanine:glyoxylate aminotransferase and the effects of mutations associated with primary hyperoxaluria type I: an overview. Biochim Biophys Acta 2015;1854(9):1212–9. doi: 10.1016/j.bbapap.2014.12.029. 81. Cellini B, Montioli R, Paiardini A, Lorenzetto A, Maset F, Bellini T, …, Voltattorni CB. Molecular defects of the glycine 41 variants of alanine glyoxylate aminotransferase associated with primary hyperoxaluria type I. Proc Natl Acad Sci USA 2010;107(7):2896–901. doi: 10.1073/pnas.0908565107. 82. Coulter-Mackie MB, Lian Q. Consequences of missense mutations for dimerization and turnover of alanine:glyoxylate aminotransferase: study of a spectrum of mutations. Mol Genet Metab 2006;89(4):349– 59. doi: 10.1016/j.ymgme.2006.07.013. 83. Santana A, Salido E, Torres A, Shapiro LJ. Primary hyperoxaluria type 1 in the Canary Islands: a conformational disease due to I244T mutation in the P11L-containing alanine:glyoxylate aminotransferase. Proc Natl Acad Sci USA 2003;100(12):7277–82. doi: 10.1073/pnas.1131968100. 84. Cellini B, Lorenzetto A, Montioli R, Oppici E, Voltattorni CB. Human liver peroxisomal alanine:glyoxylate aminotransferase: different stability under chemical stress of the major allele, the minor allele, and its pathogenic G170R variant. Biochimie 2010;92(12):1801–11. doi: 10.1016/j.biochi.2010.08.005.

Vitamin B6-dependent enzymes and disease

85. Cellini B, Montioli R, Paiardini A, Lorenzetto A,Voltattorni CB. Molecular insight into the synergism between the minor allele of human liver peroxisomal alanine:glyoxylate aminotransferase and the F152I mutation. J Biol Chemis 2009;284(13):8349–58. doi: 10.1074/jbc.M808965200. 86. Mesa-Torres N, Tomic N, Albert A, Salido E, Pey AL. Molecular recognition of PTS-1 cargo proteins by Pex5p: implications for protein mistargeting in primary hyperoxaluria. Biomolecules 2015;5(1):121– 41. doi: 10.3390/biom5010121. 87. Oppici E, Roncador A, Montioli R, Bianconi S, Cellini B. Gly161 mutations associated with primary hyperoxaluria type I induce the cytosolic aggregation and the intracellular degradation of the apo-form of alanine:glyoxylate aminotransferase. Biochim Biophys Acta 2013;1832(12):2277–88. doi: 10.1016/j.bbadis.2013.09.002. 88. Pey AL, Salido E, Sanchez-Ruiz JM. Role of low native state kinetic stability and interaction of partially unfolded states with molecular chaperones in the mitochondrial protein mistargeting associated with primary hyperoxaluria. Amino Acids 2011;41(5):1233–45. doi: 10.1007/s00726-010-0801-2. 89. Mesa-Torres N, Betancor-Fernández I, Oppici E, Cellini B, Salido E, Pey AL. Evolutionary divergent suppressor mutations in conformational diseases. Genes 2018;9(7)doi: 10.3390/genes9070352. 90. Convertino M, Das J, Dokholyan NV. Pharmacological chaperones: design and development of new therapeutic strategies for the treatment of conformational diseases. ACS Chem Biol 2016;11(6):1471– 89. doi: 10.1021/acschembio.6b00195. 91. Hopper ED, Pittman AMC, Fitzgerald MC, Tucker CL. In vivo and in vitro examination of stability of primary hyperoxaluria-associated human alanine:glyoxylate aminotransferase. J Biol Chem 2008;283(45):30493–502. doi: 10.1074/jbc.M803525200. 92. Montioli R, Oppici E, Dindo M, Roncador A, Gotte G, Cellini B, Borri Voltattorni C. Misfolding caused by the pathogenic mutation G47R on the minor allele of alanine:glyoxylate aminotransferase and chaperoning activity of pyridoxine. Biochim Biophys Acta 2015;1854(10 Pt. A):1280–9. doi: 10.1016/j.bbapap.2015.07.002. 93. Oppici E, Fargue S, Reid ES, Mills PB, Clayton PT, Danpure CJ, Cellini B. Pyridoxamine and pyridoxal are more effective than pyridoxine in rescuing folding-defective variants of human alanine:glyoxylate aminotransferase causing primary hyperoxaluria type I. Human Mol Genet 2015;24(19):5500–11. doi: 10.1093/hmg/ddv276. 94. Degenhardt TP, Alderson NL, Arrington DD, Beattie RJ, Basgen JM, Steffes MW, …, Baynes JW. Pyridoxamine inhibits early renal disease and dyslipidemia in the streptozotocin-diabetic rat. Kidney Int 2002;61(3):939–50. doi: 10.1046/j.1523-1755.2002.00207.x. 95. Williams ME, Bolton WK, Khalifah RG, Degenhardt TP, Schotzinger RJ, McGill JB. Effects of pyridoxamine in combined phase 2 studies of patients with type 1 and type 2 diabetes and overt nephropathy. Am J Nephrol 2007;27(6):605–14. doi: 10.1159/000108104. 96. Andy V, Horváth P, Wanders RJ. Aminooxy acetic acid: a selective inhibitor of alanine:glyoxylate aminotransferase and its use in the diagnosis of primary hyperoxaluria type I. Clin Chim Acta 1995;243(2):105–14 Available from: http://www.ncbi.nlm.nih.gov/pubmed/8747487. 97. Oppici E, Montioli R, Dindo M, Maccari L, Porcari V, Lorenzetto A, …, Cellini B. The chaperoning activity of amino-oxyacetic acid on folding-defective variants of human alanine:glyoxylate aminotransferase causing primary hyperoxaluria type I. ACS Chem Biol 2015;10(10):2227–36. doi: 10.1021/ acschembio.5b00480. 98. Madoux F, Janovick JA, Smithson D, Fargue S, Danpure CJ, Scampavia L, …, Conn PM. Development of a phenotypic high-content assay to identify pharmacoperone drugs for the treatment of primary hyperoxaluria type 1 by high-throughput screening. Assay Drug Dev Technol 2015;13(1):16–24. doi: 10.1089/adt.2014.627. 99. Macauley SL. Combination therapies for lysosomal storage diseases: a complex answer to a simple problem. Pediatr Endocrinol Rev 2016;13(Suppl. 1):639–48 Available from: http://www.ncbi.nlm.nih. gov/pubmed/ 27491211. 100. Chen J-C, Chuang C-H, Wang J-D, Wang C-W. Combination therapy using chelating agent and zinc for Wilson’s disease. J Med Biol Eng 2015;35(6):697–708. doi: 10.1007/s40846-015-0087-7. 101. Mosli MH, Feagan BG. Combination therapy for the treatment of Crohn’s disease. Expert Opin Biol Ther 2015;15(10):1429–42. doi: 10.1517/14712598.2015.1065249. 102. Patel L, Grossberg GT. Combination therapy for Alzheimer’s disease. Drugs Aging 2011;28(7):539–46. doi: 10.2165/11591860-000000000-00000.

219

220

Protein Homeostasis Diseases

103. Coughlin CR, van Karnebeek CDM, Al-Hertani W, Shuen AY, Jaggumantri S, Jack RM, …, Van Hove JLK. Triple therapy with pyridoxine, arginine supplementation and dietary lysine restriction in pyridoxine-dependent epilepsy: neurodevelopmental outcome. Mol Genet Metab 2015;116(1–2):35– 43. doi: 10.1016/j.ymgme.2015.05.011. 104. Castello R, Borzone R, D’Aria S, Annunziata P, Piccolo P, Brunetti-Pierri N. Helper-dependent adenoviral vectors for liver-directed gene therapy of primary hyperoxaluria type 1. Gene Ther 2016;23(2):129–34. doi: 10.1038/gt.2015.107. 105. Salido EC, Li XM, Lu Y, Wang X, Santana A, Roy-Chowdhury N, …, Roy-Chowdhury J. Alanineglyoxylate aminotransferase-deficient mice, a model for primary hyperoxaluria that responds to adenoviral gene transfer. Proc Natl Acad Sci USA 2006;103(48):18249–54. doi: 10.1073/pnas.0607218103. 106. Salido E, Rodriguez-Pena M, Santana A, Beattie SG, Petry H, Torres A. Phenotypic correction of a mouse model for primary hyperoxaluria with adeno-associated virus gene transfer. Mol Ther 2011;19(5):870–5. doi: 10.1038/mt.2010.270. 107. Dutta C, Avitahl-Curtis N, Pursell N, Larsson Cohen M, Holmes B, Diwanji R, …, Lai C. Inhibition of glycolate oxidase with dicer-substrate siRNA reduces calcium oxalate deposition in a mouse model of primary hyperoxaluria type 1. Mol Ther 2016;24(4):770–8. doi: 10.1038/mt.2016.4. 108. Lai C, Pursell N, Gierut J, Saxena U, Zhou W, Dills M, …, Brown BD. Specific inhibition of hepatic lactate dehydrogenase reduces oxalate production in mouse models of primary hyperoxaluria. Mol Ther 2018;26(8):1983–95. doi: 10.1016/j.ymthe.2018.05.016. 109. Milliner D, Hoppe B, Groothoff J. A randomised phase II/III study to evaluate the efficacy and safety of orally administered Oxalobacter formigenes to treat primary hyperoxaluria. Urolithiasis 2018;46(4): 313–23. doi: 10.1007/s00240-017-0998-6. 110. Grujic D, Salido EC, Shenoy BC, Langman CB, McGrath ME, Patel RJ, …, Margolin AL. Hyperoxaluria is reduced and nephrocalcinosis prevented with an oxalate-degrading enzyme in mice with hyperoxaluria. Am J Nephrol 2009;29(2):86–93. doi: 10.1159/000151395. 111. Martin-Higueras C, Luis-Lima S, Salido E. Glycolate oxidase is a safe and efficient target for substrate reduction therapy in a mouse model of primary hyperoxaluria type I. Mol Ther 2016;24(4):719–25. doi: 10.1038/mt.2015.224. 112. Moya-Garzón MD, Martín Higueras C, Peñalver P, Romera M, Fernandes MX, Franco-Montalbán F, …, Díaz-Gavilán M. Salicylic acid derivatives inhibit oxalate production in mouse hepatocytes with primary hyperoxaluria type 1. J Med Chem 2018;61(16):7144–67. doi: 10.1021/acs.jmedchem.8b00399.

CHAPTER 11

Galactosemia: opportunities for novel therapies Thomas J. McCorviea, David J. Timsonb

Section of Structural Biology, Department of Medicine, Imperial College, London, United Kingdom School of Pharmacy and Biomolecular Sciences, University of Brighton, Brighton, United Kingdom

a

b

Outline Introduction: four types of galactosemia Causes of pathology and current treatment Potential novel therapies Pharmacological chaperones for GALT deficiency Pharmacological chaperones for other types of galactosemia Conclusions References

221 224 226 229 233 236 237

Abbreviations ALS  amyotrophic lateral sclerosis ERT  enzyme replacement therapy GALE  UDP-galactose 4’-epimerase GALK1  galactokinase GALK2  UDP-N-acetylgalactosamine kinase GALM  galactose mutarotase GALT  galactose 1-phosphate uridylyltransferase GHMP  galactokinase, homoserine kinase, mevalonate kinase, phosphomevalonate kinase IQ  Intelligence quotient PGM  Phosphoglucomutase ROS  Reactive oxygen species SOD  superoxide dismutase UPR  Unfolded protein response

Introduction: four types of galactosemia Galactosemia describes a group of inherited metabolic diseases of carbohydrate metabolism.1–4 The common feature of these diseases is an inability to metabolize the aldose hexose monosaccharide galactose. Galactose is the C-4 epimer of glucose (Fig. 11.1A). This single difference in the stereochemistry in the sugar means that it is not recognized by the first enzyme of glycolysis, hexokinase.5 Thus, most organisms have a short pathway

Protein Homeostasis Diseases. http://dx.doi.org/10.1016/B978-0-12-819132-3.00011-7 Copyright © 2020 Elsevier Inc. All rights reserved.

221

222

Protein Homeostasis Diseases

Figure 11.1  Galactose anomers and Leloir pathway. (A) The two anomers of d-galactose shown in Haworth projections. The two anomers differ in the configuration of the hydroxyl group attached to carbon-1 (right-most carbon atom in these structures).6,15 (B) The Leloir pathway of galactose metabolism, including the initial anomerization (mutarotation) of galactose. Enzymes are shown in capitals. The types of galactosemia associated with each enzyme are shown below the enzyme names. GALM, galactose mutarotase; GALK1, galactokinase; GALT, galactose 1-phosphate uridylyltransferase; GALE, uridine diphosphate [UDP]-galactose 4’-epimerase.

that enables the carbon atoms in galactose to be used in energy metabolism and biosynthesis (Fig. 11.1B).The Leloir pathway is generally considered to begin with the site- and stereospecific phosphorylation of α-d-galactose at position 1. This reaction is catalyzed by galactokinase (Enzyme Commission number (EC) 2.7.1.6; GALK1 in humans). In solution, galactose exists as an equilibrium mixture of α- and β-anomers.6 Hydrolysis of a common dietary source of galactose, the disaccharide lactose, initially results in β-dgalactose. Since only α-d-galactose can be processed by GALK1, the rapid attainment of anomeric equilibrium is catalyzed by galactose mutarotase (EC 5.1.3.3; aldose 1-epimerase; GALM).7–9 Galactose 1-phosphate is converted to glucose 1-phosphate in a reaction catalyzed by galactose 1-phosphate uridylyltransferase (EC 2.7.7.12; GALT). This also generates uridine diphosphate (UDP)-galactose as a product that is isomerized back to UDP-glucose in a reaction catalyzed by UDP-galactose 4’-epimerase (EC 5.1.3.2; GALE). This completes what is classically considered to be the Leloir pathway.10–12 However, one more reaction is required to convert glucose 1-phosphate to a glycolytic intermediate: this compound is isomerized to glucose 6-phosphate in a reaction catalyzed by phosphoglucomutase (EC 5.4.2.2). The net result of this reaction is the conversion of galactose into glucose 6-phosphate with no overall change in the amounts of UDP-galactose and

Galactosemia: opportunities for novel therapies

UDP-galactose.The pathway also has biosynthetic roles. Both UDP-galactose and UDPglucose are used in the synthesis of the oligosaccharide moieties in glycoproteins and glycolipids. GALE also catalyzes the interconversion of UDP-N-acetylgalactosamine and UDP-N-acetylglucosamine, both of which are also important precursors in glycoprotein and glycolipid synthesis.13,14 Classical (or type I; Online Mendelian Inheritance in Man (OMIM) #230400) galactosemia was the first form of the disease to be associated with a specific enzyme in the Leloir pathway, GALT.16–18 This is probably the most common type of galactosemia in the human population. The most severe forms manifest shortly after birth when a newborn baby takes their first milk meal. Sickness, diarrhea and jaundice often occur and are life-threatening if the disease is not diagnosed rapidly and the baby placed on a galactose-free diet. Many cases of type I galactosemia result in progressive damage to the liver, kidneys, brain, and ovaries. Movement disorders also occur in some patients.19,20 Thus, the pathology and disability increase throughout the early years of life.21,22 Some adult female galactosemia patients are infertile.23,24 Most patients have cognitive as well as physical disabilities. These can manifest as low intelligence quotient (IQ) and behavioral problems.25–27 However, a recent report documents higher IQ scores for some adult galactosemics and the graduation of a patient with type I galactosemia from a university with a degree.28,29 The diversity of symptoms observed in type I galactosemia results partly from environmental variation and also from the great variety of mutations that can cause dysfunction in GALT. Over 300 disease-causing mutations are currently documented.These are largely point (missense) mutations that alter amino acid residues in the protein. However, there are also some deletions and frameshifts.30,31 In the 30 years following the discovery of the association of GALT mutations with type I galactosemia, it became clear that there are rarer forms associated with mutations in the GALK1 and GALE genes.32–36 GALK1 deficiency (type II galactosemia; OMIM #230200) is typically mild, with the only confirmed symptom being early-onset cataracts. There are no confirmed reports that this type of galactosemia is associated with organ damage or cognitive impairment.37,38 However, there are currently no studies following patients with type II galactosemia over their entire lifespan. GALE deficiency (type III galactosemia; OMIM #230350) is probably the rarest type and has the widest spectrum of symptoms. In its mildest form, it results in perturbation of blood chemistry with no reported symptoms at the whole-organism level. The most severe forms resemble type I galactosemia, with similar symptoms, including organ damage and cognitive impairment.39 In 2018, a fourth type of galactosemia was discovered.Type IV galactosemia results from mutations in the gene encoding GALM. The symptoms appear to be relatively mild, resembling those of type II galactosemia.40,41 To date, only a very small number of patients (