[1st Edition] 9781483217000

Table of contents :
Content:
Contributors to Volume 1Page ii
Front MatterPage iii
Copyright pagePage iv
List of ContributorsPage v
PrefacePage viiBERNARD L. HORECKER, EARL R. STADTMAN
Conformational Aspects of Enzyme RegulationPages 1-27D.E. KOSHLAND JR.
Limitation of Metabolite Concentrations and the Conservation of Solvent Capacity in the Living CellPages 29-43DANIEL E. ATKINSON
The Role of Equilibria in the Regulation of MetabolismPages 45-55H.A. KREBS
Regulation of the Biosynthesis of the Branched-Chain Amino AcidsPages 57-76H.E. UMBARGER
On the Roles of Synthesis and Degradation in Regulation of Enzyme Levels in Mammalian Tissues*Pages 77-124ROBERT T. SCHIMKE
The Regulation of the Biosynthesis of α-1,4 Glucans in Bacteria and PlantsPages 125-160JACK PREISS
Allosteric L-Threonine Dehydrases of MicroorganismsPages 161-182W.A. WOOD
The Aspartokinases and Homoserine Dehydrogenases of Escherichia coliPages 183-231GEORGES N. COHEN
Pyruvate Dehydrogenase ComplexPages 233-251LESTER J. REED
Pyruvate Carboxylase*Pages 253-296MERTON F. UTTER, MICHAEL C. SCRUTTON
Author IndexPages 297-310
Subject IndexPages 311-314

Citation preview

Contributors to Volume 1 DANIEL E. ATKINSON GEORGES N. COHEN D. E. KOSHLAND, JR. H. A. KREBS JACK PREISS LESTER J. REED ROBERT T. SCHIMKE MICHAEL C. SCRUTTON H. E. UMBARGER MERTON F. UTTER W . A. W O O D

CURRENT TOPICS IN

Cellular Regulation edited by Bernard L. Horecker Albert Einstein College of Medicine Bronx, New York

·

Earl R. Stadtman National Institutes of Health Bethesda, Maryland

Volume I 7969

ACADEMIC PRESS New York and London

COPYRIGHT ©

1969,

BY ACADEMIC PRESS,

INC.

ALL RIGHTS RESERVED. NO PART OF THIS BOOK MAY BE REPRODUCED I N ANY FORM, BY PHOTOSTAT, MICROFILM, RETRIEVAL SYSTEM, OR ANY OTHER MEANS, WITHOUT WRITTEN PERMISSION FROM THE PUBLISHERS.

ACADEMIC PRESS, INC. I l l Fifth Avenue, New York, New York 10003

United Kingdom Edition published by A C A D E M I C P R E S S , I N C . (LONDON) L T D . Berkeley Square House, London W1X 6BA

LIBRARY OF CONGRESS

CATALOG CARD N U M B E R : 72-84153

PRINTED I N THE UNITED STATES OF AMERICA

List of Contributors Numbers in parentheses indicate the pages on which the authors' contributions begin.

E. ATKINSON (29), Biochemistry Division, Department of Chemistry, University of California, Los Angeles, California GEORGES N. COHEN* (183), Laboratoire d'Enzymologie, Centre National de la Recherche Scientifique, 91 Gif-sur-Yvette, France D. E. KOSHLAND, JR. (1), Department of Biochemistry, University of California, Berkeley, California H. A. KREBS (45), Metabolic Research Laboratory, Nuffield Department of Clinical Medicine, Radcliffe Infirmary, Oxford, England JACK PREISS (125), Department of Biochemistry and Biophysics, University of California at Davis, Davis, California LESTER J. REED (233), Clayton Foundation Biochemical Institute and Department of Chemistry, University of Texas at Austin, Austin, Texas ROBERT T. SCHIMKE (77), Departments of Pharmacology and Biological Sciences, Stanford University, Stanford, California MICHAEL C. SCRUTTON (253), Department of Biochemistry, Rutgers Medical School, Rutgers University, New Brunswick, New Jersey H. E. UMBARGER (57), Department of Biological Science, Purdue University, Lafayette, Indiana MERTON F. UTTER (253), Department of Biochemistry, Case Western Reserve University, School of Medicine, Cleveland, Ohio W. A. WOOD (161) Department of Biochemistry, Michigan State University, East Lansing, Michigan DANIEL

* Present address: Service de Physiologie Microbienne, Institut Pasteur, Paris, France.

Preface Recent years have witnessed rapid advances in our knowledge of the basic mechanisms involved in the regulation of diverse cellular activities such as intermediary metabolism, the transfer of genetic information, membrane permeability, and cellular differentiation and other organ functions. Information gained from the detailed analyses of a large number of isolated enzyme systems, together with results derived from physiological investigations of metabolic processes in vivo, constitutes an ever-increasing body of knowledge from which important generalized concepts and basic principles of cellular regulation are beginning to emerge. However, so rapid are the present advances in the general area of cellular regulation and so diverse are the disciplines involved, that it has become a formidable task for even the expert in a specialized area to keep abreast of the progress in his field. These considerations have prompted us to undertake the publication of a new series of volumes concerned with the recent developments in various areas of cellular reg­ ulation. The new serial publication will be entitled Current Topics in Cellular Regulation. We do not intend that it will consist of compre­ hensive annual reviews of the literature. We hope rather that it will constitute a medium which will, on the one hand, provide contributing authors with an opportunity to summarize progress in specialized areas of study that have undergone substantial developments and, on the other hand, serve as a forum for the enunciation of general principles and for the formulation of provocative theories and novel concepts. To this end editorial review of individual contributions will be concerned primarily with the clarity of presentation and conformity to publication policies. It is hoped that this new series will bring together current knowl­ edge of various aspects of cellular regulation and thereby serve both to enlighten the uninformed and to provide a base of knowledge for those engaged in research in this subject. September, 1969

BERNARD

L.

HORECKER

EARL R. STADTMAN

VII

Conformational Aspects of Enzyme Regulation I

D. E. KOSHLAND, JR.

I I I

Department of Biochemistry University of California Berkeley, California

I. II. III. IV. V.

Introduction The Symmetry Model The Ligand-Induced or Sequential Model Comparison of the Models Observed Systems A. Hemoglobin B. Aspartyl Transcarbamylase C. Yeast Glyceraldehyde-3-phosphate Dehydrogenase . . . . D. Rabbit Muscle Glyceraldehyde-3-phosphate Dehydrogenase . E. Isocitric Dehydrogenase F. Thioredoxin Reductase G. Phosphoenol Pyruvate Carboxylase H. CTP Synthetase I. Aspartokinase VI. Evaluation VII. Future Developments References

1 2 6 16 18 18 19 20 20 21 22 22 22 22 23 24 26

I. Introduction

The control of enzyme action can occur in a number of ways—by product inhibition, by the depletion of substrate, by the failure to syn­ thesize an enzyme, by covalent modification, or by destruction of the enzyme. Although all these mechanisms are available and used by the cell, perhaps the most ubiquitous and efficient process for the regulation of complex metabolic sequences is by the use of effector molecules that are not themselves consumed in the reaction. When it was recognized that the flexibility of an enzyme was an essential feature in its action, the way was paved for the understanding at the molecular level of the process of regulation of enzyme activity by effector molecules that are not themselves consumed in that specific reaction (20-23). This led in turn to the elaboration of the concept of a distinct separate site which played a role in the feedback regulation so prevalent in bacterial systems (11, SO). Feedback inhibition, feedback activation, and the interactions of subunits are today considered to be all part of a common property caused by the conformational interactions within a protein. The evidence that regulatory control operates through effector sites 1

2

D. E. KOSHLAND, JR.

by means of conformational changes led to the necessity of defining these relationships more clearly. The "molecular biology" of regulation therefore seeks the correlation of the kinetic properties of the protein with its structural architecture. This review will emphasize molecular models that attempt to provide such a correlation. Recent reviews that have covered other aspects of the regulatory control, such as the interac­ tion of metabolic pathways, various types of end product inhibition, the evidence for conformational changes, in this volume and elsewhere (3, Jf., 24, 38), should be consulted for other aspects of these subjects. Two comprehensive molecular theories that are capable of explaining allosteric regulation in quantitative terms have been proposed (19, 25, 31). In this review we shall briefly describe these models and attempt to evaluate their present status and future use. ABBREVIATIONS AND DEFINITIONS

Fx = fraction of total sites specific for ligand X which are occupied by ligand X ; L = (T0/R0) of Monod-Wyman-Changeux model = (B 4 /A 4 ) of Koshland-Nemethy-Filmer model; S0.5 is substrate concentration at F s = 0.5; # s or cooperativity index is (S0.9/S0.1) as defined in (25). Saturation curve is plot of F x versus (X) or Fx versus log (X). Heterotropic and homotropic represent interactions between unlike and like ligands, respectively, as defined in reference (31). Protomer is the minimal repeating combination of subunits as defined in reference (31). K$Ai ifÏB, etc. represent association constants of ligand S, I, etc., for conformations A, B, etc., as defined in reference (25). KtAB, KtBC, etc., represent the equilibrium constant for the change from conformation A to conformation B, i.e., (B)/(A) or conformation B to conformation C, i.e., (C)/(B) in the absence of ligand as defined in reference (25). if AB, KBC, etc., represent the strength of the subunit interactions relative to KAA = 1 as defined in reference (25). Conformation A will refer to the conformation of the protomer in the absence of any bound ligand. KSR and if sT are the asso­ ciation constants of S to the R and T states, respectively, and are the reciprocals of K*. and KT of reference (31). Νχ = the number of moles of ligand X bound per moles of protein, = ηΥχ where n is the number of protomers per molecule of protein. II. The Symmetry Model

The symmetry model of Monod, Wyman, and Changeux (31) is an elegant, simple, and imaginative proposal to explain the cooperative kinetics of allosteric proteins in terms of the subunit structure of the

CONFORMATIONAL ASPECTS OF ENZYME REGULATION

3

molecule. The model is described in the following series of statements quoted from the original article. The general properties of allosteric systems may then be stated as follows: (1) Most allosteric proteins are polymers, or rather oligomers, involving several identical units. (2) Allosteric interactions frequently appear to be correlated with alterations of the quaternary structure of the proteins (i.e. alterations of the bonding between subunits). (3) While heterotropic effects may be either positive or negative (i.e. co-operative or antagonistic), homotropic effects appear to be always co-operative. (4) Few, if any, allosteric systems exhibiting only heterotropic effects are known. In other words, co-operative homotropic effects are almost invariably observed with at least one of the two (or more) ligands of the system. (5) Conditions, or treatments, or mutations, which alter the heterotropic interac­ tions also simultaneously alter the homotropic interactions. The model is described by the following statements: (1) Allosteric proteins are oligomers the protomers of which are associated in such a way that they all occupy equivalent positions. This implies that the molecule possesses at least one axis of symmetry. (2) To each ligand able to form a stereospecific complex with the protein there corresponds one, and only one, site on each protomer. In other words, the symmetry of each set of stereospecific receptors is the same as the symmetry of the molecule. (3) The conformation of each protomer is constrained by its association with the other protomers. (4) Two (at least two) states are reversibly accessible to allosteric oligomers. These states differ by the distribution and/or energy of inter-protomer bonds, and therefore also by the conformational constraints imposed upon the protomers. (5) As a result, the affinity of one (or several) of the stereospecific sites towards The corresponding ligand is altered when a transition occurs from one to the other state. (6) When the protein goes from one state to another state, its molecular sym­ metry (including the symmetry of the conformational constraints imposed upon each protomer) is conserved.

A schematic illustration of this model for a tetramer is shown in Fig. 1, where S is assumed to bind only to the R state and I only to the T state. The saturation equation for this case is given by Eq. y

=

S

(S)g 8 ,[l + ^ S R ( S ) ] 3

L[l + tflT(I)]« + [1 +KsR(S)V

m

V ;

(1) where L is the equilibrium constant for the transition from the state R to the state T when no ligand is bound to either state. KSR

4

D. E. KOSHLAND, JR. R State

T State

It S

II S S

SIS

si

s s sis FIG. 1. Schematic illustration of symmetry model of Monod, Wyman, and Changeux for tetrameric protein in which there is exclusive binding of S to the R ( K S R = finite, Ks T = 0) state and of I to the T state.

is the intrinsic association constant of S for an individual site on the R state of the protein. If an activator J binds exclusively to the R state in a manner which is not competitive with S, then Eq. (1) becomes: ΫΒ =

tfSR(S)[l + XSR(S)]3 Γ1+ΚΙΤ(Ι)1*

[l+tfjR(J)J

(2)

+ [1 + KSR(S)V

Some typical curves illustrating the variations in these parameters are shown in Fig. 2. The fraction of the molecules in the R state is given by Eq. (3) for the case in which the binding of S alone is being considered =

[1 + if9R(S)]< L + [1 + ^ S K ( S ) ] 4

^

;

The assumption of exclusive binding as shown in the illustrative ex­ ample is not a necessary requirement of the symmetry model (or of the sequential model described below) but preferential binding is. Equa­ tion (1) describes this mathematically because the addition of ligand

I

2

1

/

I

/ï

V\^l— ^

-

0

0.5

Ls]0

£SU

1 1

(A)

1

/

>/

/L

/L = \000

/ί=ιοο /

1

/

10

1

J

A

^=0.00

=10,000

1

W

/

-

r"°

^

10

^

1

^ ι

/ ß"°

y*\00l

(B)

^

0

2

/7 = 4

1 20

L =1000

i—i

/7=4

1

\

Z. = I000

y^=25

|

FIG. 2. Effect of varying parameters on the shape of the saturation curves of the symmetry model. (A) Effect of L on the saturation curves. (B) Effect of added inhibitor and activator on the shape of the saturation curve [ß = üCiT(I), 7 = KjR(J)]. (C) The effect of preferential binding on the shape of the saturation curve where C is the ratio of KQT/KSR; a represents (β)Κβη.

Y

1.0

Cn

CONFORMATIONAL ASPECTS OF ENZYME REGULATION

6

D. E. KOSHLAND,

JR.

will shift the equilibrium from the T to the R state if the ligand binds preferentially to the R state. If the ligand binds equally well to both states, no such shift will occur and no cooperativity would be observed. The effect on such curves when preferential binding instead of exclusive binding is observed has been calculated (31, 37). A typical example of the effect of nonexclusive binding is shown in Fig. 2C. A qualitative description of this model for the usual case in which it is applied could be given as follows. In the absence of ligand the protein exists largely in one state, i.e, the T state, with only a small fraction preexisting in the R state. As the substrate is added, it binds preferentially to the R state, shifting the equilibrium in that direction thereby making available more high affinity sites for S. The combination of multiple sites on the stabilized R state and the shift in equilibrium results in a sigmoid saturation curve. Addition of the activator J which binds preferentially to the R state would have the same effect. However, if an excess of activator were already present so that the protein was in the R state prior to addition of S, the addition of S would cause no further shift in the protein equilibrium and therefore MichaelisMenten kinetics would be observed. By similar reasoning the saturation curve of an inhibitor I which binds exclusively to the T state would follow a typical Michaelis-Menten saturation curve when added to the unliganded enzyme (almost exclusively in the T state) but would be sigmoid if added to an enzyme which had been largely converted to the R state because of excess substrate. Thus this model will predict the possibility of both positive and negative heterotropic effects (e.g., the effect of A on S or the effect of I on S) but will lead only to positive homotropic effects (e.g., the effect of S on S or I on I ) . The three essential features of this model are seen to be: (a) the existence of an equilibrium between at least two states differing in affi­ nities for ligand and/or catalytic activity of the molecule, (b) the prefer­ ential binding of a ligand to one state, and (c) the conservation of symmetry during any conformational changes of the protein, i.e., all in­ dividual protomers change identically. The affinity of each ligand for a particular site is independent of the number of other sites that are occupied. The sigmoid nature of the saturation curve arises entirely from the shift of equilibrium and the preferential affinity. III. The Ligand-lnduced or Sequential Model

The sequential model of Koshland, Nemethy, and Filmer (25) at­ tempts to explain the kinetic and regulatory properties of proteins from the viewpoint of the induced-fit theory (20-22). I t assumes that ligand-

COXFORMATIONAL ASPECTS OF ENZYME REGULATION CI'

No*

Cl"

7

Na+

O O - E D - FIG. 3. Alternative forms of the change in subunit interactions as result of ligand-induced conformational changes. H stands for hydrophobic bonds that attract the two subunits to each other and are, in this case, presumed not to change during the conformation changes. + and — represent electrostatic charges but could equally well represent hydrogen bond attractions or hydrophobic attractions. I n the top line the situation in which the changes in conformation do not cause any net change in subunit interactions are illustrated. Sodium ions and chloride ions are depicted to indicate t h a t distant -f- and — charges can be neutralized by counterions in the solution. In this case the electrostatic charges are too far apart to provide stabilization of the subunits. Thus, KAB = 1 because there is no change in the subunit interactions. In line 2, the + and — charges are brought near enough to each other to give added stabilization by the change in one subunit, but there is no change in the shape of the unliganded subunit. I t is presumed that counterions, such as sodium and chloride ions, are present in this system also, but the intersubunit charges in this case are close enough to have some positive interaction. In the third line there is again a net stabilization because of the added attraction of the adjacent electostatic charges, b u t the adjacent unliganded subunit is changed in shape as well as stability. In the fourth line, the neighboring subunit is changed in shape so that it is essentially identical to that of the liganded subunit, because of the strong interactions of the adjacent charges. I n the final line, the induced conformation separates attractive opposite charges so that subunit interaction is decreased.

8

D. E. KOSHLAND, JR.

induced conformational changes occur within a subunit. This distortion in one subunit can affect the stability and shape of the neighboring subunits in much the way that distortions in one part of a peptide chain can alter the conformation of adjacent parts of the same peptide chain. When a ligand induces a conformation change in a polymeric enzyme, the stability relationships with neighboring subunits can be changed in various ways as shown schematically in Fig. 3. The strength of the subunit interactions may be increased, decreased or remain the same. The binding of a molecule of ligand will depend therefore not only on parameters expected for a monomeric protein but also on the changes in the subunit interactions. The sequential model analyzes these effects in terms of molecular parameters that explain the cooperativity and kinetic effects in molecular terms. These molecular parameters are used to calculate the molecular species present by considering the amount of each species relative to the concentration of the unliganded oligomer, which is designated as A2 in the dimer case. In this way quantitative values can be placed on the qualitative features of this model of "ligandinduced conformational changes." The equilibrium constant for the process shown in Eq. 4 will be the product of the equilibrium constant for the change from the circle (or A

OO + * =~ dD (A 2)

(ABS)

() '

conformation) to the square (or B conformation) designated as KtAB, mul­ tiplied by the ratio of the affinity constants for the induced versus the original conformation KSB/KSA, multiplied by the change in the subunit interactions, KAB/KAA. In addition a statistical factor of 2 must be included to allow for the fact that there are two forms of ES (squarecircle, circle-square) and only one form of A2. This gives the equilibrium constant (cf. Eq. 5) for the reaction of Eq. (4) : jT _

K

-

2

^t A B^ABÄ"8 B (S)

κΓΑ

,-x

w

If one repeats this process for the binding of the second molecule of ligand as shown in Eq. (6)

[Uo ♦ s

* [Us]

(ABS)

(6)

( B 2S 2)

the derived equation is _ ^tAB^BB^SB(S) 2ÄAB

/,-x

V

CONFORMATIONAL ASPECTS OF ENZYME REGULATION

9

Comparison of this equation with Eq. (6) shows that (after correction for the statistical factor) the binding constant of the second molecule of S will equal the binding constant of the first molecule only if the subunit interaction terms are identical in both. If KBB/KAB is greater than KAB/KAA) however, the binding of the second molecule will proceed more readily than the first and positive cooperativity will be observed. If KBB/ KAB is less than KAB/KAA, negative cooperativity occurs. It is convenient to define a standard state as that state in which the protein contains no bound ligands and to define the protomer interactions KAA in the A state to be equal to 1. This involves no loss in generality but makes it possible to eliminate the repetitive use of KAA in many expressions. It readily becomes apparent that an increase in AB over AA subunit interactions is reflected by a value of KAB greater than 1, a decrease by a value less than 1, and no change, a KAB value equal to 1. An illustration of similar calculations for other molecular species expected in a dimer is shown in Table I. This table illustrates several features of the model. First, the approach is general and is readily ap­ plicable to cases in which several conformations are present. Second, the methods requires no a priori assumptions about the magnitudes of the subunit interactions or the subunit geometry. Third, for any given set of values of these constants certain species will be present in insignificant quantities whereas others will be present in significant amounts. Thus, in the example given, the molecular species present at 10~3 M substrate concentration will be essentially the B 2 S 2 and ABS species. The species A2S and B2S will be present only to the extent of about 1%, and the species B 2 S to a fraction of a percent. As experience is gained with a particular system and the systems in general, good first approximations as to the species likely to be present will reduce the complexity of the mathematical derivations. One can then consider additional species and calculate whether they contribute significantly to the overall behavior. If there is a net stabilization realized by two adjacent subunits in the B conformation over that of adjacent A conformations or adjacent A and B conformations (KAB = 1, KBB > 1), then the binding of the second molecule of S will occur more readily than the binding of the first and a positive homotropic effect will be observed. On the other hand, if there is a repulsion between adjacent subunits in the B conformation (KBB < 1, KAB = 1), then negative homotropic effects will be observed. When there is no change in the subunit interactions (KAB = KBB = 1), a hyperbolic saturation curve or Michaelis-Menten kinetics will be observed.

B2S B2S2

—* H D

—^HH]

2KtABKtACKBC

KhABKBBK>SB(S)*

2K\ABKBBKa(S)

2KtABKtACKBCKSB(S)

2KtABKtACKBCKSc(S)

1

2 X IO"3

4 X IO"2

4 X lO-io

4 X IO"*

0.6

3 X IO" 4

2 X 10-2

Ratio (new species)/(A 2 ) a t (S) — 1 0 - 3 M for values of molecular parameters listed below 6 · 0

Ξ A

» C U = B»

and

t>

Ξ c

Molecular parameters used in specific example: KsA = 10, KSB = 106, KSc = 10 - 2 , KtAB = 10" 4 , 2£tAc = 10~2> ^ A B = 1.5, KBC = 20, K B B = 100, and (S) = 10" 3 . ' c Fraction of total molecular species present will be this number divided b y sum of all species—in this case 2.66 from (A2) = 1, plus all species in column.

6

concentration when S in complex. Conformations : 0

using KAA = 1 as reference state; (d) affinity of ligand for conformation, i.e., KsB is affinity constant for Q ; and (e) substrate

Ratio calculated b y multiplying terms for (a) statistical factor', i.e., two possible species fsTQ ♦ C d ] ; (b) conformation energy term, i.e., KtAB means change from O — " O î ( c ) change in interaction between subunits, i.e., KBC for subunit ΓΤ>

a

BC BCS CBS

O

— * [|>

Π>

—

—

2KtABKABKSB(S)

A B S

2tf S A (S)

*[|]0

*S 2KtABKAB

A

AB

©O

Ratio of concentration of new species relative to (A 2 ) a

—► D O

OO —

Stoichiometric change

New species formed

D I M E R UNDERGOING LIGAND-INDUCED CONFORMATIONAL CHANGES

CALCULATION OF A M O U N T S OF VARIOUS MOLECULAR SPECIES PRESENT FOR A

TABLE I

D. E. KOSHLAND, JR.

11

CONFORMATIONAL ASPECTS OF ENZYME REGULATION

oo

S S K

t A B K BB K S B iS Γ

ΞΟ 2K t

t AB

Κ ΔΒ Κ ς (S) AB

SB

K? »AB

K œœK j

J (J) B

2

□0 Ξ0 AB

2K,, 0 K.clV')

2K t

K.

*AC

*AB

KBC B C KSS K, (S)( I ) B

'C

«0

2K,AeKABKJe(J)

2K tT

O

K AB A B K S5 K,J Î S ) ( J ) B

B

2

F were being carried out nonenzymatically, an energy profile like that of Fig. 2A might be imagined. The high energy barriers would necessitate the use of high temperature or other extreme conditions. For the profile shown, the concentration of intermediate C must be much greater than that of F, so even if the sequence proceeds because of constant removal of product F, the steady-state level of C must be high. A schematic representation of the corresponding metabolic conversion of A to F is shown in Fig. 2B. The role of enzymes, of course, is to supply a path with low activation energy barriers—one in which the

34

DANIEL E .

ATKINSON

FIG. 2. Schematic representation of the conversion of compound A to compound F to illustrate typical differences between a nonbiological pathway (A) and a metabolic sequence (B). The vertical coordinate is a free energy scale.

free energies of the highest-energy transition states A'*, L*, M*, and N : are relatively low. At the same time, and almost equally importantly, various cofactors function in supplying a pathway in which the lowestenergy stable intermediates are at relatively high free energy. Rather than the free compounds B, C, and D, activated derivatives L, M, and N serve as the intermediates in this sequence. Their steady state concen­ trations need be no higher than those of A and F, and in real cases may be considerably lower. The joint result of these two effects—lower­ ing of energy barriers by enzymes and raising of energy pits by activa­ tion of intermediates—is a metabolic sequence that can proceed rapidly at physiological temperature and pH, without accumulation of unde­ sirable concentrations of any intermediates. The smoothing effect caused by enzymatic catalysis and by activation of intermediates in the conversion of pyruvate to acetyl-CoA is shown in Fig. 3. This figure is not drawn to any scale, and is only a rough schematic illustration. The nonbiological pathway (Fig. 3A, solid line), is characterized by high activation barriers, especially between pyruvate and acetaldehyde, and by a very deep pit at acetate in which, as discussed above, nearly all of the carbon of the system would accumulate. Enzymes

35

LIMITATION OP METABOLITE CONCENTRATIONS

pyruvate AcSCoA

AcOH

pyruvate HETPP

Ac lipoate

-AcSCoA

FIG. 3. Schematic comparison (not to scale) of the conversion of pyruvate to acetyl-CoA by three pathways. A, solid line, a hypothetical nonbiological pathway; A, broken line, an enzymatic pathway involving acetate, like the lower pathway of Fig. 1; B, the metabolic route shown in the upper pathway of Fig. 1.

and thiamine pyrophosphate might convert this profile to that shown by the broken line of Fig. 3A. Here the reactions could go rapidly because the enzymes have provided different reaction pathways with transition states of much lower energy, but the end result would be merely a more rapid conversion of pyruvate to acetate. Virtually no acetyl-CoA could be formed, no matter how effective the enzyme for the last step. The metabolic situation, featuring enzymatic catalysis of the reactions of activated intermediates, is illustrated in Fig. 3B. Here the sequence proceeds smoothly and without accumulation of intermediates. III. Levels of Enzyme Activities

Nonenzymatic reactions are typically first order with respect to reactant. The linear dependence of reaction velocity on reactant concentra­ tion shown in Fig. 4A corresponds, in a steady-state system, to a linear dependence of reactant concentration on velocity (Fig. 4B). Thus when reaction sequence (5) is proceeding at steady state, if back reactions ka

AH

kj

kh

G->H-> J — L-*P

(5)

36

DANIEL E . ATKINSON

FIG. 4. Generalized illustration of the relationship between reaction velocity and substrate concentration for a typical first-order reaction (A and B), a typical enzymatic reaction following simple Michaelis kinetics (C and D), and a typical regulatory enzyme (E and F). In C and D, curves a and b correspond to different levels of enzyme, and the broken line is the maximal velocity attainable at satura­ tion of the lower amount of enzyme. In E and F, the curves correspond to different levels of a feedback modifier: h, high; n, normal; I, low.

can be ignored, the velocity is given by Eq. (6). vel = kG (G) = fcH (H) = kj (J) = kL (L)

(6)

From this we obtain directly the familiar fact that steady-state concen­ trations of the intermediates in a sequence of first-order reactions are inversely proportional to the velocity constants of the following reactions : (G) = vel//cG;

(H) = vel/fcH;

(J) = vel/fcj;

(L) = vel//cL

In such a system, an increase in flux of material through the system is accompanied by a proportionate increase in the concentrations of all intermediates. Enzymes are surface catalysts with a limited number of specific cata­ lytic sites. The well-known consequence of this fact is that a simple

LIMITATION OF METABOLITE CONCENTRATIONS

37

enzymatic reaction may be anywhere between first order and zero order with respect to reactant, depending on the extent to which the catalytic sites of the enzyme are saturated. This relationship is seen in the familiar Michaelis plot of velocity as a function of substrate concentration (Fig. 4C). As the catalytic sites approach saturation, the order of the reaction with respect to substrate approaches zero. When the enzyme catalyzes a reaction in a sequence at steady state, a plot of substrate concentration as a function of velocity is more appropriate (Fig. 4D). This figure shows that, in evolving the effective surface catalysts that are essential to permit an active metabolism under mild conditions, the organism has developed a new problem. Increase in velocity is accompanied by a disproportionate increase in substrate concentration. Indeed, if the flux of material through the sequence should approach the maximal velocity of one of the enzymes in the sequence (Vm in Fig. 4D), the concentration of the substrate for this reaction would tend to rise without limit (curve a ) . This situation would be potentially catastrophic. Of course, no metabolite concentration would be expected in fact to reach extremely high levels. Two types of factors would operate to limit the concentration actually attained. As the concentration rises, the reverse reaction may become significant; thus simple mass action or equilibrium considerations may limit the concentration of the intermediate and the maximal rate at which the sequence can operate. The efficacy of mass action control will evidently depend on the thermodynamics of the situa­ tion (especially on the equilibrium constant of the preceding reaction). This is probably not in general an effective limitation; in any case the problem is only moved back one reaction, since now the concentration of the preceding intermediate will tend to rise. The second factor limiting the concentration to which an intermediate can accumulate is dissipation by side reactions. As the concentration of a reactive intermediate rises, so does its tendency to react with other intermediates or with macromolecular components of the cell. Nearly all such side reactions will be deleterious to the cell, and if at all extensive they will probably upset the cell's economy to an unacceptable degree, and cause death. As a sort of variant on removal by side reactions, the intermediate might either leak passively out of the cell or be actively excreted. Espe­ cially in the case of unicellular organisms, this might lead to no more serious consequence than wastage of the primary metabolites and meta­ bolic energy that were used in synthesis of the excess intermediate. But, in general, large increases in the concentrations of intermediates seem likely to be highly undesirable. The most obvious way of avoiding harmful increases in the concentra-

38

DANIEL E. ATKINSON

tions of intermediates is to increase the amount of enzyme, as indicated by curve b of Figs. 4C and 4D. Clearly the rate represented by Vm, which with the lower level of enzyme could be approached only at the expense of unacceptable increases in substrate concentration, is easily attained at moderate substrate concentration when the higher level of enzyme is present. Such considerations emphasize the advantages to the organism of maintaining a reasonably constant ratio of activities of the enzymes in any sequence. It would be highly dangerous for the first enzyme of a sequence to be capable of catalyzing a significantly higher flux than could be handled by all other enzymes of the sequence. Such mecha­ nisms as coordinate derepression, coordinate induction, and sequential induction have presumably evolved as safeguards against the harmful or even fatal consequences that would necessarily follow if only the first enzyme were induced or derepressed. IV. Michaelis Constants It follows from the discussion in the previous section that the Michaelis constant of an enzyme is not a chance characteristic, but one that is of great metabolic importance and that must therefore have been subject to stringent evolutionary control. Since the steady-state concentration of each intermediate must depend on the rate at which the substrate is supplied (the flux through the sequence) and on the Michaelis constant and the activity level of the enzyme that catalyzes the reaction involving that substrate, the Michaelis constants of enzymes must play a major role in establishing the concentrations of metabolic intermediates. It is evident from Fig. 4D that the amount of each enzyme present should be such that the steady-state level of its substrate will be of the order of (but probably appreciably lower than) the Michaelis constant of the enzyme. If the concentration of the substrate were greater than the Michaelis constant of the enzyme that utilizes it, the dynamic safety factor (the factor by which the reaction rate could safely increase without a dangerous rise in substrate concentration) would be consider­ ably less than 2. Probably few metabolic sequences operate with safety factors as small as this. On the other hand, the concentration of substrate could be very much lower than the Michaelis constant only if the frac­ tion of the active sites binding substrate at any given time were very low. This situation would indicate a wastefully high concentration of enzyme.

LIMITATION OF METABOLITE CONCENTRATIONS

39

V. Modulation of Enzyme Activity

Although end-product feedback control and other types of regulation through modulation of enzymatic activity are often thought of pri­ marily as means for ensuring that the concentrations of building-block metabolites such as amino acids and nucleoside triphosphates will not fall below working levels, it is equally important that the concentrations of these compounds and their precursors not rise above the normal range. The first committed step in a reaction sequence is the best site for control of the sequence from any point of view, but perhaps the most compelling reason against control at a later step is the marked increase in concentra­ tion of the intermediates before the regulated step that such control would cause. A typical relationship between reaction velocity and substrate con­ centration for the first enzyme of a biosynthetic sequence is illustrated in Figs. 4E and 4F. The familiar sigmoid shape of the curve indicates that velocity is a higher-than-first-order function of substrate concentra­ tion. Line n represents the properties of the enzyme in the presence of the normal physiological level of the regulatory end product, and the other lines indicate the range through which these properties may be modulated as the end-product concentration varies from low (I) to high (A). The enzyme catalyzing the first committed step in a biosynthetic sequence necessarily competes with one or more other enzymes for their common supply of the branch-point metabolite, and modulation by the end product affects the outcome of this competition. The conditions of competition will, however, vary greatly among different situations. The simplest case is represented schematically in Fig. 5. Here the branch­ point metabolite X is an intermediate in a major pathway, and the amount of X bled off into the sequence leading to product P is small compared to the amount used in the major pathway. In such a case the element of competition is minimized, and the concentration of X will be affected very little by fluctuations in the rate of synthesis of P. The first enzyme need only catalyze its specific reaction at a faster or slower rate as the concentration of end-product P tends to fall or rise. The operation of this control is indicated in the graphs of Fig. 5. For convenience in illustration, the reaction is shown as proceeding at only slightly less than half of the maximal rate when the concentration of P is normal (curve n). (Probably in a real case this normal rate would

40

DANIEL E. ATKINSON

FIG. 5. Schematic illustration of regulation at a metabolic branch point byaction of a feedback modifier. The broken line indicates the steady-state concentra­ tion of substrate. Curves I and h, as in Fig. 4F, represent the extremes of enzyme response as the concentration of the modifier varies from low (I) to high (h). Curve n represents the normal level of the feedback modifier. The curves drawn with heavy lines indicate that the modifier concentration is normal in graph A, but somewhat below normal in graph B.

be considerably lower than half maximal, as discussed in the preceding section, but this would not affect the present discussion in any significant way.) If the concentration of end product falls slightly, the change in fraction of modifier sites occupied by product will result in a change in the properties of the enzyme (heavy curve of Fig. 5B). I t is clear that a relatively small change in affinity for substrate will lead to a large change in velocity of the reaction catalyzed. Thus the typical sigmoid shape of the characteristic curve for regulatory enzymes is ideally suited for very high sensitivity of control. A more general and slightly more complex control situation is illus­ trated in Fig. 6. Here the branch-point metabolite X serves only as the starting point for two (or more) synthetic sequences that proceed under normal conditions at about the same rate. In such a case, the concentration of X may be affected significantly by changes in properties

LIMITATION OF METABOLITE CONCENTRATIONS

41

of the enzymes that utilize it. In the interests of simplification, the concentration of product Q is assumed to remain constant, so that the consequences of variation in P can be considered in isolation. Obviously in a real case both product concentrations will tend to vary, so that both enzymes I and II will be modulated simultaneously. In Fig. 6A, both products, P and Q, are considered to be at the midpoints of their normal ranges. A slight decrease in the concentration of P would cause the properties of enzyme I to change (heavy line of Fig. 6BI). The more rapid use of metabolite X by this enzyme might cause the concentration of X to decrease, thus reducing the rate at which

FIG. 6. Schematic illustration of some aspects of the competition between two regulatory enzymes (I and II) for a branch-point metabolite ( X ) . T h e broken line in each graph indicates the level of the common substrate, X. The curves represent the behavior of the enzymes when the concentration of the feedback modifier ( P or Q) varies from low (I) to high (h). (A) Normal level of P ; normal level of Q. (B) Below normal level of P ; normal level of Q. (C) High level of P ; normal level of Q.

42

DANIEL E. ATKINSON

this compound can be used by enzyme II. I t is evident that enzyme with response curves like those shown here must compete sharply for their common intermediate. If product P were available from the environment at a level more than sufficient to meet metabolic demands, the regulatory sites of enzyme I would be saturated and the enzyme would respond as shown by the heavy curve of Fig. 6CI. Almost no X would enter the pathway leading to P, but the synthesis of Q could proceed unaffected. The level of sub­ strate X is shown as rising slightly because of the decrease in demand caused by cessation of synthesis of P. The concentration of X itself will, of course, also be controlled by enzyme modulation. Presumably in some cases this control will be so effective that changes in concentration caused by fluctuating rates of use will be negligible. The type of competition between enzymes repre­ sented in Fig. 6 is in all essential aspects independent of whether the concentration of the common precursor varies appreciably or not, and slight variation was shown in the figure for the sake of generality. It is evident that, whether the concentration of X is virtually fixed or is slightly variable, response curves of the type illustrated are uniquely well suited to lead to very sensitive competition between enzymes, and that the outcome of the competition will be affected strongly by even slight modification of either response curve caused by changes in the concentration of its feedback modifier. The system thus assures that the concentrations of P and Q, and of all intermediates between X and these products, will be held low, and at the same time provides that the concentrations of P and Q will be protected from falling below the physiologically desirable level. VI. Summary Conservation of low concentrations of metabolites, both individually and collectively, is one of the most fundamental requisites for a viable metabolizing system. The rate of flow of material into a metabolic sequence is typically controlled by modulation of the enzyme catalyzing the first committed step (by end product, adenine nucleotides, etc.). If concentrations of intermediates are to be kept low, it is necessary in addition that the Michaelis constant of each enzyme involved in a sequence have an appropriate value and that the activity of each enzyme be high enough to handle the maximal flux through the sequence without the substrate concentration rising much above this Michaelis value. Evidently if the first enzyme is derepressed (or induced), the activities of the other enzymes of the sequence must be correspondingly

LIMITATION OF METABOLITE CONCENTRATIONS

43

increased if harmful increases in concentrations of intermediates are to be avoided; thus coordination of derepression or induction of the enzymes involved in a metabolic sequence is necessary. These kinetic controls would be insufficient to maintain low concentrations of inter­ mediates if some of the reactions involved markedly unfavorable equi­ libria. For such a reaction, a useful level of the product could be produced only if the substrate concentration were very high. This situation is avoided through the use of activated derivatives that, even at low con­ centrations, are thermodynamically capable of conversion to products at useful levels. These considerations lead to the conclusion that activa­ tion of metabolic intermediates, coordinate derepression or induction, and enzyme modulation all contribute importantly to stability and sur­ vival through aiding in the conservation of low concentrations within the cell.

The Role of Equilibria in the Regulation of Metabolism I

H. A. KREBS

I I I I |

Metabolic Research Laboratory Nuffield Department of Clinical Medicine Radcliffe Infirmary Oxford, England

I. Introduction II. The Role of the Lactate Dehydrogenase System in the Control of Gluconeogenesis from Lactate III. The Role of Equilibria in the Alcohol Dehydrogenase System in the Metabolism of Ethanol IV. Extrahepatic Ketogenesis V. Equilibria in the Metabolism of Propionate VI. Summing Up References

45 46 49 50 52 54 55

I. Introduction

For the analysis of the factors that determine the overall rates of metabolic pathways, it has been useful to classify the individual steps into two categories: steps that, on account of the high activity of the enzyme concerned, are so rapid that they establish near-equilibrium be­ tween starting materials and end products of the step; and steps that, for one reason or another, do not establish equilibrium (5, 16, 22). Exam­ ples of near-equilibrium and nonequilibrium steps are shown in Table I for glycolysis and in Table II for the tricarboxylic acid cycle and related reactions. It is generally held that the enzymes of nonequilibrium steps play the predominant role in the control of the overall rate of pathways because the very high potential activity of the "equilibrium enzymes" is far in excess of the normal flux rate. This excess is assumed to make control of flux, i.e., variations of rates, impracticable. For example, a 90% inhibition of highly active enzymes, such as hexosephosphate isomerase or fumarase, is not likely to affect flux rates appreciably. The study of enzymatic control mechanisms has therefore concentrated on "nonequilibrium enzymes/' the activity of which is relatively low and therefore potentially rate limiting. When an enzyme normally functions at near its maximum capacity any variations in this capacity are liable to be transmitted to the whole pathway. The study of such rate-limiting enzymes, for example of phosphofructokinase, pyruvate kinase, citrate synthase, and pyruvate dehydrogenase has proved very fruitful in the 45

46

H. A. KREBS TABLE I EQUILIBRIUM AND NONEQUILIBRIUM STEPS I N GLYCOLYSIS

Equilibrium

Nonequilibrium

Hexose-P isomerase Triose-P isomerase Aldolase (?) Glyceraldehyde-P dehydrogenase Phosphoglycerate kinase Phosphoglycerate mutase Enolase Lactate dehydrogenase

Hexokinase Phosphofructokinase Pyruvate kinase

TABLE II EQUILIBRIUM AND NONEQUILIBRIUM STEPS IN THE TRICARBOXYLIC ACID CYCLE AND RELATED REACTIONS

Equilibrium

Nonequilibrium

Aconitase Fumarase Malic dehydrogenase Transaminases Isocitrate dehydrogenase (NADP)

Citrate synthase α-Oxoglutarate dehydrogenase Succinic dehydrogenase Pyruvate dehydrogenase Acyl thiokinase Isocitrate dehydrogenase (NAD)?

analysis of control mechanisms. It has revealed the feedback inhibitions and activations which correlate the activities of these enzymes to physio­ logical needs. However, these considerations do not imply that equilibrium steps cannot also make essential contributions to the regulation of metabolic processes, for the simple reason that the rate of enzymatic reactions is not solely dependent on the amount of active enzyme. No less impor­ tant can be the concentration, and the rate of supply, of substrate, because many enzymes of intermediary metabolism are not saturated with substrate in situ. In fact the rate of many intermediary steps is determined by the rate of supply of the substrate. II. The Role of the Lactate Dehydrogenase System in the Control of Gluconeogenesis from Lactate

An example, the synthesis of glucose from lactate in liver or kidney, illustrates the principle. The initial steps of gluconeogenesis from lactate

47

EQUILIBRIA IN METABOLIC REGULATION

are shown in Fig. 1. Rate-limiting are the steps between pyruvate and phosphopyruvate, catalyzed by pyruvate carboxylase and phosphopyruvate carboxykinase. The main, but not the only, rate-limiting enzyme is pyruvate carboxylase, which is a nonequilibrium enzyme. Although the first step, catalyzed by lactic dehydrogenase, is so rapid that it establishes near-equilibrium between lactate, pyruvate, NAD, and NADH, it can also limit the rate of gluconeogenesis because this equi­ librium determines the concentration of pyruvate, which is a critical factor in the pyruvate carboxylase reaction. The Km of pyruvate carboxylase is near 10~4 M. According to Scrutton and Utter- (26) it Lactate (lactate dehydrogenase)

I[ Pyruvate

(pyruvate carboxylase)

I

+ C0 2 + ATP

Oxaloacetate (phosphopyruvate carboxykinase)

I

+ GTP

Phosphopyruvate + C0 2

Glucose FIG. 1. Initial pathway of gluconeogenesis from lactate.

is 4.4 X IO-4 M for purified chicken liver pyruvate carboxylase (25° ; pH 7.0) ; Krebs and Stubbs (20) found 4.4 X 10 4 M for rat liver pyr­ uvate carboxylase (37°; pH 7.5). The latter measurements were made in crude homogenates where the activity of pyruvate carboxylase was measured according to the method of Ballard and Hanson (2) with varia­ tions in the pyruvate concentration. The pyruvate concentrations in liver and other animal tissues are of the same order as the Km and may vary between, say, 0.04 and 0.5 mM. Two factors are responsible for the variations. Either the sum of lactate plus pyruvate changes, e.g., after physical exercise when it can increase 10-fold, or the [NAD] : [NADH2] ratio changes at a constant sum of [lactate] plus [pyruvate]. When this occurs the [lactate] : [pyruvate] ratio moves in parallel with the [NADH2] : [NAD] ratio because the components of the lactate dehydro­ genase system adjust themselves to near-equilibrium concentrations on account of the high dehydrogenase activity. The relation therefore holds, [lactate]

_ 1_

[pyruvate] " ~K X

[NADHJ [NAD]

48

H. A. KREBS

where K is the equilibrium constant and [NADH 2 ] and [NAD] refer to the concentrations of the free pyridine nucleotides. Thus at a constant sum of [lactate] plus [pyruvate] a rise of the [NADH 2 ] : [NAD] ratio causes a fall of [pyruvate]. Changes in the [NADH 2 ] : [NAD] ratio can experimentally be pro­ duced in the liver by ethanol. Ethanol inhibits gluconeogenesis from lactate in the liver (17, 18), and this inhibition is related to, but not directly caused by, the reduction of NAD in the alcohol dehydrogenase reactions. The result is a rise in the [NADH 2 ] : [NAD] ratio. The im­ mediate cause of the inhibition of gluconeogenesis is not the change of the redox state of the NAD couple, but the related fall of the steadystate concentration of pyruvate. To sum up: the equilibrium in the lactate dehydrogenase system, by determining the concentration of pyr­ uvate, plays a decisive role in controlling the rate of gluconeogenesis. A physiological situation, where the lactate dehydrogenase equilibrium operates as a controlling factor, is physical exercise. The lactate released by the muscles causes a rise of [lactate] in the blood and liver, and at a constant redox level of the pyridine nucleotides the pyruvate concen­ tration rises as a consequence of the equilibrium. Evidence that it is not the redox state of the pyridine nucleotides that controls the rate of gluconeogenesis is the fact that the rates of gluconeogenesis are, within wide ranges, independent of the [lactate]: [pyruvate] ratio. Such variations can be produced experimentally by the addition of relatively high concentrations (10 mM) of either lactate or pyruvate. The rates of gluconeogenesis with 10 mikf lactate and 10 mM pyruvate are approximately equal, although the [lactate] : [pyr­ uvate] ratios may differ by a factor of 10 and more. For example, when lactate, in one dose, was added to the medium perfusing rat liver to bring the concentration of lactate to 10 mM, the [lactate] : [pyruvate] ratio in the tissue 20 minutes later was 10.7. When pyruvate was added instead, the ratio was 1.1. Yet the rates of gluconeogenesis were the same (about 1.0 pinole of glucose per minute per gram wet weight) (#4). Even greater differences in the [lactate] : [pyruvate] ratio can be pro­ duced experimentally with mouse liver slices (19a) and with kidney cortex slices (18a) without affecting the rates of gluconeogenesis. Further evidence that the concentration of pyruvate is the crucial factor is pro­ vided by the fact that the inhibition of gluconeogenesis by ethanol can in part be overcome by a high lactate concentration, which raises the pyruvate concentration (18). The concentration of pyruvate, as established by the equilibrium in the lactate dehydrogenase system, also determines the extent of

EQUILIBRIA IN METABOLIC REGULATION

49

the pyruvate dehydrogenase reaction in the liver or kidney. This follows from the fact that at the relatively low concentrations of pyruvate gen­ erated by the addition of 10 mM lactate to the perfused starved liver, the conversion of lactate, via pyruvate, to glucose is quantitative (11). This means that the pyruvate arising from lactate is quantitatively carboxylated to oxaloacetate and that none is dehydrogenated to serve as a fuel for the tricarboxylic acid cycle. The acetyl-CoA required for the cycle is provided in the starved liver by fatty acids. If, however, an excess of pyruvate (e.g., 10 mM) is added to the perfused starved liver, pyruvate also supplies acetyl-CoA for the tricarboxylic acid cycle and for ketone body formation. This effect of high concentrations of pyruvate can also be demonstrated in slices of mouse liver (19a) and of rat kidney cortex (18a). Thus the activity of pyruvate dehydrogenase depends on the steady-state concentration of pyruvate; this, in turn, is determined by equilibrium reactions. III. The Role of Equilibria in the Alcohol Dehydrogenase System in the Metabolism of Ethanol

Similar in importance to the equilibrium state of the lactate dehy­ drogenase system is the equilibrium in the alcohol dehydrogenase system in the liver. The rate-limiting reaction in the conversion of ethanol to acetic acid (the main metabolic product of ethanol in the liver) is deter­ mined by the steady-state concentration of acetaldehyde, which in turn depends on the equilibrium position of the alcohol dehydrogenase system. When the concentrations of ethanol and acetaldehyde in the perfused liver were determined 30 minutes after the addition of 10 mM lactate and 10 mM ethanol the fethanol] : [acetaldehyde] ratio had about half the value of the [lactate] : [pyruvate] ratio, and the values of these ratios were reasonably constant (Table I I I ) . From the known equilib­ rium constants of the ethanol (1) and lactate (32) dehydrogenase reac­ tions it can be calculated that the theoretical ratio of the two ratios should be 2.0, which is the same, within the limits of error, as the ob­ served value of 2.42 ± 0.43 (4 observations). Thus the reactants of the alcohol dehydrogenase system are in equilibrium with those of the lactate dehydrogenase system. Further relevant findings are the following. Ethanol removed by the perfused liver was almost quantitatively recovered as acetate (neglecting the small amount of acetaldehyde that accumulated in the liver), and the rate of the acetate formation in rat liver was of the order of 0.93 db 0.38 ^mole/minute per gram. The capacity of alcohol dehydro-

50

H. A. KREBS

genäse at 37°C in the presence of 10 mM ethanol is considerably greater at the physiological pH. The aldehyde dehydrogenase ac­ tivity (combined NAD-linked dehydrogenase and flavoprotein-linked en­ zyme), at 37°C, was, under the test conditions, about the same as the rate of acetate formation at the concentration of acetaldehyde produced from ethanol. Hepatic ethanol metabolism and its regulation may therefore be sum­ marized thus: ethanol is metabolized in the liver by two steps leading via acetaldehyde to acetate. The first step establishes equilibrium in TABLE III CONCENTRATIONS OF LACTATE, PYRUVATE, ETHANOL, AND ACETALDEHYDE IN PERFUSED RAT LIVER 30 MINUTES AFTER ADDITION OF LACTATE (10 mM) AND ETHANOL (10 mM°)

Substrate Lactate Pyruvate Ethanol Acetaldehyde [Lactate] : [pyruvate] [Ethanol] : [acetaldehyde] [Lactate] : [pyruvate] [ethanol] : [acetaldehyde]

Concentration6 4.32 0.178 4.21 0.397 25 11.5 Found: 2.42 Calculated: 2.0

±0.55 ± 0.024 ±0.17 ± 0.056 ± 1.22 ±2.35 ± 0.43

° Unpublished data of Krebs and Hems obtained with freeze-clamped tissue. b The values are expressed as micromoles per gram of liver, means and standard error of the mean (4 observations). the alcohol dehydrogenase system. The rate of the second step depends on the concentration of acetaldehyde determined by the first step. This first step is therefore the rate-limiting factor in the removal of ethanol. IV. Extrahepatic Ketogenesis A third example illustrating the importance of equilibrium reactions concerns extrahepatic ketogenesis. It has been known for some time that tissues other than liver, especially kidney cortex and cardiac muscle, can form ketone bodies (7, 8, 13, 19, 30, 31), though the liver is by far the most important site of ketogenesis. In ruminants the epithelium of the intestinal tract can also make major contributions to ketogenesis {12,23). It is now clear that, in general, extrahepatic ketogenesis [with the probable exception of rumen and intestinal epithelium (12) ] is caused

EQUILIBRIA I N METABOLIC

51

REGULATION

by a mechanism different from that of hepatic ketogenesis and is the consequence of equilibrium reactions. Hepatic ketogenesis is brought about via the hydroxymethyl glutaryl-CoA pathway discovered by Lynen (21 ; see also 25, 34) in 1958 (Fig. 2). A key enzyme of this pathway, HMG-CoA synthase, is virtually absent from kidney cortex and cardiac muscle (33). On the other hand, these tissues possess 3-oxoacid-CoA-transferase, an enzyme absent from the liver. In view of the weakness of acetoacetyl-CoA deacylase in kidney (33), the most 2 Acetyl-CoA thiolase 1 Γ Acetoacetyl-CoA + CoA HMG-CoA synthase

I|

+ acetyl-CoA

Hydroxymethyl-glutaryl-CoA HMG-CoA lyase

1

Acetoacetate + acetyl-CoA FIG. 2. Pathway of hepatic ketogenesis

(21).

likely pathway of renal ketogenesis consists of the following two reactions : thiolase

2 Acetyl-CoA ^

acetoacetyl-CoA + CoA

3-oxoacid-CoA-transf erase

Acetoacetyl-CoA + succinate ^

acetoacetate + succinyl-CoA

The activities of thiolase and 3-oxoacid-CoA-transferase are high enough to account for the observed rates of ketogenesis (33). This mechanism was first discussed for heart muscle by Beinert and Stansly (3), Stern (27), and Hartmann and Lynen (10). Whenever acetyl-CoA accumu­ lates, a proportion is converted to acetoacetyl-CoA and made available for the transferase reaction. The physiological role of the above two reactions probably lies solely in their reverse reactions which convert acetoacetate to acetyl-CoA. Heart and kidney readily utilize acetoacetate as a fuel of respiration (30, 35) and this process of utilization is initiated by the transferase and thiolase reactions. Once the enzymes are present and establish nearequilibrium, ketone bodies are bound to arise when there is a surplus of acetyl-CoA. Such a surplus occurs experimentally when unphysiologically high concentrations of butyrate or oleate are added to the perfused, sliced, or homogenized tissue. This situation is not likely to arise in vivo because the concentration of plasma fatty acids is so regulated

52

H . A. KREBS

as to prevent the occurrence of the above two reactions to an appreciable degree in the direction of acetoacetate formation. It might be thought that the unfavorable equilibrium position of the thiolase system (the equilibrium constant (9) being about 5 X 10~5) [acetoacetyl-Co A] [Co A] [acetyl-CoA]2

makes a sufficient formation of acetoacetyl-CoA unlikely. However, this is not a valid argument because the thiolase reaction in the unfavorable direction is also a step in hepatic ketogenesis. In this tissue the unfavor­ able equilibrium is compensated by the kinetic and thermodynamic char­ acteristics of the HMG-CoA synthase reaction. This appears to have a low Michaelis constant for acetoacetyl-CoA and is virtually irreversi­ ble; it therefore pulls the thiolase reaction in the unfavorable direction, i.e., the formation of acetoacetate. Analogous considerations apply to the coupling of the thiolase and transferase reactions in extrahepatic tissues. V. Equilibria in the Metabolism of Propionate

In the liver and in kidney cortex, propionate is readily converted to glucose. This process is initiated in the cytoplasm by the reactions Propionate + CoA + ATP

propionyl-CoAsynthetase

=± propionyl-CoA + AMP + PP

carnitine acyl

Propionyl-CoA + L-carnitine

= ^ propionyl-L-carnitine + CoA

transferase

Propionyl carnitine enters the mitochondria where it is converted to succinyl-CoA by the following reactions: Propionyl-L-carnitine + CoA ^ Propionyl-CoA + HCO*" + ATP

carnitine acyl

propionyl-CoA + L-carnitine

(1)

methylmalonyl-CoA + ADP + Pi

(2)

transferase

propionyl-CoAcarboxylase

-J

methylmalonyl-CoA-

Methylmalonyl-CoA *

racemase + mutase

succinyl-CoA

Succinyl-CoA —» succhiate + CoA

The formation of glucose from succinate involves another twelve steps, which are not of immediate concern in the present context.

53

EQUILIBRIA IN METABOLIC REGULATION

The overall rate of glucose synthesis from propionate in kidney cortex slices depends in a characteristic way on the concentrations of propionate, carnitine, phosphate, and bicarbonate (29). Propionate above 5 mAf inhibits the synthesis. Carnitine has a large stimulating effect in phos­ phate-buffered saline, but not in bicarbonate-buffered saline (Table IV) ; or, what amounts to the same, bicarbonate accelerates the rate of syn­ thesis when no carnitine is added and phosphate inhibits it. At higher concentrations (above 5 mAf) carnitine inhibits glucose synthesis, and this inhibition varies with the propionate concentration (29). The followTABLE IV EFFECT OF BICARBONATE AND CARNITINE ON GLUCONEOGENESIS FROM PROPIONATE IN KIDNEY CORTEX"

Saline medium

Glucose formed6

No bicarbonate, 6 mM phosphate No bicarbonate, 6 mAf phosphate, 4 mAf carnitine 25 mAf bicarbonate, no phosphate 25 mAf bicarbonate, no phosphate, 4 mM carnitine

21 73 73 89

a

Kidney cortex slices were incubated in phosphate-buffered or bicar­ bonate-buffered saline in the presence of 2.5 mAf propionate. For details see (29). 6 Micromoles per gram dry weight per hour.

ing considerations show how these effects can be explained on the basis of the fact that the reactions leading from propionate to methylmalonylCoA are all reversible (6, 28). The sum of the reactions (1) and (2) is: Propionyl-L-carnitine + CoA + ATP + HC0 3 _ ^ L-carnitine + ADP -f Pi + methylmalonyl-CoA

At equilibrium the following relation holds : _ [L-carnitine] ~ [propionyl-L-carnitine]

[ADP] [Pi] [ATP]

[methylmalonyl-CoA] [CoA][HC03"]

or [Methylmalonyl-CoA] = K X

[CoA][HCQ3-][ATP] rAFIP1rp1

[ADP] [Pi]

X

[propionyl-L-carnitine] ; :^—j [L-carnitine]

. (3)

(

It is very likely that the concentration of methylmalonyl-CoA limits the rate of glucose formation from propionate because the rate of glucose synthesis from succinate is much higher than from propionate. If this

54

H. A. KREBS

is the case it would be expected from Eq. (3) that bicarbonate accelerates, and carnitine and Pi inhibit, glucose synthesis since at equilibrium the concentration of methylmalonyl-CoA rises with that of bicarbonate and falls with those of carnitine and Pi. Thus several of the kinetic charac­ teristics of propionate metabolism indicate that equilibria play a role in determining the rate of propionate utilization. Independent evidence for the reversibility of the reactions of pro­ pionate metabolism has been supplied by experiments of B0hmer (4), who found that substances that give rise to a formation of succinyl-CoA all cause an increase in the concentration of propionyl-carnitine in rat liver mitochondria (Table V) and suggests that the raised concentration TABLE V EFFECT

OF

METABOLITES

ON

THE

FORMATION

OF

PROPIONYL CARNITINE BY RAT LIVER MITOCHONDRIA ACCORDING TO B 0 H M E R

Additions (3.3 milf ) None Citrate Oxoglutarate Succinate Malate Pyruvate

(4)

Propionyl carnitine found (nmoles/mg protein) 0.1 4.7 1.6 1.8 0 0.1

of succinyl-CoA might be the reason for the increased formation of propionyl-carnitine, though he does not discuss the nature of the mecha­ nism. Since in addition to the reactions involved in Eq. (3) the conver­ sion of methylmalonyl-CoA to succinyl-CoA is also readily reversible [the equilibrium concentration is about 1:20 in favor of succinyl-CoA (15) ], an increase in the concentration of succinyl-CoA would necessarily lead to an accumulation of propionyl-carnitine. VI. Summing Up It is customary, and rightly so, to single out enzymes taking part in metabolic sequences as "regulatory" enzymes because they are en­ dowed with allosteric (feedback) control properties. It is pointed out that the effectiveness of such regulatory enzymes depends not only on their potential catalytic activity, but also on the concentration of their substrates, which is regulated by the equilibrium state of other enzyme

EQUILIBRIA IN METABOLIC REGULATION

55

systems. This principle is illustrated by examples taken from the metabo­ lism of pyruvate, ethanol, propionate, and ketone bodies. REFERENCES

1. Bäcklin, K.-L, Ada Chem. Scand. 12, 129 (1958). 2. Ballard, F. J., and Hanson, R. W., Biochem. J. 104, 866 (1967). 5. Beinert, H., and Stansly, P. G., J. Biol. Chem. 204, 67 (1953). i. B0hmer, T., Biochim. Biophys. Ada 164, 487 (1968). 6. Bücher, T., and Rüssmann, W., Angew. Chem. 75, 881 (1963). 6. Fritz, I. B., Schultz, S. K., and Srere, P. A., J. Biol. Chem. 238, 2509 (1963). 7. Geyer, R. P., and Cunningham, M., J. Biol. Chem. 184, 641 (1950). 8. Geyer, R. P., Cunningham, M., and Pendergast, J., J. Biol. Chem. 185, 461 (1950). 9. Goldman, D. S., J. Biol. Chem. 208, 345 (1954). 10. Hartmann, G., and Lynen, F., in "The Enzymes" (P. D. Boyer, H. Lardy, and K. Myrbäck, eds.), 2nd Ed., Vol. 5, p. 381. Academic Press, New York, 1961. 11. Hems, R., Ross, B. D., Berry, M. N., and Krebs, H. A., Biochem. J. 101, 284 (1966). 12. Hird, F. J. R., and Symons, R. H., Biochem. J. 84, 212 (1962). 15. Jowett, M., and Quastel, J. H., Biochem. J. 29, 2181 (1935). 14. Kaziro, Y., and Ochoa, S., Advan. Enzymol. 26, 283 (1964). 16. Kellermeyer, R. W., Allen, S. H. G., Stjernholm, R., and Wood, H. G., J. Biol. Chem. 239, 2562 (1964). 16. Klingenberg, M., and Bücher, T., Ann. Rev. Biochem. 29, 669 (1960). 17. Krebs, H. A., Advan. Enzyme Regulation 6, 467 (1968). 18. Krebs, H. A., Freedland, R. A., Hems, R., and Stubbs, M., Biochem. J. 112, 117 (1969). 18a. Krebs, H. A., Gascoyne, T., and Notton, B. M., Biochem. J. 102, 275 (1967). 19. Krebs, H. A., and Johnson, W. A., Biochem. J. 31, 647 (1937). 19a. Krebs, H. A., Notton, B. M., and Hems, R., Biochem. J. 101, 607 (1966). 20. Krebs, H. A., and Stubbs, M., unpublished data. 21. Lynen, F., Henning, IL, Bublitz, C , Sorbo, B., and Kròplin-Rueff, L., Biochem. Z. 330, 269 (1958). 22. Newsholme, E., and Gevers, W., Vitamins Hormones 25, 1 (1967). 23. Pennington, R. J., Biochem. J. 51, 251 (1942). 24. Ross, B. D., Hems, R., and Krebs, H. A., Biochem. J. 102, 942 (1967). 25. Sauer, F., and Erfle, J. D., J. Biol Chem. 244, 30 (1966). 26. Scrutton, M. C , and Utter, M. F., J. Biol. Chem. 240, 1 (1965). 27. Stern, J. R., in "Methods in Enzymology" (S. P. Colowick and N. O. Kaplan, eds.), Vol. 1, p. 573. Academic Press, New York, 1955. 28. Tietz, A., and Ochoa, S., /. Biol. Chem. 234, 1394 (1959). 29. Weidemann, M. J., and Krebs, H. A., Biochem. J. I l l , 69 (1969). 30. Weidemann, M. J., and Krebs, H. A., Biochem. J. 112, 149 (1969). 31. Weil-Malherbe, H., /. Soc. Chem. Ind. (London) 55, 838 (1936). 32. Williamson, D. H., Lund, P., and Krebs, H. A., Biochem. J. 103, 514 (1967). S3. Williamson, D. H., D. Phil, thesis Oxford Univ., Oxford, England, 1967. 34. Williamson, D. H., Bates, M. W., and Krebs, H. A., Biochem. J. 108, 353 (1968). 36. Williamson, J. R., and Krebs, H. A., Biochem. J. 80, 540 (1961).

Regulation of the Biosynthesis of the Branched-Chain Amino Acids H. E. UMBARGER

Department of Biological Science Purdue University Lafayette, Indiana I. Historical Introduction II. Regulation of Metabolic Flow by End-Product Inhibition . . . A. Inhibition of Threonine Deaminase B. Inhibition of Acetohydroxy Acid Synthetase C. Inhibition of α-Isopropylmalate Synthetase III. Control of Enzyme Level in the Pathways to the Branched-Chain Amino Acids A. Repression of the Leucine Biosynthetic Enzymes B. Repression of the Isoleucine and Valine Biosynthetic Enzymes . IV. The Inhibition of Growth of Eschenchia coli Strain K12 by Valine Note Added in Proof References

57 58 59 64 67 68 68 70 73 74 75

I. Historical Introduction For a long time the branched-chain amino acids have been considered under a common heading in biochemistry textbooks. The basis for this grouping was solely their structural relationship since, before the advent of isotopie tracer techniques and mutant methodology, so little was known of their catabolism and nothing was known of their biosynthesis. Indeed, on the basis of their catabolism, they might have been considered separately for studies with the diabetic dog revealed that leucine was a "ketogenic" amino acid, valine was "glycogenic" whereas isoleucine was both. One of the earliest rational reasons for suspecting that they might be metabolically related was an antagonism observed by Gladstone (21) in 1939. He noted that the addition of any one of the three branchedchain amino acids to an otherwise sufficient medium prevented the growth of the anthrax bacillus. The three amino acids added together, however, were stimulatory. Gladstone suggested that each of the amino acids might have interfered with either the utilization or the formation of the other two. Although this case has not been reexamined, examples could be cited today that illustrate both possibilities suggested by Gladstone. 57

58

H . E. UMBARGER

With the elucidation of the biosynthetic pathways leading to the branched-chain amino acids by isotopie, enzymatic, and genetic studies, it became quite clear that the three amino acids were indeed quite closely related to each other with respect to their biosynthesis. Although it is not within the scope of this chapter to review the evidence that led to the demonstration of these pathways, reference will be repeatedly made to the responsible enzymes and the reactions they catalyze, which are schematically represented in Fig. 1. For a fairly complete review a-Ketoisocaproate -

-Leucine

/?-Isopropylmalate

À 7 a-lsopropylmalate

Pyruvate -

l-/3-Dihydroxyisovalerate

-►a-Ketoisovalerate -

►Valine

- a - Acetohydroxy- -U-a-£-Dihydroxy/3-Methyl valerate butyrate

- ► a - K e t o - £ - Methyl valerate

-Isoleucine

-a-Acetolactate -

Active Acetaldehyde

a-Ketobutyrate-

Threonine

FIG. 1. Biosynthetic pathways leading to the branched-chain amino acids. The enzymes catalyzing the reactions are indicated by arabic numerals; 1, threonine deaminase; 2, acetohydroxy acid synthetase; 3, acetohydroxy acid isomeroreductase ; 4, dihydroxy acid dehydrase; 5, transaminase B and, for the valine pathway only, valine-alanine-aminobutyrate transaminase; 6, a-isopropylmalate synthetase; 7, isopropylmalate isomerase; 8, ß-isopropylmalate dehydrogenase; 9, transaminase B and a glutamate-leucine transaminase with an undefined specificity.

of that evidence, the reader is referred to an older review (54) and, since only isotope data supported the pathway to leucine at the time of its preparation, to a series of three papers documenting that pathway in Neurospora and Salmonella (4, 23, 29). II. Regulation of Metabolite Flow by End-Product Inhibition

Because the control of the formation of the branched-chain amino acids by end-product inhibition of enzyme action can be considered more simply than the control of their formation by repression of enzyme syn-

BRANCHED-CHAIN AMINO ACIDS

59

thesis, the former will be considered first. Although the two mechanisms can be considered separately, one should not assume a priori that the two are completely independent. As shall be shown, the two can be closely interrelated, and even though the examples of their interrelation­ ship to be cited are trivial, one should not yet assume even for these pathways that end-product repression and end-product inhibition are not more fundamentally intertwined in mechanism. A. Inhibition of Threonine Deaminase

One of the earliest described examples of the almost, but not quite, universal pattern of control of metabolite flow by end-product inhibition is the inhibition of threonine deaminase by isoleucine (51 ). This pattern, inhibition of the first specific enzyme in the pathway by the end product, is also found in the pathways to valine and leucine. Thus far, in all organisms examined the same patterns are found, although in certain details the nature of the interactions appear to vary from one organism to another. The early studies were made on threonine deaminase of Escherichia coli and were characterized by the fact that saturation curves for sub­ strate ([V] vs. [S]) and for inhibitor (percent inhibition vs. [I]) were sigmoid and that nearly straight lines could be obtained in Lineweaver and Burk or Dixon plots of the kinetic data if the plots were made against 1/[substrate] 2 or against [inhibitor] 2 , respectively (51, 52). In other words, there were cooperative effects in the binding of both sub­ strate and inhibitor implying multiple binding sites on the enzyme with interactions between the sites. Furthermore, the inhibition by isoleucine appeared competitive with substrate and thus enhanced the cooperative interactions (sigmoidicity) seen in the saturation of the enzyme by substrate. Changeux (10, 11), investigating further the specific question of the mechanism by which the cooperative interactions and the competitive antagonism between two dissimilar amino acids occurred, made the very important observation that heating the enzyme, or treating it with mer­ curic salts, resulted in a loss of sensitivity to inhibition by the end product and a loss of the cooperativity observed in saturation by sub­ strate. This observation was interpreted to indicate that the binding sites for substrate and inhibitor were distinct, the competitive interaction notwithstanding. Another important observation made by Changeux in these studies was that valine overcame the substrate-substrate interaction (i.e., "nor-

60

H. E. UMBARGER

malized" the kinetics) and antagonized the inhibition by isoleucine. Thus, under certain conditions (particularly at low substrate levels), valine stimulated the activity of threonine deaminase and was considered to be an "activator." Studies with the enzyme from S. typhimurium, which is almost certainly similar to the Escherichia coli enzyme, revealed that valine did not activate either desensitized threonine deaminase or a mutant threonine deaminase which was not inhibited by isoleucine and which exhibited "normal" kinetics (hyperbolic substrate saturation curves) (18). Thus, valine, the activator, appeared to overcome sub­ strate-substrate interactions, and isoleucine, the inhibitor, appeared to enhance them. With the results of Changeux' experiments and the somewhat similar, but more detailed, studies of Gerhart and Pardee (20) on the aspartate transcarbamylase of E. coli as examples, Monod and Jacob (38) pointed out the general importance of the kind of regulatory properties exhibited by these enzymes as a basis not only of end-product inhibition, but also of a way to achieve a physiological coordination of reactions that were not biochemically linked. Specifically, it was pointed out that for valine to stimulate the flow of metabolite along the pathway parallel to that by which it was synthesized when isoleucine was limiting (see Fig. 1) might be as important as the inhibition of that flow when isoleu­ cine was in excess. Although it has not yet been possible to determine whether the interaction of valine with threonine deaminase is physiologi­ cally significant, the concept of activation by physiologically related but chemically unrelated metabolites is now matter of factly accepted, and numerous examples have been recognized both in catabolism and biosynthesis. The second generally important feature of Changeux' results high­ lighted by Monod and Jacob in their essay was that the competitive inhibitor, isoleucine, was not an isosteric one (binding at the same site as did substrate) but was an allosteric one (binding at a different site) and that therefore its effect was transmitted through the tertiary or perhaps through the quaternary structure of the protein. On the basis of the concept of Koshland (32) that proteins are flexible and undergo induced changes in structure upon binding small molecules, this idea seemed quite plausible, although quite vague in terms of mechanism. The concept of an allosteric protein has since become more formalized and restricted with respect to a mechanism that accounts for cooperative ligand binding (which most regulatory proteins do) in terms of a multi­ mene protein (which most regulatory proteins are). The present concept, introduced by Monod in collaboration with Wyman and Changeux (39),

BRANCHED-CHAIN AMINO ACIDS

61

is that allosteric proteins are composed of identical protomers (containing at least one polypeptide) which are symmetrically arranged and that all the subunits can exist in two states which differ in their affinity for the allosteric ligand. Furthermore, the change of state or allosteric transition is a concerted one in which hybrid enzymes do not exist as stable intermediates. The allosteric ligand therefore is considered as a trapping agent which preferentially binds one form of the enzyme. Inter­ estingly, the mathematical equations describing the ligand binding pre­ dicted by this model can account for a surprising number of real systems (including, as will be discussed below, the steady-state kinetic data of threonine deaminase). It currently seems unlikely that the allosteric model in its pure form accounts for all the properties of regulatory proteins, but the concept has probably had a greater impact upon the field of enzymology than any single idea since Summer's crystallization of urease demonstrated the protein nature of an enzyme. Many of the earlier observations on threonine deaminase can indeed be interpreted in terms of the allosteric model of Monod et al. {89). Thus, the enzyme from E. coli and S. typhimurium appeared to be nor­ mally in the inactive form and could be stabilized in that form by isoleucine (a negative effector). Activation occurred by adding valine or substrate (positive effectors). It must be recalled in considering the early results that, owing to difficulties in the purification of the enzyme, it was always examined in crude extracts or in partially purified preparations. Such extracts might have had some variable amount of valine or isoleucine or both retained by the cells during the harvesting of the cells and the prepara­ tion of the extract, or even liberated from extract protein by protease activity after dialysis. Furthermore, small amounts of isoleucine were often added which were subinhibitory at high (saturating) levels of substrate but which at low levels of substrate (below S0.5) could have been inhibitory. It was therefore of interest that in this author's laboratory, different preparations of E. coli and S. typhimurium extracts yielded substrate saturation curves with varying degrees of sigmoidicity. The first report which was based upon results with a fairly pure preparation of threonine deaminase from S. typhimurium was that of Maeba and Sanwal {35). These workers did, in fact, observe a noncooperative binding of threonine in the absence of isoleucine, but the usual cooperative binding when isoleucine was present. In retrospect, it may be that these experiments were conducted with enzyme that was free of isoleucine. However, Burns and Zarlengo {5) have shown that the effect of very low levels of iso-

62

H. E. UMBARGER

leucine (sufficient to render the substrate saturation curves sigmoid) can be overcome in buffer of the ionic strength of that used by Maeba and Sanwal (85). It is now clear, however, that the purified threonine deaminase of S. typhimurium exhibits cooperative binding of substrate only in the presence of isoleucine. Thus, it now appears that, in terms of the allosteric model of Monod et al. (39), the equilibrium between the active state and inactive state of threonine deaminase strongly favors the active form so that normal, Michaelis-Menten kinetics are observed except in the presence of the negative effector (isoleucine). There have been several examples of threonine deaminase which, in crude systems, appeared to be different from those of S. typhimurium and E. coli in that the enzyme in crude systems exhibited "normal" kinetics except when isoleucine was added to the assay system. An exam­ ple is the activity found in extracts of Rhodospirillum rubrum W). In this laboratory the same was found to be true of the activity in Micrococcus denitrificans and Bacillus subtilis. The enzyme from the latter was purified and studied in some detail on the assumption that it was different from the S. typhimurium enzyme then being purified at Duke. It now appears, however, that the differences between the two enzymes are rather trivial and, although different types of experiments have been done with each enzyme, many of the findings are interprétable by a common model. Considering first the findings already published on the S. typhimurium enzyme, the enzyme appears to be composed of four identical subunits, with a molecular weight of about 200,000 (60). In view of the fact that it was possible to demonstrate only two pyridoxal phosphate sites on the enzyme, it would appear that the enzyme is actually a dimer of subunits which contain two identical peptides arranged so that only one substrate site on each subunit is functional. Compatible with this view is the observation that the value of n in an empirical "Hill plot" (12) never exceeds 2 for either substrate or inhibitor. The B. subtilis enzyme is very nearly the same size and has been observed in the electron microscope to be composed of two, nearly spheri­ cal, subunits (27). Half-molecules, but not quarter molecules, have been obtained. It has not yet been possible to demonstrate a reassociation of the half-molecules into active enzyme. Kinetic studies have been more extensive with the enzyme from B. subtilis. However, enough comparative studies have been done with the enzyme from S. typhimurium to suggest that the differences will be found to be minor. Using the purified B. subtilis enzyme, Hatfield (27) found that the steady-state kinetic data (i.e., steady-state velocities

BRANCHED-CHAIN AMINO ACIDS

63

in the presence of varying amounts of substrate and inhibitor) could be readily interpreted, with the proper choice of constants, by the equa­ tions describing the allosteric model of Monod et al. (89). To apply these equations, which involve binding data, to kinetic data, it was neces­ sary to employ the convention introduced by Frieden (19), which as­ sumed that isoleucine was purely a "K-system" inhibitor (i.e., it affected substrate binding but not 7 m a x ), an assumption which seemed justified at least in the ranges of substrate and inhibitor employed. Essentially, the conversion consists of equating the binding function, YF, of the origi­ nal equation with the observed v/VmSLX values. Presumably, too, con­ stants could be chosen so that other models could be used to account for the data. For example, the linear model of Koshland et al. (83), which can readily be applied to a two-subunit protein and which involves a sequential, rather than a concerted, change in conformation of the subunits, would probably also account for the kinetic data obtained with the B. sub tilis enzyme. It is very important to emphasize that it was the steady-state kinetic data that could be explained by the allosteric model of Monod et al. (39). This limitation was necessary because it was observed that the transition from the inhibited form of the enzyme to the active form was a slow process requiring time periods that were readily demonstrable on an ordinary recording spectrophotometer when a coupled assay (in which the product was reduced by lactic dehydrogenase) was employed. While the transition could be faster (less than 10 seconds) than could be conveniently observed with such an instrument, it could require 15 or 20 minutes to reach a steady-state velocity. The transition was examined in detail under the rather special condi­ tions in which all the enzyme was initially in the inhibited state, i.e., enzyme was preincubated with isoleucine at a concentration of about 30 times the Ki. A 5-minute preincubation period was sufficient to con­ vert the enzyme, which was in the active state in its isolated form, to the inactive state. The rate of increase in activity was then determined after adding an amount of substrate that would reverse completely the inhibition. It was observed that the rate constant describing the conver­ sion of inactive enzyme to active enzyme was directly proportional to substrate concentration (27). Although experiments of this kind were not useful in deciding between a concerted or a sequential transition of the individual subunits, they were useful in deciding between two other alternatives distinguishing the models of Monod et al. (39) and of Koshland et al. (33), namely, whether the transition was induced or spontaneous. Specifically, two reaction sequences, which for simplicity

64

H. E. UMBARGER

concerned only one of the two presumed subunits, were considered: ki

ki

ht

koht

FI — F ^ E ^ ES k-i k\

k-i ki

kt

FI — FIS — ESI ^± ES k-\

k-2

>P

(1)

k-i fce»t

>P

(2)

k-i

It was further assumed that during the transition period the interme­ diates in either sequence would be close to zero concentrations. The integrated differential equation describing the appearance of active en­ zyme in sequence (1) revealed that the proportionality of rate of activa­ tion and substrate concentration was compatible with the concept of a spontaneous reaction. However, the terms in the equation which con­ tained 1/S also contained I. Thus, the rate constant describing the acti­ vation should also have been related (inversely) to inhibitor concentra­ tion. When this prediction was tested, the rate constant was shown to be independent of inhibitor concentrations over the ranges of substrate and inhibitor concentration chosen for study. Variants of sequence (1), such as introducing an activator site for substrate or assuming the F ^± E transition to be immeasurably fast, did not alter the general form of the equation. In contrast, the equation describing the appearance of active enzyme via reaction sequence (2) was compatible with the dependence of the rate of activation upon substrate concentration and its independence of inhibitor concentration. Thus, the experiments favored the concept of an induced transition. An apparently noncatalyzed activation could also be observed upon removal of inhibitor by gel filtration. However, even at room temperature, the transition was extremely slow, requiring about 75 minutes under the conditions employed. The nature of the transition is unknown. It does not appear to involve an association or dissociation reaction, however. Finally, it might be mentioned that threonine deamination can also serve a degradati ve role in many microorganisms. Obviously, an enzyme with regulatory properties as rigidly related to isoleucine biosynthesis as the one described here could not catalyze a catabolic deamination. An enzyme in E. coli that is catabolic is described elsewhere in this volume (59). It, too, is a regulatory protein with properties that make it uniquely suited to its presumed catabolic role. B. Inhibition of Acetohydroxy Acid Synthetase

The enzyme catalyzing the first step in the biosynthesis of valine from the key intermediate, pyruvate, and the one that is inhibited by

BRANCHED-CHAIN AMINO ACIDS

65

valine is acetohydroxy acid synthetase (53). The effectiveness of this inhibitory effect in most strains of bacteria would be expected to be very low, since the enzyme catalyzes not only the synthesis of acetolactate, the valine precursor, but also that of acetohydroxybutyrate, the isoleucine precursor. Thus, quenching of the valine pathway by exogenous valine would be expected to be accompanied by a quenching of the isoleucine pathway as well. As will be discussed below, in most strains there is probably a compensatory process (multivalent repression) that would prevent the quenching of isoleucine biosynthesis but that com­ pensatory response would lead to increased valine production as well. Furthermore, there is no evidence that the failure of valine to prevent isoleucine formation in these strains is related to that compensatory process. Paradoxically, although in vitro experiments pinpoint a control point for regulation of valine biosynthesis, there is little evidence ob­ tained with growing cells that the sensitivity of the enzyme to valine is physiologically significant. Indeed, one of the few experiments that bears directly on the question was one performed by the Biophysics Group at the Carnegie Institution of Washington's Department of Ter­ restrial Magnetism during their isotope competition studies on strain B of E. coli (44). These workers found that the amount of carbon flow over the pathway to valine was virtually the same whether exogenous valine was supplied or not. Thus, isotope competition in this case was probably due to a "swamping" of the endogenously formed pool of valine with exogenous valine. In contrast, flow over the isoleucine pathway was quenched very efficiently by exogenous isoleucine. Nevertheless, the acetohydroxy acid synthetase of E. coli strain B is inhibited by valine in vitro. The exception to the ineffectiveness of valine in interfering with iso­ leucine biosynthesis is found in the K12 strain of E. coli, a strain in which valine inhibits growth unless isoleucine or one of its six-carbon precursors is added to the medium (2, 34). Although the sensitivity of the enzyme to inhibition by valine is not the only factor involved in the inhibition of growth (see below), one mechanism of valine resis­ tance in mutants of the K12 strain is the formation of an enzyme less sensitive to valine than that in the wild type (34, 4%) · Before acetolactate was proposed by Strassman et al. (49) as an intermediate in the biosynthesis of valine, it was clearly implicated as an intermediate in the formation of acetoin, a catabolic product of glu­ cose dissimilation by Aerobacter aerogenes (30). The enzymes responsible for the two roles for acetolactate formation, however, are different (26). The valine-inhibited enzyme with a pH optimum of about 8.0 is repressed

66

H. E. UMBARGER

in rich medium, whereas the other is formed in rich medium containing sugars, but only after the pH drops to near its optimal pH (6.0). This feature is compatible with its role in acetoin formation by A. aerogenes when the fermentation mixture becomes acidic. Halpern and Even-Shoshan (25) have described mutants of A. aerogenes that lack the pH 6.0 enzyme as well as mutants that lack the pH 8.0 enzyme. Only the absence of the pH 8.0 enzyme could be cor­ related with auxotrophy for isoleucine and valine. Studies by Störmer (46) have clearly demonstrated the fact that the two physiologically distinct enzymes in A. aerogenes are also biochemi­ cally different. The pH 6.0 enzyme was crystallized and shown not to be a flavoprotein (47). In contrast, the valine-sensitive enzyme exhibits a requirement for FAD and even in crude extracts, it is almost com­ pletely resolved for this compound (48). In passing, it might be stated that, thus far, in the author's laboratory attempts to purify the valinesensitive enzyme from E. coli or S. typhimurium have been unsuccessful. Stabilizing the enzyme has been a problem, the enzyme appearing to be most stable under conditions that are optimal for catalytic activity (M. Kisumi, personal communication). For all practical purposes, E. coli and S. typhimurium do not have the pH 6.0 enzyme found in A. aerogenes. However, in extracts in which the valine-sensitive acetohydroxy acid synthetase is strongly repressed, another activity with an optimum at about pH 6.0 is readily apparent. Evidence that this activity can function in valine biosynthesis in E. coli was provided by a series of experiments by Ramakrishnan and Adelberg (4$), who obtained the only acetohydroxy acid synthetasenegative mutants yet reported in E. coli, an observation that allowed them to locate the position of the structural gene and an adjacent ele­ ment, which was presumably the operator region. The procedure these workers employed to obtain acetohydroxy acid synthetase-negative mutants was first to select mutants that were re­ sistant to inhibition by a-aminobutyrate (43). Some of the resistant mutants had a derepressed level of acetohydroxy acid synthetase, a derepression attributed to an 0C mutation of an "operator" region specific for the acetohydroxy acid synthetase structural gene. Selection of α-aminobutyrate-sensitive mutants from one of these strains led to the isolation of mutants with lesions in the acetohydroxy acid synthetase structural gene. Although acetohydroxy acid synthetase belongs to the generic group of proteins that can clearly be termed "regulatory," the enzymes from E. coli and S. typhimurium do not exhibit cooperative binding of either

BRANCHED-CHAIN AMINO ACIDS

67

substrate or inhibitor. Thus, neither enzyme falls into the generic group of allosteric proteins as rigidly defined by Monod et al. (39). The enzyme from E. coli exhibits a competitive antagonism between valine and pyruvate (53). On the other hand, valine inhibits the enzyme from S. typhimurium noncompetitively with respect to substrate (34). C. Inhibition of α-lsopropylmalate Synthetase

The third end-product-sensitive enzyme in the pathways to the branched-chain amino acids is the leucine-inhibited enzyme a-isopropylmalate synthetase. This enzyme was highly purified from Neurospora by Webster and Gross (57), who demonstrated that the saturation of the enzyme by the two substrates, acetyl coenzyme A and a-ketoisovalerate was noncooperative, i.e., normal Michaelis-Menten saturation curves were seen. At pH 7.5, it was observed that leucine inhibited the enzyme competitively with respect to acetyl coenzyme A. At lower pH, the antagonism was of the "mixed" type (i.e., both Km and 7 max were affected) with respect to acetyl coenzyme A, as it was with respect to a-ketoisovalerate at all pH values examined. The interaction of inhibi­ tor with enzyme, however, was cooperative, with n values in "Hill plots" of about 1.5. Physical studies on the enzyme pointed to a structure with a molecu­ lar weight of 143,000 which upon treatment with guanidine yielded subunits of about 48,000 (58). Analysis of tryptic peptides pointed to a trimeric structure of the enzyme. Gross and Webster (24) described some mutants that were resistant to 5',5',5'-trifluoroleucine and which contained α-isopropylmalate synthetases that were resistant to inhibition by leucine. These mutants were of interest because, as will be discussed below, the subsequent two enzymes in the pathway, α-isopropylmalate isomerase and ß-isopropylmalate dehydrogenase, were derepressed in these mutants. In yeast, similar mu­ tants containing α-isopropylmalate synthetases resistant to inhibition by leucine do not have derepressed levels of the subsequent two enzymes (45). The yeast enzyme, however, has been studied only in crude ex­ tracts. In each organism, such mutants overproduce and excrete leucine. The enzyme from S. typhimurium has recently been purified by Kohlhaw et al. (31). The saturation curves for both substrates reveal normal (noncooperative) kinetics. However, in the presence of leucine, the apparent binding of acetyl coenzyme A, which was competitive with leucine, became increasingly cooperative. As was found with the Neu­ rospora enzyme, the enzyme was more strongly inhibited by leucine at pH 6.5 than at pH 8.5, the optimal pH for the Salmonella enzyme.

68

H. E. UMBARGER

The inhibition by leucine is not complete, particularly at higher acetyl coenzyme A concentrations. The binding of leucine is cooperative, with n values in "Hill plots" of close to 2. The binding of leucine to the enzyme leads to a reversible retardation of the enzyme on a Sephadex G-100 column that is indicative of a reduction in molecular size, presumably to "half molecules." Disc gel electrophoresis of the purified enzyme yielded two bands of active en­ zyme, but in the presence of leucine only one (the faster migrating) band was found. The nature of the leucine effect is not understood but is undoubtedly related to binding at the inhibitor site since a leucineresistant mutant enzyme does not exhibit the leucine-dependent shifts. The a-isopropylmalate synthetase of yeast is also inhibited by leu­ cine (45). Trifluoroleucine-resistant mutants with leucine-insensitive enzymes have been isolated. The enzyme has not been purified, however. III. Control of Enzyme Level in the Pathways to the Branched-Chain Amino Acids

The second mode of control of biosynthetic function, regulation of enzyme amount, can now be rather clearly defined on the physiological level for the biosynthesis of the branched-chain amino acids in E. coli and S. typhimurium. For other forms of bacteria, and for N. crassa and yeast, less is known although some comparisons are possible. It would appear that, in bacteria, the differences are rather trivial whereas there are probably basic differences between the physiological patterns of control of enzyme amount in prokaryotic and in eukaryotic cells. In neither system, however, can much be stated definitively regarding the molecular mechanism of regulation of the levels of these enzymes (i.e., at the level of gene expression). Because of its greater simplicity from a comparative point of view, the leucine pathway will be considered first. A. Repression of the Leucine Biosynthetic Enzymes

The regulation of the level of leucine biosynthetic enzymes conforms to the "classical" pattern that would be predicted on the basis of the Jacob-Monod model (28) with a few modifications here and there. How­ ever, alternative models would be equally apt, and it should be empha­ sized that the facts presently known do not warrant the assumption that regulation does indeed occur by a simple, repressor-operator model. There is, however, in Salmonella a functional unit of four genes that define the structure of the three specific enzymes required for leucine biosynthesis (86). That the cluster does function as a unit is indicated

BRANCHED-CHAIN AMINO ACIDS

69

by the amounts of enzyme found under conditions of repression and derepression. The expression of the cluster does seem to be coordinate, although, owing to a differential stability of the three enzymes, coordina­ tion is somewhat difficult to demonstrate (3). Analysis of the regulation of this (leu) gene cluster has been possible because of the isolation of several different kinds of mutants in which regulation of the cluster was altered. One of these is a mutant in which there is repression of function of the entire cluster by some as yet unidentified component of normal cytoplasm. The mutant, strain leu 500, is a leucine auxotroph only in cells containing an intact supX gene (40). (SupX was so named because mutations affecting it suppress the "0X" or "xenesthetic" muta­ tion in strain leu 500). Many of the mutations which suppress the leu 500 mutation are deletion mutations affecting the adjacent cysB region and the tryptophan operon. [The latter are of considerable interest since they provide a mechanism for the direct selection of deletion mutations which extend into or even through the tryptophan operon from the opera­ tor end (87)]. A set of mutants affecting expression of the leu operon were obtained by Calvo (8), who isolated a large collection of trifluoroleucine-resistant mutants of S. typhimurium. One group of these was shown to have elevated levels of the leucine biosynthetic enzymes and, as a result, was found to overproduce and excrete leucine. The lesion in these mu­ tants was found to be linked to the leu operon and was shown to lie in what must be the operator region of the leu operon (7). The lesions that have been studied in detail appear to lie between the position of the leu 500 site and what has been recognized as the most operator-proxi­ mal mutational site in the leuA cistron. A second group of trifluoroleucine-resistant mutants studied by Calvo and his co-workers (6) comprised those in which the lesions were un­ linked to the leu operon. These were characterized phenotypically by being excretors not only of leucine, but of isoleucine and valine as well. These mutants exhibit derepressed levels of both the leucine biosynthetic enzymes and the isoleucine and valine biosynthetic enzymes. While these mutants might be compared superficially to mutants in which the repressor of the Jacob-Monod model (28) has been modified, analysis of their function does not support so simple an interpretation. Actually, two classes of such mutants have been identified. One, which exhibits some repressibility when the level of exogenous leucine is very high, has a leucyl-tRNA synthetase with a reduced affinity for leucine (Calvo, per­ sonal communication). The second class exhibits very little repressibility even with high exogenous leucine levels. The biochemical lesion in this class has not yet been elucidated.

70

H. E. UMBARGER

The present picture of repression of the leu operon is thus unclear. For leucine to trigger repression, its activation does appear necessary. Whether leucine tRNA is itself involved in repression or is necessary for one step leading to the formation of the repressor cannot be stated with certainty. Until now, no mutants have been described which appear to lack the postulated repressor of the Jacob-Monod model. Thus, it is possible neither to support nor to supplant the operator-repressor model for the regulation of the leu operon in Salmonella. The relative effectiveness of end-product inhibition and repression in leucine biosynthesis in Salmonella was demonstrated in some novel experiments by Calvo and Calvo (9). Trifluoroleucine-resistant mutants which contain normal isopropylmalate synthetase, but have derepressed levels of the enzyme, excrete leucine. The same is true for the mutant that contains a leucine-insensitive enzyme but has lower enzyme levels because of repression by endogenously formed leucine. For both the growth rate is normal. Thus, repressibility of the leucine operon is suffi­ cient to prevent overproduction of leucine only if the first enzyme is end-product sensitive. When both lesions were combined in the same organism, excretion of leucine was extremely heavy, and the growth rate was actually reduced unless valine was added. In Neurospora, the studies of Webster and Gross (57) on factors regulating the level of enzymes in the pathway to leucine have revealed a physiological pattern that is indeed different from that found in Sal­ monella. As mentioned earlier, the first enzyme in the pathway is in­ hibited by leucine (57). It is also repressed by leucine. On the basis of his analysis using mutants lacking the first enzyme and mutants in which end-product sensitivity of the first enzyme was lost, Gross (22) has postulated that α-isopropylmalate, the product of the first en­ zyme, is the inducer of the second and the third enzymes in the pathway. The physiological pattern in yeast is still different in that incorpora­ tion of excess leucine into the medium represses the second and third enzymes, but actually increases the amount of the first enzyme (45). The first enzyme, however, is repressed when threonine as well as leucine are present in excess. At the present time, no satisfactory molecular models for the regulation of the leucine biosynthetic enzymes in either yeast or Neurospora have been postulated. B. Repression of the Isoleucine and Valine Biosynthetic Enzymes

The trifluoroleucine-resistant mutants that excrete leucine as well as valine and isoleucine are of special interest when the regulation of the isoleucine-valine-forming enzymes is under consideration, since the

BRANCHED-CHAIN AMINO ACIDS

71

repression of these enzymes in Salmonella is multivalent; i.e., for repres­ sion to occur, leucine, valine, and isoleucine must be present in excess (16). Thus, the trifluoroleucine-resistant mutants in which the enzymes of both pathways are derepressed appear to be altered in a step that is essential for leucine to exert repression on the leucine operon and is essential also for leucine to participate in multivalent repression of the isoleucine-valine (ilv) gene cluster. As implied above in the discussion of an operator-like region specific for the structural gene (ilv B) for acetohydroxy acid synthetase, the ilv gene cluster does not constitute a single operon. A second distinct operator in E. coli was identified by Ramakrishnan and Adelberg (43), which upon mutation to the 0C (operator constitutive) state resulted in the derepressed formation of the ilv A, D and E gene products (en­ zymes 1, 4 and 5 in Fig. 1). Mutants containing a derepressed ilv C gene (which lies between the ilv B and ilv A genes) have not yet been described, so there is no direct evidence for a third operator region. However, because the isomeroreductase is formed at repressed or dere­ pressed levels in a pattern that is independent of (i.e., not coordinate with) the other four isoleucine- and valine-forming enzymes, there must be some controlling element that provides for the recognition of re­ pressing levels of the branched-chain amino acids or their active derivatives. Because the mutations in Salmonella leading to a leucyl tRNA syn­ thetase with reduced affinity for leucine resulted in faulty recognition of leucine in the multivalent repression of the isoleucine- and valineforming enzymes, it might be expected that the activating enzymes cor­ responding to valine and isoleucine might also be involved in the forma­ tion of the hypothetical "multivalent repressor." Indeed, the involvement of valyl tRNA synthetase in multivalent repression had been shown even earlier in the demonstration by Eidlic and Neidhardt (14) that under conditions of a limited capacity to charge tRNA val with valine, there was a derepression of the ilv genes. Further augmenting the idea that multivalent repression might involve activated derivatives of the three branched-chain amino acids was the observation of Szentirmai et al. (50) on a thiaisoleucine-resistant mutant of E. coli. Extracts of this mutant exhibited low levels of isoleucyl tRNA synthetase activity which had a reduced affinity for isoleucine and an even more reduced affinity for the analog, thiaisoleucine. Thus, the altered synthetase re­ sulted in an increased discrimination against the analog. However, it is not clear whether this property or the concomitant derepression of the ilv A-D-E operon was the basis for resistance of the mutant.

72

H. E. UMBARGER

It should be pointed out that one feature attributed to an operon, viz., the coordinate expression of component genes, was not observed in the experiments with E. coli (13). Whether this failure was due to a technical difficulty or was an indication of the existence of internal sites of initiation (of either transcription or translation) in the ilv A-D-E operon is not clear. That only three of the five enzymes were found to be derepressed in the mutants was in keeping with the fact that in E. coli strain K12, enzymes 2 and 3 in the sequence (see Fig. 1) are not derepressed when isoleucine is limiting (13). This difference from the behavior of the same enzymes in S. typhimurium will be discussed below. Whereas the occurrence of the his S, U, and W mutants in S. typhi­ murium points to direct involvement of histidyl tRNA in the regulation of the his operon, analogous evidence is not strong for the ilv operon in either E. coli or S. typhimurium. An interesting correlation between the effects of two analogs of valine differing in their reaction with valyl tRNA synthetase and their effects upon the level of the isoleucine-valine-forming enzymes was reported by Freundlich (15). One analog, α-aminobutyrate, competed with valine for activation but was not itself transferred to tRNA val . This analog led to a derepression of the ilv operon. On the other hand, α-amino-ß-chlorobutyrate, which was itself transferred to tRNA val , led to repression of the isoleucine-valine-forming enzymes. These two observations are indeed compatible with the idea that valyl tRNA formation is a necessary reaction in the formation of the "multivalent repressor." However, the observations do not allow such an unequivocal interpretation. Thus, as shown by Ramakrishnan and Adelberg (43) and by Freundlich and Clarke (17), a-aminobutyrate mimics the effect of valine in inhibiting acetohydroxy acid synthetase in vitro. If aminobutyrate were an effective inhibitor of this enzyme in vivo, there might be sufficient quenching of valine biosynthesis to lead to a derepression of the ilv genes because of limiting valine. Again, the question of the significance of reduced levels of the isoleucine-valine biosynthetic enzymes in the presence of aminochlorobutyrate is difficult to interpret. The difficulty lies in deciding whether the reduced activity of the isoleucine-valine biosynthetic enzymes is due to repression or is a result of formation of inactive protein. Freundlich and his co-work­ ers, however, have taken some pains to demonstrate the continued forma­ tion of other enzymes when the isoleucine-valine enzymes were appar­ ently repressed by the substitution of aminochlorobutyrate for valine. It is not clear in any of these experiments whether aminochlorobutyrate has "replaced" valine in the formation of the "multivalent repressor."

BRANCHED-CHAIN AMINO ACIDS

73

On the other hand, until some evidence is obtained for or against the existence of such an entity, discussions of the role played by aminochlorobutyrate may be premature. Like the pathway to leucine, the question of mechanism of regulation of the pathway to isoleucine and valine is an open one. In each case, it seems likely that the structural genes are adjacent to receptor elements comparable to the operator of the Jacob-Monod (28) model that would provide for recognition of the repressor. The idea that the repressore are not the three branched-chain amino acids themselves is evident from the need for activation (13, 14). A cheap hypothesis would be that the repressor of the Jacob-Monod model is "activated" not by the amino acids, but by some derivative of them, such as amino acyl tRNA's. Until some genetic or biochemical evidence is obtained for the existence of an intermediary element comparable to the repressor, the cheap hy­ pothesis is also a worthless one since it does not lead to any experimental tests. It is clear from the discussion that the writer is prejudiced in favor of some sort of negative control mechanism. It should be pointed out that the question of a "positive" control system (i.e., a mechanism that promotes a "turn on" of the transcription of the ilv gene cluster) is not completely ruled out. IV. The Inhibition of Growth of Escherichia coli Strain K12 by Valine In this final section it might be appropriate to discuss and reiterate some of the observations pertaining to the phenomenon that originally stimulated interest in the biosynthesis of the branched-chain amino acids. That phenomenon was, of course, the inhibitory effect of valine on the growth of E. coli strain K12 and the reversal of that inhibition by isoleucine (2). Until it was realized that the repression of the isoleucine and valine enzymes was multivalent, it appeared that the greater valine sensitivity of the acetohydroxy acid synthetase of the K12 strain was a sufficient explanation of the quenching of isoleucine biosynthesis and the resulting inhibition of growth (34). Compatible with this view is the fact that the six-carbon precursors of isoleucine reverse valine inhibition whereas the four-carbon precursors do not. However, if the enzymes of the iso­ leucine and valine biosynthetic pathway were multivalently repressed, the quenching of isoleucine biosynthesis should lead to a derepression of the very enzyme that is inhibited by valine (16). Indeed, derivatives of the K12 strain in which this enzyme is genetically derepressed are valine-resistant (43). However, in this respect, the K12 strain of E. coli is unique in that only three of the five enzymes needed for isoleucine and valine biosynthesis are derepressed on limiting isoleucine (13, 55).

74

H. E. UMBARGER

Thus, the addition of valine to a minimal medium not only inhibits the second step in the sequence of reactions leading to valine, but also causes a repression of the sensitive enzyme and the one following it in the sequence. The molecular basis of the repression signal is, of course, unknown. On the metabolic level, there yet remains one paradoxical observation to be explained. The apparent mutation of the operator locus controlling the ilv ADE cluster to the operator-constitutive state leads to a marked (about 25-fold) derepression of the three corresponding enzymes and to a very high level of resistance to valine (4$). Furthermore, the valine resistance accompanied the postulated 0+ -> 0C mutation only if the threonine deaminase structural gene was intact. From this fact it might be inferred that a high rate of threonine deamination results in a reversal of the inhibitory effect of valine. However, the addition of a-ketobutyrate or a-aminobutyrate (which can enter the cell sufficiently well to support the growth of threonine deaminase-deficient mutants) to the medium does not reverse the inhibition. (α-Aminobutyrate in higher concentra­ tions is itself inhibitory, presumably by mimicking the effect of valine itself.) In addition, cells in which the same three enzymes are derepressed by a mechanism not involving the operator locus are still valine sensitive (SO). Although the level of derepression of threonine deaminase in these cells is only 6-fold, there is not even a low level of resistance to valine. A possible, though rather unlikely, metabolic mechanism by which a high rate of threonine deamination might allow resistance to valine would be the "removal" of the entering valine via a transamination reaction with the a-ketobutyrate produced by the derepressed enzyme. Examination of the internal amino acid pool should readily reveal whether this possibility is tenable. Perhaps the basis of the resistance that ac­ companies the 0+ -» 0C mutation lies in the expression of the gene cluster itself. There may be, for example, an isoleucine-valine-synthetizing par­ ticle ("ilvasome") such as postulated for Neurospora by Wagner et al. (56). It may be that the 0C mutation allows the formation of an ilvasome in which isoleucine and valine biosynthesis is more efficiently integrated with the needs of the cell and which functions without interruption by exogenous valine. If such were the case, it may then be that the answer to this 23-year-old question must await the technical breakthrough that will permit the study of gene expression in subcellular systems. NOTE ADDED IN PROOF

In view of the earlier failures to obtain evidence of an operator region that specifically controlled the structural gene for acetohydroxy

75

BRANCHED-CHAIN AMINO ACIDS

acid isomeroreductase (ilv C), the recent observations of Dr. Stuart Arfin and Mr. Barry Ratzkin in this laboratory are of interest. They have found that the isomeroreductase is induced by either of the two sub­ strates even in the presence of repressing levels of isoleucine, valine, and leucine. It would therefore appear that earlier observations on the repres­ sion and derepression of this enzyme by the three branched-chain amino acids may have been a secondary consequence of the effect of these amino acids on repression and inhibition of acetohydroxy acid synthetase. REFERENCES

1. Bauerle, R. H., Freundlich, M., Stürmer, F. C , and Umbarger, H. E., Biochim. Biophys. Ada 92, 142 (1964). 2. Bonner, D. M., J. Biol Chem. 116, 545 (1946). 8. Burns, R. 0., Calvo, J., Margolin, P., and Umbarger, H. E., J. Bacteriol. 91, 1570 (1966). 4. Burns, R. 0., Umbarger, H. E., and Gross, S. R., Biochemistry 2, 1053 (1963). 5. Burns, R. 0., and Zarlengo, M. H., J. Biol Chem. 243, 178 (1968). 6. Calvo, J. M., Freundlich, M., and Umbarger, H. E., J. Bacteriol 97, 1272 (1969). 7. Calvo, J. M., Margolin, P., and Umbarger, H. E., Genetics 61, 777 (1969). 8. Calvo, J. M., and Umbarger, H. E., Federation Proc. 23, 377 (1964). 9. Calvo, R. A., and Calvo, J. M., Science 156, 1107 (1967). 10. Changeux, J.-P, J. Mol Biol 4, 220 (1962). 11. Changeux, J.-P., Cold Spring Harbor. Symp. Quant. Biol 26, 313 (1961). 12. Changeux, J.-P., Cold Spring Harbor Symp. Quant. Biol. 28, 497 (1963). 13. Dwyer, S. B., and Umbarger, H. E., J. Bactenol 95, 1680 (1968). H. Eidlic, L., and Neidhardt, F. C , Proc. Nati Acad. Sci. U.S. 53, 539 (1965). 15. Freundlich, M., Science 157, 823 (1967). 16. Freundlich, M., Burns, R. O., and Umbarger, H. E., Proc. Nati Acad. Sci. U.S. 48, 1804 (1962). 17. Freundlich, M., and Clarke, L. P., Biochim. Biophys. Ada 17Ç, 271 (1968). 18. Freundlich, M., and Umbarger, H. E., Cold Spring Harbor Symp. Quant. Biol 28, 505 (1963). 19. Frieden, C , / . Biol. Chem. 242, 4045 (1967). 20. Gerhart, J. C , and Pardee, A. B., J. Biol. Chem. 237, 891 (1962). 21. Gladstone, G. P., Brit. J. Exptl. Pathol 20, 189 (1939). 22. Gross, S. R., Proc. Nati Acad. Sci. Uß. 54, 1538 (1965). 23. Gross, S. R., Burns, R. O., and Umbarger, H. E., Biochemistry 2, 1046 (1963). 24. Groas, S. R., and Webster, R. E., Cold Spring Harbor Symp. Quant. Biol. 28, 543 (1963). 25. Halpern, Y. S., and Even-Shoshan, A., Biochim. Biophys. Ada 139, 502 (1967). 26. Halpern, Y. S., and Umbarger, H. E., J. Biol. Chem. 234, 3067 (1959). 27. Hatfield, G. W., Ph.D. thesis. Purdue Univ., Lafayette, Indiana, 1968. 28. Jacob, F., and Monod, J., Cold Spring Harbor Symp. Quant. Biol 26, 193 (1961). 29. Jungwirth, C , Gross, S. R., Margolin, P., and Umbarger, H. E., Biochemistry 2, 1 (1963). 30. Juni, E., J. Biol Chem. 195, 715 (1952).

76 31. 32. 33. 34. 35. 36. 37.

H. E. UMBARGER

Kohlhaw, G. B., Leary, R , and Umbarger, H. E., J. Biol. Chem. 244, 2218 (1969). Koshland, D. E., Jr., J. Cellular Comp. Physiol. 54, Suppl. 1, 245 (1959). Koshland, D. E., Jr., Nemethy, G., and Filmer, D., Biochemistry 5, 365 (1966). Leavitt, R. I., and Umbarger, H. E., J. Bacteriol. 83, 624 (1962). Maeba, P., and Sanwal, B. D., Biochemistry 5, 525 (1966). Margolin, P., Genetics 48, 441 (1963). Margolin, P., and Bauerle, R. H., Cold Spring Harbor Symp. Quant. Biol. 31, 311 (1966). 38. Monod, J., and Jacob, F., Cold Spnng Harbor Symp. Quant. Biol. 26, 389 (1961). 39. Monod, J., Wyman, J., and Changeux, J.-P., J. Mol. Biol. 12, 88 (1965). 40. Mukai, F. H., and Margolin, P., Proc. Nati. Acad. Sci. U.S. 50, 140 (1963). 41. Ning, C, and Gest, H., Proc. Nati. Acad. Sci. U.S. 56, 1823 (1966). 42. Pittard, J., Loutit, J. S., and Adelberg, E. A., J. Bacteriol. 85, 1394 (1963). 43. Ramakrishnan, T., and Adelberg, E. A., J. Bacteriol. 89, 654 (1965). 44- Roberts, R. B., Abelson, P. H., Cowie, D. B., Bolton, E. T., and Britten, R. J., Carnegie Inst. Wash. Pubi. 607 (1955). 45. Satyanarayana, T., Umbarger, H. E., and Lindegren, G. J. Bacteriol. 96, 2018 (1968). 46. Stornier, F. C , / . Biol. Chem. 242, 3735 (1968). 47. Stornier, F. C, J. Biol. Chem. 242, 3740 (1968). 48. St0rmer, F. C , and Umbarger, H. E., Biochem. Biophys. Res. Commun. 17, 587 (1964). 49. Strassman, M., Thomas, A. J., and Weinhouse, S., J. Am. Chem. Soc. 75, 5135 (1953). 60. Szentirmai, A., Szentirmai, M., and Umbarger, H. E., J. Bacteriol. 95, 1672 (1968). 51. Umbarger, H. E., Science 123, 848 (1956). 52. Umbarger, H. E., and Brown, B., J. Bacteriol. 73, 105 (1957). 53. Umbarger, H. E., and Brown, B., J. Biol. Chem. 233, 1156 (1958). 64. Umbarger, H. E., and Davis, B. D., in "The Bacteria" (I. C. Gunsalus and R. Y. Stanier, eds.), Vol. Ill, pp. 167-251. Academic Press, New York, 1962. 55. Umbarger, H. E., and Freundlich, M., Biochem. Biophys. Res. Commun. 18, 889 (1965). 56. Wagner, R. P., Berquist, A., Barbee, T., and Kiritani, K , Genetics 49, 865 (1964). 57. Webster, R. E., and Gross, S. R., Biochemistry 4, 2309 (1965). 58. Webster, R. E., Nelson, C. A., and Gross, S. R., Biochemistry 4, 2319 (1965). 59. Wood, W. A., Current Topics Cellular Regulation, this volume. 60. Zarlengo, M. H., Robinson, G. W., and Burns, R. O., J. Biol Chem. 243, 186 (1968).

On the Roles of Synthesis and Degradation in Regulation of Enzyme Levels in Mammalian Tissues* ROBERT T.

SCHIMKE

Departments of Pharmacology and Biological Sciences Stanford University Stanford, California I. Introduction II. Properties of Protein Turnover in Rat Liver III. Theoretical Formulation of a Model for Changing Enzyme Levels in Animal Tissues IV. Control of Synthesis and Degradation of Specific Enzymes . . . A. Nutritional Control of Rat Liver Arginase B. Hormonal and Substrate Control of Tryptophan Oxygenase (Pyrrolase) C. Effects of Degradation Rate Constants on the Apparent Re­ sponse of Enzymes to Altered Rates of Enzyme Synthesis . . D. Mutations Affecting Rates of Synthesis and Degradation of Specific Enzymes V. On Mechanisms Controlling Synthesis and Degradation of Specific Enzymes A. Control of Enzyme Synthesis B. Control of Enzyme Degradation VI. Concluding Remarks References

77 80 84 86 87 92 98 102 110 Ill 114 119 120

I. Introduction

Information concerning the regulation of enzyme concentrations or levels has been obtained largely from studies with microbial systems, most specifically with Escherichia coli (64). The simplified organization uniform cell population, and ready availability of mutants have allowed for a rapid advancement of knowledge of mechanisms involved in such regulation. An increasing wealth of evidence indicates that in mammals marked changes in enzyme levels occur in response to a variety of physi­ ologic, hormonal, nutritional, and pharmacologie stimuli. The nature of some mechanisms responsible for the regulation of enzyme levels in mammalian tissues, in particular as studied in the author's laboratory, will be the subject of this review. At the outset it should be emphasized that an individual animal * Some of the studies described in this paper were supported by research grant GM14931 from the National Institute of General Medical Sciences, and research grant P-427 from the American Cancer Society. 77

78

ROBERT T. SCHIMKE

cell is not comparable to the unicellular organisms on which most current concepts of regulation of enzyme levels have been based. Thus the types of regulatory phenomena that have evolved in bacteria can be considered to be geared to achieving a maximal rate of cell proliferation consistent with available nutritional sources. In contrast, the individual cell of a multicellular organism is more often dividing slowly, if at all, and is variable in degree of organizational complexity and in the type of specific function it carries out. Such cells are associated with both similar and dissimilar cells into tissues and organs, functioning as an integrated whole in the protection and propagation of specialized reproductive cells. In such a different environment, then, new regulatory problems have arisen, and new solutions have been found or superimposed on those mechanisms common to all organisms. Among the problems encountered by more complex organisms is that involved in the integrated function and development of diverse cell types with differing functions. A solution to this problem has been the elaboration of a complex system of hormones. As will be discussed, many hormones have been demonstrated to be important in regulating the content of specific enzymes in various target tissues. Another problem involves that of how to effect changes in the metabolic machinery, i.e., specific enzymes, in response to environmental and nutritional changes or as part of a developmental sequence. Such changes include removal of unneeded enzymes as well as the synthesis of those newly required. In bacteria the removal process can involve dilution during phases of rapid growth. In animal tissues, on the other hand, where little cellular division takes place, the process of protein degradation becomes increas­ ingly more significant as a means of removing unneeded metabolic ma­ chinery, and therefore as a means for controlling enzyme levels. It is of interest that the rate of protein degradation is greatest in liver, the organ specifically involved in conversion of ingested nutrients to a form that maintains the constancy of the internal milieu, and the organ whose enzyme profile undergoes major changes under various physiological and nutritional conditions. The interaction between the synthesis and degradation of specific enzymes in controlling enzyme levels in animal tissues is reflected in a general comparison of differences in the course of enzyme induction in animal tissues and in bacteria (Fig. 1). Thus in bacteria the total enzyme activity in a culture increases following the introduction of an inducer. Similarly, in animal tissues the level of enzyme can be increased by agents such as hormones, substrates, or changes in diet. However, in contrast to the general finding of stability of induced enzymes in

79

ENZYME LEVELS IN MAMMALIAN TISSUES

bacteria (91 ), the universal finding in adult, or differentiated mammalian tissues, has been that enzyme activity returns to a basal level along an exponential time course once the stimulus is removed (see ref. 124). Although one may consider that these changes in enzyme activity result simply from activation and inactivation of preexisting enzyme protein, such as with muscle and liver phosphorylase (59, 142), this is not gener­ ally the case. Thus protein synthesis has been implicated in the increases in many liver enzymes, since drugs that inhibit protein synthesis prevent BACTERIAL

ANIMAL

TIME

FIG. 1. Schematic time courses of enzyme induction in bacteria and animal tissues.

the increases in enzyme activity (23, 50, 69, 110). More convincing is an increasing number of studies which have combined immunologie and radioisotope techniques in demonstrating both an increased content of immunologically reactive protein, and an active net uptake of radioac­ tive isotope into specific enzyme protein (66, 69, 122, 125, 129). More important, perhaps, is the fact that the changes in enzyme levels take place against a background of continual synthesis and degradation of protein, as initially documented by Schoenheimer and his co-workers (128) and studied more recently in several laboratories (16, 122, 143). This continual turnover is to be contrasted with the lack of demonstrable degradation of protein in exponentially growing E. coli (61, 81, 119), a finding that led to the suggestion that the protein "degradation" ob­ served in animal tissues may result from cellular turnover and the secre-

80

ROBERT T. SCHIMKE

tion of plasma protein, rather than a degradation of intracellular protein (61). More recently, however, Mandelstam has shown that there is an active degradation of protein in nongrowing bacteria (90). It would appear, then, that the animal cell might be more correctly compared to the nongrowing bacteria with respect to protein degradation. II. Properties of Protein Turnover in Rat Liver

Since the regulation of enzyme levels takes place against a back­ ground of continual synthesis and degradation, certain properties of this overall process will be discussed first as a basis for understanding subse­ quent sections. Studies from a number of laboratories have shown that the replace­ ment of liver protein is rapid and extensive. One such experiment, de­ signed to answer questions about the rate and extent of turnover of total liver protein and one specific enzyme, arginase, is shown in Fig. 2 300 r-

600

1200

Liver protein

r-

'S 8 0 0

TCA extroct

400

H

o

2

100 h

2

& 400

200

6 8 10 12 ,4 DAYS ON LYSINE- C DIET

28

FIG. 2. Incorporation of continuously administered L-lysine-14C into total protein, arginase, and trichloroacetic acid (TCA) soluble extracts of rat liver. OsborneMendel rats, weighing 250-275 gm each, were maintained for 7 days on a diet con­ sisting of 25% complete amino acid mixture. After a 12-hour period without food, they were placed on a diet containing L-lysine-14C (specific activity, 1100 cpm^mole). At the intervals specified one rat was killed ; the liver was then removed and divided into two weighed portions. One sample was made into acetone powder from which arginase was isolated by immunological techniques; the other was treated with 10% TCA and divided into protein and supernatant fractions. Radioactivity of the TCAsoluble counts is expressed as counts per minute per extract from 1 gm of liver, wet weight ( O O ) · Counts in total liver protein are expressed as counts per minute per milligram of protein (φ φ ) . Counts in the arginase represent the total number of counts precipitated ( A A)

81

ENZYME LEVELS IN MAMMALIAN TISSUES

{122). In this experiment rats were fed an amino acid diet containing L-lysine-14C of a constant specific activity for up to 28 days. The rate and extent of protein turnover were estimated from how rapidly and to what extent total cellular protein and arginase were replaced from the dietary source. As shown in Fig. 2, the free lysine pool, as estimated by counts present in the TCA-soluble pool, approached maximal labeling in approximately 24-36 hours. The incorporation of lysine into total liver protein, expressed as counts per minute per milligram of protein, was initially rapid, and then slowed markedly after 5-6 days. The rate TABLE I SPECIFIC ACTIVITY OF L-LYSINE-14C ISOLATED FROM TOTAL LIVER PROTEIN AND ARGINASE FOLLOWING 20 DAYS OF CONTINUOUS ADMINISTRATION OF L-LYSINE- U C DIET"

Source of lysine Diet Total liver protein Arginase

Radioactivityrecovered as Lysine specific activity lysine (cpm//nmole) (%) 98 95 66

1105 829 1006

Percent replacement of lysine

75 91

α The liver from three rats maintained on a L-lysine-14C diet for 20 days as described in Fig. 2 was pooled, and arginase was purified as outlined elsewhere {122). L-Lysine-14C of the purified arginase and a sample of all discarded protein (total liver protein), as well as of the initial L-lysine-14C amino acid mixture, was isolated following hydrolysis in 6 N HC1 by chromatography on Amberlite CG-50 (NH 4 + form) columns.

of incorporation of lysine into arginase, expressed as total counts per minute in enzyme as isolated with an antibody that specifically precipi­ tated arginase, was slower. This experiment allows for a measure of the extent to which the cellular proteins are replaced. Thus the lysine-14C incorporated may have represented anywhere from 1% to 100% replacement of the lysine residues of the protein. Such information can be obtained by comparing the specific activity of the lysine in the protein with that of the dietary source. These results are shown in Table I. The specific radioactivity of the lysine isolated from the original diet was 1105 cpm//xmole. After 20 days of labeling, approximately 76% of the lysine residues of total liver had been replaced; i.e., the specific activity of the lysine of liver

82

ROBERT T. SCHIMKE

protein was 829 cpm//miole. In addition, virtually all the lysine of arginase had been replaced in this time. From this experiment we can conclude: 1. The replacement of protein in rat liver is extensive and rapid. At least 50% of the protein is replaced in 4-5 days. Similar conclusions as to the rapid and extensive nature of such turnover have been made by Buchanan (16) and by Swick and his co-workers (143, 146)2. Most of the protein degradation is intracellular, rather than inter­ cellular. This statement is based on the results of studies indicating that the life-span of cells in liver is 160-400 days (16, 97, 145). It follows, then, that since the vast majority of liver protein is replaced within 20 days, the degradation that occurs must be largely intracellular, rather than the result of cell replacement. That the replacement does not represent simply secretion of plasma proteins is again indicated by the fact that after 20 days about 75% of the protein of liver has been replaced. Since the steady-state concentration of the major plasma pro­ tein, albumin, in liver has been estimated to be about 1% of total liver protein (17), it is clear that such a protein cannot account for the bulk of the replaced protein. Furthermore, one intracellular protein, i.e., arginase, is replaced. 3. There is a marked heterogeneity of degradation rate constants of different cell organelles and individual intracellular enzymes. This is already evident from the experiment of Fig. 1, since arginase is re­ placed more slowly than a large portion of the liver protein. In general, information on degradation rates of specific enzymes and proteins is extremely limited, a situation due in part to problems involved in obtain­ ing valid measurements (see ref. 124). All available evidence, however, indicates that rates of degradation are markedly heterogeneous. Table II shows the heterogeneity of degradation rates, or more correctly, degradation rate constants (see subsequent section for definition), of various cell fractions and organelles as isolated by standard techniques (3). L-Arginine-guanidino-^C was used in these studies, since, as shown by Swick and Handa (144), this isotope is not subject to extensive reutilization in liver as are most other amino acids, because of the large amount of arginase present. It is evident that there are marked differences in the mean degradation rate constants of these fractions. Of particular interest is the finding that the proteins of the membrane fractions, i.e., smooth and rough endoplasmic reticulum and the plasma membrane, are in the greatest state of flux. It should be pointed out that all experi­ ments in which the decay of pulse-labeled proteins are studied are subject to the problems of isotope reutilization (80, 115) ; hence, estimated

83

ENZYME LEVELS IN MAMMALIAN TISSUES

degradation rate constants can at best be considered minimal estimates. Recently Swick et al. (I46) have used a method for determining degrada­ tion rate constants that involves the continuous administration of carbonate- 14 C and subsequent isolation of protein arginine. With this method, which is not subject to the problem of isotope reutilization, these workers have found that mitochondrial protein has a half-life of 5 days, a value even lower than that obtained in the studies of Table I, and far less than half-lives obtained in previous studies (6, 37). There is an even more marked heterogeneity of degradation rate constants among individual enzymes in rat liver. A variety of methods TABLE II MEAN HALF-LIVES OF PULSE-LABELED PROTEINS OF SUBCELLULAR FRACTIONS OF RAT LLVER*

Fractions Homogenate Nuclear Mitochondrial Lysosomal Microsomal Supernatant Smooth endoplasmic reticulum Rough endoplasmic reticulum Plasma membrane

Mean half-life (days) 3.3 5.1 6.8 7.1 3.0 5.1 2.1 2.0 1.8

° See Arias el al. (3) for details.

have been employed for determining such rate constants, each with cer­ tain assumptions and limitations (see ref. 124) · For instance, the decay of enzyme activity after an initial increase to high levels by administra­ tion of a glucocorticoid has been used to obtain half-lives of 2-3 days for glutamic-alanine transaminase (129), and 2-4 hours for tryptophan oxygenase (34) and tyrosine transaminase (9). The decay of enzyme activity after prevention of further protein synthesis by puromycin ad­ ministration has a given half-life of 60-75 minutes for δ-aminolevulinic acid synthetase (94). By use of an irreversible inhibitor of catalase, a half-life of approximately 30 hours has been obtained (111). Each of the above methods involves the administration of agents that by themselves may affect rates of degradation. Degradation rate constants of arginase (122), tryptophan oxygenase (125), and tryosine transaminase (71) under basal conditions have been determined by use of combined

84

ROBERT T. SCHIMKE

immunologie and isotopie techniques. Arginase has a half-life of 4-5 day (see Fig. 1 and also Fig. 5). Values obtained for tyrosine transaminase and tryptophan oxygenase of approximately 2-4 hours are simi­ lar to those obtained by less rigorous techniques. The heterogeneity of degradation rate constants of proteins extends to those proteins associated with the endoplasmic reticulum of the rat hepatocyte. Omura et al. (104) found that the half-life of cytochrome 65 as determined by the decay of protein pulse-labeled with leucine-14C was approximately 110 hours, whereas in the same experiment, that of another membrane-associated protein, cytochrome c reductase, was 80 hours. Arias et al. (3) have confirmed this finding, using a double isotope technique to determine rates of degradation of proteins. These workers have found, in addition, that the heterogeneity of degradation rates of membrane associated proteins is extensive. III. Theoretical Formulation of a Model for Changing Enzyme Levels in Animal Tissues

In view of the fact that there is a continual synthesis and degradation of essentially all proteins of liver, any formulation for expressing changes in enzyme levels must consider both synthesis and degradation. A simple formulation for this has been developed in several laboratories (9, 111, 129). Thus a change of an enzyme level is expressed by: dE/dt = k9 - kdE

(1)

where E is the content of enzyme, expressed as units per mass, ks is a zero-order rate constant of synthesis,* expressed as units time-1 mass-1, and kd is a first-order rate constant for degradation^ expressed as time-1. In general there is little, if any, change in total mass of a tissue, * It is obvious that the rate of synthesis of a specific protein will be determined by a number of factors, including the number of ribosomes, amount of messenger RNA, levels of amino acids and tRNA, availability of initiation and transfer factors, etc. In this simplified model the separate roles of such variables have not been factored, since they are largely unknown in mammalian tissues. Hence all such variables have been included under a general notation of a rate of enzyme synthesis. t The rate of degradation of a protein is expressed in terms of a first-order rate constant because in all cases studied, except for that of the red blood cell and its hemoglobin (181), the rate of degradation of specific, intracellular proteins, especially those in liver, has followed first-order kinetics. This includes both the decay of enzyme activity after that activity has been increased to a high level by treatment with various agents (34, 129; see ref. 124), and the decay of specific pulse-labeled proteins under conditions where the total amount of enzyme remains constant, i.e., a steady state (71,104,122,126).

85

ENZYME LEVELS IN MAMMALIAN TISSUES

e.g., liver, during an experimental time period, and consequently an ex­ pression for a change in total mass of tissue is not included. In the'steady state, i.e., when dE/dt = 0, then k8 = kdE

(2)

E

= Γ (3) kd Thus in the steady state the amount of enzyme is a function both of the r^te of synthesis and the rate of degradation.*"* An alteration in either rate, therefore, can affect the level of E. Let us, then, consider the time course of change in enzyme level that may result from various hormonal, nutritional, or physiological manip­ ulations or treatments, where ks is changed to fc/ and kd is changed to

fc/.

The time course describing the approach of E to a new steady state, defined by the new values fc/ and fc/ is given by :

El

Eo

=

J?L- _ ( k · ' kd'Eo

\kd'Eo

e-kd't

(4)

* A certain confusion exists in the use of the term "turnover" and "degradation." I n the literature on bacteria protein "turnover" is used to denote degradation of protein (91). I n the case of animal tissues "turnover" has been used to denote the general phenomenon whereby tissue constituents are continually synthesized and degraded, as well as used to denote the more limited process of degradation. I n this paper "turnover" is used only to denote the overall renewal process. In this context, then, the concept of a "turnover rate" has a limited meaning, inasmuch as only in the steady state is there justification for considering such a rate. Since the steady state is a special, limited instance of the more general state wherein levels of specific tissue proteins are changing continuously (76), the more specific and meaningful terms, "rates of synthesis" and "rates of degradation" are used. Confusion also arises in the meaning of the term, "rate of degradation." I t is evident from the above formulation t h a t an "absolute rate of degradation" of an enzyme, expressed as units/time, will always be a function of the level of E, i.e., kdE. Hence the term "rate of degradation" as applied to experimental situations in which the absolute amount of enzyme is varying is not very meaningful if it is used to denote an absolute rate. Since one is generally concerned about degradation rates, especially as they relate to mechanisms of altered enzyme levels, the important question is whether the rate constant oj degradation is the same for different proteins, and whether that rate constant has been altered by a specific treatment. Thus, when one uses the term, "rate of degradation," it is used to denote a rate constant of degradation as defined above [Eq. ( 1 ) ] . Rate constants of degradation are often expressed in terms of a half-life. The half-life is given by : 0.693 tH = -7— kd

(&)

86

ROBERT T. SCHIMKE

where Et is the activity at any time, £, and where E0 is the enzyme activity under steady state conditions defined by ks and kd. As written if E0 were taken as 1, the equation would represent the "fold" increase in enzyme activity, an expression commonly used in studies on manmalian tissues. Equation (4) shows that, although a new steady state is defined by the new values for fc/ and fc/, the half-time required to shift from one steady state to another is determined only by the rate constant of degradation, fc/ (9). The significance of this will be discussed later. Several general conclusions can be drawn from this formulation: 1. Whereas in bacteria enzyme induction involves an increase in the rate of synthesis, the accumulation of enzyme protein in mammalian tissues may result from two general mechanisms: (a) an increase in the rate of enzyme synthesis, and/or (b) stabilization of existing enzyme in the presence of continued enzyme synthesis. 2. Any condition that reduces the stability of an enzyme, either by directly labilizing it, or by removing a stabilizing factor, could result in decreasing enzyme levels in a manner that would mimic enzyme re­ pression as found in bacteria. 3. The fact that an agent causes an increase in the activity of one enzyme relative either to total protein or to another enzyme at some finite time does not necessarily indicate a specific effect on that enzyme. Thus, as indicated in Eq. (4), the time required to approach a new steady state is solely a function of the rate constant of degradation. Since there is a great heterogeneity of values for kd among different proteins, as discussed previously, the time course of increase in activity of each enzyme will vary, even if the rates of synthesis of all proteins are increased to the same extent. IV. Control of Synthesis and Degradation of Specific Enzymes

As mentioned in the Introduction there are many instances in which the amount of an assayable enzyme activity has been modified by a variety of nutritional, hormonal, or pharmacologie manipulations. Exam­ ples of some of these will be described below. In studying the mechanisms involved in such changes, a logical sequence of questions and answers occurs. Thus the first question is whether the differences in assayable enzyme activity result from differences in content of enzyme protein. Once this has been established, the question of whether the rate of syn­ thesis, the rate constant of degradation, or both, have been altered can be asked. In the author's laboratory we have attempted to answer these questions as directly as possible by isolating the specific enzyme protein

87

ENZYME LEVELS IN MAMMALIAN TISSUES

in question. Generally this has involved the use of immunologie tech­ niques as adjuncts to isolation and purification of the protein. We have preferred this approach, with its limitations, to the more indirect ap­ proaches using drug inhibitors, such as puromycin, cycloheximide, and actinomycin D, since it is not entirely clear in the intact animal to what extent the results observed can be related to a primary effect of inhibiting specific protein or RNA synthesis, and to what extent the results are second- or third-order effects of these toxic agents (44, 62, 88,105,117,134). A. Nutritional Control of Rat Liver Arginase 1. STEADY STATE CONDITIONS

As is the case with many liver enzymes that are involved in the catabolism of amino acids (77), a direct relationship has been demonTABLE III EFFECT OF D I E T ON STEADY STATE LEVELS, S Y N T H E S I S , AND DEGRADATION OF RAT LIVER ARGINASE 0

Diet

Activity (/xmoles/gm wet weight X 10-3)

Half-life6 (days)

kd (day-»)

(units /gm/ day X IO"3

8% Casein 30% Casein 70% Casein

20.2 + 1.0 36.7 + 1.3 56.1 ± 1.1

5.2 4.8 4.6

0.13 0.14 0.15

2.6 5.2 8.4

a

b

*.

D a t a from Schimke {122).

See Fig. 5.

strated between the levels of such enzymes and the caloric intake pro­ vided in the form of protein. Among such enzymes are those of the urea cycle, i.e., carbamyl phosphate synthetase, ornithine transcarbamylase, argininosuccinate synthetase, argininosuccinase, and arginase (122). We have studied arginase particularly because it can be purified readily to a homogeneous state and is capable of eliciting precipitating antibodies when administered to rabbits. As shown in Table I I I there is a 2- to 3-fold difference in the specific activity of liver arginase be­ tween animals maintained 14 days on a diet containing either 8% or 70% of the diet by weight. That this difference in activity results from a difference in the amount of enzyme protein is supported by studies using an antibody specific for arginase (122). The arginase used to im-

88

ROBERT T. SCHIMKE

munize rabbits was purified to the point of homogeneity as indicated by sedimentation velocity studies in an analytical ultracentrifuge and by the presence of a single protein band on acrylamide gel electrophoresis. Figure 3 shows a typical Ouchterlony double diffusion pattern of such an antiserum (center well) reacting with arginase from various

FIG. 3. Ouchterlony double diffusion patterns of rat liver arginase. A, Center well, contains antiserum against arginase (specific activity, 5480 units per mg of protein) ; well 1 contains crude liver extract (specific activity, 10 units per milligram of protein) ; well 2, arginase (specific activity, 220 units per milligram of protein) (from heat step) ; well 3, arginase (specific activity, 1010 units per milligram of protein) ; well 4, protein which contains no arginase activity, nonadsorbed protein from a CM-cellulose column. In wells 1 to 3 were placed 250 units of arginase activity ; 250 mg of nonarginase protein were placed in well 4 U%&) ·

stages of purification. In all cases a single, continuous line of precipita­ tion is seen. This antiserum also shows a single line of precipitation when crude extracts from livers of animals maintained on different diets with differing amounts of protein are studied. Figure 4 shows quantita­ tive precipitin reactions of the antiserum with highly purified enzyme (5480 units/mg), as well as with crude liver extracts with specific activi­ ties varying by 4-fold (10 and 39 units/mg). These crude extracts were from livers of animals maintained on 8% or 70% dietary protein, re-

89

ENZYME LEVELS IN MAMMALIAN TISSUES

spectively. The amount of enzyme activity neutralized (precipitated) by a constant amount of antiserum was the same, despite the fact that the total amount of protein added varied over a 500-fold range in the three arginase preparations. Furthermore, as indicated by the amount of protein precipitated, the quantitative nature of the precipitation reac­ tions were similar with the highly purified enzyme and with the crude

2

4

6

ARGINASE UNITS X IO - 2 ADDED

FIG. 4. Quantitative precipitin reactions of purified arginase and liver extracts. Arginase preparations were of three sources: (a) purified arginase (specific activity, 5480 units per milligram of protein), A A ; (b) a crude arginase preparation from animals maintained on 8% dietary casein (specific activity, 10 units/mg), Δ Δ ; (c) a crude arginase from animals maintained on 70 % dietary casein (specific activity, 39 units/mg), # φ . Volumes and protein concentrations were made constant by equalizing the specific activity of all preparations by addition of suitable amounts of bovine serum albumin. See Schimke {122) for details.

extracts of differing specific activities. These results, then, demonstrate that the differences seen in the amount of assayable enzyme activity in livers of animals maintained on diets with differing protein contents represent differences in the amount of enzyme protein as determined immunologically. The next question to be answered is whether the difference in enzyme content resulting from maintaining the animals on diets with differing

90

ROBERT T. SCHIMKE

proportions of protein results from a more rapid rate of synthesis, or a less rapid rate of degradation, i.e., from Eq. (3), is ks or kd altered? This question was answered by determining experimentally the values for kd at three different steady-state levels of arginase. In this experi­ ment, the details of which are given in the legend to Fig. 5, animals were pulse-labeled with arginine-guanidino- 14 C, and thereafter the decay of specific radioactivity of labeled total protein and arginase was deter­ mined with time. It can be seen in Fig. 5 that the degradation rate TURNOVER 8% CASEIN

OF

ARGINASE

30%CASEIN

70% CASEIN 1

1

1

1

1

'^PROTEIN tp 2.6]

0

2

4

6

8 "0 2 4 6 8 "0 2 4 6 8 DAYS AFTER GUANID0 J4 C-ARGININE INJECTION

-^ARGINASE

FIG. 5. Turnover of total liver protein and arginase determined by the decay of radioactivity after single administration of L-arginine-guanidino-14C. Following main­ tenance of 20 rats for 14 days on each of three diets containing 8, 30, or 70% casein, each rat was given a single intraperitoneal injection of 25 μϋί of L-arginine-guanidino14 C. One hour later, and at 2-day intervals, four animals from each dietary group were killed and the four pooled livers were subjected to purification of arginase. The protein fractions discarded during the arginase purifications were pooled to constitute total liver protein. Results are expressed as counts per minute per milligram of protein. O O , total liver protein; % Φ , arginase. See Schimke {122) for details.

constant of arginase, here indicated by a half-life value, is essentially the same in the three steady states, i.e., t^ = about 5 days. From such information, then, it is possible to calculate a rate of synthesis of argi­ nase, as shown in Table III. Such a calculation shows that the rate of enzyme synthesis, rather than the rate of enzyme degradation, is affected by variations in dietary protein content. 2. CHANGING NUTRITIONAL CONDITIONS

During abrupt and extensive changes in the nutritional status of animals, differences in rates of degradation of arginase also occur. Two experiments illustrative of this principle are summarized in Fig. 6. In

91

ENZYME LEVELS IN MAMMALIAN TISSUES

one of these, animals that had been maintained previously on an 8% protein diet were starved for a total of 6 days. This resulted in a net increase in total arginase of about 2-fold. This net increase in total arginase results from a continued enzyme synthesis in the absence of CHANGE FROM 70% TO 8% CASEIN

STARVATION UJ

9

-

M

^r

o z6 2Ec5 ^3 -

iS 0.5

¥> 0

(SYNTHESIS)

i . I I

§0.5 1.0

1

X -L

L·

(DEGRADATION)

CO

ospr*ory^e

0.4 *

-m.2

hl Υ*Φ*\

ι

1

ι

1 L

I

10 Hours

■

ι

12

■

1

14

ι

1

ι

1

ι

16

FIG. 6. Changes in glycogen, ADP-glucose pyrophosphorylase, However, this has not been investigated further. The Ka value for AMP was shown by Hirata et al. {20) also to vary with the L-threonine concentration. Thus, the biodegradative de­ hydrase displays a mutual heterotropic interaction between AMP and L-threonine, but is devoid of homotropic interactions. 2. MOLECULAR EFFECTS

Phillips and Wood {41) first observed that AMP causes a drastic change in molecular weight of the dehydrase, which can be observed with partially purified fractions either by sucrose density gradient centrifugation or by gel filtration {39, 53). The sedimentation velocity values increase smoothly from 3.2 S in the absence of AMP and at low protein concentrations to 8.0-8.2 S in the presence of AMP and at moderately high protein concentrations. Assuming that both the dehydrase and the internal standards behave as typical globular proteins, the molecular weight range would be 40,000-160,000. These molecular weight calcula­ tions from sedimentation velocity have been confirmed by gel filtration {53). Even in the fresh crude extract in the absence of AMP, the sedimen­ tation coefficient (s) value and molecular weight were low (ca 3.5 S and 40,000, respectively). This indicates that the monomeric form of 40,000 is a naturally occurring and spontaneously formed species. With as little as 0.2 unit of dehydrase, the sedimentation velocity was 3.2 S in the absence of AMP and 7.3 S in the presence of AMP {53). This amount of dehydrase is in the linear range of the spectrophotometric

L-THREONINE DEHYDRASES OF MICROORGANISMS

165

assay for activity and, thus, the change in sedimentation behavior can be correlated with the change in Km. Increasing concentrations of AMP caused a smooth increase in sedimentation velocity (4I) and a continuous decrease in the Km for threonine. These observations lead to the conclu­ sion that conversion of monomer to tetramer is linked to, or causes the change in, Km for L-threonine. However, the possibility that these are only coincidentally related has not been eliminated by definitive experiments. Since a single peak is observed in sucrose gradients and on gel filtra­ tion columns over a wide range of conditions, it is believed that the various molecular species from monomer to tetramer are in rapid equilib­ rium and that the s value observed and corresponding molecular weights are a statistical measure of the relative amounts of the various forms. A theoretical basis for this viewpoint has been provided by Gilbert (17). Under certain limited conditions, however (low protein concentration), two molecular species may be observed, and, after artificial oxidation of the dehydrase and reduction with dithiothreiotol, many activity peaks are observed in sucrose gradients and on Sephadex columns. For many of the minor peaks, the calculated molecular weights fall lower than 40,000 and between the 40,000, 80,000, and 120,000 peaks; this suggests the possibility of subunits that can recombine to give nonintegral molecu­ lar weights (39, 53). As described below, this behavior is commonly observed in C. tetanomorphum (54). In the absence of AMP, the equilibrium among forms is shifted in the direction of the tetrameric species by increased protein concentrations (53). However, when traces of AMP are rigorously excluded, the protein concentration must be high (i.e., 2 mg/ml) to cause an appreciable pro­ tein concentration-directed accumulation tetramer. Plots of the log of units applied to the gradient versus sedimentation velocity increased linearly in both the presence and absence of AMP, but with greatly differing slopes. The lines, if extended, would intersect at 8.2-8.5 S for high protein concentration and, in the absence of AMP, reach a limiting value of 3.2 S at low protein concentration. It has been possible to show by assays of crystalline dehydrase (42) at high protein concentration that the higher molecular weight species are formed in the absence of AMP and that there is no change in Km value from that of the low affinity form normally observed in catalytic assays in the absence of AMP. If there is rapid equilibration among the forms, the observed increases in sedimentation velocity to 6.7 S under these conditions would indicate that some appreciable proportion of the molecules are in the tetramer form and that a sizable change in

166

W. A. WOOD

Km should be observed. Since this was not the case, these higher molec­ ular weight forms, very likely including tetrameric species, must differ from those produced by AMP and indicate that binding of AMP rather than oligomer formation per se is required to elicit the Km change (42). Oxidation with either oxygen or GSSG inactivates the dehydrase, converts all species to monomer, and abolishes the ability of AMP to cause tetramer formation (39, 53). The oxidized dehydrase, while able to reform active dehydrase by reduction, is more unstable than the re­ duced form. After incubation with dithiothreitol, activity returns, as does the property of AMP-directed tetramer formation, but, as noted above, multiple peaks may be observed by centrifugation and gel filtration. B. Pyridoxal Phosphate

Pyridoxal phosphate is more firmly bound than in many enzymes and is partially resistant to carbonyl reagents (50). After resolution, the dehydrase becomes more unstable and EDTA must be added to preserve activity. After treatment with cysteine and EDTA, pyridoxal phosphate is completely removed, and the s value drops from 3.2 S to 2.6 S, whereas the molecular weight remains at 40,000 (53). AMP will not promote tetramer formation with the inactive resolved dehy­ drase. Addition of pyridoxal phosphate results in virtually complete re­ covery of activity. The s value increases to the 3 to 4 S region in the absence of AMP and, upon addition of AMP, the s value rises to 7.8 S. Resolution is prevented by inorganic phosphate or AMP (39, 53), both of which are effective in preserving activity during storage. The homogeneous apodehydrase has been shown to be more susceptible to inactivation under a number of conditions (44)- Thus, conformation is important in resolution (53), and a conformational change results from removal of pyridoxal phosphate which affects both stability (44) and the ability to form tetramer (53). Figure 1 summarizes the molecular changes observed. I t is considered that an active monomeric species of molecular weight 40,000 with low affinity for L-threonine is in facile equilibrium with two tetrameric forms. AMP and protein concentration shift the equilibrium in the direction of one tetrameric form which has a high affinity for L-threonine. Protein concentration alone shifts the equilibrium in the direction of a second tetrameric form which has a low affinity for L-threonine. The equilibrium position for the second tetramer is far in the direction of monomer. By removal of pyridoxal phosphate, the monomer also may be reversibly converted to inactive apomonomer of molecular weight 40,000 but of

167

L-THREONINE DEHYDRASES OF MICROORGANISMS

RESOLVED MONOMER 40,000 s=2.6S

Î

INACTIVE

i

OXIDIZED MONOMER 40,000 s=3.0-3.5S

TETRAMER 1

^

1 ACTIVE 1 MONOMER 40,000 HIGH Km