Volume 26, number 2 MOLECULAR • CELLULAR BIOCHEMISTRY July 31, 1979

ARGINYL RESIDUES A N D ANION BINDING SITES IN PROTEINS*

James F. R I O R D A N

Biophysics Research Laboratory, Department of Biological Chemistry, Harvard Medical School; Division o[ Medical Biology, Peter Bent Brigham Hospital Division, Affiliated Hospitals Center, Inc. Boston, Massachusetts 02115, U.S.A.

(Received March 22, 1979)

Summary Introduction

The functions of a number of amino acid residues in proteins have been studied by chemical modification techniques and much useful information has been obtained. Methods using dicarbonyl compounds for the modification of arginine residues are the most recent to have been developed. Since their introduction about 10 years ago, they have led to the identification of a large number of enzymes and other proteins that contain arginine residues critical to biological function. These reagents are discussed in terms of their chemical reactivity and mechanisms of action and in relation to the unique chemical properties of the guanidinium group. Butanedione, phenylglyoxal and cyclohexanedione are the most commonly employed arginyl reagents, and their relative advantages are examined. A survey of the functional role of arginine residues in enzymes and other proteins is presented in which nearly 100 examples are cited. The prediction that arginine residues would be found to serve a general role as anionic binding sites in protein has obviously been validated. The genetic and physiological implications of the selection of arginine for this important function are discussed.

* This work was supported by Grant-in-Aid GM-15003 from the National Institutes of Health.

Current views on the evolution of protein structure are generally based on the constancy as well as universality of the genetic code. For many millions of years, or at least since the emergence of the code in its present form, proteins have been constructed from the same twenty amino acids that we know today. These amino ac ids- actually their side cha ins- must encompass all the physicochemical properties necessary for the formation of polypeptide macromolecules that have relatively stable, three-dimensional structures and can exhibit biological function. In broad terms, of the twenty different side chains, eight are hydrophobic and twelve are hydrophilic. Of those that are hydrophilic, five are charged and the rest are uncharged. Of those that are charged, three are positive and two are negative. This all inclusive categorization is obviously sufficient to meet the needs of functional proteins.

We probably do not know all of the functions subserved by the various amino acid residues in proteins. Those with hydrocarbon side chains may do little more than generate the entropic forces needed for folding and provide the hydrophobic regions for inter- and intramolecular interactions. The more chemically reactive residues can not only stabilize structure and assist in the binding of ligands but also enter directly into the bond making and breaking steps associated with e.g.

Dr. W. Junk b.v. Publishers- The Hague, The Netherlands 71

enzyme catalysis. Much has been learned about the latter group of residues particularly by means of selective chemical modification reactions.

Chemical modification as an approach to studying the biological function of proteins was an outgrowth of the classical derivatization techniques of organic chemistry. Many of the reagents now in use were first employed with proteins long before the chemical composition of these macromolecules was known and, in fact, even before proteins were known to be macromolecules. A considerable degree of clarification took place during the 1940's and 1950's, largely due to the systematic work of HERRIOTr and FRAENKEL-CONRAT, and in fact, present day protein chemists are still reasonably well served by the methods described in their reviews 1"2. With a few exceptions, most of the advances made in the last twenty or so years have resulted from increased reagent specificity and selectivity. The amino acid analyzer, radioactive and chromophoric reagents and the knowledge of protein structure contributed by X-ray crystallography have greatly assisted this process.

Early in this century, modifications were carried out largely to gain understanding of the chemical composition of proteins. Later, this approach became useful to study protein function. For example, by the mid-1950's almost 100 enzymes had been classified as sulfhydryl enzymes because they were inactivated by compounds known to react with the thiol group of cysteine 3. HARTLEY categorized the proteolytic enzymes as acidic, sulfhydryl, or seryl based in part on their susceptibility to specific chemical reagents 4. As knowledge of protein structure grew studies became more sophisticated. Carefully designed reagents could be used to probe details of reaction mechanisms and to detect the changes in protein conformation that accompany function. The list of reagents that have been employed to modify proteins has never ceased to lengthen, though as indicated above, most additions are just variations of earlier themes. Yet, in spite of the tremendous effort that has accompanied this lengthening process it is still only possible to modify about half of the twenty different amino acid residues in proteins. Some of the types of reactions that are used for amino

72

TABLE I

Reactions for ModificatiQn of Functional Groups in Proteins

Functional Grou E Reactions

Amino of lysine

Guanidiaium of arginine

Carboxyl of aspartic and glutamic acids

Thiol of cysteine

lmidazole of histidine

Thioether of methionine

Phenol of tyrosine

Indole of tryptophan

Hydroxyl of serine and threonine

Acylation, aklylation, arylation, reaction w~th canbonyls

Condensation with dicarbonyls

Esterification, amide formation with carbodiimides, reduction

Oxidation, arylatian, alk~vlation, B-elimination, heavy metal derivatives

Alk.vlation. diazonium coupling, iodination, oxidation, photooxidation

Oxidation, alkylation

Acylation, alkylation, todination, nitration oxidation, diazonium coupling

Alkylatio~formylatiDn~ oxidation, ozonlysis

Esterification, phosphorylation, alkylation

acid modification are given in Table I. Most of these reactions are non-specific; they occur with more than a single residue or even a single type of residue. Fortunately, certain residues, because of their microenvironment within the three- dimensional structure of the protein, may exhibit reactivities different from those expected on the basis of studies with model systems. These are often the residues that participate in biological function and, in fact, the structural features generating function are closely related to those giving rise to unusual chemical reactivity. Because of this unique reactivity, they can be modified selectively thereby facilitating structure-function correlation studies. Moreover, it becomes possible to employ a variety of reagents to modify the same residue. With tyrosine, for example, reactions can be carried out either on the phenolic oxygen or at the 3-position of the aromatic ring 5. Structural circumstances that activate one of these sites usually activate the other as well. Depending on the nature of the 3-position substituent, the pK of the phenolic hydroxyl group can vary over several units. This provides a means to explore the pH dependence of function as it relates to degree of ionization of the specific tyrosine in question.

Once a method has been developed for modification of a given type of amino acid side chain there soon follows a series of reports demonstrating the functional role of such a residue in a number of different proteins. HARTLEY identified the "seryl proteases ''4 by virtue of the fact that they all were inactivated

by diisopropylphosphofluoridate (DFP), a reagent that phosphorylates a specific seryl residue at the active site of these enzymes. DFP cannot be called a seryl reagent, however, since it does not react with seryl residues in general, but only with those at the active sites of certain enzymes. On the other hand, a large number of reagents has been used to identify the "sulfhydryl enzymes". These enzymes are inactivated by heavy metal ions such as Ag + and Hg 2+, p-mercuribenzoate, N-ethylmaleimide, and iodoacetate, among others, which are indeed sulfhydryl reagents. However, in many of these instances, loss of activity is not due to the selective modification of an active site cysteine residue but rather to effects on structure or various alternatives that have been discussed elsewhere 6. Thus, classification of proteins based on functional site residues is operationally convenient but can be mechanistically misleading. With all protein chemical modifications it is important to understand the specificity of the reagent employed and its effects on protein structure before making functional deductions.

One of the few amino acids for which a selective reagent had not been proposed by the 1940's was arginine. Its guanidinium group does not react with substances that modify lysine, for example, at least not under the mild conditions (neutral pH, aqueous solution, low temperature) that are preferred for studying functional proteins. Nevertheless, the value of a suitable arginine reagent had been recognized for some time, primarily because it would enhance the proteolytic specificity of trypsin and thereby facilitate protein sequence studies. Since protein sequencing does not require an intact native structure, the first arginine methods to be reported 7,s were actually carried out in strongly acidic or basic media. These reports proved to be catalytic, however, and milder methods appeared soon thereafter 9'I°'11'12. In the past decade, arginine modification has become almost commonplace and an extraordinary number of proteins has been found to contain essential arginyl residues. This review will describe the development of reagents specific for the chemical modification of arginyl residues, survey the current status of "arginyl enzymes", and examine the general role of arginyl residues as anion recognition sites in

proteins. In addition, other roles for arginine will be discussed especially as they relate to the frequency of occurrence of this amino acid in proteins and to the evolution of the genetic code.

I. The chemical properties of the guanidinium group Among the various reactions listed in Table I, the most recent to have been discovered are those cited for the chemical modification of arginine residues. One reason for this is the unique chemistry of the guanidinium group. Its two -NH2 functions do not exhibit the same chemical reactivity as the e-NH2 group of lysine. The latter can generally be modified under slightly alkaline conditions where some of the free base form is present but the guanidinium group is protonated over the entire pH range of protein stability. In fact, guanidine is the strongest organic base known. The guanidinium ion can be represented by three equivalent canonical forms, as opposed to the nonequivalent hybrid forms of guanidine. The difference in resonance energy of 6-8 kcal/mol accounts for the stability of the protonated species 13. A somewhat smaller amount of resonance stabilization occurs in arginine and other alkylated guanidines but the effect is sufficient to give a pK of about 12.5.

Another feature of the guanidinium group is the coplanarity of the three nitrogens and the central carbon atom brought about by the partial double bond character of each of the C--N bonds. This structure has been seen clearly by X-ray crystallographic analysis of proteins and of methyl guanidinium salts TM. It enables the guanidinium side chain of arginine to enter into extended patterns of hydrogen bonding that are unique, thus perhaps explaining the evolutionary selection of arginine as a constituent of proteins.

II. Chemical modification of arginine residues A large number of a-dicarbonyl compounds (dialdehydes, ketoaldehydes and diketones) have been employed during the past 15 years to modify arginyl residues in proteins. Initial interest in these reagents stemmed from the studies of ITANO and coworkers 7'15 who showed that benzil and 1,2-cylcohexanedione react with arginine in strong alkaline solution. At about

73

the same time KING 8 demonstrated that malondialdehyde in very strong acid could be used for this purpose and later it was found that with nitromalondialdehyde optimal reaction occurred between pH 12 and 1416. All of these procedures were satisfactory for blocking arginines for sequence work but the extreme conditions were hardly appropriate for structure-function studies. Consequently, efforts were made to find a dicarbonyl compound that could be used under mild conditions.

A. Butanedione In 1966, YANKEELOV and coworkers 9 reported that 2,3-butanedione solutions react with guanidinium salts. Their evidence suggested that an initial self-condensation of the reagent was necessary before it would react. Condensation took place by incubating the butanedione solution in phosphate buffer at pH 8 for a day or two. GROSSBERG and PRESSMAN 17 used such a preparation to study arginyl residues at antibody combining sites. Subsequently, YANKEELOV was able to prepare the crystalline dimer and trimer of butanedione and demonstrated their reactivity with arginine, both free and in proteins TM.

This reviewer first employed butanedione to investigate the active site of carboxypeptidase A 11A9"2°. The specificity of this enzyme called for a positively charged residue to be at the active site serving as a receptor for the terminal carboxylate group of peptide and depsipeptide substrates. At the time, X-ray crystallography had not yet identified this residue as Arg-145. For our initial experiments, a 15% aqueous solution of butanedione was incubated overnight in buffer at pH 8 to generate the reactive species. Since carboxypeptidase is inhibited by phosphate, other buffers, e.g. borate, Tris, Veronal, bicarbonate and Hepes, were examined for their suitability. Borate and Tris were found to interact with butanedione as evidenced by an immediate fall in pH on addition of the reagent to the buffer solution. The rest did not appear to react with butanedione nor to have any other specific effects. After the incubation period at pH 8 the borate, solution remained pale yellow in color but the others had become very dark, probably due to oxidation and oligomerization. An investigation of the optimal incubation time

?H, 9H, c., H + H O . ' . O - G - N H . c,.o #,,~-c--N.. ~ .o-~-N.. _*

coo + .o-~-N.- ~-~" ---" .O'~'O-~-N~ "°° "~ CH 3 CH= (I) OH 3 (It)

CHEOH CH, C-NH '" H2C=C-NH + HaC-C-N½ ÷ ~_NH~ -C=NHR ~ HO_~_ NH ;C=NHR O=~_ NH ~c=NHR

~H 3 (IV) C:H~ (11[)

SCHEME I

revealed to our surprise that waiting was actually unnecessary and that freshly prepared solutions were perfectly adequate for reacting either with free arginine or with carboxypeptidase 19"2°. In fact, monomeric butanedione was found to react even faster than the trimer. In addition, modification in borate buffer specifically enhanced the rate of reaction and also stabilized the product.

This effect of borate turned out to be critical for establishing both the specificity and mechanism of the modification reaction. The reaction of butanedione with arginine is analogous to its reaction with benzamidine 21'22 which proceeds via the formation of a 4,5- dimethyl-4,5-dihydroxy-2-imidazoline (Scheme I, I). A similar reaction scheme has been proposed for the condensation of glyoxal or benzil with urea 23. The formation of a borate complex (Scheme I, II) with such an intermediate could then account for the specific buffer enhancement effects. According to this scheme butanedione reacts reversibly with the guanidino group of arginine to form the cis-diol, dihydroxyimidazoline derivative which then complexes rapidly and reversibly with borate 24. Thus, in the presence of borate the reaction proceeds faster due to product stabilization. Intermediate I is moderately stable: if the butanedione-carboxypeptidase reaction is carried out for an hour in the absence of borate and borate is then added, formation of complex II occurs instantaneously as evidenced by the effect on enzymatic activity (Fig. 1, arrow). On the other hand, if after an hour the reaction mixture is gel filtered to remove excess reagent, the intermediate dissociates, arginine is regenerated, and native enzymatic activity is restored. If gel filtration is performed in the presence of borate, of course, the borate-diol complex is stabilized and there is no return of native activity over many hours. Reaction of carboxypeptidase with butanedione in the

74

30C

o o

20C

ioo:

I.-

i-- 50

A L fj [3

f . . • m m " m " " /" ° ° "" . . . . . Z Y- - l - - - - . . . . . ""

/ ~ ' O ~

°~o ;/

I Ibb[ 30 60 0 MINUTES

t I 60 120

Fig. 1. (A) Changes in esterase (11, []) and peptidase (O, O) activities on modification of carboxypeptidase A (0.15 mM) with butanedione in 50 rnM borate- lu NaC1, pH 7.5 (9 m~ reagent, dosed symbols) or in 20 mM Veronal-lM NaC1, pH 7.5 (75 mM reagent, open symbols), 20 °. The changes in activity immediately on addition of borate after 1 hr. to the sample reacted in Veronal buffer are indicated by the ar- rows. 03) Changes in activities of the samples reacted in borate buffer subsequent to gel filtration through Bio-Gel P-4 equilibrated with either 50 mM borate-lM NaC1, pH 7.5 ( - - - - ) or with 20 mM Veronal-lM NaC1, pH 7.5 ( - - - - ) . Activities are expressed as the ratio of that of the modified enzyme, V, and the unmodified control, Vc, times 100.

presence or absence of borate for periods longer than an hour leads to the formation of another derivative. Thus, if the reaction is allowed to continue overnight the resultant product can no longer be reactivated by gel filtration and addition of borate at that time has no apparent effect. It is likely that intermediate I rearranges to a nondissociable product (III) by a pinacol- type process analogous to that reported earlier 22. A closely related product was obtained by reacting arginine with cyclohexanedione under strong alkaline conditions is.

Evidence for the glycol nature of I is given by the effect of periodate on butanedione-modified carboxypeptidase. Although periodate does not alter the activities of the native enzyme, it prevents the restoration of activity to the enzyme modified in the absence of borate, which would otherwise occur on dialysis or gel filtration. Presumably, periodate acts by cleaving the glycol.

Intermediate I can undergo yet another type of alteration indicated by the third reaction pathway in Scheme I. Under the conditions employed for protein hydrolysis prior to amino

acid analysis, the intermediate is probably dehydrated to form a methylene species which then undergoes an anionotropic shift to give the hydroxymethylimidazoline (IV) 21. The product does not break down to regenerate arginine and, hence, it is possible to measure the degree of arginine modification in proteins by routine amino acid analysis.

The specific enhancement effect of borate buffer together with the reversibility of the modification (within the time limitations mentioned) and prevention of reversibility by borate provide a convenient means to recognize arginine modification in proteins. It should be stressed, however, that the concentrations of butanedione and borate buffer optimal for modification of a given protein must be established empirically. In the case of carboxypeptidase, for example, using 10 mM butanedione the best buffer concentration was found to be 50 mM. On varying the buffer concentration above or below this value, the changes in enzymatic activity are not as great. In fact, in 0.5 M borate there is virtually no effect on activity at all (Fig. 2). This is not due to borate binding at the active site of the enzyme and, thus, preventing modification, but rather to an interaction of borate with butanedione. This interaction is apparent from the marked drop in pH that occurs on adding the reagent to borate buffer. Further, the absorption maximum of an aqueous solution of butanedione at 408 nm (e 1.15) is abolished by addition of borate (Fig. 3) while the intensity of

2° 25o /

2 O 0 K

f50 >_"

_> 100 1-

5o ~ ' ~ -

0.1 0 .2 0.3 0.4 0 5

I B O R " T E ] , ill

Fig. 2. Changes in esterase (ll) and peptidase (.) activities on modification of carboxypeptidase A (0.21raM) with butanedione (30 raM) in NaC1, pH 7.5, 20 °, with different concentrations of borate buffer. Aliquots were removed after 15 roan. and assayed,

75

Z ~C

05

0.3

03

250

I

300 350 400 450

k ,nm

Fig. 3. Absorption spectra of 30 mM butanedione in water at pH 7.5. The solutions contained borate at the following concentrations 1) no borate present, 2) 50 mM, 3) 200 rnM, 4) 500 mM.

the band at 284 nm (s 9.75) is reduced by__about 75%.

In general, the best reaction conditions will be a function of the concentration of protein being modified and the reactivity of its arginyl residues which, in turn, will determine the concentration of butanedione required to give an appreciable extent of reaction within a reasonable period of time. Borate will be required to stabilize the glycol intermediate but too much borate will reduce the reagent concentration and slowdown the reaction. It has been reported 25 that the borate complex of 1,2-cyclohexanedione is the species that reacts with arginine. Complex formation is thought to occur selectively with one of the hydrated isomers of the dione which then undergoes condensation with the guanidino group. This scheme seems unlikely in the case of butanedione, at least when it reacts with carboxypeptidase, since the same final activity is obtained if borate is present at the start of the reaction or if it is added just at the end (Fig. 1). It is, thus, more reasonable to assume that the effect of borate is to stabilize the reaction product rather than direct the reaction specificity. The kinetics of inactivation of a- aconitase 26 and D-serine dehydratase 27, among others, by butanedione/borate are consistent with this conclusion.

Butanedione is quite selective for modification of arginine residues in proteins. Depending on the conditions employed, lysine modification can be negligible 2°. In the absence

76

of borate, high concentrations of reagent are required to achieve extensive derivatization and under these circumstances, particularly at alkaline pH, lysine or a-amino group modification becomes more prominent. It has been reported that a-dicarbonyl compounds may react with free sulfhydryl groups 28 but protection experiments a9 and the lack of effect of added thiols 3° would indicate that this is not likely to be a problem. There is no evidence for modification of tyrosine, tryptophan or other amino acids under ordinary conditions. It is important to note, however, that dicarbonyls are prone to photosensitization 31'32 and, hence, their reaction with proteins should be carried out in the dark. This is especially true for reactions that are run for many hours or days. Details of the mechanism of interaction of butanedione with arginine are, as yet, not fully known and the possibility of a free-radical process has not been ruled out. Therefore, unless precautions are taken to avoid photoactivation, the possibility of tyrosine, tryptophan, or some other residue being altered must always be considered.

Quantitation of arginine modification by butanedione is accomplished by amino acid analysis. Probably no more than 10% of the modified product is converted back to free arginine on hydrolysis in 6 N HC1. The initial product is unstable under these conditions and is converted to a substance that is not detectable with ninhydrin. Hence, quantitation is done by difference, i.e. by measuring the loss of arginine. This procedure is not capable of detecting changes of one or a very few residues in proteins that have a very high arginine content. While it would be convenient to measure the degree of modification directly, e.g. by using radioactive reagent, it is unfortunately, not possible to obtain 14C-labeled butanedione. The product is very unstable and rapidly polymerizes via a process thought to be initiated by the /3-emission from radioactive decay. Tritiated butanedione might be stable but there are no reports describing its preparation or use.

Some dicarbonyl compounds react with proteins to generate a chromophoric species that can be quantitated by spectrophotometry. In fact, the earliest study of the effect of butanedione on arginyl residues in proteins 33 was undertaken to examine the origin of the

pink color and green fluorescence that accompanies the Voges-Proskauer reaction, an early test developed to distinguish B. coli and B. aerogenes 34. HARDEN and NoRris demonstrated that the color arises from the reaction of arginine or arginyl residues with butanedione in the presence of strong alkali. Subsequently, this reaction became the basis for the well-known Sakaguchi method for measuring arginine 35. Color development requires not only high pH but also the addition of a-naphthol. Under the conditions described here for protein modification, there is no evidence for changes in either the visible or ultraviolet absorption spectrum.

B. Phenylglyoxal Phenylglyoxal, first introduced as an arginyl reagent by TAKAHASHI in 196812, reacts with these residues under mild conditions to give a product that contains two phenylglyoxal moieties per guanidino group (Scheme II). This product is relatively stable at acidic pH's and dissociates to regenerate arginine only on prolonged incubation in neutral and alkaline media. In addition, 14C-phenylglyoxal can be obtained commercially. Thus, the reagent offers the advantage of direct quantitation even with proteins containing many arginyl residues and the stability of the product ought to be helpful in identifying which arginyl residues in the primary structure are modified. On the other hand, the lack of reversibility can be a disadvantage when trying to establish the relationship between modification and effects on enzymatic activity. Moreover, phenylglyoxal appears to be somewhat less selective for

© © C= 0 H2N \ + H O - C - N H \ + I + C = N H R ~ I H~.C = NHR C= 0 H 2 N / H O - - C - - N I I

" //" ©

0 / O - C - - N H \ + - -C I C = NHR

I ~ ' O - C - N H / H I

H

S C H E M E II

arginine than butanedione and quantitation must be based on an assumed 2 : 1 stoichiometry which does not always pertain 1a'3°. Phenylglyoxal reacts rapidly with the a-amino group of peptides, probably via Schiff base formation followed by transamination, to give ct-keto acyl peptides 12. This is a potentially important side reaction that has not received major attention 36. Prolonged treatment of ribonuclease with 14C- phenylgiyoxal led to the incorporation of many more molecules of reagent than could be accounted for by the loss of arginine and it was suggested that Schiff base formation with lysine or an arginine stochiometry of greater than 2 to 1 might be a possible explanation 12.

A recent study illustrates a problem that can be encountered when trying to isolate peptides containing modified arginyl residues. Phenylglyoxal modifies two arginyl residues per subunit in horse liver alcohol dehydrogenase abolishing both enzymatic activity and NADH binding 29. Arginine modification also reduces the rate of alkylation of Cys-46 by iodoacetate leading to the conclusion 37 that inactivation of alcohol dehydrogenase is due to modification of Arg-47, i.e. the residue adjacent to the active site cysteine. In order to obtain direct evidence that this arginyl residue was indeed modified, enzyme labeled with 14C-phenylglyoxal was subjected to peptide analysis 38. Only a single radioactive peptide could be isolated from the chymotrypsin digest of the enzyme, in spite of two arginyl residues originally being modified per subunit. The label was located on Arg-84, not Arg-47. Failure to isolate a labeled peptide from the region around Arg-47 was attributed to the instability of the modified residue and it was pointed out that unless conditions are particularly favorable, labeled peptides will not survive fragmentation and isolation procedures 38. Lability of modified active center arginyl residues has been observed in other systems and it may be that features contributing to the unusual reactivity of such residues toward chemical reagents might also affect the stability of the reaction product.

An unusual reaction involving phenylglyoxal was observed recently in studies on rhodanese 39. This enzyme catalyzes the transfer of a sulfane sulfur atom from an anionic donor, such as thiosulfate, to a thiophilic acceptor, such as

77

cyanide. The reaction involves a stable, covalent sulfur-rhodanese intermediate. Treatment of sulfur-rhodanese with phenylglyoxal results in almost complete inactivation due to modification of Arg-186, likely a constituent of the substrate binding center. In contrast, phenylglyoxal inactivation of free rhodanese, which is generated from sulfur-rhodanese by addition of cyanide, is due to a novel reaction in which disulfide bonds are formed very rapidly between Cys-247 and either Cys-254 or Cys- 263. Only stoichiometric quantities of phenylglyoxal are required and there is an absolute dependence on the presence of cyanide. The mechanism of this reaction is unknown but is under investigation.

As with butanedione, the details of the reaction of phenylglyoxal with arginyl residues in proteins are not understood. Model studies with guanidine and N-acetylarginine have demonstrated a pH dependence similar to that seen with proteins, the rate increasing as pH increases from 7.0 to 11.540 . Further, the reaction is faster in bicarbonate buffer than in N-ethylmorpholine, borate, phosphate or Tris. The specific effect of bicarbonate is thought to be due to complex formation between bicarbonate and the guanidinium group, which lowers the pKa of and thus promotes nucleophilic attack by the guanidine group at the carbonyl carbon of phenylglyoxal.

C. Cyclohexanedione The characteristics of butanedione and phenylglyoxal just described make them suitably complementary for most structure-function studies of proteins. One is reversible but not easily quantitated, the other can be made radioactive but is not reversible. Yet a third reagent has been offered for modifying arginyl residues, 1,2-cyclohexanedione 41. Under the appropriate conditions, cyclohexanedione reacts selectively with arginyl residues and it combines reversibility with commercially available radioactivity. Optimal conditions are pH 9 in borate buffer, usually at temperatures from 30-40 °. A single product (Scheme III, I) is formed which is analogous to that proposed for the butanedione reaction and for the first step of the phenylglyoxal reaction. At higher pH values several other products are detected as well but at very high pH a diazospiro derivative

78

C=O •/•-C- NH\ ÷ + HzN; c® N+HR ~" I C-

HzN ~ - C I -NH / NHR ( I )

BORATE / ~ pH> 12

A l,.,- - "~'n "-'~G--NH\ O= C-NHx HO~ H O ~ . , B ~ I C= NHR [ C • NR _ C--NH/" C-NH"

t i t ) (ili}

SCHEME III

of ornithine (Scheme IIi, II) is the sole product. The presence of borate serves both to accelerate the reaction and to stabilize the product (Scheme III, III) as in the case of the butanedione reaction. In other buffers the reaction is less specific and reaction with lysine is observed. This side reaction is characterized by the formation of a yellow product that absorbs at 440 nm 42.

The principal advantage of cyclohexanedione is the stability of the dihydroxy adduct either in borate buffer or in acidic media. While it is destroyed by hydrolysis in 6 N HCL at 110 ° for 24 hours it is stable indefinitely in 30% acetic acid. This allows the selective cleavage of polypeptides by trypsin at lysyl residues and also the identification of specific arginyl residues modified by the reagent 43.

Removal of borate results in the spontaneous slow regeneration of arginine and the process can be accelerated by the addition of a strong nucleophile, such as hydroxylamine, which traps the liberated cyclohexanedione. Complete regeneration occurs within about 8 hours at 37 ° in the presence of 0.5 M NH2OH. The dioxime of cyclohexanedione reacts with nickel ions to give a red complex, which can be used as the basis for detecting peptides containing modified arginyl residues after paper chromatography or electrophoresis.

D. Other Dicarbonyl Compounds An interesting new reagent that has been suggested for arginyl modification is the 3-nitro- 4-hydroxy- derivative of phenylglyoxal introduced by BORDERS et al. 44. This reagent gives pH-dependent absorption spectra typical of nitrophenols. The spectrum of the related 4- methoxy- derivative is insensitive to pH

changes. Hence both offer the possibility of being chromophoric reporter groups if they can be selectively incorporated into proteins. The 4-hydroxy reagent has been shown to inactivate creatine kinase with a 1 : 1 stoichiometry. Importantly, the inactivation is not readily reversible adding further assurance that the chromophore will remain in place during structure-function studies.

GILBERT and O'LEARY 45 have investigated the use of the/3-diketone, 2,4-pentanedione, for modifying both arginine and lysine residues in proteins. The reaction with primary amines proceeds readily but is reversible at low pH or on treatment with hydroxylamine or other nucleophiles. The guanidinium group reacts much more slowly and complete modification of arginyl residues in proteins may require anywhere from 10 to 100 hours at pH 9.0 and 20 °. Lysines will also be modified during this period and it is necessary to incubate the product with 1.0 M hydroxylamine in order to achieve selective arginine modification. These somewhat unfavorable reaction conditions will likely prevent widespread acceptance of this or any other /3-diketone as an arginyl reagent, but one valuable characteristic should be emphasized. The product of the reaction is an N-substituted 2-amino-4,6-dimethylpyridine (Scheme IV) which has an absorption maximum at 300 nm (e 3.4 x 103 M -1 cm -1) and a pK of approximately 5. Protonation shifts the hmax to 310 nm. Quantitation can thus be accomplished by spectral analysis. The ease with which 2,4- pentanedione modifies lysine and particularly the reversibility of this reaction suggests that this may be the application of greatest practical utility for the reagent in studies of proteins. It

C - 0 IHO,,, / C H ~ I HzNx + ,C - -NH~ + ~ H 2 + / C = N H R ~ H z C " C = N H R

HzN [ IC,,-- NH ~' [ CI : 0 |HO CH 3

L CH 3

/ H,.C\

4,.C -- N % -I- H C ~ c _ N f C - NHR

/ H3C

SCHEME IV

should not be forgotten, however, that 2,4- pentanedione has been used for some time as a convenient agent for blocking guanidinium side chains in arginine-containing oligopeptides so that they can be volatilized for gas chromatography-mass spectrometric analysis 46. The pryimidylornithine nature of the reaction product was established from its mass spectrum. A similar product has been obtained on reacting guanidinium compounds with nitromalondialdehyde 16. In this case the derivative exhibits an absorption maximum near 335 nm with a molar absorptivity of about 16,000. An especially interesting feature of this reaction is the lack of susceptibility of the product to the hydrolytic action of trypsin. However, reduction with sodium borohydride gives the tetrahydropyrimidyl ornithine derivative which, remarkably, restores susceptibility to trypsin, in fact, to a level even greater than with arginine itself. This unique reversibility may provide novel opportunities for chemical modification of proteins.

III. The functional role of arginine residues in proteins

A. Enzyme Active Sites

1. Arginines and Carboxyl Binding Sites

Chemical Modification studies and X-ray crystallography have led to the recognition of a number of different amino acid side chains as participants in the interaction of enzymes with substrates 6. Cysteines, serines, tyrosines, histidines and lysines, as well as the carboxyl groups of glutamic and aspartic acid have all been identified as components of many active sites. With the advent of reagents specific for arginine side chains the importance of this residue, long inaccessible to modification, has become firmly established. In fact, more enzymes are known to contain arginine than to contain any other specific amino acid. This is because arginine generally serves as an anion binding site, particularly for phosphate and carboxylate groups, and more than two-thirds of all the known enzymes either act on anionic substrates or require anionic coenzymes. Hence, arginyl enzymes are ubiquitous.

Bovine carboxypeptidase was the first enzyme in which arginyl residues were shown to be involved in substrate binding and specificity 11'2°.

79

Modification with butanedione in the presence of borate reduces its peptidase activity to between 10 and 20% of that of the native enzyme concomitant with the loss of approximately two arginyl residues. On dialysis or gel filtration to remove both excess reagent and borate, activity is recovered but only a single arginyl residue reappears. In the mechanism of carboxypeptidase action, one arginyl residue, Arg-145, has been revealed by X-ray crystaUography 47 to bind the terminal carboxyl group of the pseudosubstrate, Gly-L- Tyr, thus accounting for the specificity of the enzyme. Other arginyl residues, Arg-127 and Arg-71, have been located within the peptide substrate binding groove and ab initio molecular orbital calculations have indicated that these residues may serve as initial binding sites for the terminal carboxyl group of the substrate, sliding it smoothly to its final complex with Arg-14548. Converting any one of these residues to a butanedione-borate derivative would introduce a negative charge and likely prevent substrate binding. Stopped-flow fluorescent studies indeed demonstrate that loss of peptidase activity is accompanied by loss of peptide binding 2°.

However, while the peptidase activity of carboxypeptidase A is abolished by arginine modification, its esterase activity is not. In the absence of borate there is no change in esterase activity and with borate added esterase activity actually increases to about 300% of the control. This is clear evidence that the ester substrate does not bind to the same carboxyl recognition site as the peptide and is entirely consistent with observations that the metal atom at the active center of carboxypeptidase functions primarily in the catalytic step of peptide hydrolysis but primarily in the binding step of ester hydrolysis 49. This would be expected if ester binding involved a metal-carboxyl interaction rather than arginine binding.

Similar results have been observed with the closely related enzyme, carboxypeptidase B 5°'51. Using phenylglyoxal, it was found that peptidase activity is markedly decreased as arginine residues are modified, as is activity toward the ester hippurylphenyllacetate. However, activity toward the basic ester, hippurylargininic acid, is slightly increased because the productive binding mode of the peptide substrate is abolished while that of the basic ester is

80

preserved. In other words, the peptides and the basic ester have different productive binding modes. It is of interest that three peptides containing a modified arginyl residue were isolated from 14C-phenylglyoxal-treated carboxypeptidase B. Two of the peptides probably came from the same sequence of the protein but none of them correspond to the region containing Arg-145 (using the numbering sequence for carboxypeptidase A). However, approximately half of the radioactive label was lost from the peptide mixture during the isolation process, and it is not clear if the loss occurred by conversion of the 2 : 1 phenylglyoxal-arginine adduct to a 1 : 1 adduct or by complete regeneration of a modified residue to free arginine. A 50% loss of radioactivity has also been observed with phenylglyoxal-modified D-serine dehydratase 27 and the loss was independent of the presence of borate during treatment with the reagent.

Two other carboxypeptidases have been shown to contain functional arginyl residues. Yeast carboxypeptidase C loses peptidase activity on reaction of a single arginyl residue with phenylglyoxal but its esterase activity is unchanged s2. The ester substrate used with this enzyme is N-acetyl-L-tyrosine ethyl ester which lacks the free terminal carboxyl group of typical peptide substrates. It is probably hydrolyzed by a mechanism that shares certain features with that employed for peptides but, as in the case of the metallocarboxypeptidases, the two types of substrates must bind in different regions of the active site.

The angiotensin converting enzyme is a dipeptidylcarboxypeptidase that requires zinc for catalytic function. It acts on peptide bonds penultimate to the carboxyl terminus of polypeptides to release dipeptides. Its peptidase activity is also drastically reduced on reaction with butanedione and the effect is promoted by borate buffer 53. A suitable ester substrate has not been examined in this case but the enzyme does exhibit many of the characteristics that seem to be common to most metaUoproteases. It contains essential carboxyl and tyrosyl residues in addition to arginine and the active site zinc atom 53, and is therefore similar to carboxypeptidase A and B. It also contains a critical lysyl residue that is not present in the other carboxypeptidase but the function of this

residue has yet to be established. Moreover, the converting enzyme is similar to another zinc protease, thermolysin, which together with carboxypeptidase A is thought to represent an example of convergent evolution. It may be that all zinc exo and endoproteases share common catalytic mechanisms and this may extend to the serine carboxypeptidases as well.

Arginyl residues serve as substrate sites for other enzymes that act on amino acid substrates, providing positive loci for interaction with carboxyl groups. For example, the inactivation of D-aspartate oxidase by butanedione or phenylglyoxa154 clearly indicates that arginyl residues bind the substrate to the active site. Also the asparatate aminotransferases from pig and chicken heart supernate and beef kidney mitochondria are all inactivated by arginyl reagents 5s-57. These enzymes catalyze reactions involving the dicarboxylic acids asparatate, a-ketoglutarate, oxaloacetate and glutamate. In addition, they employ pyridoxal phosphate as a cofactor. It has been proposed 58 that there are three different positively charged binding sites at the active center of these enzymes, one each for the two carboxyl groups of the substrate and one for the phosphate group of the coenzyme, pyridoxal phosphate. In all three enzymes, loss of activity is prevented by substrates and inhibitors. The modified enzymes still bind pyridoxal phosphate and modification of apoenzyme does not prevent reconstitution to holoenzyme. Thus, it appears that the phosphate group of pyridoxal

cis -ACONITATE

/7 , ,

0 " " C \ .,.H / I + C = N

H c /C~ 'coo- . . .H . -NH

-OOC / z H,,S

.J.

11 D - I S O C I T R A T E

phosphate does not interact with an arginyl group of the enzyme. Loss of activity is likely due to blocking an arginine that interacts with the ~o-carboxyl group of the substrate. This is substantiated by the fact that the modified enzyme can still bind and undergo a transamination half-reaction with alanine without impairment 56'57.

Substrate carboxyl groups interact with enzyme arginyl residues in porphobilinogen deaminase 59, cystathionase 6°, and aconitase 61. In the last case a single arginine reacts with butanedione in borate and the kinetics of inactivation make it clear that the effect of the buffer is to stabilize the enzyme-butanedione complex. The functional arginine is believed to serve as more than just a positive charge. It can provide a side arm for rotation. If two adjacent substrate carboxyls bind to the same guanidinium group, the conversion of citrate to isocitrate would only require the 180 ° rotation around an axis that is centered through the alkylated nitrogen and central carbon of that guanidinium group (Scheme V).

Isocitrate is also a substrate for the NADP- dependent isocitrate dehydrogenase from pig heart. This enzyme has been modified with butanedione, the reaction being carried out in MES buffer at pH 6.562. An unusual feature of this reaction is the lack of a specific borate effect and the failure of the modified enzyme to recover activity on dialysis even though amino acid analysis indicates the only residue modified is arginine. It may be that the low pH of the

% ,coo-. . . , - , , \

-ooc / o . . . c + H /C,,, _ /

H " COO . . -H -NH

H\ S

C I T R A T E

H C = N"

SCHEME V

8l

reaction promotes a rapid rearrangement of the intermediate glycol (Scheme I), if it even forms at all under these conditions.

Butanedione inactivates the enzyme and the loss in activity is largely prevented by the prior addition of isocitrate. Remarkably, the rate of inactivation is increased almost 5-fold by addition of coenzyme, NADP. Hence, the critical arginine(s) susceptible to butanedione lie within the isocitrate binding site and not the nucleotide binding site.

The NAD-dependent isocitrate dehydrogenase from pig heart is also inactivated by butanedione under acidic conditions 63. There is slight protection by NAD, rather than activation, and by isocitrate, but if Mn 2+ is added the protection by isocitrate is increased dramatically. This enzyme also exhibits allosteric activation by ADP, an effect that is abolished by arginine modification. Complete protection against both modification effects is given by isocitrate, Mn 2+ and ADP. Thus, in this case, arginine residues appear to be important for both the catalytic function and the allosteric activation of isocitric dehydrogenase.

Fructose-l,6-diphosphatase is another allosteric enzyme whose activity is altered by chemical modification with butanedione 64'6s. Although this enzyme does not interact with substrate carboxyl groups, its catalytic site binds fructose-l,6-diphosphate, its regulatory site is senstive to the negative ettector, AMP, and it has a site for binding activating monovalent cations. If the enzyme from rabbit muscle is assayed in the presence of AMP it exhibits only a small fraction (about 3%) of the toal activity obtained in the absence of AMP. Monitoring the time course of butanedione modification by assaying with and without AMP reveals that arginyl residues are present at both the AMP and the substrate binding sites 64. The presence of substrate prevents abolition of activity but does not prevent elimination of AMP sensitivity. On the other hand, the presence of AMP protects against modification of arginyl residues at the regulatory site. In addition, modification of the enzyme from pig kidney causes it to lose its sensitivity to activation by monovalent cations 66. It is unlikely that arginyl residues would be directly involved in cation binding, but obviously the phenomenon of cation activation is complex.

2. Arginines and Nucleotide Binding Sites. For many years, the identity-of the binding sites for NADH in dehydrogenases remained essentially unknown. However, the availability of reagents for arginine modification has changed all that and in many instances it is now clear that arginines are likely common to the majority of such sites. For example, the alcohol dehydrogenases from yeast and from horse and human liver all have arginines at their coenzyme binding sites 29. Modification of the horse liver enzyme with butanedione or phenylglyoxal abolishes enzymatic activity concomitant with loss of coenzyme binding and with modification of two arginyl residues per subunit. Both arginyl residues need not be involved in coenzyme binding, however, and it is now known that one of them, Arg-84, is in a region of the protein that is particularly reactive but remote from the active site 39. The other is almost surely Arg-47, which has been shown by X-ray analysis to interact with the pyrophosphate group of ADP- ribose when bound to the active site 67.

Studies of the alkylation of Cys-46 with iodoacetate provide indirect evidence for the reaction of Arg-47 with butanedione 37. The speed and selectivity of carboxymethylation with iodoacetate compared to iodoacetamide plus the saturation kinetics of the reaction led to the suggestion that the carboxyl group of the alkylating agent binds to a positive charge in the vicinity of Cys-46. Arginine modification markedly reduces the overall rate of carboxymethylation and prevents alkylation of Cys-46. X-ray analysis of carboxymethylated enzyme crystals shows that Cys-46, an active site zinc ligand in the native enzyme, is still coordinated to the metal through the sulfur atom 68. It is not known why a labeled Arg-47 peptide could not be isolated from 14C- phenylglyoxal-treated alcohol dehydrogenase but the lability of active site arginine derivatives has been observed in several instances (vide supra).

The inhibition of mitochondrial malate dehydrogenase by butanedione correlates with the modification of two arginyl residues per mole or one per active center 69. Similarly, cytoplasmic malic dehydrogenase is also inactivated by this reagent 7°. In both cases fluorescence titrations indicate that modification has little effect on NADH binding but that

82

formation of the enzyme-NADH- hydroxymalonate ternary complex is markedly decreased. Studies on the effect of butanedione on lactate dehydrogenase 71 gave essentially the same results i.e. the partially inactivated enzyme retained most of its NADH binding but the resulting binary complex bound inhibitors and pyruvate much less than the native enzyme- NADH complex. The conclusion is that these enzymes all utilize an active site arginyl residue to bind the carboxylate group of the substrate to the enzyme-coenzyme complex.

Another dehydrogenase with functional arginyl residues is yeast D-glyceraldehyde-3- phosphate dehydrogenase 72. It has also been inactivated with butanedione while retaining its capacity for interacting with coenzyme. In this case binding is reduced by about an order of magnitude and the molar' adsorptivity of the binary complex is markedly reduced, a signal of a possible conformational change. Although ternary complexes were not examined, loss of activity is likely due to loss of substrate binding. Somewhat analogous results were obtained with rat muscle glyceraldehyde-3-phosphate dehydrogenase which was reversibly inactivated by treatment with butanedione due to modification of two arginines per subunit 73. Inactivation was prevented by addition of inorganic phosphate suggesting that the arginine(s) reacts with the phosphate moiety of the substrate. Surprisingly, the rate of inactivation was enhanced by the addition of NAD contrary to its protective effect with the yeast enzyme. This is analogous to results obtained with the NADP-dependent isocitric dehydrogenase 62.

Arginine residues have been found to be involved in coenzyme binding to the glutamic dehydrogenases from Neurospora and bovine liver TM. However, the modified liver enzyme still binds substrates and coenzymes with no change in Michaelis constants 75. The ability of the liver enzyme to be activated by ADP is abolished by reaction with butanedione, an effect which is prevented by the presence of ADP. Hence, arginyl residues are involved in both the regulation and catalytic function of liver glutamic dehydrogenase and likely act by providing a positively charged environment. Similar results were obtained with the NAD- specific isocitrate dehydrogenase 63. The effector

molecules ADP and GTP also protect two arginines in beef heart glutamic dehydrogenase 76 from modification with hutanedione. In this case the reaction is reversible on removing borate. It is suggested that one or two arginines are involved in coenzyme binding and may also serve as a positive charge to direct the alkylation of a critical sulfhydryl group.

Glucose-6-phosphate dehydrogenase from L. mesenteroides is rapidly inactivated by butanedione; substrates and coenzymes, NAD as well as NADP, protect the enzyme 77. Although borate stimulates the rate of inactivation, the reaction is not reversible by dialysis. Since borate can react with NAD, as was pointed out by LANGE et al., 29, these studies were carried out in barbital buffer as well, with essentially the same results. The modified enzyme still binds coenzyme, NADPH somewhat more tightly and NAD somewhat less tightly than does the native enzyme. Although an unequivocal assignment could not be made, it was suggested that one of the two arginines modified per subunit binds the phosphate of glucose-6-phosphate. The implication was that the other one is not involved in coenzyme binding but may control the active site conformation. Arginine does seem to be involved in binding coenzyme to aldehyde reductase 78 and so it appears that while all dehydrogenases studied have functional arginyl residues, the precise role of these residues varies from protein to protein.

ROSSMAN and coworkers have pointed to the conservation of the basic nucleotide-binding domain among various dehydrogenases including alcohol, malate, lacate and glyceraldehyde-3-phosphate dehydrogenases 79. Thus, it is not surprising that these enzymes should behave similarly toward modification with butanedione. However, the specific functions of the active site arginyl residues vary from direct to partial to almost no participation in coenzyme binding, on the one hand, to just the opposite in terms of substrate, on the other. Arg-47 of alcohol dehydrogenase has been shown to interact with the pyrophosphate bridge of the ADP-ribose moiety of the coenzyme 67'68. Three arginyl residues interact with the NAD- pyruvate complex in lactate dehydrogenaseS°; Arg-101 with thq pyrophosphate bridge of the

83

coenzyme, and Arg-109 and -171 with pyruvate. In forming the ternary complex the guanidinium group of Arg 101 moves 13 from its position in the apoenzyme while that of Arg-109 moves 23 A. It should be noted that modification of arginine with dicarbonyls does not necessarily abolish the positive charge 41. Hence, unless a complex with borate is formed salt bridges to the coenzyme could still be possible. The negative charge introduced by the borate would act as a repellant. In most of-the instances cited, however, the lack of reversibility on removing borate suggests that modification has not led to the formation of a dihydroxyimidazoline derivative. Thus, until all of the X-ray analyses of dehydrogenases are complete the role of arginyl residues in binding coenzymes will be uncertain.

The kinases constitute another class of enzymes that interact with anionic coenzymes, in this case ATP. As with the pyridine nucleotide dependent dehydrogenases, the identity of the coenzyme binding sites of kinases remained unknown prior to the introduction of ar'ginyl reagents. Now it has been shown, for example that modification of a single, very reactive arginyl residue in creatine kinase virtually abolishes MgATP or MgADP binding concomitant with loss of enzymatic ac- tivity 8~. Similarly, hexokinase 82, yeast phos- phoglycerate kinase s3-85, pyruvate kinase s6 and adenylate kinase s7 are all inactivated by arginyl reagents.

Both glutamine synthetase and carbamyl phosphate synthetase utilize ATP in synthetic reactions and both lose activity when treated with phenylglyoxa188. ATP but not other substrates protects against inactivation indicating that arginyl residues are involved in ATP binding. They are also involved in binding dUMP to thymidylate synthetase 89 and ATP to a regulatory site in lysyl-t-RNA synthetase 9°.

3. Arginine and Phosphate Binding Sites

A number of ATPases have been shown to contain functional arginines at their active sites. The ATPases from mitochondria 91 and the (Na++ K+)-ATPase from rabbit kidney outer medulla 92 are inactivated by butanedione due to modification of arginines involved in nucleotide binding. One type of arginine is necessary for

ATP hydrolysis and another for ATP-Pi exchange in Complex V and the activity of the soluble F1-ATPase is abolished by phenylglyoxa193.

Each of the various classes of nucleotide polymerases has been reported to be inactivated by arginine modification. The DNA-dependent DNA-polymerase 194 and DNA-dependent RNA-polymerase 9s of E. coli, and the RNA- dependent DNA-polymerase 0ti avian myeloblastosis virus 96 are all arginyl enzymes. The DNA-photoreactivating enzyme of yeast is inactivated with phenylglyoxal 9v as is the purine nucleoside phosphorylase from calf spleen or human erythrocytes 9s. No details of the modes of inhibition are available but it seems likely that arginines participate in substrate binding by interacting with phosphate moieties.

There is a host of other enzymes that are inactivated by dicarbonyl compounds and all of these enzymes act on phosphorylated substrates. All but one of the enzymes of the glycolytic pathway are inactivated by butanedione- borate 64. Inorganic pyrophosphatase 99, alkaline phosphatase from E. coli 100 or pig kidney lm, prostatic acid phosphatase 1°2, phosphoglycerate mutase 1°~-1°5, enolase lO6 and aldolase 1°7't°8 are all arginyl enzymes. There is an arginyl residue at the carbamyl phosphate binding site of as- partate transcarbamylase m9 and others that are involved in its allosteric regulatory proper- ties n°. An arginine interacts with the phos- phate group of pyridoxal phosphate in D-serine dehydratase 27, and with the phosphorylated substrates of ornithine transcarbamylase m , phospholipase C n2, stearylcoenzyme A de- saturase 113 and prenyl transferase t~4. Both the catalytic and regulatory functions of E. coli phosphoenolpyruvate carboxylase are affected by butanedione modification us. Ribulosebis- phosphate carboxylase from tobacco n6 and spinach 117, propionyl coenzyme A carboxyl- ase ~8, /3-methylcrotonyl coenzyme A carboxyl- ase 118 and glutamate decarboxylase ~19 all re- spond to one or another dicarbonyl compound. In some of these cases the evidence that ar- ginyl residues are involved in function is cir- cumstantial i.e. loss of activity on treatment with the dicarbonyl. But it is abundantly clear that there is a very large number of enzymes that respond to these agents and the initial supposition 2° that arginyl residues might serve

84

a general function as anion recognition sites in enzymes would seem to be valid.

B. Functional arginyl residues in other proteins One class of non-enzyme proteins that has been studied rather extensively in terms of arginyl residues is the trypsin inhibitors. These inhibitors are classified as either lysine- or arginine-dependent according to the effect of specific chemical modification reagents on their biological activity. Thus, for example, soybean inhibitor and chicken ovomucoid are of the arginine type while lima bean inhibitor, turkey ovomucoid and human aa-antitrypsin are of the lysine type 25'42'~2°. It is of interest to note that these inhibitors have been particularly useful in defining the specificity of some of the dicarbonyls that have been employed for arginine modification. Both cyclohexanedione and phenylglyoxal have been found to react with lysyl residues 121'~22 though with the former this effect can be eliminated by using borate buffer zS.

Antibody molecules have been examined for the presence of functional arginine residues. Antibodies directed against negatively charged haptens such as p-azobenzoate, p- azobenzenearsonate and succinamate lose their specific binding activity on arginine modification whereas those directed against a positively charged hapten are unaffected ~7. Antibodies directed against the p-azobenzenearsonate hapten could actually be separated into two different populations, those containing lysyl residues at the combining site and those with arginine TM.

While antibody molecules exhibit an extremely high degree of specificity and bind only a single type of determinant, serum albumin exhibits an extremely broad specificity and can bind a very large number of different substances. In this way albumin can serve as a general purpose transport protein and there is much interest in the nature of these various ligand binding sites. Some of these sites have been shown to contain positive charges ~22 and modification with cyclohexanedione has demonstrated that the binding of diazepam but not of bilirubin requires an arginyl residue ~23.

Biologically active oligopeptides can usually be investigated for structure-activity relationships by direct synthesis. However,

chemical modifications have been particularly helpful in deciphering the critical features of parathyroid hormone. Since the amino terminal 34 residue sequence of the 84 residue hormone exhibits full biological activity it is well suited for such modification studies. Reaction of this peptide with cyclohexanedione completely blocks the two arginyl residues at positions 20 and 25 and reduces biological activity to, at most, 16% of the control TM. Ovine pituitary lutropin is also inactivated by treatment with cyclohexanedione 12s. In both instances the modification and loss of activity are reversed by treatment with hydroxylamine.

Blocking Arg-13 of cytochrome C reduces its ability to transfer electrons to cytochrome C oxidase 126 while modification of Arg-95 abolishes the ability of F-actin to interact with tropomyosin 127. The binding of the protein core of purified proteoglycan to hyaluronidase requires intact arginine residues ~28 and the conformation of parvalbumin III is dependent on the single arginine that is present in every parvalbumin studied thus far 129. Arginines are essential for photophosphorylation by chloroplasts 13° and have been shown to be essential for the action of chloroplast coupling factor TM. The bicarbonate binding site of transferrin seems to contain one critical arginine 132 and the same is true for the nucleotide binding site of elongation factor G from E. coll.

One additional application of arginyl reagents has been to study their effects on the growth of HeLa cells TM. The experiments were based on the observation that there is an increase in arginine at cell surfaces during the division phases of the cell cycle, and on the supposition that these positive charges might control growth. Phenylglyoxal inhibits HeLa cell growth specifically in the S, G2 and M phases of the cell cycle. It may be that arginine-rich proteins on the cell surface are modified by this treatment but the specificity of the reagent does not permit such unequivocal deductions.

C. Functional arginyl residues in proteins- a summary It is obvious that there are numerous reports of proteins that contain functional arginyl residues. The above survey, while extensive, is certainly not complete. By now more than a hundred

z 85

different arginyl proteins have been recognized and the list continues to grow almost exponentially. In fact, approximately 80% of the arginyl proteins have been identified in the past four years, with 28 publications appearing in 1978 alone. Each category of the IUB system of enzymes has at least one example of an arginyl enzyme. There are probably more enzymes that utilize arginyl residues as active site components than those using any other specific amino acid residue. Based on the approximation that two thirds of all enzymes act on anionic substrates of coenzymes it would appear that at least that many of them will turn out to be arginyl enzymes. No other class of reagents has provided more information about the active sites of enzymes than the dicarbonyl compounds.

III. Reactivity of arginyl residues In virtually all of the above examples the chemical modification reaction is not only limited to arginyl residues but to a selected few of the total arginines present per molecule (Table II). A good example of this selective modification is found with creatine kinase 3°. Only one of the 18 arginyl residues per subunit reacts with phenylglyoxal. Clearly, in the native enzyme this particular arginine possesses some feature which distinguishes it from the others in terms of reactivity. It reacts 10-15 times faster than the free amino acid. Conversely, most or all of the 17 arginines that do not get modified must react at a rate significantly slower than free arginine. The selectivity, therefore, arises from both an increased reactivity of the

TABLE I I

Selective Modif icat ion of Ar~inyl Residues

Arg Total Enzyme Reagent Modified Arg*

Creatine kinase 30 Phenylglyoxal 1 18

AldQlase I08 Phenylglyoxal l 14

Malate dehydrogenase 70 Butanedione l 8

A s p ~ t } o ~ . . . . . . . bamy- Phenylglyoxal l 24

G1utamic dehydrogenase 74 Cyclohexanedione l 17

Phosphoglycerate mutase I03 Butanedione l 16

Adenylate kindle B7 Phenylglyoxal 1 I I

Aspartate aminotrans- ferase 55 Phenylglyoxal 2 26

Alcohol dehydrogenase 2g Butanedione 2 I I

Carboxypeptidase 20 But~nedione 2 Ig

*Per monomeric unit

essential arginine and a decreased reactivity of most or all of the others. Similar enhancements of reactivity have been noted in other instances as well. The rate at which phenylglyoxal reacts with the essential arginyl residue in aspartate transcarbamylase 1°9 is about 3 to 4 times faster than the corresponding rates in glutamine synthetase 8s, alcohol dehydrogenase 29, and aldolase 1°8 and substantially faster than the rates for the other arginines in the same enzyme. Cyclohexanedione reacts with arginyl residues in glutamic dehydrogenase much more rapidly than with free arginine or even with the most reactive residue in ribonuclease 74. Several factors can be involved in determining the reaction selectivity. Presumably, most of the highly polar arginyl groups would be exposed on the surface of the proteins. A few of these may be involved in polar interactions with other amino acid side chains that would preclude modification. Others will be solvated and their reaction might be slow. Some, however, will be in a chemical environment that enhances their reactivity and that may be essential to their biological function. The unique juxtaposition of residues in the active site that is critical to catalytic function could well give rise to enhanced chemical reactivity of the constituent side chains 6.

Selective binding of reagent may also contribute to reaction specificity. The dependence of the rate of inactivation of horse liver alcohol dehydrogenase on butanedione concentration 29 suggests that the reagent binds reversibly to the enzyme to form a Michaelis- type complex, with an apparent binding constant of about 14 mM. This saturation kinetics behavior has also been observed in the reaction of butanedione with phosphoglycerate kinase 83. It has been suggested 2s that the specificity-enhancing effect of borate on the cyclohexanedione reaction is due to the formation of a reagent-borate complex that is more readily directed to the reactive arginine. Kinetic evidence does not favor this view, however. Nevertheless, in specific instances such active site binding could well account for unusual reactivity.

IV. Arginyl residues as anionic binding sites: genetic implications We return now to the point made in the

86

introduction that the functions subserved by all the various amino acid residues in proteins are not known. From the above discussion, however, it should be clear that arginyl residues are well suited to serve as anionic binding sites in proteins. Moreover, it appears that the incidence of these residues in proteins has been restricted by evolutionary pressures in order that they might be better able to carry out this function. In this regard it might be helpful to examine some current views on the role of evolution in determining the amino acid (particularly arginine) composition of proteins.

One of the pieces of evidence cited in support of the non-Darwinian model of evolution of species is the relationship between the occurrence of amino acids in proteins and their numbers of synonymous codons 135. This hypothesis predicts that a significant proportion of the amino acids present in proteins has arisen by random neutral mutation and drift rather than by natural selection. A plot of the frequency of occurrence of amino acids observed in 53 vertebrate polypeptides versus the frequencies predicted by the genetic code reveals a good correlation for most amino acids. The one conspicuous exception is arginine which is only present to the extent of 4.2% as compared with an anticipated frequency of 10.7%. A more recent tabulation 136 of 208 proteins from various species indicated an average arginine content of 4.39%. A number of theories have been proposed to account for this obvious selection against arginine. JUKES 137

has suggested that arginine replaced ornithine in protein synthesis during evolution of the genetic code, i.e. that arginine entered the code as an intruder. This replacement might have occurred if arginine had a greater affinity for the ornithine tRNA-aminoacyl ligase system than ornithine itself. The intrusion could have conveyed certain special advantages but, at the same time it must have had an even greater number of disadvantages that were subsequently offset by negative selection for arginine and positive selection for lysine, i.e. homo-ornithine. Alternat ively, HOLMQUIST 138 has suggested that selection against arginine might be a result of it becoming obsolescent rather than it being an intruder.

WALLIS 139 has rejected the intruder hypothesis, arguing that the arginine/lysine ratio

of proteins does not correlate with their rates of evolution as the hypothesis might predict. Only a limited number of proteins was examined from this point of view, however, and, besides, such a correlation would pertain only if evolution proceeded in the absence of functional or structural constraints, an unlikely possibility. Thus, the arginine/lysine ratio in a given protein will depend on the degree to which the presence of these amino acids is required or tolerated in the structure and specific function of that protein 14°.

WALLIS also contends that the intruder hypothesis appears to assume that arginine subserved an important function other than in protein synthesis and because of this its intrusive entry into proteins was tolerated. Therefore, he finds the lack of a important role for arginine in prokaryotes, other than in protein synthesis, to be in conflict with, and hence an obstacle to the acceptance of, the intruder hypothesis. However, as JUKES has noted 14°, such an important role may have long since become outmoded and, therefore eliminated through evolution from currently existing prokaryotes.

Even if this function were to be found, WALUS still believes that the capture by arginine of all six codons from ornithine would have been an unlikely and remarkable event. Instead he has proposed several alternative explanations for the relative scarcity of arginine, all of which are based on special properties of the amino acid. Thus, he points out that arginine has been recognized as an important site for the action of specific proteases involved in physiological protein turnover. Restricting the arginine content of a protein by selection pressure would limit and to some extent control the degradation process. It might be expected that under these circumstances there would be a direct correlation between turnover rates and arginine contents. Such a correlation has not been reported but DICE and GOLDBER6141 have found a highly significant correlation between the in vivo degradative rates of 22 rat liver proteins and their isoelectric points (F = 0.224; P < 0.01), those with the lowest pI values turning over most rapidly. This would seem to suggest an inverse relationship between arginine content and protein turnover, on the assumption that proteins with low pI values would have

87

i50

n- I00

3' I-

50

• • t

I i I I 1 0 i a

2 4 6

ARGININE CONTENT, M O [ . % ,I

Fig. 4. The relationship between protein half-lives and ar- ginine content for 15 proteins. Data taken from DICE and GOLDBERG TM and from REECK 136.

correspondingly low arginine contents• This assumption is probably incorrect, however. In fact, using data compiled by R E E C K 136 together with those of DICE and GOLDBERG 141 it appears that proteins turning over rapidly do have relatively high arginine contents (Figure 4). Insufficient data are available to substantiate this hypothesis but it certainly merits additional investigation.

Another possibility suggested by WALLIS is that the arginine codons serve a role as "modulating triplets" in the control of translation. He points out that there are many instances of the use of CGN codons for arginine but no definitive evidence for the use of AGG or AGA. On the other hand, the doublet CpG, which is present in four of the six arginine codons, is relatively scarce in the DNA of certain organisms ~42 while AGR codons, the remaining two that specify arginine, have been established in viral RNA sequences 143. Moreover, their presence has been inferred from mutations involving E. coli and tobacco mosaic virus proteins. Thus, it is likely that all six arginine codons are utilized in protein synthesis and the view that certain of them are restricted to a regulatory role remains

88

speculative. A third suggestion by WALLIS is that the

presence of arginine in proteins has been suppressed because of the nature of the guanidinium group which, having a pK of 12.5, is always positively charged under physiological conditions. As a consequence, it would be perhaps the most difficult side chain group to be accommodated within the interior of a protein. If protein evolution involved an increase in size, more residues per molecule would have to be internalized and arginine would, therefore, be rejected•

In order to verify this hypothesis it would be necessary to compare the arginine content of proteins with molecular weight, particularly of proteins whose three-dimensional structure is known so that the arginyl residues can be located properly. In general, however, it appears that the arginine content of proteins, determined as mol%, tends to increase rather than decrease with increasing molecular weight (Figure 5). This may, in fact, be related to the observation 141 that protein turnover in vivo correlates with subunit molecular weight but

? o

x 150

O3 z o I.- .J ,¢[

~_" I 0 0

J O =E } -

Z

m 50

03 • o l •

OLD • I

1 • D 0 Q •

I I I I . I I 2 4 6

A R G I N I N E C O N T E N T , M O L %

Fig. 5. The relationship between subuni t molecular weight and arginine content for 32 proteins ( l l = proteins with more than 4 subunits). Data on molecular weights are taken from DARNALL and KLOTZ 146.

does not support the view that large proteins cannot accommodate many arginyl residues.

It may well be that genetic pressure has been applied to restrict arginyl residues to certain important biological functions and one such function, as we have shown, is to interact with anions, particularly phosphorylated metabolites. Studies have revealed that the guanidinium group is ideally suited for interaction with phosphate-containing substances by virtue of its planar structure and its ability to form multiple hydrogen bonds with the phosphate moiety 14. Lysyl residues might also have been selected to serve this function and, indeed, are known to be important in a number of enzymes acting on phosphorylated substrates. However, because of resonance stabilization, the guanidinium group is a poor proton donor (pKa> 12) and, hence, would probably not function as a general acid catalyst for the hydrolysis of the phosphorylated intermediates. Recent model studies T M have indeed shown that complexation between guanidinium ions and p-nitrophenyl phosphate actually lowers the rate of ester hydrolysis, likely by reducing the P-O bond order. This would lead to an increased activation energy for the formation of the metaphosphate transition state during ester hydrolysis. Moreover, CNDO/2 molecular orbital calculations of the energy states of phosphate and triphosphate analogs of ATP suggest that formation of a charge transfer complex with arginine (or lysine) lowers their energy state 145. One consequence of this, for example, would be to further ensure maximum utilization of substrate phosphate for the synthesis of ATP and thereby contribute to the overall efficiency of glycolysis. Thus, the phosphorylation of glycolytic metabolites con- fers an extra degree of specificity in the proper interaction between the metabolites and en- zyme arginyl residues. The selection of arginyl residues for this function minimizes nonspecific hydrolysis and optimizes the metabolic process. Importantly, arginyl residues could assist in the hydrolysis of phosphate esters if the mechanism involves nucleophilic attack rather than the for- mation of a metaphosphate intermediate 144.

In the energetics of living systems, ATP occupies a critical position. It is not known at what stage of evolution this unique conductor of chemical energy first emerged. If it was prior to the acquisition of a macromolecular coding

system, then evolutionary pressure could have restricted the role of arginine in proteins to anion, particularly phosphate, binding. Optimal binding specificity could then be achieved by limiting the number of arginyl residues in proteins. If the ATP system emerged subsequent to the coding system, the code might have required amendment to include arginine, perhaps by adopting the transfer RNA of some other amino acid, for example, ornithine 137, that is less suitable for binding phosphate. Late appearance of arginine codons could then account for the relative scarcity of this residue in proteins. Although present investigations do not favor either alternative, they clearly emphasize an important generality, that arginyl residues constitute positively charged binding sites for enzymes acting on anionic substrates and for all proteins that interact with anions.

References

1. Herriott, R., 1947. Advances in Protein Chem. 3, 169-225.

2. Olcott, H. S. and Fraenkel-Conrat, H., 1947. Chem. Rev. 41, 151-197.

3. Boyer, P. D., 1959 in The Enzymes (Boyer, P. D., Lardy, H. A, and Myrback, K, Eds.) 2nd Ed. pp. 511-588. Academic Press, New York.

4. Hartley, B. S., 1960. Annu. Rev. Biochem. 29, 45-72. 5. Riordan, J. F. and Sokolovsky, M., 1971. Accounts

Chem. Res. 4, 353-360. 6. Vallee, B. L. and Riordan, J. F., 1969. Annu. Rev.

Biochem. 38, 733-794. 7. Itano, H. A. and Gottlieb, A. J., 1963. Biochem.

Biophys. Res. Commun. 12, 405-408. 8. King, T. P., 1966. Biochemistry 5, 3454-3459. 9. Yankeelov, J. A., Jr., Kochert, M., Page, J. and

Westphal, A., 1966 Fed. Proc. 25, 590. 10. Yankeelov, J. A., Jr., Mitchell, C. D., and Crawford,

T. H., 1968. J. Am. Chem. Soc. 90, 1665-1666. 11. Vallee, B. L. and Riordan, J. F., 1968. Brookhaven

Syrup. Biol. 21, 91-119. 12. Takahashi, K., 1968. J. Biol. Chem. 243, 6171-6179. 13. Gund, P., 1972. J. Chem. Ed. 49, 100-103. 14. Cotton, F. A., Day, V. W., Hazen, E. E., Jr., and

Larsen, S., 1973. J. Am. Chem. Soc. 95, 4834-4840. 15. Toi, K., Bynum, E., Norris, E. and Itano, H. A., 1967.

J. Biol. Chem, 242, 1036-1043. 16. Signor, A., Bonora, G. M., Biondi, L., Nisato, D.,

Marzotto, A. and Scoffone, E., 1971. Biochemistry 10, 2748-2752.

17. Grossberg, A. L. and Pressman, D., 1968. Biochemistry 7, 272-279.

18. Yankeelov, J. A., Jr., 1970. Biochemistry 9, 2433- 2439.

89

19. Riordan, J. F , 1970. Fed. Proc. 29, 462. 20. Riordan, J. F., 1973. Biochemistry 12, 3915-3923. 21. Cornforth, J. W. and Huang, H. T., 1948. J. Chem.

Soc. 731-735. 22. Diels, O. and Schleich, K., 1916. Ber. 49, 1711-1721. 23. Pauly, H. and Sauter, H. 1930. Ber. 63, 2063-2069. 24. Roy, G. L., Laferriere, A. L. and Edwards, J. O.

1957. J. Inorg. Nucl. Chem. 4, 106-110. 25. Dietl, T. and Tschesche, H., 1976. Hoppe-Seyler's Z.

Physiol. Chem. 357, 657-665. 26. Gawron, O. and Jones, L., 1977. Biochim. Biophys.

Acta 484, 453-464. 27. Kazarinoff, M. N. and Snell, E. E., 1976. J. Biol.

Chem. 251, 6179-6182. 28. Schubert, M. P.,1935. J. Biol. Chem. 111,671-678. 29. Lange, L. G., III, Riordan, J. F., and Vallee, B. L ,

1974. Biochemistry 13, 4361-4370. 30. Borders, C. L. and Riordan, J. F., 1975. Biochemistry

14, 4699. 31. Ryang, H.-S. and Wang, S. Y., 1978. J. Am. Chem.

Soc. 100, 1302-1303. 32. Rubin, M. B. and Inbar, S., 1978. J. Am. Chem. Soc.

100, 2266-2268. 33. Harden, A. and Norris, D. 1911. J. Physiol. 42,

332-336. 34. Voges, O. and Proskaner, B., 1898. Z. Hyg. InfektKr.

28, 20. 35. Sakaguchi, S., 1925. J. Biochem., Tokyo 5, 25, 133. 36. Dixon, H. B. F. and Fields, R., 1972. Methods in

Enzymol. 25,409-419. 37. Lange, L. G., III, Riordan, J. F., Vallee, B. L. and

Briinden, C.-I., 1975. Biochemistry 14, 3497-3502. 38. JiSrnvall, H., Lange, L. G., III, Riordan, J. F. and

Vallee, B. L., 1977. Biochem. Biophys. Res. Commun. 77, 73-78.

39. Weng, L., Heinrikson, R. L. and Westley, J., 1978. J. Biol. Chem. 253, 8109-8119.

40. Cheung, S.-T. and Fonda, M. L., 1978. Fed. Proc. 37, 1331.

41. Patthy, L. and Smith, E. L.,1975. J. Biol. Chem. 250, 557-564; 565-569.

42. Liu, W. H., Feinstein, G., Osuga, D. T., Haynes, R. and Feeney, R. E., 1968. Biochemistry 7, 2886-2892.

43. Austen, B. M. and Smith, E. L., 1976. J. Biol. Chem. 251, 5835.

44. Borders, C..L., Jr,, Pearson, L. J., McLaughlin, A. E., Vasiloff, J., Gustafson., M. E. and Morgan, D. J., 1978. Fed. Proc. 37, 1325.

45. Gilbert, H. F. and O'Leary, M. H., 1975. Biochemistry 14, 5194-5198.

46. Vetter-Deichtl, H., Vetter, W., Richter, W. and Biemann, K., 1968. Experienfia 24, 340-341.

47. Lipscomb, W. N., 1970. Accounts Chem. Res. 3, 81-89.

48. Nakagawa, S. and Umeyama, H., 1978. J. Am. Chem. Soc. 100, 7716-7725.

49. Auld, D. S. and Holmquist, B., 1974. Biochemistry 13, 4355-4361.

50. Werber, M. M. and Sokolovsky, M., 1972. Biochem. Biophys. Res. Commun. 48, 384-390.

51. Werber, M. M., Moldovan, M, and Sokolovsky, M., 1975. Eur. J. Biochem. 53,207-216.

9 0

52. Kuhn, R. W., Walsh, K. A. and Neurath, H., 1976. Biochemistry 15, 4881-4885.

53. Biinning, P., Holmquist, B. and Riordan, J. F., 1978. Biochem. Biophys. Res. Commun. 83, 1442-1449.

54. Crifo, C. Santoro, L., Rinaldi, A. and DeMarco, C., 1977. Mol. Cell. Biochem. 17, 7-9.

55. Riordan, J. F. and Scandurra, R., 1975. Biochem. Biophys. Res. Commun. 66, 417--423.

56. Gilbert, H. F. and O'Leary, M. H., 1975. Biochem. Biophys. Res. Commun. 67, 198-202.

57. Azaryan, A. V., Mekhanik, M. L., Egorov, Ts. A. and Torchinskii, Yu. M , 1978. Biokhimiya 43,686-695.

58. Braunstein, A. E., 1973. In The Enzymes (Boyer, P.D. Ed.) 3rd Ed. pp. 379-481. Academic Press, New York.

59. Pollack, S. E. and Russell, C. S., 1978. FEBS Lett. 90, 47-50.

60. Chatagner, F. and Pierre, Y., 1977. FEBS Lett. 81, 335-338.

61. Gawron, O. and Jones, L., 1977. Biochim. Biophys. Acta 484, 453-464.

62. Ehrlich, R. S. and Colman, R. F., 1977. Biochemistry 16, 3378-3383.

63. Hayman, S. and Colman, R. F., 1978. Biochemistry 17, 4161-4168.

64. Riordan, J. F., McElvany, K. D. and Borders, C. L., Jr., 1977. Science 195, 884-886.

65. Marcus, F., 1976. Biochemistry 15, 3505-3509. 66. Marcus, F., 1975, Biochemistry 14, 3916-3921. 67. Branden, C. I , Eklund, H., Nordstrom, B., Boiwe, T.,

Soderlund, G., Zeppezauer, E., Ohlsson, I., and Abeson, ~. , 1973. Proc. Nat. Acad. Sci. U.S. 70, 2439-2442.

68. Zeppezauer, E., Jornvall, H., Ohlsson, F., 1975. Eur. J. Biochem. 58, 95-104.

69. Foster, M. and Harrison, J. H., 1974. Biochem. Biophys. Res. Commun. 58,263-267.

70. Bleile, D. M., Foster, M., Brady, J. W. and Harrison, J. H., 1975. J. Biol. Chem. 250, 6222-6227.

71. Yang, P. C. and Schwert, G. W., 1972. Biochemistry 11, 2218-2224.

72. Nagradova, N. K. and Asryants, R. A., 1975. Biochim. Biophys. Acta 386, 365-368.

73. Nagradova, N. K., Asryants, R. A., Benkevich, N. V. and Safronova, M. I., 1976. FEBS Lett. 69,245-248.

74. Blumenthal, K. M., and Smith, E. L., 1975. J. Biol. Chem. 250, 6555-6559.

75. Pal, P. K. and Colman, R. F., 1976. Eur. J. Biochem. 68, 437-443.

76. David, M., Rashed, I. and Sund, H., 1976. FEBS Lett. 62, 288-292.

77. Levy, H. R., Ingulli, J. and Afolayan, A., 1977. J. Biol. Chem. 252, 3745-3751.

78. Davidson, W. S. and Flynn, T. G., 1978. Fed. Abstr. 37, 1320.

79. Buehner, M., Ford, G. C., Moras, D., Olsen, K. W. and Rossman, M. G., 1973. Proc. Natl. Acad. Sci. U.S. 70, 3052-3054.

80. Adams, M. J., Buehner, M., Chandrasekhar, K., Ford, G. C., Hackert, M. L., Liljas, A. Rossman, M. G., Smiley, I. E., Allison, W. S., Everse, J., Kaplan, N. O. and Taylor, S. S., 1973. Proc. Natl. Acad. Sci. U.S. 70,

1968-1972. 81. Borders, C. L., Jr., and Riordan, J. F., 1975.

Biochemistry 14, 4698-4704. 82. Borders, C. L., Jr., Cipollo, K. L., Jorkasky, J. F. and

Neet, K. E., 1978. Biochemsitry 17, 2654--2658. 83. Hjelmgsen, T., Strid, L., and Arvidsson, L., 1976.

FEBS Lett. 68, 137-140. 84. Rodgers, K. and Weber, B. H., 1977. Arch. Biochem.

Biophys. 180, 19-25. 85. Philips, M., Roustan, C., Fattoum, A. and Pradel, L.

A., 1978. Biochim. Biophys. Acta 523, 368-376. 86. Berghaeuser, J., 1977. Hoppe-Seyler's Z. Physiol.

Chem. 358, 1565-1572. 87. Berghaeuser, J., 1975. Biochem. Biophys. Acta 397,

370-376. 88. Powers, S. G. and Riordan, J. F., 1975. Proc. Natl.

Aead. Sci. U.S. 72, 2616-2620. 89. Cipollo, K. L. and Dunlap, R. B., 1978. Biochem.

Biophys. Res. Commun. 81, 1139-1144. 90. Kim, J. P. and Mehler, A, H., 1978. Fed. Proc. 37,

2537. 91. Marcus, F., Schuster, S. M. and Lardy, H. A., 1976, J.

Biol. Chem. 254, 1775-1780. 92. DePont, J. J. H. H. M., Schoot, B. M., Van Prooijen-

Van Eeden, A. and Bonting, S. L., 1977. Biochim. Biophys. Acta 482, 213-227.

93. Frigeri, L., Galante, Y. M., and Hatefi, Y., 1978. J. Biol. Chem, 253, 8935-8940.

94. Salvo, R. A., Serio, G. F., Evans, J. E. and Kimball, A. P., 1976. Biochemistry 15, 493-497.

95. Armstrong, V. W., Sternbach, H. and Eckstein, F., 1976. FEBS Lett. 70, 48-50.

96. Borders, C. L., Jr., Riordan, J. F. and Auld, D. S., 1975. Biochem. Biophys. Res. Commun. 66, 490-495.

97. Hill, J. A., Davies, R. J. H. and Elmore, D. J., 1975. Biochem. Soc. Trans. 3, 1262-1263.

98. Jordan, F. and Wu, A., 1978. Arch. Biochem. Biophys. 190, 699-704.

99. Cooperman, B. S. and Chiu, N. Y., 1973. Biochemistry 12, 1676-1682.

100. Daemen, F. J. M. and Riordan, J. F., 1974. Biochemistry 13, 2865-2871.

101. Woodroofe, M. N. and Butterworth, P. J., 1978. Biochem. Soc. Trans. 6, 562-564.

102. McTigue, J. J. and Van Etten, R. L., 1978. Biochim. Biophys. Acta 523,422-429.

103. Borders, C. L., Jr. and Wilson, B. A., 1976. Biochem. Biophys. Res. Commun. 73, 978-984.

104. Fothergill, L. A., 1977. Biochem. Soc. Trans. 5, 774-776.

105. Sasaki, R., Utsumi, S. and Chiba, H., 1977. J. Biochem. 82, 1173-1175.

106. Borders, C. L., Jr., Woodall, M. L. and George, A. L., Jr., 1978. Biochem. Biophys. Res. Commun. 82, 901-906.

107. Lobb, R. R., Stokes, A. M., Hill, H. A. O. and Riordan, J. F., 1975. FEBS Lett. 54, 70-72.

108. Lobb, R. R., Stokes, A. M., Hill, H. A. O. and Riordan, J. F., 1976. Eur. J. Biochem. 70, 517-522,

109. Kantrowitz, E. R. and Lipscomb, W. N., 1976. J. Biol. Chem. 251, 2688-2695.

110. Kantrowitz, E. R. and Lipscomb, W. N., 1977. J. Biol.

Chem. 252, 2873-2880. 111. Kalousek, F. and Rosenberg, L. E., 1978. Fed. Proc.

37, 236. 112. Aurebekk, B. and Little, C., 1977. Int. J. Biochem. 8,

757-762. 113. Enoch, H. G. and Strittmatter, P., 1978. Biochemistry

17, 4927-4932. 114. Barnard, G. F., Parker, T. S. and Popjak, 1978. Fed.

Proc. 37, 1324. 115. Kameshita, I., Tokushiga, M. and Katsuki, H., 1978. J.

Biochem. 84, 795-803. 116. Chollet, R., 1978. Biochem. Biophys, Res. Comrnun.

83, 1267-1274. 117. Schloss, J. V., Norton, I. L., Stringer, C. D. and

Hartman, F. C., 1978. Fed. Proc. 37, 233. 118. Wolf, B. and Kalousek, F., 1978. Fed. Proc. 37, 234. 119. Tunnicliff, G. and Ngo, T. T., 1978. Experimentia 34,

989-990. 120. Busby, T. F. and Gan, J. C., 1976. Arch. Biochem.

Biophys. 177, 552-560. 121, Freedman, M. H., Grossberg, A. L. and Pressman, D.,

1968. J. Biol. Chem. 243, 6186-6195. 122. Gambhir, K. K. and McMenamy, R. H., 1973. J. Biol.

Chem. 248, 1956-1960. 123. Roosdorp, N., Wiinn, B. and Sj6holm, I., 1977. J.

Biol. Chem. 252, 3876-3880. 124. Rosenblatt, M., Shepard, G. L., Tyler, G. A. and

Potts, J. T,, Jr,, 1978. Biochemistry 17, 3188-3191, 125. Sairam, M. R., 1976. Arch. Bioehem. Biophys. 176,

197-205. 126. Margoliash, E., Ferguson-Miller, S., Tulloss, J., Kang,

C. H., Feinberg, B. A., Brantigan, D. L. and Morrison, M., 1973. Prox. Natl. Aead. Sei. U.S. 70, 3245-3249.

127. Johnson, P. and Blazyk, J. M., 1978. Biochem. Biophys. Res. Commun. 82, 1013-1018.

128. Hardingham, T. E., Ewins, R. J. F., and Muir, H., 1976. Biochem. J. 157, 127-143.

129. Gosselin-Rey, C., Bernard, N. and Gerday, C., 1973. Biochim. Biophys. Acta 303, 90-104.

130. Schmid, R., Jagendorf, A. J. and Hulkower, S., 1977. Biochim. Biophys. Acta 462, 177-186.

131. Vallejos, R. H., Viale, A. and Andreo, C. S., 1977. FEBS Lett. 84, 304-308.

132. Rogers, T. B., Boerresen, T. and Feeney, R. E., 1978. Biochemistry 17, 1105-1109.

133. Rohrbach, M. S. and Bodley, J. W., 1977. Biochemistry 16, 1360-1363.

134. Stein, S. M. and Berestecky, J. M., 1974. Cancer Res. 34, 3112-3116.

135. King, J. L. and Jukes, T. H., 1969. Science, 164, 788-797.

136. Reeck, G., 1976. In Handbook of Biochemistry and Molecular Biology (G. D. Fasman, Ed.) 3rd Ed. vol. III, pp. 504-519, CRC Press, Cleveland.

137. Jukes, T. H., 1973. Biochem. Biophys. Res. Commun. 53,709-714.

138. Cited in ref. 140. 139. Wallis, M., 1974. Biochem. Biophys. Res. Commun.

56, 711-716. 140. Jukes, T, H., 1974. Biochem. Biophys. Res. Commun.

58, 80-84.

91

141. Dice, J. F. and Goldberg, A. L., 1975. Proc. Natl. Acad. Sci. U.S. 72, 3893-3897.

142. Subak-Sharpe, H., Bark, R. P,., Crawford, L. V., Morrison, J. M., Hay, J. and Kier, H. M., 1966. Cold Spring Harbor Symp. Quant. Biol. 31,737.

143. Contreras, R., Ysebaert~ M., Min Jou, W. and Fiers, W., 1973. Nature New Biology 241, 99.

144, Cotton, F. A., laCour, T., Hazen, E. E., Jr., and Legg, M. J., 1977. Biochim. Biophys. Acta 481, 1-5.

145. Laki, K., Seel, M. and Ladik, J., 1977. J. Theor. Biol. 67, 489-498.

146. DarneU, D. W. and Klotz, I. M., 1976. In Handbook of Biochemistry and Molecular Biology (Fasman, G. D., Ed.) 3rd Ed. vol. II, pp. 325-371, CRC Press, Cleveland.

92