115:408/508 Proteins and Enzymes spring 2002

Chemical Modification II, Active-Site Directed Modification

Now, to turn to active-site-directed chemical modification. This is in general the reaction of a specific residue at a rate much greater than that of other residues of the same kind, such that it can be labeled specifically and thus identified. Usually this procedure is used to identify a residue at the active site of an enzyme, or other specific site on a protein such as an effector site or a site where binding to another protein or nucleic acid occurs. It can also be described as a combination of reversible competitive inhibition and irreversible chemical modification.

Such specific modification can occur with ordinary reagents like those on the lists I gave out last week. This can be for one of two reasons:

First, the residue may be hyperreactive, i.e. it reacts much faster than is normal for a residue of that amino acid under those conditions, because it is activated by other nearby amino acid residues. The classic case is serine in serine proteases and esterases, which readily and generally reacts with reagents such as diisopropyl fluorophosphate, DFP for short, and phenylmethanesulfonyl fluoride. Cysteine in papain reacts with a-iodopropionic acid at a rate 1000x that for free cysteine, even though the pK_a of the cysteine is little lower than that for free cysteine or ordinary cysteines in proteins. Lysine in glutamate dehydrogenase, tyrosine⁸⁵ in staphylococcal nuclease similarly react much faster than the ordinary amino acid, free or in small peptides. The hyperreactivity of course depends on the native structure of the protein, disappearing in 8M urea which denatures the protein; and when staphylococcal nuclease combines with the inhibitor deoxythymidine 3',5'-bisphosphate tyr¹¹⁵ is hyperreactive rather than tyr⁸⁵, presumably due to a conformational change. Of course steric protection (burial) may render other groups unreactive, but that doesn't make the reactive amino acid more reactive than the free amino acid! Hyperrreactivity is often inferred to indicate involvement in catalysis, which is reasonable but not necessarily true.

Secondly, the reagent may be selectively adsorbed, i.e. form a complex with the enzyme by noncovalent binding, and then react covalently in the complex. For example, at pH 5.5 iodoacetate reacts with N¹ of his¹¹⁹ or N³ of his¹² of ribonuclease A, eight times with his¹¹⁹ to once with his ¹², but never with both, and the reaction is considerably faster than with free histidine. The carboxyl of the reagent is forming an ion pair - salt interaction - with the protonated form of one histidine, while the unprotonated form of the other displaces iodine from the other end of the reagent.

Selective adsorption is shown by what is usually called a saturation effect: the rate of the reaction is determined, either by loss of biological or enzymatic activity or by incorporation of the reagent, and as usual plotting ln % unmodified or still active vs. time (assuming that the concentration of reagent is much greater than that of protein). The rate, the slope of the plot, can also be approximated by plotting reagent incorporated or % incorporated in the early part of the reaction. The rate is determined for a number of concentrations of the reagent, and plotted vs. reagent concentration. If the reaction is an ordinary second-order chemical reaction, the rate will increase linearly with reagent concentration, indefinitely. If the reaction proceeds via formation of a non-covalent complex, the rate will level off, just as the rate of an enzyme-catalyzed reaction plotted vs. substrate concentration levels off at high substrate conc., and for the same reason: the 'active site' is saturated, fully occupied, and the actual covalent reaction cannot occur any faster as the concentration of free reagent is increased further. Of course when the enzyme + reagent is diluted for assay, reagent not covalently bound diffuses off the active site because the concentration of free reagent is now much lower, so that inactive enzyme is only covalently modified enzyme. Similarly, a plot of 1/rate of inactivation vs. 1/reagent conc. will be a straight line, just like a Lineweaver-Burk plot, with the intercept being 1/the rate at saturation and the slope the dissociation constant of the E·I complex, where the dot denotes a noncovalent complex. Of course you could also use the other transformations of the Michaelis-Menten equation, again substituting rate of inactivation for rate of catalysis.

The active-site-directed reagent is one designed to react in this way, i.e. to bind non-covalently - or sometimes covalently, as a substrate analog - and then react covalently from this complex. Because it reacts in this intimate complex, rather than by an occasional random hit, the active-site-directed reagent typically is chemically much less reactive than general modification reagents; one uses an a-bromo or a-chloro carbonyl compound rather than iodoacetate or iodoacetamide, iodide being the most easily displaced halide from an aliphatic carbon and fluoride the most difficult. By using the less reactive compound one avoids general chemical reaction with all the available residues of that amino acid and restricts modification to an amino acid in the immediate residue of where the reagent binds. It is to be emphasized that the reagent is then a reagent for the binding site, not for a specific type of amino acid. Any nucleophile which is sterically close to the reactive part of the reagent can react, closeness being more important than inherent reactivity. For instance, while tosyllysine chloromethyl ketone reacts with histidine at the active site of trypsin, p-guanidinophenacyl bromide, a chemically similar but shorter reagent, reacts with serine, to which its reactive a-CH₂Br is presumably closer. One may thus be able to identify an amino acid in the active site which is not chemically the most reactive, as Elliott Shaw identified his⁴⁶ in the active site of trypsin and chymotrypsin.

Thus the first criterion for demonstrating that a reagent is in fact site-directed is saturation kinetics, the approach of reaction rate to a maximum as reagent concentration increases indefinitely, as already explained for general reagents which form a non-covalent complex before reacting. Typically the K_d for a reagent designed to be active-site-directed will be considerably lower than for one happened on by chance. For instance, the classic active-site-directed reagent tosylphenylalanine chloromethyl ketone resembles a substrate, tosylphenylalanine methyl ester, except for the chloromethyl group; it has the acylated a-amino group, the aromatic side chain and the carbonyl of the substrate.

A further aspect of non-covalent binding prior to covalent modification is that if the covalent modification is slow and initial velocity in the ordinary assay can first be determined, one may observe that the reagent is an ordinary competitive inhibitor vs. the substrate it resembles when added to an assay. However, after some covalent modification has taken place, the reagent will appear to be a noncompetitive inhibitor, because it has simply removed some enzyme from catalysis and decreased V_max. Conversely, the substrate, or an ordinary competitive inhibitor, will protect the protein against the reagent; but this is also true of ordinary chemical modification, it is not a special criterion for active-site-directed modification. It does demonstrate, however, that the reagent is probably reacting at the same site where the substrate binds.

The residue modified will also be hyperreactive toward the reagent, i.e. reaction of that type of amino acid free or in a small peptide will be very much slower. Also, a chemically similar reagent without binding specificity will also react very much slower - for instance, bromoacetamide would react with a specific residue only very slowly, depending on random hits, while a specific a-bromoamide would react very rapidly.

There are two general purposes for using active-site-directed reagents. One we have already mentioned, the identification of residues at the active site, and thence, we hope, elucidation of the chemical mechanism of the enzymic catalysis. Site-directed mutagenesis is now widely used for the same purpose, but to use it one really needs not only the cloned gene but the three-dimensional structure of the protein. In contrast, active-site-directed modification homes in on the active site without knowing anything about it, and when followed by taking the modified protein apart with proteases and identifying the reacted residue it is a route to identifying the active site and amino acids in it. One might similarly label the active site for X-ray crystallography, as well as in some cases introducing a heavy atom such as iodine at a specific site as needed to solve the phase problem. This is particularly important for proteases which will digest themselves while you are attempting to crystallize them if they have not been inactivated.

The other purpose is the inactivation of the target protein for some practical reason - most commonly inactivation of an enzyme of a pathogenic microorganism, virus or cancer cell. This purpose is identified with the late B.R. Baker, who wrote a book entitled Design of Active-Site-Directed Irreversible Inactivators. He hoped to develop reagents specific enough to distinguish between the enzymes of tumor cells and those of normal cells, without much success; but now people try to devise inactivators of specific enzymes of HIV-1. Reagents of this sort have been developed by microorganisms - the antibiotic azaserine (diazoacetyl-serine), for instance, is a reactive analog of glutamine, and it and the chemically synthesized analog DON (6-diazo-5-oxonorleucine) inhibit purine biosynthesis by inactivating the enzyme 2-formamido-N-ribosylacetamido-5'-phosphate: l-glutamine amidoligase, and generally will inhibit enzymes which use glutamine as donor of an amino group. The antibiotic chloromycetin [1-(p-nitrophenyl)-2-deoxy-2-dichloracetamido-glycerol] is probably an active-site-directed reagent, though I don't know what it reacts with. Many other natural toxins are more or less active-site-directed reagents, and I lecture about them in Dr. Cooper's Biochemical Mechanisms of Toxicology course. For instance, anatoxin a(s) from the cyanobacterium Anabæna flos-aquæ is a very good irreversible inhibitor of mammalian acetylcholinesterase and has killed dogs drinking water with a heavy bloom of the bacterium.

The practical active-site-directed reagent, unlike the investigative reagent, need not have its reactive group where the substrate portion on which the enzyme acts is - though the difference in rate between specific and non-specific reaction will probably be greater if it is. For instance, Baker observed that isophthalic acid is a competitive inhibitor of glutamate dehydrogenase, attached the iodoacetamide group to it, and other analogs with longer chains between the iodoacetamido group and the isophthalate. This type of reagent is termed exo, vs. endo for reaction within the catalytic site.

Larger reagents can be devised to achieve greater specificity. For instance, the serine protease elastase does not react with simple reagents such as tosylalanine chloromethyl ketone, even as it doesn't act on simple substrates such as tosylalanine methyl ester or amide; but alanyl-alanyl-alanyl chloromethyl ketone does inactivate, even as tetrapeptides are substrates. Ala-phe-lys chloromethyl ketone inhibits plasma kallikrein, a protease which cuts the blood-pressure-raising peptide out of a protein, and plasmin, the clot-dissolving protease, but not thrombin, the clot-forming protease, or plasminogen activator. The last two are however inhibited by phe-ala-lys chloromethyl ketone. One can also vary the reactivity of the reagent; a less chemically reactive reagent will generally be more specific, inactivating the enzyme to which it binds most tightly, but not other enzymes. So there has been much interest in pharmaceutical companies in devising specific enzyme active-site-directed inactivating reagents as drugs.

Two related topics are photoaffinity labeling and suicide substrates. The only thing different about photoaffinity labeling is that the compounds are not reactive until illuminated with strong UV or violet light, whereupon they break down to give nitrenes or carbenes, electron-deficient compounds to which almost anything adds - even CH₃ groups. They are useful in identifying active sites when nothing reacts with a more conventional reagent, but the yield with any one group of the protein is generally low; however, the more tightly fixed the reagent is in the active site the more specific the reaction should be. They are particularly useful in a getting a reporter group such as a nitrophenol attached. Diazomethyl ketones yield carbenes: RCOCHN₂ Æ RCOCH: + N₂. Arylazides yield nitrenes: NO₂FN₃ Æ NO₂FN:, + RNH₂ Æ RNHNHFNO₂, a hydrazine. In some cases, such as dimethylcarbamylphenylazide, the nitrene rearranges to a seven-membered ring, a dihydroazepine, which again readily adds amines.

'Suicide substrates' are compounds not initially reactive, but converted by the action of an enzyme into reactive intermediates; they are only reactive at the enzyme's active site, and must react faster than they dissociate to be really specific. The classic example is the enzyme b-hydroxydecanoyl-CoA dehydrase, which removes the hydroxy group to generate an a, b double bond. However, the analogous compound with a b,g triple bond is rearranged by the enzyme to an allene, with both a,b and b,g double bonds, and this is very reactive and an enzyme SH adds to it, inactivating the enzyme.

Now, finally, let me say a few things about the chemistry of modification reactions. Most are with the basic form of an amino acid, reacting as a nucleophile. The cysteine sulfhydryl, reacting as S^-, is the most reactive, but also, in order of decreasing reactivity, the uncharged forms of histidine and lysine, the anions of tyrosine and the carboxylic acids, the uncharged sulfur of methionine, and, rarely, the indole nitrogen of tryptophan. The rate of reaction of any of these which are protonatable will therefore decrease as the group is protonated; indeed, the pH dependence of the rate of chemical modification is a very good way of determining the pK_a of the group in the free enzyme, as the pH dependence of V_max shows the pK_a of the group in the enzyme-substrate complex, while the pH dependence of V_max/K_m generally shows the pK_a in the free enzyme, but may be influenced by other binding effects (and by the pK_as, if any, of the substrate). Among protonatable nucleophiles, the higher the pK_a, in general, the better the nucleophile. Thus specificity, for rather non-specific reagents such as many alkylating reagents, depends on pH; iodoacetamide may react with methionine at pH 5, histidine at pH 7, and lysine at pH 9; cysteine, however, is always most reactive, and will react first with such reagents. It is sometimes necessary to block cysteine with a reversible modification in order to investigate the effect of modifying some other amino acid. Arginine is protonated under all normal circumstances, and does not then react as a nucleophile; however, there is a special category of reagents for arginine, generally with two adjacent carbonyl groups, as butane-2,3-dione, phenylglyoxal; some of these give a modification which is stable only in the presence of borate, which makes them conveniently reversible by removing excess borate.

Acylation gives amides with lysine, thioesters with cysteine, acylimidazoles with histidine, esters with tyrosine and rarely serine and threonine. The modifications of cysteine, histidine and tyrosine are not very stable, and are removed quantitatively by neutral hydroxylamine, leaving the acylated lysines. Anhydrides such as maleic anhydride, with the other carboxyl group in the acyl-lysine nearby, come off again at more acid pH by reformation of the anhydride, which can be useful - the pH at which they come off depends on the exact structure of the acyl group.

Amidination and guanidination are reactions only of lysines; they maintain a positive charge on the group, whereas acylation neutralizes it. Sometimes having the positive charge is necessary for stability of the protein. Positive charge is also maintained when the amino group is merely methylated, by Schiff base formation with formaldehyde and borohydride reduction, as I showed last time. Amidination and guanidination tend to require rather high pH, depending on the reagent.

Also shown on the list are some reagents with some specificity for histidine and tyrosine. Cyanuric fluoride could be called either acylating or arylating.

Alkylation is the widest class of reactions, the most used. They depend on attack of a nucleophilic group of the protein on an activated carbon atom, usually next to a carbonyl group - even in an acid or amide - or to an aromatic nucleus. Usually the nucleophile displaces a halogen atom from the activated carbon, though in some cases it adds to a double bond, or a three-membered ring as in ethyleneimine and ethylene oxide. Iodine is the most easily displaced, so iodoacetate and iodoacetamide are general alkylating agents, usually of cysteine but occasionally of histidine, lysine, methionine and carboxylic acids. Bromine and chlorine are successively less easily displaced - which means that they generally react only when the reagent binds non-covalently to the enzyme - and fluorine is displaceable only from sulfonyl fluorides, like the well known inactivator of serine proteases, phenylmethane sulfonyl fluoride. The products of these reactions are generally acid-stable, and some give products which are well characterized in amino acid analysis, such as carboxymethylcysteine.

Arylation is similar, but the nucleophile reacts directly with the aromatic nucleus; here fluorine is the most easily displaced, in the familiar 1-fluoro-2,4-dinitrobenzene (Sanger's reagent).

Reduction of disulfides is usually by excess small molecule sulfhydryl compound, such as 2-mercaptoethanol, or 1,4-dithiothreitol, whose oxidized form is a 6-membered ring, which pulls the reaction in that direction and allows a much lower concentration to be used - 1 mm dithiothreitol is as good as 50 mm mercaptoethanol and much less smelly. A few other reagents have been used; NaBH₄ selectively reduced specific SS bonds in the reference given.

Disulfide interchange is used either to measure free SH, with Ellmann's reagent, 5,5-dithiobis(2-nitrobenzoate), the released 5-thio-2-nitrobenzoate having a high absorbance, e₄₁₂ = 13,600, or to put a removable blocking agent on free SH groups before modifying other groups with a reagent which otherwise would react with the SH groups. Tetrathionate and methyl methanethiosulfonate are used for this purpose.

Sulfhydryl groups can be oxidized to several states - the first state, a sulfinic acid, -SOH, is not stable in small molecules, which dismute to a sulfenic acid -SO₂H and a sulfhydryl, but can occur when well isolated on a protein. The usual oxidation is all the way to cysteic acid, which is easier to measure by amino acid analysis than cysteine, which after acid hydrolysis is a mixture of cysteine, cystine and mesocystine, cystine with one a carbon inverted in configuration to give a compound separable on the amino acid analyzer. It is easier to oxidize all the cysteine to a single product. Cysteine of course also reacts with mercury compounds - the reaction with p-chloromercuribenzoate was the old way of measuring free SH groups, by a change in absorbance at 255 nm. Mercury compounds are removable by a large excess of mercaptoethanol, or even better by passage through an immobilized SH column; similarly, proteins may be adsorbed on an immobilized mercurial column and eluted with mercaptoethanol, or the fluorescent dansyl group may be attached through a mercury. Pierce now sells tris(2-carboxyethyl)phosphine attached to a gel as an immobilized reducing agent, with greater stability than immobilized SH, for reduction of disulfide bonds without a need to dialyze the protein solution after reduction to remove free small SH compounds.

Certain positively charged reagents react by electrophilic substitution on tyrosine, next to the OH, and histidine. These include tetranitromethane, which generates NO₂⁺ in solution, iodine, which generates I⁺, and diazonium compounds such as diazonium 1H-tetrazole.

There are various ways of oxidizing tryptophan and methionine - though they also oxidize cysteine. You probably know about cleavage of the peptide bond of methionine with cyanogen bromide; 3-bromo-2(2-nitrophenylsulfenyl)skatole does the same thing at tryptophan. Oxidative reactions are otherwise rather hard to control, and typically do not yield a single identifiable product.

There are a few strong reducing agents - Raney nickel, which reduces cysteine to alanine, and diborane and alkylboranes, which can reduce carboxyl groups to CH₂OH. These then decrease the size of the side chain. Unfortunately the reaction tends to be rather incomplete, and they are not often used.

Carboxyl groups are usually modified by ester or amide formation. Some reagents form the ester directly, such as p-bromophenacyl bromide reacting with active-site carboxyls in pepsin, but more usually the carboxyls are activated with a water-soluble carbodiimide, or rarely Woodward's K reagent, and then react with a nucleophile such as aminomethanesulfonic acid, which can be measured on the amino acid analyzer. The first chemical modification of a protein was Frankel-Conrat's esterification of the carboxyls of ribonuclease by refluxing in methanolic HCl! RNAse is a very stable protein.