Lysine 87 in the beta subunit of tryptophan synthase that forms an internal aldimine with pyridoxal phosphate serves critical roles in transimination, catalysis, and product release.

This study provides valuable insights into the functions of the lysine residue that forms an internal aldimine with pyridoxal phosphate in the beta subunit of tryptophan synthase from Salmonella typhimurium. Our spectroscopic and kinetic studies demonstrate that a mutant alpha 2 beta 2 complex having beta subunit lysine 87 replaced by threonine forms external aldimines with several amino acids including L-serine, beta-chloro-1-alanine, L-tryptophan, and D-tryptophan. Because the rates of aldimine formation are very slow, we conclude that one role of lysine 87 in the wild type enzyme is to facilitate formation of external aldimines by transimination. Lysine 87 is an essential catalytic residue because the mutant alpha 2 beta 2 complex has no measurable activity in reactions catalyzed by the beta subunit and does not convert external aldimines to products. The mutant enzyme carries out two slow partial beta-elimination reactions: the conversion of beta-chloro-L-alanine and L-serine to enzyme-bound aminoacrylate. The reaction with L-serine is catalyzed by ammonia, which partially replaces the deleted epsilon-amino group. Lysine 87 is important for substrate and product release because L-serine, L-tryptophan, and aminoacrylate dissociate very slowly from the mutant alpha 2 beta 2 complex. Our ability to prepare very stable derivatives of the mutant alpha 2 beta 2 complex containing tightly bound aldimines with a substrate, a product, or a reaction intermediate provides valuable materials for ongoing x-ray crystallographic investigations and future kinetic analyses of the allosteric activation of the alpha subunit by beta subunit ligands.

This study provides valuable insights into the functions of the lysine residue that forms an internal aldimine with pyridoxal phosphate in the 8 subunit of tryptophan synthase from Salmonella typhirnurium. Our spectroscopic and kinetic studies demonstrate that a mutant azbz complex having j3 subunit lysine 87 replaced by threonine forms external aldimines with several amino acids including L-serine, 8-chloro-1-alanine, L-tryptophan, and D-tryptophan. Because the rates of aldimine formation are very slow, we conclude that one role of lysine 87 in the wild type enzyme is to facilitate formation of external aldimines by transimination. Lysine 87 is an essential catalytic residue because the mutant az@z complex has no measurable activity in reactions catalyzed by the subunit and does not convert external aldimines to products. The mutant enzyme carries out two slow partial 8-elimination reactions: the conversion of 8-chloro-L-alanine and L-serine to enzyme-bound aminoacrylate. The reaction with L-serine is catalyzed by ammonia, which partially replaces the deleted e-amino group. Lysine 87 is important for substrate and product release because L-serine, L-tryptophan, and aminoacrylate dissociate very slowly from the mutant a& complex. Our ability to prepare very stable derivatives of the mutant a2& complex containing tightly bound aldimines with a substrate, a product, or a reaction intermediate provides valuable materials for ongoing x-ray crystallographic investigations and future kinetic analyses of the allosteric activation of the a subunit by @ subunit ligands.
The aim of the present work was to determine the functional roles of the lysine residue that binds pyridoxal phosphate in the p subunit of tryptophan synthase. All pyridoxal phosphate-dependent enzymes bind the carbonyl group of pyridoxal phosphate through an internal aldimine with the 6 -* Preliminary reports of portions of this work were presented at the International Workshop on New Aspects of Biocatalysis Research, June 12-13,1992, Kyoto, Japan and at the 21st FEBS Meeting, August 9-14, Dublin, Ireland. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ll TO whom correspondence and reprint requests should be addressed Bldg. 8, Rm. 2A09, NIH, Bethesda, MD 20892. Tel.: 301-496-2763;Fax: 301-402-0240. amino group of a lysine residue. The lysine residue has been postulated to serve one or more roles in the enzymatic reactions: cofactor binding, formation of enzyme-substrate intermediates, catalysis, or product release. The roles of the lysine residue that forms an internal aldimine with pyridoxal phosphate have been investigated in L-aspartate aminotransferase (1-11), in closely related D-aminO acid aminotransferases (12,13), and in L-histidine decarboxylase (14). L-Aspartate aminotransferase belongs to the most thoroughly studied class of pyridoxal phosphate enzymes and has been extensively investigated by x-ray crystallography (15). The lysine residue that binds pyridoxal phosphate in tryptophan synthase may serve different roles because this enzyme belongs to a second, important class of pyridoxal phosphate enzymes and has a three-dimensional structure that is quite different from that of L-aspartate aminotransferase (16).
The bacterial tryptophan synthase azPz complex (EC 4.2.1.20) catalyzes the final reaction in the biosynthesis of . This reaction is termed the ap reaction because it is essentially the sum of two half-reactions: the a reaction, which is catalyzed at the active site of the a subunit, and the ( 3 reaction, which is catalyzed at the active site of the p subunit. a reaction: indole-3-glycerol phosphate u indole + D-glyceraldehyde 3-phosphate reaction: indole + L-serine + L-tryptophan + Hz0 + D-glyceraldehyde 3-phosphate + H,O The isolated / 3 subunit usually exists as a dimer and catalyzes a number of pyridoxal phosphate-dependent P-replacement reactions and &elimination reactions (20). These reactions proceed through a series of pyridoxal phosphate intermediates shown in Scheme I. Early evidence that pyridoxal phosphate forms an internal aldimine with lysine 87 (21-23) ( E in Scheme I) has been confirmed by x-ray crystallography of the tryptophan synthase azp2 complex from Salmonella typhirnuriurn (16). We have previously reported that replacing lysine 87 with threonine yields an inactive form of the subunit (K87T), which binds the a subunit, pyridoxal phosphate, and L-serine (24). Circular dichroism spectra of the mutant enzyme and its complex with L-serine have been reported (25). In the present work, we have investigated the rates of formation of enzyme-substrate intermediates by the K87T a& complex and the occurrence of partial reactions. The results provide evidence that lysine 87 in the wild type enzyme serves critical roles in transimination, catalysis, and product release. SCHEME I chloro-L-alanine (hydrochloride) were freshly prepared and adjusted to pH 7.8 immediately before use. Indole-3-glycerol phosphate was prepared enzymatically and purified as described (26). Buffers-Buffer B (0.05 M N,N-bis(2-hydroxyethyl)glycine containing 1 mM EDTA) was used for all spectroscopic studies at pH 7.8 with the exception of one study in which the pH was varied (see Fig.  4). Enzymes that were stored in Buffer B containing pyridoxal phosphate (0.02 mM) or 2-mercaptoethanol (0.01 M), or both, were converted into Buffer B by gel filtration on a PD-10 column (Pharmacia LKB Biotechnology, Inc.) before spectroscopic studies with amino acids to avoid aldimine formation between free pyridoxal phosphate and free amino acids in solution.
Enzymes-Wild type and mutant forms of the tryptophan synthase a2f12 complex from 5' . typhimuriurn were purified and crystallized from extracts of a host strain Escherichia coli (CB 149) that lacks the trp operon and harbors a high copy plasmid carrying wild type or mutant trpA and trpB genes from S. typhimurium as described previously (24).' The preparation of the K87T olz@z complex that was first reported contained about 0.4 mol of L-serine/mol of pyridoxal phosphate (24).* The spectroscopic properties of that enzyme indicated that the L-serine was bound as the external aldimine with pyridoxal phosphate (24). The preparation isolated for the present studies by essentially the same purification procedure contains a much lower concentration of L-serine (<0.1 mol/mol of pyridoxal phosphate) and was used without further treatment for some experiments. This new preparation exhibits an absorbance maximum near 400 nm and has a low fluorescence (see "Results"). For some experiments, the residual L-serine-pyridoxal phosphate aldimine and pyridoxal phosphate were removed by a modification of a procedure for removal of pyridoxal phosphate from the wild type a& complex (27). This method employs 1 M KSCN to facilitate dissociation of the a and @ subunits and 0.01 M hydroxylamine to react with pyridoxal phosphate. Although the reaction of pyridoxal phosphate with hydroxylamine is very rapid in the wild type enzyme in the presence or absence of KSCN, the reaction of the pyridoxal phosphate-L-serine adduct with hydroxylamine in the K87T a& complex is slower in the presence of KSCN (tlIz = 4 min) and much slower in the absence of KSCN (t1I2 = 11 min).3 These results indicate that the L-serine aldimine is bound very strongly to the K87T az@z complex and that ' E , coli CB 149 harboring a high copy plasmid (pSTB7) carrying wild type S. typhimurium trpA and trpB genes, which was originally developed by Dr. Ronald Bauerle, has been deposited at the American Type Culture Collection for distribution.
The serine content was determined by amino acid analysis of the K87T a& complex after removal of salts by gel filtration with elution by water and acid precipitation of the protein. Serine was the only amino acid detected. E. W. Miles, unpublished results.
some "loosening" of the enzyme structure is necessary to facilitate the displacement of L-serine from the external aldimine (ES I in Scheme I) by hydroxylamine. In the modified method for preparing L-serine-free K87T a~p~ complex, the enzyme (2 ml at 10 mg/ml) was incubated for 2 h at 25 "C in Buffer B containing 1 M KSCN and 0.01 M NHzOH, submitted to gel filtration on a PD-10 column equilibrated with Buffer B containing 1 M KSCN, and dialyzed for 16 h against Buffer B. The resulting apo-K87T az@2 complex was converted to the holo-K87T a& complex by incubation with 0.1 mM pyridoxal phosphate for 90 min at 25 "C followed by gel filtration on a PD-10 column equilibrated with Buffer B. This unpublished method was used to prepare the holo-K87T a& complex used for circular dichroism studies (25). Enzyme Assays-One unit of activity in any reaction is the formation of 0.1 pmol of product in 20 min at 37 "C. The activities of the ~Z @ Z complex in the CY and a@ reactions were measured by spectrophotometric assays coupled with ~-glyceraldehyde-3-phosphate dehydrogenase (28). Activities in @-replacement reactions with indole and 40 mM L-serine were measured by a direct spectrophotometric assay (29). The activity of the a& complex in @-elimination reactions with 40 mM L-serine was determined by a spectrophotometric assay coupled with lactic dehydrogenase; the reaction mixture contained 0.18 M sodium chloride or ammonium chloride (30) as indicated.
Spectroscopic and Analytical Methods-Absorption spectra, difference absorption spectra, and time course measurements at single wavelengths were made using a Hewlett-Packard 8452 diode array spectrophotometer. Fluorescence spectra and single wavelength measurements were made using a Perkin-Elmer model MPF-44B fluorimeter. Data were analyzed where indicated using the PC-MLAB version (Civilized Software) of the MLAB modeling system (31). Circular dichroism spectra were taken on a Jasco 5-600 spectropolarimeter. The concentration of the purified ~Z @ Z complex was deteimined from the specific absorbance at 278 nm (E'" = 6.0) (29). Protein concentrations of some modified a&, complexes were determined by the BCA protein assay reagent (Pierce Chemical Co.) using purified K87T a& complex as standard. The phenylhydrazones of pyruvate and pyridoxal phosphate were prepared by the method for pyridoxal phosphate (32) that has also been applied to a-keto acids (33). Standards and enzyme solution were treated with the acidic reagent for 45 min at 25 'C (33). The acid-denatured proteins were removed by centrifugation, and the absorption spectra were determined on the supernatant solutions. Concentrations were determined from the absorbance of the pyridoxal phosphate phenylhydrazone at 412 nm (E412nm = 2.47 x lo' M" cm") (32) and the pyruvate phenylhydrazone at 322 nm (E32znrn = 1.8 X lo4 M-' cm") (33). When the concentrations of pyruvate and pyridoxal phosphate were determined in mixtures, the absorbance at 322 nm was corrected for the absorbance due to the pyridoxal phosphate phenylhydrazone at this wavelength (E32znm = 0.71 X lo' M" cm").

RESULTS
T o define the functional roles of the lysine residue that forms an internal aldimine with pyridoxal phosphate ( E in Scheme I) in the active site of the p subunit of tryptophan synthase, we compared some catalytic and spectroscopic properties of the wild type a& complex and a mutant a& complex in which lysine 87 of the p subunit is replaced by threonine (K87T). Spectroscopic Properties of the Wild Type and K87T ad32 Complexes: Reactions with Amino Acids- Fig. 1 shows absorption and circular dichroism spectra of t h e wild type and K87T a& complexes in the presence and absence of amino acids that give complementary information about the formation of enzyme-substrate intermediates. The bands in these spectra are identified with the structures in Scheme I on the basis of previous studies of the wild type enzyme as described in the legend to Fig. 1 (25,27,[34][35][36][37][38][39][40][41]. The circular dichroism spectra reflect the asymmetry of bound intermediates and help in the resolution of the bands in the absorption spectra. For example, the rather broad E S I11 peak or shoulder in t h e absorption spectra in Fig. 1A is seen as a sharper negative trough in t h e circular dichroism spectra in Fig Comments on spectra and of assignments of bands to structures in Scheme I are as follows. A , the spectrum of the wild type enzyme plus L-serine reflects an equilibrium mixture of intermediates ES 1-111 dominated by E S 111 with a major peak at 340 nm (35). The finding that the intensity of the 340 nm peak is lower with (3-chloro-L-alanine than with L-serine is not currently understood. Many factors including pH and ligands affect the equilibrium distribution of intermediates formed with L-serine and (3-chloro-L-alanine (63, 64). These reactions are under investigation in several laboratories using presteady-state and steady-state kinetics. It is possible that the chloride product can back up the reaction. B, the negative ellipticity band in the spectrum with L-serine has been reported (25, 39, 65). C, the L-serine band with Amax = 424 nm is characteristic of E S I that is formed as a transitory intermediate by the wild type a& complex and as a transitory intermediate by the wild type ( 3 subunit (38,411. D , the L-serine complex bound to the K87T a& complex (ES I) exhibits a positive ellipticity band at 424 nm that is much more intense than the ellipticity band at 400 nm of the bound pyridoxal phosphate ( E ) ; the increased intensity is probably caused by the more rigid orientation of the coenzyme (25). The weak, negative band at 460 nm in the spectrum of E is an experimental artifact because it did not appear in the previous study. E and F, absorption spectra for the wild type azpz complex with L-tryptophan and D-tryptophan and circular dichroism spectra with L-tryptophan have been reported (39,40,61,62). The L-tryptophan band with Amax = 476 nm has been assigned to the quinonoid E S IV, and the D-tryptophan band with A,, = 446 nm has been assigned to E S V.
forms external aldimines (ES I with L-serine and ES V with L-and D-tryptophan) that have absorption and circular dichroism properties similar to those of the corresponding external aldimines formed by the wild type az@z complex. The observation that the formation of the external aldimines of L-serine and D-tryptophan by the mutant enzyme greatly increases the very weak intensity of the coenzyme ellipticity band indicates that pyridoxal phosphate becomes more rigidly oriented when its carbonyl group forms a covalent bond with the bound substrate.
A key difference between the wild type and K87T a& complexes is that the wild type enzyme converts L-serine to an equilibrium mixture of intermediates dominated by E S 111, whereas the mutant enzyme forms E S I but not E S 111.
Similarly, the wild type enzyme converts L-tryptophan to a mixture of E S V and E S IV, whereas the mutant enzyme forms E S V but not E S IV. The results provide partial evidence that the reaction of the mutant enzyme with L-serine or L-tryptophan is blocked at steps 2 or 5, respectively, in which the a-proton is removed. In contrast, the K87T mutant does form E S I11 from 0-chloro-L-alanine, a highly reactive amino acid with a strong chloride leaving group. This reaction is described below and in Fig. 5.
Rates of Reactions of the K87T a z P 2 Complex with Amino Acids-The spectral changes that accompany the reaction of the wild type enzyme with amino acids reach a steady-state p Subunit equilibrium within about 1 s (35). Because changes with the mutant enzyme are much slower, spectra for the mutant enzyme in Fig. 1 were recorded after 1 h. Fig. 2 shows that the rate of reaction of the mutant enzyme with D-tryptophan depends on the time of incubation and on the amino acid concentration. The rates of reaction of the K87T a& complex with D-tryptophan follow first-order kinetics (Fig. 2B). The rate constants at each concentration of D-tryptophan have been determined in two ways: 1) from the plots in Fig. 2B,  and 2) regression techniques (see Fig. 2C for the data at 0.5 mM Dtryptophan). Analysis of the derived rate constants assuming a hyperbolic dependence on D-tryptophan concentration yields an apparent association constant for D-tryptophan of 0.53 & 0.2 mM (Fig. 2C, inset). Table I lists the apparent dissociation constants for the reaction of the K87T a& complex with several amino acids and the times for halfmaximal reaction at saturating amino acid concentration.
Occurrence of an Isomerization-The reaction of L-serine (Fig. 3) or D-serine (not shown) with the K87T azPn complex exhibits more complex kinetics. A semilogarithmic plot (Fig.   3B) of the change in difference absorbance versus time after addition of 0.1 mM L-serine shows a fast phase (tlIz = -20 min) and a slow phase (tip = -200 min). The results of similar plots at several other concentrations of L-serine (data not shown) show that the fast phase is dependent on the Lserine concentration whereas the slow phase appears to be independent of the amino acid concentration. These two phases may correspond to the rapid formation of an E S complex followed by a slow isomerization process leading to a slowly dissociating ES* complex, as reported for slow binding inhibitors (42-46).
fast slow The occurrence of an isomerization is supported by the presence of an isosbestic point in the early stages of the conversion of E to E S I (Fig. 3A, 0-60 min), which is followed by a slow increase in the extinction coefficient for E S I in the second phase. It is also supported by our finding that the apparent dissociation constant for L-serine decreased from 0.1 mM after 2 h to 0.01 mM after 24 and 96 h (Fig. 3c).
Partial Reactions-The product of reaction of the K87T a& complex with L-serine ( E S I) is very stable because the absorption spectrum was essentially unchanged after gel filtration (Fig. 4A,  a The wavelengths of maximum and minimum absorbance are taken from the difference spectra in Fig. 5 B, and the others are not shown. The two sets of wavelengths with p-chloro-L-alanine are for the first and second steps of the reaction as described in Fig 5 and the text. The difference absorbance (AhmaJ -A h e n ) was used to follow the reaction of each amino acid as illustrated in Figs. 2, 3, and 5.
*Apparent dissociation constants were calculated from rates of reaction determined at several concentrations of each amino acid as illustrated for D-tryptophan in Fig. 2 and for L-serine (fast phase) in Fig. 3B. e The time for half-maximal reaction at saturating amino acids, the value in parentheses was at 1 mM p-chloro-L-alanine.  ( E S 111). The rate of disappearance of E S I in the presence of 1 M ammonium chloride is much more rapid at pH 8.8 (tl12 = 20 min) than at pH 7.5 (t112 = 110 min) (Fig.   4B) and is linearly dependent on the concentration of ammonium chloride at pH 8.5 (Fig. 4C).
The reaction of the K87T cy& complex with p-chloro-Lalanine (Fig. 5) occurs in at least two steps: formation of the external aldimine ( E S I with Amax = 424 nm) followed by the partial conversion of E S I to E S I11 with A, , , = 340 nm. The plots of two difference absorbance values (AAmax=440nm -Axmin = 380 nm) and (Axma, = 470 nm -Aimin = 400 nm) versus time of reaction with 1 mM P-chloro-L-alanine (Fig. 5A, inset) show the progress of the first and second reaction steps, respectively. The addition of 0.5 M ammonium chloride decreases the rate of the first step and has no effect on the rate of the second step (data not shown). The spectrum of the enzyme The solid line shows the best fit of the data to Equation 2 usingMLAB. The inset shows the apparent dissociation constants (K'd) derived from this analysis and from analysis of the data at 2,7, and 24 h. The inset shows the time course of the difference absorbance at the indicated wavelengths that were selected from the difference absorption spectra in B. B, the spectrum of the K87T aZp2 complex alone was subtracted from the spectrum of the enzyme immediately after addition of P-chloro-L-alanine (0' -K87T) and 60 min after addition of 0-chloro-L-alanine (60' -K87T). Values of X , , and Amin for each curve are shown in Table I and used in the inset in A to follow the two stages of the reaction. C, absorption spectra of the K87T-P-chloro-~-alanine complex after isolation by gel filtration.

Site-directed Mutagenesis
The K87T a& complex (-30 p~) was first incubated with 40 mM 8chloro-L-alanine for 3 h at 25 "C and then isolated by gel filtration on a PD-10 column in Buffer B at pH 7.8. Absorption spectra of the isolated K87T-P-chloro-~-alanine (-15 p~) complex were recorded at 0,24, and 72 h; the spectrum of the untreated K87T a& complex is shown for comparison. after reaction with P-chloro-L-alanine was unchanged after gel filtration (Fig. 5C, 0 H R ) but changed slowly over a period of 72 h to give a final spectrum similar to that of the untreated K87T a& complex.
To confirm the presence of aminoacrylate bound to the K87T azP2 complex after treatment with p-chloro-L-alanine or with L-serine in the presence of ammonium chloride, we have determined the pyruvate and pyridoxal phosphate content of the treated enzyme solutions with the phenylhydrazine reagent under acidic conditions, which denature and precipitate the protein (Table I1 and Fig. 6) (32,33). Table I11 shows that the K87T a&-L-serine complex that had been incubated in the presence of ammonium chloride (Experiment 3) contained pyruvate and pyridoxal phosphate in a ratio of 0.9:l.O; the K87T-P-chloro-~-alanine complex (Experiment 4) contained both pyruvate and pyridoxal phosphate and in a ratio of 0.5:l.O.
Specific Activities for Reactions Catalyzed by the Wild Type and K87T a& Complexes-Although the homogeneous K87T aZ& complex has no measurable activity in the p reaction (Table 111) as reported previously (24), the specific activity of the K87T a& complex in the a reaction is 65% that of the wild type a2p2 complex and much higher than the very low intrinsic activity of the a subunit alone in the a reaction (17,47). This finding that the K87T mutant p subunit stimulates

PLP) and pyruvate (Pyr) content of the K87T a& complex before and after reaction with L-serine or 0-chloro-Lalanine
Exp Pyridoxal phosphate and pyruvate content were determined on the treated or untreated enzymes as described under "Experimental Procedures." * A solution of the K87T a& complex was pretreated with 40 mM L-serine for 64 h, isolated by gel filtration, and then incubated in the presence or absence of 0.2 M ammonium chloride for 96 h as described in Fig. 3A.
A solution of K87T aZp2 complex was pretreated with 40 mM pchloro-L-alanine for 3 h as described in Fig. 4 A , isolated by gel filtration as described in Fig. 5B, and treated with phenylhydrazine reagent within 30 min after gel filtration. the activity of the a subunit provides evidence for the conformational integrity of the mutant p subunit.

DISCUSSION
It is important to understand how reactions catalyzed by pyridoxal phosphate enzymes differ in rate and specificity from reactions catalyzed by pyridoxal or pyridoxal phosphate in model reactions (48,49). Our results show that lysine 87, which forms an internal aldimine with pyridoxal phosphate p Subunit 8733 in the @ subunit of tryptophan synthase, serves critical roles in transimination, catalysis, and product release. Lysine 87 does not play an essential role in cofactor binding because replacement of this residue by threonine does not prevent pyridoxal phosphate binding (24,25). Studies with several other pyridoxal phosphate enzymes have shown that chemical modification of the active site lysine (1, 4, 10) or amino acid replacement (2,3,6,10,(12)(13)(14) does not prevent pyridoxal phosphate binding.
Lysine 87 Facilitates Transimination-Our finding that the K87T a2P2 complex forms external aldimines ( E S I in Scheme I) with several amino acids very slowly (Figs. 3 and 4 and Table I) is evidence that lysine 87 facilitates transimination in the wild type enzyme. These results support an early proposal that formation of internal aldimines facilitates transimination (50). An alternative explanation of our results is that the mutation has affected the reactivity of the enzyme by changing the conformation of the active site or by altering the position of another base such as His-86 or Glu-109. Although exogenous amines facilitate the formation of external aldimines by an analogous mutant form of D-amino acid transiminase (12), added ethanolamine or ammonium chloride inhibits the rate of formation of the external aldimine by the K87T a& complex with amino acids (Fig. 2 A ) .
Lysine 87 Is an Essential Catalytic Residue-The K87T a& complex in inactive in @-replacement and @-elimination reactions with L-serine (Table 111) and does not catalyze the turnover of the external aldimine of L-serine ( E S I) during a period of many days in the presence or absence of indole ( Fig.  4 and text). These results are strong evidence that lysine 87 catalyzes either removal of the a-proton of L-serine or protonation of the weak hydroxide leaving group of L-serine (51), or both steps. The active site lysine residue in aspartate aminotransferase catalyzes proton transfer in the tautomerization of the external aldimine to the ketimine (9, 15).
Partial Reactions Yield Insight into Mechanism-The external aldimine of @-chloro-L-alanine bound at the active site of the K87T a& complex ( E S I) is slowly converted to the tightly bound aldimine of aminoacrylate ( E S 111) (Figs. 5 and 6 and Table 11). This slow, partial reaction probably occurs in the absence of catalysis by lysine 87 because @-chloro-Lalanine is a highly reactive amino acid that is converted to HC1, pyruvate, and ammonia slowly in the absence of enzyme or more rapidly in model reactions with pyridoxal phosphate or pyridoxal (49,(52)(53)(54). We propose that pyridoxal phosphate bound at the active site of the K87T a& complex catalyzes the @-elimination of HCl from the aldimine of @-chloro-Lalanine in the mutant enzyme lacking the normal catalytic lysine residue and that the rate of this nonenzymatic reaction is increased by the favorable orientation of the labile Ca-H aldimine bond perpendicular to the pyridine ring (11, 55). Because (3-chloro-L-alanine undergoes a slow, partial reaction with the K87T a& complex whereas L-serine does not, lysine 87 may catalyze the protonation of the hydroxide leaving group of L-serine. Abeles and co-worker (51) have suggested that @-elimination of the weak hydroxide leaving group of Lserine requires protonation whereas @-elimination of the strong chloride leaving group of @-chloro-L-alanine does not. Because addition of ammonium chloride results in the slow conversion of the external aldimine of L-serine at the active site of the K87T a& complex ( E S I) to the tightly bound aldimine of aminoacrylate ( E S 111) (Fig. 4 and Table 11), ammonium chloride partially substitutes for the deleted tamino group of lysine 87. The pH dependence of the observed pseudo-first order rate constants for this reaction (Fig. 4B) indicates that the rate is dependent on the concentration of the free base form of the ammonia catalyst. The results imply that NH, serves as the acceptor of the a-proton of L-serine and that lysine 87 serves this role in the wild type a& complex. "Chemical rescue" of a mutant form of aspartate aminotransferase by amines has been extensively investigated (5, 11,56,57) and has also been demonstrated with one mutant form of the thermostable D-amino acid transiminase (13). The studies by Kirsch and colleagues (5,11,56,57) demonstrate that the amine nitrogen directly abstracts the aproton and that this catalysis is increasingly effective at the higher pH values at which the amine is unprotonated. Assuming that the concentration of NH3 is 0.2 M in 1 M NH4C1 at pH 8.5, the calculated rate of reaction of NH3 with the isolated K87T-~-serine complex in Fig. 4C ( k = 0.0024 s-' M-') is very much slower than the rate of reaction of NH3 with the cysteine sulfinate complex of the K258A mutant of aspartate aminotransferase ( k = 16 s-') calculated from Fig. 4 in Ref. 11.
Lysine 87 Serves a Critical Role in Product Release-The final step in all reactions catalyzed by pyridoxal phosphate enzymes is displacement of the a-amino group of the product by the e-amino group of lysine with regeneration of the original enzyme ( E S V + E and E S VI -+ E in Scheme I).
This is the reverse of the first step in substrate binding ( E S I + E ) . The external aldimines of L-serine and D-and Ltryptophan are bound very tightly to the K87T a& complex because each of these complexes can be isolated by gel filtration and shown to remain stable for days (Fig. 4A, and analogous data for D-and L-tryptophan not shown). Tightly bound L-serine can only be removed by the stringent treatment with KSCN and hydroxylamine described under "Experimental Procedures." The external aldimine of aminoacrylate ( E S 111) that is produced by the reaction of the K87T a& complex with @-chloro-L-alanine is also bound very tightly because it is stable to gel filtration and decays slowly over a period of days (Fig. 5C). Thus, lysine 87 in the wild type enzyme plays a critical role in substrate binding and product release. Ammonium chloride does not substitute for lysine 87 in this role because the isolated D-tryptophan complex is stable in the presence of 0.5 M ammonium chloride at pH 8.0 for 6 h (data not shown). Replacement of the active site lysine in aspartate aminotransferase by alanine also results in very tight binding of L-aspartate (10).
The slow binding (Fig. 3) and slow dissociation of L-serine by this mutant enzyme resemble the slow binding and slow dissociation of inhibitors and reaction intermediate analogs that have been observed with a large number of enzymes (42)(43)(44)(45)(46) and may involve a conformational change that results in tighter binding.
Conclusions and Future Directions-Our most significant conclusions are that lysine 87 is an essential catalytic residue in @-elimination and @-replacement reactions and serves important roles in transimination and product release. An important consequence of this work is that we can now prepare very stable, covalent enzyme-substrate intermediates with the K87T a~@2 complex. The L-serine and L-tryptophan complexes with the K87T a& complex (ES I and E S V in Scheme I) have been crystallized and analyzed by x-ray crystallography at high resolution: The structure of the L-tryptophan complex reveals a striking "collapse" of the tunnel that extends from the a site to the @ site in the unliganded wild type ~P @ P complex (16). This structural alteration may be partially responsible for the observed tight binding of L-tryptophan. These enzyme-substrate complexes will also be valuable for studies of the effects of bound ligands on the stability of the p Subunit az/3z complex and for examination of the allosteric effects of ,8 subunit ligands on the activation of the a subunit. Previous studies provide indirect evidence that aminoacrylate formation is the trigger that activates indole-3-glycerol phosphate cleavage at the a site (58)(59)(60). We may be able to obtain direct evidence for this activation by studying the rate of cleavage of indole-3-glycerol phosphate by the isolated K87T aZp2 complex that contains the bound external aldimine of aminoacrylate (ES 111).