Mutation of an active site residue of tryptophan synthase (beta-serine 377) alters cofactor chemistry.

To better understand how an enzyme controls cofactor chemistry, we have changed a tryptophan synthase residue that interacts with the pyridine nitrogen of the pyridoxal phosphate cofactor from a neutral Ser (beta-Ser377) to a negatively charged Asp or Glu. The spectroscopic properties of the mutant enzymes are altered and become similar to those of tryptophanase and aspartate aminotransferase, enzymes in which an Asp residue interacts with the pyridine nitrogen of pyridoxal phosphate. The absorption spectrum of each mutant enzyme undergoes a pH-dependent change (pKa approximately 7.7) from a form with a protonated internal aldimine nitrogen (lambdamax = 416 nm) to a deprotonated form (lambdamax = 336 nm), whereas the absorption spectra of the wild type tryptophan synthase beta2 subunit and alpha2 beta2 complex are pH-independent. The reaction of the S377D alpha2 beta2 complex with L-serine, L-tryptophan, and other substrates results in the accumulation of pronounced absorption bands (lambdamax = 498-510 nm) ascribed to quinonoid intermediates. We propose that the engineered Asp or Glu residue changes the cofactor chemistry by stabilizing the protonated pyridine nitrogen of pyridoxal phosphate, reducing the pKa of the internal aldimine nitrogen and promoting formation of quinonoid intermediates.

An important question in investigations of enzyme structure and function is how enzymes have evolved different reaction and substrate specificities. Pyridoxal phosphate (PLP) 1dependent enzymes are attractive targets for addressing this question because they catalyze a wide variety of reactions of amino acids (1,2). The enzyme protein directs and restricts the catalytic potential of the bound PLP to provide the substrate and reaction specificity and the enhanced reaction rate of the enzyme (3,4). Thus, it is important to understand how the protein structure controls the cofactor chemistry and enhances catalytic rates.
Information on how the protein structure controls PLP-de-pendent reactions is beginning to emerge from x-ray crystallography and from sequence comparisons designed to establish evolutionary relationships (5,6). One of the most important and best studied interactions between the cofactor and the protein active site of all PLP enzymes is that between the pyridine nitrogen of PLP and an amino acid side chain (see Fig.  1) (7). In L-aspartate aminotransferase (EC 2.6.1.1) (8) and tryptophanase (EC 4.1.99.1), 2 this interaction is a hydrogen bond/salt bridge between the N-1 proton of PLP and a negatively charged aspartate side chain. These two enzymes are classed in the ␣ family (5) or fold type I (6). The PLP-binding site of the tryptophan synthase ␤ subunit (EC 4.2.1.20), 3 a representative of the ␤ family or fold type II, has the neutral hydroxyl of Ser 377 interacting with PLP N-1 (9,10). Alanine racemase (EC 5.1.1.1; fold type III) is unique in having a positively charged arginine near N-1 of PLP (7). Although the enzymes in the three different fold types have unrelated threedimensional structures, the two enzymes in fold type I have related structures. These four enzymes catalyze their different reactions by the pathways illustrated in Fig. 1. The primary reactions of tryptophan synthase and tryptophanase are the synthesis and degradation of L-tryptophan by ␤-replacement and ␤-elimination reactions, respectively. In this work, we have used site-directed mutagenesis to change Ser 377 to Asp (␤S377D) or Glu (␤S377E) and have determined the effects of these mutations on some kinetic and spectroscopic properties. We have reported briefly (11) that mutation of ␤-Ser 377 to Ala, Asp, or Glu results in a Ͼ100-fold decrease in the rate of conversion of L-serine and indole to tryptophan and that the ␤S377D and ␤S377E ␣ 2 ␤ 2 complexes display some spectral properties similar to those of tryptophanase and aspartate aminotransferase.

EXPERIMENTAL PROCEDURES
Chemicals and Buffers-PLP and ␤-chloro-L-alanine hydrochloride were from Sigma. L-Serine was purchased from Fluka. Solutions of ␤-chloro-L-alanine hydrochloride were freshly prepared and adjusted to pH 7.8 with sodium hydroxide immediately before use. Buffer A (50 mM sodium Bicine containing 1 mM EDTA at pH 7.8) was used for spectroscopic studies unless otherwise specified. The other buffer used was composed of 50 mM triethanolamine/Bicine (pH 7.8) containing 0.2 M NaCl, 0.2 M CsCl, or 0.2 M KCl.
Mutagenesis-The expression vector pEBA-10 was used as the template for quick and convenient mutagenesis by megaprimer polymerase chain reaction (PCR) (12). Mutagenic primers used in the construction of the missense mutations were as follows (where base changes are * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  underlined): S377D, 5Ј-GTC-AAT-CTC-GAT-GGC-CGC-GGA-GT-3Ј; and S377E; 5Ј-GTC-AAT-CTC-GAA-GGC-CGC-GGA-GTA-3Ј. Other primers are described (12). The cloning and expression plasmid for each mutant tryptophan synthase was constructed as described (12). Briefly, each mutagenic primer and PE2 (which contains an XbaI restriction site) were used to amplify the first round of PCR with the pEBA-10 template plasmid using Pfu DNA polymerase (Stratagene). The first PCR fragments were purified and used directly as primers together with the alternate primer PE5 (which contains an SphI restriction site) to amplify the second round of DNA synthesis with the pEBA-10 template plasmid. Deoxyadenosine (dA) was added to the newly amplified second round PCR products by the non-template-dependent activity of Taq polymerase. The second round PCR fragments were purified and directly inserted into the linearized pCRII sequencing plasmid (Invitrogen), which has single 3Ј-deoxythymidine (dT) residues. After confirmation of the mutated genes by DNA sequencing, the inserted DNA fragment was restricted with SphI and XbaI restriction enzymes (Promega) and ligated into the original parent plasmid (pEBA-10), which had also been digested with SphI and XbaI.
Enzymes-Escherichia coli CB149 (13), which lacks the trp operon, was used as a host strain for plasmid pEBA-10 (12) that expresses the wild type and mutant ␤ subunit forms (S377D and S377E) of the Salmonella typhimurium tryptophan synthase ␣ 2 ␤ 2 complex. Cultures of the host harboring wild type or mutant plasmid were grown, and enzyme expression was induced with isopropyl-1-thio-␤-D-galactopyranoside as described (12). Purification of wild type and mutant ␣ 2 ␤ 2 complexes utilized crystallization from crude extracts followed by recrystallization (14). The amounts of purified enzymes obtained from 1-liter cultures were 1000 mg (S377D) and 270 mg (S377E). Analysis of the purified enzymes by SDS-polyacrylamide gel electrophoresis reveal that the S377D enzyme contained a lower content of ␣ subunit (ϳ50%) than the wild type ␣ 2 ␤ 2 complex and that the S377E enzyme contained only ␤ subunit (data not shown). The wild type and S377D ␤ 2 subunits were obtained by heat precipitation of the ␣ subunit from the ␣ 2 ␤ 2 complex (15). Plasmid pEBA-4A8 was used to express the wild type ␣ subunit in E. coli CB149 (12). The ␣ subunit was purified as described (16) with a slight modification; after DEAE-Sephacel column chromatography, all fractions were analyzed by SDS-polyacrylamide gel electrophoresis, and the fractions showing the single band corresponding to the ␣ subunit were combined and concentrated. A 1-liter culture yielded ϳ1 g of homogeneous ␣ subunit. Protein concentrations were determined from the specific absorbance at 278 nm using A cm 1% ϭ 6.0 for the holo-␣ 2 ␤ 2 complex, A cm 1% ϭ 6.5 for the holo-␤ 2 subunit, and A cm 1% ϭ 4.4 for the ␣ subunit (15).
Spectroscopic Methods-Absorption spectra were measured in a Hewlett-Packard 8452 diode array spectrophotometer thermostatted at 25°C. Circular dichroism measurements (mean residue ellipticity in degrees cm 2 /dmol) were made in a Jasco J-500C spectrophotometer equipped with a DP-500N data processor (Japan Spectroscopic Co., Easton, MD).
Data Analysis-The effects of pH on absorbance at 416 nm were analyzed by Equation 1, where EH ϩ and E represent the enzyme with a protonated (416 nm species) and deprotonated (334 -338 nm species) internal Schiff base, respectively. The pK a value for the dissociation of EH ϩ was derived from nonlinear least squares fit to Equation 2, where A is the absorbance at 416 nm and A min and A max are the minimum and maximum absorbance values, respectively. The interaction of the ␣ and ␤ subunits was characterized by measuring the absorbance maxima at 504 nm in the presence of L-tryptophan as a function of ␣ subunit concentration. The data were modeled assuming that the ␣ and ␤ subunits associate in a noncooperative fashion. Under this assumption, the ␣-␤ interaction can be simply modeled according to Equation 3.
The position of the equilibrium of Equation 3 can be monitored by absorbance measurements, where the maximum absorbance A 0 is the maximum absorbance of the ␤ 2 subunit alone with L-tryptophan (S377D and S377E), and A max is the intrinsic maximum absorbance of the ␣ 2 ␤ 2 complex with L-tryptophan.

The concentration of [␣␤] at any given [␣] tot /[␤] tot ratio can be solved explicitly from Equation 4,
. The unique solution for [␣␤] is described by Equation 5, and the variation of maximum absorbance (A) with the fraction of added ␣ subunit (f ␣ ) is given by Equation 6.
The dissociation constants for the ␣ and ␤ subunits in the presence of L-tryptophan were obtained from Equation 6 and are termed apparent dissociation constants (K d(␣␤) ) because they were determined in the presence of ligands including L-tryptophan and cations.

RESULTS
We have altered the active site of the tryptophan synthase ␤ subunit by changing a residue near the pyridine nitrogen (N-1) of PLP from a neutral Ser to a negatively charged Asp or Glu as found in aspartate aminotransferase and tryptophanase (Fig.  1). The engineered mutant ␤ subunits were expressed in high yield by a vector that also expresses the wild type ␣ subunit. Purification of the mutant ␤ subunits by a method that has been used to purify the wild type ␣ 2 ␤ 2 complex (12) and ␤ 2 subunit (14,16) resulted in a partial loss of ␣ subunit from the S377D ␤ subunit and a complete loss of ␣ subunit from the S377E ␤ subunit, as described under "Experimental Procedures." The results suggest that association of the S377D and S377E ␤ subunits with the ␣ subunit is weaker than that of the wild type ␤ subunit, as demonstrated below.
Effects of pH on the Absorbance and Ellipticity Properties of the S377D ␤ 2 Subunit and ␣ 2 ␤ 2 Complex-The absorption spectra of the pyridoxal phosphate cofactor of the tryptophan syn- with the ⑀-amino group of the enzyme lysine; N-1 of PLP interacts with residue X, which is Asp in aspartate aminotransferase (AspAT) and tryptophanase (TPase), Ser in tryptophan synthase (TSase), and Arg in Ala racemase (see the Introduction). The reaction catalyzed by each enzyme proceeds through a series of PLP intermediates, where E-S is the enzyme-substrate intermediate and E-Q 1 and E-Q 2 are quinonoid intermediates (see Fig. 3 for structures). Tryptophan synthase and tryptophanase catalyze the ␤-elimination of OH Ϫ from L-serine to form E-AA, the aminoacrylate intermediate, which either is hydrolyzed to pyruvate and NH 3 or reacts with indole (IND) to form E-Q 2 , followed by protonation to form E-Trp, the enzyme-product complex. The different pathways illustrate the reaction specificity of each enzyme. It is clear that the efficacy of each enzyme lies not only in its ability to accelerate the required reaction at each stage, but as far as possible to prevent all the other alternatives (4). PMP, pyridoxamine phosphate. thase ␤ 2 subunit and ␣ 2 ␤ 2 complex from E. coli (17) and of the ␣ 2 ␤ 2 complex from S. typhimurium (18) are pH-independent between pH 6 and 10. The absorption spectra of the S377A ␤ 2 subunit and ␣ 2 ␤ 2 complex are also pH-independent (11). In contrast, the absorption spectra of the S377D ␤ 2 subunit ( Fig.  2A) and of the S377E ␤ 2 subunit (data not shown) exhibit two pH-dependent absorption bands with maxima at 334 and 416 nm. The sharp isosbestic point indicates that there are only two significantly populated species involved.
Analysis of plots of absorbance against pH at these wavelengths (Fig. 2B) shows pK a values of 7.63 Ϯ 0.06 for this ionization for the S377D ␤ 2 subunit and of 7.89 Ϯ 0.08 for the S377D ␣ 2 ␤ 2 complex. The finding that the pK a value for the S377E ␤ 2 subunit (7.78; data not shown) is 0.15 pH units higher than that for the S337D ␤ 2 subunit (7.63; Fig. 2B) is consistent with the higher pK a value for Glu (4.5) compared with Asp (4.1) in polypeptides and uncharged derivatives of Asp and Glu (19). The CD spectra of the S377D ␣ 2 ␤ 2 complex show that the ellipticity band with a maximum at 416 nm is also pH-dependent (Fig. 2C). The absorption band centered at 334 nm appears to be optically inactive.
Spectroscopic Properties of the S377D ␣ 2 ␤ 2 Complex: Accumulation of Quinonoid Intermediates-The wild type ␣ 2 ␤ 2 complex and ␤ 2 subunit catalyze ␤-replacement reactions with L-serine and indole or ␤-mercaptoethanol that proceed through a series of PLP intermediates (Fig. 1), which have characteristic absorption spectra (20). Two quinonoid-type intermediates occur in this pathway, the first (E-Q 1 ) after removal of the ␣-proton of L-serine and the second (E-Q 2 ) after the ␤-addition of a nucleophile to the aminoacrylate intermediate (E-AA). E-Q 1 has been observed in rapid kinetic studies of the ␣ 2 ␤ 2 complex as a transitory intermediate with maximum absorbance at 460 nm (21), but does not accumulate under equilibrium conditions. E-Q 2 accumulates as a stable intermediate under steady-state conditions in reactions of the wild type ␣ 2 ␤ 2 complex with L-serine and nucleophiles, including ␤-mercaptoethanol and indole (17), and with the product, L-tryptophan (22,23).
Mutation of ␤-Ser 377 to Ala, Asp, or Glu results in a Ͼ100fold decrease in the rate of conversion of L-serine and indole to tryptophan by the mutant ␣ 2 ␤ 2 complexes (11). 4 Nevertheless, 4 The activity of the S377D ␣ 2 ␤ 2 complex is also very low in the reaction with ␤-chloro-L-alanine and indole and in ␤-elimination reactions with L-serine or ␤-chloro-L-alanine. The activities of the S377D ␣ 2 ␤ 2 complexes are rather insensitive to pH between 7 and 9.5 (K.-H. Jhee, unpublished results).  Table I. FIG. 2. Effect of pH on the absorbance and ellipticity properties of the S377D ␤ 2 subunit and ␣ 2 ␤ 2 complex. A, absorption spectra of the S377D ␤ 2 subunit (1 mg/ml in Buffer A) made at 25°C at the indicated pH values. B, pH dependence of the absorbance of the S377D ␤ 2 subunit at 334 nm (E) and 416 nm (Ⅺ) and of the S377D ␣ 2 ␤ 2 complex at 334 nm (q) and 416 nm (f) measured in Buffer A in the presence of a 3-fold excess of ␣ subunit. Derived pK a values are 7.63 Ϯ 0.06 for the S377D ␤ 2 subunit and 7.89 Ϯ 0.08 for the S377D ␣ 2 ␤ 2 complex. C, CD spectra of the S377D ␣ 2 ␤ 2 complex made at the indicated pH values under the conditions described for B. addition of substrates to the S377D ␣ 2 ␤ 2 complex in the presence of different monovalent cations results in formation of new absorption bands near 500 nm that exhibit maximum absorbance within 2.5-9 min (Fig. 3 and Table I). The pronounced band that accumulates in the reaction with L-serine in the presence of Cs ϩ can be attributed to E-Q 1 . More intense bands are observed in the reactions with L-tryptophan or with Lserine and ␤-mercaptoethanol. These bands can be attributed to E-Q 2 and exhibit absorption maxima at longer wavelengths (Table I) than quinonoid bands observed with the wild type ␣ 2 ␤ 2 complex. The structure of the quinonoid (E-Q) is shown in Fig. 3. The tryptophan quinonoid has absorbance maxima at 476 and 504 nm with the wild type and S377D ␣ 2 ␤ 2 complexes, respectively (Table I), whereas the quinonoid formed from Lserine and ␤-mercaptoethanol exhibits absorbance maxima at 468 and 508 nm with the wild type and mutant enzymes, respectively. The L-tryptophan quinonoid has a much greater maximum absorbance with the S377D ␣ 2 ␤ 2 complex (⑀ 504 nm ϭ 49.2 mM Ϫ1 cm Ϫ1 ) than with the wild type ␣ 2 ␤ 2 complex (⑀ 476 nm ϭ 1.4 mM Ϫ1 cm Ϫ1 ) (23) ( Table I).
The quinonoid bands formed differ in stability (Table I). The half-times for disappearance range from 3 min for the L-serine intermediate in the presence of K ϩ to 24 h for the L-tryptophan quinonoid in the presence of Cs ϩ ( Table I). The disappearance of the L-serine intermediate probably reflects either the slow conversion to pyruvate by the very low catalytic activity of the S377D ␣ 2 ␤ 2 complex or the occurrence of irreversible inactivation of the S377D ␣ 2 ␤ 2 complex by a reaction intermediate or both (11).
␣ Subunit Is Required for Tryptophan Quinonoid Formation by the S377D ␤ 2 Subunit-The S377D ␤ 2 subunit alone forms no quinonoid intermediate from L-tryptophan in the presence of Cs ϩ (Fig. 4). Addition of 0.5-3.5 molar eq of ␣ subunit results in formation of increasing amounts of the quinonoid intermediate.  Table I was obtained with more freshly prepared enzyme.) Titration of the S377E ␤ 2 subunit with the ␣ subunit by the same method gave values of K d(␣␤) ϭ 32 Ϯ 3.6 M and of 1750 Ϯ 60 M Ϫ1 cm Ϫ1 for the maximum absorptivity. Thus, association of the S377E ␤ subunit with the ␣ subunit is weaker than that of the S377D ␤ subunit, consistent with the greater loss of ␣ subunit during purification of the S377E ␤ subunit from extracts containing ␣ subunit.
We cannot directly compare the dissociation constant for the S377D ␤ subunit with that for the wild type ␤ subunit because the constants have not been determined under the same conditions. The apparent dissociation constant for the S377D ␤ subunit in the presence of L-tryptophan and Cs ϩ (K d(␣␤) ϭ 7.0 Ϯ 0.16 M) is 137-fold higher than the value of K d(␣␤) ϭ 0.051 Ϯ 0.005 M determined for the wild type ␣ and ␤ subunits from measurements of enzymatic activity in the presence of Cs ϩ in the reaction of L-serine with indole to form L-tryptophan (46) and 3.5-fold higher than the value of K d(␣␤) ϭ 2 M determined from sedimentation equilibrium in the absence of ligands (24). The presence of L-serine is known to tighten the association between the ␣ and ␤ subunits (25). The weaker association of the S377D ␤ subunit with the ␣ subunit is consistent with loss of approximately one ␣ subunit during purification. DISCUSSION The experiments described above were aimed at assessing the functional role of ␤-Ser 377 in the PLP-binding site of tryp-  I Spectroscopic properties of quinonoid intermediates formed by the S377D ␣ 2 ␤ 2 complex Absorption spectra were recorded as described in the legend to Fig. 3 at intervals after mixing the S377D ␤ subunit (23.3 M) in the presence of a 3-fold excess of ␣ subunit with substrate: L-tryptophan (10 mM), ␤-chloro-L-alanine (10 mM), or L-serine (50 mM) in the presence or absence of ␤-mercaptoethanol (50 mM) and in the presence of 0.17 M NaCl, CsCl, KCl, or NH 4 Cl. The maximum absorptivity (⑀ max ) of each quinonoid band at the maximum wavelength ( max ) is given at the time at which maximum absorbance was obtained (t max ). The half-time of disappearance of each quinonoid band (t 1/2 ) is also shown. Substrate Fig. 3  tophan synthase and the effects of replacing this residue with a negatively charged Asp or Glu residue, which is found in several other PLP-dependent enzymes including aspartate aminotransferase and tryptophanase (see Fig. 1). Our most important results are that the mutant enzymes display pH-dependent spectral changes and exhibit enhanced formation of quinonoid intermediates. We discuss these results in relation to the chemistry of PLP and PLP-dependent enzymes. All known PLP-dependent enzymes bind PLP as an internal aldimine (Schiff base) with the ⑀-amino group of an enzyme lysyl residue (see E in Fig. 1). Although the equilibria and absorption spectra of PLP Schiff bases have been investigated in model systems (26), binding of PLP to the active site of an enzyme affects the electronic and spectral properties of the PLP derivatives. The interaction of a residue X with the pyridine nitrogen (N-1) of PLP influences the electronic state of the cofactor, the equilibrium distribution between the different reaction intermediates in Fig. 1, and the pathway of catalysis. Effects of a negatively charged carboxylate (X ϭ Asp or Glu) near N-1 of PLP are discussed here with reference to the probable structures of the low and high pH forms of internal aldimines shown in Fig. 5.
A Nearby Carboxylate Raises the pK a of N-1 of PLP-Interaction of the carboxylate stabilizes the proton on N-1 (Ha) (see Fig. 5) and increases the pK a by 2-5 units (7,27). Although the position of the proton Ha cannot be determined by crystallography, semiempirical calculations of absorption spectra of aspartate aminotransferase suggest that the N-1-Asp 222 pair may be present in both the charged and neutral states, i.e. Ha may be located on the carboxyl or on N-1 (27).
Although the electron-accepting properties of the pyridine ring are enhanced by protonation (7), the presence of a negatively charged Asp near N-1 would reduce the ability of the pyridine ring to attract electrons. Semiempirical calculations of the electronic absorption spectra of PLP derivatives of mitochondrial aspartate aminotransferase (in which X ϭ Asp 222 ) have been carried out using information from x-ray data (28). Calculations of the molecular orbital energy level for PLP in the presence of various protein residues showed that inclusion of Asp 222 alone raises the molecular orbital energies by ϳ3 eV relative to those of PLP. However, inclusion of groups hydrogen-bonded to Asp 222 (a water molecule and the His 143 -Ser 139 hydrogen-bonded pair) substantially lowers the molecular orbital energies. Thus, in the case of aspartate aminotransferase, the charge density on active site Asp is modulated by hydrogenbonding to other groups. The crystal structure of the wild type tryptophan synthase ␣ 2 ␤ 2 complex shows the presence of no water molecules or other residues that might interact with the carboxylate that has been introduced in the ␤S377D mutant enzyme. It is possible that the low activities observed for the mutant enzymes (11) 4 result from the absence of residues that modulate the charge density of the introduced carboxylate.
A Carboxylate near N-1 of PLP Reduces the pK a of the Schiff Base Nitrogen-Although the pK a of the Schiff base nitrogen is usually well over 11 for model Schiff bases with PLP, a value close to 9.6 is observed for the Schiff base between valine and N-methyl-PLP, which has a positive charge on N-1 (29). This result provides evidence that the protonated state of the pyridine nitrogen (N-1) reduces the pK a of the Schiff base nitrogen by ϳ2.5 units. The pH-dependent changes observed with tryptophanase and aspartate aminotransferase have been attributed to dissociation of the proton (Hb) on the Schiff base nitrogen (see Fig. 5). Tryptophanase undergoes a pH-dependent change (pK a ϭ 7.2) from a protonated form ( max ϭ 420 nm) to a deprotonated form ( max ϭ 337 nm) in the presence of K ϩ (30). Aspartate aminotransferase exhibits a pH-dependent conversion (pK a ϭ 6.7) from a protonated internal form ( max ϭ 430 nm) to a deprotonated form ( max ϭ 358 nm) (31,32). The finding that mutation of aspartate aminotransferase Asp 222 to Ala (D222A) or Asn (D222N) yields enzymes with pH-independent absorption spectra provides evidence that Asp 222 is indeed responsible for reducing the pK a of the Schiff base nitrogen (31,33).
Although the spectra of tryptophan synthase are pH-independent (18,34), the absorption spectra of the mutant enzymes engineered in this work are pH-dependent (Fig. 2). These results provide evidence that the engineered ␤-Asp 377 or ␤-Glu 377 interacts with the protonated N-1 of PLP and lowers the pK a of the imine nitrogen of the internal aldimine, as proposed for aspartate aminotransferase (31,32). Our finding that the wild type and S377A ␤ 2 subunits (11) have pH-independent spectra similar to those of the D222A mutant of aspartate aminotransferase gives further evidence that the charge on the residue near N-1 of PLP affects the electron distribution of the cofactor.
Probable structures for the high and low pH forms of the S377D and S377E mutant enzymes are shown in Fig. 5. The low pH form (structure I) with max ϭ 416 nm has two protons, Ha on N-1 and Hb on the Schiff base nitrogen. Ha may reside on the carboxylate some of the time as discussed above. The high pH form (structures IIa and IIb) with max ϭ 334 nm results from loss of one proton. The structure could be either a neutral form (structure IIa), resulting from dissociation of Ha and transfer of Hb to the phenolic hydroxyl, or a dipolar ionic form (structure IIb), resulting from dissociation of Hb. The observed max ϭ 334 nm is consistent with that of the neutral enolimine form of the Schiff base, which is in equilibrium with the resonance-stabilized ketoenamine form in the wild type tryptophan synthase (35). The enolimine form predominates following a thermally induced reversible conformational transition of the ␤ 2 subunit (35). The high pH form of tryptophanase also has max ϭ 337 nm, consistent with an enolimine structure (36 -38). The dipolar ionic form (structure IIb) is the structure that has been suggested for the high pH form of aspartate aminotransferase, which has max ϭ 360 nm (39). Consequently, the enolimine form (structure IIa) is the more likely structure for tryptophan synthase. The locations of the protons shown in Fig. 5 cannot be established by crystallography, but can, in favorable cases, be determined by 1 H NMR spectroscopy (40). Hb dissociates around a pK a value of 6.4 with aspartate aminotransferase (40).
A Carboxylate near N-1 of PLP Promotes Quinonoid Formation-According to the generally accepted mechanism of catalysis by PLP enzymes, formation of the initial enzyme-substrate intermediate (E-S in Fig. 1) is followed by withdrawal of an electron pair from the ␣-carbon into the pyridine ring (1). The electron-accepting properties of the pyridine ring are enhanced by protonation of the ring nitrogen (N-1), as discussed above. If the ring nitrogen is protonated, cleavage of the C-␣-H bond leads directly to the quinonoid E-Q 1 in Fig. 1 (for structure, see Fig. 3). Evidence that the presence of a proton or methyl group on N-1 of PLP facilitates quinonoid formation is provided by studies of quinonoid formation in model and enzymatic systems with pyridoxal and N-methylpyridoxal (41). Investigations of the spectra of quinonoids derived from O-methyl-PLP and PLP suggest that the proton may have migrated from the Schiff base nitrogen to O-3Ј (41,42), as shown in the structure in Fig.  3.
Conclusions-We conclude that replacing tryptophan synthase ␤ subunit Ser 377 with a negatively charged Asp or Glu in the PLP-binding site alters the electron distribution in the PLP derivatives. Some of the spectroscopic properties of the mutant enzymes become similar to those of tryptophanase and aspartate aminotransferase, enzymes that also have an Asp residue near N-1 of PLP, but have very different overall structures.