Detection of Open and Closed Conformations of Tryptophan Synthase by 15N-Heteronuclear Single-Quantum Coherence Nuclear Magnetic Resonance of Bound 1-15N-L-Tryptophan*

1-15N-l-Tryptophan (1-15N-l-Trp) was synthesized from 15N-aniline by a Sandmeyer reaction, followed by cyclization to isatin, reduction to indole with LiAlH4, and condensation of the 15N-indole with l-serine, catalyzed by tryptophan synthase. 1-15N-l-Trp was complexed with wild-type tryptophan synthase and β-subunit mutants, βK87T, βD305A, and βE109D, in the absence or presence of the allosteric ligands sodium chloride and disodium α-glycerophosphate. The enzyme complexes were observed by 15N-heteronuclear single-quantum coherence nuclear magnetic resonance (15N-HSQC NMR) spectroscopy for the presence of 1-15N-l-Trp bound to the β-active site. No 15N-HSQC signal was detected for 1-15N-l-Trp in 10 mm triethanolamine hydrochloride buffer at pH 8. 1-15N-l-Trp in the presence of wild-type tryptophan synthase in the absence or presence of 50 mm sodium chloride showed a cross peak at 10.25 ppm on the 1H axis and 129 ppm on the 15N axis as a result of reduced solvent exchange for the bound 1-15N-l-Trp, consistent with formation of a closed conformation of the active site. The addition of disodium α-glycerophosphate produced a signal twice as intense, suggesting that the equilibrium favors the closed conformation. 15N-HSQC NMR spectra of βK87T and βE109D mutant Trp synthase with 1-15N-l-Trp showed a similar cross peak either in the presence or absence of disodium α-glycerophosphate, indicating the preference for a closed conformation for these mutant proteins. In contrast, the βD305A Trp synthase mutant only showed a 15N-HSQC signal in the presence of disodium α-glycerophosphate. Thus, this mutant Trp synthase favored an open conformation in the absence of disodium α-glycerophosphate but was able to form a closed conformation in the presence of disodium α-glycerophosphate. Our results demonstrate that the 15N-HSQC NMR spectra of 1-15N-l-Trp bound to Trp synthase can be used to determine the conformational state of mutant forms in solution rapidly. In contrast, UV-visible spectra of wild-type and mutant Trp synthase in the presence of l-Trp with NaCl and/or disodium α-glycerophosphate are more difficult to interpret in terms of altered conformational equilibria.

Bacterial tryptophan (Trp) 1 synthase (EC 4.2.1.20) is a pyridoxal 5Ј-phosphate (PLP)-dependent enzyme that catalyzes the final two steps of the biosynthesis of L-Trp (1)(2)(3)(4). The enzyme is an ␣ 2 ␤ 2 complex, consisting of two ␣-subunits bound to a central ␤ 2 -dimer. Each ␣-subunit independently catalyzes the reversible aldol cleavage of indole-3-glycerol phosphate to indole and D-glyceraldehyde 3-phosphate (␣-reaction, Reaction 1). The ␤-subunit catalyzes the conversion of indole and Lserine to L-Trp, with PLP as the cofactor (␤-reaction, Reaction 2). The overall reaction is the combination of the ␣-reaction and the ␤-reaction (␣␤-reaction, Reaction 3). Indole is not observed as a free intermediate in the ␣␤-reaction (5)(6)(7)(8)(9), suggesting that indole is not released from the ␣-subunit of Trp synthase into solution during turnover; hence, Reaction 3 is the physiological reaction of Trp synthase. X-ray crystallographic investigations of Trp synthase from Salmonella typhimurium have revealed a 25-30-Å long interenzyme tunnel linking the ␣and ␤-active sites, through which it has been proposed that indole travels to the ␤-active site directly from the ␣-active site after formation (10 -14). Moreover, intermediate complexes of the ␣and ␤-reactions reciprocally regulate catalysis at the other site in an allosteric manner (11,(15)(16)(17). These allosteric interactions coordinate the forward progress of the ␣␤-reaction.
In the absence of substrate, PLP is bound to ␤Lys87 in the form of the internal aldimine (E A ) (Scheme 1) (18). The reaction of L-Ser at the ␤-site, promoted by the presence of indole-3glycerol phosphate bound at the ␣-active site, forms an external aldimine (E EA-Ser ), which is converted to an ␣-aminoacrylate Schiff's base (E AA ) with elimination of water, by the action of ␤Lys87 as a general base (19). In return, formation of the ␣-aminoacrylate Schiff's base with PLP at the ␤-site activates the ␣-reaction (20). D-Glyceraldehyde 3-phosphate bound to the ␣-site keeps the ␣-site closed, blocking the release of indole into solution and facilitating transfer of indole from the ␣and ␤-active sites through the interenzyme tunnel. Nucleophilic addition of indole to the ␣-aminoacrylate intermediate at the ␤-site then affords the quinonoid intermediate of L-Trp (E Q ). ␤Glu109 activates the addition of indole, a weak nucleophile, probably by acting as a general base or H-bond acceptor to the N-H of the reacting indole (21). At this point, ␤Lys87 acts as a general acid, providing the proton for the conversion of E Q to the L-Trp external aldimine (E EA-Trp ) (19). Release of D-glyceraldehyde 3-phosphate from the ␣-site results in an open conformation, which then allows release of the L-Trp product from the ␤-active site. Nucleophilic attack of ␤Lys87 on E EA-Trp concomitant with release of L-Trp (19) leads to regeneration of the internal aldimine to begin another catalytic cycle.
Evidence of allosteric regulation of Trp synthase comes from both x-ray data (22)(23)(24)(25) demonstrating conformational changes associated with bound intermediates and from kinetic and spectroscopic data obtained with wild-type and several mutant forms of Trp synthase, where residues postulated to be associated with allosteric regulation have been mutated to residues bearing less complementary side chains. ␤Asp305 forms an ion pair with ␤Arg141 in the closed conformation and thus is proposed to help stabilize the closed conformation of Trp synthase (26). Monovalent cations have also been shown to influence the distributions of external aldimine, quinonoid, and aminoacrylate intermediates (22,(27)(28)(29).
In the present study, L-Trp enriched with 15 N at the 1-position of the indole ring was synthesized and used as an NMR probe to obtain conformational information on wild-type and several ␤-site Trp synthase mutants, ␤K87T, ␤E109D, and ␤D305A, in both the absence and presence of Na ϩ , an activating monovalent cation, and an analog of the ␣-reaction product, disodium ␣-glycerophosphate (GP). This novel NMR method permits the simple and rapid determination of the conformational state of wild-type and mutant Trp synthase in solution.

EXPERIMENTAL PROCEDURES
Materials-L-Ser, L-Trp, PLP, and disodium ␣-GP were obtained from Sigma Chemical Co. or U. S. Biochemical Corp. 15 N-Aniline was obtained from Isotec, Inc. Triethanolamine was a product of Fisher Scientific.

Synthesis of 1-15 N-L-Trp
15 N-Isonitrosoacetanilide-Isonitroacetanilide was prepared by the method of Sandmeyer (30,31). 15 N-Aniline (0.500 g, 0.0053 mol) was placed in a 100-ml, round-bottomed flask with 5 ml of water and 0.5 ml of concentrated HCl. One g (0.0051 mol) of chloral hydrate in 15 ml of water was added to this solution, followed by 15 g of sodium sulfate. A solution consisting of 1.2 g (0.05 mol) of hydroxylamine hydrochloride in 5 ml of water was added after 20 min of stirring at room temperature. The reaction mixture was heated at reflux in an oil bath for 1 h. The flask containing the reaction mixture was immediately removed from heat, allowed to cool to room temperature, and then placed in an ice bath. 15 N-Isonitrosoacetanilide crystallized from the ice-cold solution and was collected by filtration. The product was recovered as a white, crystalline powder, which was dried in vacuo to afford 0.770 g (88%). 15 N-Isatin-15 N-Isonitrosoacetanilide (0.500 g, 0.003 mol) was added to 5 ml of concentrated sulfuric acid at 40°C. The reaction mixture was slowly heated to 65°C and stirred at that temperature for 1 h. Upon completion of the reaction, the opaque, purple mixture was poured over 200 g of crushed ice. The ice quickly melted to reveal a bright orange precipitate, which was collected by filtration to afford 0.412 g (92%) of 15 N-isatin. 15 N-Indole-Isatin is reduced to indole with LiAlH 4 (32). 15 N-Isatin (0.350 g, 0.00237 mol) was combined with 6 ml of freshly distilled diethyl ether and cooled to Ϫ78°C with a dry ice/acetone bath. Four equivalents of lithium aluminum hydride in 10 ml of diethyl ether were cooled to Ϫ78°C and added dropwise to the 15 N-isatin solution with vigorous stirring. This mixture was allowed to warm to room temperature over a period of 6 h. Ethanol was then added until gas evolution stopped. Water (5 ml) was then added to the reaction mixture, followed by 1.0 ml of 1.0 M HCl. The aqueous layer was separated, and the ether layer was extracted with 5 ml of 0.5 M HCl. The ether layer was distilled in steam, and 15 N-indole was isolated from the distillate after concentration by rotary evaporation. The product, 15 N-indole, was purified by cold-finger sublimation to afford 0.185 g (66%) of white solid. 1

Preparation of Trp Synthase
Wild-type Trp synthase used in chemoenzymatic reactions was expressed from Escherchia coli CB149 cells containing plasmid pSTB7 (18). Cells were grown, collected by centrifugation, and sonicated, and REACTION 1.
REACTION 3 the enzyme was purified as described previously (18). All samples of Trp synthase, wild-type and mutant ␤K87T, ␤E109D, and ␤D305A, used in the NMR experiments were purified as described previously (18,33).

NMR Sample Preparation
Pyridoxal 5Ј-phosphate (0.5 mg) was added to 1.0 ml of enzyme solution (ϳ40 mg/ml), and it was allowed to stand at room temperature for 1 h. The solution was then applied to a PD-10 (Pharmacia) gel filtration column equilibrated with 10 mM triethanolamine hydrochloride (TEA-HCl), pH 8, eluted with 10 mM TEA-HCl, pH 8, and then concentrated to 0.5 ml by ultrafiltration in a PM-3 Amicon ultrafiltration cell over a YM-30 membrane. The final concentration of Trp synthase was ϳ75 mg/ml (1 mM). Three or four NMR spectra were then obtained for each enzyme sample. The initial data acquisition was made with 500 l of enzyme in 10 mM TEA-HCl, pH 8, plus 50 l of D 2 O for lock. The subsequent spectra were collected after addition to the sample of 25 l of a solution containing 0.02 M 15 N-L-Trp for the second spectrum, 25 l of 1 M NaCl for the third spectrum, and addition of 25 l of 0.5 M disodium ␣-GP for the fourth spectrum.

NMR Data Collection
NMR data were collected with a Varian Inova 500 spectrometer (499.8 MHz, 1 H) with a triple resonance probe. Water signal suppression was done using flip-back pulses (34) and pulse-field gradients (35). Heteronuclear single-quantum coherence (HSQC) experiments (36) were carried out with sensitivity-enhanced, gradient coherence selection (37), and the data were processed with NMRPipe (38).

UV-Visible Spectroscopic Measurements
The UV-visible spectra were obtained with a Cary 1 spectrophotometer. The samples, in a total of 0.6 ml, contained 1 mg/ml of Trp synthase in 10 mM TEA-HCl, pH 8. Spectra were collected of the enzyme alone and together with 1 mM L-Trp and 50 mM NaCl or disodium ␣-GP from 300 to 550 nm at 25°C.

Synthesis of 1-15 N-L-Trp-
The methodology chosen for the synthesis of 1-15 N-L-Trp affords a high yield starting from 15 N-aniline (Scheme 2). Several total syntheses of 15 N-L-Trp have been reported previously in the literature (39 -41), and each used the conversion of 15 N-indole and L-Ser to 1-15 N-L-Trp using Trp synthase, either by reaction in cell cultures modified to express this enzyme in high yield or by the free enzyme in aqueous buffer. However, none of the syntheses reported previously used 15 N-aniline, either as a step or starting material, as the source of the label. We considered the various ways to synthesize indole from aniline, e.g. the Fischer indole synthesis, Japp-Klingeman synthesis, and via the Sandmeyer synthesis of isatin, and the latter was chosen for its overall efficiency. A Sandmeyer reaction of 15 N-aniline with chloral hydrate and hydroxylamine afforded 15 N-isonitrosoacetanilide, which gave 15 N-isatin, as a brilliant orange powder, after ring closure with SCHEME 2 SCHEME 1 sulfuric acid. Subsequent reduction of isatin with four equivalents of lithium aluminum hydride and purification by steam distillation and crystallization produced pure 1-15 N-indole in good yield. 15 N-L-Trp was synthesized from 1-15 N-indole by reaction with one equivalent of L-Ser and a catalytic amount of PLP using Trp synthase. All steps gave greater than 80% yield, except for the reduction of isatin to indole, which gave 66% yield.
HSQC NMR of Trp Synthase-15 N-HSQC NMR measurements were performed on wild-type Trp synthase, K87T, E109D, and D305A mutants, with 1-15 N-L-Trp, with and without a monovalent cation activator, NaCl, and the ␣-subunit ligand, disodium ␣-GP, in 10 mM TEA-HCl, pH 8. Control experiments showed no detectable 15 N-HSQC signal from 1-15 N-L-Trp in the same conditions, in the absence of Trp synthase, because of the rapid exchange of the N-1 H with solvent at pH 8. 1-15 N-L-Trp in the presence of wild-type Trp synthase exhibited a cross peak at 10.2 ppm on the 1 H axis and 128.8 ppm on the 15 N axis (Fig. 1A). The coupling constant for 15 N and 1 H is J NH ϭ 99.8 Hz. Addition of a monovalent cation, NaCl, resulted in a slight increase in signal intensity (Fig. 1B), but 50 mM ␣-GP resulted in a 2-fold increase in peak intensity (Fig. 1C). In contrast, the mutant enzymes, ␤K87T (data not shown) and ␤E109D (Fig. 2), exhibited HSQC cross peaks with 1-15 N-L-Trp that were equal in intensity in the presence or absence of NaCl or disodium ␣-GP. For ␤E109D, the coupling constant for 15 Table I.
UV-Visible Spectra of Wild-type and Mutant Trp Synthase with L-Trp-Addition of 1 mM L-Trp to solutions of wild-type Trp synthase in 10 mM TEA-HCl had only minimal effects on the spectrum of the PLP cofactor of the enzyme (Fig. 3A,  dashed line). There was little change in the spectrum if 50 mM NaCl was added. However, if 50 mM disodium ␣-GP was included, there was a red shift in the peak at about 410 nm, and a new peak at 476 nm, attributable to a quinonoid intermediate, was observed as well as a weak new band at 330 nm. ␤E109D Trp synthase exhibited only a slight change in the spectrum in the presence of L-Trp alone or together with NaCl, but a more prominent peak at 330 nm was observed when disodium ␣-GP was added (Fig. 3B) (21). ␤D305A mutant Trp synthase also showed little change in spectrum with L-Trp alone or with NaCl, but with disodium ␣-GP, a prominent peak at 330 nm and a very weak shoulder at 476 nm were observed (Fig. 3C). DISCUSSION NMR spectroscopy is a very useful technique for the study of enzyme structure and function. Whereas x-ray crystallography is time-intensive and is only capable of providing details of protein structure in a crystalline sample, observation of proteins by NMR spectroscopy, with the use of selectively labeled ligands, is rapid and easy to perform with modern instrumentation. A problem in the measurement of 15 N-NMR spectra in the past has been the very low sensitivity of the 15 N-nucleus, because of the low resonance frequency of 15 N, combined with the low natural abundance of 15 N. The problem of natural abundance can be overcome by enrichment of the compound of interest, but the inherently low sensitivity of the 15 N nucleus remains. 15 N-HSQC NMR overcomes this problem by observation of a 15 N signal by detecting the attached proton(s) (36). Because this method observes the 15 N nucleus indirectly via polarization transfer with the attached hydrogens, an N-H proton must be present, and not exchanging rapidly with the bulk solvent, for a NMR signal to be seen.
Trp synthase is an ideal enzyme for study by NMR with a labeled ligand for several reasons. First, Trp synthase has been the subject of considerable biochemical investigation, and a number of x-ray crystal structures of the wild-type enzyme, several mutant forms, and complexes with ligands have been reported (10,(22)(23)(24)(25). Thus, the information provided by NMR observation of a labeled ligand can be correlated with known crystallographic structural data. Once structural correlations are made, the NMR data can provide new information for systems that lack x-ray structural data. Second, various intermediate complexes of this enzyme allosterically regulate enzyme conformation, catalysis, and intermediate stability of the two active sites, and NMR is a tool that is well-suited for measuring these subtle conformational and electronic changes.  (Fig. 1A). The observation of this signal is dependent on the slow exchange of the hydrogen with the solvent, because a signal from 1-15 N-L-Trp is not observed under these conditions in the absence of Trp synthase. Thus, the observation of the NMR signal in Fig.  1A implies that 1-15 N-L-Trp is bound at the ␤-site and that Trp N-H is at least partially sheltered from solvent exchange. The presence of Na ϩ , which can influence the conformational equilibrium of open and closed conformations, did not have much effect on the HSQC NMR spectrum (Fig. 1B). The binding of ␣-site ligands, disodium ␣-GP or D-glyceraldehyde 3-phosphate, results in a closed conformation of the ␣-site, which is transmitted allosterically to the ␤-active site and shifts the equilibrium at the ␤-site to favor the closed conformation (11). In the presence of disodium ␣-GP, the NMR signal was about twice as intense (Fig. 1C). This result is consistent with the hypothesis that the ␤-active site of Trp synthase with L-Trp bound is in equilibrium between open and closed forms in the absence of disodium ␣-GP and shifts to a predominantly closed conformation in the presence of disodium ␣-GP. The results of the 15 N-HSQC experiments with wild-type Trp synthase are summarized in Table I. Although no x-ray structure of Trp synthase with ␣-GP bound to the ␣-site and L-Trp bound to the ␤-site has been determined, the structure of Trp synthase with D-glyceraldehyde 3-phosphate in the ␣-site and dihydroisotryptophan bound to the ␤-site has been determined, and it is in a closed conformation. 2 The UV-visible spectra of wild-type Trp synthase with L-Trp present (Fig. 3A) showed little change with NaCl, but an increase in the content of the quinonoid intermediate, absorbing at 476 nm, was seen in the presence of disodium ␣-GP, consistent with a closed conformation (11).
The ⑀-amino side chain of ␤Lys87 binds PLP in the form of the internal aldimine, which undergoes nucleophilic attack by the incoming substrate, L-Ser, to form the external aldimine (E A-Ser ). Mutation of ␤Lys87 to threonine (K87T) results in an inactive enzyme, which can form an external aldimine (E A-Ser ) but which cannot form an ␣-aminoacrylate intermediate (19). Addition of ammonia to the K87T L-Ser complex can form an ␣-aminoacrylate intermediate, suggesting that Lys-87 functions as a general base in the mechanism (20). Furthermore, addition of L-Trp to ␤K87T under typical reaction conditions leads to the slow formation of the L-Trp external aldimine.
Because the lysine ⑀-amino group is not present in K87T, this external aldimine is very stable, the internal aldimine cannot form, and hence L-Trp is bound slowly but not released. This property of ␤K87T has been useful for the study of stable complexes of Trp synthase by x-ray crystallography. Currently, x-ray crystal structures of ␤K87T Trp synthase have been 1-15 N-L-Trp are consistent with these x-ray crystallographic results. A strong 15 N signal is observed for 1-15 N-L-Trp with ␤K87T Trp synthase, either with or without the presence of disodium ␣-GP (Table I), as expected, because the closed conformation is seen in the ␤K87T-L-Trp crystal structure, even without an ␣-site ligand present.
The second mutant Trp synthase that we examined by 15 N-HSQC NMR was ␤E109D. ␤Glu109 in the ␤-active site is seen in the structure of the ␤K87T-L-Trp complex (23) to form a hydrogen bond to the indole N-H of the product. Substitution of this glutamate residue by aspartate reduces the reach and conformational freedom of the carboxylate functionality. Ki-netic studies of ␤E109D have indicated that allosteric communication between the ␣and ␤-active sites is unaffected by this mutation (21). D-Glyceraldehyde-3-phosphate bound to the ␣-site activates the reaction of L-Ser at the ␤-site of ␤E109D, as it does with the wild-type enzyme. Furthermore, formation of the ␣-aminoacrylate intermediate is unaffected, as well as its allosteric communication to the ␣-site to release glyceraldehyde-3-phosphate. 15 N-HSQC NMR spectra of 1-15 N-L-Trp with ␤E109D Trp synthase revealed a strong signal, with equal intensity either with or without the presence of NaCl or disodium ␣-GP (Fig. 2). This result suggests that the L-Trp external aldimine of ␤E109D Trp synthase adopts a predominantly closed conformation, as does the ␤K87T mutant Trp synthase. Because formation of the ␣-aminoacrylate intermediate is essentially unaffected by the ␤E109D mutation, the critical role of this residue must be catalysis of indole addition to the aminoacrylate (21). Indeed, kinetic experiments reveal an accumulation of indole at the ␤-active site for ␤E109D Trp synthase (14), whereas this accumulation does not occur with the wild type, indicating a slower reaction of indole with the aminoacrylate for ␤E109D Trp synthase. Furthermore, the ␤E109D mutant enzyme demonstrates a marked preference for indoline as opposed to indole as the ␤-site nucleophile, perhaps suggesting a change in the steric environment of the active site (21). The N-H coupling constant, J NH , is reduced from 99.8 Hz for wild-type to 97.3 Hz for ␤E109D Trp synthase. This is consistent with a stronger hydrogen bond with the N-H of L-Trp for ␤E109D Trp synthase, which may be responsible for the apparent increased preference for a closed conformation. The 15 N-HSQC NMR results with ␤E109D Trp synthase are summarized in Table I. The UV-visible spectra of E109D Trp synthase in the presence of L-Trp did not show much difference with disodium ␣-GP (Fig. 3B) added. There is a small increase in the band at 330 nm with disodium ␣-GP present (21), which could be due to a gem-diamine complex or to an enolimine form of the external aldimine. The enolimine form of a PLP Schiff's base is generally favored by a hydrophobic environment. In contrast to the NMR data, these UV-visible data are more difficult to interpret in terms of altered conformational equilibria.
The third mutant Trp synthase that we studied by 15 N-HSQC NMR was ␤D305A Trp synthase. Unlike the other mutant proteins, ␤K87T and ␤E109D, residue ␤D305 is not located within the ␤-active site. The x-ray data of different conformations of Trp synthase suggest that ␤Asp305 may be important for allosteric communication between the ␣and ␤-subunits via the monovalent cation binding site (29,44), because it forms an ion pair with ␤Arg141 in the closed conformation. ␤D305A Trp synthase has a broader substrate specificity for nucleophiles than wild-type Trp synthase, suggesting that it favors a more open, solvent-accessible active site than wild-type Trp synthase (26). The results of our experiments given in Table I suggest that although the conformational equilibrium in the absence of ␣-ligands indeed preferentially favors the open structure, allosteric communication in ␤D305A is similar to that of the wild-type enzyme. The complete absence of a 15 N-HSQC signal for 1-15 N-L-Trp with ␤D305A Trp synthase without disodium ␣-GP is consistent with previous conclusions that the ␤D305A protein may adopt a more open conformation than wild-type Trp synthase (26,29,44). However, the observation of the 15 N-HSQC cross peak with disodium ␣-GP present implies that the closed conformation of the L-Trp complex can still form, at least partially, when an ␣-ligand is present. This is consistent with the observation that a closed complex with L-Ser can be formed by ␤D305A Trp synthase in the presence of disodium GP or Cs ϩ but not with Na ϩ  (44). There were some differences in the UV-visible spectra of ␤D305A Trp synthase with L-Trp and disodium ␣-GP (Fig. 3C). As with ␤E109D Trp synthase, a band appeared at 330 nm, and similar to wild-type Trp synthase, there was absorption at 476 nm, but very much weaker than for the wild-type enzyme. It is not clear from these UV-visible spectra that a closed conformation is formed by ␤D305A Trp synthase in the presence of disodium ␣-GP.
An important concern with x-ray crystallography is whether the conformational state of a protein in solution is the same as that seen in the crystalline state. An advantage of NMR spectroscopy is that it measures the properties of proteins in solution and thus provides direct information on the solution structure of the protein. Our NMR results with wild-type and ␤K87T Trp synthase, for which x-ray crystal structural data are available, are in agreement with the crystal structures. However, no crystal structures have been obtained yet for ␤E109D and ␤D305A Trp synthase. The 15 N-HSQC NMR results reported herein suggest that the conformational equilibrium of ␤E109D Trp synthase has a preference for the closed conformation with L-Trp bound, more so than does wild-type Trp synthase, whereas the L-Trp complex of ␤D305A Trp synthase favors the open conformation more than wild-type Trp synthase.