Importance of conserved and variable C-terminal residues for the activity and thermal stability of the beta subunit of tryptophan synthase.

To assess the functional roles of helix 13 and of the conserved and variable residues in the C-terminal region (residues 378-397) of the tryptophan synthase β subunit, we have constructed four C-terminal truncations and 12 point mutations. The effects of these mutations on kinetic and spectroscopic properties and thermal stability are reported here. The mutant β subunits all form stable α2β2 complexes that have been purified to homogeneity. The mutant α2β2 complexes are divided into two classes on the basis of activity in the reaction of L-serine with indole to form tryptophan. Class I enzymes, which have mutations at Arg-379 or Asp-381 or truncations (384-397 or 385-397), exhibit significant activity (1-38% of wild type). Class II enzymes, which have mutations at Lys-382 or Asp-383 or truncations (382-397 or 383-397), exhibit very low activity (<1% of wild type). Although Class II enzymes have drastically reduced activity in the reaction of L-serine with indole and an altered distribution of enzyme-substrate intermediates in the reaction of L-serine with β-mercaptoethanol, they retain activity in the reaction of β-chloro-L-alanine with indole. Correlation of the results with the three-dimensional structure of the α2β2 complex suggests that Lys-382 and Asp-383 serve important roles in a proposed “open” to “closed” conformational change that occurs in the reactions of L-serine. Because mutant β subunits having C-terminal truncations (383-397 or 384-397) undergo much more rapid thermal inactivation at 60°C than the wild type β subunit, the C-terminal helix 13 stabilizes the β subunit.

(residues 53-84 and 205-397 shown in cyan and red in Fig.  1A). The pyridoxal phosphate coenzyme is located in the interface between these two domains and interacts with residues from both domains. Portions of the two domains possess a high level of structural homology and are nearly superimposable, suggesting that they may have evolved by gene duplication and fusion. The core region of the C-domain terminates at residue 377 and thus excludes the C-terminal residues 378 -397. Residues 383-393 form a helix (helix 13) that protrudes from the center of the ␤ subunit into solvent. Three residues in helix 13 (Ile-384, His-388, and Leu-391) are involved in ␤/␤ interaction (Fig. 1B). The side chains of Lys-382 and Glu-350 form a salt bridge located below the plane of the pyridoxal phosphate ring in the active center of the ␤ subunit ( Fig. 1B and Fig. 5 in Ref. 6).
Alignment of the sequences of the tryptophan synthase ␤ subunit or ␤ domain from 24 species (see supplementary material in the electronic appendix to Ref. 7) reveals that residues corresponding to 384 -397 of the ␤ subunit from S. typhimurium are variable, whereas the preceding residues 378 -383 are identical (Fig. 2). Several other pyridoxal phosphatedependent enzymes, including dehydratases and synthases, have also been aligned with the tryptophan synthase ␤ subunit and assigned to the ␤ family (8) or to Fold type II (7). The sequence similarity of these other enzymes with the ␤ subunit is very low in the C-terminal region (beyond Glu-350 in the ␤ subunit from S. typhimurium).
To investigate the roles of helix 13 and of the conserved and variable residues in the C-terminal region, we have constructed four C-terminal truncations and 12 point mutations (Fig. 2) using methods based on PCR 1 (6,9). The effects of these mutations on kinetic and spectroscopic properties and thermal stability are reported here. Our results show that residues in the C-terminal region are important for the thermal stability and activity of the ␣ 2 ␤ 2 complex but are not essential for catalysis. Alteration or deletion of Lys-382 or Asp-383 drastically reduces activity in the reaction of L-serine with indole and alters the substrate specificity and spectroscopic properties. We correlate these results with data on the three-dimensional structure of the ␣ 2 ␤ 2 complex (5, 10) 2 and with models depicting ligand-mediated conformational changes of the tryptophan synthase ␣ 2 ␤ 2 complex (11)(12)(13)(14)(15)(16). Our results suggest that Lys-382 and Asp-383 are involved in an open to closed conformational transition that activates the ␣ 2 ␤ 2 complex. Initial aspects of portions of this work have been reported (6,9).
Bacterial Strains and Plasmids, Growth of Cells, and Purification of Enzymes-The Escherichia coli host strain CB149 lacks the trp operon (17). Plasmids pEBA-10, pEBA-6, and pEBA-4A8 (6) express the S. typhimurium tryptophan synthase ␣ 2 ␤ 2 complex, ␤ subunit, and ␣ subunit, respectively. Cultures of the host harboring wild type or mutant plasmids (see below) were grown, and enzymes were induced with IPTG as described (6). Enzymes were purified from the disrupted cells from 100-ml cultures. Purification of the wild type and mutant ␣ 2 ␤ 2 complexes (18) utilized crystallization from crude extracts followed by recrystallization. The combined yields of ␣ and ␤ subunits were greater than 70% of the total soluble proteins, as reported recently for the five complexes with mutations in ␤ subunit residue 382 expressed under the same conditions (6). The amounts of purified ␣ 2 ␤ 2 complex obtained from 100-ml cultures ranged from 23 to 90 mg. Each purified ␣ 2 ␤ 2 complex gave only two bands on SDS-polyacrylamide gel electrophore-FIG. 2. Sequence of C-terminal residues 378 -397 of the tryptophan synthase ␤ subunit from S. typhimurium and structural elements found in the three-dimensional structure (5) (r, random coil; h, helix; ?, cannot be seen). Multiple alignment of 24 ␤ subunit or domain sequences (7) shows that residues corresponding to 378 -383 (underlined) of the ␤ subunit from S. typhimurium are identical, whereas 384 -397 are variable. We have constructed and characterized the four indicated C-terminal truncation mutant proteins and 12 mutant proteins having single amino acid replacements at positions 379, 381, 382, and 383 marked by *; the construction and initial characterization of five mutant ␤ subunits altered at position 382 has been reported recently (6).

FIG. 1.
A, ribbon diagram of the three-dimensional structure of the tryptophan synthase ␣ 2 ␤ 2 complex from S. typhimurium (5). The ␣ subunits are in green, the N-domain of the ␤ subunit (residues 1-52 and 85-204) is in yellow; the C-domain residues 53-84 and 205-382 are in cyan, and residues 378 -393 are in red; C-terminal residues 394 -397 are not visible in the structure and are not shown. The pyridoxal phosphate coenzyme is in black. B, stereo ribbon diagram of the central section from A. Structural elements are labeled according to Ref. 5: ␣, ␣-helix; ␤, ␤ strand. Side chains that are deleted or mutated in this work (see Fig. 2) are shown in lavender. Side chains of Glu-350 and Gln-142 are shown in green.
sis that correspond to the ␣ and ␤ chains (data not shown; see Fig. 3 in Ref. 6 for analogous results) and an absorption spectrum showing the presence of bound pyridoxal phosphate (see below). Our finding that each overexpressed mutant ␤ subunit gave a sharp band on SDSpolyacrylamide gel electrophoresis and was obtained in high yield indicates that there was no degradation of any of these overexpressed proteins overall or from the carboxyl end.
The ␣ and ␤ subunits were purified as described (19). Although the ⌬385-397 mutant protein was expressed as the ␤ subunit by the plasmid pEBA-6 (⌬385-397) as described below, this ␤ subunit was purified as the ␣ 2 ␤ 2 complex for three reasons as follows: 1) higher yields of ␣ 2 ␤ 2 complex than ␤ subunit are obtained by the procedures used; 2) a mutant ␤ subunit may be stabilized by interaction with the ␣ subunit; and 3) ␣ 2 ␤ 2 complexes were used in the experiments reported. The ⌬385-397 ␤ subunit was purified as the ␣ 2 ␤ 2 complex after mixing the harvested cells from two 100-ml cultures (E. coli CB 149/pEBA-6 (⌬385-397) and E. coli CB 149/pEBA-4A8), which express the mutant ␤ subunit and the wild type ␣ subunit, respectively.
PCR-based Mutagenesis-The expression vectors pEBA-10 and pEBA-6 were used as templates for quick and convenient mutagenesis by megaprimer PCR (6). Mutagenic primers used in the construction of the missense mutations and C-terminal truncations were as follows, where base changes are underlined and mixed bases at the same position are in parentheses: R379P, R379K, or R379Q, 5Ј-T CTC TCT GGC (CA)(CA)A GGA GAT AAA G-3Ј; D381S, D381N, or D381Y, 5Ј-AT CTC TCC GGA CGC GGA (TA)(AC)T AAA GAC AT-3Ј (n.b. TCC GGA introduces an AccIII site); ⌬382-397, 5Ј-C CGC GGA GAT TAG GAC ATC TTT A-3Ј; D383A, 5Ј-GA GAT AAA GCC ATC TTT ACC G-3Ј; ⌬383-397, 5Ј-C GGA GAT AAA TAG ATC TTT ACC G-3Ј; ⌬384 -397, 5Ј-A GAT AAA GAC TAG TTT ACC GTA C-3Ј; ⌬385-397, 5Ј-AAA GAC ATC TAG ACC GTA CAC G-3Ј. The TCTAGA in the oligomer for ⌬385-397 introduces an XbaI site. As a result, the mutagenized PCR fragment has two XbaI sites that prevent recloning into pEBA-10. Therefore, the mutagenized fragment was cloned into pEBA-6 after additional manipulation. The fragment was first inserted into the linear pCRII as described (6). The product was cut with EcoRI, and the resulting small fragment was filled in by Klenow and then cut by SphI to yield a fragment that has an SphI site at one end and EcoRI/fill in at the other end. This fragment was then introduced into the SphI and HindIII/fill in sites of pEBA-6. The mutant plasmids are designated as parent (mutation) (e.g. pEBA-6 (⌬385-397) and pEBA-10 (R379Q)). The construction of five mutant proteins with amino acid replacements at position 382 was as described (6).
Enzyme Assays-One unit of activity is the amount of enzyme that gives rise to formation of 0.1 mol of product per 20 min at 37°C. Activities of the ␣ 2 ␤ 2 complex in the conversion of indole (0.2 mM) and L-serine or ␤-chloro-L-alanine (40 mM) to L-tryptophan were determined by a spectrophotometric assay in the presence of an approximately 3-fold excess of ␣ subunit (16,20). CsCl (0.5 M) was added in some assays where indicated. The activities of the ␣ 2 ␤ 2 complex in the ␣ reaction (conversion of indole-3-glycerol phosphate to indole and D-glyceraldehyde 3-phosphate) and in the ␣␤ reaction (conversion of indole-3-glycerol phosphate and L-serine to tryptophan and D-glyceraldehyde 3-phosphate) were measured by a spectrophotometric assay coupled with D-glyceraldehyde-3-phosphate dehydrogenase in the presence of excess ␣ subunit (21).
Spectroscopic Methods-Absorption spectra utilized a Hewlett-Packard 8452 diode array spectrophotometer. Fluorescence measurements were made in a ⌬-Scan 1 (Photon Technology International) dual excitation spectrofluorimeter. Fluorescence titrations of enzymes with Lserine measured the increase in fluorescence emission at 510 nm (excitation at 420 nm) upon incubation of the ␣ 2 ␤ 2 complex (0.26 -0.52 M ␣␤ pair) with 0.01-10 mM L-serine at 25°C (22)(23)(24). K d values for L-serine were obtained by hyperbolic curve fitting of the change in fluorescence ⌬F using Equation 1: K d values for ␤-chloro-L-alanine were obtained from the competitive inhibition by ␤-chloro-L-alanine of the fluorescence of the enzyme complexes with L-serine. In Method 1, several concentrations of ␤-chloro-Lalanine were used (0, 0.2, 0.5, 1, and 3 mM for the wild type and 0, 10, 20, 40, and 60 mM for the K382N ␣ 2 ␤ 2 complex), and K i values were obtained by linear fit of K obs /K d versus [␤-chloro-L-alanine] using Equation 2: In Method 2, a single concentration of ␤-chloro-L-alanine was used (0.5 or 4 mM), and K i values were obtained from double-reciprocal plots of ⌬F versus [L-serine] using Equation 3: Thermal Inactivation-Enzyme solutions (2 mg/ml ␣ 2 ␤ 2 complex in ϳ10 mM sodium N,N-bis(2-hydroxyethyl)glycine containing 0.1 mM pyridoxal phosphate, 0.1 M potassium phosphate, and 1 mM EDTA at pH 7.8) were incubated in Eppendorf tubes in a water bath at 60°C for various times. Tubes were chilled to 4°C for 30 min or more and then centrifuged. Aliquots of supernatant solutions were assayed for enzymatic activity in the ␤ reaction with indole and L-serine in the presence of a 3-fold excess of wild type ␣ subunit.

RESULTS
To investigate the functional and structural roles of the C-terminal region of the tryptophan synthase ␤ subunit, we have engineered 12 point mutations of the conserved residues Arg-379, Asp-381, Lys-382, and Asp-383 and 4 C-terminal truncations. We did not investigate the effects of mutations altering the conserved residues Gly-378 and Gly-380 in this TABLE I Specific activities of wild type and mutant ␣ 2 ␤ 2 complexes Specific activities in the ␣, ␤, and ␣␤ reactions were determined as described under "Experimental Procedures" in the presence of a 3-fold excess of wild type ␣ subunit. The activity of the ␣ subunit alone in the ␣ reaction was 0.6 units/mg or 1-2% that of the wild type ␣ 2 ␤ 2 complex. IGP, indole-3-glycerol phosphate; G3P, D-glyceraldehyde 3-phosphate; WT, wild type; Ind, indole; Rxn, reaction. Activities in the ␣ reaction of ␣ 2 ␤ 2 complexes containing mutant ␤ subunits and wild type ␣ subunit range from 15 to 91% that of the wild type ␣ 2 ␤ 2 complex (Table I). Thus, the mutant ␤ subunits all stimulate the catalytic activity of the ␣ subunit. This result provides partial evidence for the structural integrity of the mutant ␣ 2 ␤ 2 complexes.
Activity Measurements Distinguish Two Classes of Mutant ␤ Subunits- Table I gives the specific activities of the wild type and mutant ␣ 2 ␤ 2 complexes in the synthesis of L-tryptophan from L-serine and indole (␤ reaction, Equation 5) and from L-serine and indole-3-glycerol phosphate (␣␤ reaction, Equation 6).
The ␣␤ reaction is essentially the sum of the ␣ and ␤ reactions and requires coupled catalysis by the ␣ and ␤ subunits. Reac-tion of L-serine at the ␤ site in the ␣ 2 ␤ 2 complex stimulates the rate of indole-3-glycerol phosphate cleavage approximately 20fold as shown by the ratio of ␣␤ reaction:␣ reaction in Table I.
We have arbitrarily classified the mutant ␣ 2 ␤ 2 complexes on the basis of the effects of the mutations on the activity in the ␤ reaction, the reaction of L-serine with indole. "Class I" enzymes, which have mutations at positions 379 and 381 or truncations after residue 383 (384 -397 or 385-397), exhibit significant activity (1-38% of wild type). "Class II" enzymes, which have mutations at positions 382 or 383 or truncations including residues 382 or 383 (382-397 or 383-397), exhibit very low activity (Ͻ1% of wild type). The absence of L-serine stimulation of indole-3-glycerol phosphate cleavage by Class II enzymes (Ratio ␣␤ reaction/␣ reaction ϭ ϳ1 in Table I) can be attributed to the lack reaction of L-serine at the ␤ site.
Some mutant forms of tryptophan synthase have considerably higher activity in the overall ␣␤ reaction than in the individual ␤ reaction (25,26). These results were partly due to a reduced association of the mutant ␣ subunit with the ␤ subunit; addition of indole-3-glycerol phosphate during catalysis of the ␤ reaction or ␣␤ reaction increased association and activity. The finding that the activity of each mutant protein in Class I is quite similar in the ␣␤ and ␤ reactions (Table I) shows that Class I mutations have no effects on association between the ␣ and ␤ subunits that can be detected by these assays of enzymatic activity.
Class II Mutant ␣ 2 ␤ 2 Complexes Have Altered Substrate Specificity- Table II compares the activities of the wild type and mutant enzymes in the reaction of ␤-chloro-L-alanine with indole to activities in the reaction of L-serine with indole. It is striking that all of the mutant enzymes, except the Class I enzymes with mutations at position 379, have rather high activities with ␤-chloro-L-alanine (42-140% of the wild type enzyme). Thus, the Class II mutations, which lead to loss of TABLE II Substrate specificities and spectroscopic properties of wild type and mutant ␣ 2 ␤ 2 complexes Specific activities in the conversion of indole and either L-serine or ␤-chloro-L-alanine to tryptophan were determined as described under "Experimental Procedures" and expressed as percent of activity of the wild type ␣ 2 ␤ 2 complex with L-serine in the absence of CsCl (1100 units/mg, see Table I) or in the presence of 0.5 M CsCl (880 units/mg, see Ref. 6). ␤ Rxn, ␤ reaction, ␤-ME, ␤-mercaptoethanol, WT, wild type. Values of max were obtained from absorption spectra recorded as described in Fig. 3 activity with L-serine and indole, do not prevent the analogous reaction with ␤-chloro-L-alanine. Consequently, the substrate specificity ratio, which is defined as the ratio of activity with ␤-chloro-L-alanine and indole to activity with L-serine and indole, is much higher for the Class II enzymes (25-500) than for the wild type ␣ 2 ␤ 2 complex (0.24) or the Class I mutant proteins (0.11 to 8.7). Table III compares the binding constants of L-serine and ␤-chloro-L-alanine for the wild type and several mutant ␣ 2 ␤ 2 complexes. The results were obtained by determining the effect of L-serine concentration on the change in fluorescence emission at 510 nm in the presence or absence of fixed concentrations of ␤-chloro-L-alanine, a competitive inhibitor that does not form a fluorescent external aldimine (22)(23)(24). The K d values of L-serine and ␤-chloro-L-alanine are similar for all of the enzymes except K382N. Although the binding constant of L-serine (3 mM) for the K382N ␣ 2 ␤ 2 complex is markedly elevated, this enzyme should be saturated by the concentration of L-serine (40 mM) used in the assay. The K382N ␣ 2 ␤ 2 complex exhibits high activity with ␤-chloro-L-alanine and indole even though the K d for ␤-chloro-L-alanine (78 mM) is higher than the concentration of ␤-chloro-L-alanine (40 mM) used in the assay. Thus, the effects of the Class II mutations on substrate specificity are not attributable to changes in the binding of L-serine or of ␤-chloro-L-alanine.

Effects of Mutations on Binding Constants of L-Serine and ␤-Chloro-L-alanine-
CsCl Partially Repairs Activities of the Mutant ␣ 2 ␤ 2 Complexes-Because CsCl activates certain mutant ␣ 2 ␤ 2 complexes (11), including those having single amino acid replacements of ␤ subunit Lys-382 (6), we have also determined activity in the ␤ reaction in the presence of 0.5 M CsCl (Table II). CsCl activates all of the mutant ␣ 2 ␤ 2 complexes that have low activity but has little effect on the R379Q and R379K ␣ 2 ␤ 2 complexes that have relatively high activities. CsCl results in a striking 16-fold activation of the R379P ␣ 2 ␤ 2 complex, which has a much lower activity than R379Q and R379K under standard conditions. CsCl also activates the wild type ␤ subunit (11).
Spectroscopic Properties of the Mutant ␣ 2 ␤ 2 Complexes-The wild type ␣ 2 ␤ 2 complex and ␤ subunit catalyze a ␤-replacement reaction of L-serine with ␤-mercaptoethanol that proceeds through a series of pyridoxal phosphate-substrate intermediates that have characteristic absorption spectra (see Fig. 3 for structures and abbreviations) (27,28). Whereas an equilibrium mixture of these intermediates accumulates with the wild type ␣ 2 ␤ 2 complex, the E-Ser intermediate predominates with the wild type ␤ subunit and most of the mutant ␣ 2 ␤ 2 complexes (data not shown). Conversion of E-Ser to E-AA is rate-limiting with the wild type ␤ subunit (29). Because Cs ϩ and NH 4 ϩ alter the rate of this conversion by the ␤ subunit (29) and alter the rates and spectroscopic properties of some mutant enzymes (11), we have determined the absorption spectra obtained upon reaction of L-serine and of L-serine and ␤-mercaptoethanol in the presence of 0.5 M CsCl (Fig. 3 and Table II) . Fig. 3, A and B, shows the spectra obtained in the presence of 0.5 M CsCl with the R379Q ␣ 2 ␤ 2 complex (a representative of Class I) and with the ⌬383-397 ␣ 2 ␤ 2 complex (a representative of Class II), respectively. The three spectra obtained with the R379Q ␣ 2 ␤ 2 complex are essentially identical to those of the wild type ␣ 2 ␤ 2 complex under the same conditions. The enzyme alone exhibits a major peak centered at 410 nm due to the internal aldimine (E). Reaction with L-serine yields a complex spectrum with a major peak centered at 340 -350 nm, which is ascribed to the external aldimine between pyridoxal phosphate and aminoacrylate (E-AA). Reaction of L-serine and ␤-mercaptoethanol yields a major peak at 468 nm, which is ascribed to a quinonoid (E-Q) intermediate formed upon addition of ␤-mercaptoethanol to the aminoacrylate intermediate. The ⌬383-397 enzyme alone also exhibits a peak at 410 nm (E). Addition of L-serine yields a major peak at 424 nm which is ascribed to the external aldimine between pyridoxal phosphate and L-serine (E-Ser). The spectrum is unchanged by the further addition of ␤-mercaptoethanol. Table II presents the wavelength of maximum absorbance in the presence of L-serine or of L-serine plus ␤-mercaptoethanol for the wild type ␣ 2 ␤ 2 complex and ␤ subunit and for each mutant ␣ 2 ␤ 2 complex in the presence of CsCl. In general, the three spectra for each mutant ␣ 2 ␤ 2 complex in Class I were similar to those of the wild type ␣ 2 ␤ 2 complex, whereas the spectra for each mutant ␣ 2 ␤ 2 complex in Class II were similar  to those of the ⌬383-397 ␣ 2 ␤ 2 complex and of the wild type ␤ subunit. Thus, CsCl partially restores the wild type spectral properties of the Class I mutant ␣ 2 ␤ 2 complexes but not of the Class II mutant ␣ 2 ␤ 2 complexes.
Attempted Peptide Rescue of the C-terminal Truncation Mutant ⌬383-397-Addition of a synthetic peptide corresponding to residues 1-9 of the tryptophan synthase ␤ subunit partially restores the activity of a mutant ␤ subunit having residues 1-9 deleted (30). To test whether a peptide corresponding to the deleted helix 13 could restore the activity and spectral properties of the mutant ␤ subunit having residues 383-397 deleted, we mixed the ⌬383-397 ␣ 2 ␤ 2 complex (7.8 M) with the peptide corresponding to residues 383-393 (1.05 mM) in Buffer B in the absence and presence of 0.5 M CsCl at 25°C for 90 min. Assays of the activity with L-serine and indole in the presence of CsCl (as described in Table II in the absence of peptide) showed that addition of the peptide had no effect on the activity. Similarly, the peptide did not affect the absorption spectra of this mutant enzyme and its enzyme-substrate complexes either in the absence of CsCl or in the presence of CsCl (as described in Fig. 3B in the absence of peptide).
C-terminal Truncation Mutant ␣ 2 ␤ 2 Complexes Are Thermolabile- Fig. 4 shows the rates of thermal inactivation of the wild type and three mutant ␤ subunits upon incubation of the corresponding ␣ 2 ␤ 2 complexes at 60°C. Because the ␣ subunits in the holo-␣ 2 ␤ 2 complexes undergo thermal inactivation at 60°C (31), the activities of the ␤ subunits were determined in the presence of added excess ␣ subunit. The results show that the ␤ subunits in the two C-terminal truncation mutant ␣ 2 ␤ 2 complexes examined (⌬383-397 and ⌬384 -397) were much more thermolabile than the wild type ␤ subunit. The D381Y ␤ subunit was somewhat more thermolabile than the wild type ␤ subunit.

DISCUSSION
The experiments described above were aimed at assessing the functional roles of the conserved and variable residues and of helix 13 in the C-terminal region of the ␤ subunit. Because the C-terminal residues 378 -397 are outside of the core region of the C-terminal domain (5) and are not conserved in related enzymes (7), these residues may have evolved to have special functions, such as stabilizing the ␤ 2 dimer.
The four ␤ subunit C-terminal truncations and 12 point mutations were all expressed in high yield as stable ␣ 2 ␤ 2 complexes ("Experimental Procedures" and Ref. 6), contained bound pyridoxal phosphate that forms enzyme-substrate intermediates ( Fig. 3 and Table II), and activated indole-3-glycerol phosphate cleavage by the ␣ subunit (Table I). Thus, the residues deleted (⌬382-397) or altered (379, 381, 382 and 383) are not required for overall folding of the ␤ subunit, for association of the ␤ subunit with the ␣ subunit to form an ␣ 2 ␤ 2 complex, or for activation of the ␣ subunit. These results document the structural integrity of these mutant ␤ subunits. The ␤ subunit C-terminal helix 13 may serve a structural role because the mutant proteins having residues 383-397 or 384 -397 deleted were much more thermolabile (Fig. 4).
Residues 379 and 381-397 are not essential for catalysis because all of the mutant enzymes exhibit some activity in the reaction of L-serine with indole (0.01-38% of wild type) and quite significant activity in the reaction of ␤-chloro-L-alanine with indole (10 -140% of wild type) (Table II). We have divided the mutant proteins into two classes, I and II, based on activity in the reaction of L-serine with indole. Class I mutant proteins are more similar to the wild type enzyme than Class II mutant proteins in activity with L-serine and indole, in substrate specificity, and in spectroscopic properties (Tables I and II and Fig. 3). The results with Class I mutant proteins show that Arg-379, Asp-381, and residues 384 -397 are not very important for catalytic activity. Although substitution of Arg-379 by Lys or Gln has minor effects on the catalytic properties, substitution by Pro (R379P) has more drastic effects. This result is consistent with the finding that E. coli cells having this mutation in the trpB gene are unable to convert indole to tryptophan (32). Our results do not explain why the R379P mutant cells were found to produce more indole than cells with a trpB mutation at position 382 (32). The activity in the ␣ reaction is in fact lower than that of any of the other mutant proteins. The proline replacement may alter the backbone geometry and have a more deleterious effect on the conformation of the ␤ subunit than the more conservative Lys and Gln replacements.
Class II mutant proteins have extremely low activities in the reaction with L-serine and indole but have activities close to that of wild type in the reaction of ␤-chloro-L-alanine with indole (Tables I and II). These mutant proteins all have a substitution or deletion of Lys-382 or of Asp-383. These results support and extend observations that the mutation of Lys-382 results in inactivation (1,32,33). The D383A and ⌬383-397 mutant proteins lack the conserved residue Asp-383. The finding that these mutant proteins have extremely low activity in the reaction of L-serine and indole provides the first evidence that Asp-383 is very important for the catalysis of this reaction. The effects of the Class II mutations on activity in the reaction of L-serine with indole are not attributable to changes in the binding of L-serine because the K d values of L-serine for the mutant enzymes (Table III) are all lower than the concentration of L-serine (40 mM) used in the assay. The importance of Lys-382 for tight binding of L-serine and ␤-chloro-L-alanine may be attributable to formation of a salt bridge between Lys-382 and Glu-350 (see Fig. 1B and Ref. 6). This salt bridge could stabilize the substrate binding site. An additional role for Lys-382 in a substrate-induced conformational change is discussed below.
Relation of Class II Mutant Proteins to a Model for Ligandmediated Conformational Changes-Class II mutant proteins exhibit altered substrate specificity and an altered distribution of enzyme-substrate intermediates in the reaction of L-serine and ␤-mercaptoethanol (Table II and Fig. 3). These results are now discussed in relation to previous studies of the conforma- tional states of tryptophan synthase. Dunn and co-workers (12)(13)(14)(15) have drawn upon data from a number of investigations of tryptophan synthase by different groups of investigators to formulate a model depicting ligand-mediated conformational changes that occur during the course of the ␣␤ reaction. The model postulates that the ␣ and ␤ subunits undergo transitions between open and closed conformations which function to coordinate the catalytic activities and promote diffusion of the indole intermediate. Recent investigations using 8-anilino-1naphthalenesulfonate binding as a probe have identified three distinct conformations of the ␣ 2 ␤ 2 complex associated with different liganded states of the ␣ and ␤ subunits (13). The conformation of the ␤ subunit stabilized by E-AA and E-Q has been designated as closed, whereas the conformation stabilized by E-Ser and E-S-hydroxyethylcysteine has been designated as open (see diagram at the top of Fig. 3) (13).
In the reaction of L-serine with ␤-mercaptoethanol, the E-AA  (Table II and Fig. 3). This result indicates that the mutations hinder the ligandmediated transition from an open to a closed structure either by stabilizing the open structure or by destabilizing the closed structure. The E-Ser intermediate also predominates in several other ␣ 2 ␤ 2 complexes having mutations in the ␤ subunit (6,11,16). We have postulated that the wild type ␤ subunit and these mutant ␣ 2 ␤ 2 complexes have a conformation (termed "open") that results in a poor alignment of the weak hydroxyl leaving group of L-serine for protonation and ␤-elimination. These enzymes may have much higher activity with ␤-chloro-L-alanine because this substrate has a strong leaving group that does not require protonation. The finding that addition of CsCl (0.5 M) activates a number of open mutant proteins and the wild type ␤ subunit suggests that Cs ϩ may stabilize a more active alternative conformation of these enzymes (6,11). CsCl also increases the very low activities of the D383A, ⌬382-397, and ⌬383-397 mutant proteins 10 -80-fold (Table II). CsCl may stabilize the more active, closed conformation of the ␤ subunit and partially shift the equilibrium from the open to the closed form (11).
Relation of the Mutation Studies to Crystallographic Results- Fig. 1, A and B, shows the locations of residues that have been altered or deleted in the present investigation in the three-dimensional structure of the wild type ␣ 2 ␤ 2 complex. The region containing altered residues (residues 378 -393) is shown in red. The C-terminal residues 394 -397 are not clearly visible in the structure and are not shown. The C-terminal helix 13 from each ␤ subunit is located near the ␤/␤ interaction site and protrudes from the center of the ␤ subunit into solvent. Three residues in helix 13 are involved in ␤/␤ interactions (Fig. 1B). These interactions are Ile-384 to Pro-144, His-388 to Phe-147, and Leu-391 to Phe-147. Two of the conserved residues mutated (Arg-379 and Asp-381) also make ␤/␤ interactions: the two Arg-379 residues are stacked with each other; Asp-381 interacts with Arg-148. Lys-382 forms an internal salt bridge with Glu-350 of the same ␤ subunit. Although the four Cterminal truncations remove the three ␤/␤ interactions made by Ile-384, His-388, and Leu-391, the mutant ␤ subunits all form stable ␣ 2 ␤ 2 complexes. Thus, disruption of interactions between six pairs of residues in the ␣ 2 ␤ 2 complex is not sufficient to cause disruption of the ␤/␤ interface. Similarly, single amino acid substitution of Arg-379 or Asp-381 does not cause disruption of the ␤/␤ interface.
How are ligand-mediated conformational changes in tryptophan synthase related to structural changes? Crystallographic investigations are only beginning to reveal structural effects of ligands. Recent studies reveal that exchange of K ϩ or Cs ϩ for Na ϩ induces local and long range changes in the three-dimensional structure of the tryptophan synthase ␣ 2 ␤ 2 complex (10). Studies of a mutant form of the enzyme (␤K87T) having an external aldimine of L-serine at the ␤ site and indole-3-propanol phosphate or DL-␣-glycerol 3-phosphate at the ␣ site show conformational changes that may be related to formation of the closed conformation in solution. 2 Conformational changes observed in these structures of the residues investigated here include movements of Arg-141 toward Thr-386 and of Gln-142 toward Lys-382 and Asp-383. The possible involvement of these interactions in the closed conformation may explain why the mutation or deletion of Asp-383 or of Lys-382 in Class II mutant enzymes appears to destabilize the closed form of the enzyme and prevent the open to closed transition.
Because proteins are stabilized by a large network of interactions, mutations are quite likely to interfere with this complex network and destabilize the protein or alter the conformation of the enzyme needed for optimal activity (34). Some proteins are stabilized by better attachment of the N and C termini to the rest of the molecule to prevent "fraying" (34). Deletion of one of these termini may remove these stabilizing interactions and promote fraying. Although the C-terminal helix 13 of the ␤ subunit is partially exposed to solvent, it does make several interactions in the ␤/␤ interface. The removal of helix 13 may destabilize the ␤ subunit to heat and loosen ␤/␤ interaction by removing these interactions as discussed above. Addition of a synthetic peptide corresponding to residues 383-393 did not restore the catalytic activity or the spectroscopic properties of the mutant ␤ subunit having residues 383-397 deleted.
Conclusions-Our results show that helix 13 is important for thermal stability and that residues 379, 381, 382, and 383 in the C terminus of the ␤ subunit are important, but not essential, for catalytic activity. Lys-382 and Asp-383 appear especially important for stabilization of the closed conformation of the enzyme that has optimal activity with L-serine.