Crystal Structure of the βSer178 → Pro Mutant of Tryptophan Synthase

The catalytic activity of the pyridoxal 5′-phosphate-dependent tryptophan synthase α2β2 complex is allosterically regulated. The hydrogen bond between the helix βH6 residue βSer178 and the loop αL6 residue Gly181 was shown to be critical in ligand-induced intersubunit signaling, with the α-β communication being completely lost in the mutant βSer178 → Pro (Marabotti, A., De Biase, D., Tramonti, A., Bettati, S., and Mozzarelli, A. (2001) J. Biol. Chem. 276, 17747–17753). The structural basis of the impaired allosteric regulation was investigated by determining the crystal structures of the mutant βSer178 → Pro in the absence and presence of the α-subunit ligands indole-3-acetylglycine and glycerol 3-phosphate. The mutation causes local and distant conformational changes especially in the β-subunit. The ligand-free structure exhibits larger differences at the N-terminal part of helix βH6, whereas the enzyme ligand complexes show differences at the C-terminal side. In contrast to the wild-type enzyme loop αL6 remains in an open conformation even in the presence of α-ligands. This effects the equilibrium between active and inactive conformations of the α-active site, altering k cat andK m , and forms the structural basis for the missing allosteric communication between the α- and β-subunits.

digm for the analysis of intersubunit regulatory signals. The enzyme is formed by two ␣and ␤-subunits, arranged in a linear ␣␤␤␣ geometry (1), and catalyzes the last two steps in the biosynthesis of L-tryptophan in a highly concerted mode. The isolated ␣and ␤-subunits exhibit an activity about 2 orders of magnitude lower that in the tetramer, indicating that the formation of the complex is accompanied by subunit conformational changes (2,3). Further levels of regulation are achieved by ligand-induced intersubunit signals that are pH-, temperature-, and cation-dependent and keep the catalytic activities of ␣ and ␤-active sites in phase (4 -11). The mechanism of the ␣-␤ activation and allosteric regulation is based on an open-close transition of both subunits (7,12,13) involving the movement of several parts of the ␣and ␤-subunits. In particular, signal transduction from the ␣to the ␤-active site involves residues belonging to loop ␣L2, loop ␣L6, and the helix ␤H6 (14,15), the latter structural element being part of the ␤Ϫsubunit COMM domain, as defined by Schneider et al. (15). The question whether there is a unique or multiple pathways of communication and whether these pathways are specialized in controlling defined functional properties of the enzyme was addressed by investigating the activity and regulatory properties of several mutants of the ␣and ␤-subunits. It was found that mutants in loop ␣L2 and mutants in helix ␤H6 that interact with loop ␣L2 exhibit altered ␣ and ␤ activities (16 -18), whereas mutants in loop ␣L6 and mutants in helix ␤H6 interacting with loop ␣L6 also exhibit altered allosteric regulation (19 -22). In particular, the mutant ␤S178P exhibits a 2-fold decrease in ␤-subunit activity and a completely impaired ("knock-out") allosteric regulation (21). This finding was explained by the loss of a key hydrogen bond between the carbonyl oxygen of Ser 178 , located at the end of helix ␤H6, and ␣Gly 181 , located in the center of loop ␣L6. To determine the structural basis of the loss of regulation, and, thus, to acquire further information on the mechanism of intersubunit communication, the three-dimensional structure of the mutant ␤S178P was determined in the absence and presence of the allosteric effectors indole-3-acetylglycine (IAG) (23) and DL-␣glycerol-3-phosphate (GP). The structural findings indicated that the equilibrium between active and inactive forms of the ␣-subunit might be affected by the ␤S178P mutation. Therefore, the steady kinetics of the forward and reverse ␣-reaction were analyzed in detail. The crystal structures of wild-type tryptophan synthase complexed with IAG and other novel ␣-subunit ligands are reported in the accompanying paper (24).

MATERIALS AND METHODS
Crystallization and Complex Preparation-IAG and GP were obtained from Sigma and were used without further purification. The ␤S178P mutant of TRPS was purified as described previously (21). The protein was stored at a concentration of 10 mg/ml in a solution containing 50 mM Na-Bicine, pH 7.8, 10 mM Na-EDTA, 1 mM dithioerythritol, and 20 M pyridoxal 5-phosphate. Crystals were grown in the dark at room temperature using the hanging drop geometry and mixing equal volumes (2-4 l) of protein and reservoir solutions. The latter contained 9 -12% polyethylene glycol 8000, 1.5 mM spermine, 1 mM EDTA, 50 mM Bicine buffer, pH 7.8. Prior to flash-cooling in liquid nitrogen the crystals were rinsed briefly in a cryo-protectant solution consisting of 100 mM Bicine buffer, pH 7.8, 15% (w/v) polyethylene glycol 8000, and 20% (v/v) glycerol. The ␤S178P GP and ␤S178P IAG complexes were generated by soaking crystals for 10 min in the cryo-protectant solution containing 50 mM GP or IAG, respectively, before flash-cooling in liquid nitrogen.
X-ray Data Collection and Refinement-Diffraction data of the ␤S178P mutant were collected at beamline ID14-1 at the European Synchrotron Radiation Facility, Grenoble, France using a wavelength of 0.934 Å and a MAR CCD detector. The crystals were kept at 100 K during measurements. The diffraction data were processed with XDS and scaled with XSCALE (25). Data statistics are given in Table I. Refinement was started to 2.5-Å resolution with CNS 1.0 (26) by performing rigid-body and simulated-annealing steps. The coordinates of the wild-type TRPS IPP structure (Protein Database accession code 1Q0P (27)) were used as a starting model, omitting the coordinates of loops ␣L2 and ␣L6, IPP, the cofactor pyridoxal phosphate, and all water molecules. The mutated residue was modeled as alanine to avoid model bias. The final model was obtained by cyclic rounds of manual model building with the program O (28) and Maximum-Likelihood refinement with the program REFMAC (29) using all reflections to maximum resolution. Water molecules were incorporated by ARP (30) using the automatic cut-off option. All waters were checked manually and removed if displaying unusual hydrogen-bonding geometry. Apart from residues at the C termini of both polypeptide chains, the only part of the ␤S178P mutant structures that remains too disordered to be build into electron density was the loop ␣L6 (␣L177-␣L191) and, in the ligandfree ␤S178P and the ␤S178P GP structure, the side chain of ␣F212. In all three structures, for several amino acids the electron density is consistent with a second and, in one case, third side chain conformation. Data and refinement statistics are given in Table I. The coordinates and structure factor amplitudes have been deposited with the Protein Database (31) (accession codes 1K7X, 1K8Z, and 1K8Y).
Structure Superposition-To investigate the influence of the proline mutation and of the different ligands of the ␣-active site on the allosteric communication, we superimposed the new ␤S178P mutant structures with each other and also with the ligand free (Protein Database accession code 1BKS (32)) and IAG wild-type (24) TRPS structures. The superposition was done with the program O (28) using all common C ␣ -atom coordinates of both structures except the C ␣ atoms belonging to the COMM domain (␤Gly 102 -␤Gly 189 ). The resulting r.m.s. deviations were calculated with the program BRAGI (33).
Assays of Enzymatic Activities-IGP was synthesized according to the procedures described by Kawasaki et al. (34), and its concentration was determined spectroscopically. DL-glyceraldehyde 3-phosphate (G3P) was obtained from Sigma and used without further purification in the acidic form. The ␣ activities were measured spectroscopically in 100 mM bis-Tris-propane, pH 7.8, 100 mM NaCl, utilizing the difference in absorption between indole and IGP at 290 nm (35,36). Spectra were taken at 37°C using 0.2-mm path length quartz cuvettes. The forward ␣-reaction was measured using a protein concentration of 3.61 M and IGP concentrations between 0.03 and 1.5 mM. The solution was preincubated at 37°C for 15 min, and the reaction was started with the addition of TRPS. The reverse ␣-reaction was measured using a protein concentration of 63 nM, an indole concentration of 2 mM, and G3P concentrations between 0.3 and 3.0 mM. The solution was preincubated at 37°C for 15 min, and the reaction was started with the addition of G3P. The data were fitted to a hyperbolic curve with the program GRAFIT.

RESULTS AND DISCUSSION
Since the influence of the proline mutation in helix ␤H6 of the COMM domain on the TRPS structure is most pronounced in the ligand-free enzyme, the ligand free ␤S178P structure is discussed first in more detail. Then, this structure serves as a basis for the comparison with the IAG and GP complexes.

Ligand-free ␤S178P Structure
The crystal structure of the ligand-free ␤S178P TRPS mutant was solved at 1.7-Å resolution and refined to crystallo- a Completeness, R sym , and ͗I/(I)͘ are given for all data and for the highest resolution shell: ␤S178P GP , 1.6 -1.5 Å; ␤S178P and ␤S178P IAG , graphic R-factors of R work ϭ 18.7% and R free ϭ 23.9% (Table I).
The overall topology of the mutant structure is the same as for the wild-type (15). Besides the C-terminal residues of both polypeptide chains, weak or no electron density was found for the residues of loop ␣L6 (␣Leu 177 -␣Ala 190 ) and for the side chain of phenylalanine ␣Phe 212 . The latter residue was modeled as alanine. The missing parts of the protein structure are not due to poor x-ray data quality, since the ␤-subunit core residues ␤Met 187 , ␤Lys 219 , and ␤Ser 301 exhibit clear features for double side chain conformations. The observation of multiple side chains, which also occurs in the ␤S178P IAG and ␤S178P GP structures (see Table I), indicates high x-ray data quality. Moreover, loop ␣L6 was also not detectable in other structures of the wild-type and mutant (1,14,(37)(38)(39)(40), confirming that this region of the ␣-subunit is particularly mobile. A close-up of the mutation site (Fig. 1A) shows  Table II for the bond length. The hydrogen bond between the carboxylate of the active conformation of ␣Glu 49 and of IAG has also been observed in the TRPS IAG complex (24). C, the ␣-active site of the final ␤S178P GP model. The hydrogen bonds involving glycerol phosphate are shown as dashed lines. The figure was prepared using "BOBSCRIPT" (46), "MOLSCRIPT" (47), and "RASTER3D" (48,49). 3) evidences local and distant conformational changes caused by the mutation. Special features of the ligand-free ␤S178P mutant structure and differences with the wild-type are described and discussed below.
␤S178P ␣-Subunit-The structure of the ligand-free ␣-subunit is very similar to the wild-type enzyme, indicating that the mutation in the ␤-subunit does not perturb the ␣-subunit. Larger structural differences between the ␤S178P ␣-subunit and the wild-type were found in loop ␣L5 at ␣Asn 157 , as indicated by a large r.m.s. deviation ( Fig. 2A, left panel, peak A). Since this loop region has ambiguous electron density in most structures determined by us (15,27), the structural difference, indicated by a high r.m.s. deviation for ␣Asn 157 , might be caused by distinct interpretations of this part of the electron density map. In one of these possible map interpretations ␣Asn 157 is able to connect the COMM domain to loop ␣L6, the most mobile region of the TRPS enzyme. On one side, in the mutant structure the side chain of ␣Asn 157 is able to form hydrogen bonds to the hydroxyl group of ␤Tyr 181 and to the carbonyl oxygen of ␤Ile 20 . On the other side, in the wild-type structure the carbonyl oxygen of the preceding amino acid ␣Pro 156 has a distance of approximately 4.5 Å to the amide nitrogen of ␣Leu 190 . This distance is too long for a favorable hydrogen bond. However, since this part of the TRPS structure shows the highest flexibility and, therefore, the highest coordinate error, water molecules may be difficult to identify. With the present structural data we cannot exclude the presence of a water molecule mediating the connection between the COMM domain and loop ␣L6 via ␣Asn 157 . Hiraga and Yutani (41) have studied the subunit association of several TRPS interface mutants by titration calorimetry, including two ␣Asn 157 mutants, and found that these mutations did not affect the stimulatory activity, indicating that this subunit interaction is not of crucial importance for the allosteric regulation. This supports our believe that the structural difference at the ␣Asn 157 containing surface loop is of no relevance for the observed different allosteric communication properties of the ␤S178P mutant.
The second structural difference ( Fig. 2A, left panel, peak B) within the ␣-subunit of the ␤S178P mutant is found at ␣Phe 212 , which is modeled as alanine, as this part of the structure also has weak electron density. However, the electron density of the final structure has well defined density for a water molecule, forming hydrogen bonds with the amide nitrogen atoms of ␣Phe 212 and ␣Gly 234 .
␤S178P ␤-Subunit-In the case of the ␤-subunit, several significant differences between the wild-type and mutant structures are observed. Apart from different conformations in a surface loop ( Fig. 2A, right panel, peak A) and the C-terminal end of the ␤-polypeptide chain ( Fig. 2A, right panel, peak D), the ␤Ser 178 3 Pro mutation introduces a different backbone conformation of the neighboring residues of helix ␤H6 (Fig. 2A,  right panel, peak C). In contrast to the molecular simulation FIG. 3. A, C ␣ -atom trace representation of the superposition of the wild-type and ␤S178P mutant structures focused on the ␣-active site, the mutation site and the ␣/␤-subunit interface. The wild-type C ␣atom trace is colored in gray, and loops ␣L2 and ␣L6 of the ␣-subunit are colored in cyan. The ␤S178P mutant C␣-atom trace is colored in red for the ␣-subunit and in yellow for the ␤-subunit. The amino acids at position ␤178 are shown in ball-and-stick representation. B, C ␣ -atom trace representation of the superposition of the wild-type TRPS IAG and ␤S178P IAG structures focused on the ␣-active site, the mutation site, and the ␣/␤-subunit interface. The wild-type C ␣ -atom trace is colored in gray, and loops ␣L2 and ␣L6 of the ␣-subunit are colored in cyan. The ␤S178P mutant C␣-atom trace is colored in red for the ␣-subunit and in yellow for the ␤-subunit. The IAG molecule of the ␤S178P IAG structure and the amino acids at position ␤178 are shown in ball-andstick representation. The figure was prepared using MOLSCRIPT (47) and RASTER3D (48,49). (21), that indicated the presence of only localized structural changes leading to the loss of the hydrogen bond between the native ␤Ser 178 and ␣Gly 181 , the x-ray structure shows that the change at position ␤178 disturbs the normal ␣-helix backbone conformation of helix ␤H6 (amino acids ␤Thr 165 -␤Tyr 181 (15)). The normal hydrogen bond between the amide proton of ␤Ser 178 and the preceding arginine ␤Arg 175 carbonyl oxygen is no longer possible (Fig. 1). The / torsion angles for ␤Arg 175 changed from Ϫ62.8°/Ϫ44.4°in wild-type TRPS to Ϫ75.7°/ Ϫ36.1°in the mutant. By this rotation an unfavorable contact of the new proline ring is avoided. This conformational change is further transmitted to the rest of the N-terminal part of helix ␤H6. Therefore, compared with the wild-type structure the largest C ␣ -r.m.s. deviation differences are found in this region of the mutant ( Fig. 2A, right panel, peak C), whereas the changes are insignificant at the C-terminal end of helix ␤H6 ( Fig. 2A, right panel, and Fig. 3A). A similar conformational change in helix ␤H6 is observed in all three ␤S178P complexes: ligand-free, IAG, and GP. The N-terminal part of helix ␤H6 and the adjacent structural elements of the COMM domain are linked to the metal binding loop (␤Asp 305 -␤Pro 307 ) via a watermediated hydrogen bond network and some hydrophobic side chain interactions. In the ligand-free ␤S178P structure ␤Leu 166 shifted toward the side chain of ␤Phe 306 , and the aspartate ␤Asp 305 side chain points toward the ␣-active side, forming a water-mediated contact to ␤Asp 138 , which is in direct neighborhood to the N-terminal part of helix ␤H6.
In case of the ligand-free mutant structure the "disturbance" caused by the mutation is further transmitted by the neighboring secondary structure elements strand ␤S5 and helix ␤H5 ( Fig. 2A, right panel, peaks B and C, Fig. 3A) to ␤Glu 109 , which is believed to play a crucial role in the catalytic activity and the substrate specificity of the TRPS ␤-reaction (42,43). This may explain the 2-fold lower ␤ activity of the ␤S178P mutant compared with the wild-type enzyme (21).

␤S178P IAG Structure
The structure was solved at a resolution of 1.7 Å. As in the uncomplexed structure, the amino acids of loop ␣L6 (␣Leu 178 -␣Ala 190 ) are not detectable. In addition to the ␣and ␤-subunits, the final model contains 377 water molecules, a sodium ion, and an indole-3-acetylglycine bound to the ␣-active site. This model was refined to R-factors of R work ϭ 21.0% and R free ϭ 26.5%, respectively (Table I). Although all three x-ray data sets are of the same quality (Table I), a B-factor analysis of the structures (data not shown) shows that the sodium binding loop ␤L8, loop ␣L2, and the C-terminal part of the COMM domain (loop ␤L3, strand ␤S4, helix ␤H5, strand ␤S5, and helix ␤H6) have higher mobility in ␤S178P IAG . Furthermore, critical inspection of the final SigmaA-weighted (44) (2mFo Ϫ DFc and mFo Ϫ DF c ) electron density maps indicates a second backbone conformation (which is not included in the final model) for these regions and the loop between ␤Leu 294 and ␤Ser 297 that is consistent with the ligand-free ␤S178P structure. For the following reasons we believe that it is a reduced IAG and not a reduced sodium occupancy that causes the conformational changes within the sodium binding loop. A series of sodium titrations (data not shown) with the wild-type and the ␤S178P mutant in the absence and presence of IAG showed no influence of the ␣-ligand on the sodium affinity, and in contrast to the findings of Marabotti et al. (21), the ␤S178P mutant has the same sodium affinity as the wild-type. Moreover, a cross-check for the ligand occupancy showed a slightly lower IAG occupancy (data not shown), but the resolution of the x-ray data does not allow further refinement.
IAG binds to the mutant in the same manner as to the wild-type (24), but loop ␣L6 is not closed. The indole nitrogen forms a hydrogen bond with aspartate ␣Asp 60 and the acetyl carboxylate group mimics the IGP/IPP/GP phosphate group. The second catalytically important amino acid ␣Glu 49 is modeled in two conformations and forms a hydrogen bond with the IAG acetyl oxygen atom. The implications of this interaction for the ␣-reaction are discussed in the accompanying paper (24). The hydrogen bonding patterns and distances are shown in Fig.  1B and in Table II, respectively. ␤S178P/␤S178P IAG Comparison-The C ␣ -r.m.s. deviation structure comparison of the ligand-free ␤S178P and the ␤S178P IAG structure shows, apart from a large difference in the phosphate binding loop ␣L8 at the ␣-active site (Fig. 2C, left  panel, peak A), which is also observed in case of the analogous wild-type structures (data not shown), only minor differences in the ␣-subunit and also in the COMM domain (Fig. 2C). However, some larger deviations are found at the metal binding site within ␤L8 (Fig. 2C, right panel, peak A). Although the water network at the sodium binding site is similar in both structures, in the ␤S178P mutant structure the ␤Asp 305 side chain points toward the ␤-active site, a conformation also found in the wild-type IAG complex (24). In contrast, in the ␤S178P IAG structure the aspartate side chain is pointing away from the ␤-active side, forming a new hydrogen bond with the carbonyl oxygen of ␤Ser 297 . The origin of the different conformation of the metal binding loop can be traced back to IAG bound at the ␣-active site by the following hydrogen bonding network: the ␤Pro 307 carbonyl oxygen atom, which is also involved in the metal binding, is linked to the ␤Met 287 amide proton. The neighboring glutamine ␤Gln 288 side chain forms an intersubunit hydrogen bond to ␣Asn 104 , and the close by ␣Tyr 102 side chain interacts with the aspartate ␣Asp 60 side chain. Besides this intersubunit pathway starting from glutamine ␤Gln 288 , a second one exists via the ␤Gln 288 carbonyl oxygen atom, which is linked to the amide proton of ␤Glu 295 . The conformational changes are further transmitted by a back- bone shift of the adjacent residues ␤Gln 293 and ␤Ile 294 , which form hydrogen bonds with the loop ␣L2 amino acid ␣Ser 55 and finally via a backbone shift of loop ␣L2 residues with aspartate ␣Asp 60 . The latter forms a hydrogen bond to the indole nitrogen atom of IAG. ␤S178P IAG /WT IAG Comparison-The comparison of mutant and wild-type IAG complexes shows large C ␣ -r.m.s. deviation within helix ␤H6 and also in the sodium binding loop at ␤Asp 305 (Fig. 2B, right panel, peak C, and Fig. 3B), as described above. Surprisingly, the major differences occur at the C-terminal side of helix ␤H6 (Fig. 2B, right panel, peak B), whereas they occur at the N-terminal side in the ligand free structures, taking the mutation site as origin. This difference is caused by the ␣-ligand modifying the conformation of loop ␣L2 via the interaction between the nitrogen atom of the indole ring of IAG and the ␣Asp 60 carboxylate. In this conformation loop ␣L2 residues interact with the N-terminal part of helix ␤H6 (15), and the helix adopts in the ␤S178P IAG structure the same conformation as in TRPS IAG . The proline substitution at ␤178 prevents the formation of a hydrogen bond with the ␣Gly 181 carbonyl oxygen atom, thereby interrupting the ␣ 3 ␤-subunit communication upon IAG binding (see Fig. 3B). On the Cterminal side of the mutation the ␤S178P IAG structure adopts a different conformation than the wild-type, since the prolyl ring causes a slight backbone shift. It is transmitted via residues of the following strand ␤S6 to the surface residue lysine ␤Leu 103 (Fig. 2B, right panel, peak A). The remaining part of the ␤-subunit, especially the ␤-active site, is not altered with respect to the wild-type complex.
The comparison of the mutant ␤S178P IAG and wild-type TRPS IAG structures shows no large differences in the core of the ␣-subunit (Fig. 2B, left panel), indicating that the ␤Ser178 3 Pro mutation does not affect the ␣-subunit grossly. However, although the IAG molecule binds to the mutant in the same manner as to the wild-type enzyme, loop ␣L6 is not closed in the mutant. The most obvious reason for the open loop is the point mutation at serine ␤Ser 178 . As the structure superposition shows no steric hindrance by the introduced ␤Pro 178 , removal of the single hydrogen bond between ␤Ser 178 and ␣Gly 181 prevents stabilization of the closed ␣L6 conformation, which is linked via ␣Thr 183 to the ␣-active site residue ␣Asp 60 . As a result, allosteric signals are no longer able to travel from the ␣to the ␤-active site, and the intersubunit regulation is lost (21).
In the context of an open ␣L6 in ␤S178P IAG it is interesting to note that ligand binding to the ␣-subunit appears to be a two-step process consisting of an initial binding step and a subsequent isomerization to an activated form E* (8) with E representing an open conformation of the ␣-subunit, while E* represents the closed conformation (20). A shift in the equilibrium between E and E*, e.g. by a mutation, would effect k cat correspondingly. A reduction of the occupancy of ␣L6 by a factor of two to three would be enough for it not to be visible in the electron density. To gain more insight in the influence of the ␤S178P mutation on the equilibrium distribution at the ␣-active site, we analyzed the steady state kinetics of the forward and reverse ␣-reactions of the ␤S178P mutant. As can be seen in Table III, there are differences in K m and k cat between mutant and wild-type enzymes, the latter being faster in both reactions. There is no difference in K m for G3P and IGP in the reverse and forward reactions for the mutant, whereas they differ by a factor of 6 in the wild-type. In the latter case the binding energy of the indolyl ring is used for enzymatic rate acceleration (Table III). The rate-limiting step in the forward ␣-reaction is the isomerization from the catalytically inactive (E IGP ) to the active IGP complex (E* IGP ) (8). This step is slowed by a factor of two in the ␤S178P mutant, correspondingly the equilibrium between inactive (open) and active (closed) conformations shifted by a factor of two, which explains the lack of electron density for ␣L6 in the ␤S178P IAG complex. These findings agree with the hypothesis that an open ␣L6 corresponds to the inactive conformation of the ␣-active site (e.g. TRPS IGP ), whereas the closed ␣L6 corresponds to the active conformation (e.g. TRPS IPP ) (20,27). Thus, the combination and comparison of the structural information of TRPS IAG and ␤S178P IAG and the kinetic data allows the direct correlation of the equilibrium distribution between open and closed conformations of ␣L6 with k cat .

␤S178P GP Structure
The mutant GP structure was solved at a resolution of 1.5 Å. As in the ␤S178P IAG structure the amino acids of loop ␣L6 (␣Arg 179 -␣Pro 192 ) are missing in the ␤S178P GP structure. Besides protein atoms, the final model contains 658 water molecules, a sodium ion, and a glycerol 3-phosphate molecule bound to the ␣-active site. This model was refined to R-factors of R work ϭ 17.1% and R free ϭ 20.9%, respectively (Table I). The ␤S178 GP structure (Fig. 1C) shows how GP binds to the ␣-active site. The orientation of the phosphate moiety is similar to that of the IAG carboxylate group, pointing toward ␣S235. The hydrogen bond network between GP and ␣-active site residues is shown in Fig.  1C. The comparison of the distances between selected atoms of GP or IAG and amino acid residues (Table II) indicates a close agreement of the binding modes of both allosteric effectors, despite their chemical differences. Another common feature with the TRPS IAG structure is the lack of loop ␣L6 closure, indicating again the structural basis for the absence of an allosteric effect of GP on the ␤-subunit activity (21). Since a wild-type TRPS GP structure is not available, the ␤S178P GP structure was only compared with the ligand-free mutant structure (Fig. 2D). The C ␣ -r.m.s. deviation plots show that larger differences occur at the ␣-active site at loop ␣L7 (Fig. 2D, left panel, peak A) and loop ␣L8 (Fig. 2D, left panel, peak B). Amino acids of both loops are involved in the binding of the GP phosphate moiety. In case of the ␤-subunit, the C ␣ -r.m.s. deviation plot of the ligand-free and GP mutant structure (Fig. 2D,  right panel) shows only minor differences. Thus, the conforma- tion of the mutant ␤-subunit seems to be independent of GP binding to the ␣-active site, confirming the kinetic finding that the ␤S178P mutant has lost the signaling capability between the ␣and ␤-subunits (21). It is interesting that loop ␣L6 is in the closed conformation in the ␤K87T mutant complexed with GP (␣-site) and serine (external aldimine, ␤-site) (Protein Database code 2TSY (14)). The ␤K87T mutant has no measurable activity in the ␤-subunit but retains ␣-subunit activity (45).

CONCLUSIONS
The correlation of the three-dimensional structure of the ␤Ser 178 3 Pro mutant of tryptophan synthase, reported herein, with the functional and regulatory properties described previously (21) allows to unequivocally attribute a key role in the transmission of ligand binding information between ␣and ␤-sites to the hydrogen bond between ␤Ser 178 and ␣Gly 181 . This is supported by equilibrium studies of a mutant in which ␣Gly 181 was mutated to proline. Also in this case, the allosteric properties of the enzyme are knock-out. 2 The structural element that mediates the communication is loop ␣L6 that, in the wild-type enzyme, closes on the ␣-active site residues upon ␣-ligand binding, whereas it remains open in the ␤S178P mutant. This not only affects the allosteric ␣-␤ interactions but also shifts the equilibrium between active and inactive conformations of the ␣-active site, resulting in different k cat and K m values. In tryptophan synthase, the data presented here indicate that the intersubunit hydrogen bond between the ␣Gly 181 amide proton and the ␤Ser 178 carbonyl oxygen atom represents an "informational pivot point" for ␣ 7 ␤ communication, and a loss of this functionality is enough to breakdown allosteric regulation. The importance of this interaction is also reflected in it being mediated by backbone atoms, since these are invariant toward mutations.