Cryo-crystallography of a True Substrate, Indole-3-glycerol Phosphate, Bound to a Mutant (αD60N) Tryptophan Synthase α2β2 Complex Reveals the Correct Orientation of Active Site αGlu49*

The reversible cleavage of indole-3-glycerol by the α-subunit of tryptophan synthase has been proposed to be catalyzed by αGlu49 and αAsp60. Although previous x-ray crystallographic structures of the tryptophan synthase α2β2 complex showed an interaction between the carboxylate of αAsp60 and the bound inhibitor indole-3-propanol phosphate, the carboxylate of αGlu49 was too distant to play its proposed role. To clarify the structural and functional roles of αGlu49, we have determined crystal structures of a mutant (αD60N) α2β2 complex in the presence and absence of the true substrate, indole-3-glycerol phosphate. The enzyme in the crystal cleaves indole-3-glycerol phosphate very slowly at room temperature but not under cryo-conditions of 95 K. The structure of the complex with the true substrate obtained by cryo-crystallography reveals that indole-3-glycerol phosphate and indole-3-propanol phosphate have similar binding modes but different torsion angles. Most importantly, the side chain of αGlu49 interacts with 3-hydroxyl group of indole-3-glycerol phosphate as proposed. The movement of the side chain of αGlu49 into an extended conformation upon binding the true substrate provides evidence for an induced fit mechanism. Our results demonstrate how cryo-crystallography and mutagenesis can provide insight into enzyme mechanism.

Various biochemical and physical methods, including sitedirected mutagenesis, kinetic analysis, and x-ray crystallography, have revealed many mechanistic aspects of the bifunctional bacterial tryptophan synthase ␣ 2 ␤ 2 complex (EC 4.2.1.20) (for reviews see Refs. 1 and 2). These studies provide evidence that ␣-subunit residues Asp60 and Glu49 catalyze the reversible cleavage of IGP 1 to glyceraldehyde 3-phosphate and indole (␣-reaction; see Scheme 1) (3)(4)(5)(6). Recent x-ray crystallographic studies, however, demonstrated that whereas the side chain of ␣Asp60 interacts with the bound inhibitor IPP, the side chain of ␣Glu49 is too distant (ϳ6 Å) from IPP to play its proposed role (7). However, IPP lacks the hydroxyl groups of the true substrate IGP, and this could affect the orientation of ␣Glu49. To investigate interactions between active site residues of the ␣-subunit and IGP, we have formed the enzyme-IGP complex using the mutant (␣D60N) ␣ 2 ␤ 2 complex, in which the other catalytic residue ␣Asp60 is replaced with Asn. This enzyme has no measurable activity in the reaction catalyzed by the ␣-subunit but retains substantial ␤-subunit activity (5).
Here we present crystal structures obtained by cryo-crystallography of the mutant (␣D60N) tryptophan synthase in the presence and absence of the bound true substrate IGP.

EXPERIMENTAL PROCEDURES
The expression and purification of the mutant (␣D60N) tryptophan synthase ␣ 2 ␤ 2 complex from Salmonella typhimurium has been described (5). Although four different substitutions were made at position 60, only the ␣D60N ␣ 2 ␤ 2 complex yielded crystals suitable for further study. Crystals of the ␣D60N ␣ 2 ␤ 2 complex were grown under the conditions used previously for crystallization of the wild-type enzyme (50 mM N,N-bis(2-hydroxyethyl)glycine, 1 mM Na-EDTA, 0.8 -1.5 mM spermine, and 12% polyethylene glycol 8000 adjusted to pH 7.8 with NaOH) (8) and belong to the space group C2. ␣D60N crystals grown in the presence of Na ϩ were soaked for 1-2 days in a standard K ϩ soaking solution containing 100 mM N,N-bis(2-hydroxyethyl)glycine (pH 7.8 titrated with KOH), 1 mM EDTA, and 20% polyethylene glycol 8000 (9), and these K ϩ -soaked crystals were used for further soaking experiments with ligands (see below). Preliminary x-ray diffraction data collected at room temperature from IGP-soaked ␣D60N crystals indicated that there was no electron density for IGP bound to the ␣-active sites of the enzyme and suggested that the crystalline enzyme still catalyzes the slow cleavage of IGP. Therefore, to eliminate low enzymatic activity of crystalline tryptophan synthase, we have flash frozen substrate-bound ␣D60N crystals and then collected diffraction data at 95 K.
Substrate IGP was prepared enzymatically with tryptophan synthase from indole and glyceraldehyde 3-phosphate (10). IGP was introduced into ␣D60N crystals by soaking the crystals in a IGP soaking solution (0.4 mM IGP with the standard solution) for 1 day. The crystals were then transferred to a IGP soaking solution for 30 min into each of a series of solutions having 5, 10, 15, 20, and 25% glycerol as cryoprotectant and then were flash frozen for data collection. To evaluate the effects of the ␣D60N mutation on the structure, the unliganded ␣D60N crystals prepared as above were also flash frozen and subjected to data collection. Diffraction data were collected at 95 K on a Raxis IIC imaging plate system mounted on a Rigaku RU-200 rotating anode x-ray generator operating at 50 kV and 100 mA. All diffraction data were integrated with DENZO and scaled with SCALEPACK (11). Table I summarizes data statistics and refinement statistics. In refining these structures with X-PLOR (12), the 2.0 Å wild-type structure determined in the presence of K ϩ at room temperature (Protein Data Bank entry 1TTQ) (9) served as a starting model. The starting model was divided into three substructures (corresponding to the ␣-subunit, and N-and C-terminal domains of the ␤-subunits) and subject to a rigid body refinement followed by simulated annealing refinement. Subsequent manual rebuilding was carried out using the program O (13). At this stage, F o Ϫ F c maps of ␣D60N-IGP revealed the bound IGP in the ␣-subunit. An idealized model of IGP was modeled using QUANTA and fitted to the density then refined again by simulated annealing refinement followed by positional and temperature factor refinements.

RESULTS AND DISCUSSION
␣Glu49 Interacts with the True Substrate IGP-Structural comparisons indicate that ␣D60N and the starting model (the wild-type structure) are almost identical within root mean square deviation of 0.52 Å for the main chain atoms of all residues, suggesting that the mutation does not produce any significant structural perturbations. Subsequent comparisons between ␣D60N and ␣D60N-IGP also indicated that there are no noticeable conformational changes induced by the binding of IGP. The ␣D60N-IGP structure is similar to that of the wild type-IPP complex (14) in that loop 6 (residues 179 -191) is invisible and loop 2 (residues 53-62 including residue 60) is highly disordered. Fig. 1A shows the electron density map for IGP and nearby residues in the ␣D60N-IGP structure. The overall binding site is almost identical to that of other structures complexed with IPP that have been determined previously (7,14). However, the ␣D60N-IGP structure reveals that the side chain of ␣Glu49 adopts an extended conformation and is within 2.8 Å from the hydroxyl group of C3Ј of IGP (Fig. 1B). This interaction has been proposed in the mechanism of ␣-reaction (Scheme 1) but has not been seen in other structures with the bound substrate analog IPP (7,14). The positions of Asn60 in the ␣D60N-IGP structure is almost identical to that of the IPP-complexed structure, but Asn60 is very mobile (refined temperature factors for the all atoms in Asn60 are about 80 Å 2 ).
Despite the similarity in the overall binding site, there are differences in positions (0.9 -1.4 Å) of the corresponding atoms  including phosphate, C1Ј, C2Ј, and C3Ј but not in positions of indole ring (Fig. 1B). The torsion angle around the C2Ј-C3Ј bond is 169°for IGP but 28°for IPP. Because there is relatively weak electron density for the C1Ј atom of IGP, we initially modeled IGP according to the torsion angles of IPP. This fitting keeps the hydroxyl group at C2Ј away from its corresponding density, suggesting that the current C1Ј atom represents the reliable position in IGP. Interaction between ␣Glu49 and the C3Ј hydroxyl group is unambiguous based on clear density. The C2Ј hydroxyl group is in D-enantiomeric configuration but does not form any hydrogen bonds with active site residues.
Other Active Site Residues around Bound IGP- Fig. 1 also shows that the phenolic hydroxyl of ␣Tyr175 interacts with the C3Ј hydroxyl group of IGP. However, the finding that the mutant enzyme in which ␣Tyr175 is replaced by Phe (Y175F) has substantial activity indicates that ␣Tyr175 is not essential for catalysis or substrate binding (5). Early studies showed that whereas the Y175C mutant was inactive, a second site revertant (Y175C/G211E) exhibited partial activity in the ␣-reaction (15). Later biochemical investigations of this mutant enzyme combined with computer graphics modeling of the substrate binding site of the ␣-subunit (5) led to the conclusion that the partial restoration of the ␣-subunit activity in the doubly altered second site revertant results from restoration of the proper geometry of the substrate binding site. This conclusion supports the view that Tyr ␣175 serves a structural role but is not an essential catalytic residue.
Catalytic Role of ␣-Subunit ␣Glu49 -The structure of the inactive mutant (␣D60N) with bound IGP reveals two new structural features that were not observed in structures in the presence of substrate analog IPP. These features are the interaction between ␣Glu49 and the hydroxyl group at C3Ј and the location of the hydroxyl groups at C2Ј and C3Ј.
There is substantial evidence from site-directed mutagenesis studies that ␣Glu49 (4, 6) and ␣Asp60 (5) are catalytic bases in the reaction catalyzed by the ␣-subunit. The mechanism of this reaction (Scheme 1), based on previous proposals (3)(4)(5), suggested that ␣Asp60 facilitates tautomerization of the indole ring of IGP (I) to form II and that ␣Glu49 abstracts a proton from the C3Ј hydroxyl group of the glycerolphosphate moiety to form III.
Although previous x-ray crystallographic structures of the tryptophan synthase ␣ 2 ␤ 2 complex demonstrated interaction between the carboxylate of ␣Asp60 and the indole nitrogen of the bound inhibitor (IPP), the side chain of ␣Glu49 was too distant to play its proposed role (7). The carboxylate of ␣Glu49 was folded away from the IPP and was located approximately 6.1 Å from the modeled C3Ј hydroxyl group (Fig. 1B). In the new structure, the side chain of ␣Glu49 interacts with the C3Ј hydroxyl group of IGP as proposed. This interaction is made possible by the movement of the side chain of ␣Glu49 into an extended form in the presence of IGP. These results provide an example of an induced fit mechanism in which an active site residue adopts a catalytically correct orientation when a substrate is bound to the active site. This type of induced fit is much smaller and more localized than the larger changes observed in some other structures which involve domain movement, loop closure, or conversion from an open to a closed conformation. SCHEME 1. Proposed mechanism of the ␣ reaction. The cleavage of indole-3-glycerol phosphate (I) is activated by tautomerization of the indole ring to yield the indolenine tautomer (II), which undergoes aldol cleavage to yield III. These steps are catalyzed by three putative residues, B 1 , B 2 , and B 3 (5). The binding geometry of indole-3-propanol phosphate suggests that ␣Asp60 may serve as B 2 residue, which abstracts the proton on N-1 of the indole ring or forms a strong hydrogen bond to polarize the nitrogen atom. ␣Glu49 is thought to serve as B 3 , which accepts the proton on the hydroxyl group. ␣Glu49 may also serve as B 1 that protonates the C3 position of the indole moiety. (Reproduced with permission from Ref. 2).