Structural Basis for Activation of the Thiamin Diphosphate-dependent Enzyme Oxalyl-CoA Decarboxylase by Adenosine Diphosphate*

Oxalyl-coenzyme A decarboxylase is a thiamin diphosphate-dependent enzyme that plays an important role in the catabolism of the highly toxic compound oxalate. We have determined the crystal structure of the enzyme from Oxalobacter formigenes from a hemihedrally twinned crystal to 1.73 Å resolution and characterized the steady-state kinetic behavior of the decarboxylase. The monomer of the tetrameric enzyme consists of three α/β-type domains, commonly seen in this class of enzymes, and the thiamin diphosphate-binding site is located at the expected subunit-subunit interface between two of the domains with the cofactor bound in the conserved V-conformation. Although oxalyl-CoA decarboxylase is structurally homologous to acetohydroxyacid synthase, a molecule of ADP is bound in a region that is cognate to the FAD-binding site observed in acetohydroxyacid synthase and presumably fulfils a similar role in stabilizing the protein structure. This difference between the two enzymes may have physiological importance since oxalyl-CoA decarboxylation is an essential step in ATP generation in O. formigenes, and the decarboxylase activity is stimulated by exogenous ADP. Despite the significant degree of structural conservation between the two homologous enzymes and the similarity in catalytic mechanism to other thiamin diphosphate-dependent enzymes, the active site residues of oxalyl-CoA decarboxylase are unique. A suggestion for the reaction mechanism of the enzyme is presented.

appears to have evolved from an FAD-binding site, which is present in the related enzymes pyruvate oxidase (POX) (21) and acetohydroxyacid synthase (AHAS) (22). Kinetic studies stimulated by this structural observation support the hypothesis that ADP is a high affinity activator of the enzyme, a finding that may be of physiological relevance in O. formigenes.

MATERIALS AND METHODS
Expression and Purification-A BL21(DE3) Escherichia coli expression strain transformed with pET-9a carrying the wild type oxalyl-CoA decarboxylase (oxc) gene from O. formigenes (20) was generously supplied by Dr. Harmeet Sidhu (Ixion Biotechnology, Inc., Alachua, FL). Cells were grown in LB broth to an A 600 of 0.3-0.6, and protein expression was induced by the addition of isopropyl 1-thio-␤-D-galactopyranoside to a final concentration of 0.4 mM. Cells were harvested by centrifugation after 3 h at 37°C (A 600 1.8 -2.0). Cell pellets from 1 liter of culture were suspended in 50 ml of lysis buffer (100 mM KH 2 PO 4 , pH 7.2, 1 mM dithiothreitol, 10 mM MgCl 2 ) and sonicated. The lysate was clarified by centrifugation, and supernatant was loaded onto a 20-ml Blue-Sepharose fast flow affinity column equilibrated with Buffer A (25 mM NaH 2 PO 4 , pH 7.2, 0.1 M NaCl) at a flow rate of 4 ml/min. The column was washed with Buffer A, and OXC was then eluted with Buffer B (25 mM NaH 2 PO 4 , pH 7.2, 2 M NaCl). Fractions containing OXC were combined, and buffer was exchanged using a Sephadex G25 HiPrep 26/10 desalting column equilibrated with Buffer C (25 mM NaH 2 PO 4 , pH 6.5) at a flow rate of 10 ml/min. OXC was eluted with Buffer C, and fractions containing the enzyme were loaded on a Q-Sepharose high performance anion exchange column equilibrated with Buffer C at a flow rate 5 ml/min. Purified OXC was eluted using a 10 -40% gradient of Buffer D (25 mM NaH 2 PO 4 , pH 6.5, 0.5 M NaCl), and fractions were pooled. Purified OXC was concentrated in an Amicon Stirred Cell and stored at Ϫ80°C as previously described (23).
Crystallization-OXC was dialyzed against 50 mM MES buffer, pH 6.5, and 10 mM ThDP and 10 mM MgCl 2 , and 1 mM coenzyme A was added. The solution was concentrated to a final protein concentration of 5 mg/ml before crystallization by the hanging drop vapor-diffusion technique. Full details about crystallization screening and optimization have been published elsewhere (23). Useful diffraction-quality crystals were obtained after ϳ4 days, with a precipitating solution containing 27% polyethylene glycol 550 monomethyl ether, 100 mM Bis-Tris buffer, pH 6.5, and 50 mM CaCl 2 .
Data Collection and Processing-Data were collected to 1.73 Å in a nitrogen stream at 110 K at beamline ID23-1, European Synchrotron Radiation Facility, Grenoble, France, as described (23). No additional cryoprotectant was needed, and the crystals were directly flash-frozen in the nitrogen stream. All images were processed using MOSFLM (24), and the unit-cell parameters were determined using the autoindexing option. The data set was scaled using the program SCALA (25) (TABLE  ONE).
Structure Solution and Crystallographic Refinement-Full details about the structure solution procedure, including how twinning was characterized qualitatively and quantitatively, have been published elsewhere (23). Briefly, the data showed 622 symmetry, but hemihedral twinning was detected using the intensity statistics and distributions obtained from the CCP4 programs TRUNCATE (25,26) and DETWIN (25,27). The Yeates and Fam Merohedral Crystal Twinning Server was used as an additional indicator and for calculations of the twin fraction where I i is the intensity measurement for a reflection, and ͗I͘ is the mean value for this reflection.

Crystal Structure of Oxalyl-CoA Decarboxylase
(␣) (28). An incomplete low resolution model of OXC has previously been determined from crystals in space group P4 2 2 1 2 diffracting to 4.1 Å resolution (23). This model, covering 468 of 568 residues, was obtained by molecular replacement using a polyalanine model of AHAS (22), omitting loops and the C-terminal domain. The low resolution model of OXC was subsequently used for molecular replacement in the high resolution data with the program MOLREP (25,29) searching for a dimer in all possible true space groups. The molecular replacement search was initially performed in twinned data without any success and was, therefore, continued in detwinned data generated by DETWIN. The true space group P3 1 21 was deduced based on packing considerations and statistical comparisons. The structure of OXC in the P3 1 21 crystal form was refined to 1.73 Å with a fixed twin fraction of ␣ ϭ 0.43 using the protocol for twinned data in the CNS software (30). A test set of 4.6% of the reflections was excluded to monitor the R free value. Care was taken to keep the twin-related reflections together in either the test or work set in order not to bias the refinement. Refinement with simulated-annealing, individual B-factor, and energy minimization procedures was interspersed with rounds of manual model building in A -weighted 2F o Ϫ F c electron density maps with the graphics program O (31) (TABLE TWO). Initially, non-crystallographic symmetry restraints were applied but were later abandoned. All ligand library files were created with the Dundee PRODRG2 server (32). Water molecules were added automatically in CNS at F o Ϫ F c difference density peaks Ͼ3 followed by manual inspection and additional assignment in the molecular graphics program Coot (33). After convergence of the refinement in CNS, refinement of both the coordinates and the twin fraction was performed in SHELXL (34). The test set was kept intact during all refinement steps. The refined twin faction ␣ ϭ 0.440 was used in the final refinement steps in CNS when double conformations for some side chains were added.
Structural Analysis-The final model was validated with the PRO-CHECK (35,36) and CNS (30) programs. Structural comparison with homologous structures were carried out using the SSM superposition in Coot (33) and the least square facility in O using default parameters (31). All figures portraying protein models were prepared using PyMOL (www.pymol.org).
Enzymatic Assay-Recombinant OXC (0.775 g/ml) was assayed in 60 mM KH 2 PO 4 , pH 6.7, 60 M ThDP, 6 mM MgCl 2 , with a final reaction volume of 100 l. The reaction was started by adding 10 l of oxalyl-CoA at appropriate concentrations to reaction mixture preincubated at 30°C for 2.5 min. Oxalyl-CoA was synthesized and purified following literature protocols (14,37) and diluted in 50 mM NaH 2 PO 4 , pH 4.5. The reaction was quenched by adding 11.1 l of 20% acetic acid to assay mixture, and the initial rate of formyl-CoA formation was analyzed by reverse-phase HPLC using a modification of a previously published procedure (14). Briefly, 75-l aliquots of quenched reaction mixture were injected onto a C 18 analytical column (Dynamax Microsorb 60 -8 C 18 , 250 ϫ 4.6 mm) equilibrated with 98% Buffer F (25 mM sodium acetate, pH 4.5) and 2% Buffer G (20% Buffer F; 80% CH 3 CN) at a flow rate of 1 ml/min. After injection, the proportion of Buffer G was raised to 6% over a 12-min period followed by a step to 95% Buffer G that was continued for 2 min before returning to 2% Buffer G. Under these conditions ThDP eluted after 3.7 min, oxalyl-CoA eluted after 6.1 min, free CoA eluted after 9.9 min, and formyl-CoA eluted after 11.3 min. The concentration of CoA derivatives in the aliquots was determined as described previously (14).
Activation and Inhibition Studies-Recombinant OXC was assayed as described above in the presence of varying concentrations of ADP, ATP, and FAD suspended in 100 mM KH 2 PO 4 , pH 6.7, and free CoA diluted in 50 mM NaH 2 PO 4 , pH 4.5.

RESULTS AND DISCUSSION
Expression, Purification, and Characterization of Recombinant OXC-The recombinant, wild type enzyme was expressed in BL21(DE3) E. coli cells and purified using a two-step purification procedure that was significantly different from that published previously (15). Thus, after an initial dye-affinity chromatography step, wild type OXC could be purified to near homogeneity by anion exchange chromatography in yields of typically 5-10 mg/liter of culture (Supplemental Table S1). Gel filtration experiments were performed to determine the oligomeric state of the recombinant OXC using a size exclusion chromatography column calibrated with a variety of molecular mass standards. OXC eluted in a single peak with a retention time corresponding to 243 kDa ( Fig. 2A). Because the OXC monomer has a molecular mass of 60,691 Da (20), this finding suggests that the purified enzyme is a tetramer in solution rather than the homodimer suggested in earlier studies of recombinant OXC (20).
The activity of the recombinant, wild type OXC was measured using an HPLC-based assay rather than the coupled assay used to characterize the native enzyme (15) so as to measure formyl-CoA production directly. All reactions were performed at pH 6.7 to minimize the rate of uncatalyzed thioester hydrolysis. Under these conditions, product formation was linear over a period of 3 min (Fig. 2B), permitting the derivation of Michaelis-Menten steady-state kinetic parameters by fitting to standard equations (Fig. 2C) (39). Under our assay conditions, the apparent K m of oxalyl-CoA was 23 Ϯ 3.5 M, which is 10-fold lower than reported for native OXC (15), and the turnover number (k cat ) was 88 s Ϫ1 . This difference in the apparent K m values likely reflects technical difficulties in the coupled assay used in the earlier study (15) because formate dehydrogenase activity appears to be affected by the presence of CoA thioesters.
Sequence alignments (Supplemental Fig. S1) are consistent with the hypothesis that OXC is evolutionarily related to AHAS (22,40), the ThDP-dependent enzyme that plays an important role in the biosynthesis of branched chain amino acids (41) and which is the target for sulfonylurea-based herbicides (42). This observation was intriguing in that AHAS contains a "vestigial" FAD-binding site even though the redox co-factor plays no role in the catalytic mechanism and presumably is present merely to stabilize the three-dimensional fold of the protein (43,44). Because electronic spectroscopy provided no evidence for the presence of bound flavin in OXC, 3 we hypothesized that the FAD-binding site in AHAS had been "co-opted" as a binding site for the CoA portion of the substrate and examined the ability of free CoA and FAD to inhibit OXC activity. These experiments showed that although FAD does not inhibit OXC (data not shown), free CoA is a mixed inhibitor of OXC with respect to oxalyl-CoA, exhibiting K i and K i Ј values of 400 and 270 M, respectively (Fig. 2D).
Significant functional insights have been provided by the availability of crystal structures of many different ThDP-dependent enzymes, of which some of the most extensively studied are transketolase (45), POX (21), AHAS (22), the E1-component of pyruvate dehydrogenase (46), branched-chain ␣-keto acid dehydrogenase (47), benzoylformate decarboxylase (BFD) (48), and the pyruvate decarboxylases from Zymomonas mobilis (zPDC) (49) and yeast (yPDC) (50). To provide a high resolution three-dimensional structure of OXC from O. formigenes, we have developed crystallization conditions for the enzyme and were able to phase the collected x-ray data, as reported elsewhere (23).
Quality of the Electron Density Map and the Model-Electron density maps calculated from the crude models at different stages of refinement allowed tracing of the gaps in the polypeptide chain and assignment of the sequence where not previously clear. The asymmetric unit contains a homodimer of two OXC monomers related by a non-crystallographic 2-fold symmetry parallel to the c axis. The C ␣ atoms of the two monomers superimpose with a root mean square deviation of 0.22 Å. The model contains 2 ϫ 546 amino acids (comprising residues 7-552), two 3 P. Moussatche, unpublished results. thiamin-2-thiazolone diphosphate molecules, two Mg(II) ions, and two ADP molecules. A total of 655 ordered water molecules and 10 residues with alternative conformations were built. No density could be seen for the 6 first and 16 last residues, which are likely disordered. The structure of OXC was refined to a twinned R-factor 4 of 15.1% and an R free of 17.5% with good stereochemistry (TABLE TWO). 89.6% of the amino acid residues in the final model were located in the most favorable regions of the Ramachandran plot. We note that one residue in the active site, Tyr-483, is located in the disallowed region in both subunits but is well defined in the electron density.
Overall Structure-The monomer of OXC shows the general ThDP binding fold containing three ␣/␤-type domains, designated PYR (residue 1-192), R (residue 193-368), and PP (residue 369 -568) (51). Each domain is composed of a central six-stranded parallel ␤-sheet surrounded by ␣-helices on both sides (Fig. 3A). The functional dimer arrangement is common to all ThDP-dependent enzymes (51) and is conserved also in OXC. In the dimer the C-terminal ends of the ␤-strands of the PYR domain in one subunit faces the C-terminal ends of the ␤-strands of the PP-domain in the second subunit, thus forming two equivalent active sites (Fig. 3B). The overall structure is highly similar to that observed for other members of the ThDP-dependent enzyme family. OXC shows the highest sequence identity, 23%, to AHAS (see Supplemental Fig. 1), and despite this moderate overall sequence homology, superimposition of the C ␣ trace of a subunit of OXC with the holo-structure of AHAS from Saccharomyces cerevisiae (PDB code 1jsc) matches 465 residues with an root mean square deviation of 1.74 Å (Fig.  4A) and 898 C ␣ atoms with an root mean square deviation of 1.89 Å for the dimer. Comparison of OXC with BFD (1bfd), zPDC (1zpd), and POX (1pow) shows slightly less similarity (1.88 Å for 432 C ␣ atoms, 1.96 Å for 414 C ␣ atoms, and 1.99 Å for 443 C ␣ atoms, respectively). Because of the marked structural similarity between OXC and AHAS, we investigated if OXC was also a target for herbicide inhibition. Under our assay conditions, no changes in enzyme activity were observed in the presence of chlorimuron ethyl or sulfometuron methyl (data not shown). These results were consistent with the absence in OXC of the amino acids required for herbicide inhibition of AHAS, Arg-380, Lys-251, Trp-586, Pro-192, and Val-583 (52,53). Modeling of the herbicide chlorimuron ethyl into the OXC crystal structure also resulted in stereochemical clashes.
The structure of a mobile loop, comprised of residues 479 -504, is unique to OXC in that it is opened up and lacks any discernible secondary structure. In other homologous ThDP-dependent decarboxylase structures, this region folds down over the active site, forming a helixloop structure. Close to where the helix-loop is found in the homologous structures, the last ordered C-terminal residue of OXC is located (Fig. 4B). The remaining 16 residues are disordered. In AHAS, the helixloop as well as the C terminus is disordered in the holo structure (22), but in complex with several different herbicides these regions become ordered and cover the active site (54). This finding on AHAS was followed up in a recent study where the "mobile" loop and C-terminal "lid" were deleted and shown to be important for stabilization of the active dimer and ThDP binding (55). It is probable that, during the catalysis of oxalyl-CoA decarboxylation, the mobile loop in OXC changes conformation, and the 16 C-terminal disordered residues fold up and form a lid over the opening to the active site, forming the required hydrophobic environment.
Quaternary Structure-There are differences in the overall quaternary structure between ThDP-dependent enzymes. For example, although AHAS (22) and transketolase (46) are dimers, POX (21), zPDC (49), and BFD (48) are all tetramers. As described above, gel filtration experiments suggest that the oligomeric state of OXC is a tetramer in solution. This proposal is further substantiated by the crystal structure in which the crystallographic 2-fold axis generates the biological tetramer from the dimer in the asymmetric unit (Fig. 3B). The arrangement of the tetramers differs among POX, BFD, and zPDC (21,48,49). Superimposition of the OXC tetramer with each of these three structures shows that OXC forms the same compact dimer of dimers as POX and BFD.
ThDP Binding-The ThDP cofactor is bound in a cleft between the PYR and PP domain from two different subunits (Fig. 3B). A strong electron density feature close to the C2 position on the thiazolium ring was interpreted as the thiazolone derivative of ThDP (Fig. 5A) (56), which is likely a result of exposure to the intense x-ray radiation and is, therefore, an artifact of the experiment. Despite considerable differences in quaternary structure and lack of sequence homology, the ThDPdependent enzymes bind ThDP and the divalent Mg 2ϩ ion cofactor within a similar framework (57), which is crucial for proper orientation of the coenzyme in the V-shaped conformation. All of the identified common family features are shared by OXC. The pyrophosphate is bound to the PP domain at the cleft formed when the loops from two ␤-strands cross over to different sides of the sheet and the N termini of the following two helices provide hydrogen bonds. Further stabilization is obtained through hydrogen bonds to the polar residues Asn-402 and Tyr-377 and via the Mg 2ϩ ion, which is bound to the PP domain by the conserved ThDP fingerprint (58,57). The thiazolium ring is located between the subunits, and Met-428 contributes with a conserved hydrophobic interaction that stabilizes the V conformation. The pyrimidine ring is predominantly bound by the PYR domain. The binding includes only two conserved interactions, a hydrogen bond from a glutamic acid (Glu-56 in OXC) to the N1Ј of the ring and a hydrogen bond from N4Ј to the carbonyl of Gly-426 in the PP domain.
Active Site Residues-Although the mode of cofactor binding and fold is highly conserved among ThDP-dependent enzymes throughout evolution, there are significant differences in the active sites, and no conserved residues other than the few involved in cofactor binding can be found. The putative active site of OXC (see below) contains three residues that could possibly contribute to catalysis; Tyr-120, Glu-121, and Tyr-483 are all located in the proximity of the catalytic site (Fig. 6). All these residues are conserved among oxalyl-CoA decarboxylases from different bacterial species, with the exception of Glu-121, which is exchanged for a glutamine in Mycobacterium tuberculosis. The corresponding residues that have been implicated to assist in the reaction in PDC are His-113, His-114, and Glu-473 (49,50), but in BFD, another decarboxylating ThDP-dependent enzyme of known structure, there are no corresponding polar residues.
Catalytic Mechanism-The first steps in OXC-catalyzed decarboxylation, as in all ThDP-dependent enzymes, involves activation of the cofactor by the enzyme (59 -62) (Fig. 7). Thus, the conserved Glu-56-N1Ј hydrogen bond interaction stabilizes the tautomeric form of the cofactor in which the N4Ј amino group is converted to an imino group, and in the process one proton is displaced from the N4Ј atom to the solvent, perhaps assisted by Glu-121. The V conformation of the ThDP, constrained by Met-428, positions the 4Ј-imino group sterically close for direct deprotonation of the C2 atom of the thiazolium ring. A conserved hydrogen bond between the formed imino group and the carbonyl group of Gly-426 may further assist by positioning the lone pair of the imino group favorably for proton abstraction at C2 and thereby form the highly nucleophilic ylid structure.
After the C2 carbanion is formed, the cofactor attacks the carbonyl carbon of the thioester in oxalyl-CoA, and the covalently attached intermediate can be modeled into the active site (Fig. 6). The carboxyl group of the intermediate forms hydrogen bonds to Tyr-483 and to the main chain amino group of Ile-34. Upon formation of the covalent bond, the developing negative charge of the carbonyl oxygen on the ␣ carbon atom of the substrate has to be electrostatically stabilized. In the model there are two candidate functional groups that might act to perform this task, the Tyr-120 side chain and the protonated 4Ј-imino group of ThDP. The 4Ј-NH 2 was proposed earlier to stabilize the negative charge of this oxygen in transketolase and other enzymes (61,63).
Cleavage of the substrate with the formation of CO 2 gives a covalent intermediate, the ␣-carbanion/enamine. The resonance contribution of the reactive high energy ␣-carbanion form and the low energy planar enamine to this enzyme intermediate has long been debated in ThDP chemistry (60) and might indeed be different for ThdP enzymes catalyzing different reactions. Recent studies of BFD suggest that the enzyme prevents the intermediate from taking the planar enamine form by avoiding orbital overlap through hydrogen bonding to the hydroxyl group on the ␣ carbon atom (64). In OXC this stabilization can be performed by Tyr-120 and the 4Ј-NH 2 . From this, it would follow that the intermediate stays partly tetrahedral, and protonation of the ␣-carbon in the next step has to take place from the same direction as the CO 2 previously left. None of the three residues, Tyr-120, Glu-121, and Tyr-483, is in proper distance from the ␣-carbon to be involved; however, a bound water molecule is suitably positioned for the ␣-carbon to attack and abstract a proton. This water molecule is anchored by hydrogen bonds to the side chains of Tyr-120 and Glu-121 and to the carbonyl of Ile-34, with the proceeding Pro-35 in cis-conformation, which is unique to OXC. A bound water molecule in the same position is found in zPDC (49), where it forms hydrogen bonds with Asp-27, His-114, and Thr-72 (zPDC numbering). The yPDC double mutant D28A/E91D (yPDC numbering) is not able to release acetaldehyde from the intermediate (65), which suggests that the same water molecule is involved in this protonation event in PDC. In BFD, the N⑀2 of His-70 is close to the same position as this water molecule but is suggested to be involved in protonation of the carbonyl oxygen of the intermediate before decarboxylation, i.e.
In the last step before the second product, formyl-CoA, leaves, the hydroxyl group of the ␣ carbon has to return the proton. The acceptor is most probably the same group that donated the proton, the 4Ј-imino group. In the suggested mechanism, none of the residues in the active site (except Glu-56) is strictly needed for catalysis. This is in accordance with results on yPDC for which mutations of any single active site res-  idue never led to more than a 100 -1000-fold reduction in k cat or k cat /K m (67).
ADP Binding and OXC Activation-These structural studies confirm that OXC is homologous to AHAS and, therefore, belongs to the family of ThDP-dependent enzymes that are related to pyruvate oxidase (43). As described above, most enzymes in this family possess a FAD-binding site (68), the exception being some variants of acetolactate synthase (69) in which the cognate structural cavity is filled by amino acid side chains (38). Although FAD plays a role in electron transfer in the POX-catalyzed formation of acetyl-CoA from pyruvate (21), this co-factor appears merely to stabilize the structure of other enzymes, such as AHAS (70) and glyoxylate carboligase (71), in this evolutionary family. We anticipated that OXC would also contain this vestigial FAD-binding site, and the absence of bound flavin in the purified protein, therefore, suggested (given the structural similarity of the ADP-like, distal regions of FAD and CoA) that this site had been co-opted during OXC evolution (72)(73)(74) to function as a binding pocket for the CoA moiety of the substrate. Therefore the presence of strong electron density in the OXC R domain at a region cognate to the binding site of the non-catalytic FAD in AHAS (22) was interesting, especially because a surplus of CoA was included in the crystallization mixture. A more detailed examination, however, showed that the density could only be interpreted as an ADP molecule bound within the Rossmann fold (Fig. 5B) at the same site as the ADP part of the non-catalytic FAD in AHAS (22) (Fig. 4A). Thus, ADP is buried in a cleft between the Pyr domain and the R domain of the same subunit, with the latter contributing most of the interactions (Fig. 8). The adenine ring is wedged between two isoleucine side chains; the ribose hydroxyl groups make hydrogen bonds to the side chains of Asp-306 and Arg-160, with the pyrophosphate bound by water molecules and the positively charged side chains of Arg-282 and Lys-222.
Our original hypothesis that the site could bind CoA was investigated. Modeling oxalyl-CoA into the ADP-binding site does permit the thioester of the extended substrate to be located adjacent to the C2 position of ThDP without further adjustment. However, these studies also showed that the 3Ј-ribose phosphate of the CoA moiety can be accommodated only after substantial conformational rearrangement of the Arg-16, Arg-282, and Asp-306 side chains (Fig. 8).
Given the presence of ADP in the OXC crystal structure despite the fact that no ADP was added during enzyme purification or crystallization, we assumed that this ligand must have been supplied by the E. coli cells during expression. We, therefore, tested the effects of this small molecule on the kinetics of OXC-catalyzed decarboxylation (Fig. 9). In these experiments, which were carried out in the presence of saturating oxalyl-CoA, endogenous ADP reproducibly increased the specific activity of the recombinant enzyme at micromolar concentrations, whereas ATP had no effect on the enzyme activity. Because millimolar levels of ADP are estimated to be present in E. coli cells, we therefore believe that recombinant OXC is saturated with ADP before purification. This also FIGURE 8. ADP binding to OXC. Stereoview of interacting residues and ordered water molecules. explains the observation that OXC exhibits high specific activity in cell lysates (Supplemental Table S1) that is attenuated by subsequent purification and storage, 3 presumably due to the release of ADP from the protein. Although the mechanism by which ADP stimulates activity remains to be established, it is probable that ADP binding stabilizes the active conformation of the enzyme. The stimulation of decarboxylase activity by ADP may have physiological importance since oxalyl-CoA decarboxylation is an essential step in ATP generation in O. formigenes.
If ADP binds within the vestigial FAD-binding site of the enzyme, this would imply that there is a second site for oxalyl-CoA. Unfortunately, despite several trials we have not been able to obtain co-crystals exhibiting electron density corresponding to this substrate analogue. The precise location of binding site for oxalyl-CoA, therefore, remains elusive, although we note that there is a large cavity formed by the R domain and the PP domain of the same subunit that might accommodate the substrate (Fig. 4). This cavity contains several lysine and arginine side chains that could participate in binding of the phosphate groups, and binding of oxalyl-CoA in this site would not prevent either the mobile loop or the conformationally flexible C-terminal residues to close down on the active site during catalysis. Importantly, this cleft is smaller in AHAS, where the isoalloxazine ring of FAD and an extra loop of seven residues (beginning at AHAS residue 264) partly fill the cognate cavity.
Implications of the Structure for Half-site Reactivity in OXC-Recent experiments on the ThDP-dependent E1 component of pyruvate dehydrogenase have suggested that there is communication between the two active sites in the homodimer (75,76). More specifically, the existence of a network of hydrogen bonding interactions has been identified as a "proton wire" that mediates the shuttling of a proton between active sites, giving rise to "half-site" reactivity (75). Taken together with findings on other ThDP-dependent enzymes (77,78), such as differential kinetics of co-factor binding to the two sites (79) and co-crystal structures in which substrate analogs are observed to bind in only one of the possible active sites (80,81), this work on pyruvate dehydrogenase raises questions concerning the generality of this molecular mechanism in mediating active site communication. There is no evidence, however, for the existence of such a proton wire in the crystal structure of the OXC homodimer. In pyruvate dehydrogenase the active sites are linked by an acidic solvated tunnel with six glutamic acid residues, two aspartic acid residues, and a magnesium ion. The corresponding solvated tunnel in OXC includes six ionizable residues, two glutamic acids, and four histidine residues, but no hydrogen bond network connects the active sites.