Triosephosphate isomerase (TIM, E.C. 126.96.36.199) catalyzes the reversible isomerization of dihydroxyacetone phosphate to D-glyceraldehyde 3-phosphate in the glycolytic pathway.1 It is a well-studied enzyme conserved in function across eukarya, bacteria, and archaea. The “TIM barrel” represents one of the most common protein folds and has been found in variety of proteins with different functions.2 All known TIM structures comprise approximately 250 amino acid residues per monomer and function as homodimers.3 It has been proposed that the isomerization reaction proceeds through an enediolate intermediate formed by substrate deprotonation.4 Crystal structures of TIM and its complexes with transition-state analogs have revealed conformational changes associated with substrate binding and a glutamate residue conveniently positioned to abstract and transfer a proton from one carbon of the substrate to another.5, 6
The crystal structure of TIM from Caenorhabditis elegans (ceTIM) presented here was solved as a part of the structural genomics project on the C. elegans genome with 19,099 predicted genes.7 We employed a recently described phasing approach based on derivatization with halides, which seems to be applicable for high-throughput crystallographic projects.8 Both crystallographically independent ceTIM molecules have been found in the closed conformation with one sulfate and one acetate ion in the substrate-binding site. The sidechain of the catalytic residue Glu164 has been refined in a dual conformation, which is relevant for the mechanism of proton transfer.
The protein expression in Escherichia coli and purification were performed as reported previously9 (also see http://sgce.cbse.uab.edu). As a result of Gateway cloning,10 the protein was produced with a hexahistidine tag and an eight-amino-acid peptide at both the N-terminus and the C-terminus. Initial crystallization conditions were obtained with the screening kit WIZARD I (Emerald BioStructures) and also with NATRIX (Hampton Research). Diffraction-grade crystals were grown at 22°C by vapor diffusion in hanging drops consisting of 3 μL protein solution and 3 μL well solution (2 M ammonium sulfate and 50 mM sodium acetate buffer, pH 5.5). The protein stock solution at 16 mg/mL was in 15 mM sodium N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), pH 7.5. The crystals are monoclinic, space group P21, a = 36.40 Å, b = 64.37 Å, c = 105.71 Å, β = 91.5°, and the asymmetric unit contains two protein molecules.
One such crystal was soaked for 10 min in a mother liquor with 0.5 M sodium iodide, dipped in mother liquor with 25% (v/v) glycerol, and flash-frozen in liquid nitrogen. Anomalous data to 2-Å resolution were collected at −170°C on an Raxis IV using the CuKα radiation generated by a rotating anode (Table I). No attempts were made to collect the Bijvoet pairs close in time. The data were processed in Denzo/Scalepack,11 and the structure was solved by the single-wavelength anomolous diffraction (SAD) method in SOLVE.12 The anomalous Patterson map revealed 12 iodide sites with partial occupancies that were used for protein phasing. After density modification in RESOLVE,12 the refined phases provided a quality map suitable for automatic model building of approximately 75% of the amino acid residues by RESOLVE. A native diffraction data set was collected to 1.7 Å at the Stanford Synchrotron Radiation Laboratory (SSRL) beam line 9-1, with the wavelength at 0.976 Å. The native data were processed in HKL200011 and used for further structure refinement in Crystallography & NMR System (CNS).13 The final model [Protein Data Bank (PDB) code 1MO0] consists of two protein chains with a total of 502 amino acid residues, 461 water molecules, 5 sulfate ions, and 2 acetate ions. The crystallographic R-factor is 18.3%, and R-free is 21.3%, with no cutoffs. The model was validated in MolProbity14 and meets standards for the high-resolution structures.
Table I. Data Collection and Refinement Statistics
The R-free was calculated using a 5% subset of randomly selected reflections.
Soaking with iodide proved to be a quick and robust phasing method that can even use data collected on an in-house X-ray source. The refinement against native data resulted in defined protein chains without breaks in the 2Fo-Fc electron density map at 1.1σ level. However, the N- and C-terminal peptides from the expression vector were just partially resolved in chain A and completely disordered in chain B. In course of the refinement, both active sites revealed separated electron density residuals that were clearly compatible with sulfate and acetate ions acquired from the crystallization. The acetate ions were positioned with respect for optimal hydrogen bonding, whereas positioning of sulfate ions was prompted by the shape of electron density residuals. In addition, a significant residual in the proximity of Glu164 was consistent with a double conformation of the sidechain of Glu164.
The overall structure and active site of ceTIM are, as expected, very similar to other fully liganded TIMs (Fig. 1). A least-squares fit of Cα atoms between the homodimers of ceTIM (PDB code 1MO0) and a complex of chicken TIM with phosphoglycolohydroxamate15 (PDB code 1TPH) resulted in a root-mean-square (RMS) deviation of 0.94 Å. Both ceTIM molecules assume the “closed” conformation, which is most apparent at loop 6 (residues 164–176), which moves by almost 7 Å toward the active site compared to the “open” conformation of an unliganded TIM (PDB code 1YPI). It has been shown previously that the sulfate alone can elicit the closed TIM conformation16; however, the sidechain of the catalytic glutamate was not found in the well-defined “swung-in” conformation observed in high-resolution structures of TIM complexes with transition-state analogs.15, 16
Remarkably, in this structure, both sulfate and acetate ions are bound in a mode that closely resembles binding of transition-state analogs, particularly phosphoglycolic acid (not shown) and phosphoglycolohydroxamate [Fig. 1(b)]. In both subunits of ceTIM, the sulfate ion occupies the phosphate-binding site, and the acetate simulates the sugar part of the substrate, interacting with active site residues Asn10, Lys12, His94, and Glu164. The key residue Glu164 has been refined with 75% of the side chain in a nearly swung-in conformation and 25% in a new conformation, which has not yet been observed in TIM structures. Superpositions with other TIM structures with transition state analogs suggest that, whereas the carboxylate of Glu164 in the swung-in conformation is positioned to abstract a proton from one carbon atom of the substrate, as proposed earlier, the new conformation shown here brings the carboxyl group of Glu164 within a convenient distance for the proton readdition to the next carbon atom of the substrate, as required for the isomerization. It has also been shown that there is a proton exchange, which includes solvent molecules and possibly other groups of the active site.17, 18 Thus, most protons are not transferred from one carbon atom to another in a single step. However, the sidechain of Glu164 is observed in conformations that can apparently mediate the initial and final steps of the proton transfer between carbon atoms of the substrate.
We thank the Structural Genomics of Caenorhabditis elegans (SGCE) production group for providing the protein.