Portal de Periódicos da CAPES

Crystal structure of a component of glycine cleavage system: T-protein from Pyrococcus horikoshii OT3 at 1.5 Å resolution

2004; Wiley; Volume: 58; Issue: 3 Linguagem: Inglês

10.1002/prot.20345

1097-0134

N.K. Lokanath, Chizu Kuroishi, Nobuo Okazaki, N. Kunishima,

Alcoholism and Thiamine Deficiency

The glycine cleavage system (GCS; EC 2.1.2.10) is a multicomponent enzyme system that catalyzes the cleavage of glycine to yield carbon dioxide, ammonia, the methylene carbon unit in 5,10-CH2-tetrahydrofolate,1 and reduced nicotinamide adenine dinucleotide (NADH) + H+. This system is composed of 4 proteins, namely, a pyridoxal phosphate enzyme (P-protein), a carrier protein containing covalently bound lipoic acid (H-protein), a folate dependent enzyme (T-protein), and a protein exhibiting lipoamide dehydrogenase activity (L-protein). The oxidative cleavage of glycine, catalyzed by the GCS system, is a major catabolic pathway in various organisms. In the first step, the P-protein catalyzes the decarboxylation of the glycine molecule and subsequent transfer of the residual methylamine to the oxidized H-protein (Hox), generating the methylamine-loaded H-protein (Hmet). In the second step, the T-protein, a folate dependent enzyme, catalyzes the transfer of a methylene carbon from Hmet to tetrahydrofolate, resulting in the release of ammonia and the generation of reduced H-protein (Hred). In the last step, the hydrolipoyl group of Hred is oxidized by the L-protein and Hox is generated, thereby completing the catalytic cycle. These sequential reactions are reversible.1 The biochemical properties of GCS in bacteria and mitochondria of plants and mammals have been extensively studied.2, 3 In humans, a genetic disease termed nonketotic hyperglycinemia, caused by the absence of CGS (due to a number of mutations in the T-protein), results in a dramatic accumulation of glycine in blood, leading to metabolic and neurological disorders.4 Although the H-protein5-9 and the L-protein10-12 from several sources have been well studied by X-ray crystallography, the three-dimensional (3D) structures of the P-protein and the T-protein have not yet been reported. Here, we report the crystal structure of the T-protein from Pyrococcus horikoshii OT3. The native protein was purified and crystallized as reported elsewhere.13 Derivative crystals were obtained under the same condition as native crystal. A bromine derivative was prepared by soaking the crystals in artificial mother liquor containing 1 M sodium bromide for 5 s. Multiple-wavelength anomalous diffraction (MAD) data at peak, edge, and remote from the bromine derivative were collected at 100 K on the Rigaku R-axis V imaging plate detector at the beamline BL26B1 of SPring-8, Japan. The data were processed and scaled using HKL200014 (Table I). Forty Br atoms were located by the program SnB15 using anomalous differences. The anomalous phasing and refinement of the Br positions, and the subsequent improvement of electron density by solvent flattening and 2-fold noncrystallographic symmetry averaging were done using the CCP4 suite.16 The automatic model building with RESOLVE17 in combination of REFMAC18 correctly placed 55% of the residues of the 2 monomers in the asymmetric unit. After phase extension to 1.5 Å resolution, the remaining residues were built manually with the program QUANTA (Accelrys, San Diego, CA). Two glycerol molecules and a polyethylene glycol molecule were identified and included in the refinement using the Crystallography & NMR System (CNS).19 Alternate conformations were adopted for the residues Arg19, Ser117, Arg162, and Glu362. The final refinement statistics are shown in Table I. Atomic coordinates and structure factors are available from the Protein Data Bank (PDB) under accession code 1V5V. The figures were prepared by MOLSCRIPT,20 BOBSCRIPT,21 RASTER3D,22 and GRASP.23 The crystal structure of the P. horikoshii T-protein [Fig. 1(A)] was determined at 1.5 Å resolution using the MAD technique. Data collection, model, and refinement statistics are summarized in Table I. The Ramachandran plot produced by PROCHECK24 shows that 92% of the residues are in the most favored region, and 7.7% are in the additionally allowed region. With the exception of one residue, Ala46, which fits neatly into well-defined electron density (ϕ = −137.2°, ψ = −101.6°), no other residue lies in the generously allowed or disallowed regions. (A) Ribbon diagram of the T-protein, showing the structure of a protomer. Domains I, II, and II are colored light blue, yellow, and red, respectively. The central hole, the long loop in domain II, the dimerization helix, and N- and C-termini are labeled. (B) Two orthogonal views of the T-protein dimer. Molecular symmetry is depicted as crystallographic symbols. Perspective of the protomer (red) in the left view is similar to that in (A). The asymmetric unit contains 2 polypeptide chains of the T-protein. This apparent homodimeric state is consistent with our previous report,13 where it was shown by dynamic light scattering experiments that the T-protein is a dimer in solution. The protomer comprises 3 domains, and appears as a cloverleaf-like or a ring-like arrangement around a central hole [Fig. 1(A)]. Domains I and II have similar folds, suggesting plausible gene duplication. The root-mean-square deviation (RMSD) for the structurally equivalent 70 Cα atoms of the 2 domains is 2.1 Å, although their sequence homology is low (8%). Domain I (residues 2–55 and 148–258) is composed of a 6-stranded β-sheet covered by 4 helices and an N-terminal β-hairpin with a short helix. Similarly, domain II (residues 56–147 and 259–310) is composed of a 5-stranded antiparallel β-sheet covered by 2 helices and a long loop (residues 259–299) followed by a helix. The major β-sheets in both the domains are associated to form a distorted 8-stranded β-barrel. The C-terminal domain III (residues 311–401) comprises a modified 6-stranded β-barrel perpendicular to the distorted β-barrel of N-terminal domains. The 2 protomers are dimerized through an interface involving residues of domain II [Fig. 1 (B)]. The dimer interface is mainly composed of a part of the long loop (269–275) and a helix in domain II (120–137) designated as the “dimerization helix” [Fig. 1(A)]. The dimer interface is extensive, with a surface area of 1845 Å2, and consists of a hydrophobic core and 11 hydrogen bonds. Pairwise superposition of the 2 subunits in the dimer yields a RMSD value of 0.3 Å, indicating that the organization of the domains in both subunits is almost identical. The search for homologous proteins, performed with the coordinates of the T-protein using the DALI25 server, suggested strong structural similarity with the C-terminal region of the dimethylglycine oxidase (DMGO) from Arthrobacter globiformis (PDB code: 1PJ5).26 The RMSD between the T-protein and the C-terminal region of DMGO is 1.6 Å over 330 aligned residues. Superimposed parts of the structures shows a residue identity of 26%. The structural alignment of these 2 proteins reflects similar folds [Fig. (2A)]. However, the DMGO structure lacks two loops corresponding to residues 211–224 and 270–278 of the T-protein and has a 30-residue insertion between residues 34 and 35 of the T-protein. Also, Yfgz protein from Escherichia coli (PDB code: 1NRK, not published) shows reasonable structural similarity with the T-protein, yielding an RMSD of 2.6 Å over 276 aligned residues. (A) Stereo view of the superposition of the T-protein (red) and the C-terminal region of the dimethylene oxidase from A. globiformis (PDB code: 1PJ6). Perspective is similar to that in Figure 1(A). Inserted and deleted segments are distinguished by labeling and light coloring. (B) Molecular surface representation of the t-protein protomer. Perspective on the left is similar to that in Figure 1(A). Picture on the right is viewed from the rear left side. Conserved residues in the T-protein orthologs are mapped with coloring: dimer interface residues in magenta; putative folate-binding residues in yellow; the cross-linking residue Lys323 between the H- and T-proteins in blue; putative H-protein interface residues in brown; conserved sites I, II, and III residues in red, cyan, and light blue, respectively; others are in green. Boundary of the dimer interface is depicted by dotted lines. The T-protein is a tetrahydrofolate-dependent enzyme and catalyzes the release of ammonia and the transfer of methylene carbon unit to tetrahydrofolate from the amino-methyl moiety of glycine attached to the H-protein. The initial step of the methylene transfer reaction is considered to be the nucelophilic attack on the methylene carbon of the H-intermediate by the N10 atom of the tetrahydrofolate. The structural similarity with the DMGO, which contains a tetrahydrofolate binding site, allowed us to deduce the ligand-binding site in the T-protein. By analogy with the crystal structure of DMGO-folate complex,26 we conclude that the role of Glu100 might be the activation of N10 atom by increasing nucleophilic character of N10. The acidic nature of this residue is well conserved among the T-protein family (Asp or Glu). Another active site residue, Glu205, which is invariant in the T-protein family, might be important for binding the pterin group of folate. We also identified other highly conserved residues Tyr87, Leu102, Phe179, Gly197, Tyr198, Tyr207, Arg251, and Phe289 [colored yellow in Fig. 2(B)] as important for the binding of folate. In order to locate functionally important regions of the T-protein, we mapped the highly conserved residues in orthologs onto the molecular surface [Fig. 2(B)]. By docking studies with the crystal structure of the H-protein (PDB code: 1HPC),27 we could identify the putative H-protein–T-protein interface. The interface includes invariant residues Arg327, Tyr330, and Arg376 of domain III; and Lys20, Phe24, Gly26, Ser35, Ile36, Leu248, Gly249, Asp252, Thr253, Arg255, and Leu262 of domains I and II [colored brown in Fig. 2(B)]. A well-conserved but not invariant residue Lys323, colored blue in Fig. 2(B), has been reported to be involved in the binding interface between the H- and T-proteins by cross-linking experiments on E. coli proteins.28 The location of Lys323 is on the perimeter of the identified H-protein interface, in agreement with the cross-linking experiment. We also found a number of invariant residues in a part of the dimer interface adjacent to the central hole: Lys122 and Trp126 at N-terminal part of the dimerization helix; and Tyr263, Gly264, Glu271, Leu273, and Gly299 on a part of the long loop [colored magneta in Fig. 2(B)]. This observation suggests that the dimer formation is important for the function of this protein. Furthermore, we identified 3 other clusters of invariant residues designated as sites I, II, and III [Fig. 2(B)]. From the docking study with the H-protein, site I residues [colored red in Fig. 2(B)] comprising Glu276, Ile277, Gln306, Lys307, Glu308, Gly310, Arg313, and Lys314 on the dimer counterpart might be important for the conformational change of lipoate in the H-protein. Of these residues, Arg313 on the dimer counterpart is likely to interact with the C-terminal part of the H-protein. This invariant arginine residue has been suggested to have a role in participating the onset of nonketotic hyperglycinemia; the mutation of Arg313 to histidine has been reported to result in the loss of either the catalytic activity or the ability of binding the folate.26 On the other hand, residues Lys307, Glu308, and Lys314 might be important for interactions with other GCS components (P- or L-protein) because of their proximity to the bound lipoate of the H-protein. In contrast with the site I, biological roles of other sites are currently not so clear. Residues at the site II, Pro349, Leu351, Val385, Phe389, and Tyr390 [colored cyan in Fig 2 (B)], which are near the putative H-protein interface and residues at site III, Gln70, Asn75, Asp76, Ser78, Tyr111, Lys295, and Lys300 [colored light blue in Fig. 2(B)] are both located in the vicinity of the central hole on the folate accessing side, and they might be involved in the interfaces for unknown factors. The crystal structure reported here represents the first description of the 3D structure of the GCS component T-protein from any organism. The fold is similar to that of DMGO, where 3 domains make a cloverleaf-like or ring-like structure around a central hole accommodating folate. From a comparison of the crystal structures, we could identify the potential ligand binding site and the interfaces for other components. We also found that the dimeric state might be essential for the function. This work will provide a structural basis for understanding the structure–function relationship of the GCS multienzyme complex. Our thanks to M.R.N. Murthy for a critical reading of the manuscript and to the beamline staff for assistance during data collection at beamline BL26B1 of SPring-8. N. K. Lokanath solved the structure and wrote the article, C. Kuroishi contributed the large-scale protein production, N. Okazaki contributed the automated crystallization, and N. Kunishima supervised this work.