Crystal Structure of T-protein of the Glycine Cleavage System

The glycine cleavage system catalyzes the oxidative decarboxylation of glycine in bacteria and in mitochondria of animals and plants. Its deficiency in human causes nonketotic hyperglycinemia, an inborn error of glycine metabolism. T-protein, one of the four componentsoftheglycinecleavagesystem,isatetrahydrofolatedependent aminomethyltransferase. It catalyzes the transfer of the methylene carbon unit to tetrahydrofolate from the methylamine group covalently attached to the lipoamide arm of H-protein. To gain insight into the T-protein function at the molecular level, we have determined the first crystal structure of T-protein from Thermotoga maritima by the multiwavelength anomalous diffraction method of x-ray crystallography and refined four structures: the apoform; the tetrahydrofolate complex; the folinic acid complex; and the lipoic acid complex. The overall fold of T-protein is similar to that of the C-terminal tetrahydrofolate-binding region (residues 421–830) of Arthrobacter globiformis dimethylglycine oxidase. Tetrahydrofolate (or folinic acid) is bound near the center of the tripartite T-protein. Lipoic acid is bound adjacent to the tetrahydrofolate binding pocket, thus defining the interaction surface for H-protein binding. A homology model of the human T-protein provides the structural framework for understanding the molecular mechanisms underlying the development of nonketotic hyperglycinemia due to missense mutations of the human T-protein.

The glycine cleavage system (GCS) 1 serves an important biochemical function by catalyzing the oxidative decarboxylation of glycine in the mitochondria of animals and plants as well as in bacteria (1). It is composed of four components: P-protein (EC 1.4.4.2); H-protein; T-protein (EC 2.1.2.10); and L-protein (EC 2.1. 8.1.4). Inherited deficiency of the human mitochondrial GCS causes nonketotic hyperglycinemia (NKH), an inborn error of glycine metabolism (2,3). It is characterized by elevated levels of glycine in blood and cerebrospinal fluid. In patients with NKH, convulsive seizures, coma, and respiratory distress develop within a few days after birth (4). A defect in any component of the GCS can abolish the overall activity of the GCS. Up to 15% NKH patients have defects in T-protein, and most other patients have P-protein defects, whereas Hprotein and L-protein deficiencies are rare (5).
P-protein of the GCS catalyzes the pyridoxal phosphate-dependent decarboxylation of glycine and transfer of the residual methylamine moiety to the lipoyl-lysine arm of the oxidized H-protein, generating a methylamine-loaded H-protein. When methylamine-bound, the lipoamide arm of H-protein is pivoted and is tightly bound into a cleft at the protein surface (6). T-protein is a tetrahydrofolate (H 4 folate)-dependent aminomethyltransferase. It catalyzes transfer of the methylene carbon unit to T-protein-bound H 4 folate from the methylamine group covalently attached to the lipoamide arm of H-protein, releasing ammonia and producing the reduced H-protein and 5,10-methylene-H 4 folate. The resulting dihydrolipoyl residue of H-protein is reoxidized by L-protein, thereby completing the reaction cycle (7). During the course of the GCS catalytic cycle, H-protein commutes successively among P-, T-, and L-proteins with its lipoamide arm visiting all three active sites of other proteins.
Small angle x-ray scattering (8) and cross-linking experiments (9) indicated that T-protein and H-protein form a stable complex in a 1:1 ratio. NMR spectroscopic studies of the complex between H-protein and T-protein indicated that the interaction surface of H-protein is localized on one side of the cleft where the lipoate arm is positioned (6). This work also suggested that the role of T-protein is not only to locate H 4 folate in a position favorable for a nucleophilic attack on the methylene carbon but also to destabilize the methylamine-loaded H-protein to facilitate unlocking of the arm and to initiate the reaction (6). The lack of the N-terminal 16 residues in Escherichia coli T-protein caused a loss of catalytic activity (10). Further N-terminal deletion mutant studies suggested that the N-terminal region of T-protein is essential for the conformational change of T-protein that accompanies its interaction with Hprotein (9). Results of limited proteolysis studies on mutants of the E. coli T-protein suggest that the N-terminal region of T-protein functions as a molecular "hasp" to hold T-protein in the compact form required for the proper association with H-protein (11).
Until now, three-dimensional structures of two proteins of the GCS have been determined, H-protein in four different forms (apoform, oxidized, methylaminated, and reduced) (12)(13)(14)(15)(16) and L-protein (15). Both P-protein and T-protein have been crystallized (17,18). In this study, we have determined the first three-dimensional structure of T-protein by the multiwave-length anomalous diffraction (MAD) method. Here we report the crystal structure of T-protein from Thermotoga maritima (Tm) in four forms: the apoform; the folinic acid complex; the H 4 folate complex; and the reduced lipoic acid complex. This study provides essential structural information on cofactor/ inhibitor binding and useful insights into H-protein recognition by T-protein. It also provides the structural framework for understanding the molecular mechanisms of how some missense mutations of the human T-protein lead to its inactivation and NKH (2, 3), a metabolic disorder with severe, frequently lethal, neurological symptoms in the neonatal period (4).

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The gcvT gene (TM0211) encoding Tm T-protein was cloned into the expression vector pET-21a(ϩ) (Novagen). The intact protein without any purification tag was overexpressed in E. coli B834(DE3) cells using terrific broth culture medium. Protein expression was induced by 1 mM isopropyl 1-thio-␤-D-galactopyranoside, and the cells were incubated for an additional 30 h at 15°C following growth to mid-log phase at 37°C. The cells were lysed by sonication in 50 mM Tris-HCl (pH 7.2) and 200 mM NaCl. Following heat treatment at 80°C for 10 min, the sample was centrifuged at 18,000 rpm for 60 min. The supernatant was applied to a HiLoad 26/10 Q-Sepharose column (Amersham Biosciences), which was previously equilibrated with 50 mM Tris-HCl (pH 7.2). Upon eluting with a gradient of NaCl in the same buffer, Tm T-protein was eluted at 250 -300 mM NaCl concentration. The protein was further purified by gel filtration on a HiLoad XK-16 Superdex 200 prep-grade column (Amersham Biosciences), which was previously equilibrated with 50 mM Tris-HCl (pH 7.2) and 200 mM NaCl. The procedure for preparing the selenomethionine (SeMet)-substituted protein was the same except for the presence of 10 mM dithiothreitol in all of the buffers used during the purification steps. When overexpressing the SeMet-substituted protein in E. coli B834(DE3) cells, we used the M9 cell culture medium that contained extra amino acids including SeMet.
Crystallization-Crystals were grown by the hanging-drop vapor diffusion method at 24°C by mixing equal volumes (2 l each) of the protein solution (27 mg ml Ϫ1 concentration in 50 mM Tris-HCl (pH 7.2) and 200 mM NaCl) and the reservoir solution. To grow crystals of the native protein in the apoform, we used a reservoir solution consisting of 15-20% (w/v) polyethylene glycol 3350 and 200 mM sodium dihydrogen phosphate (pH 4.25). The crystals grew to approximate dimensions of 0.2 ϫ 0.2 ϫ 0.3 mm within a few days. To grow crystals of the native protein complexed with folinic acid, 1.0 M folinic acid solution (dissolved in 50 mM Tris-HCl (pH 7.2) and 200 mM NaCl) was mixed with the protein solution in a 1:20 volume ratio, resulting in an ϳ67-fold molar excess of folinic acid over the T-protein monomer. The protein mixed with folinic acid was incubated for 30 min at 4°C before crystallization. Crystals of the folinic acid complex grew under identical conditions as the apocrystals. The SeMet-substituted protein was crystallized by microseeding techniques under crystallization conditions identical to those for the native crystals except for the presence of 10 mM dithiothreitol in the protein solution. The crystals of the SeMet-substituted protein grew up to approximate dimensions of 0.2 ϫ 0.2 ϫ 0.3 mm within a few days. The crystals of the H 4 folate complex was obtained by soaking crystals of the SeMet-substituted protein in a H 4 folate-saturated solution (20% (w/v) polyethylene glycol 3350, 200 mM sodium dihydrogen phosphate (pH 4.25), and 10 mM dithiothreitol) for 24 h before cryoprotection. To obtain the reduced lipoic acid complex, the crystals of the SeMet-substituted protein were soaked in 50% (v/v) dimethyl sulfoxide solution containing 10 mM dithiothreitol, which was previously saturated with lipoic acid, for 2 min before cryoprotection.
X-ray Data Collection and Structure Determination-A crystal of the SeMet-substituted protein was frozen using a cryoprotectant solution containing 25% (v/v) glycerol in the crystallization mother liquor. X-ray diffraction data were collected at 100 K on a Bruker CCD area detector system at the Beamline-6B experimental station of Pohang Light Source. For each image, the crystal was rotated by 1°and the crystalto-detector distance was set to 360 mm. The raw data were processed and scaled using the program suite HKL2000 (19). The SeMet-substituted crystal belongs to the space group P2 1 2 1 2 1 with unit cell parameters of a ϭ 52.61 Å, b ϭ 54.16 Å, and c ϭ 149.44 Å. Table I summarizes the statistics of MAD data collection. All of the eleven expected selenium atoms of a monomer in each crystallographic asymmetric unit were located with the program SOLVE (20), and the selenium sites were used to calculate the phases with RESOLVE (21). Phasing statistics are summarized in Table I. X-ray diffraction data of the H 4 folate complex and folinic acid complex were collected as above. Data of the reduced lipoic acid-bound crystal were collected at 100 K on a Quantum 315 CCD detector (Area Detector Systems Corporation, Poway, CA) at the Beamline-5A experimental station of Photon Factory, whereas the data of the apoform were collected at the Beamline-18B experimental station on an ADSC Quantum 4R CCD detector. The raw data were processed and scaled using the program suite HKL2000 (19).
Model Building and Refinement-Excellent quality of the electron density map allowed automatic model building by the program RE-SOLVE (21), giving an initial model that accounted for ϳ70% of the backbone of the polypeptide chain with much of the sequence assigned. Subsequent manual model building was done using the program O (22). The model was refined with the program CNS (23), including the bulk solvent correction. 10% of the data were randomly set aside as the test data for the calculation of R free (24). Several rounds of model building, simulated annealing, positional refinement, and individual B-factor refinement were performed. Subsequently, this model was used to refine structures of the apoform, the folinic acid-bound form, the H 4 folate-bound form, and the lipoic acid-bound form. Refinement statistics are summarized in Table I. All of the models have excellent stereochemistry (Table I) as evaluated by the program PROCHECK (25).
Homology Modeling of Human T-protein-A structural model of the human GCS T-protein (Leu 33 -Phe 394 ) was built by the homology modeling server (swissmodel.expasy.org/) using the crystal structure of Tm T-protein as template. The N-terminal (Met 1 -Val 32 ) and C-terminal (Val 395 -Lys 403 ) regions of the human T-protein were not modeled due to a lack of significant sequence homology.

RESULTS AND DISCUSSION
Overall Structure-We have determined the crystal structure of Tm T-protein by the MAD method and refined four structures: (i) the apoform at 1.84 Å; (ii) the complex with (S)-folinic acid (5-formyl-5,6,7,8-tetrahydrofolic acid) at 2.0 Å; (iii) the complex with H 4 folate at 2.4 Å; and (iv) the complex with reduced lipoic acid at 1.95 Å. The latter three models account for residues 1-362 of one T-protein monomer in an Every tenth residue is marked by a black dot, and every twentieth residue is labeled. Three signature sequence motifs are highlighted by thick lines: TGYTGE-XGXE motif (residues 186 -195) in magenta; PXGLGARDXXRhEAXXXLYG motif (residues 221 and 240) in blue; and GXh(T/S)(S/T)GXXSPTL motif (residues 306 -317) in green, respectively. The Nterminal region (residues 14 -35) of domain 1, which plays a crucial role in Hprotein interaction, is also highlighted by orange thick lines. asymmetric unit, whereas Arg 362 is additionally missing from the apostructure. The missing residues have no electron density. Tm T-protein is monomeric and is oblate-shaped with approximate dimensions of 55 ϫ 50 ϫ 30 Å (Fig. 1). It is tripartite. Its three domains are positioned in a cloverleaflike arrangement. Domain 1 (residues 1-51 and 140 -240) consists of a predominantly antiparallel, six-stranded ␤-sheet that contains a single Greek-key motif packed on one side by three ␣-helices (␣1, ␣5, and ␣6) and on the other side by two ␣-helices (␣2 and ␣7). Domain 2 (residues 52-139 and 241-280), which includes a long excursion from domain 1, has a five-stranded antiparallel ␤-sheet with flanking ␣-helices. The two antiparallel ␤-sheets from domains 1 and 2 are loosely packed against each other. The C-terminal domain 3 (residues 281-362) forms a distorted six-stranded jelly roll that packs perpendicular with the ␤-sheets of domains 1 and 2. The C-terminal tail (residues 354 -362) of domain 3 covers part of domain 1 (Fig. 1, front side) Figs. 1 and 2). Both H 4 folate and folinic acid adopt a kinked conformation (Fig. 2). The H 4 folate-bound structure is also nearly identical to the apostructure with an r.m.s. deviation of 0.23 Å for 361 C␣ atoms (Met 1 -Arg 361 ), and the central hole has similar solvent accessible pocket volumes in both the apostructure and the H 4 folate complex structure. This finding suggests that Tm T-protein has a rigid H 4 folate binding pocket. The mouth opening of the central hole is more open on the C-terminal side or the glutamate tail side (Fig.  1A, front side) than the N-terminal side or the H-protein interaction side (Fig. 1A, back side) (Fig. 2A). Five of these residues (Asp 96 , Tyr 100 , Tyr 188 , Glu 195 , and Arg 227 ) are well conserved among bacterial T-proteins (Fig. 3). Tyr 83 , Tyr 168 , Tyr 236 , Leu 237 , and Tyr 239 interact with H 4 folate indirectly through water molecules or through protein main chain atoms. Two of them (Tyr 83 and Tyr 239 ) are well conserved among bacterial Tproteins (Fig. 3) Tm T-protein resembles that of folinic acid binding to the C-terminal region (residues 430 -827) of Arthrobacter globiformis dimethylglycine oxidase (27). Some other structural features around H 4 folate are noteworthy ( Fig. 2A). Asn 112 and Tyr 83 interact with Asp 96 . Both Asn 112 and Tyr 83 are highly conserved in bacteria, whereas only Asn 112 is conserved in the human T-protein (as Asn 145 ) and Tyr 83 is substituted with Leu 116 in human. The separation between the Tyr 83 O1 atom and the Asp 96 O␦2 atom is 2.97 Å, whereas the distance between the Asn 112 N␦2 atom and the Asp 96 O␦1 atom is 2.79 Å. One water molecule is additionally hydrogen- . Arrows above the sequences denote ␣-helices and cylinders ␤-strands. Blue circles above the sequence indicate the residues that interact with H 4 folate with the exception of Asn 112 , which interacts with folinic acid only. Orange squares below the sequences represent the residues that are close to the bound lipoic acid. Red triangles below the sequences are the missense mutation sites of the human T-protein associated with NKH. Three signature sequence motifs are enclosed by colored boxes: TGYT-GEXGXE motif (residues 186 -195) in magenta; PXGLGARDXXRhEAXXXLYG motif (residues 221 and 240) in blue; and GXh(T/S)(S/T)GXXSPTL motif (residues 306 -317) in green, respectively. The N-terminal region (residues 14 -35) of domain 1, which plays a crucial role in H-protein interaction, is also enclosed by a dotted orange box. This figure was drawn with ClustalX (38) and GeneDoc (39). bonded to Asp 96 (2.89 and 3.17 Å to the O␦1 and O␦2 atoms of Asp 96 , respectively).
In the folinic acid complex, the same residues that contact H 4 folate interact with folinic acid. Additionally, the oxygen atom of the N5-formyl group of folinic acid interacts with the side chain of Asn 112 (Fig. 2B). H 4 folate lacks a formyl group and thus does not interact with Asn 112 (Fig. 2A). The N5-formyl group is also in proximity of the side chain of Tyr 188 with its aromatic ring nearly perpendicular to the p-aminobenzoic acid group (Fig. 2B). In the case of the folinic acid complex, we observe an additional binding of the second folinic acid at a narrow cleft on the surface of domain 2 near the interface between domains 1 and 2 (Fig. 1A). The second folinic acid takes an extended conformation, and the residues that interact with the second folinic acid (the side chains of Glu 106 , Glu 160 , Lys 173 , Ile 175 , and Glu 180 ; main chain atoms of Met 197 and Leu 198 ) are not highly conserved. Furthermore, the average B-factor of the second folinic acid (41.8 Å 2 ) is much higher than the first (18.1 Å 2 ), suggesting a weaker binding of the second folinic acid. This secondary binding of folinic acid is probably a crystallization artifact.
It was reported that a pool of polyglutamate forms of folate is dominated by tetraglutamate (25%) and pentaglutamate (55%) in the pea leaf mitochondria (28). The binding affinity of H 4 folate polyglutamates for pea leaf T-protein was found to increase with increasing number of glutamates up to six residues (29). This observation may be explained by the presence of a surface patch with highly positive electrostatic potential due to clustering of nine positively charged residues (Lys 80 , Arg 185 , Lys 280 , Lys 328 , Lys 352 , Lys 353 , Arg 357 , Arg 361 , and Arg 362 ) in the vicinity of the glutamate tail of H 4 folate bound to Tm T-protein (Fig. 4A). Five of them (Lys 80 , Arg 185 , Lys 280 , Lys 352 , and Arg 357 ) are well conserved in bacterial T-proteins. The first glutamate moiety is bound to Tm T-protein in essentially identical manners in both complexes of H 4 folate and folinic acid (Fig. 2). to the binary complexes with H 4 folate, folinic acid, or lipoic acid, we incorporated H 4 folate of the H 4 folate complex into the model of the lipoic acid complex to facilitate discussion ( Figs. 1 and 4C).

Lipoic Acid Binding Reveals Insights into H-protein Recognition-To
Lipoic acid is bound in an ϳ15-Å deep pocket adjacent to the H 4 folate binding pocket with its carboxylate group pointing toward the bulk solvent ( Figs. 1 and 4C). The S6 sulfur atom makes a hydrogen bond with the side chain of Asp 228 at a distance of 2.90 Å. Arg 227 (N2 atom) is close to the S8 sulfur atom (3.23 Å). Phe 20 , Tyr 188 , Leu 224 , and Leu 238 as well as the aliphatic part of the Arg 227 side chain surround the aliphatic part of lipoic acid (Fig. 4C). Leu 224 , Arg 227 , Asp 228 , and Leu 238 are all strictly conserved in bacterial T-proteins, and they belong to the highly conserved sequence motif PXGLGAR-DXXRhEAXXXLYG between positions 221 and 240 in domain 1 of Tm T-protein (boxed in blue in Fig. 3; highlighted in thick blue lines in Fig. 1B), where X stands for any amino acid and h is a hydrophobic residue. This motif is the longest signature sequence of T-protein (H 4 folate-dependent aminomethyltransferase). Tyr 188 belongs to another conserved sequence motif TGYTGEXGXE between positions 186 and 195 in domain 1 of Tm T-protein (boxed in magenta in Fig. 3; highlighted in thick magenta lines in Fig. 1B). This motif also contains the strictly conserved residue Glu 195 , which directly contacts H 4 folate (Fig. 1B). Bacterial T-proteins have a third conserved sequence motif GXh(T/S)(S/T)GXXSPTL between positions 306 and 317 in domain 3 of Tm T-protein (boxed in green in Fig. 3; highlighted in thick green lines in Fig. 1B), where two possible residues are grouped within parentheses. This signature motif in domain 3 is not directly involved in catalysis or the binding of H 4 folate and lipoic acid. It appears to be important for H-protein recognition (further discussed below).
Two structurally conserved water molecules are bound between lipoic acid and H 4 folate (Wat47 and Wat114 in Fig. 4C). They are present in all four structures except in the folinic acid complex where Wat114 is absent, because the N5-formyl group occupies the site of Wat114. Wat114 is hydrogen-bonded to the S8 atom of lipoic acid at a distance of 2.83 Å and to the O␦1 atom of Asn 112 at 2.88 Å. Wat47 is hydrogen-bonded to the backbone nitrogen atom of Tyr 239 at a distance of 2.74 Å and to the O␦1 atom of Asp 96 at 2.84 Å. If a methylamine group were covalently attached to the S8 atom of lipoic acid, Asn 112 (O␦1 oxygen) and Tyr 188 (O1 oxygen) would be within hydrogenbonding distances from the methylamine group. Asn 112 and Tyr 188 are strictly conserved in T-proteins (Fig. 3).  (Figs. 1B and 4B). There is a deep cleft between domains 1 and 3 on this side of Tm T-protein (Fig. 4D). Along this cleft around lipoic acid, many conserved residues are clustered, including Phe 20 , Leu 224 , Arg 227 , Asp 228 , Leu 238 , and Tyr 239 (Fig. 4B). This cleft appears to be the site of H-protein binding. Deletions of the N-terminal 4, 7, 11, and 16 residues from the E. coli T-protein led to reduction in the activity to 42, 9, 4, and 0%, respectively, relative to the wild-type enzyme (9,10). Our Tm T-protein structure revealed that the N-terminal region of domain 1 (circled in Fig. 4B) contributes to one side of this cleft and that removal of the N-terminal residues 1-13 would seriously affect the folding of the sequence region 14 -35 (boxed in orange in Fig. 3; highlighted in thick orange lines in Fig. 1B), causing the distortion of the H-protein binding surface. Therefore, the loss of E. coli T-protein activity caused by a deletion of the N-terminal 16 residues is due to disruption of the H-protein-binding site. It is also apparent that the signature sequence motif GXh(T/S)(S/T)GXXSPTL in domain 3 (boxed in green in Fig. 3; highlighted in thick green lines in Fig. 1B) provides another side of the H-proteinbinding cleft (Fig. 1B).
Cross-linking experiments indicated that Lys 288 of E. coli T-protein is close to Asp 43 of E. coli H-protein when they form a 1:1 complex (9). E. coli Lys 288 corresponds to Tm Lys 289 and is conserved as lysine or arginine in bacterial T-proteins (Fig.  3). The location of Lys 289 in Tm T-protein is shown in Fig. 4, B and D. The lipoyl moiety is covalently attached to Lys 63 (indicated by a green arrow in Fig. 4D) of T. thermophilus H-protein (16), which has a 57% sequence identity with Tm H-protein. All of these pieces of information allowed us to build a crude but reasonable model of the complex between T-protein and H-protein (Fig. 4D). In this model, H-protein is positioned along the cleft between domains 1 and 3 of T-protein on the Nterminal side, Glu 42 (corresponding to E. coli Asp 43 ) of T. thermophilus H-protein (marked by a half-transparent red dot in Fig. 4D) is close to Lys 289 of Tm T-protein (corresponding to E. coli Lys 288 ), and Lys 63 of T. thermophilus H-protein points toward the lipoic acid binding pocket (as indicated by a green arrow in Fig. 4D). It was suggested that H-protein undergoes a small conformational change upon binding to T-protein so that the lipoyl arm carrying the methylamine group is exposed (6). Upon interaction with H-protein, T-protein may also undergo a small structural alteration such as limited domain rearrangement. To characterize the possible structural changes in both T-protein and H-protein that accompany the complex formation, further structural studies are required.
Structural Understanding of Nonketotic Hyperglycinemia-A considerable level of sequence similarity exists between the human T-protein and Tm T-protein (Fig. 5A): 33% identity between the human T-protein (Leu 33 -Phe 394 ) and Tm T-protein (Met 1 -Phe 355 ). Thus, the homology-modeled structure of the human T-protein is highly similar to that of Tm T-protein in its core. Only the surface regions (Ser 68 -His 71 , Gly 171 -Ala 178 , Gly 217 -Val 220 , His 242 -Ile 247 , Leu 292 -Ala 299 , and Gln 311 -Arg 315 of human) that are not directly associated with the catalytic machinery show significant structural deviations.
Among the missense mutation sites, Asn 145 (human)/Asn 112 (Tm) and Tyr 225 (human)/Tyr 188 (Tm) are strictly conserved among both bacterial and human T-proteins (Figs. 3 and 5A). Asp 276 (human)/Asp 242 (Tm) is conserved as either Asp or Glu (Figs. 3 and 5A). Arg 320 (human)/Leu 281 (Tm) is strictly conserved in bacterial T-proteins only (Fig. 3). Gly 269 (human)/ Thr 235 (Tm) is variable in the sequence but is within the signature sequence motif PXGLGARDXXRhEAXXXLYG in domain 1, whereas Asp 276 (human)/Asp 242 (Tm) is right after this motif (Fig. 3). Tyr 225 (human)/Tyr 188 (Tm) belongs to the signature sequence motif TGYTGEXGXE in domain 1 (Fig. 3). The side chains of both Asn 112 (Tm) and Tyr 188 (Tm) are close to the oxygen atom of the N5-formyl group of folinic acid (Fig.  5D) as mentioned above. Thus, the mutations of their equivalents in the human T-protein would cause a considerable loss in the T-protein activity. Mutation of Asn 145 (human)/Asn 112 (Tm) into histidine will affect H 4 folate binding both directly and indirectly by disturbing the network involving Asp 96 , which interacts with the N10 atom of H 4 folate (Fig. 5D). Mutation of Tyr 225 (human)/Tyr 188 (Tm) into cysteine will seriously alter the THF binding pocket.
His 42 (human)/His 10 (Tm) and Gly 47 (human)/Ala 15 (Tm) are semi-conserved among both bacterial and human T-proteins (Figs. 3 and 5A). His 10 (Tm), Ala 15 (Tm), and Asp 242 (Tm) are slightly separated from either the H 4 folate binding pocket or the lipoic acid-binding site (Fig. 1B). Mutations at these three sites would have no direct effect on the binding of H 4 folate or lipoic acid, but they appear to alter the H-protein binding interface. Interestingly, the residues interacting with His 10 (Tm) and Asp 242 (Tm) (Ala 15 , Pro 26 , Tyr 29 , and Asp 47 ; Arg 231 , Thr 309 , and Ser 310 ) are well conserved in the T-protein family (Fig. 3). Details of the interactions around His 10 (Tm) and Asp 242 (Tm) are shown in Fig. 5, B and C, respectively. In Tm T-protein, Asp 47 and Tyr 29 stabilize the imidazole ring of His 10 by hydrogen bonding and the side chain of Pro 26 lies close to the imidazole ring of His 10 . The side chain of Ala 15 points toward the hydrogen-bonding network around His 10 (Fig. 5B). Thus, the H42R and G47R mutations in the human T-protein could possibly disrupt the network involving these residues, thus altering the proper conformation of the N-terminal portion of domain 1, which provides the interface for interaction with H-protein as discussed above. In Tm T-protein, Arg 231 , Thr 309 , and Ser 310 are clustered around Asp 242 , forming a hydrogenbonding network and a salt bridge (Fig. 5C). Arg 231 is part of the signature motif PXGLGARDXXRhEAXXXLYG in domain 1, whereas Thr 309 and Ser 310 belong to the GXh(T/S)(S/T)GXX-SPTL motif in domain 3 (Fig. 3). Mutation of Asp 276 (human)/ Asp 242 (Tm) into histidine could possibly perturb the hydrogenbonding network around this residue and might alter the surface features that are crucial for H-protein binding.
Val 176 (Tm), corresponding to Val 212 (human), makes a hydrophobic core with neighboring residues (Leu 155 , Val 159 , Val 183 , and Leu 196 of Tm), of which Leu 155 , Val 159 , and Val 183 are conserved in human, whereas Leu 196 is replaced with Ile 233 in human (Fig. 5A). Thus, it is expected that the mutation of Val 212 (human)/Val 176 (Tm) into alanine would destabilize this hydrophobic core. Leu 281 (Tm), corresponding to Arg 320 (human), is conserved among bacterial T-proteins only. It is not part of the H 4 folate-or lipoic acid-binding site but is close to the conserved Asp 242 (Tm). The environment around Asp 242 (Tm) is shown in Fig. 5C. In the human T-protein model, Arg 320 makes an additional salt bridge with Asp 276 (Fig. 6B). This interaction may be necessary for the proper function or stability of the human T-protein, thus explaining why the R320H mutation causes the impaired T-protein activity.