Biochemical Journal

Research article

Structural and functional studies of Bacillus stearothermophilus serine hydroxymethyltransferase: the role of Asn341, Tyr60 and Phe351 in tetrahydrofolate binding

Vinitha R. Pai, V. Rajaram, Shveta Bisht, B. S. Bhavani, N. Appaji Rao, M. R. N. Murthy, H. S. Savithri


SHMT (serine hydoxymethyltransferase), a type I pyridoxal 5′-phosphate-dependent enzyme, catalyses the conversion of L-serine and THF (tetrahydrofolate) into glycine and 5,10-methylene THF. SHMT also catalyses several THF-independent side reactions such as cleavage of β-hydroxy amino acids, transamination, racemization and decarboxylation. In the present study, the residues Asn341, Tyr60 and Phe351, which are likely to influence THF binding, were mutated to alanine, alanine and glycine respectively, to elucidate the role of these residues in THF-dependent and -independent reactions catalysed by SHMT. The N341A and Y60A bsSHMT (Bacillus stearothermophilus SHMT) mutants were inactive for the THF-dependent activity, while the mutations had no effect on THF-independent activity. However, mutation of Phe351 to glycine did not have any effect on either of the activities. The crystal structures of the glycine binary complexes of the mutants showed that N341A bsSHMT forms an external aldimine as in bsSHMT, whereas Y60A and F351G bsSHMTs exist as a mixture of internal/external aldimine and gem-diamine forms. Crystal structures of all of the three mutants obtained in the presence of L-allo-threonine were similar to the respective glycine binary complexes. The structure of the ternary complex of F351G bsSHMT with glycine and FTHF (5-formyl THF) showed that the monoglutamate side chain of FTHF is ordered in both the subunits of the asymmetric unit, unlike in the wild-type bsSHMT. The present studies demonstrate that the residues Asn341 and Tyr60 are pivotal for the binding of THF/FTHF, whereas Phe351 is responsible for the asymmetric binding of FTHF in the two subunits of the dimer.

  • L-allo-threonine
  • crystal structure
  • folate binding
  • 5-formyl tetrahydrofolate (FTHF)
  • pyridoxal 5′-phosphate
  • serine hydroxymethyltransferase (SHMT)
  • ternary complex


SHMT (serine hydroxymethyltransferase; EC, a member of the α-family of PLP (pyridoxal 5′-phosphate)-dependant enzymes [1], plays a pivotal role in one-carbon metabolism. It catalyses the reversible interconversion of L-serine and THF (tetrahydrofolate) into glycine and 5,10-CH2-THF (5,10-methylene THF). 5,10-CH2-THF is a key intermediate for the biosynthesis of purines, thymidylate, choline and methionine [2], and is a vital link between amino acid and nucleotide metabolism. In addition, SHMT levels have been found to increase by 5–10-fold in cancer cells [3,4]. Therefore it is considered as a suitable target for cancer chemotherapy [5].

SHMT also catalyses several alternative reactions which are THF-independent. These include cleavage of β-hydroxy amino acids such as L-allo-threonine and β-phenylserine, decarboxylation, racemization and transamination [69]. The enzyme has been purified from several sources. Crystal structures of SHMT, as well as its complexes with substrates and inhibitors, have been elucidated [1014]. In addition, structures of a few mutants of active-site residues have been determined [1517]. An examination of the ternary complex of bsSHMT (Bacillus stearothermophilus SHMT) with glycine and FTHF (5-formyl THF), a structural analogue of THF, has revealed that N-1 and N-8 of FTHF are at hydrogen bonding distance from the amide group of Asn341 and Glu53 interacts with O-3 and O-10 of FTHF. On the other hand, Tyr60 stacks against the PABA (p-amino benzoic acid) moiety of FTHF and its hydroxy group is hydrogen bonded to the monoglutamate carboxy group of FTHF. Furthermore, in the ternary complex, the side chain of the FTHF monoglutamate tail is ordered in only one of the subunits. In this subunit, the phenyl ring of Phe351 is displaced to avoid a severe short contact with the carboxy group of the FTHF monoglutamate (Figure 1) and residues 342–352 are displaced so as to form a hydrogen bond between the Oγ of Ser349 and Oε of FTHF monoglutamate [14].

Figure 1 Stereo diagram illustrating the superposition of residues Asn341, Tyr60, Phe351 and FTHF (FON) in the two subunits of the bsSHMT–glycine–FTHF ternary complex

Studies on mutants of Glu53 in bsSHMT [17] and its equivalent Glu75 in rabbit liver cytosolic SHMT [15] have shown that this residue is essential for interactions with the THF substrate. As expected, mutation of Glu53 to glutamine led to complete loss of THF-dependent activity; however, the THF-independent activity was enhanced. In the present study, the role of other residues that are involved in the binding of FTHF/THF was studied by mutational analysis. In particular, Tyr60, Asn341 and Phe351 were targeted for mutagenesis. The structural and biochemical analyses of Y60A and N341A bsSHMTs show that these residues are directly involved in FTHF/THF binding and consequently affect the THF-dependent activity. The structure of the ternary complex of F351G bsSHMT shows that the monoglutamate part of FTHF is ordered in both the subunits of the dimer, unlike the wild-type ternary complex, consistent with the proposed role of Phe351 in the asymmetric binding of FTHF.


Site-directed mutagenesis

Plasmids were prepared by the alkaline lysis procedure using the DH5α strain of Escherichia coli [18]. Preparation of competent cells and transformation were carried out according to the method of Alexander [19]. The N341A bsSHMT mutant was constructed using a PCR-based sense–antisense primer method [20], with pRSH (B. stearothermophilus SHMT gene cloned into the pRSET C vector) as a template using appropriate sense (5′-GTGAATAAAGCCACCATTCCG-3′) and antisense (5′-CGGAATGGTGGCTTTATCAC-3′) primers and Deep Vent Polymerase (New England Biolabs). The nucleotides in italics indicate the mutations introduced. The mutations were confirmed by DNA sequencing. The F351G bsSHMT and Y60A bsSHMT mutants were also generated by a similar procedure using the following primers: F351G bsSHMT sense, 5′-GAAAGCCCGGGTGTCACGAGC-3′ and antisense 5′-GCTCGTGACACCCGGGCTTTC-3′; Y60A bsSHMT sense, 5′-GGCCGCCGCGCTTATGGCGGC-3′ and antisense 5′-GCCGCCATAAGCGCGGCGGCC-3′.

Expression and purification of bsSHMT and the mutant enzymes

pRSH and the mutant constructs were transformed into E. coli BL21(DE3) pLysS strain. A single colony was grown in 50 ml of LB (Luria–Bertani) medium containing 50 μg·ml−1 ampicillin for 14 h and inoculated into 500 ml of terrific broth containing 50 μg·ml−1 ampicillin at 30 °C. After sufficient growth (D600=0.6 at 3 h), the cells were induced with 0.3 mM IPTG (isopropyl β-D-thiogalactoside) and allowed to grow further for 5 h. The expression of the mutant constructs was as good as the wild-type clone (pRSH). The expressed proteins were present predominantly in the soluble fraction and purified by a procedure identical with that used for the wild-type enzyme [21]. The cells were harvested, resuspended in 60 ml of buffer A [50 mM potassium phosphate buffer (pH 7.4), 1 mM 2-mercaptoethanol, 1 mM EDTA and 100 μM PLP] and sonicated. The supernatant was subjected to 0–65% ammonium sulfate precipitation; the pellet obtained was resuspended in 20–30 ml of buffer B [20 mM potassium phosphate buffer (pH 8.0), 1 mM 2-mercaptoethanol, 1 mM EDTA and 50 μM PLP] and dialysed for 24 h against the same buffer (1 litre with two changes). The dialysed sample was loaded on to a DEAE-cellulose column (2 cm×15 cm), which was previously equilibrated with buffer B. The column was washed with 500 ml of buffer B, and the bound protein was eluted with 50 ml of buffer C [200 mM potassium phosphate buffer (pH 6.4), 1 mM EDTA, 1 mM 2-mercaptoethanol and 50 μM PLP]. The eluted protein was precipitated at 65% ammonium sulfate saturation, and the pellet was resuspended in buffer D [50 mM potassium phosphate buffer (pH 7.4), 1 mM EDTA and 1 mM 2-mercaptoethanol] and dialysed against the same buffer (2 litres with two changes) for 24 h. The protein concentration was estimated using the Lowry method using BSA as a standard [22]. For structural studies, the purified enzyme, after the ammonium sulfate precipitation, was resuspended in 0.1 M Hepes (pH 7.5) with 0.2 mM EDTA and 5 mM 2-mercaptoethanol and the ammonium sulfate was removed by washing with the same buffer using Amicon centricon filters.

Enzyme assays

L-[3-14C]Serine (Amersham Pharmacia Biotech) and THF were used to study the THF-dependent conversion of L-serine into glycine [23]. One unit of enzyme activity is the amount of enzyme required to form 1 μmol of formaldehyde per min at 37 °C. The specific activity was expressed as units per mg of protein. THF-independent cleavage of L-allo-threonine to glycine and acetaldehyde was monitored by recording the decrease in absorbance at 340 nm in a NADH-dependent alcohol-dehydro-genase-coupled assay [6] using a Jasco V-530 UV–visible spectrophotometer. The NADH consumed in the reaction was calculated using a molar absorption coefficient of 6220 M−1·cm−1. The kinetic constants Km and kcat were calculated using double reciprocal plots. The pseudo first-order rate constant for the THF-independent transamination of D-alanine was calculated from the time course of the reaction [6].

Spectroscopic methods

UV–visible absorption spectra of the enzymes were recorded in a Jasco V-530 UV–visible spectrophotometer in buffer D at 25±2 °C using 1 mg·ml−1 (25 μM) of the enzyme. CD measurements were made in a Jasco J-500A automated recording spectropolarimeter. Spectra were collected at a scan speed of 10 nm·min−1 and a response time of 16 s. Visible CD spectra were recorded from 550 to 300 nm using a protein concentration of 1 mg·ml−1 in buffer D with or without substrates (L-serine/glycine, THF/FTHF).

Thermal stability

Thermal denaturation of bsSHMT and the mutants was performed in a Jasco V-530 UV–visible spectrophotometer [18]. The absorbance change was monitored at 287 nm. The first derivative of the thermal denaturation profile obtained using the software Sigma Plot, was used to evaluate the apparent melting temperatures for the proteins.

Crystallization, data collection and processing

The mutant protein crystals were grown using the hanging drop vapour diffusion method using crystallization buffer containing 50% MPD (2-methyl 2,4-pentanediol) as the precipitant with 0.2 mM EDTA, 5 mM 2-mercaptoethanol in 100 mM Hepes (pH 7.5) [14]. A drop containing 4 μl of protein (18 mg·ml−1) and 4 μl of crystallization buffer was placed on a siliconized coverslip, which was then inverted and sealed over the well containing crystallization buffer. Plates were kept at 4 °C. Crystals started appearing within 4–5 days and grew to a maximum size in 5–10 days. The crystals were stable for a longer period of time when stored at 4–8 °C. The mutant protein crystals complexed with L-serine, glycine and L-allo-threonine were obtained under the same conditions as that of the native protein, except that the crystallization buffer also contained 10 mM of the respective ligand. The ternary complex crystals of SHMT with glycine/L-serine and FTHF were obtained by incubating the protein with FTHF (at a final concentration of 2 mM), and glycine (10 mM) was added to the crystallization buffer. The crystals were washed with the crystallization buffer and flash frozen in a stream of liquid nitrogen at 100 K for data collection. No additional cryoprotectant was added as crystallization buffer contained 50% MPD which acts as a cryoprotectant. X-ray diffraction data were collected using a Rigaku RU-200 rotating-anode X-ray generator (Cu-Kα radiation) equipped with a MAR research imaging-plate detector system. The crystals diffracted to a maximum resolution of approx. 2.7–1.7 Å (1 Å=0.1 nm). DENZO and SCALEPACK of the HKL suite [24] were used for indexing, integration, data reduction and scaling. The unit cell parameters and data collection statistics are given in Supplementary Tables S1–S3 at

Structure determination and refinement

The wild-type bsSHMT structure (PDB number: 1KKJ) was used as the initial model for refining the internal aldimine structures of all the mutants [14]. Of the total unique reflections, 5% were used as a test set to calculate Rfree in order to monitor the progress of the refinement and avoid model bias [25,26]. PLP and water molecules were omitted from the initial model. The structure was initially subjected to a rigid body refinement and then to restrained refinement using REFMAC5 [27] of the CCP4 suite of programs [28]. The output MTZ and PDB files of the refinement were used for the calculation of electron-density maps. Adjustment of the models was carried out using the program COOT [29]. The electron-density map clearly showed negative density for the side chains at the sites of mutation, i.e. Tyr60, Asn341 and Phe351. These residues were mutated to alanine, alanine and glycine respectively using COOT. The model was readily interpretable, and ligand and water molecules were then added for further refinement. Differences between the structures were detected visually and by calculating the distance between corresponding Cα atoms using the program CONTACT [28]. The mean B-factor and Ramachandran plot statistics were calculated using the BAVERAGE [28] and PROCHECK [30] programs of CCP4. Structural superposition of wild-type and mutants was achieved using the program ALIGN [31]. Wild-type glycine external aldimine (PDB number: 1KL1) was used as the starting model for the binary complexes of the mutants with glycine, L-serine and L-allo-threonine. The wild-type ternary complex was used as the starting model for the structure determination of the ternary complex of F351G bsSHMT with glycine and FTHF. Mutant–ligand complexes were then refined in the same manner as described above. The refinement statistics of all the mutants and their ligand complexes are listed in Supplementary Tables S1–S3. Figures were prepared using the program PyMOL (DeLano Scientific;


THF-dependent activity

N341A bsSHMT had a specific activity of 0.04 units·mg−1 (0.76%) compared with 5.2 units·mg−1 observed for the wild-type bsSHMT, suggesting that the mutation had resulted in a considerable loss of activity. On the other hand, F351G bsSHMT had a specific activity (5.0 units·mg−1) comparable with that of the wild-type enzyme. Y60A bsSHMT had no detectable activity. Since the activity was very low, kinetic constants such as kcat and Km could not be determined for the N341A and Y60A bsSHMTs. F351G bsSHMT had a kcat of 3.9 s−1 and a Km of 1.0 mM, values comparable with those of the wild-type enzyme (Table 1). As the crystal structure of the bsSHMT–glycine–FTHF ternary complex reveals the involvement of Asn341 and Tyr60 in the binding of FTHF [14], it is possible that the mutants are inactive owing to their inability to bind THF.

View this table:
Table 1 Kinetic parameters of bsSHMT and its mutants N341A, F351G and Y60A using L-serine, L-allo-threonine and D-alanine as substrates

NDA, no detectable activity; ND, not determined.

THF-independent activity

The cleavage of L-allo-threonine to glycine and acetaldehyde and transamination of D-alanine, both of which are THF-independent reactions, were assayed as described in the Experimental section. All the three mutants (N341A, Y60A and F351G bsSHMTs) were as active as the wild-type enzyme for the L-allo-threonine reaction and their kinetic parameters were comparable (Table 1). However, when D-alanine was added to N341A bsSHMT, the mutant enzyme precipitated and hence the reaction could not be monitored. This could be due to the instability of the mutant enzyme under the assay conditions. The rate constant for the transamination of D-alanine for Y60A and F351G bsSHMTs was the same (0.02 s−1) as that of the wild-type enzyme. These results suggest that these mutations have a negligible effect on the THF-independent activity.

Spectral and structural studies on the mutants

Internal aldimines

The presence of intermediates with characteristic absorption maxima is a convenient handle to follow SHMT-catalysed reactions [33]. As can be seen from Figure 2, all mutant enzymes have an absorbance maximum at approx. 425 nm, similar to that of bsSHMT. The three-dimensional structure of N341A, Y60A and F351G bsSHMTs showed that the PLP was bound at the active site as an internal aldimine forming a Schiff base with Lys226. All Cα atoms of N341A bsSHMT, Y60A bsSHMT and F351G bsSHMT superposed extremely well with the corresponding atoms of bsSHMT with rms (root mean square) deviations of 0.08, 0.10 and 0.08 Å respectively. There was no significant change in the conformation of any residue, either in the active site or near the site of mutation. In N341A bsSHMT, a water molecule was found close (1.4 Å) to the position corresponding to OD1 of Asn341. These results clearly demonstrate that the mutations did not affect the overall structure of the enzyme.

Figure 2 Absorbance changes on the addition of glycine and THF to (a) bsSHMT, (b) F351G bsSHMT, (c) N341A bsSHMT and (d) Y60A bsSHMT

bsSHMT gave an absorbance maximum at 425 nm (—, solid line), addition of glycine (50 mM) did not change the absorbance maximum at 425 nm (…, broken dotted line), and addition of THF (1.8 mM) to bsSHMT+glycine shows a prominent peak at 495 nm (---, broken dashed line) in (a) and (b), a small peak in (d), but not in (c). All of the experiments were carried out in buffer D at a protein concentration of 1 mg·ml−1.

Binary complexes with glycine/L-serine

The addition of the substrates glycine or L-serine to bsSHMT resulted in the formation of an external aldimine, which also absorbed maximally at approx. 425 nm. The mutants also had similar absorption spectra (Figure 2). In the structure of the binary complex of N341A bsSHMT with glycine, PLP was found at the active site as an external aldimine forming a Schiff base with the amino group of glycine. Two water molecules were found at positions close (approx. 1 Å) to the side chain (OD1 and NE2) of Asn341 of the wild-type binary complex.

In Y60A and F351G bsSHMT–glycine complexes, PLP was covalently bonded essentially to Lys226, showing that they are predominantly in the internal aldimine form, although glycine was bound at the active site (Figure 3). The electron density for glycine was unambiguous. The refined temperature factor for glycine, however, was higher than those of the surrounding atoms by a factor of 1.5. This implies that either the glycine residue has partial occupancy or it has a higher thermal factor. Only in the Y60A bsSHMT binary complex, weak density connecting C4A of PLP and the glycine amino group was observed, suggesting the presence of a small fraction of gem-diamine or an external aldimine form. The CD spectrum showed a decrease in the molar ellipticity at 430 nm upon formation of external aldimine. As shown in Figure 4, addition of glycine to Y60A bsSHMT resulted in a decrease in the molar ellipticity at 430 nm, suggesting the formation of external aldimine in solution. Similar results were obtained with N341A and F351G bsSHMTs. In addition, the Y60A and F351G bsSHMT mutants showed an increase in thermal stability upon the addition of glycine (Table 2), unlike the wild-type and N341A bsSHMTs. With L-serine, all of the mutants showed enhanced thermal stability, as in the wild-type enzyme. In the crystal structures of mutant–serine binary complexes, L-serine was present at the active site as an external aldimine, forming a Schiff linkage with PLP.

Figure 3 Stereo view of the electron density difference map (FoFc contoured at 3σ) of the region corresponding to PLP in Y60A bsSHMT–glycine (grey) superimposed on to the bsSHMT–glycine external aldimine (black)
Figure 4 Changes in the visible CD spectra of (a) bsSHMT and (b) Y60A bsSHMT on addition of glycine (50 mM)

The visible CD spectra of the bsSHMT and Y60A bsSHMT (1 mg·ml−1) showed a peak at 428 nm (−, solid line). Addition of glycine (50 mM) showed a decrease in molar ellipticity [θ] at 428 nm (…, broken line). N341A and F351G bsSHMTs (1 mg·ml−1) showed a similar decrease in molar ellipticity.

View this table:
Table 2 Apparent Tm values calculated for bsSHMT and its mutants N341A, F351G and Y60A in the absence and presence of the substrates, either glycine or L-serine

Binary complex with L-allo-threonine

All of the three mutants were also crystallized in the presence of 10 mM L-allo-threonine. The structure of N341A bsSHMT with L-allo-threonine revealed only a glycine external aldimine at the active site. This suggested that L-allo-threonine was converted into the product glycine under the conditions of crystallization. This is in agreement with the biochemical observation that N341A bsSHMT has THF-independent activity (Table 1). The activity was also observed in the buffer conditions used for crystallization.

The crystal structure of Y60A bsSHMT and F351G bsSHMT obtained in the presence of L-allo-threonine had L-allo-threonine as a Schiff base with PLP (Figure 5), unlike in wild-type and N341A bsSHMT, where only glycine external aldimines were observed. This was in disagreement with the solution studies where both Y60A bsSHMT and F351G bsSHMT were shown to be active for the L-allo-threonine cleavage reaction. As the crystals were washed with the crystallization buffer containing L-allo-threonine prior to flash freezing in a stream of nitrogen at 100 K, it is possible that the crystals contained unreacted L-allo-threonine although the enzyme is active. To resolve this discrepancy, experiments were repeated where the crystals were washed with reservoir solution without L-allo-threonine and were flash frozen for data collection.

Figure 5 Stereo diagram illustrating the electron density difference map (FoFc contoured at 3σ) corresponding to the external aldimine formed by F351G bsSHMT in the presence of L-allo-threonine (ATH)

The structures obtained had only glycine and not L-allo-threonine at the active site. Interestingly, the density connecting the PLP and amino group of glycine was significantly stronger than that connecting PLP and Lys226 (results not shown). This is in contrast with the structures of glycine binary complexes obtained in presence of glycine, where the enzyme existed predominantly in the internal aldimine form (Figure 3). In addition, in these structures, there was a water molecule at the position of the side chain hydroxy group of L-serine/L-allo-threonine. The observed structure probably represents a mixture of internal aldimine, external aldimine and gem-diamine forms. These observations emphasize the importance of conditions of flash-freezing for studying enzyme–substrate complexes.

Mutant complexes in the presence of glycine and THF/FTHF

The addition of THF to a mixture of bsSHMT and glycine resulted in a pink-coloured product with an absorbance maximum of approx. 495 nm, which is characteristic of a quinonoid intermediate (Figure 2a). A similar increase in absorbance at 495 nm was observed in the case of F351G bsSHMT (Figure 2b). Addition of THF to Y60A bsSHMT–glycine resulted in the appearance of a very small peak at 495 nm (Figure 2d), whereas no peak was observed in the case of N341A bsSHMT–glycine (Figure 2c). A similar result was obtained when FTHF, a structural analogue of THF, was used. FTHF is a slow, but tight binding, inhibitor of SHMT.

The structure of the wild-type enzyme complexed to glycine and FTHF has been determined previously [14]. Attempts to crystallize N341A bsSHMT in the presence of glycine and FTHF resulted in crystals of glycine external aldimine, suggesting that the ternary complex was not formed. These results suggest that, although the overall structure is not affected by the N341A mutation, FTHF and probably THF have lower affinity for the mutant and hence the mutant is inactive for the THF-dependent physiological reaction. Crystallization of N341A bsSHMT in the presence of L-serine and FTHF also resulted in the formation of only L-serine external aldimine complex. Even in the case of the wild-type enzyme, FTHF does not bind in the presence of L-serine, probably due to the steric clash of the formyl group of FTHF with the -CH2OH side chain of the PLP-bound L-serine. Similarly, in Y60A bsSHMT also, there was no electron density for FTHF in the crystals obtained in presence of glycine/L-serine and FTHF.

Addition of FTHF and glycine to F351G bsSHMT resulted in a change of colour of the solution from yellow to pink, as in the case of the wild-type enzyme. Crystals were obtained for F351G bsSHMT in the presence of glycine and FTHF. These crystals were also pink in colour and diffracted to a resolution of 2.7 Å. The crystals belonged to a monoclinic space group with a dimer in the asymmetric unit. The cell parameters were very similar to those of the wild-type ternary complex. These observations suggested that FTHF could be bound to this mutant. Indeed, when the structure was determined, electron density corresponding to FTHF could be clearly seen (Figure 6). The overall structure of F351G bsSHMT ternary complex is similar to that of the wild-type ternary complex. Superposition of the two subunits of F351G bsSHMT and wild-type ternary complexes gave a rms deviation of 0.36 Å for all Cα atoms. In the wild-type ternary complex, the glutamate side chain of FTHF is ordered in the B subunit, leading to the displacement of residues around Phe351 of the same subunit. Such a displacement does not occur in the same region of the A subunit because the glutamate arm of FTHF is not ordered in this subunit. As a result, this region is in different conformations in subunits A and B of the wild-type ternary complex [14]. The structure was examined carefully and it was verified that the crystal contacts are not responsible for this asymmetry in binding of FTHF. However, in F351G bsSHMT, FTHF is bound to both of the subunits with ordered monoglutamate side chains. The position of the residues in the region around Gly351 in both of the subunits agree with that of subunit B of the wild-type ternary complex as shown in Supplementary Figure S1(A) (at This conformation is stabilized by a hydrogen bond between Oγ of Ser349 and Oε of the carboxy group of the ordered FTHF monoglutamate tail. Therefore, the overall rms deviation for the superposition of subunit A of the bsSHMT ternary complex with that of F351G bsSHMT ternary complex is higher (0.40 Å) when compared with the corresponding rms deviation of the B subunits (0.27 Å) (Supplementary Figure S1B). Although FTHF is bound to both of the subunits of F351G bsSHMT, there are small differences in the conformation of FTHF between the two binding sites. This arises mainly due to a small rotation of the pteridine ring and results in greater differences towards the PABA moiety and the monoglutamate part of FTHF (Figure 6). The glutamate side chain of FTHF also has a slightly different conformation and different contacts in the two subunits (Supplementary Table S4 at Another feature observed in the F351G bsSHMT ternary complex is that the density for PLP is not well defined in one of the subunits. In spite of this, F351G bsSHMT is nearly as active as the wild-type enzyme in both THF-dependent and THF-independent reactions.

Figure 6 Stereo diagram illustrating the electron density (2FoFc at 1σ) corresponding to FTHF in subunit A (grey) and superposed on the FTHF in subunit B (black) of F351G bsSHMT


The results obtained from the structural and biochemical studies on N341A bsSHMT clearly suggest that the interaction of Asn341 with THF is essential for THF binding and therefore the mutation of Asn341 to alanine makes the enzyme inactive for the THF-dependent conversion of L-serine into glycine. The mutation of Tyr60 to alanine prevents the binding of FTHF/THF due to the disruption of the stacking interaction of Tyr60 and the PABA ring of FTHF/THF and loss of a hydrogen bond between the tyrosine hydroxy group and monoglutamate carboxylate. From the ternary complex structure of F351G bsSHMT it is evident that the mutation results in the binding of FTHF/THF with an ordered glutamate side chain in both of the subunits and hence decreases the asymmetry observed between the subunits of the wild-type enzyme upon FTHF/THF binding. The mutant is as active as the wild-type enzyme for both THF-dependent and independent functions. Thus, Asn341 and Tyr60 are critical for the binding of THF and hence for the THF-dependent activity, whereas Phe351 is responsible for the asymmetry in the binding of THF.


This work was supported by the Indian Council of Medical Research [grant number 63/168/2001-BMS] and the Department of Biotechnology (DBT) of the Government of India [grant number BT/PR5739/BRB/10/396/2005 (to H. S. S. and M. R. N. M.)], a DBT post-doctoral fellowship to V. R. P.; the Council for Scientific and Industrial Research, Government of India fellowships to V. R. and B. S. B.; and the Indian Institute of Science fellowship to S. B.


We thank Dr V. Prakash for his valuable suggestions during discussions in the preparation of this manuscript. We thank Dr Ambili for her help in cloning. Diffraction data were collected at the X-ray facility for Structural Biology at the Molecular Biophysics Unit, Indian Institute of Science, supported by the Department of Science and Technology and the Depatment of Biotechnology. We thank Mr Babu and Mr James for their help during data collection. We also thank the anonymous reviewers for helpful suggestions.

Abbreviations: 5,10-CH2-THF, 5,10-methylene THF; FTHF, 5-formyl tetrahydrofolate; MPD, 2-methyl 2,4-pentanediol; PABA, p-amino benzoic acid; PLP, pyridoxal 5′-phosphate; rms, root mean square; SHMT, serine hydoxymethyltransferase; bsSHMT, Bacillus stearothermophilus SHMT; THF, tetrahydrofolate


View Abstract