Isolation and studies of the structure and interactions at the active site of sheep liver serine hydroxymethyltransferase

Abstract

Serine hydroxymethyltransferase can be considered as the first enzyme in the pathway for the interconversion of folate coenzymes (Fig. 1) and is the source of C1 fragments for the biosynthesis of a large number of biologically important end products such as nucleic acids, ubiquinone, etc. (Fig. 2) in mammalian systems (5). Efforts to selectively inhibit enzymes of folate metabolism in cancer chemotherapy were directed towards the inactivation of dihydrofolate reductase and thymidylate synthetase, notwithstanding the problems of drug resistance and side effects of these drugs (12–24). Recent studies revealed that the levels of SHMT as well as its kinetic properties were altered in neoplastic tissues (25, 27). These results suggested that this enzyme could also be exploited as an alternate target in cancer chemotherapy (10).

SHMT resembles other PLP dependent enzymes in its mechanism of action and active site structure. PLP enzymes catalyze a variety of reactions of amino acids (Fig. 4, Table 3) involving transamination, decarboxylation, racemization, and elimination and replacement reactions and aldolization (35, 41–43). A common property of all these enzymes is the presence of PLP at the active site in a Schiff base linkage with the amino group of a lysine residue (Fig. 7). Studies with model systems which were extended to PLP enzymes and the use of several sophisticated techniques revealed that the mechanism proceeded via the intermediate formation of aldimine or ketimine form of a substrate–PLP Schiff base complex (Figs. 4, 8). A study of the detailed stereochemistry of these reactions revealed that the reaction pathway followed (Fig. 5) depended upon the nature of the dissymmetry of the orientation of the substrate at the active site (41). The dynamic model (Fig. 9) developed to explain the mechanism of aspartate transaminase (113, 125) has formed the basis for understanding the mechanisms of other PLP enzymes.

SHMT exhibited in addition some unique features (Fig. 13) which distinguished it from the other PLP enzymes (150). The involvement of folate in the reaction facilitated the addition and removal of HCHO from the active site. The detailed mechanism of the substrate C –C bond cleavage at the active site is not yet well understood.

Mechanism based inactivators of PLP enzymes have provided insights into the events in the catalytic cycle and it is believed that these inhibitors could be used to selectively inactivate the enzymes in vivo (183–185). However, success has not been commensurate with the effort made in this direction. Some encouraging results were obtained for the inhibition of SHMT by D cycloserine (27); a detailed understanding of the properties of the active site of SHMT is important for the design of suitable selective inhibitors of the enzyme.

The objective of this investigation was to derive some insight into the structure and function of SHMT. This was hoped to be achieved by isolating SHMT from sheep liver and examining its interactions with substrates, allosteric effectors and reversible and irreversible inhibitors as well as by chemical modification of essential amino acid residues. The regulatory nature of the enzyme from pig kidney, mung bean and monkey liver reported earlier were recently attributed to an artifact of the assay procedure used (203). It was therefore also an objective of this thesis to resolve this controversy.

The experimental methods employed in this thesis included chemical, enzymatic and spectral estimation of H4 folate, a radioactive assay for SHMT, techniques of protein purification, electrophoretic and immunochemical methods, chemical modification of active site residues and absorbance, CD and fluorescence spectroscopy.

The cytosolic SHMT was purified from sheep liver by ammonium sulfate fractionation, CM Sephadex ion exchange chromatography, gel filtration on Ultrogel AcA 34 and Blue Sepharose affinity chromatography (Table 4). The special feature of this procedure was the absence of a heat denaturation step commonly employed in the purification of this enzyme and the use of Blue Sepharose affinity chromatography. The purification procedure gave a yield of 16% and a fold purification of 130. The specific activity of the purified enzyme was about 6.2 units and was the highest reported for the enzyme from any source. The recovery was also higher than by other procedures.

The homogeneity of the purified enzyme was rigorously established by the criteria of PAGE, SDS PAGE, isoelectrofocusing, gel filtration, sedimentation, immunodiffusion and immunoelectrophoresis (Figs. 15–18).

The enzyme had a molecular weight of 210,000 ± 5,000 and was a homotetramer with a subunit molecular weight of 50,000 ± 5,000 (Figs. 19, 20). The s° ,w of the enzyme was 6.8.

The reaction rates were linear up to 25 g enzyme per ml reaction mixture (Fig. 21A) and up to 15 min of assay (Fig. 21B). The enzyme functioned optimally at a temperature of 60 °C (Fig. 21C) and an activation energy of 9.6 kcal/mol was calculated. A pH optimum of 7.2 (Fig. 21D) and a catalytic center activity (at 37 °C and pH 7.4) of 315 min ¹ were also obtained for the enzyme. The PLP bound to the holoenzyme was dissociated by treatment with L cysteine and ammonium sulfate followed by dialysis. The apoenzyme was inactive and was fully reconstituted with PLP (Table 5). The presence of a reducing agent such as 2 ME was required for protecting the enzyme against inactivation on storage.

The enzyme had visible absorbance and CD peaks at 425 and 432 nm respectively, due to bound PLP (Figs. 22, 23). The holoenzyme (but not the apoenzyme) was reduced by [³H] NaBH indicating that PLP was bound as a Schiff base at the active site (Table 5).

The enzyme also had a protein absorbance peak at 278 nm. The aromatic residues in the enzyme gave a characteristic fluorescence excitation maximum at 285 nm and an emission maximum at about 335 nm when excited at 295 or 285 nm (Fig. 26A–C). The near UV CD spectrum showed negative bands at 261, 268, 280 and 300 nm due to aromatic residues and sulfhydryl groups (Fig. 24).

The far UV CD spectrum showed negative bands at 210 and 220 nm. The contents of helical, sheet and random coil structures in the enzyme were calculated (199) to be 20%, 35% and 45% respectively (Fig. 25).

The sheep liver SHMT, purified by this procedure, thus resembled the enzyme obtained from other sources, in its physico chemical and catalytic properties. The enzyme also exhibited the general spectral and chemical properties of PLP enzymes.

The saturation of the enzyme with L serine was hyperbolic, with a Km value of 0.9 mM (Fig. 27).

The H4 folate saturation of the enzyme was sigmoidal, with an nH value of 2.8 and a K0.5 of 0.7 mM (Fig. 28). These values were independent of temperature. Preincubation of the enzyme with L serine reduced the cooperativity of H4 folate interactions to nH = 1.6–1.8 (Fig. 28). The effect of serine preincubation was reversed by dialyzing out serine from the enzyme.

NAD and NADH were negative and positive allosteric effectors of the enzyme, respectively (Table 6, Figs. 29, 30). High concentrations of NADH did not completely abolish the cooperative interactions of H4 folate with the enzyme.

Recently, the homotropic cooperative interactions of H4 folate with SHMT were attributed to oxidation of H4 folate in the assay, at low concentrations (203).

A major difference between the studies demonstrating cooperativity (10, 25, 227) and the recent criticism of these results (203) was the use of different concentration ranges of H4 folate to determine the saturation pattern i.e., 0.1–2.7 mM H4 folate in the former study, as compared with 0.02–0.2 mM in the latter studies. In addition, in the studies demonstrating cooperativity a radioactive assay (188) was used whereas a coupled spectrophotometric assay (203) was used in the other study.

Oxidation of H4 folate at one of the lowest concentrations (0.25 mM) used in this thesis was ruled out by spectral, chemical and enzymatic estimation of H4 folate after incubation under assay conditions (Table 7, Fig. 31). The H4 folate saturation pattern was independent of the time of incubation (between 1–15 min) for estimating the velocity of the enzyme (Table 8) and was unaltered upon carrying out the assays in an atmosphere of nitrogen (Fig. 32). These results ruled out the possibility that oxidation of H4 folate was responsible for the cooperativity observed.

It was also pointed out that the nH value of SHMT varied from 1–3.9 depending upon the source of the enzyme and that the sigmoidicity could be altered by specific effectors emphasizing the specificity of these interactions. It was concluded that cooperative interactions of the enzyme with H4 folate were an inherent property of the enzyme and not an artifact of assay.

L Serine, H4 folate and NAD caused significant quenching of the visible CD spectrum of the enzyme (Fig. 34) indicating a changed environment of the bound PLP at the active site upon the binding of these ligands to the enzyme. The specificity of the environment of the bound PLP was indicated by comparison with the CD spectrum of the enzyme with that of a PLP–BSA complex (Fig. 37). Serine also altered the near UV CD spectrum of the enzyme (Fig. 35). Since the fluorescence properties of the enzyme were not altered in the presence of L serine it was concluded that the change in the near UV CD spectrum might be due to altered environment of the bound PLP.

L Serine, but not H4 folate, protected the enzyme against heat inactivation, indicating that serine provided conformational stability to the enzyme (Fig. 38).

Cibacron blue was a complete inhibitor of the enzyme (Fig. 39). Both NAD and H4 folate protected the enzyme against this inhibition (Figs. 39, 40A, B). These observations and the similarity of the dye difference spectrum of the enzyme–dye complex (Fig. 43) to that of dihydrofolate reductase, which had a dye binding site overlapping the nucleotide and folate binding sites (224), as well as the nature of the visible CD change of the sheep liver SHMT in the presence of the dye (Fig. 41) suggested that the allosteric site and the active site of the enzyme were probably close to each other and that the dye was bound in an overlapping manner to both these sites.

The presence of multiple interacting sites on the enzyme was further evidenced by the partial inhibition produced by dihydrofolate, aminopterin and phospho L serine while folic acid, dichloromethotrexate, DL O methylserine and O acetyl L serine were complete inhibitors (Figs. 44–48).

The kinetic mechanism for the interaction of D cycloserine with the enzyme was established by measuring changes in the activity (Fig. 55), absorbance (Figs. 51–53) and CD (Figs. 56–58) spectra of the enzyme. DCS inhibited the enzyme irreversibly and this interaction was characterized by three irreversible steps: an initial rapid step, followed by two successive steps with rate constants of 5.4 × 10 ³ s ¹ and 1.4 × 10 s ¹ (Fig. 61).

The products of this interaction of DCS with the enzyme were identified by absorbance (Figs. 54, 51), CD (Figs. 59, 56, 58) and fluorescence (Fig. 60) spectra as the apoenzyme and a PLP–DCS Schiff base complex, which was expelled from the active site of the enzyme.

The proposed mechanism (Fig. 61), which is different from those operative in other PLP enzymes, probably accounts for the selective inhibition of SHMT by the drug in vivo. The findings also suggested the requirement for ionization of the p hydroxy group of the substrate amino acid for C –C bond cleavage and also the importance of the substrate carboxyl and hydroxyl groups for anchoring and proper orientation of the amino acid–PLP complex at the active site.

The PLP bound at the active site of the enzyme did not cause any gross conformational change in the enzyme as indicated by the structural similarity between the apo and holo enzymes as determined by a comparison of the difference spectra obtained upon Cibacron blue binding to the apo and holo enzymes (Fig. 64), protein fluorescence excitation and emission spectra (Fig. 65A–D), iodide quenching of tryptophyl fluorescence (Fig. 66), fluorescence polarization (Table 9), far UV CD spectra and thermal denaturation (Fig. 68).

The larger value of the intensity of negative near UV CD spectra in the holoenzyme as compared with the apoenzyme (Fig. 36) was probably due to the bound PLP rather than changes in the environment of aromatic residues. The absence of tyrosine or tryptophan in the active site was indicated by identical I quenching of the fluorescence of holo and apo enzymes, the absence of any effect of L serine on the protein fluorescence and the absence of energy transfer between the PLP at the active site and aromatic residues.

Chemical modification of the enzyme using phenylglyoxal (Figs. 69, 70; Table 10), NEM (Figs. 73, 74; Table 11) and DEPC (Figs. 76–78; Tables 12, 13) indicated that at least one residue each of arginine, highly reactive cysteine and histidine were essential for activity. The second order rate constants for the inactivation by these reagents were calculated to be 0.016 min ¹·mM ¹ for phenylglyoxal, 0.52 min ¹·mM ¹ for NEM and 0.06 min ¹·mM ¹ for DEPC. Differential modification of these residues in the presence and absence of substrates and cofactor (Figs. 71, 75, 78) and the spectra of the modified protein (Figs. 72, 80) suggested that these residues might occur at the active site of the enzyme.

The pH independence of the binding of the carboxyl of the substrate amino acid (168) suggests that arginine might bind this group at the active site; arginine might also bind the carboxyl of the substrate H4 folate (Fig. 83). The histidine residue probably functions as the general base (168, 173) in the enzyme catalysis (Fig. 83). The role of the essential sulfhydryl group is yet to be understood.

It can be concluded from the results presented in this thesis that (i) SHMT from sheep liver is a regulatory protein exhibiting both positive homotropic and positive as well as negative heterotropic interactions with substrate and effectors; (ii) the allosteric and active sites are probably located very close to each other and there occur multiple interacting sites on the enzyme for folate and serine analogues; (iii) D cycloserine inactivates the enzyme by converting it into apoenzyme and this mechanism could reveal several mechanistic features at the active site and help in the design of inhibitors of greater therapeutic value; (iv) the apoenzyme and the holoenzyme are structurally similar, emphasizing that in this case PLP has predominantly a catalytic function; and (v) arginine, cysteine and histidine are essential active site residues of the enzyme.