Crystal structure and solution characterization of the activation domain of human methionine synthase Kirsten R. Wolthers 1, *, Helen S. Toogood 1, *, Thomas A. Jowitt 1 , Ker R. Marshall 2 , David Leys 1 and Nigel S. Scrutton 1 1 Faculty of Life Sciences, University of Manchester, UK 2 Department of Biochemistry, University of Leicester, UK Human methionine synthase (EC 2.1.1.13; 5-methyl- tetrahydrofolate homocysteine methyltransferase, hMS) plays a vital role in folate metabolism and the recyc- ling of homocysteine. It is the only enzyme that liberates tetrahydrofolate (H 4 folate) from methyltetra- hydrofolate (CH 3 -H 4 folate), which is a key metabolite for protein and nucleic acid biosynthesis. The enzyme contains a cobalamin cofactor, and in the highly nucle- ophilic cob(I)alamin state, the cofactor abstracts a methyl group from CH 3 -H 4 folate to form H 4 folate and methylcob(III)alamin (Scheme 1, 1a) [1,2]. The methyl group is subsequently transferred from methylcob(III)alamin to homocysteine to generate methionine and cob(I)alamin (Scheme 1, 1b) [3]. Keywords activation domain; cobalamin-dependent enzyme; methionine synthase; methionine synthase reductase; S-adenosyl-methionine Correspondence N. S. Scrutton, Manchester Interdisciplinary Biocentre and Faculty of Life Sciences, University of Manchester, 131 Princess Street, Manchester M1 7ND, UK Fax: +44 161 306 8918 Tel: +44 161 306 5152 E-mail: nigel.scrutton@manchester.ac.uk Database The atomic coordinates and structure fac- tors (202K) have been deposited in the Pro- tein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers Univer- sity, New Brunswick, NJ, USA (http:// www.rcsb.org) *These authors contributed equally to this work (Received 3 October 2006, revised 22 November 2006, accepted 28 November 2006) doi:10.1111/j.1742-4658.2006.05618.x Human methionine synthase (hMS) is a multidomain cobalamin-dependent enzyme that catalyses the conversion of homocysteine to methionine by methyl group transfer. We report here the 1.6 A ˚ crystal structure of the C-terminal activation domain of hMS. The structure is C-shaped with the core comprising mixed a and b regions, dominated by a twisted antiparallel b sheet with a b-meander region. These features, including the positions of the active-site residues, are similar to the activation domain of Escheri- chia coli cobalamin-dependent MS (MetH). Structural and solution studies suggest a small proportion of hMS activation domain exists in a dimeric form, which contrasts with the monomeric form of the E. coli homologue. Fluorescence studies show that human activation domain interacts with the FMN-binding domain of human methionine synthase reductase (hMSR). This interaction is enhanced in the presence of S-adenosyl-methionine. Binding of the D963E ⁄ K1071N mutant activation domain to the FMN domain of MSR is weaker than with wild-type activation domain. This suggests that one or both of the residues D963 and K1071 are important in partner binding. Key differences in the sequences and structures of hMS and MetH activation domains are recognized and include a major reorien- tation of an extended 3 10 -containing loop in the human protein. This struc- tural alteration might reflect differences in their respective reactivation complexes and ⁄ or potential for dimer formation. The reported structure is a component of the multidomain hMS : MSR complex, and represents an important step in understanding the impact of clinical mutations and poly- morphisms in this key electron transfer complex. Abbreviations AdoMet, S-adenosyl-methionine; AUC, analytical ultracentrifugation; FLD, flavodoxin; FNR, ferredoxin-NADP + oxidoreductase; hMS, human methionine synthase; MALLS, multiangle laser light scattering; MS, methionine synthase; MSR, methionine synthase reductase. 738 FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS The latter half of the reaction highlights additional roles of MS in cell homeostasis (a) the production of methionine, which is an essential amino acid and a precursor in the biosynthesis of S-adenosyl-methionine (AdoMet); and (b) the recycling of homocysteine, which is cytotoxic to vascular endothelial cells and an independent risk factor in coronary arterial disease [4,5]. High total plasma homocysteine coupled with diminished folate pools has also been associated with an increased incidence of neural tube defects in new- borns and Down’s syndrome [6,7]. These medical conditions may arise from either a vitamin deficiency or inborn errors in the gene enco- ding hMS or the gene encoding the enzyme involved in the reactivation of hMS, human methionine synthase reductase (hMSR) [8]. The activity of hMS ceases after % 1–2000 catalytic turnovers with the one-electron oxi- dation of cob(I)alamin (Scheme 1, 2) [3,9]. Human MSR binds to hMS forming a ‘reactivation complex’ and an NADPH-derived electron is shuttled to cob(II)alamin via the FAD and FMN cofactors of MSR (Scheme 1, 3) [9]. Transfer of a methyl group from AdoMet accompanies reduction by MSR, thus converting cob(II)alamin to methylcob(III)alamin and returning MS to the primary catalytic cycle. Reactiva- tion of the Escherichia coli homologue of hMS, MetH, also involves FAD and FMN redox centres; however, the cofactors are components of individual proteins: ferredoxin-NADP + oxidoreductase (FNR) and flavo- doxin (FLD), respectively [10]. To date, the majority of information on the struc- tural and functional behaviour of hMS has been derived from biochemical and biophysical research on MetH [11]. Although the 3D structure of the full- length MetH has not been determined, structures of three individual functional modules have been solved. The N-terminal module, determined from MetH of Thermotoga maritima, consists of two b,a barrels, which each house substrate-binding pockets for homo- cysteine and CH 3 -H 4 folate [12]. The cobalamin is sandwiched between two domains in the central mod- ule [13]. Finally, the C-terminal domain, termed the activation domain, binds AdoMet and forms part of the reactivation complex with hMSR [14]. The struc- ture of the E. coli MetH activation domain is C-shaped, with a twisted antiparallel b sheet as a cen- tral feature. AdoMet binds near the centre of the inner surface of the domain and is held in place by interac- tions with both side-chain and backbone atoms [14]. MetH, and by extension MS, are envisioned to be highly dynamic proteins as both substrate-binding pockets on the N-terminal module (separated by 50 A ˚ ) and the activation domain have to form discrete com- plexes with the cobalamin-binding domain in order to catalyse each of the transmethylation reactions of the primary catalytic cycle and reactivation process [15,16]. hMS has an added level of complexity compared with MetH, as its redox partner, MSR is itself a multi- domain protein; MSR is modelled on the structural family of diflavin reductases, of which cytochrome P450 is the prototype [17]. Enzymes belonging to this class of proteins have a NADPH ⁄ FAD-binding domain, tethered to an FMN-binding protein, which are related to bacterial FNR and FLD, respectively. Studies have shown that with the two-component system of E coli, FLD forms mutually exclusive com- plexes with MetH and FNR [18]. If the FMN-binding region of MSR behaves similarly, this domain would pivot between hMS and the FAD domain of MSR to facilitate electron transfer. Alternatively, the FMN domain is relatively immobile and is sandwiched between the FAD domain and hMS during electron transfer to cob(II)alamin. In this case, the binding interfaces between the activation domain of hMS and MSR may be sufficiently different from that of the MetH activation domain and FLD. Evidence for notable differences in the binding inter- face between MetH and hMS is supported by the poor ability of the FNR ⁄ FLD-reducing system to reactivate hMS and the complete inability of MSR to reactivate MetH [19]. Structural information on the hMS activa- tion domain will help establish those key components on these proteins that make them specific for their respective redox partners. In addition, structural infor- mation will help identify how particular clinical and polymorphic (e.g. P1173L) variations appearing in the MS activation domain compromise the activity of the enzyme and lead to various clinical states [20,21]. Here, we provide insight into the biophysical proper- ties of human MS ⁄ MSR system and we report the crystal structure of a D963E ⁄ K1071N double-mutant CH 3 -H 4 -folate H 4 -folate Cob(I)alamin Methylcob(III)alamin Methionine Homocysteine Cob(II)alamin e AdoMet AdoHyc e 1a 2 3 1b Scheme 1. K. R. Wolthers et al. Human MS activation domain structure FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS 739 of the 38 kDa activation domain of hMS to 1.6 A ˚ resolution. Solution studies demonstrate binding of both the wild-type and mutant activation domains to the FMN domain of the physiological partner protein, MSR. We show that the human activation domain exists as a distribution of monomeric and dimeric forms, with the monomer comprising the main compo- nent in solution. Key structural differences between the hMS activation domain and the E. coli homologue are discussed. Results Crystal structure determination The mutant activation domain structure of hMS was determined by molecular replacement using the struc- ture of the corresponding domain of MetH as a model [14]. The asymmetric unit contains two protein mole- cules, related by near-perfect translational symmetry. The crystallographic and final refinement statistics are summarized in Table 1. The last 1–2 residues are not visible in the electron-density map. Similarly, residues T1001–G1003 and D1072–A1074, which are located in flexible loops, are not visible. In humans, residue Gln1041 in both subunits is in a disallowed region in the Ramachandran plot. It is one of only two residues in the sharp turn between b2 and b3 in the b meander. In several organisms, including E. coli, this residue is glycine (G1008) which can accommodate this geometry. The structure of the activation domain monomer of human MS is C-shaped, the central feature comprising a twisted antiparallel b sheet (Fig. 1A), similar to the E. coli structure [14]. This is not surprising as the two MS proteins share 48% sequence identity in the activa- tion domain, with the human activation domain con- taining several insertions (Fig. 1B). The core of the structure comprises mixed a + b regions, dominated by an antiparallel b sheet, with an overall topology that does not resemble any other AdoMet-binding protein structure [14]. An antiparallel b sheet (b1, b2, b5 and b8) forms the upper part of the structure along with strands b3 and b4, which form a b meander. On the opposite side of the sheets, after the meander, is a region of six a helices and a 3 10 -helix. Helices a2–a4 connect the strands b1 and b2, whereas helices a5–a7 follow strand b5. The long helix a6 is surrounded by the central b sheet on one side, b6–b7 and the short helix a7 on the other side. The C-terminus is dominated by two short helices a10–a11 [14]. The two mutations D963E and K1071N are located after b1 and b4, respectively, in regions distant from the AdoMet-binding site. Owing to their location in flexible surface loops, these muta- tions are not thought to have a major impact on the overall structure, but rather have only localized effects. Structural comparison of the hMS and E. coli MetH activation domains Figure 2A shows a superimposition of the structures of the hMS activation domain and corresponding MetH domain in E. coli. Using the program dalilite [22], the rmsd value of the superimposition is 2 A ˚ with a Z-score of 41.9 (314 residues aligned by C a). Although the two structures look very similar, there are significant differences between them, most strik- ingly in the region of helices a3–a4. In the human enzyme, helix a3 is extended by a further four amino acids followed by the insertion of an extra five amino acids between a3 and the 3 10 2-helix. This has resulted in a dramatic reorientation of these latter two helices, relative to the E. coli enzyme, beginning at residue Leu984 (Leu956 in MetH) and ending at the beginning of the a4 helix. In the E. coli enzyme, this region is oriented so the 3 10 -helix is within 4 A ˚ of helix a7, and within 7 A ˚ of Ile1126 in the AdoMet-binding region. In the human structure, the longer a3 helix ends at the beginning of the extra five amino acids. The 3 10 2- helix is oriented below the beginning of helix a4 and is Table 1. Crystallographic data and refinement values for hMS activation domain double-mutant D963E ⁄ K1071N. All numbers in parentheses represent last outer shell (1.66–1.60 A ˚ ). D963E ⁄ K1071N mutant Data collection Resolution limits (A ˚ ) 20–1.6 (1.66–1.60) Space group P212121 Unit cell (A ˚ )a¼ 77.8 b ¼ 90.1 c ¼ 123.0 Total reflections 110417 Unique reflections 106145 Redundancy 5.3 Completeness (%) 96.3 (96.5) I ⁄ rI 20.1 (3.2) R merge (%) 6.4 (40.2) Solvent content (%) 58 Refinement No. of residues 666 No. of water molecules 1205 R fac ⁄ R free (%) 20.9 ⁄ 22.3 (27.0 ⁄ 29.9) rmsd bond lengths (A ˚ ) 0.006 rmsd bond angles (°) 1.0 Average B-factor (A ˚ 2 ) 25.2 Ramachandran plot Most favoured regions (%) 94.1 Additionally allowed regions (%) 5.6 Generously allowed regions (%) 0.0 Disallowed regions (%) 0.3 (Q1041 both subunits) Human MS activation domain structure K. R. Wolthers et al. 740 FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS close to the three amino acid turn between helices a5– a6 (Fig. 2A). This positions the 3 10 -helix at one of the tips of the C-shaped molecule, preventing it from being in a position near the AdoMet-binding region as in MetH. This creates a new hydrophobic cluster not found in the E. coli enzyme between residues Phe993 and Phe997 and the a3 and a6 helices. Other hydro- phobic residues involved include Trp982, Val1009, Tyr1121, and Ile1124. Part of this region is disordered in the human enzyme, with clear electron density lack- ing for residues Thr1001–Gly1003. The role of this loop)3 10 region in E. coli is uncer- tain, but owing to its location near the AdoMet-bind- ing site it may be involved in interacting with its FLD partner. Recent cross-linking and NMR studies of MetH show that Lys959 (Lys987 in hMS) is located close to Glu61 of FLD [18,23] in the activation domain–FLD complex, supporting the notion that this is a key interaction in the formation of the reactivation complex. hMS interacts with a much larger partner, MSR, which is a cytochrome P450 reductase-like pro- tein constructed by the fusion of two genes encoding a FMN-containing FLD and an FAD-containing ferre- doxin oxidoreductase separated by a large interdomain linker [9,17]. The change in position of this loop)3 10 region of the human enzyme in addition to the shift in position of Lys987 from the equivalent MetH residue Lys959 of 7.9 A ˚ (Fig. 2A) suggests a different mode of interaction between human MS and MSR compared with the E. coli complex. Because no structure is avail- able for MSR, or any structure of a MS–partner com- plex, the exact nature of these interactions remains to be determined. The region containing b3–b5, known as the b mean- der [14], contains a six amino acid insertion in the human structure (Fig. 1A). The b-meander begins with a twist in b2 at the conserved residue Pro1036 (Pro1003 in E. coli), as indicated by an asterisk in Figs 1B and 2A. This meander is oriented % 90° to the central sheet [14]. In the human enzyme, three of these extra residues are located between b3 and b4, forming an additional solvent-exposed 3 10 -helix (3 10 3), absent in the E. coli enzyme. The other three extra amino acids are located in the more disordered region between b4–b5, next to A B Fig. 1. (A) Stereoview of the overall struc- ture of the hMS activation domain. The car- toon is drawn as a gradient from blue at the N-terminus to red at the C-terminus. Red spheres, location of the mutations D963E and K1071N; magenta sphere, clinical muta- tion P1173L. (B) Sequence alignment of human and E. coli MS activation domains, confirmed by structural superimposition. Secondary structure elements were assigned using DSSP [28]. Pink, secondary structure in humans; green, major differ- ences in secondary structure in E. coli compared with humans; orange, additional 3 10 -helices in humans; asterisk, Pro1036 (beginning of b-meander); blue, residues interacting with AdoMet in E. coli. K. R. Wolthers et al. Human MS activation domain structure FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS 741 the mutation K1071N. The equivalent lysine residue in E. coli activation domain (Lys1035) is suggested to be involved in the interaction between MS and FLD [18]. We note that in the two regions of the protein where the major differences occur between the E. coli and human structures, both contain a residue which in MetH is known to interact with FLD. This suggests that these differences between both structures of the activation domains might reflect structural differences in their respective reactivation complexes. Like MetH, the structure of the hMS activation domain contains some well-ordered water molecules in two cavities which interact with residues in the b-mean- der region. This region, containing helices a10–a11, is one of the more conserved regions of the activation domain. It contains many buried ionic and polar resi- dues, forming an extended network of salt bridges and hydrogen bonds similar to the E. coli enzyme [14]. Active-site region A superimposition of the two structures at the MetH AdoMet-binding site shows a high similarity of both residue type and position (Fig. 2B). AdoMet binds near the centre of the inner surface of the domain of the E. coli structure (Fig. 2A). It is held in place by hydrophobic interactions and hydrogen bonds, through both side chain and backbone interactions, and is partly solvent exposed [14]. This is one of the more conserved regions of the activation domain and contains the consensus sequence Arg-X-X-X-Gly-Tyr critical for the binding of AdoMet [14]. The human enzyme structure does not contain AdoMet, although the positions of the active site residues are strikingly similar. Only two of the residues (Tyr1190 and Ala1141) known to directly interact with AdoMet in the E. coli protein are different in human activation domain (i.e. Phe1228 and Ser1179, human numbering). However, these residues in the E. coli protein interact with Ado- Met via backbone interactions only. The main struc- tural difference in the active site residues is the reorienting of the Tyr1177 side chain away from where AdoMet would bind. This suggests only minimal chan- ges in the positions of the active site residues are required for binding of AdoMet. A B Fig. 2. Stereoview of a superimposition of human and E. coli MS activation domains. (A) Overall structures: blue and green car- toons are human and E. coli proteins, respectively; AdoMet from the E. coli struc- ture (1MSK) [14]; is shown as red sticks; Ca atoms of K987 and K959 of human and E. coli activation domains, respectively, are shown as red spheres. (B) AdoMet-binding region: E. coli residues interacting with Ado- Met, and the equivalent human residues, are shown as atom-coloured sticks with green and orange carbons, respectively; AdoMet is shown as purple lines. Human MS activation domain structure K. R. Wolthers et al. 742 FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS Monomer interactions in the crystalline state The crystal structure reveals that both monomers within the asymmetric unit are involved in close inter- actions with symmetry-related molecules, leading to a putative dimeric form. Figure 3A shows two views of the structure of this human MS activation domain ‘dimer’. The contact surface between the two mono- mers, consisting of two nearly identical interaction sites, is % 609 A ˚ 3 with a shape complementarity (S c ) value of 0.769, which is in the range of S c values for surfaces within known dimeric proteins and protein–lig- and interactions [24]. There were no other interactions of sufficient size between monomers of the activation on the structure, and their symmetry-related molecules to be able to accurately calculate S c values. Both the size of the contact surface, the S c value and the fact both monomers in the asymmetric unit form near iden- tical dimers suggests dimer formation could be physio- logically relevant rather than simply a consequence of the crystallization conditions and ⁄ or crystal packing. The central cavity of the dimer consists of a large elliptical-shaped central groove, with extensions at the base of the dimer forming a cross shape. This exten- sion of the cavity is due to the side chains of several residues of the extended a3 helix of both subunits pointing directly into the cavity. The volume of this cavity is calculated to be % 7000 A ˚ 3 using the program CASTp [25]. The smallest part of the cavity (24 · 5A ˚ ) is at the lower region of the dimer (Fig. 3A) close to where AdoMet binds in the E. coli structure [14]. Two AdoMet molecules can be modelled to bind within the dimer, with the large size of the cavity easily allowing substrate entry. However, the side chains of residues Tyr988 and Lys987 of the second molecule of the ‘dimer’ clash with the region where the adenosyl group of AdoMet is bound in the MetH structure. While the side chain of Lys987 can move to accommodate Ado- Met, the side chain of Tyr988 cannot, suggesting the dimeric form is not compatible with substrate binding. Superimposition of the E. coli structure onto the human activation domain ‘dimer’ shows significant cla- shes due to the reorientation of the a 3–3 10 loop region (Fig. 3A). Clearly, the E. coli activation domain is unli- kely to form a human-like dimer. Figure 3B shows the residues involved in forming the dimer interface of the human activation domain. Although there is a signifi- cant contact surface between the monomers, the number of direct interactions is small. The key interacting resi- due is Arg991. The NH2 group of Arg991 forms a hydrogen bond with the carbonyl oxygen of Glu1077 (2.7 A ˚ ). The NH1 and NH2 groups are also within van A B Fig. 3. Dimeric nature of human MS activa- tion domain. (A) Two views of the human activation domain dimer, with the mono- mers shown as green and blue cartoons. AdoMet is superimposed onto the structure as red sticks in the equivalent position found in the E. coli enzyme [11]. E. coli ‘dimer’ superimposed on the human dimer is represented by an olive cartoon. (B) The dimer interface of human activation domain shown as green and blue cartoons for monomers A and B, respectively. Side chains of residues directly interacting at the dimer interface are shown as atom-coloured sticks. Side chains of residues lining a hydrophobic pocket are shown as magenta sticks. K. R. Wolthers et al. Human MS activation domain structure FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS 743 der Waal’s distance of the hydroxyl group of Tyr1079, as is the NE group to the backbone oxygen atoms of Pro1078 and Glu1077. The OD2 atom of Asp1120 forms a salt bridge with the NH2 group of Arg927 (2.7 A ˚ ). The interface is furthermore formed by hydro- phobic interaction between Tyr988 of one subunit with Pro1173, Tyr1177, Pro1178 and Tyr1227 of the second subunit. As the equivalent residues of the E. coli activa- tion domain are located in significantly different posi- tions, these dimer interface interactions are not present. Molecular mass determination and dimer formation A key question to address is whether the dimer species as seen is a significant species in solution. To answer this question, we carried out multiangle laser light scattering (MALLS) and analytical ultracentrifugation (AUC) to determine the proportion of dimer formed in solution under both high and low salt conditions. Table 2 shows the results of the sedimentation and light-scattering analyses. Both wild-type and the dou- ble-mutant activation domains show near identical sedimentation profiles (Fig. 4A) with corrected sedi- mentation coefficients (s 0 20;W ) of 2.87 ± 0.16 and 2.84 ± 0.2, respectively, in 0.4 m sodium acetate buf- fer (i.e. similar to the crystallization conditions). How- ever, in NaCl ⁄ P i and Tris ⁄ HCl buffers in the absence of sodium acetate, the sedimentation coefficient was increased to 2.98 ± 0.2. Thus, the molecular mass of the two proteins are identical, as shown by light scat- tering, and there is no detectable difference in shape between the wild-type and mutant proteins in the same buffer systems (Fig. 4B). However, differences in the solution properties are seen for both proteins in differ- ent buffers. The addition of 0.4 m sodium acetate to the buffers increases the presence of the dimer species (14% of the total protein analysed). The estimated frictional ratios for the monomeric species using the sedimentation coefficient distribution (c(s)), where f ⁄ f 0 ¼ estimated frictional ratio, for 2D size and shape distributions showed that there is a difference in the apparent length or flexibility of the molecules in the two different buffers. In 0.4 m sodium acetate, the molecules exhibited a more extended or more flexible conformation with a frictional ratio of 1.45 compared with a value of 1.31 representing a more compact structure in NaCl ⁄ P i . This corresponds to a difference in hydrodynamic radius of % 3A ˚ (Table 2). Global analysis of mutant and wild-type protein in high salt buffer at different concentrations was per- formed to ascertain whether there was evidence of reversible association occurring between monomeric and dimeric species. The results fitted well to a mono- mer–dimer model. However, only a very small amount of dimer is present. This was true even at very high concentrations of 2 mgÆmL )1 , giving a very low associ- ation constant. This indicates that, although there may be dimer species present within the samples in 0.4 m sodium acetate, there is no real evidence for a dynamic associating system. The molecular mass obtained throughout these experiments was lower than expected (Table 2) for the sequence molecular masses when taking into account the calculated mass for hydrated protein. This might be attributed to a small amount of protease cleavage during purification. The molar mass distribution is near identi- cal for the wild-type and mutant proteins giving an aver- age monomeric molecular mass of 41 600 ± 1020 Da. The high polydispersity detected is similar to the results gained using sedimentation equilibrium indicating that a proportion of the molecules are slightly smaller than sequence molecular mass (Fig. 4C). Some of the dimer species can also be seen using this technique. Analysis of dimer interactions by chemical cross-linking Inspection of the dimer interface reveals that Lys925 of one subunit of the activation domain ‘dimer’ is Table 2. Solution studies of wild-type and mutant activation domain to determine oligomeric states in high- and low-salt buffers. NaCl ⁄ P i ;M, observed hydrated mass of the monomer; S, sedimentation coefficient; R H , radius of the structure; f ⁄ f o , estimated frictional ratios; ND, not determined; NA, not applicable. R Experimental M (kDa) a M (kDa) b S 0 20,W R H a (nm) R H b (nm) f ⁄ f 0 Bead modelling Monomer R H (nm) S 0 20,W Dimer R H (nm) S 0 20,W WT NaCl ⁄ P i ND 43.10 ± 1.5 2.98 ± 0.2 2.83 2.7 ± 0.3 1.27 2.97 3.25 3.56 5.0 Mutant NaCl ⁄ P i ND 43.0 ± 1.5 3.0 ± 0.2 2.92 2.6 ± 0.2 1.31 ND ND ND ND WT HS c 36.20 ± 2.2 40.80 ± 0.8 2.87 ± 0.2 3.23 2.7 ± 0.1 1.45 NA NA NA NA Mutant HS c 35.12 ± 2.0 41.60 ± 1.0 2.84 ± 0.1 3.18 3.0 ± 0.2 1.43 NA NA NA NA a Determined by sedimentation studies. b Determined by light scattering. c 100 mM Tris ⁄ HCl pH 7.0 + 400 mM Na acetate. Human MS activation domain structure K. R. Wolthers et al. 744 FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS likely to be in close proximity to the N-terminal amino group of the second subunit. To determine if this spe- cific surface interaction occurs frequently in solution, an attempt was made to cross-link these two specific primary amine groups with a homo-bifunctional imi- doester cross-linker. Mass analysis of in-gel trypsinized samples showed that although the 80-kDa band observed in SDS–PAGE was indeed a dimer of activa- tion domain molecules, the location of the cross-link indicated a ‘back-to-back’ orientation for the two monomers and is thus inconsistent with the crystal structure (results not shown). This suggests that either: (1) the concentration of the dimer species under cross- linking conditions is very small; (2) the location of the N-terminus in solution is not consistent with models of the dimer (the N-terminus is not visible in the crystal structure of the activation domain); and ⁄ or (3) there is some proteolytic removal of the N-terminus during purification. Analysis of the interaction between MS activation domain and the FMN domain of MSR Titration of the activation domain of MS (both the D963E ⁄ K1071N mutant and wild-type proteins) with the FMN domain of human MSR resulted in a quenching of the fluorescence emission spectra of the flavin cofactor. The decrease in fluorescence intensity showed a hyperbolic dependence on the concentration of the activation domain (Fig. 5). A fit of Eqn (1) to the data yielded the apparent dissociation constant for the complex formed between activation domain and the FMN-binding region of MSR (Table 3). The wild-type activation domain has an apparent K d value 1.3 lm for the FMN domain, which is 10-fold lower than the apparent K d (11.9 lm) for binding of the D963E ⁄ K1071N to the flavin-binding protein. The presence of AdoMet in the fluorescence bind- ing assays resulted in a twofold decrease in the apparent dissociation constant for both the wild- type (0.7 lm) and the mutant (4.7 lm) activation domains. Discussion hMS is important for maintaining adequate levels of methionine and AdoMet, preventing the accumulation of cytotoxic homocysteine, and is essential in methi- onine metabolism. Elevated levels of homocysteine in the blood have been linked to an increased likelihood of developing cardiovascular disease, birth defects, Down’s syndrome and affecting the development of some types of cancer [4–7,26]. Functional deficiency of MS or MSR results in diseases such as homocys- tinuria, hyperhomocysteinemia and hypomethionine- mia [8,20]. A P1173L mutation (magenta sphere in A 0246810 0246810 0.0 0.2 0.4 0.6 0.8 1.0 1.2 c(s) c(s) Sedimentation coefficient (s) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Sedimentation coefficient (S) 0 50000 100 000 150000 200 000 0.0 2.0x10 -5 4.0x10 -5 6.0x10 -5 8.0x10 -5 1.0x10 -4 1.2x10 -4 c(M) Molar Mass (Da) B 10 12 14 16 0 20 000 40 000 60 000 80 000 10 0000 Refractive Index Molar Mass (Da) Volume (ml) 0.0 0.2 0.4 0.6 0.8 1.0 C Fig. 4. (A) Sedimentation coefficient distribution c(s) of wild-type (solid line) and mutant (dotted line) in 10 m M Tris, 150 mM NaCl pH 7.0. Inset shows the same protein run in 0.4 M sodium acetate. (B) Wild-type molar mass distribution c(M) using the estimated fric- tional ratio (f ⁄ f 0 ) of 1.45. (C) Wild-type (solid line) and mutant (dot- ted line) elution from a Superdex 200 gel filtration column in NaCl ⁄ P i +0.4 M sodium acetate, with the absolute molar mass superimposed. The thick dotted line within the major peak shows the degree of polydispersion of the enzyme in NaCl ⁄ P i . K. R. Wolthers et al. Human MS activation domain structure FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS 745 Fig. 1B) in the activation domain of human MS is commonly found among patients exhibiting hype- rhomocysteinemia [21]. This residue is located between Arg1172 and Ala1174, both of which interact directly with AdoMet in the E. coli structure (Fig. 2B). P1173L is located at the start of a loop in the active site and which contains four proline resi- dues of the sequence P-X-P-X-X-P-X-X-P. This sequence is highly conserved among MS enzymes sug- gesting an evolutionary pressure to retain this struc- ture in the active site of the activation domain. Other known clinical mutations of the human activation domain include H920D, E1204X (early termination), and insertion ⁄ deletion mutations [21]. The suggestion of a dimer of human activation domain in the crystal structure as well as in some solu- tion studies raises some important issues. As the acti- vation domain is the C-terminal domain of MS, separated from the rest of the protein by a 38-residue linker region, a key question to be addressed is the possibility of the full-length enzyme forming a dimeric structure. Also, the possibility of binding Ado- Met ± the FMN domain of MSR influencing the like- lihood of dimer formation needs to be considered. That the major differences between the hMS and MetH activation domains occur in those regions involved in dimer association is intriguing. This might reflect the need to recognize different redox partners, i.e. FLD versus MSR, but a role in dictating the oligo- meric state of MS cannot be ruled out. That said, only a small proportion of dimer is found in solution stud- ies, and this is most prevalent under high salt condi- tions. Clearly, further structural analysis of other components of the MS holoenzyme is required to ascertain if full-length hMS possesses higher order quaternary structure. Fluorescence titration assays have demonstrated an interaction between the FMN-binding domain of MSR and the activation domain of MS. The appar- ent K d value determined from these binding assays is similar to that reported for the interaction between E. coli FLD and its redox partner, MetH [23]. The decrease in dissociation constant for the complex with the wild-type and mutant form of the activation domain in the presence of AdoMet suggests that binding of the substrate for the transmethylation reaction might effect a conformational change in the activation domain to increase its affinity for the FMN domain of MSR. The 10-fold higher dissociation constant measured for the mutant activation domain ⁄ FMN domain inter- action indicates that Asp963 and ⁄ or Lys1071 are important residues in this interaction. Several lysine mutants of E coli FLD have also showed a marked decrease (3- to 70-fold) in affinity towards MetH [23], suggesting that salt bridges are a key recognition 02468101214 0 2 4 6 8 10 [Act domain] μM A 0 5 10 15 20 25 30 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Δ Fluorescence (AU) Δ Fluorescence (AU) [Act domain] μM B Fig. 5. Fluorescence titration of the FMN domain of MSR with the wild-type (A) and D963E ⁄ K1071N double-mutant (B) of the MS acti- vation domain. The FMN domain at 0.25 l M was titrated with both forms of the purified activation domain under conditions described in the Experimental procedures. The binding assays were per- formed in the absence (d) or presence (s)of1m M S-adenosyl methionine. The change in the FMN fluorescence intensity was plotted versus the concentration of the activation domain and the curves show the best fit of the data to the quadratic Eqn (1). Table 3 lists the calculated dissociation constants for the wild-type and D963E ⁄ K1071N alone and in the presence of AdoMet. Table 3. Fluorescence titration of the FMN domain of MSR with the wild-type and D963E ⁄ K1071N double-mutant of the MS activa- tion domain. The FMN domain at 0.25 l M was titrated with both forms of the purified activation domain under conditions described in the Experimental Procedures. Enzyme K d (lM) No AdoMet 1mM AdoMet Wild-type 1.3 ± 0.1 0.7 ± 0.1 D963E ⁄ K1071N 11.9 ± 1.5 4.7 ± 0.6 Human MS activation domain structure K. R. Wolthers et al. 746 FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS feature at the binding interface in both the E. coli and human systems. While this study describes the structure of a double- mutant of the activation domain, the location of the mutations are on flexible, surface-exposed loops. We conclude from this observation, combined with the MALLS and AUC data which showed that the wild- type and mutant activation domains have identical shapes, that the structure of the mutant domain resem- bles closely that of the wild-type activation domain. To further understand the process of MS reactiva- tion, a structure of the cob(I)alamin ± activation domains in complex with at least the FMN domain of MSR is needed (studies that we are currently pursuing). The structure of the activation domain reported here has allowed us to gain insight into the likely mode of AdoMet binding, and provide atomic level insight into the effect of clinical mutations on the activity of MS. Experimental procedures Cloning and mutagenesis The cDNA encoding the activation domain of methionine synthase gene was cloned by PCR amplification, using nondegenerate oligonucleotides based on the published sequence. Total RNA was isolated and purified from whole human blood using a High Pure RNA isolation kit (Roche Diagnostics, Welwyn Garden City, UK). cDNA was gener- ated by reverse transcription using the Titan RT PCR sys- tem (Roche Diagnostics). The MS gene was cloned into the plasmid vector pET15b, and the 1184 bp PCR product cor- responding to the 3¢-terminus of the MS gene was gener- ated with Pfu turbo DNA polymerase (Stratagene, La Jolla, CA) and ligated into a pGEMT vector (Promega, Madison, WI) to generate the vector pGEMMB. The sequence encoding the activation domain (residues 925– 1265) was subsequently amplified by PCR from the pGEMMB vector, incorporating the restriction sites NcoI and HindIII into the 5¢- and 3¢-regions, respectively. These restriction sites were used to subclone the PCR product into pET23d, generating clones containing a C-terminal His-tag. Sequencing of the resulting vector, pACT, revealed two missense mutations that resulted in conversion of an Asp at position 963 to a Glu and a Lys to an Arg at posi- tion at 1071. (At this stage we cannot say if the identified differences to the published sequence of human methionine synthase (GenBank accession number Q99707) represent polymorphic variation of the MS gene.) The pACT vector was subsequently mutated to revert the sequence back to wild-type. Both the wild-type gene and the uncorrected mutant D963E ⁄ K1071N were subsequently transformed into competent E. coli BL21(DE3) (Stratagene). MSR FMN domain was cloned as a glutathione-S-transferase fusion protein according to the method of Wolthers et al. [27]. Protein production and purification Wild-type and mutant activation domain proteins were expressed in Terrific broth containing 100 lgÆmL )1 ampi- cillin. Cells were lysed by sonication in buffer A (20 mm K 2 HPO 4 ⁄ KH 2 PO 4 buffer, pH 7.4, and 0.5 m NaCl) also containing 1 mm MgCl 2 , EDTA-free Complete protease inhibitor tablets (Roche) and Benzonase (Merck Bio- sciences Ltd., Nottingham, UK). The proteins were puri- fied by running through Ni-NTA resin contained in 5 mL HisTrap column (Amersham Biosciences, GE Healthcare, Little Chalfont, UK) in Buffer A. Proteins were eluted in a gradient from 0 to 0.5 m imidazole. For the final puri- fication step, the protein was loaded onto HiPrep Q Sepharose (Amersham Biosciences, GE Healthcare) in 50 mm Tris ⁄ HCl pH 8.0 (Buffer B) and eluted in a gradi- ent from 0 to 1 m NaCl. The expression and purification of the FMN domain of MSR was carried out according to a modification of the method of Wolthers et al. [27]. An additional purification step was performed using Resource Q (Amersham Bio- sciences, GE Healthcare) in Buffer B. The protein was elut- ed in a gradient of 0–0.5 m NaCl. Crystallogenesis and data collection Crystals of the wild-type and mutant MS activation domains were grown using the sitting drop vapour diffusion method at 19 °C. The enzyme (10 mgÆmL )1 ) was desalted into 10 mm Tris pH 7.0 containing 0.1 mm EDTA and 0.5 mm dithiothreitol. The reservoir solution comprised 0.1 m Tris ⁄ HCl pH 7.5 containing 0.1–0.4 m sodium acetate, 10–12% poly(ethylene glycol) 8000 and 10–12% poly(ethy- lene glycol) 1000. Crystals appeared between 2 and 14 days. Crystals were soaked in mother liquor supplemented with 5% poly(ethylene glycol) 200 as a cryorotectant, before being flash-cooled in liquid nitrogen. A full 1.6 A ˚ data set was collected on a single crystal of the mutant MS activa- tion domain at the European Synchrotron Radiation Facil- ity (Grenoble, France) on ID14-EH1 using an ADSC Q4 CCD detector. Owing to problems associated with twin- ning, a model could not be obtained for data collected from wild-type crystals. Structure determination and refinement Data were processed and scaled using the HKL package programs denzo and scalepack [28]. The structure was solved via molecular replacement using the program amore [29] and the deposited structure of the MS activa- tion domain from E. coli MetH [14] as the search model K. R. Wolthers et al. Human MS activation domain structure FEBS Journal 274 (2007) 738–750 ª 2006 The Authors Journal compilation ª 2006 FEBS 747 [...]... a 3.0 A X-ray structure of B12-binding domains of methionine synthase Science 266, 1669–1674 Human MS activation domain structure 14 Dixon MM, Huang S, Matthews RG & Ludwig M (1996) The structure of the C-terminal domain of methionine synthase: presenting S-adenosylmethionine for reductive methylation of B12 Structure 4, 1263–1275 15 Bandarian V, Pattridge KA, Lennon BW, Huddler DP, Matthews RG & Ludwig... constant for the FMN domain activation domain complex qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DF ¼ Fo þ 2DFmax ðE0 þ L þ Kd Þ À ðE0 þ L þ Kd Þ2 À 4E0 L ð1Þ In Eqn (1), DF represents the change in fluorescence intensity, F0 is the measure of the fluorescence intensity in the absence of the activation domain, E0 is the concentration of the FMN domain, L denotes the concentration of the activation domain. .. added to the sample, and Kd represents the dissociation constant for the FMN domain activation domain complex Sc ¼ ðfSA!B g þ fSB!A gÞ=2 Where the braces denote the median (50th percentile) of the distribution of SA fi B and SB fi A values over two surfaces A and B, respectively A full discussion of the calculations can be found in of Lawrence and Colman [24] These calculations were done on both the dimer.. .Human MS activation domain structure K R Wolthers et al (PDB code 1MSK) Positional and B-factor refinement was performed using refmac5 [30] with alternate rounds of manual rebuilding of the model in turbo-frodo [31] Partial rebuilding of the model and building of waters was carried out using arp/warp [32] The quality of the model was checked using the program procheck [33] Determination of the oligomeric... set at 10 and 5 nm, respectively After addition of activation domain (in 2–10 lL aliquots), the sample was mixed and incubated for 1 min before the stable fluorescence emission spectra were recorded The change in the fluorescence intensity at the emission maxima (529 nm) was plotted against the concentration of the activation domain and the data were fit to a quadratic binding Eqn (1) to obtain the dissociation... standard conditions using a  value of m 0.734 calculated from the amino acid sequence of the activation domain Sedimentation equilibrium experiments were carried out using six-sector cells at speeds of 5881, 16 336 and 32 000 g at concentrations of 1, 2.5 and 5 lm Global analysis of the results was performed using sedphat [35] The shape complementarity of the dimer interface was investigated using the. .. the oligomeric state of human activation domain in solution The absolute hydrated molecular mass of hMS activation domain was determined by MALLS coupled to gel-filtration chromatography Approximately 50 lg of both wildtype and mutant activation domains were loaded onto a Superdex 200 24 ⁄ 30 gel filtration column (Amersham Pharmacia Biotech, Piscataway, NJ) at 0.5 mLÆmin)1 in either NaCl ⁄ Pi buffers... cblG): clinical and biochemical heterogeneity Am J Med Genet 34, 427– 434 9 Olteanu H & Banerjee R (2001) Human methionine synthase reductase, a soluble P-450 reductase-like dual flavoprotein, is sufficient for NADPH-dependent methionine synthase activation J Biol Chem 276, 35558–35563 10 Fujii K, Galivan JH & Huennekens FM (1977) Activation of methionine synthase: further characterization of flavoprotein... [24] The shape correlation between interacting surfaces A and B can be defined as: Cross-linking of MS activation domain The homo-bifunctional imidoester cross-linker, dimethyl pimelimidate 2 HCl (Pierce, Rockford, IL) with a spacer ˚ arm of 9.2 A, was used in cross-linking reactions with purified activation domain A 10- and 25-fold molar excess of dimethyl pimelimidate was added to the activation domain, ... between MS activation domain and MSR FMN domain Fluorescence quenching experiments were performed in 50 mm Tris ⁄ HCl pH 7.5 at 25 °C in a 3 mL volume on a Cary Eclipse Fluorescence Spectrophotometer (Varian, Oxford, UK) The FMN domain (0.25 lm) was excited at 450 nm, the absorbance maxima for the FMN cofactor, and emission spectra were recorded between 500 and 600 nm with the excitation and emission . domain and the FMN domain of MSR Titration of the activation domain of MS (both the D963E ⁄ K1071N mutant and wild-type proteins) with the FMN domain of human. Crystal structure and solution characterization of the activation domain of human methionine synthase Kirsten R. Wolthers 1, *, Helen