Structure and function study of the complex that synthesizes S-adenosylmethionine

MAT2 complex is expressed in nearly all tissues and is essential in providing the necessary SAMe flux for methylation of DNA and various proteins including histones. X-ray crystallography and solution X-ray scattering structures of this complex show an unexpected stoichiometry, offering a unique mechanism of regulation and thus providing a gateway for structure-based drug design in anticancer therapies including liver disease.

S-Adenosylmethionine (SAMe) is the principal methyl donor of the cell and is synthesized via an ATP-driven process by methionine adenosyltransferase (MAT) enzymes. It is tightly linked with cell proliferation in liver and colon cancer. In humans, there are three genes, mat1A, mat2A and mat2B, which encode MAT enzymes. mat2A and mat2B transcribe MAT2 and MAT enzyme subunits, respectively, with catalytic and regulatory roles. The MAT2 complex is expressed in nearly all tissues and is thought to be essential in providing the necessary SAMe flux for methylation of DNA and various proteins including histones. In human hepatocellular carcinoma mat2A and mat2B genes are upregulated, highlighting the importance of the MAT2 complex in liver disease. The individual subunits have been structurally characterized but the nature of the complex has remained elusive despite its existence having been postulated for more than 20 years and the observation that MAT is often colocalized with MAT2. Though SAMe can be produced by MAT(2) 4 alone, this paper shows that the V max of the MAT2 complex is three-to fourfold higher depending on the variants of MAT that participate in complex formation. Using X-ray crystallography and solution X-ray scattering, the first structures are provided of this 258 kDa functional complex both in crystals and solution with an unexpected stoichiometry of 42 and 2V2 subunits. It is demonstrated that the N-terminal regulates the activity of the complex and it is shown that complex formation takes place surprisingly via the C-terminal of MATV2 that buries itself in a tunnel created at the interface of the MAT(2) 2 . The structural data suggest a unique mechanism of regulation and provide a gateway for structure-based drug design in anticancer therapies.

Introduction
Transmethylation, the transfer of a methyl group between molecules, plays a central role in fundamental biological processes such as cell growth, gene expression and apoptosis (Lu & Mato, 2012). The S-adenosylmethionine (SAMe) molecule (Landgraf & Booker, 2013), which is synthesized by methionine adenosyltransferase (MAT), is the main source of methyl groups in all living organisms. MAT enzymes are conserved from bacteria to mammals, thus highlighting their essential regulatory function in maintaining the appropriate levels of SAMe.
Mammals express three MAT genes, mat1A, mat2A and mat2B; the former two encode two homologous MAT catalytic subunits, MAT1 and MAT2, and the last one encodes the regulatory subunit MAT. Association of MAT and MAT subunits results in the formation of the MAT complex. Recently, it was shown that there exist two major splicing variants of the mat2B gene that encode two proteins MATV1 and MATV2. Both isoforms interact with MAT2 (Yang et al., 2008;Xia et al., 2010), revealing the existence of two new complexes: MAT2V1 and MAT2V2 (Fig. 1). MAT2 expression confers a cell growth advantage and is considered increasingly important for differentiation and apoptosis (Lu & Mato, 2012), for instance in human hepatocellular carcinoma (HCC) (Yang et al., 2008), colon cancer (Chen et al., 2007) and leukaemic cells (Attia et al., 2008). MAT2 interacts with a large variety of proteins both in the nucleus and cytoplasm of mammalian cells, including HuR (Xia et al., 2010), an mRNAbinding protein known to stabilize the mRNA of cyclins; GIT1 (Peng et al., 2013), a scaffold protein that activates ERK1/2; MafK (Katoh et al., 2011), a Maf family transcription factor; and many others. Global histone methylation decreases upon MAT knockdown in Caenorhabditis elegans causing the release of heterochromatin from the nuclear periphery (Towbin et al., 2012). Recently it has been demonstrated that threonine and SAMe metabolism are coupled in pluripotent stem cells, resulting in regulation of histone methylation (Shyh-Chang et al., 2013). Similarly, disruption in mice of GNMT, the enzyme that catalyses the methylation of glycine to synthesize sarcosine, and overexpression in ovary cancer cells of NNMT, the enzyme that synthesizes 1-methylnicotinamide by methylation of nicotinamide, cause aberrant DNA, histone and phospholipid methylation by altering cellular SAMe (Martínez-Chantar et al., 2008;Ulanovskaya et al., 2013). GNMT knockdown has been shown to attenuate prostate cancer (Sreekumar et al., 2009). This has led to an emerging paradigm in enzyme regulation, where MAT2 synthesizes SAMe locally to support specific methylation reactions (Kaelin & McKnight, 2013;Gibson & Kraus, 2011;Martinez-Una et al., 2013).
The synthesis of SAMe has been proposed to follow an SN 2 catalytic mechanism (Markham et al., 1987) that has gained structural support from the crystallographic analyses of MAT catalytic subunits bound to various ligands (Komoto et al., 2004;Gonzá lez et al., 2003;Shafqat et al., 2013). Briefly, the reaction is initiated by the sulfur atom of the methionine which carries a nucleophilic attack against the C5 0 atom of ATP followed by the hydrolysis of the tripolyphosphate (PPPi) (Supplementary Fig. S1A). Despite the central role of MAT2, the nature of the complex has remained elusive and its structural elucidation a challenge. Here we report the crystallographic and solution X-ray scattering structures as well as biophysical analysis of the MAT2 complexes, providing the basis for fresh insight into this widely utilized system in biology.

Results and discussion
2.1. MATb isoforms interact with MAT(a2) 2 through the C-terminal region Crystallographic structures of the 258 kDa MAT2V2 complex determined in different forms to a resolution ranging from 2.35 to 3.3 Å reveal that the complex consists of a 185 kDa MAT2 tetramer which is flanked by two MATV2 subunits of 36.5 kDa each (Fig. 2a). The complex has four active sites located at the interface of the MAT2 dimers. The oligomeric state of our crystallographic structure is different to the suggested tetrameric form [MAT(2) 2 () 2 ] (Kotb & Kredich, 1985) or the recently proposed model in which MAT was assumed to be a trimer [MAT(2) 2 () 1 ] .
In the first study Kotb and Kredich used analytical ultracentrifugation analysis to estimate the molecular weight of the complex whereas in the second study Gonzales et al. used ITC (isothermal titration calorimetry) to investigate its stoichiometry which matches our model in a 2:1 ratio of MAT2:MAT subunits. In order to confirm that the crystallographic oligomer is representative of the solution state and is thus physiologically relevant, we performed small-angle X-ray scattering (SAXS) experiments using the HPLCintegrated SAXS set-up at the SOLEIL synchrotron. The scattering curve of the complex in solution obtained by SAXS is in agreement with the theoretical curve calculated from our 2.35 Å crystal structure ( 2 = 3.2) (Fig. 2b), confirming that the MAT2 complex in solution does indeed have the same composition and overall conformation as observed in our crystal structures, i.e. MAT(2) 4 (V2) 2 (Fig. 2c).  Schematic representation of the oligomeric states of mammalian MAT enzymes. The mammalian genes mat1A and mat2A produce the catalytic subunits MAT1 and MAT2, respectively, sharing 84% sequence homology. The MAT1 and MAT2 subunits can be found organized as dimers and tetramers. The MAT1 dimer and tetramer are known as MATIII and MATI, respectively. On the other hand, mat2B encodes the regulatory subunit for which there are two major isoforms, MATV1 and MATV2. The MAT(2) 4 (V2) 2 complex consists of a MAT2 tetramer flanked by two MATV2 subunits. In this study we were able to assemble in vitro three different MAT complexes: MAT(2) 4 (V1) 2 , MAT(2) 4 (V2) 2 and MAT(1) 4 (V1) 2 (the last two complexes are described for the first time).
The crystallographic structure of the MAT(2) 4 (V2) 2 complex reveals that MATV2 interacts with MAT2 through the insertion of the C-terminal tail of the subunit into a cavity created at the interface of the MAT2 dimer. Specifically, residues K315 to H323 of MATV2 establish extensive hydrophobic and polar interactions with side chains of both MAT2 monomers (Fig.  3a). In the complex, the usually disordered tail of the MATV2 C-terminal folds into a helical structure (Fig. 3b), and within the MAT(2) 2 binding cavity the interaction generates a dilation of the cleft without any change in the orientation of the side chains consistent with a 'lock-and-key' mechanism. In order to establish that the C-terminal tail is indeed the key region of the interaction, we generated a truncated version lacking the last 15 residues at the C-terminus in both variants of MAT, MATV1 and MATV2. These deletions, though preserving the secondary structure, as confirmed by circular dichroism (CD) spectra ( Table 1). Taken together, these data show that MATV2 and MATV1 interact with MAT(2) 2 through the insertion of the C-terminal tail of the subunit. Interestingly, the tunnel created at the interface of the MAT2 dimer has a symmetry that allows two possible conformations of the C-terminal MATV2 (Figs. 4a and 4b). The interchangeability of these two orientations could allow certain rotational flexibility of the MATV2 subunit.

NADP-binding site
Previously, it has been shown that NADP binds to a conserved glycine-rich GXXGXXG motif (G 24 ATG 27 LLG 30 ) at the N-terminal domain of MAT (Shafqat et al., 2013) (Figs. S1C and 5a). The mutants lacking residues involved in NADP binding of the subunit have been shown to be able to form a complex with MAT2 supporting the idea that NADP is not needed to form the complex (Gonzá lez et al., 2012). In our case, co-crystallization and soaking experiments with an excess of NADP yielded crystals of the MAT(2) 4 (V2) 2 complex without NADP. This observation prompted us to evaluate whether  Table 1 Thermodynamic parameters of MAT complex formation.

Figure 2
Structure of the MAT(2) 4 (V2) 2 complex. (a) Crystallographic structure of the MAT(2) 4 (V2) 2 complex. Two MAT2 monomers with visible gating loops are coloured in slate whereas the other two MAT2 monomers are coloured in red. MATV2 is coloured in green with the first visible residues at its N-terminal in orange. The reaction product (SAMe) is represented by spheres. (b) SAXS of the MAT(2) 4 (V2) 2 complex, the experimental spectrum (blue) is shown with a simulated fit (red) obtained from the crystal structure (q max = 0.3 Å À1 ) ( 2 = 3.2). The radius of gyration, R g = 50.1 AE 0.05 Å , was estimated from the low-angle scattering region by a Guinier plot. The distance distribution function, P(r), with maximum linear dimension D max = 187 Å . We interpret the higher 2 value as arising in part from the flexibility of the interaction and the missing residues at the N-terminus of both MAT subunits in the crystal structure. This is supported when SAXS data of the MAT2V1Á16 mutant (slightly better quality data, data not shown) are used with the model from the crystal structure ( 2 = 1.7). (c) The ab initio shape reconstruction of MAT(2) 4 (V2) 2 by DAMMIN using P1 symmetry, showing a good agreement between the predicted molecular shape (light green mesh) and the crystal structure (R g = 50.7) (cartoon), after alignment by SUPCOMB (Kozin & Svergun, 2001) (NSD = 0.93).

Table 2
Thermodynamic parameters of NADP binding to MAT isoforms. (2) Table 2). The presence of NADP had no effect on MAT(2) 4 (V2) 2 complex formation as observed by gel filtration and native PAGE experiments (Figs. S2C and S2D). The overall structure of MATV2 within the MAT(2) 4 (V2) 2 complex has no significant structural changes as compared with the NADPbound MAT structure (r.m.s.d. = 0.42 Å ) (Figs. 5a and 5b). These data taken together suggest that after complex forma-tion the NADP is displaced from its binding pocket. It is possible that NADP binding to MATV2 is relevant for other functions, such as the interaction with HuR, GIT1, MEK, ERK or MafK (Xia et al., 2010;Peng et al., 2013;Katoh et al., 2011). Furthermore, the observation that MAT(2) 4 does not block the binding pocket ( Fig. 5c) suggests that the NADP could bind to the MAT(2) 4 (V2) 2 complex in the context of the interaction with other proteins. In fact, there is evidence suggesting that MAT enzymes are involved in the regulation of many pathways, some of which are chromatin based and some may be independent of SAMe.

Interaction of MATb isoforms with MATa1
Looking at the structure of the MAT(2) 4 (V2) 2 complex it is difficult to understand why, in vivo, MAT has so far been found to only interact with MAT2, even though MAT1 and MAT2 have a homology of 84% and share the same folding (Fig. S3). A plausible explanation may be that in vivo when the expression of mat1A is switched on, mat2A and mat2B expression is switched off. Thus, whereas in adult hepatocytes mat1A is highly expressed and the expression of mat2A and mat2B is low, in HCC where the expression of mat2A and mat2B is turned on, as well as the expression of other proteins that interact with MAT such as HuR, GIT1, MEK, ERK or MafK (Xia et al., 2010;Peng et al., 2013;Katoh et al., 2011), mat1A expression is low or absent (Lu & Mato, 2012). Accordingly, we observed by gel filtration and ITC the formation of the MAT(1) 4 (V1) 2 complex upon incubation of MATV1 with MAT1 (Figs. S4A and S4B; Table 1). Remarkably, we did not observe a strong interaction by gel filtration between MATV2 with MAT1 ( Fig. S4C), indicating a role of the N-terminus of MAT in providing the stability of the MAT1 complex. To confirm this hypothesis, we generated a truncated version of MATV1 lacking the first 16 amino acids (MATV1Á16). This mutant produces a much smaller amount of complex with MAT1, supporting the hypothesis that the MAT N-terminus is indeed important for the formation of stable MAT1 complexes (Fig. S4D).

Active site and enzymatic activity
After incubation of the MAT(2) 4 (V2) 2 complex with its product SAMe, its substrate MET, ATP or AMPPNP (nonhydrolysing ATP analogue), the presence of SAMe, adenosine (ADO) or PPNP [(--imido)triphosphate] was clearly observed in different crystals at the active site (Figs. 6a, 6b and 6c) as supported by the corresponding omit maps (Figs. 6d, 6e  and 6f). In all of the structures the adenine group makes astacking interaction with F250 of MAT2, supporting the hypothesis that the substrate (ATP) and product (SAMe) can occupy the active site in similar orientations. The structures show that two of the four active sites are occupied by SAMe or adenosine whereas the other two are empty, thus providing details of structural differences that accompany SAMe formation by comparison of empty and occupied sites.
A comparison of the active site in two different states reveals that loops flanking the empty catalytic pockets are disordered (Fig. 2a), in particular the 'gating loop' (residues 113-131) that has been proposed to act as a dynamic lid controlling access to the active site (Komoto et al., 2004). The    flexibility of this loop induces two different conformations of the catalytic subunit. The loop is disordered in the open conformation causing the entrance to the active site to be opened. The entrance to the active site is blocked in the closed conformation and the gating loop becomes well ordered (Fig. 7a). In the case of the catalytic subunit MAT1, S-nitrosylation of residue C121 in the 'gating loop' promotes the inactivation of the enzyme (Pé rez- Mato et al., 1999). A comparison of the SAMe-bound MAT(2) 4 (V2) 2 complex with SAMe-bound MAT(2) 4 (PDB entry 2p02; Shafqat et al., 2013) shows that in the absence of MAT(V2) 2 the four active sites of MAT(2) 4 are in a closed conformation. The complex formation with MAT(V2) 2 causes an asymmetry in MAT(2) 4 , in which two sites are found in an open state while the other two are in a closed conformation (Fig. 1a). In the absence of SAMe, the apo-structures exhibit no density for the gating loop, indicating its flexible nature (Fig. 7b). However in MAT(2) 4 (V2) 2 the open active sites show two additional flexible loops, near the inserted MATV2 C-terminus (Fig.  7c). In addition, the N-terminal loop (residues 1-13) of both MATV2 in the complex, which is orientated to the same side of the open active sites, is disordered. This observation raises the question of whether the N-terminus of MAT could regulate the gating loop of MAT2. If this was the case, the differences between the N-terminus of MAT2V1 and MAT2V2 should affect their enzymatic activities. Thus, we compared the activity of the MAT(2) 4 (V1) 2 , MAT(2) 4 (V2) 2 and MAT(2) 4 (V1Á16) 2 complexes. Notably, the presence of either MATV1 or MATV2 increased the V max of MAT(2) 4 without altering the K m for methionine. Additionally, the V max of MAT2 was 34% higher in the presence of MATV1 than with MATV2, thus emphasising that differences at the N-terminus affect the activity of MAT(2) 4 . Furthermore, the MAT(2) 4 (V1Á16) 2 complex, in which MAT has the shortest Nterminus, also has the lowest V max which is still higher than the catalytic subunit alone ( Fig. 7d and Table 3). Therefore, the increasing N-terminal length correlates with an increase in V max of the complex activity, confirming the role of the N-terminal in the regulation of the activity.
A more extensive mutational analysis shows by gel filtration that the minimum motif required for the formation of the MAT(2) 4 (V2) 2 complex comprises three residues at the end of the C-terminal of MATV2 (Val 321 Phe 322 His 323 ). This motif interacts with the loop that recognizes the tripolyphosphate of the ATP at the active site, suggesting a possible allosteric mechanism that could be responsible for the observed increase of the V max of MAT(2) 4 (V1Á16) 2 in comparison with the catalytic subunit alone (Fig. 4c).
In summary, we propose that the Nterminal loop of MAT acts in a concerted manner with the C-terminus motif, helping the release of the product by making the active site solventaccessible for the next substrate to be processed. MAT2 activity and SAMe  Table 3 Kinetic properties of MAT2 complexes. MAT(2) 4 MAT (2)   levels appear to be tightly linked with cell proliferation, e.g. upregulation of MAT2 and/or MAT provides a growth advantage in hepatoma and colon cancer cells (Attia et al., 2008;Yang et al., 2008;Lu & Mato, 2012;Xia et al., 2010;Chen et al., 2007). Similarly, T-leukaemic cells exhibit higher MAT2 activity but, remarkably, when MAT2 expression is inhibited SAMe levels decrease and there is more apoptosis (Attia et al., 2008;Jani et al., 2009). In this regard, regulating SAMe production might be an option for potential anticancer therapies. The structure of the MAT(2) 4 (V2) 2 complex presented here has direct implications for a broad range of SAMe-based biochemistry (Landgraf & Booker, 2013;Kim et al., 2013). Our results show that the complex MAT(1) 4 (V1) 2 is stable in vitro, raising the possibility that this complex may exist during the transition of the expression between different isoforms. It also provides a gateway for structure-based drug design with the aim of searching lead compounds that regulate the levels of SAMe without interfering with the catalytic reaction for its synthesis.

Protein expression and purification
MAT1, MAT2 constructs were kindly provided by SGC Oxford and MATV1, MATV2 constructs were from the laboratory of SCL. MATV1 and MATV1Á16 isoforms were cloned in the HIS-parallel vector via NcoI and XhoI sites, and MATV2 was cloned in the pET-28a(+) vector via Ndel and Xhol sites. The expression of MATV1 was carried out in Escherichia coli BL21(HC41) and the expression of the other three proteins in E. coli BL21(DE3) strain. Cells were grown in LB medium at 37 C to an A 600 = 0.6-0.8 at which protein expression was induced by the addition of 1 mM isopropyl -d-1-thiogalactopyranoside (IPTG) (GoldBio), at 20 C overnight.
Cell pellets were lysed at 4 C using high-pressure homogenization at 27 Kpsi (1 Kpsi = 69 Â 10 3 MPa) (Constant System Ltd, UK) in lysis buffer (500 mM NaCl, 5% glycerol, 5 mM imidazole, 10 mM BME (-mercaptoethanol)   cell homogenate was clarified by centrifugation at 20 000 rev min À1 for 40 min. The clarified supernatant was loaded onto nickel resin equilibrated with lysis buffer. The column was washed with lysis buffer and then with wash buffer (500 mM NaCl, 5% glycerol, 30 mM imidazole, 10 mM BME). Proteins were eluted with elution buffer (500 mM NaCl, 250 mM imidazole, 10 mM BME) and the His tag was then cleaved from MAT2, MAT1 and MATV1 by incubation overnight with Tobacco etch virus (TEV) protease and with thrombin in the case of MATV2. MAT2, MAT1 and MATV2 were then loaded onto an ion-exchange chromatography column (HiTrap Q HP column, GE Healthcare), and MATV1 on a HiTrap S HP, that were pre-equilibrated with buffer A (50 mM NaCl, 5 mM BME); purification was then performed using an isocratic gradient from 0.05 to 1 M NaCl. Selected fractions of MAT2 and MAT1 were then concentrated and loaded onto a HiLoad 16/60 Superdex 200 gel-filtration column (GE Healthcare) and fractions of MATV2 and MATV1 onto a HiLoad 16/60 Superdex 75. Finally, fractions containing pure protein were pooled and stored at À80 C. The buffer used in each purification step was 25 mM HEPES pH 7.5.

Complex formation, crystallization and data collection
In order to assemble the complexes, MAT and MAT were incubated together for 1 h at 4 C in 50 mM HEPES buffer pH 7.5 containing 10 mM MgCl 2 , 50 mM KCl, 300-500 mM NADP. The complex was then loaded onto a Superdex 200 10/300 column and eluted with buffer consisting of 200 mM NaCl, 25 mM HEPES pH 7.5, 1 mM MgCl 2 , 5 mM KCl, 1 mM TCEP [tris(2-carboxyethyl)phosphine]. Crystals appeared at 25 C within 1-2 d in drops containing 2 ml MAT(2) 4 (V2) 2 complex at 5.8 mg ml À1 mixed with 1 ml precipitant solution of 100 mM MES/imidazole buffer pH 6.5, 10% ethylene glycol, 20% PEG 8K. Before crystallization the MAT(2) 4 (V2) 2 complex was incubated with its product SAMe (1 mM), its substrate ATP (1 mM) or AMPPNP (250 mM) and MET (1 mM). These compounds were added to the precipitant and cryosolution. Different data sets were collected at the PROXIMA1, XALOC and I04 beamlines at SOLEIL (St Aubin, France), ALBA (Barcelona, Spain) and Diamond (Oxford, England) synchrotron centres, respectively. Data reduction was carried out with the HKL-2000 (Otwinowski & Minor, 1997) and XDS programs (Kabsch, 2010). The phases were calculated with Phaser (McCoy et al., 2007) using MAT2 (PDB entry 2p02) and MAT (PDB entry 2ydy, Shafqat et al., 2013) as search models for molecular replacement. Model building and refinement were performed using Coot (Emsley & Cowtan, 2004), PHENIX (Adams et al., 2010) and REFMAC (Murshudov et al., 2011). Datacollection and refinement statistics are summarized in Table 4.  SAXS data were collected on the SWING beamline at the SOLEIL synchrotron, using the HPLC-integrated SAXS setup with a two-dimensional AVIEX CCD detector over an angular range q min = 0.01 Å À1 to q max = 0.5 Å À1 . 80 ml MAT(2) 4 (V2) 2 complex at 5 mg ml À1 was loaded onto a pre-equilibrated Shodex KW-402.5-4F 150 kDa SEC (sizeexclusion chromatography) column, 250 frames of SAXS data were taken over the course of protein elution. Data averaging and reduction were carried out with the Foxtrot suite, developed at SOLEIL for the SWING beamline. Further analyses were performed with the ATSAS suite (Petoukhov et al., 2012). In order to generate an ab initio model ten runs of DAMMIN (Svergun, 1999) were performed, and after averaging and filtering a model with 732 beads was produced.

Isothermal titration calorimetry (ITC)
In order to address the association constant (K a ) of MAT complexes, MAT2, MAT1, MATV1, MATV2 and the mutant variants were buffer-exchanged by gel filtration on a Superdex 200 10/300 GL column equilibrated with 200 mM NaCl, 20 mM HEPES pH 7.5 buffer before ITC analysis. Subsequently, MAT isoforms and mutants were injected into MAT2 or MAT1 solution in aliquots of 10 or 20 ml, respectively (Table 1). To verify the interaction of NADP with different MAT isoforms, each protein was buffer-exchanged by gel filtration on a Superdex 200 10/300 GL column equilibrated in 5 mM MgCl 2 , 5 mM KCl, 20 mM HEPES pH 7.5, 200 mM NaCl buffer before ITC analysis. NADP was also diluted in the same buffer. NADP was injected into MATV2 in aliquots of 15 ml and into MATV1 in aliquots of 10 ml ( Table 2). All ITC measurements were carried out at 25 C on a VP-ITC Microcalorimeter (MicroCal/GE Healthcare). The ITC data were processed using Origin software (OriginLab Corp., USA).

Activity assays
MAT activity was addressed by measuring production of SAMe using a fixed concentration of ATP (1 mM) and different concentrations of methionine (5-200 mM). The concentration of MAT2 and MAT complexes was optimized to evaluate the activity at the linear region of the Michaelis curve. The final concentrations of proteins in the reaction were 50, 25 and 12.5 nM for MAT2, MAT(2) 4 (V1) 2 and MAT(2) 4 (V2) 2 , respectively.
For each reaction the protein, methionine and buffer were pre-incubated for 15 min prior to the addition of ATP. The reaction mixtures were thermostated and agitated (37 C, 1400 rev min À1 ). After 10 min the reactions were terminated by the addition of 800 ml of 75% acetonitrile and 1.2% formic acid. To ensure the reaction stopped after the addition of acetonitrile/formic acid solution all samples were shaken (4 C, 1400 rev min À1 ) before centrifugation (14 000 rev min À1 ) and transferred to 96-well plates for UPLC-MS analysis.
Samples were injected in a randomized order for the detection and quantification of SAMe and methionine. Briefly, upon injection, polar metabolites bind to the UPLC column and are then eluted on a polarity gradient. Each fraction was subjected to mass-spectroscopy analysis to determine the production of SAMe and the remaining methionine levels. A ten-point calibration curve with exponentially spaced concentrations of methionine was used for quantization of each sample (van Liempd et al., 2013). The rate of SAMe formation was calculated (pmol s À1 per nmol of complex or pmol s À1 per nmol of MAT2 tetramer) for each substrate concentration. R (R Core Team, 2013) software was used to fit the enzyme kinetic data with the Michaelis-Menten equation for calculation of V max and K m values. Each reaction was performed in triplicate.