S-Adenosylmethionine-Binding Properties of a Bacterial Phospholipid N-Methyltransferase^,

Abstract

The presence of the membrane lipid phosphatidylcholine (PC) in the bacterial membrane is critically important for many host-microbe interactions. The phospholipid N-methyltransferase PmtA from the plant pathogen Agrobacterium tumefaciens catalyzes the formation of PC by a three-step methylation of phosphatidylethanolamine via monomethylphosphatidylethanolamine and dimethylphosphatidylethanolamine. The methyl group is provided by S-adenosylmethionine (SAM), which is converted to S-adenosylhomocysteine (SAH) during transmethylation. Despite the biological importance of bacterial phospholipid N-methyltransferases, little is known about amino acids critical for binding to SAM or phospholipids and catalysis. Alanine substitutions in the predicted SAM-binding residues E58, G60, G62, and E84 in A. tumefaciens PmtA dramatically reduced SAM-binding and enzyme activity. Homology modeling of PmtA satisfactorily explained the mutational results. The enzyme is predicted to exhibit a consensus topology of the SAM-binding fold consistent with cofactor interaction as seen with most structurally characterized SAM-methyltransferases. Nuclear magnetic resonance (NMR) titration experiments and ¹⁴C-SAM-binding studies revealed binding constants for SAM and SAH in the low micromolar range. Our study provides first insights into structural features and SAM binding of a bacterial phospholipid N-methyltransferase.

INTRODUCTION

Phosphatidylcholine (PC) is the most abundant phospholipid in eukaryotes and the key building block of membrane bilayers. As a major source of lipid second messenger, it plays an important role in signal transduction (15). Although most prokaryotes lack PC, it is present in substantial amounts in membranes of rather diverse bacteria. It has been estimated that about 10% of all bacterial species possess PC (47). Intriguingly, many of these bacteria interact with eukaryotes, and a number of pathogenic and symbiotic PC-containing bacteria, such as Agrobacterium tumefaciens, Bradyrhizobium japonicum, Brucella abortus, and Legionella pneumophila, require PC for an efficient interaction with their respective hosts (2, 11, 12, 28, 33, 52). Two PC biosynthesis pathways operate in prokaryotes. In the PC synthase (Pcs) pathway, choline is directly condensed with CDP-diacylglycerol (CDP-DAG) to form PC in a reaction catalyzed by Pcs. In the methylation pathway, PC is formed by three consecutive methylations of phosphatidylethanolamine (PE) via the intermediates monomethylphosphatidylethanolamine (MMPE) and dimethylphosphatidylethanolamine (DMPE). Depending on the bacterial species, the transmethylation reactions are catalyzed by one or more phospholipid N-methyltransferases (Pmt) using S-adenosylmethionine (SAM) as a methyl donor (18, 28, 47). Many eukaryotes also use a methylation pathway for PC formation (8, 13, 51). However, their enzymes differ from bacterial Pmt enzymes in sequence and structure.

Phospholipid N-methyltransferases belong to the SAM-dependent methyltransferase (SAM-MTase) family. Members of this diverse class of enzymes catalyze the transfer of the methyl group from the ubiquitous cofactor SAM to proteins, nucleic acids, lipids, or small molecules (30, 32, 42). Consistent with these structurally diverse substrates, the substrate-recognition domain in the C terminus of SAM-MTases is highly variable (27, 30). On the other hand, SAM-MTases contain a conserved N-terminal SAM-binding fold comprised of a central seven-stranded β-sheet, flanked by three α-helices on each side. Except for a few key residues, amino acids involved in SAM binding are diverse. So far, about 10 sequence motifs (SAM-I to -X) important for SAM binding and catalysis have been described (27, 29). The only highly conserved residues are the glycine-rich sequence E/D-X-G-X-G-X-G (SAM-I) and an acidic loop between the second β-strand and the following α-helix (SAM-II). Motifs I, II, III, and X are responsible primarily for SAM binding, and motifs IV, VI, and VIII in the active site are involved in catalysis (10, 42).

PC in the plant pathogen A. tumefaciens is synthesized via the Pcs and the methylation pathway. In the latter, all three methylation steps are catalyzed by a single Pmt enzyme called PmtA (24, 52). An A. tumefaciens strain lacking both PC biosynthesis pathways is unable to elicit plant tumors (52). Recently, we produced recombinant PmtA by using Escherichia coli and characterized the enzyme properties in vitro (1, 26). PmtA acts as a monomeric enzyme of 22.3 kDa and is inhibited by the end products PC and S-adenosylhomocysteine (SAH). SAM binding depends strictly on the presence of phospholipid substrate (1). In a step toward understanding the catalytic mechanism of bacterial Pmt enzymes, we measured SAM-binding affinities and constructed a series of PmtA point mutants in residues thought to be critical for SAM binding. A three-dimensional homology model agrees well with the mutagenesis data.

MATERIALS AND METHODS

Materials.

3-sn-Phosphatidyl-ethanolamine was purchased from Sigma Aldrich. S-[methyl-¹⁴C]Adenosyl-l-methionine (1.806 GBq; 48.8 mCi/mmol) was obtained from Hartmann Analytic. HAWP02500 filters for SAM-binding assays were purchased from Millipore, and Hybond-C Extra membranes for protein-lipid overlay assays were from Amersham. The high-performance thin-layer chromatography (HPTLC) silica gel 60 plates were from Merck. All other reagents were of the highest standard commercially available.

Bacterial strains and growth conditions.

Bacterial strains and plasmids used in the present study are listed in Table 1. E. coli cells were grown at 37°C in Luria-Bertani (LB) broth or on LB agar plates supplemented with kanamycin (Km) at a final concentration of 50 μg/ml. E. coli DH5α was used as the host for all cloning procedures. E. coli BL21(DE3) served as the host for overproduction of PmtA variants.

Table 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristic(s)a	Reference or source
Bacterial strains
E. coli DH5α	Cloning host	19
E. coli BL21(DE3)	Expression host	50
Plasmids
pET28b(+)	Plac-T7, Km; high-copy-number His-tag expression vector	Novagen, Darmstadt, Germany
pBO832	wt PmtA in pET28b	1
pBO1222	PmtAE58A in pET28b	This study
pBO1228	PmtAG60A in pET28b	This study
pBO1223	PmtAP61A in pET28b	This study
pBO1229	PmtAG62A in pET28b	This study
pBO1230	PmtAG64A in pET28b	This study
pBO866	PmtAE84A in pET28b	This study
pBO871	PmtAD106A in pET28b	This study
pBO1244	PmtAD106E in pET28b	This study
pBO873	PmtAG162A in pET28b	This study

^aKm, kanamycin.

Site-directed mutagenesis.

To generate PmtA-E58A, -G60A, -P61A, -G62A, -G64A, -E84A, -D106A, -D106E, and -G162A variants, site-directed mutagenesis was conducted using a QuikChange mutagenesis kit (Stratagene) following the supplier's protocol. For the sequence of primers used in this study, see Table S1 in the supplemental material. The vector pBO832 (pET28b containing the wild-type [wt] pmtA gene) (1) was subjected to site-directed mutagenesis. Mutated pmtA variants were verified by sequencing.

Expression and purification of PmtA proteins in E. coli.

E. coli BL21(DE3) carrying the wt or mutated pmtA gene in pET28b was cultivated in LB medium containing kanamycin at 37°C until the optical density at 600 nm (OD₆₀₀) reached a value of between 0.5 and 0.8. Synthesis of PmtA was then induced by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to a final concentration of 0.4 mM, and the cultures were incubated for another 2 h at 30°C. A 1-ml culture of BL21 was harvested by centrifugation for SDS-polyacrylamide gel electrophoresis (SDS-PAGE), and 2-ml cultures were harvested for lipid analysis via thin-layer chromatography (TLC). For SDS-PAGE, cell pellets were resuspended in 1× SDS loading buffer according to the OD₆₀₀ (for an OD₆₀₀ of 1, ∼100 μl of 1× SDS loading buffer) and boiled for 10 min. Ten microliters of each sample was separated on 12.5% SDS-polyacrylamide gels, and the proteins were stained with Coomassie blue.

PmtA wt and mutant proteins were purified as described previously (1).

Analysis of PmtA reaction products via thin-layer chromatography.

The lipid composition of E. coli strains producing PmtA derivatives was determined via TLC. Cells were cultivated as mentioned above, and 2-ml cultures were harvested by centrifugation, washed with 500 μl of water, and resuspended in 100 μl of water. The lipids were extracted according to the method described by Bligh and Dyer (5), separated by one-dimensional (1D) TLC using HPTLC silica gel 60 plates (Merck), and stained with Cu₂SO₄ solution [300 mM copper(II)-sulfate-pentahydrate; 8.5% (vol/vol) phosphoric acid]. As running solvent, n-propanol/propionate/chloroform/water (3:2:2:1) was used.

Circular dichroism (CD) spectroscopy.

The CD spectra of recombinant PmtA proteins were recorded 10 times between 190 and 320 nm with a Jasco 715 spectropolarimeter at 20°C in 50 mM potassium phosphate buffer, pH 8. PmtA thermostability was measured in duplicate with 10 μM enzyme over a temperature range of 5° to 80°C in increments of 5°C. The final spectra obtained were the average results of the 10 scans, normalized against buffer. Analyzes were performed in duplicate using 10 μM enzyme.

The wt PmtA CD spectrum was analyzed using K2D2 (17, 36) and CDSSTR (53) algorithms. The experimentally estimated secondary structure content of PmtA was compared to the predicted secondary structure calculated via GORIV (16) and PHD (39) and to the homology model with the program Stride (20).

Protein lipid overlay assay.

Protein lipid overlay assays were carried out as described previously (1). One microliter of a 14-nmol/μl PE solution or 2 μl of an A. tumefaciens lipid solution in a mixture of chloroform-methanol-water (1:2:0.8) (extracted from a 2-ml A. tumefaciens cell culture according to the method described by Bligh and Dyer; 5) was spotted onto Hybond-C extra membrane strips and air-dried for 1 h at room temperature. Membranes were blocked in blocking buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.1% [vol/vol] Tween 20, 2% [wt/vol] fatty acid-free bovine serum albumin [BSA]) for 1 h at room temperature and incubated with 300 μg of recombinant PmtA protein variants in 5 ml blocking buffer for 2 h at room temperature. Membranes were washed six times for 5 min in TBST buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.1% [vol/vol] Tween 20), and His-tagged proteins bound to the lipids were detected with an anti-Penta-His horseradish peroxidase (HRP)-coupled antibody (Qiagen) and a chemiluminescence (ECL) Western blotting detection system according to the manufacturer's instructions (GE Healthcare).

Radioligand-binding studies.

SAM-binding assays were carried out as described previously (1). Briefly, 10 μM recombinant PmtA protein was incubated with 100 μM phosphatidylethanolamine liposomes (100 nm) and 0 to 200 μM S-[methyl-¹⁴C]adenosyl-l-methionine (48.8 mCi/mmol) in binding buffer (50 mM KH₂PO₄, pH 8.0, adjusted with KOH) for 10 min at 30°C (total assay volume, 50 μl). Binding assay mixtures were passed over HAWP02500 filters on a filtration funnel, and unbound S-[methyl-¹⁴C]adenosyl-l-methionine was removed by washing four times with 300 μl of binding buffer. Bound S-[methyl-¹⁴C]adenosyl-l-methionine was quantified by liquid scintillation spectrometry (LS-6000 TA; Beckman Coulter). The Michaelis-Menten equation was used for calculation of SAM-binding affinities of wt and mutated PmtA enzymes. The data were fitted by nonlinear regression using SigmaPlot 9.0.

NMR spectroscopy.

All spectra were recorded at 298 K and pH 8.0 with a Bruker DRX600 and spectrometer equipped with a pulsed-field gradient and a triple-resonance probe head. Water suppression in experiments recorded for samples in H₂O was achieved by incorporation of a Watergate sequence into the various pulse sequences (6, 14, 45). The binding studies were performed essentially as previously published (4, 35, 37, 44, 48, 49). Briefly, 125 μM PmtA was incubated with 250 μM PE-liposomes and 5% D₂O in binding buffer (50 mM KH₂PO₄, pH 8.0) in a reaction volume of 500 μl for the ¹H titration experiments with SAM and SAH. Next, a concentration of up to 200 μM of either SAM or SAH was titrated to this mixture, and ¹H nuclear magnetic resonance (¹H-NMR) spectra were recorded. For the titration with SAM, the assay was coupled to the activity of 1 μM SAH-nucleosidase (EC 3.2.2.9; G-Biosciences). Conversion of SAH, which is formed during PmtA action, to S-ribosylhomocysteine by the SAH-nucleosidase prevented the accumulation of SAH and its competition with SAM binding (1).

Development of a homology model for PmtA.

The three-dimensional structure of PmtA was predicted by the threading method using the I-TASSER online server (40, 54, 55). Protein structures with Protein Data Bank (PDB) identification codes 3fuxC, 3bkwA, 3futA, 1zg9A, 1qaqA, 1gyrA, 3ggdA, and 3futA were chosen by I-TASSER as the templates in the modeling procedure. This server produced five possible models for PmtA. The first model with the best quality of prediction (C score, −0.79; TM score, 0.61 ± 0.14; and root mean square deviation [RMSD], 7.0 ± 4.14 Å) was used here.

RESULTS

Prediction of key residues for SAM binding.

PmtA catalyzes a series of SAM-dependent methylation reactions to produce PC and SAH. In an attempt to define residues in PmtA with a critical role in SAM binding, a multiple sequence alignment of PmtA with other sinorhizobial Pmt enzymes was generated. The SAM-binding motif I (SAM-I) characterized by an acidic residue and a glycine-rich stretch (D/E-X-G-X-G-X-G) is highly conserved in all Pmt enzymes and easily discernible (Fig. 1). Regardless of the substrate, this motif is characteristic of all SAM-dependent methyltransferases and is critical for SAM binding (23, 27, 47). In contrast, an unambiguous assignment of other SAM-binding motifs is not plausible without structural guidance due to the sequence heterogeneity in these regions. On the basis of the secondary structure prediction of PmtA (I-TASSER) (40, 54), we were able to predict two additional sequence motifs (SAM-II and -III) (Fig. 1) in A. tumefaciens PmtA. Motif II includes the predicted β-strand 2 and the adjacent turn. An acidic residue, E84 in A. tumefaciens PmtA, is common at the C terminus of this strand (27). Motif III is characterized by an acidic residue close to the C terminus of β-strand 3 (27). The conserved D106 in PmtA satisfies the motif III criteria.

Fig. 1.

Comparative alignment of the amino acid sequence of A. tumefaciens PmtA and other bacterial phospholipid N-methyltransferases using ClustalW. Highly conserved similar residues are highlighted in black, ≥80% conserved similar residues in dark gray, and ≥60% conserved similar residues in light gray. The secondary structure of the A. tumefaciens PmtA model generated by the I-TASSER server (40) is depicted above the primary sequence. Predicted SAM motifs are indicated. Arrows point out residues that were exchanged with alanine or glutamic acid. The asterisks show mutated PmtA variants which were insoluble. Atum_PmtA, A. tumefaciens PmtA (GenBank accession number NP_10552); Bjap_PmtA, Bradyrhizobium japonicum PmtA (NP_767321.1); Bjap_PmtX3, Bradyrhizobium japonicum PmtX3, (NP_774806.1); Mloti_PmtA, Mesorhizobium loti PmtA (Mll4753); ML_mlr5374, Mesorhizobium loti open reading frame (ORF) (NP_106049). Smel_PmtA, Sinorhizobium meliloti PmtA (AAG10237).

Effect of alanine substitutions of putative SAM-motif residues on PmtA activity.

In order to investigate the role of the putative SAM-binding residues, we generated seven PmtA derivatives with single alanine substitutions in the predicted SAM motifs described above (Fig. 1). As a control, we constructed a PmtA mutant containing an alanine substitution outside the predicted SAM motifs (G162A). The effect of these mutations on PmtA activity was analyzed in E. coli BL21(DE3). All proteins were heterologously expressed, and successful production was confirmed by SDS-PAGE (Fig. 2 A). The G64A and D106A variants were completely insoluble and thus inactive (data not shown). A less severe mutation of D106 to D106E still rendered the protein insoluble (data not shown), suggesting that this residue is critical for proper protein folding.

Fig. 2.

Expression and activity of point-mutated PmtA derivatives after expression in E. coli. (A) Detection of PmtA synthesis in E. coli crude extracts via SDS-PAGE. (B) Lipid formation after expression of agrobacterial PmtA derivatives in E. coli BL21(DE3). Lipids of BL21(DE3) derivatives were extracted and separated by one-dimensional TLC. Phospholipids were visualized with Cu₂SO₄ solution. L, BenchMark protein ladder; V, empty vector (pET28b); WT, wild-type PmtA; PG, phosphatidylglycerol; CL, cardiolipin; PE, phosphatidylethanolamine; MMPE, monomethylphosphatidylethanolamine; DMPE, dimethylphosphatidylethanolamine; PC, phosphatidylcholine.

PmtA activity was assayed via TLC of membrane lipids. E. coli is unable to produce PC or any other methylated phospholipids (Fig. 2B, lane V). As reported previously (26), MMPE, DMPE, and PC were synthesized when wt PmtA was expressed (Fig. 2B, lane WT). PmtA-P61A and -G162A were as active as wt PmtA, suggesting that these amino acids are not critical for enzyme activity. The G60A and G62A variants produced only traces of MMPE and no DMPE and PC. The E58A and E84A variants were completely inactive, indicating an essential role of these amino acids for PmtA activity.

PmtA variants are properly folded.

In case soluble proteins were obtained, the mutant proteins were purified according to the procedures previously established for the wt enzyme (1). Their structural integrity was assessed using CD spectroscopy. The spectrum of wt PmtA showed a minimum at around 210 nm, suggestive of an extensive helical content (Fig. 3 A). Overall, the CD profiles of PmtA-E58A, -G60A, -P61A, -G62A, -E84A, and -G162A were similar to that of the wt protein, indicating that the mutated PmtA proteins were structurally intact.

Fig. 3.

(A) Circular dichroism analysis of the conformational state of wt and mutated forms of recombinant PmtA proteins. The spectrum of each protein with a concentration of 10 μM is plotted as ellipticity in millidegrees (mdeg) versus wavelength. (B) Midpoints of the thermal unfolding transition (T_m) were used to determine the conformational stability of wt PmtA. (C) Determined T_m values of mutated PmtA variants.

As a second line of evidence for normally folded proteins, we used CD spectroscopy to compare the thermal unfolding of the variants. wt PmtA exhibited a thermal unfolding transition midpoint (T_m) of 35.9°C (Fig. 3B), and the six mutated proteins were comparable, with wt PmtA having a melting point around 35°C (Fig. 3C).

The programs CDSSTR and K2D2 were used to calculate the secondary structure content of wt PmtA from the CD spectrum. The estimated α-helical (39 to 42%) and β-sheet (∼11%) content agree well with both the predicted secondary structure of the PmtA amino acid sequence and the homology model structure (Table 2).

Table 2.

Prediction and experimental estimation of PmtA secondary structure

Algorithma	Secondary structure content (%)
	α-Helices	β-Sheets	Loops and coils
GOR IV	38.07	12.69	49.24
PHD	38.58	17.77	43.65
Stride	37.06	10.15	52.79
CDSSTR	42.00	11.00	47.00
K2D2	39.37	11.52	49.11

^aGOR IV and PHD provide secondary structure predictions based on protein sequence. Stride assigns secondary structures to known structures or models. CDSSTR and K2D2 allow deconvolutions of protein CD spectra for experimental estimations of secondary structure content.

Predicted SAM-binding residues are required for SAM binding.

To analyze whether the functional defect of the putative SAM-binding motif mutants in vivo (Fig. 2B) was due to a deficiency in SAM binding, we tested their SAM-binding capacity by filter binding assays using ¹⁴C-SAM. Consistent with almost wt-like activities in E. coli (Fig. 2B), PmtA-P61A and -G162A bound SAM efficiently in vitro (Fig. 4A). The poor enzyme activity of the G62A and G60A variants (Fig. 2B) was reflected by residual SAM binding (Fig. 4A). The inactive E84A enzyme was completely unable to bind SAM. These results suggest a critical role of amino acids E58, G60, G62, and E84 in SAM binding by the methyltransferase.

Fig. 4.

SAM-binding and lipid-binding ability of wt and mutated PmtA derivatives. (A) SAM-binding activity was analyzed with 10 μM recombinant PmtA and 100 μM S-[methyl-¹⁴C]adenosyl-l-methionine (48.8 mCi/mmol) in the presence of 100 μM PE. One hundred percent S-[methyl-¹⁴C]adenosyl-l-methionine corresponds to 143,757 dpm. (B) Lipid binding of SAM-binding-defective PmtA proteins. 6His-tagged PmtA variants (4 nmol) were incubated with nitrocellulose strips displaying equal amounts of A. tumefaciens lipids (upper panel) or 14 nmol of PE (lower panel). Bound protein was detected with anti-Penta-His HRP-coupled antibody (Qiagen).

Since PmtA binds SAM only in the presence of one of its phospholipid substrates (1), we needed to exclude the possibility that the observed defects in SAM binding were caused by impaired lipid binding. Commercially available PE or total lipids extracted from A. tumefaciens were spotted on a nitrocellulose membrane. Purified His-tagged PmtA variants were incubated with the lipid-displaying membrane, and bound proteins were detected with antisera raised against the His tag. All PmtA derivatives were able to bind both PE and Agrobacterium total lipids. Consistent with previous findings (1), PmtA binding to the Agrobacterium lipid mixture was stronger than that to PE alone (Fig. 4B).

Determination of binding constants of PmtA derivatives for SAM by ¹⁴C-SAM titration experiments.

To determine equilibrium SAM-binding constants (K_D) for PmtA and its derivatives, we performed ¹⁴C-SAM titration experiments in the presence of PE as the substrate (Fig. 5). wt PmtA bound SAM with an affinity in the low micromolar range (∼25 μM) (Fig. 5A). Consistent with their in vivo and in vitro activities, PmtA-P61A and -G162A had a similar K_D (∼22 and ∼30 μM, respectively) (Fig. 5B). Since the E58A, G60A, and G62A variants bound only marginal amounts of SAM, a K_D calculation for these PmtA variants was not possible (Fig. 5C). As mentioned above, the E84A variant was completely incapable of binding SAM (Fig. 4 and 5C).

Fig. 5.

Plots of SAM-binding data for wt and mutated PmtA proteins. A reaction volume of 50 μl contained 10 μM PmtA, 100 μM PE-liposomes, and 0 to 200 μM ¹⁴C-SAM. Changes in bound ¹⁴C-SAM (dpm) were plotted against SAM concentration (μM). All data sets were fitted to the equation for one-site binding (see Materials and Methods) by nonlinear regression using SigmaPlot 9.0.

Determination of SAM- and SAH-binding constants for PmtA using ¹H-NMR titration experiments.

A 1D ¹H-NMR spectrum of 125 μM wt PmtA of A. tumefaciens in the presence of 250 μM PE-liposomes in a 50 mM potassium phosphate buffer at pH 8.0 (including 5% D₂O) was recorded at 298 K and 600 MHz in order to check the structural integrity of wt PmtA (Fig. 6A). The dispersion of the proton resonances in the aliphatic and amide region clearly shows that the protein adopts a folded conformation.

Fig. 6.

1D ¹H-NMR spectroscopy of A. tumefaciens wt PmtA. (A) The whole 1D ¹H-NMR spectrum of PmtA. The sample contained 500 μl of a 125 μM PmtA and 250 μM PE-liposomes in 50 mM potassium phosphate buffer (pH 8) and 5% D₂O, recorded at 298 K and 600 MHz. (B) NMR titration experiments with SAM/SAH. Changes in the high-field region of the 1D ¹H-NMR spectrum of PmtA with increasing SAM (left) or SAH (right) concentrations.

The NMR spectroscopy provided another option for determining the affinity constants of PmtA for SAM and its inhibitor SAH (1). Since SAM binding is impossible in the absence of lipid substrate (1), the NMR titration experiments were performed in the presence of PE. SAH nucleosidase was added to SAM titration experiments to prevent accumulation of SAH and its competition with SAM binding (see Materials and Methods). The addition of increasing amounts of SAM to PmtA resulted in a decrease in the peak intensities to between 0.2 ppm and 0.1 ppm (Fig. 6B, left). With increasing amounts of SAH titrated to the protein sample, a new peak appeared between −0.6 ppm and −0.5 ppm (Fig. 6B, right). The changes of the peak intensities or area were plotted against the SAM/SAH concentration, and the binding constants were determined as previously published (21, 22, 25). We were able to estimate a K_D in the micromolar range for SAM (∼91 μM) (Fig. 7A) and SAH (∼25 μM) (Fig. 7B).

Fig. 7.

SAM- and SAH-binding affinities (K_D) of PmtA determined via 1D ¹H-NMR titration experiments. Plots of SAM-binding (A) and SAH-binding (B) data. Changes in the relative chemical shift area/intensity were plotted against the ratio of ligand to protein concentration ([L]/[P]). The K_D for SAM and SAH were determined according to Kannt et al. (25) and Herrmann et al. (22), respectively.

DISCUSSION

In a number of symbiotic or pathogenic bacteria, including the plant pathogen A. tumefaciens, PC biosynthesis is essential for virulence (2, 11, 12, 33, 52). Despite the importance of PC for bacterial fitness, biofilm formation, and host-microbe interaction (26, 52), a detailed knowledge of the structural features of bacterial PC biosynthesis enzymes is lacking. Here, we studied SAM binding of a PmtA enzymes that catalyzes all three methylation reactions from PE to PC (1). Using recombinant protein, we determined binding affinities in the low micromolar range by two independent assays. The ¹⁴C-SAM-binding assays revealed a K_D of 25 μM. This is almost identical to the value determined for the PE N-methyltransferase from rat liver (∼30 μM) (38, 43). It is within the range of between 0.1 and 30 μM that has previously been reported for diverse SAM-dependent MTases (3, 31, 41). The K_D calculated from NMR titration experiments with PmtA was 3- to 4-fold higher (∼91 μM). This is mostly likely due to a residual SAH product that had not been converted by the SAH nucleosidase present in the assay. The determined K_D for SAH (∼25 μM) supports the observation that it is able to compete with SAM for the same binding site (1).

Site-directed point mutations of PmtA provided first insights into the mechanism of SAM binding. The conserved topology of the SAM-binding fold in SAM-MTases suggested three putative SAM-binding motifs (SAM-I, -II, and -III) in A. tumefaciens PmtA (Fig. 1). Amino acids E58, G60, P61, G62, G64 (motif I), E84 (motif II), and D106 (motif III) were predicted to form the SAM-binding pocket and were therefore mutated to alanine.

The E58A, G60A, G62A, and E84A variants were hardly able to convert PE to PC. Their inactivity did not result from failure to fold but from impaired SAM binding. The hypothesis that these amino acids form the SAM-binding pocket is supported by a homology model of PmtA. All four residues are in close proximity in the three-dimensional structure (Fig. 8A) and may interact directly with SAM (Fig. 8B). The typical SAM-MTase fold (30) consists of a central β-sheet sandwiched by α-helices (Fig. 8A). The central β-sheet is composed of five parallel β-strands and an anti-parallel β-strand with a topological organization of (6↓5↑4↑1↑2↑3). Three of the helices (αE, αF, αG) on one side of the sheet and three helices (αB, αC, αD) on the other side create a α/β/α secondary superstructure. The N-terminal helix (αA) is positioned at the top of the β-sheet.

Fig. 8.

PmtA homology model. (A) The PmtA model, shown as a ribbon diagram, was generated using the online server I-TASSER (40, 54) and visualized by PyMOL (http://www.pymol.org). The fold is characterized by a central β-sheet which is formed by five parallel β-strands (5, 4, 1, 2, and 3), preceded by the antiparallel β-strand 6. Experimentally proven SAM-binding residues are shown in stick representation and are highlighted. (B) Closer view into the SAM-binding pocket in complex with SAM. The SAM-binding cleft is made up of the C-terminal loop regions following β-strands 1 to 4. SAM is shown in stick representation in yellow. (C) Closer view of the putative ligand-binding site in the A. tumefaciens PmtA model showing the proposed interactions of the residues with SAM.

The glycine-rich motif I in PmtA (₅₈ELGPGTG₆₄) forms a tight loop between β-strand 1 and the following helix (Fig. 8B). In most MTases, this so-called G-loop was shown to be crucial in positioning the adenine ring of SAM in its correct conformation to ensure close contact with the main chain of the protein framework (10). At least one of the conserved glycines in the “GxGxG” loop (motif I) was shown to contact the carboxypropyl moiety or amino group of SAM (9). Thus, G60 and/or G62 in PmtA might allow for hydrogen bonding of the main chain atoms to the carboxyl and amino group of SAM (Fig. 8C). In analogy to various MTases of known structure and function (7, 27), we speculate that E58 of motif I coordinates SAM through a water molecule. SAM-MTases are postulated to perform a nucleophilic catalysis or SN2 reaction (10, 27, 34, 56). The water molecule coordinated by the acidic residues either could serve as a nucleophile or could aid the displacement of the bond between the sulfonium ion and the methyl group (27).

Motif II containing the crucial acidic amino acid E84 is found at the end of strand 2 (Fig. 8B). E84 is predicted to contact the ribosyl hydroxyl groups and/or adenine base of SAM (Fig. 8C) as in other SAM-Mtases (7, 10, 27).

Although the classic SAM-binding motifs (I to III) are missing in eukaryotic Pmt enzymes, two partial consensus SAM-binding motifs were identified recently in the phospholipid N-methyltransferase PEMT from mice. As with A. tumefaciens PmtA, two G and two E residues (G98, G100, E180, and E181) were found to be essential for SAM binding (46).

Based on the common architecture of SAM-MTases (29) and the order of the SAM-binding motifs (Fig. 1) we suggest that PmtA is sequentially organized into SAM binding, catalytic, and substrate-binding regions from the N to C terminus. This is supported by our results, which show that the SAM-binding pocket is located at the N-terminal region of PmtA. The C terminus may confer substrate selectivity, as shown for most other SAM-MTases (27, 30). Although the PmtA model presented here should be regarded as preliminary, it provides a useful basis for the rational design of further mutations, in particular those aimed at the identification of lipid-binding residues. Moreover, the well-refined NMR spectra presented in this study provide a promising foundation for more detailed insights into the structural characteristics of a bacterial phospholipid N-methyltransferase.