Characterization of the Staphylococcus aureus mprF gene, involved in lysinylation of phosphatidylglycerol

Staphylococcus aureus cells have three major phospholipids, phosphatidylglycerol (PG), cardiolipin (CL) and lysylphosphatidylglycerol (LPG) (White & Frerman, 1967). LPG is a basic lipid in which a lysyl group from lysyl-tRNA is transferred to PG (Lennarz et al., 1966, 1967; Nesbitt & Lennarz, 1968). This unusual lipid is identified not only in staphylococci, but also in other clinically important bacteria such as Enterococcus faecalis (Houtsmuller & van Deenen, 1965) and Pseudomonas aeruginosa (Kenward et al., 1979). Recently, it was reported that an S. aureus mprF gene mutant, in which LPG is absent, is sensitive to cationic antimicrobial peptides of the innate immune system (Peschel et al., 2001; Kristian et al., 2003). A decrease in the basic charge on the cell membrane was suggested to lead to a reduced repulsion against cationic peptides and sensitize the mprF mutant cells to these peptides (Peschel, 2002). By protecting against cationic antimicrobial peptide-like bacteriocins, LPG also plays a role in the struggle for existence among soil micro-organisms (Staubitz & Peschel, 2002). Moreover, the mprF mutation also sensitized the S. aureus cells to vancomycin, β-lactams and bacitracin, suggesting that LPG has a role in multi-drug resistance in a series of methicillin-resistant S. aureus strains (Ruzin et al., 2003).

LPG is involved in the regulatory mechanisms of the initiation of DNA replication in S. aureus (Ichihashi et al., 2003). The initiation activity of DnaA protein, the initiator of bacterial DNA replication, is regulated by its adenine-nucleotide binding (Sekimizu et al., 1987; Katayama et al., 1998; Kurokawa et al., 1999; Nishida et al., 2002). The affinity of DnaA protein for adenine nucleotides is affected by acidic phospholipids (Sekimizu & Kornberg, 1988; Castuma et al., 1993; Xia & Dowhan, 1995; Mizushima et al., 1996; Kitchen et al., 1999). LPG inhibits the interaction between DnaA protein and acidic phospholipids, PG and CL, and the S. aureus mprF mutant exhibits an increased amount of replication origins (Ichihashi et al., 2003). These findings suggest that LPG is also involved in cell cycle regulation.

LPG synthetase activity in S. aureus exists in the cell membrane fraction and catalyses the transfer of L-lysine from lysyl-tRNA to PG (Lennarz et al., 1967; Nesbitt & Lennarz, 1968), but the gene encoding the enzyme has not yet been identified. Because the mprF deletion mutant did not synthesize LPG, the mprF gene was suggested to encode the LPG synthetase (Peschel et al., 2001). In the present study, we trans-expressed the S. aureus MprF protein in Escherichia coli cells and obtained biochemical evidence that the mprF gene product encodes LPG synthetase.

Bacterial strains and plasmids.
S. aureus strains RN4220 and CK1001, an insertional deletion mutant of the mprF gene, were described previously (Novick et al., 1993; Ichihashi et al., 2003). E. coli strain JM109 was obtained from Takara Shuzo. E. coli strains W3110, JE5513Tc and YA5513Tc (JE5513Tc pgsA3) were from our laboratory stock (Miyazaki et al., 1985; Tomura et al., 1993). An arabinose-inducible expression plasmid in E. coli, pBAD24 (Guzman et al., 1995), was kindly provided by Dr S. Yasuda (Cloning Vector Collection of the National Institute of Genetics, Japan). The MprF protein-expression vector, pBADmprF, was constructed as follows. The reading frame of the mprF gene was amplified by PCR using an S. aureus RN4220 genomic DNA as a template with PfuTurbo DNA polymerase (Stratagene) and primers 5'-GAATCATGAATCAGGAAGTTAAAAACA-3' and 5'-ACGGGTCGACTCCAAGCGCTTCAGGCATAA-3'. The resultant PCR product was digested with BspHI and SalI, and cloned into pBAD24 at the NcoI and the SalI sites.

Bacterial culture conditions.
S. aureus and E. coli cells were grown at 37 °C in LuriaBertani (LB) medium [1 % tryptone (Difco), 0·5 % yeast extract (Difco) and 1 % NaCl] containing, where appropriate, 50 µg thymine ml^-1, 50 µg ampicillin ml^-1, 5 µg tetracycline ml^-1 or 12·5 µg chloramphenicol ml^-1. When bacterial cells were metabolically labelled with L-[¹⁴C]lysine (300 mCi mmol^-1, Amersham Pharmacia Biotech) or [³²P]orthophosphate (10 mCi ml^-1, Amersham Pharmacia Biotech), LB medium containing 0·5 µCi (18·5 kBq) [¹⁴C]lysine ml^-1 or 10 µCi (0·37 MBq) [³²P]orthophosphate ml^-1 was used. For preparation of the membrane fraction from S. aureus RN4220 or CK1001 cells, cells were grown at 37 °C in 5 l LB medium to an OD₆₀₀ value of 0·9 and then harvested. For preparation of the membrane fraction from E. coli JM109 cells harbouring either pBADmprF or pBAD24, cells were grown at 37 °C in 1 l LB medium to an OD₆₀₀ value of 0·5, then up to 1 % L-(+)arabinose (Tokyo Kasei) was added; the cells were incubated further for 2 h, and then harvested.

Analysis of lipids of bacteria.
For analysis of the E. coli membrane lipid composition, total lipid was extracted from 1 ml of each bacterial cell culture using a modified BlighDyer method as described previously (Sekimizu & Kornberg, 1988). Extracted lipids were separated by TLC developed with chloroform/methanol/acetic acid [65 : 25 : 10 (by vol.)] using a silica gel plate (Silicagel 60, type PK6F, Whatman). To detect lipids, TLC plates were sprayed with 100 mg CuSO₄ ml^-1 containing 8 % phosphoric acid and heated. To detect [¹⁴C]lysine- or [³²P]orthophosphate-labelled lipids, TLC plates were fluorographed using an image analyser (Fuji Film Bas2000).

Preparation of membrane fraction.
Membrane fractions from S. aureus or E. coli were prepared as described previously (Lennarz et al., 1966) with slight modifications. S. aureus cells (wet weight 11·2 g), harvested as described above, were suspended in 2·24 ml 20 mM Tris/HCl (pH 6·8), and incubated with 200 µg lysostaphin (Wako) ml^-1 at 4 °C for 30 min. The resulting lysate was sonicated for 30 s at output 2·0 using a Branson Sonifier 450. Sonicated lysate was centrifuged at 15 000 g for 15 min at 4 °C, followed by centrifugation at 100 000 g for 60 min at 4 °C. The resultant pellet was suspended in 250 µl buffer containing 20 mM Tris/HCl and 1 mM 2-mercaptoethanol. The suspension was recentrifuged at 100 000 g for 60 min at 4 °C; the pellet was suspended in 250 µl buffer containing 20 mM Tris/HCl (pH 6·8) and 1 mM 2-mercaptoethanol, and used as the membrane fraction. To prepare E. coli cell membrane fractions, harvested cells (wet weight 2·8 g) were suspended in 560 µl 20 mM Tris/HCl (pH 6·8), and lysed with 200 µg lysozyme ml^-1 at 4 °C for 30 min, followed by the addition of spermidine hydrochloride to 20 mM. Sonication and centrifugation of samples from E. coli were performed as described above. Protein concentration was determined by the Bradford method (Bradford, 1976) using BSA as a standard.

Preparation of [³H]lysyl-tRNA.
[³H]Lysyl-tRNA was prepared as described by von Ehrenstein (1967) with slight modification. To prepare the tRNA synthetase enzyme fraction, E. coli W3110 cells (wet weight 5·4 g) were suspended in 10·4 ml buffer A [1 mM Tris/HCl (pH 7·2) and 10 mM MgCl₂] and incubated with 200 µg lysozyme ml^-1, followed by the addition of spermidine hydrochloride up to 20 mM. The supernatant was centrifuged at 105 000 g for 1 h. The resultant supernatant was passed through a DE52 (Whatman) column equilibrated with buffer A to remove endogenous tRNAs, and flowthrough fractions were pooled as the tRNA synthetase enzyme fraction. The reaction mixture (0·5 ml) for [³H]lysyl-tRNA synthesis contained 100 mM Tris/HCl (pH 7·2), 10 mM MgCl₂, 10 mM KCl, 1 mM ATP, 0·4 mM [³H]lysine (100 mCi mmol^-1, Amersham Pharmacia Biotech), 200 µg E. coli tRNA ml^-1 (Sigma), and 50 µg tRNA synthetase enzyme fraction. Synthesized [³H]lysyl-tRNA was purified by salt-ethanol precipitation with two washes. To make sure that free [³H]lysine was removed in the [³H]lysyl-tRNA fraction, we simultaneously performed experiments omitting the tRNA synthetase fraction and washed the fractions until the background radioactivities disappeared.

In vitro LPG synthetase activity assay.
This assay was performed as described by Lennarz et al. (1967) with minor modifications. The reaction mixture (50 µl) contained 17·4 mM Tris/maleate (pH 7·0), 94 mM KCl, 7 mM MgCl₂, 0·24 mM L-lysine, 1·55 mM phosphatidylglycerol (Sigma), 30 µM [³H]lysyl-tRNA prepared above, and 040 µg protein from the membrane fraction. The reaction was performed at 30 °C for 10 min and terminated by the addition of 550 µl chloroform/methanol [2 : 1 (v/v)]. The mixture was vigorously vortexed, incubated at 55 °C for 3 min to denature proteins, vortexed again, incubated at 55 °C for 7 min, and filtered through a paper filter (no. 2, Toyo Roshi). The filter was washed with 100 µl chloroform/methanol [2 : 1 (v/v)] and 250 µl 0·9 % NaCl to get reproducible recovery of LPG in the organic phase. The filtrate was mixed, vigorously vortexed for 1 min, placed on ice for 5 min, and centrifuged at 3000 r.p.m. for 5 min at 4 °C. The chloroform layer was washed once with 0·9 % NaCl to extract [³H]lysyl-tRNA that distributed in the chloroform layer independently of enzyme reaction. The amount of radioactivity incorporated into the chloroform layer was measured with a liquid scintillation counter (LS3801, Beckman).

Decrease in the LPG synthetase activity of membrane fractions prepared from the S. aureus mprF deletion mutant
Because LPG was not detectable in the membrane lipid fraction of an S. aureus mprF deletion mutant, the mprF gene is considered indispensable for LPG synthesis (Peschel et al., 2001). The function of the mprF gene product, however, is not clear. Possible functions of the MprF protein other than LPG synthetase would be as an inhibitory factor for LPG degradation or an essential factor for the expression of LPG synthetase. In this study, we hypothesized that MprF protein might function as the LPG synthetase. If this is correct, LPG synthetase activity in the membrane fraction of the mprF deletion mutant should be lost. We constructed an mprF insertional deletion mutant by homologous recombination (Ichihashi et al., 2003), and compared the LPG synthetase activity in the membrane fractions of the mutant and wild-type cells. The mprF deletion mutant did not have a detectable amount of LPG (Ichihashi et al., 2003), consistent with a previous report (Peschel et al., 2001). A membrane fraction prepared from the parent S. aureus strain RN4220 had LPG synthetase activity (Fig. 1). This activity was dependent on PG and lysyl-tRNA (data not shown), consistent with previous reports (Lennarz et al., 1967; Nesbitt & Lennarz, 1968). On the other hand, [³H]lysine incorporation into the organic phase by a membrane fraction from the mprF deletion mutant was not significantly different from the background incorporation (Fig. 1). The results indicate that the mprF deletion mutant lacks LPG synthetase.

(15K): Fig. 1. Loss of in vitro LPG synthetase activity in a membrane fraction from an S. aureus mprF gene deletion mutant. Membrane fractions prepared from S. aureus RN4220 (wild-type, •) or CK1001 (ΔmprF, ) were assayed for LPG synthetase activity in vitro. The indicated amount of membrane fraction (0, 10, 20, 40 µg protein) was incubated at 30 °C for 10 min in the presence of 24 µM [3H]lysyl-tRNA and 1·55 mM PG. Reactions were terminated by the addition of chloroform, and lipids were extracted. The amount of [3H]lysine in the chloroform (organic) phase was determined. Results are mean±SD of three independent experiments. Analyses from different membrane preparations did not differ qualitatively from these.

Induction of LPG synthesis in E. coli cells via expression of S. aureus MprF protein
If the S. aureus MprF protein encodes LPG synthetase, trans-expression of MprF protein in E. coli cells, which do not have either a homologue to the mprF gene or LPG, should induce LPG synthesis in their membranes. To examine this possibility, an open reading frame of the S. aureus mprF gene was inserted into plasmid pBAD24 under the control of the arabinose promoter (Guzman et al., 1995), and E. coli JM109 was transformed by the resultant plasmid, pBADmprF. The cells were cultured in the presence or absence of arabinose, and total lipid was extracted and analysed by TLC. A lipid whose migration was consistent with LPG extracted from S. aureus cells and whose production was stimulated by the addition of arabinose to the medium was identified in total lipid from JM109/pBADmprF cells (Fig. 2a). The lipid stained positive for both the DittmerLester reagent and ninhydrin reagent (data not shown), indicating that it contained phosphates and amino groups. This lipid was not present in the total lipid from JM109 harbouring the vector plasmid (Fig. 2a). To confirm that this lipid was LPG, the E. coli strains were cultured in the presence of [¹⁴C]lysine, and total lipid was extracted and separated by TLC. A lipid labelled with ¹⁴C, whose R_F value corresponded to that of LPG extracted from S. aureus cells, was detected in the E. coli cells harbouring pBADmprF (Fig. 2b). In addition, when membrane lipids were metabolically labelled with [³²P]orthophosphate and analysed by TLC, the lipid in the E. coli cells expressing MprF protein, whose R_F value corresponded to that of LPG extracted from S. aureus cells, was also radiolabelled (Fig. 3). These results indicate that MprF protein expression in E. coli cells induces synthesis of a phospholipid to which lysine was transferred, that is, LPG.

(99K): Fig. 2. Induction of LPG synthesis by expression of the S. aureus MprF protein in E. coli cells. E. coli JM109 cells transformed either with pBAD24 (vector) or with pBAD24 harbouring the S. aureus mprF gene (mprF) under the control of the arabinose promoter were cultured in LB medium without (a) or with [14C]lysine (b) to an OD600 value of 0·1. Arabinose (1 %) was then added and the cultures were further incubated for 2 h. Total lipids were extracted and then separated by TLC. Lipids on TLC plates were visualized by spraying with CuSO4 and heating (a). 14C-labelled lipids were detected using an image analyser (Fujix Bas2000) (b). Total lipid extracted from S. aureus RN4220 cells (Sau) grown in LB medium without (a) or with [14C]lysine (b) was loaded on the far right lanes as a standard for LPG.

(97K): Fig. 3. Requirement of PG for LPG synthesis in E. coli. E. coli JE5513Tc (wild-type) and YA5513Tc (pgsA3) with pBAD24 (vector) or pBADmprF (mprF) were grown in LB medium containing [32P]orthophosphate to an OD600 value of 0·1. Arabinose (1 %) was added to the cells, which were then incubated, harvested, and the total lipids were separated as described in the legend for Fig. 2. 32P-labelled lipids were detected using an image analyser (Fujix Bas2000). Lipids extracted from S. aureus RN4220 cells (Sau) cultured in LB medium containing [32P]orthophosphate were analysed as a standard for LPG.

In vitro LPG synthetase activity requires PG as a substrate (Lennarz et al., 1967; Nesbitt & Lennarz, 1968). We examined whether LPG synthesis via MprF protein expression in cells requires PG as a substrate. When MprF protein was expressed in the E. coli pgsA3 mutant cells, which had a significantly reduced amount of PG in their cell membrane (Miyazaki et al., 1985), LPG synthesis of MprF protein was not induced (Fig. 3). The results are consistent with the notion that MprF protein catalyses LPG synthesis from PG.

When LPG was synthesized via MprF protein expression in E. coli cells, the relative amount of CL increased, and that of phosphatidylethanolamine (PE) decreased (Fig. 2a). One interpretation of the result is that a homeostatic control of the net charge might exist on the cell membrane. The present E. coli strain, in which negatively charged LPG can be induced, will be a useful tool to elucidate novel regulatory mechanisms for phospholipid biosynthesis in bacterial membranes.

Identification of MprF protein and LPG synthetase activity in the membrane fraction of E. coli cells expressing MprF protein
MprF protein is proposed to be a transmembrane protein containing 13 putative transmembrane domains (Peschel et al., 2001). Membrane fractions were prepared from E. coli cells transformed with pBADmprF or pBAD24 vector, and proteins were separated by SDS-polyacrylamide (7·5 %) gel electrophoresis. An approximately 93 kDa protein, whose molecular size was consistent with the predicted mass of the MprF protein, was induced by the addition of arabinose with dependence on plasmid pBADmprF (Fig. 4a). Repeated washing of the membrane fractions increased the content of the 93 kDa protein (data not shown). The results support the prediction that MprF protein is a 93 kDa membrane protein.

(32K): Fig. 4. Co-induction by arabinose of a 93 kDa protein and LPG synthetase activity in the membrane fraction of E. coli cells harbouring pBADmprF. Membrane fractions were prepared from E. coli JM109 harbouring either pBAD24 (vector, , ) or pBADmprF (mprF, •, ), which were grown to an OD600 value of 0·5 and further incubated for 2 h in the presence (, •) or absence (, ) of 1 % arabinose. (a) Analysis of the membrane fractions (40 µg protein) by SDS-polyacrylamide (7·5 %) gel electrophoresis with Coomassie Brilliant Blue R-250 staining. The arrow indicates an approximately 93 kDa protein, which is consistent with the predicted molecular mass of S. aureus MprF protein. The migration positions and sizes (in kDa) of molecular size markers are shown on the left. (b) In vitro LPG synthetase activity of each membrane fraction. Assays were performed as described in the legend for Fig. 1. Results are representative of three independently performed experiments.

We then analysed the in vitro LPG synthetase activity in the membrane fractions prepared above. The membrane fraction from the E. coli cells harbouring pBADmprF with arabinose induction had LPG synthetase activity measured by the transition of [³H]lysine into the organic phase (Fig. 4b). To confirm the in vitro synthesis of LPG, the lipids in the organic phase containing [³H]lysine were separated by TLC. Staining of the plate with rhodamine 6G, which detects phospholipids (Dittmer & Lester, 1964), revealed a spot whose R_F value corresponded to that of LPG from S. aureus cells (data not shown). Incorporation of reasonable amounts of ³H radioactivity into LPG was confirmed by scraping the lipids off the plate and measuring their radioactivity by liquid scintillation counting (data not shown). This LPG synthetase activity in the E. coli membrane fraction depended on both PG and lysyl-tRNA (Fig. 5), consistent with observations in the membrane fraction of S. aureus (Lennarz et al., 1967; Nesbitt & Lennarz, 1968). The K_m values for PG and lysyl-tRNA were 56 µM and 6·9 µM, respectively, corresponding to those from S. aureus (Lennarz et al., 1967; Nesbitt & Lennarz, 1968).

(14K): Fig. 5. Lysyl-tRNA- and phosphatidylglycerol-dependent synthesis of LPG in vitro by the membrane fraction prepared from E. coli JM109 cells expressing the S. aureus MprF protein. In vitro LPG synthetase activity of the membrane fraction (10 µg) prepared from these cells was assayed as described in the legend for Fig. 1 except for the [3H]lysyl-tRNA concentration of 0, 5, 10, 20 or 40 µM in (a) or except for the phosphatidylglycerol concentration of 0, 0·02, 0·04, 0·09, 0·19, 0·39, 0·77 or 1·55 mM in (b). Results are representative of three independently performed experiments.

We examined whether non-denaturing detergents could solubilize functional MprF protein from the membrane fraction. When the membrane fraction from E. coli cells expressing MprF protein was treated with 0·5 % Triton X-100 and centrifuged, a significant amount of the LPG synthetase activity was recovered in the supernatant fraction (data not shown). This result demonstrated that MprF protein could be solubilized by non-denaturing detergent and therefore supported a membrane localization of MprF protein.

Conclusion
The present study examined the LPG synthetase activity of the S. aureus mprF gene product. The membrane fraction prepared from the S. aureus mprF deletion mutant cells had lost LPG synthetase activity (Fig. 1). E. coli cells that expressed the S. aureus MprF protein synthesized LPG (Figs 2 and 3). Co-induction of a 93 kDa protein whose molecular size corresponds to that of MprF protein and LPG synthetase activity was demonstrated in the membrane fraction prepared from the E. coli cells in which MprF protein was expressed (Fig. 4). The enzyme reaction mediated by this E. coli cell membrane fraction was dependent on both PG and lysyl-tRNA, consistent with that from S. aureus cells. K_m values obtained from the E. coli membrane fraction for PG and lysyl-tRNA were consistent with the values of the LPG synthetase activity from S. aureus cells. These results suggest that MprF protein functions as the LPG synthetase in the S. aureus cell membrane.

Although this study strongly suggests that MprF protein is LPG synthetase, we cannot deny the possibility that there is a polypeptide(s) other than the MprF protein that is essential for the enzyme activity. To exclude this possibility, purification and characterization of the LPG synthetase must be performed. LPG is suggested to be involved in resistance against cationic antimicrobial peptides (Peschel et al., 2001; Kristian et al., 2003), cationic antimicrobial peptide-like bacteriocins (Staubitz & Peschel, 2002), or antibiotics (Ruzin et al., 2003), and to have a role in cell cycle regulation (Ichihashi et al., 2003). The cellular content of LPG varies depending on the growth phase and culture conditions (Houtsmuller & van Deenen, 1965; Kenward et al., 1979; Ichihashi et al., 2003). To understand the roles of LPG in various aspects of cell physiology, it is critical to elucidate the mechanism that controls the membrane LPG content. LPG localization in cells is an important factor in the regulation of LPG function. The E. coli strain expressing MprF protein constructed in the present study will serve as a useful tool for further research on this basic phospholipid.

We thank Dr Seiichi Yasuda (Cloning Vector Collection, National Institute of Genetics, Japan) for providing pBAD24. This work was supported in part by Grants-in-Aid for Scientific Research of the Japan Society for the Promotion of Science (JSPS).