The Lysis Protein E of f X174 Is a Specific Inhibitor of the MraY-catalyzed Step in Peptidoglycan Synthesis*

Coliphage f X174 encodes a single lysis protein, E, a 91-amino acid membrane protein. Dominant mutations have been isolated in the host gene mraY that confer E resistance. mraY encodes translocase I, which catalyzes the formation of the first lipid intermediate in bacterial cell wall synthesis, suggesting a model in which E inhibits MraY and promotes cell lysis in a manner analogous to cell wall synthesis inhibitors like penicillin. To test this model biochemically, we monitored the effect of E on cell wall synthesis in vivo and in vitro . We find that expression of Emyc , encoding an epitope-tagged E protein, from a multicopy plasmid inhibits the incorporation of [ 3 H]diaminopimelic acid into cell wall and leads to a profile of labeled precursors consistent with MraY inhibition. Moreover, we find that membranes isolated after Emyc expression are drastically reduced in MraY activity, whereas the activity of Rfe, an enzyme in the same superfamily, was unaffected. We therefore conclude that E is indeed a cell wall synthesis inhibitor and that this inhibition results from a specific block at the MraY-catalyzed step in the pathway. There are at least two distinct mechanisms by which phage promote destruction of the bacterial cell wall and subsequent cell lysis, the choice of which appears to be determined by the phage genome size. Bacteriophages with large genomes encode a holin-endolysin system. In the prototypic l model, the S holin protein accumulates activity assay monitoring the exchange of 3 H]UMP the UMP moiety of UDP-GlcNAc.

There are at least two distinct mechanisms by which phage promote destruction of the bacterial cell wall and subsequent cell lysis, the choice of which appears to be determined by the phage genome size. Bacteriophages with large genomes encode a holin-endolysin system. In the prototypic model, the S holin protein accumulates in the cell membrane and the R endolysin accumulates in the cytoplasm. At a genetically programmed time, S forms a membrane lesion to release the R endolysin into the periplasm where it can degrade the cell wall and cause lysis (1). In contrast, bacteriophages with small genomes can only afford to encode a single lysis protein. Three unrelated single protein lysis systems are known: the E protein from X174 (ssDNA, Microviridae), and the L and A 2 proteins from MS2 (ssRNA, group I) and Q␤ (ssRNA, group III) respectively (1)(2)(3)(4)(5). The molecular mechanism of lysis caused by any of these proteins has been elusive. No cell wall-degrading activity has been associated with any of these phages, indicating that their lytic mechanism is distinct from the holin-endolysin system of the larger phages.
Working from the premise that these phage proteins are targeting a host protein to promote lysis, we took a genetic approach to isolate lysis-resistant host mutants and identify the target genes. We recently isolated dominant mutations in Escherichia coli mraY that result in resistance to the bacteriophage X174 E lysis protein (6). mraY encodes translocase I, which catalyzes the formation of the first lipid intermediate in cell wall synthesis (7) and is a member of the UDP-GlcNAc/ MurNAc 1 :polyisoprenyl-P GlcNAc/MurNAc 1-P transferase family of enzymes, which we will refer to as the GPT enzyme family (8). Inhibition of MraY by the antibiotic mureidomycin results in lysis of Pseudomonas aeruginosa, and depletion of MraY activity from E. coli also results in concomitant cell lysis (9,10). These findings, coupled with the isolation of E-resistant mraY mutants, suggest a model in which E inhibits MraY to elicit lysis by a mechanism similar to that of antibiotics like mureidomycin and penicillin that inhibit cell wall synthesis (see Fig. 1A) (11,12). The bacterial cell wall is composed of polysaccharide chains of the repeating unit N-acetyl-glucosamine-␤-1,4-N-acetyl-muramic acid (GlcNAc␤1,4MurNAc). Attached to the MurNAc sugar is a pentapeptide side chain (in E. coli, L-Ala-␥-D-Glumeso-DAP-D-Ala-D-Ala). Adjacent polysaccharide strands are cross-linked by peptide bonds between the free amino group of meso-diaminopimelic acid (DAP) from one peptide chain and the carboxyl group of the penultimate D-Ala of an adjacent peptide. The pathway for cell wall synthesis can be divided into three phases: (i) cytoplasmic reactions, (ii) membrane reactions, and (iii) periplasmic reactions (see Fig. 2) (13). In the cytoplasm, UDP-GlcNAc is converted to UDP-MurNAc in two steps. In the first and committed step in cell wall biosynthesis, an enolpyruvyl moiety from phosphoenolpyruvate is added to C-3 by MurA. MurB catalyzes the reduction of the enolpyruvyl moiety to a lactyl group using reducing equivalents from NADPH. The pentapeptide is then extended from the lactyl group of UDP-MurNAc, one or two amino acid residues at a time, to generate UDP-MurNAc-pentapeptide. On the cytoplasmic face of the inner membrane, MraY catalyzes the transfer of P-MurNAc-pentapeptide from UMP to the lipid carrier undecaprenol-P, generating undecaprenol-P-P-MurNAc-pentapeptide or lipid I. GlcNAc is then added to lipid I from UDP-GlcNAc by MurG, generating lipid II. Lipid II is flipped, by an as yet unidentified flippase, to expose the disaccharide-pentapeptide monomer unit on the periplasmic face of the membrane. Once exposed to the periplasm, the disaccharide-pentapeptide is polymerized and cross-linked by multienzyme complexes, including the penicillin-binding proteins, to form the mature cell wall (14).
The purpose of this study is to test our model for the mechanism of E-induced lysis (see Fig. 1A) biochemically by monitoring its effect on cell wall synthesis in vivo and in vitro.

EXPERIMENTAL PROCEDURES
Media, Chemicals, and Solvent Systems-Bacterial cultures were grown in Luria-Bertani (LB) broth (15) or minimal M9 media (16) supplemented with 0.4% glucose and L-amino acids (20 g/ml His and Tyr; 30 g/ml Gly, Ile, Phe, Pro, Ser, and Val; 40 g/ml Trp; 50 g/ml Arg; 60 g/ml Ala and Asp; 70 g/ml Asn and Lys; 80 g/ml Glu; 90 g/ml Gln; 100 g/ml Met and Thr; 110 g/ml Leu). Met and Thr were present at 100 g/ml to reduce the size of the intracellular DAP pool (17). All cultures were grown in the presence of 40 g/ml kanamycin (Kan). When indicated, bacterial cultures were induced with isopropyl-␤-D-thiogalactopyranoside (IPTG) (Alexis, San Diego, CA) at a concentration of 1 mM. Bacterial Strains and Plasmids-E. coli K-12 strain ET505 (W3110 lysA::Tn10) was the host strain used in this study and was obtained from the E. coli Genetic Stock Center (New Haven, CT) on the web. The lysA mutation was required to prevent the conversion of added [ 3 H]DAP to Lys, so that [ 3 H]DAP is only incorporated into cell wall and its precursors. The plasmid pEmycZK has been described previously (6) and contains Emyc, encoding E with a C-terminal c-Myc epitope tag, cloned under control of the IPTG-inducible tac promoter (Fig. 1B). The control vector pJFlacZK is isogenic to pEmycZK, except that it does not contain Emyc (Fig. 1B). It was constructed by inserting the lacZ gene in the HindIII site of pJF118EH (18) and conversion to Kan R as described previously for pEmycZK (6). All DNA manipulations were performed according to standard procedures (19).
Cell Wall Synthesis Measurements-ET505 pEmycZK and ET505 pJFlacZK were grown in minimal M9 glucose media in 250-ml culture flasks at 37°C to an A 550 of ϳ0.3 when a portion of each culture was transferred to a small prewarmed 50-ml flask containing sufficient [ 3 H]DAP to give a final activity of 5 Ci/ml. Constant aeration of all cultures was maintained throughout the experiment. After a 10-min prelabeling period, both labeled and unlabeled cultures were induced with IPTG. Culture growth was monitored from the unlabeled culture, and [ 3 H]DAP incorporation into cell wall was monitored in the labeled culture as described previously (17) with minor modifications. Briefly, 0.2-ml aliquots were removed at the indicated times and added directly to 0.8 ml of boiling 5% SDS. The samples were boiled for 1 h and allowed to cool to room temperature before filtering through a 0.22-m mixed cellulose-ester filter (type-GS, Millipore, Bedford, MA). The filters were washed with 30 ml of distilled H 2 O and allowed to dry completely, and the radioactivity associated with the cell wall was determined by counting the filters in a Beckman LS5000TD liquid scintillation counter using EcoscintA liquid scintillation fluid (National Diagnostics, Atlanta, GA). In control experiments, label incorporation into the cell wall was linear 10 min after addition of [ 3 H]DAP, indicating that the precursor pools were in isotopic equilibrium.
Determination of Cell Wall Precursor Synthesis during Emyc Expression-Cultures were grown as described above except that induction with IPTG was done at an A 550 of 0.6. After 2 min, a portion of the culture was added to a prewarmed 50-ml flask containing sufficient [ 3 H]DAP for a total activity of 35 Ci/ml. Constant aeration of all cultures was maintained throughout the experiment. After an 8-min pulse-labeling period, prior to any observable lysis, three 1-ml aliquots of labeled culture were removed and centrifuged for 10 min at 4°C at maximum speed in a microcentrifuge. The cell pellets were washed with 1 ml of ice-cold media and resuspended in 10 l of dH 2 O. The cell suspension was spotted on Whatman 3MM paper, and labeled cell wall precursors were separated by development with solvent system A for ϳ20 h as described (20). Each lane was cut into 1-cm strips and counted as described above. Cell wall, nucleotide, and lipid intermediate fractions ran at published R F values (0, 0.1, and 0.8, respectively).
Purification of UDP-MurNAc Peptides-UDP-MurNAc-pentapeptide and -tripeptide were accumulated in Bacillus subtilis W23 in the presence of vancomycin and chloramphenicol or cycloserine and chloramphenicol, respectively, and purified through a Sephadex G-25 column (1.6 ϫ 72 cm) as described previously (21). Fractions from the G-25 column were assayed for the presence of N-acetyl sugars as described previously (22). N-Acetyl sugar-containing fractions were pooled, lyophilized, and dissolved in a small volume of dH 2 O. The partially purified material was subjected to TLC analysis on Whatman Al-SIL-G plates using solvent systems A, B, and C and yielded a single UV-absorbing spot in each case. The UDP-MurNAc-pentapeptide and tripeptide were further purified on a 1-ml Hi-Trap Source Q anion exchange column (Amersham Pharmacia Biotech, Piscataway, NJ) using a 0 -0.4 M gradient of NH 4 HCO 3 over 20 column volumes and a flow rate of 1 ml/min. For the UDP-MurNAc-pentapeptide preparation, fractions corresponding to the major UV-absorbing peak at 254 nm were pooled, lyophilized, and dissolved in 200 l of dH 2 O. Mass spectrometry at the Texas A&M Laboratory for Biological Mass Spectrometry and quantitative amino acid analysis at the Texas A&M Protein Chemistry Laboratory confirmed that the preparation contained highly purified UDP-MurNAcpentapeptide (m/z ϭ 1194 and a Ala:Glu:DAP ratio of 3:1:1). The UDP-MurNAc-tripeptide preparation was processed as described for the UDP-MurNAc-pentapeptide, and quantitative amino acid analysis confirmed its identity (Ala:Glu:DAP ratio of 1:1:1).
Accumulation of UDP-MurNAc Peptides during Emyc Expression-ET505 pEmycZK and ET505 pJFlacZK were grown in 2 liters of LB-Kan at 37°C to an A 550 of ϳ0.7 and induced with IPTG. Just before induction, MgCl 2 was added to 0.1 M to stabilize cells undergoing lysis (23). After 45 min of induction, the cells were harvested and UDP-Nacetyl sugars were purified and characterized as described above for the UDP-MurNAc-pentapeptide. For the preparation from ET505 pEmy-cZK, fractions corresponding to the major UV-absorbing peak in the anion exchange chromatogram were processed for amino acid analysis as described above. The retention time of the UDP-N-acetyl sugar that accumulated during Emyc expression was compared with those of authentic UDP-MurNAc-tripeptide and -pentapeptide standards using the anion exchange column described above with a 0 -0.6 M gradient of NH 4 HCO 3 over 20 column volumes and a flow rate of 1 ml/min.
Membrane Preparation-200 ml cultures of ET505 pEmycZK and ET505 pJFlacZK were grown in LB-Kan to an A 550 of 0.6 and induced with IPTG. After 9 min of induction, before any observable lysis, the cells were harvested by centrifugation at 9000 ϫ g at 4°C for 15 min. Cell pellets were resuspended in 2 ml of French press buffer (50 mM Tris pH 8.0, 100 mM KCl, 1 mM dithiothreitol) and disrupted in a French pressure cell (SLM Instruments Inc.) at 16000 psi. Membranes were pelleted by centrifugation at 100,000 ϫ g for 60 min at 4°C. Finally, the membrane pellets were resuspended in 300 l of French press buffer and stored in small aliquots at Ϫ80°C. Protein concentrations of the membrane preparations were determined on aliquots solubilized in 1% SDS with the Bio-Rad DC detergent-compatible protein assay kit using Methanol alone did not affect the reaction. Reactions were incubated at 37°C for 20 min and terminated by boiling for 3 min. After boiling, the membrane debris was pelleted by centrifugation at maximum speed for 5 min in a microcentrifuge. The supernatant was removed to a fresh tube, and the pellets were washed with 10 l of distilled H 2 O and centrifuged as before. The wash was pooled with the previous supernatant, and 10 l was spotted on a TLC plate (Whatman AL-SIL-G) and developed with solvent system C. UMP and UDP-MurNAc-pentapeptide (0.02 mol each) were spotted in each lane as standards. The UV-absorbing spots corresponding to UMP (R F ϭ 0.6) and UDP-Mur-NAc-pentapeptide (R F ϭ 0.3) were scraped, and the radioactivity in each spot was determined by liquid scintillation counting as described above and normalized to total protein.
Rfe Exchange Assays-Rfe activity was measured using an exchange assay monitoring the exchange of [ 3 H]UMP for the UMP moiety of UDP-GlcNAc. Rfe reaction compositions were exactly as those for MraY except that 1.5 l of 5 mM UDP-GlcNAc was added in place of UDP-MurNAc-pentapeptide, and 1 l of [ 3 H]UMP (200,000 cpm) was used. The reactions were terminated and processed as described above except that solvent system A was used for TLC. UMP and UDP-GlcNAc (0.02 mol each) were spotted in each lane as standards. The UV-absorbing spots corresponding to UMP (R F ϭ 0.3) and UDP-GlcNAc (R F ϭ 0.2) were scraped, and the radioactivity in each spot was determined by liquid scintillation counting as described above.

E Is a Cell Wall Synthesis Inhibitor-
The isolation of lysisresistant mraY mutants suggested that E is a cell wall synthesis inhibitor that blocks the MraY-catalyzed step in the pathway (Fig. 1A). To test this model, we monitored the effect of E expression on cell wall synthesis in vivo.
The mature cross-linked cell wall is the only cellular material that remains insoluble when cells are boiled in 4% SDS (17). Therefore, a convenient assay for monitoring cell wall synthesis is to measure the incorporation of [ 3 H]DAP, an amino acid unique to the cell wall, into SDS-insoluble material (17). Using this assay we found that [ 3 H]DAP incorporation into the cross-linked cell wall is inhibited 5 min after induction of pE-mycZK well before any detectable effect on growth (Fig. 3).
Cell Wall Precursor Synthesis during Emyc Expression-To determine the step in cell wall synthesis that is blocked by Emyc, we monitored the synthesis of cell wall precursors during Emyc expression from pEmycZK. After induction of Emyc, cells were pulse-labeled with [ 3 H]DAP for 8 min (Fig. 4A), and the major cell wall precursor fractions were separated by paper chromatography into cell wall, nucleotide intermediates (UDP-MurNAc-tripeptide and -pentapeptide), and lipid intermedi-ates (lipids I and II) (20). As shown in Fig. 4B, only the nucleotide intermediates are labeled with [ 3 H]DAP after Emyc induction, whereas all three fractions are labeled in the control strain. Because the ratio of the [ 3 H]DAP-labeled UDP-Mur-NAc-tripeptide and -pentapeptide species in the nucleotide fraction could not be determined in this experiment, the data are consistent with an Emyc-mediated block at either the MurF-or the MraY-catalyzed steps of cell wall synthesis (Fig. 2).
Accumulation of UDP-MurNAc-pentapeptide during Emyc Expression-For a more precise determination of which nucleotide intermediates are synthesized following Emyc induction, we purified the UDP-N-acetyl sugars that accumulate during Emyc expression and compared them to those isolated from the control strain (see "Experimental Procedures"). By N-acetlylsugar assay (22) and A 262 measurements, ϳ5 times more UDP-N-acetyl sugars were isolated after Emyc expression than in the control (1.6 versus 0.3 mol). Anion exchange chromatography of this material revealed one major UV absorbing peak eluting between 25 and 30 ml (Fig. 5A), whereas the control preparation was a complex mixture containing multiple peaks of similar absorbance (Fig. 5B). Using a slightly different elution gradient, authentic UDP-MurNAc-tripeptide and -pentapeptide standards are clearly separable, and the intermediate that accumulates during Emyc expression has an identical retention time to UDP-MurNAc-pentapeptide. However, quantitative amino acid analysis found a 2:1:1 Ala:DAP:Glu ratio for the peak fractions shown in Fig. 5B, which is consistent with an equimolar mixture of the tripeptide and pentapeptide species. The reason for this discrepancy is not known but may be due to contaminants found in the preparation from E. coli that are absent in the standard preparations from B. subtilis. Based on the chromatographic analysis, we conclude that UDP-MurNAcpentapeptide is synthesized and accumulates to high levels after Emyc induction. Therefore, the step of cell wall synthesis inhibited by Emyc must be the one catalyzed by MraY.
MraY Activity Is Dramatically Reduced in Membranes Isolated after Emyc Expression-The in vivo experiments described above indicate that it is the MraY step that is blocked upon Emyc expression. To test this directly, we measured MraY activity in vitro using membranes isolated with and without prior expression of Emyc. MraY catalyzes the reaction, where C 55 -P is undecaprenol-P. Because this reaction is readily reversible (24), addition of UDP-MurNAc-pentapeptide and [ 3 H]UMP to inner membrane vesicles, which contain the lipid substrate in addition to MraY, leads to the formation of [ 3 H]UDP-MurNAc-pentapeptide. In this reaction, the UMP released from the enzyme as a result of the forward reaction is exchanged with [ 3 H]UMP from solution, leading to the formation of [ 3 H]UDP-MurNAc-pentapeptide by the reverse reaction.
We took advantage of this exchange reaction and monitored the formation of [ 3 H]UDP-MurNAc-pentapeptide as a measure of MraY activity. As shown in Fig. 6A, MraY activity is dramatically reduced in membranes isolated after Emyc expression relative to the activity in control membranes. This level of inhibition was similar to that caused by the MraY inhibitor tunicamycin (25).
Rfe Activity Is Normal in Membranes Containing Emyc-Besides MraY, the only other known member of the GPT enzyme family in E. coli is Rfe (8,26). Rfe catalyzes the formation of the first lipid intermediate in enterobacterial common antigen synthesis, undecaprenol-P-P-GlcNAc (26). To determine whether Emyc is a specific inhibitor of MraY like the antibiotic mureidomycin (27) or a general GPT enzyme inhibitor like tunicamycin (25,26), we measured Rfe activity in the membranes described above using an exchange reaction similar to that catalyzed by MraY. Rfe activity was nearly identical in membranes isolated with and without prior Emyc expression (Fig. 6B). As expected, Rfe activity was completely inhibited by the general inhibitor tunicamycin (Fig. 6B).

DISCUSSION
In a previous report we described the isolation of dominant mutants in the host gene mraY that resulted in resistance to E-induced lysis (6). In addition, we found that overexpression of mraY from a medium-copy plasmid could also confer resistance (6). Given these results, we proposed a model in which E inhibits cell wall synthesis at the MraY-catalyzed step in the pathway to promote lysis (Fig. 1A). The results presented in this study provide compelling biochemical support for this model. By monitoring cell wall synthesis in vivo, we observed complete inhibition of [ 3 H]DAP incorporation into cell wall after only 5 min of Emyc expression. We also found that the cytoplasmic nucleotide precursors are the last intermediates labeled with [ 3 H]DAP (Fig. 4) and that UDP-MurNAc-pentapeptide accumulates to high levels during Emyc expression (Fig. 5). Taken together, the in vivo experiments indicate that it is the MraY step in the pathway that is blocked. This blockage was demonstrated unequivocally by the finding that in vitro MraY exchange activity is almost completely inhibited in membranes containing Emyc (Fig. 6A).
There are several antibiotics that block the MraY step, and they can be classified based on their mode of inhibition. They are either direct inhibitors of MraY, like mureidomycin and tunicamycin (25,28), or indirect inhibitors like amphomycin, bacitracin, and colicin M (29 -31), all of which are known to induce bacteriolysis (9,29,(31)(32)(33). The direct inhibitors all resemble the nucleotide substrate with additional hydrophobic moieties attached. They can be further subdivided as specific MraY inhibitors, like mureidomycin (27), or general GPT enzyme inhibitors like tunicamycin (25,26,34). The indirect inhibitors limit the availability of the substrate undecaprenol-P by sequestering it or interfering with its recycling (30,35). In addition to MraY, the general GPT enzyme inhibitors like tunicamycin inhibit the eukaryotic GPT enzyme involved in protein glycosylation, and the indirect inhibitors will bind its lipid substrate, dolichol-P (35,36). Therefore, only specific MraY inhibitors are likely to have potential therapeutic applications in treating bacterial infections.
E is a 91-amino acid membrane protein. Protein fusion analysis has revealed that only the amino-terminal 29 amino acids of the molecule, encompassing its putative transmembrane domain (TMD), are required for lytic function (37,38). 2 With these observations in mind, E likely inhibits MraY in one of two ways, either by direct inhibition through a TMD-TMD interaction or by one of the indirect modes described above. To distinguish between these modes of inhibition, we monitored the effect of Emyc on the activity of the only other known E. coli GPT enzyme, Rfe, which catalyzes the formation of the first lipid intermediate in enterobacterial common antigen biosynthesis, undecaprenol-P-P-GlcNAc (26). We observed no inhibition of Rfe activity in membranes containing Emyc (Fig. 6B). If Emyc were a general or an indirect GPT inhibitor, it should have also inhibited Rfe. We therefore conclude that Emyc is a specific MraY inhibitor. This apparent specificity coupled with the small size of its lethal domain and its selectable lytic phenotype make E an attractive system to probe MraY function and inhibition. It may also help in the development of new specific MraY inhibitors with therapeutic applications. Coprecipitation and inhibition studies using a more purified system are underway to investigate E inhibition of MraY in more detail.
Now that there is both biochemical and genetic evidence that cell wall synthesis inhibition is the mechanism of E-induced lysis, it is a reasonable hypothesis that other single protein lysis systems also act in the same way. We are currently taking similar genetic and biochemical approaches to investigate whether the Q␤ A 2 and MS2 L lysis proteins are also cell wall synthesis inhibitors, and if so, at which step they act.