Screen for Inhibitors of the Coupled Transglycosylase-Transpeptidase of Peptidoglycan Biosynthesis in Escherichia coli

Materials.Wheat germ agglutinin-coated (WGA)-SPA beads (RPNQ0001; polyvinyl toluidene beads) were purchased from Amersham International, Plc., United Kingdom. UDP-[³H]-N-acetylglucosamine was purchased from NEN DuPont. Flavomycin (moenomycin) was a gift from Hoechst, India. Antibiotic medium 3 was from Difco Laboratories. Chromatography materials were from Bio-Rad or Whatman. Other chemicals were from Sigma Chemical Co. Mutants of E. coli AMA1004 for PBP1b (ponB::Spc^r AMA1004) or PBP1a (ΔponA AMA1004) were generated in-house as described earlier (38); the former was grown in LB containing 50 μg/ml spectinomycin. Plasmids pBS96 and pBS98 contained the PBP1b and PBP1a genes, respectively, under the control of their native promoters (38).

Substrates.UDP-N-acetylmuramyl(pentapeptide) was purified from Bacillus cereus 6A1 as described earlier (7, 10). Briefly, cells were grown in antibiotic medium 3 to an A₅₇₈ of 0.7. Chloramphenicol was added to a concentration of 130 μg/ml, and 15 min later, vancomycin was added to a concentration of 5 μg/ml. The cells were harvested 60 min after the addition of vancomycin. A hot-water extract of the cells was purified by gel and ion exchange chromatography. Fractions were monitored by sugar estimation, and the concentration was estimated from its absorbance at 262 nm using a molar extinction coefficient of 10,000 (19).

Enzyme preparation.Membranes from Escherichia coli AMA1004 or other strains were prepared as described earlier (7). Briefly, the cells were grown to an A₆₀₀ of ∼1.8, resuspended in 50 mM Tris-HCl (pH 7.5), 0.1 mM MgCl₂, and lysed using a French pressure cell. The lysate was spun at low speed, and the supernatant obtained from this procedure was ultracentrifuged. The pellet was washed once, resuspended in the same buffer, and stored in aliquots at −70°C. The concentration of protein was estimated by use of Coomassie blue stain (Pierce Chemical Co.).

Peptidoglycan synthesis assay.Peptidoglycan synthesis was performed in membranes of E. coli ponB::Spc^r AMA1004 or the same strain transformed with a plasmid bearing the PBP1b gene [E. coli ponB::Spc^r AMA1004(pBS96)] or the PBP1a gene [E. coli ponB::Spc^r AMA1004(pBS98)] (7). E. coli membranes (4 μg protein) were incubated for 90 min at 37°C with 15 μM UDP-MurNAc(pp) and 2.5 μM UDP-[³H]GlcNAc (0.2 μCi) in a buffer of 50 mM HEPES-ammonia (pH 7.5), 10 mM MgCl₂, 8% dimethyl sulfoxide in a total volume of 25 μl. The reaction was stopped by the addition of 5 μl of 90 mM EDTA; this was followed by capture of the product on 500 μg of WGA-SPA beads in 170 μl of 50 mM HEPES-NaOH, pH 7.5, containing Sarkosyl at a final concentration of 0.2% (in 200 μl). Under these conditions, cross-linked peptidoglycan is specifically monitored (8, 24). In the case of reactions using membranes from E. coli ponB::Spc^r AMA1004, a parallel reaction to monitor the quantity of lipid II formed was carried out by the addition of beads without Sarkosyl (8). The blank used was a reaction without UDP-MurNAc(pp) captured under similar conditions.

Coupled transglycosylase-transpeptidase assay.Membranes (4 μg protein) from a PBP1b-deficient E. coli strain (ponB::Spc^r AMA10004) were preincubated for 120 min at 37°C with 15 μM UDP-MurNAc(pp) and 4.16 μM UDP-[³H]GlcNAc (0.6 μCi) in a buffer of 50 mM HEPES-ammonia (pH 7.5), 10 mM MgCl₂ in a total volume of 15 μl. Formation of lipid II in this step was monitored by the addition of WGA-SPA beads in the absence of detergent (see below) (8). To ensure that lipid II was being monitored, a parallel reaction without UDP-MurNAc(pp) was incubated under identical conditions and captured with the same beads; the difference between the level of the product in this reaction and that in the complete reaction is a measure of the lipid II formed.

Subsequently, the transglycosylase-transpeptidase reaction was carried out by adding 10 μl of a solution containing UDP-GlcNAc (to dilute out the radioactivity and prevent monitoring of lipid II synthesis during the second step) and a source of PBP1b in detergent such that the final concentration was 250 μM UDP-GlcNAc, 8% dimethyl sulfoxide, 0.04% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), and 40 mM NaCl in a final volume of 25 μl. Test compounds, i.e., potential inhibitors of the transglycosylase or transpeptidase, were added at this step. The reaction mixture was incubated for 20 min (unless specified otherwise) at 37°C. The reaction was stopped by the addition of 5 μl of 90 mM EDTA. The product of the reaction was analyzed either by the addition of WGA-SPA beads or by use of paper chromatography. All reactions were carried out in triplicate. For peptidoglycan synthesis reactions, the omission of the first sugar nucleotide, UDP-MurNAc(pp), typically is used as an enzyme blank. However, in this assay the blank used consisted of a solution that had all components except PBP1b; this solution was added at the second step. For both types of analysis, the counts per minute (cpm) obtained in the blank reaction were subtracted from those from reactions containing PBP1b as a measure of the cross-linked peptidoglycan formed by the coupled TG-TP activity of the PBP added in the second step.

For the SPA, the enzyme reaction was performed in flexible plates (1450-401) from Wallac, Finland. The product was captured by the addition of 170 μl of a suspension of WGA-SPA beads (500 μg/well) in N-laurylsarcosine-HEPES buffer such that the final concentration (in 200 μl) was 0.2% N-laurylsarcosine (Sarkosyl) and 50 mM HEPES-ammonia (pH 7.5). Under these conditions, the beads selectively measure radioactive cross-linked peptidoglycan (7, 8). Radioactivity was measured with a Microbeta Trilux instrument 3 h after the addition of the beads. The signal was quite stable, and samples could be counted from 3 to 48 h after bead addition.

For the paper chromatography, 20 μl of the reaction mixture was spotted on Whatman 3MM paper, which was then dried and chromatographed overnight in isobutyric acid:1 M ammonia (5:3 vol/vol). Peptidoglycan stays at the origin, whereas lipid I and lipid II have an R_f of ∼0.9 (5, 10, 24, 29). These regions were cut out, and after the paper had dried, the radioactivity was measured with a liquid scintillation counter (using Optiphase HiSafe 2 scintillation fluid; Wallac, Turku, Finland).

Radioactive N-acetylglucosamine was incorporated into peptidoglycan, and the quantity of peptidoglycan formed in the reaction was measured as cpm of N-acetylglucosamine incorporated into the product. For the SPA, it is difficult to determine the counting efficiency, so all results are represented as cpm. For the paper chromatography analysis, the counting efficiency was low and, for the data shown here, resulted in 350 to 600 cpm per pmol of N-acetylglucosamine.

Graphs were plotted using the Graphpad Prism software; error bars are used to indicate the standard error of the mean, but in some figures these may not be obvious, since they are smaller than the symbols. Where the purity of the starting material is not defined, compound concentrations are expressed in units other than μM.

Source of PBP1b.PBP1b was purified by affinity chromatography from E. coli ΔponA AMA1004(pBS96). Membranes made from this strain (made as described above) were treated with 1% octyl-β-glucoside, 1 M NaCl, 100 mM Tris-HCl (pH 7.5) for 60 min at room temperature. Following centrifugation at 150,000 × g for 45 min, the supernatant was loaded onto ampicillin-Sepharose. The column was washed with 0.5% octyl-β-glucoside, 2 M NaCl in the same buffer, and the PBPs were eluted with 1 M hydroxylamine, 0.5% octyl-β-glucoside, 2 M NaCl, 100 mM Tris-HCl (pH 8.0) at room temperature over 4 h. The eluate was dialyzed against 0.5% octyl-β-glucoside, 2 M NaCl, Tris-HCl (pH 7.5) and stored in aliquots at −70°C. The major PBP purified was PBP1b. Ampicillin was coupled to Affigel 10 by incubation with the gel for 2 h followed by washing with 100 mM Tris-HCl, pH 8.0.

An alternative source of PBP1b was a detergent extract of membranes from the same strain. The membranes were incubated for 60 min on ice with 1% CHAPS, 1 M NaCl, 100 mM Tris-HCl (pH 7.5) at a concentration of 5 mg/ml protein and subsequently centrifuged at 150,000 × g for 15 min. The supernatant was used as a crude source of PBP1b. This was made fresh each time, although it can also be stored in aliquots at −70°C.

Assay of PBP1b or PBP1a activity in transformants.Since the enzymes involved in the late stages of peptidoglycan synthesis are membrane associated (Fig. 1), cross-linked peptidoglycan synthesis can be monitored in E. coli membranes incubated with the sugar precursors UDP-MurNAc(pp) and UDP-[³H]GlcNAc (7). In membranes from a strain of E. coli deficient in PBP1b under the same conditions, lipid II is formed instead of peptidoglycan; lipid II can be captured by the WGA-SPA beads in the absence of the detergent (Table 1) (8, 24). However, in membranes of the PBP1b-deficient strain transformed with a plasmid bearing a copy of the PBP1b gene, cross-linked peptidoglycan was formed by the regenerated pathway (Fig. 1 and Table 1).

Cross-linked peptidoglycan was also formed in a transformant with a copy of the PBP1a gene on a plasmid (Table 1). This was surprising, since the parent strain, ponB::Spc^r AMA1004, failed to produce measurable peptidoglycan despite the presence of an intact PBP1a gene (8). PBP1a enzyme is difficult to assay, and specific conditions may be required (10, 15). This result indicates that although wild-type levels of PBP1a cannot be assayed, levels of enzyme that are increased by cloning the gene on a plasmid may permit it to be detected under conditions similar to those for PBP1b.

The cross-linked peptidoglycan formed in the transformant membranes is a result of the sequential activities of five enzymes: those of MraY, MurG, and lipid pyrophosphorylase and the TG and TP activities of PBP1a or PBP2b. However, a comparison of the effect of inhibitors on lipid II synthesis in the parent PBP1b-deficient strain to that of synthesis of cross-linked peptidoglycan in the PBP transformants can be used to specifically measure the sensitivity of the cloned PBP1b or PBP1a activity to inhibitors. Thus, inhibitors of the MraY, MurG, or lipid pyrophosphorylase should show inhibition of the parental PBP1b-deficient strain as well as of the transformants. But inhibitors of the PBPs, e.g., penicillin, should show inhibition of the transformant strain alone. This could be used to test if an unknown compound is a PBP inhibitor rather than an inhibitor of the earlier enzymes in the pathway, i.e., MraY, MurG, or lipid pyrophosphorylase.

Accordingly, we tested the effect of inhibitors on all three membranes (Fig. 2). As expected, tunicamycin, an inhibitor of MraY, inhibited both the PBP1b-deficient strain and the strain transformed with PBP1a or PBP1b. On the other hand, moenomycin and penicillin, inhibitors of transglycosylase and transpeptidase, respectively, inhibited the transformants only and not the PBP1b-deficient strain. Interestingly, penicillin inhibited peptidoglycan synthesis in the PBP1b transformant with a 50% inhibitory concentration (IC₅₀) of ∼4 ± 1.5 μM but that of the PBP1a transformant with an IC₅₀ approximately 10 times lower (<0.3 μM). This must reflect the different sensitivities of PBP1a and PBP1b to penicillin. Supporting this is the fact that the IC₅₀ of the PBP1b transformant is similar to that of wild-type E. coli, where the major activity is contributed by PBP1b (8, 10).

• Open in new tab
• Download powerpoint

FIG. 2.

Effect of inhibitors on lipid II synthesis in the E. coli AMA1004 ponB::Spc^r strain captured with WGA-SPA (circles, dotted line) versus effect on cross-linked peptidoglycan synthesis in membranes of the same strain transformed with pBS98 (encoding PBP1a; triangles) or pBS96 (encoding PBP1b; squares) or in membranes of wild-type E. coli AMA1004 (diamonds [panel B only]).

This result prompted us to test the effect of cephalexin and cephradine on peptidoglycan synthesis in membranes of the transformants. As expected, these two antibiotics, which bind with higher affinity to PBP1a and PBP3 than to PBP1b (9), showed IC₅₀ values of ∼2.7 ± 0.5 and 7.2 ± 1.2 μM, respectively, on peptidoglycan synthesis in the PBP1a transformant membranes, while having no effect on the PBP1b transformant or on wild-type membranes (8).

The IC₅₀s for moenomycin were similar with the PBP1b and PBP1a transformants (∼150 nM), but these values were more than 10 times higher than the moenomycin IC₅₀ with wild-type E. coli (Table 2) . This difference must be a reflection of the increased copy number of the PBP in membranes of the transformants. For unknown test compounds, this difference could be used as an indicator of specificity of interaction with the PBPs. A summary of IC₅₀s is shown in Table 2.

Transglycosylase-transpeptidase assay feasibility.In the above-described transformant strains, the plasmid-encoded PBPs presumably integrate into the membrane in a fashion similar to that of the native enzyme. If the radioactive lipid II that accumulates in membranes of a PBP1b-deficient strain can be converted to cross-linked peptidoglycan by the addition of an exogenous source of PBP1b, this could be used to develop a direct assay for the transglycosylase-transpeptidase activities of the PBP.

The feasibility of a two-step assay was tested (Fig. 3). In the first step, lipid II is formed by incubation of the PBP1b-deficient membranes (E. coli ponB::Spc^r AMA1004) with sugar precursors; formation of lipid II can be monitored by capture with WGA-SPA beads (24). Following this step, the radiolabel was diluted out and the coupled transglycosylase-transpeptidase assayed by the addition of purified E. coli PBP1b (0.6 μg) to the membranes and incubation for 10 min at 37°C. At the end of this incubation, any cross-linked peptidoglycan formed was captured by the addition of WGA-SPA beads and Sarkosyl.

• Open in new tab
• Download powerpoint

FIG. 3.

Schematic of two-step coupled transglycosylase-transpeptidase assay. In step 1, membranes of the E. coli AMA1004 ponB::Spc^r strain were incubated with UDP-MurNAc(pp) and UDP-[³H]GlcNAc to accumulate lipid II. In step 2, excess UDP-GlcNAc was added along with pure PBP1b (or a solubilized membrane fraction containing PBP1b) to form cross-linked peptidoglycan, which was captured by WGA-SPA beads in the presence of N-laurylsarcosine.

Cross-linked peptidoglycan was indeed formed under these conditions, indicating that the exogenously added PBP1b is able to access the lipid II formed in the membranes (Table 3). Inhibition of this reaction by moenomycin and penicillin, standard inhibitors of the transglycosylase and transpeptidase, respectively, confirmed that the reaction measured was indeed a product of these two enzymatic activities. Nisin, a MurG inhibitor, also inhibited the reaction; however, in the light of data showing that it binds lipid II (36), this is not surprising.

For an assay to be used in high-throughput screening, it is impractical to purify large quantities of PBP1b. Since PBP1b was purified from a detergent-solubilized membrane fraction, we tested whether the same reaction could be catalyzed by a CHAPS-NaCl extract of E. coli membranes overexpressing PBP1b. Indeed, the level of activity obtained was similar to that obtained with pure PBP1b, and it was also inhibited by the reference compounds (Table 3), indicating that a detergent extract of membrane could be used instead of pure PBP1b.

Reactions that were stopped at zero time by EDTA (i.e., before the addition of a source of PBP1b) gave counts similar to reactions that were incubated (step 2) for the same time without a source of PBP1b (data not shown). In addition, if UDP-MurNAc(pp) was left out of the reaction in the preincubation step, no cross-linked peptidoglycan was formed upon the addition of a source of PBP1b (compared to the reaction without PBP1b).

We have shown earlier that in wild-type E. coli membranes, the major transglycosylase-transpeptidase activity comes from PBP1b (8). In this assay, too, a detergent extract of the E. coli strain lacking PBP1b showed no conversion of lipid II to cross-linked peptidoglycan; activity could be detected in detergent extracts of wild-type E. coli membranes, although the quantity was lower than that from strains overexpressing PBP1b (data not shown). Hence, all other experiments were conducted with a detergent extract of E. coli membranes overexpressing PBP1b.

Optimization of transglycosylase-transpeptidase assay for inhibitor screening.Conditions for the coupled TG-TP assay using a detergent extract of membranes as a source of PBP1b were optimized to make it sensitive for the screening of inhibitors. The preincubation step, i.e., formation of radioactive lipid II, can be monitored by paper chromatography or, alternatively, captured by WGA-SPA beads in the absence of N-laurylsarcosine (24). Based on these results, for the first step, membranes were typically incubated at 37°C for 120 min to allow formation of lipid II (data not shown).

The transglycosylase-transpeptidase assay (step 2), i.e., cross-linked peptidoglycan formation, was studied as a function of time and also the quantity of detergent-solubilized membrane extract (Fig. 4). Based on these results, ∼0.45 μg (protein equivalent) of membrane detergent extract was used with an incubation time of 20 min at 37°C for routine assays.

• Open in new tab
• Download powerpoint

FIG. 4.

Coupled transglycosylase-transpeptidase assay. (Top left) Effect of different incubation times at step 2 of the coupled TG-TP SPA. (Top right) Dependence of activity on the quantity of PBP1b-containing detergent extract of membrane in the TG-TP SPA; the quantity of detergent extract was varied while the detergent concentration was kept constant. (Bottom) Paper chromatographic analysis of products of the coupled TG-TP assay. Peptidoglycan, both cross-linked and un-cross-linked, remains at the origin of the chromatogram at an R_f of 0, while lipid II runs at an R_f of ∼0.9. Counts per minute in 1-cm segments are shown starting 1 cm below the origin. For ease of visualizing the data, the paper chromatography analysis of the blank, complete, and inhibited reactions is split into two graphs; the blank and complete reactions are plotted in both graphs for the sake of reference. Bottom right panel symbols: P, penicillin; V, vancomycin; M, moenomycin. The inhibitors used were as follows (concentrations shown in parentheses): tunicamycin (30 μg/ml), nisin (100 μg/ml), vancomycin (1 mM), penicillin (1 mM), and moenomycin (100 nM).

By use of these conditions, the effect of inhibitors was studied (Fig. 5). Moenomycin inhibited the reaction with an IC₅₀ of 10 ± 7 nM, and penicillin did so with an IC₅₀ of 6.6 ± 1 μM as expected; the IC₅₀s are similar to those observed before for the inhibition of the peptidoglycan synthesis pathway (Table 2) (7). Nisin and tunicamycin also inhibited the coupled transglycosylase-transpeptidase reaction with IC₅₀s of 4 ± 3 μg/ml and 3 ± 2.8 μg/ml, respectively, while vancomycin had an IC₅₀ of 48 ± 14 μM.

• Open in new tab
• Download powerpoint

FIG. 5.

Effect of tunicamycin (A), nisin (B), moenomycin (C), vancomycin (D), and penicillin G (E) on the coupled TG-TP SPA. The coupled transglycosylase-transpeptidase SPA was performed as described in Materials and Methods in the presence of different concentrations of inhibitor, and the percent inhibition is plotted as a function of log inhibitor concentration. Where the starting material is not pure, the inhibitor concentrations are expressed in μg/ml.

Bacitracin (10 U) did not inhibit the reaction, nor did other drugs such as sparfloxacin (300 μg/ml) or novobiocin (100 μM), indicating that the observed inhibition by tunicamycin and other antibiotics was specific.

Validation by paper chromatography.While the inhibition by nisin can be explained by its binding to lipid II, the substrate for the transglycosylase (36), the inhibition by tunicamycin was surprising. Tunicamycin typically inhibits MraY, but since it was added at step 2, after the specific activity of UDP-[³H]GlcNAc was reduced ∼100-fold by the addition of excess UDP-GlcNAc, the inhibition observed cannot be due to the inhibition of MraY. To confirm this, the products were analyzed by paper chromatography. The inhibition of MraY should result in no radioactivity in lipid II (7, 27), but if tunicamycin was inhibiting the transglycosylase, radioactive lipid II would be formed but not converted to peptidoglycan. Thus, paper chromatography should show radioactive lipid II in the tunicamycin-inhibited reaction if the TG were inhibited.

The chromatography analysis was also used to confirm the substrate lipid II, the product of the reaction, and inhibition by other compounds. In the paper chromatogram, both cross-linked and un-cross-linked peptidoglycan remain at the origin, whereas lipid II runs near the solvent front (16, 21, 30). Thus, in the complete reaction (the product is cross-linked peptidoglycan) or in a reaction where the transpeptidase is inhibited (the product is un-cross-linked peptidoglycan), radioactivity will be seen at the origin of the chromatogram. If the transglycosylase is inhibited, however, no conversion of lipid II will occur, and this will be reflected in the lack of radioactivity at the origin but counts at the lipid II region.

In the blank reaction (time zero addition of EDTA), the major product was lipid II (R_f, ∼0.9), part of which was converted to peptidoglycan upon incubation for 20 min with a source of PBP1b (the complete reaction is shown in Fig. 4, bottom left).

Both moenomycin and nisin prevented conversion of lipid II to peptidoglycan, giving the same quantity of peptidoglycan (cpm at the origin) as the blank reaction and indicating that both inhibit the transglycosylase. Interestingly, the radioactive lipid II peak observed with nisin is more diffuse than that of the blank reaction or that inhibited by moenomycin. We have observed this in many experiments with nisin; it is probably the result of the binding of nisin to lipid II, which alters its mobility. Tunicamycin showed inhibition of peptidoglycan synthesis, with a chromatography profile similar to that of moenomycin and nisin and with no peptidoglycan at the origin. However, lipid II levels were similar to the zero time reaction, indicating that the transglycosylase was inhibited by tunicamycin and that the inhibition of MraY was not responsible for the inhibition seen in this assay.

As expected, penicillin, which inhibited the TG-TP SPA, shows no reduction in the radioactivity at the origin compared with that for the complete reaction, confirming that it inhibits the transpeptidase and not the transglycosylase. Vancomycin, which inhibited the SPA with an IC₅₀ of ∼48 μM, showed only partial inhibition of peptidoglycan synthesis in the paper chromatographic analysis at 1 mM, suggesting that the inhibition observed with the SPA is a result of the cumulative inhibition of both the transglycosylase and the transpeptidase.