|Home | About | Journals | Submit | Contact Us | Français|
Both Gram-positive and Gram-negative bacteria contain bactoprenol-dependent biosynthetic pathways expressing non-essential cell surface polysaccharides that function as virulence factors. Although these polymers are not required for bacterial viability in vitro, genes in many of the biosynthetic pathways are conditionally essential: they cannot be deleted except in strains incapable of initiating polymer synthesis. We report a cell-based, pathway-specific strategy to screen for small molecule inhibitors of conditionally essential enzymes. The screen identifies molecules that prevent the growth of a wildtype bacterial strain but do not affect the growth of a mutant strain incapable of initiating polymer synthesis. We have applied this approach to discover inhibitors of wall teichoic acid (WTA) biosynthesis in Staphylococcus aureus. WTAs are anionic cell surface polysaccharides required for host colonization that have been suggested as targets for new antimicrobials. We have identified a small molecule, 7-chloro-N,N-diethyl-3-(phenylsulfonyl)-[1,2,3]triazolo[1,5-a]quinolin-5-amine (1835F03), that inhibits the growth of a panel of S. aureus strains (MIC = 1–3 μg mL−1), including clinical methicillin-resistant S. aureus (MRSA) isolates. Using a combination of biochemistry and genetics, we have identified the molecular target as TarG, the transmembrane component of the ABC transporter that exports WTAs to the cell surface. We also show that preventing the completion of WTA biosynthesis once it has been initiated triggers growth arrest. The discovery of 1835F03 validates our chemical genetics strategy for identifying inhibitors of conditionally essential enzymes, and the strategy should be applicable to many other bactoprenol-dependent biosynthetic pathways in the pursuit of novel antibacterials and probes of bacterial stress response.
Antibiotic-resistant bacterial infections pose a major threat to public health (1). In recent years, increasing effort has been devoted to exploring unconventional antibacterial targets, including virulence factors (2). Virulence factors are bacterially expressed molecules that are non-essential for survival in vitro but are required for robust infection in a host. Small molecule inhibitors have been discovered for many virulence factors, including the type III secretion machinery, two-component signaling systems, quorum sensing, and cholera toxin (3, 4). A number of studies suggest that antivirulence factor inhibitors hold promise for limiting bacterial infections (5–7).
The cell envelope of many pathogenic bacteria contains a large and diverse group of polymers that function as virulence factors in that they are non-essential in vitro but required for pathogenicity. Although many of these polymers are potential targets for new antibacterials, the discovery of inhibitors that block their synthesis has been hampered by difficulties in reconstituting the enzymatic machinery, obtaining lipid-linked substrates, and developing high-throughput assays. A whole cell assay would circumvent some of these challenges provided it could be designed to report specifically on inhibition of the targeted polysaccharide pathway.
Many of these polysaccharide biosynthetic pathways share a common feature that suggests a strategy for inhibitor discovery: they contain conditionally essential genes. That is, the initiating genes, the products of which are responsible for the first steps of polymer synthesis, are dispensable since the polymers are non-essential in vitro, but the downstream genes are essential unless polymer initiation is prevented. Stably maintained gene deletions of conditionally essential enzymes are thus generally accompanied by suppressor mutations that prevent committed flux into these biosynthetic pathways (8–13). The conditional essentiality of genes in non-essential, bactoprenol-dependent biosynthetic pathways was first observed in 1969 for O-antigen biosynthesis in Salmonella enterica (8), and has since been extended to capsular polysaccharide biosynthesis in Streptococcus pneumoniae (9), exopolysaccharide biosynthesis in the plant pathogen Xanthomonas campestris (10), O-antigen biosynthesis in Escherichia coli and Pseudomonas aeruginosa (11, 12), as well as wall teichoic acid (WTA) biosynthesis in S. aureus (13). As with peptidoglycan biosynthesis, these polysaccharides are all assembled intracellularly on a bactoprenol carrier lipid before transfer to the final acceptor on the surface of the cell (14). It has been suggested that blocking flux once polymer synthesis has been initiated is deleterious either because toxic bactoprenol-linked polysaccharide intermediates accumulate or because bactoprenol-linked peptidoglycan precursors are depleted. The mixed gene dispensability pattern in many bactoprenol-dependent biosynthetic pathways can potentially be exploited in a high-throughput screen to discover compounds that inhibit growth of wildtype bacteria but not of strains defective in polymer initiation.
We demonstrate the utility of this screening strategy by discovering an inhibitor of wall teichoic acid (WTA) biosynthesis in the clinically relevant pathogen Staphylococcus aureus. Methicillin-resistant S. aureus (MRSA) infections have become a major problem both in hospitals and in the community (15). Further, high level vancomycin resistance in MRSA has begun to appear, creating a clear demand for antimicrobials against new targets. The WTA biosynthetic pathway is a possible target since S. aureus strains lacking WTAs are unable to colonize host tissue and exhibit a greatly diminished capacity to establish infections in animal models (16–18). No specific inhibitors of WTA biosynthesis have been previously identified, although a compound that prevents D-alanylation, a tailoring modification of both WTAs and a related family of anionic cell surface polymers (lipoteichoic acids), has been shown to affect bacterial fitness (19).
WTAs from S. aureus are disaccharide-linked anionic polymers of poly(ribitol-phosphate) that are covalently attached to peptidoglycan (Figure 1). They constitute a large percentage of the total cell wall mass (20). The primary S. aureus WTA is assembled on a bactoprenol carrier embedded in the cytoplasmic membrane by the sequential addition of two sugar residues (by TarO and TarA), two to three glycerol 3-phosphate units (by TarB and TarF), and then finally the poly(ribitol-phosphate) repeat (TarL) (Figure 2) (21, 22). WTAs are then exported through the membrane by an ABC (ATP Binding Cassette) transporter complex (TarGH) and the polymer is transferred from the bactoprenol carrier to peptidoglycan by an unidentified transferase (20).
As found in several other bactoprenol-dependent biosynthetic pathways, the first gene in the WTA pathway (tarO) can be deleted since the pathway is non-essential (13, 16). Similarly, the second gene, tarA, can also be deleted since the initiating enzyme (TarO) is readily reversible, as are other phosphosugar transferases (23). However, the downstream genes (tarBDFGHIJL) can only be deleted in a strain that cannot initiate polymer synthesis (Figure 2) (13). Therefore, we devised a high-throughput screen that involved comparing the growth of a wildtype S. aureus strain and its isogenic ΔtarO strain following addition of a small molecule. Compounds that inhibit a conditionally essential enzyme in the WTA pathway were expected to prevent the growth of the WTA-expressing wildtype strain, but not of the ΔtarO mutant, which cannot initiate polymer synthesis (see Table 1). Using this screen, we have discovered a WTA biosynthesis inhibitor, 7-chloro-N,N-diethyl-3-(phenylsulfonyl)-[1,2,3]triazolo[1,5-a]quinolin-5-amine (1835F03), that prevents the growth of both methicillin-sensitive and methicillin-resistant S. aureus. We have identified the target of this molecule as the transmembrane component (TarG) of the ABC transporter that exports WTAs to the cell surface. This is the first small molecule antibiotic that specifically targets WTA biosynthesis.
A library of 55,000 small molecules was screened in duplicate at a concentration of 38 μM against a wildtype S. aureus strain (RN4220) and the corresponding ΔtarO strain using optical density to monitor growth. We obtained 45 initial hits and confirmed three lead compounds that inhibited the growth of the wildtype strain but had no activity against the ΔtarO strain (Table 1). The most potent compound, 1835F03, had a minimum inhibitory concentration (MIC) of 1.3 μg mL−1 (3 μM) against the wildtype screening strain. The in vitro MIC values against all tested S. aureus strains, including five MRSA strains, were also in the low μg mL−1 range (Figure 3 and Supplementary Figure 1) but the compound had no activity against other Gram-positive strains that contain WTAs (MIC > 40 μg mL−1), including Streptococcus pneumoniae, Bacillus subtilis 168, and B. subtilis W23. Compound 1835F03 was taken forward for further characterization.
We first used the natural product tunicamycin to verify that growth inhibition by 1835F03 requires flux into the WTA pathway. Although toxic at high concentrations due to inhibition of MraY, an essential peptidoglycan biosynthetic enzyme (24), tunicamycin selectively inhibits TarO at low concentrations (25). Therefore, it can be used to block WTA expression at concentrations that have no effect on bacterial growth (Figure 4, panel a and Supplementary Figure 2). We reasoned that if 1835F03 inhibits a conditionally essential WTA pathway enzyme, then tunicamycin should antagonize its antibiotic activity by preventing flux into the biosynthetic pathway. The growth of S. aureus RN4220 as well as several clinical isolates was monitored in a two-dimensional matrix of different concentrations of tunicamycin and 1835F03 (26). Tunicamycin provided complete protection against 1835F03 in all S. aureus strains examined (Figure 4, panel b and Supplementary Figure 3). To confirm the pharmacological results, we constructed a strain of S. aureus that expresses tarO under the control of an isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible promoter (22). In the presence of IPTG-induced WTA biosynthesis, 1835F03 inhibited cell growth; in its absence, 1835F03 had no effect (Figure 4, panel c). These pharmacological and genetic results are consistent in showing that the antibiotic activity of 1835F03 requires flux into the WTA biosynthetic pathway; therefore, they imply that the compound inhibits one of the conditionally essential WTA enzymes.
We next tested 1835F03 against conditionally essential S. aureus WTA enzymes for which we have previously established in vitro biochemical assays – namely, TarBDFI, and L (21). None of these enzymes was inhibited by 1835F03 at concentrations well above the MIC (100 μM, see Supplementary Figures 4–8). These results narrowed the possible targets of 1835F03 to the two-component ABC transporter that exports WTAs to the bacterial cell surface (TarGH) or to the unidentified ligase that couples exported WTAs to peptidoglycan. Since the transporter has not been reconstituted in vitro, we used a genetic approach to query whether 1835F03 inhibits TarGH.
Overexpression of TarGH in S. aureus caused a reproducible increase in the MIC (Figure 5), consistent with this two-component transporter being the molecular target. To confirm this finding, the B. subtilis 168 WTA transporter complex TagGH was used for heterologous complementation. B. subtilis 168 TagG shares only 58% identity with S. aureus TarG and the transporter has been shown to have promiscuous substrate specificity, exporting poly(glycerol-phosphate) WTAs as well as other structurally distinct polymers (27, 28). We reasoned that if the resistance of B. subtilis 168 to 1835F03 is due to an intrinsically resistant ABC transporter (Figure 3, > 40 μg mL−1), then the functional expression of this transporter in S. aureus should likewise confer resistance. Indeed, expression of B. subtilis 168 TagGH in S. aureus conferred complete resistance to 1835F03 (Figure 5). These overexpression and heterologous complementation results imply that the two-component transporter TarGH is the molecular target of 1835F03.
To probe whether 1835F03 interacts with the transmembrane component (TarG) or the ATPase component (TarH) of the ABC transporter, we selected resistant mutants on agar plates (at a frequency of 1 in 106 cells) and identified a set that remained resistant after propagation in the absence of the compound. Since the first two genes in the pathway are non-essential and removing them confers resistance to 1835F03, it seemed likely that the resistant mutant population would consist of two types: null mutations in tarO or tarA that prevent initiation of WTA synthesis and mutations in the molecular target of 1835F03. Since S. aureus bacteriophage use WTAs as receptors (29), we were able to rapidly group the mutants into two classes based on their susceptibility to phage infection (Figure 6, panel a). We selected three phage-resistant and two phage-sensitive mutants for further analysis. As expected, the phage-resistant mutants did not contain extractable WTAs (Figure 6, panel b) (22). Targeted sequencing of the tar genes from these mutants showed that all three phage-resistant strains contained partial gene deletions or frameshift mutations in either tarO or tarA. In contrast, the two phage-sensitive mutants contained extractable WTAs. Targeted sequencing of the 1835F03-resistant but phage-sensitive (WTA+) mutants showed that each contained a unique point mutation in tarG (Supplementary Figure 9). The mutant tarG alleles encode different amino acid changes (TarG:F82L and TarG:W73C) on the same face of a predicted membrane spanning helix. The changes apparently impart resistance to 1835F03 while maintaining functional WTA ABC-transporter activity. To verify that the observed mutations in TarG confer resistance, we exchanged the wildtype allele with each of the mutant tarG alleles (30). The mutant strains were fully resistant, showing that the point mutations alone are sufficient to bestow 1835F03 resistance in an otherwise wildtype background (Figure 6, panel c) and implicating TarG as the target. In the presence of 1835F03, the TarG mutant strains produced WTAs identical to those produced in the wild-type strain (data not shown). These results confirm that the mutations are within the target rather than simply suppressor mutations that allow for the transport of truncated WTAs (which would occur if a conditionally essential enzyme upstream of TarG were inhibited).
We next sought to assess the antibacterial properties of 1835F03. Kill-curve analysis (Figure 7) shows that its activity is fully bacteriostatic: treated S. aureus cells immediately cease to divide but the number of viable bacteria remains constant even at high multiples of the MIC or prolonged exposure (Supplementary Figure 10). Furthermore, if 1835F03 is removed or if tunicamycin is added to block initiation of nascent WTA polymers, bacterial growth resumes after a lag (Figure 7). The prolonged but reversible stasis caused by 1835F03 is consistent with growth arrest, a bacterial response in which cell division and most other metabolic processes cease as the organism seeks to survive stressful conditions (31). Since preventing completion of WTA synthesis once it has initiated is expected to lead to an accumulation of bactoprenol-linked WTA precursors, and since bactoprenol levels are limited in bacterial membranes (32), blocking WTA synthesis would indirectly lead to a decrease in other bactoprenol-linked metabolites, including peptidoglycan precursors. Decreased pool levels of peptidoglycan precursors may elicit the observed growth stasis response. Consistent with this, growth stasis is also observed upon treatment of S. aureus with bacitracin (Supplementary Figure 11), a peptide natural product that sequesters bactoprenol in a trimolecular zinc-bactoprenol-pyrophosphate complex, which prevents its recycling (33). Likewise, depletion of mevalonate (a precursor of bactoprenol) biosynthetic genes in S. aureus triggers downregulation of primary metabolism transcripts (34), and limiting amounts of mevalonate pools in Streptococcus pneumoniae results in growth stasis (35). It is not known whether bacteria can sense levels of peptidoglycan precursors directly or sense abrupt decreases in rates of peptidoglycan synthesis in comparison with other metabolic processes.
Stasis may alternatively be a result of a programmed stress response to the intracellular accumulation of bactoprenol-linked WTA intermediates. There is evidence that the accumulation of bactoprenol-linked saccharides is deleterious in other organisms (8–12). Furthermore, Silhavy and coworkers have shown that the accumulation of bactoprenol-enterobacterial common antigen intermediates in E. coli activates two well-characterized, two-component signaling pathways involved in the response to envelope stress (36). Other bacteria may also contain signaling systems that respond to envelope stress caused by inhibition of bactoprenol-dependent metabolic pathways. Consistent with the proposal that inhibiting WTA biosynthesis elicits a stress response, Brown and coworkers have reported that depletion of TarB in Bacillus subtilis leads to increased transcription from the promoter PywaC, which drives expression of a putative GTP pyrophosphokinase (37). GTP pyrophosphokinases produce a signaling molecule, (p)ppGpp, that is linked to the stringent response (38). We are currently using 1835F03 as a chemical probe to elucidate the molecular mechanisms involved in the growth stasis of S. aureus in response to WTA inhibition.
We have described a general strategy to discover small molecules that inhibit conditionally essential enzymes in non-essential metabolic pathways. The strategy has been applied to WTA biosynthesis but can be adapted to the apparently large number of other pathways that show a similar mixed dispensability pattern. Some of these pathways are potential antibacterial or antivirulence targets and small molecule inhibitors could therefore have therapeutic utility. More generally, such inhibitors may serve as probes to investigate the cellular effects of blocking particular pathways and may provide insight into signaling pathways involved in sensing and responding to metabolic stresses.
We have demonstrated the utility of our discovery strategy by identifying a small molecule that targets a conditionally essential enzyme involved in WTA biosynthesis in S. aureus. The molecular target of the compound is the ABC transporter that exports WTAs from the cytoplasm to the cell surface. We have shown that the molecule, 1835F03, inhibits the growth of S. aureus, including MRSA, providing the first pharmacological validation of the WTA pathway as an antibacterial target. Since the pathway is non-essential, the resistance frequency in vitro due to null mutations in one of the first two genes is relatively high, but these mutants are not expected to contribute to resistance in vivo. Therefore, inhibitors of conditionally essential WTA enzymes may have efficacy in vivo and we are investigating this possibility.
Plasmids were constructed in Escherichia coli Novablue (Novagen) cells, and introduced into the restriction negative S. aureus strain RN4220 by electroporation (39). Allelic exchange was performed as previously described by Bae et al. (30) and Meredith et al. (22). All genetic manipulations were verified either by restriction digest or DNA sequencing. S. aureus was grown in tryptic soy broth, and antibiotic markers were selected with erythromycin (Em; 10 μg mL−1), tetracycline (Tc; 2.5 μg mL−1), and chloramphenicol (Cm; single copy integrated into genome 5 μg mL−1, plasmid 10 μg mL−1). Bacterial strains used are listed in Supplementary Table 1. All cloning materials are listed in Supplementary Table 2.
Although each strain was marked with a unique fluorescence marker to provide an alternative readout to bacterial growth, optical density proved to be a more robust measure of growth and was used for hit determination. For more thorough details on the fluorescence-based screen, please refer to the Supplementary Information. On the night before screening, a single colony of each strain was grown in appropriate media. On the day of screening, 384 well plates (Corning® 3710) were filled with 40 μL of media. Small molecules (10 mM in DMSO) were transferred from library plates to these culture plates using a 300 nL pin transfer. Final culture volume was 80 μL and the final concentration of compound in each plate was 38 μM. For each compound plate there were four daughter plates, wildtype and ΔtarO were done in duplicate in separate plates. Following transfer of the small molecules, each plate was inoculated with 40 μL of media containing equal number of cells. After inoculation, each plate was covered (Corning® 3009), stacked five plates high and incubated at 30 °C for 18 h. For each plate, positive control wells (Em at 10 μg mL−1) and negative control wells (media only) were included in columns 23 and 24.
The following day, the plates were read for OD600 using a Perkin Elmer Envision plate reader. Data were collected and analyzed to identify compounds that killed the wildtype strain but not the ΔtarO mutant. Data work-up included normalizing the OD data to the positive and negative controls to determine a normalized percent survival for each plate. A hit was defined as greater than 50% survival for the ΔtarO mutant and less than 10% survival for the wildtype strain based on the OD data.
For a secondary screen, hits from the primary screen were evaluated against wildtype and the ΔtarO mutant as described above, but the compound was investigated using a four point dose-response curve starting at a concentration of 50 μM and doing two-fold serial dilutions. Three inhibitors were validated. One such small molecule was 1835F03 (see Supplementary Information for compound characterization data).
1835F03-resistant mutants were selected both in liquid culture and on solid media. Mutants were selected on solid media by plating ~5 × 106 CFUs on TSB agar plates containing 1835F03 at 4× MIC. Colonies that were fully grown after 48h were passaged four times in the absence of 1835F03, evaluated for 1835F03 sensitivity and stable mutants were analyzed for the production of WTAs using phage infection and WTA extraction as previously described (22).
Mutants prepared in liquid culture were first colony-purified on TSB agar plates. Individual colonies were grown in culture and passaged four times in the absence of 1835F03. The culture was then colony purified, evaluated for 1835F03 sensitivity and analyzed for the production of WTAs using phage infection and WTA extraction as previously described (22).
S. aureus TarB, D, F, I, and L were cloned, overexpressed and purified as described previously (21). Radiometric in vitro assays were carried out as previously described (21) to test the inhibitory effects of 1835F03 on Tar B, D, F, and L. See Supplementary Information for exact reaction conditions. Briefly, enzymes were tested at 200 nM and 1835F03 was added to the reactions at 5, 50, or 100 μM. In each case, a reaction with no 1835F03 served as a negative control (active enzymatic reaction) and a reaction with heat-treated enzyme served as a positive control (no enzymatic reaction). 1835F03 was incubated with all reaction components except the substrates at room temperature. After 10 minutes, substrates were added and the reactions were allowed to proceed for 60 minutes. The reactions were quenched with DMF and the products were imaged as previously described (21). See Supplementary Figures 4–8 for results.
Supplementary Figures 1-11
Supplementary Tables 1 and 2
The authors would like to thank the ICCB/NERCE facilities for help with the small molecule screen and MIC determinations. High-throughput screening capability was provided by the National Screening Laboratory for the Regional Centers of Excellence in Biodefense and Emerging Infectious Diseases (NIAID U54 AI057159). We would also like to thank Dr. Brian Kraybill, Dr. Ben Gross, Dr. Kerry Love, Dr. Yanqiu Yuan and Yihui Wu for helpful discussions. This research was supported by the NIH (GM078477 to S.W., F3178727 to J.G.S., and EY017381/EY008289 to M.S.G), a training grant to T.C.M (T32-AI07061-30), a Mary Fieser Postdoctoral Fellowship to J.C. and a National Science Foundation Fellowship for S.B. We thank the following individuals for generously providing materials: Olaf Schneewind (Univ. of Chicago; pKOR1, RN4220, RN4220 ΔtarO), Jean C. Lee (Harvard Medical School; pLI50), Richard P. Novick (NYU; pCN33, pCN59, pCN68), David Rudner (Harvard Medical School; pDR110), Chia Lee (Univ. of Arkansas; pCL25, pCL55), and Francis Arhin (Targanta Therapeutics, Quebec, Canada; RN450, ø11).