|Home | About | Journals | Submit | Contact Us | Français|
Whole-cell screening of Mycobacterium tuberculosis (Mtb) remains a mainstay of drug discovery but subsequent target elucidation often proves difficult. Conditional mutants that under-express essential genes have been used to identify compounds with known mechanism of action by target-based whole-cell screening (TB-WCS). Here, the feasibility of TB-WCS in Mtb was assessed by generating mutants that conditionally express pantothenate synthetase (panC), diaminopimelate decarboxylase (lysA) and isocitrate lyase (icl1). The essentiality of panC and lysA, and conditional essentiality of icl1 for growth on fatty acids, was confirmed. Depletion of PanC and Icl1 rendered the mutants hypersensitive to target-specific inhibitors. Stable reporter strains were generated for use in high-throughput screening, and their utility demonstrated by identifying compounds that display greater potency against a PanC-depleted strain. These findings illustrate the power of TB-WCS as a tool for tuberculosis drug discovery.
Tuberculosis (TB) remains the leading cause of death attributable to a bacterial pathogen globally (Glaziou, et al., 2009). Although TB is treatable using the current arsenal of anti-tubercular drugs (Zhang, 2005), the therapeutic regimen is complex and lengthy (Mitnick et al., 2009) and associated with poor patient adherence. This, in turn, has promoted the selection of Mycobacterium tuberculosis (Mtb) strains that are resistant to one or more first-line drugs compounding the already daunting challenge of global TB control. The case for developing new and more effective drugs that have novel modes of action and can shorten the duration of treatment is thus compelling (Koul et al., 2011).
Existing anti-tubercular drugs act on a limited repertoire of molecular targets that are essential primarily for cell wall biosynthesis, replication, transcription or translation, and for which resistance mechanisms have already emerged (Zhang and Yew, 2009). As a consequence, the identification of compounds that inhibit new biological targets and pathways is an important goal. Genome-wide essentiality screens have yielded many potential anti-mycobacterial targets (Lamichhane, 2011; Mdluli and Spigelman, 2006); however, identifying small molecules that inhibit such targets within the cellular environment remains a challenge. On the one hand, inhibitors identified in target-based biochemical screens frequently fail to exhibit activity when tested against whole cells as a result of poor cell permeability, efflux, metabolic redundancy in the pathway of interest, and/or relative invulnerability of the target to inhibition (i.e. the target is not rate-limiting for the pathway) (Payne et al., 2007). On the other hand, traditional whole-cell screens, which test the activity of a library of compounds against live bacterial cells, are limited by the inherent insensitivity of Mtb conferred by the cell envelope and the need subsequently to determine the mode-of-action of inhibitors, which can be difficult and time-consuming (Payne et al., 2007). To overcome some of these limitations, target-based whole-cell screening (TB-WCS) has been developed to identify biologically active inhibitors of prioritized drug targets (DeVito et al., 2002; Forsyth et al., 2002; Young et al., 2006). This methodology makes use of conditional mutant strains in which the expression of a target gene is regulated using either anti-sense interference or inducible promoters. Since reducing the levels of a target should result in a commensurate increase in the sensitivity of the strain to inhibitors that act on or biochemically downstream of that target, the assay provides a simple means to identify inhibitors that are both biologically active and target- or pathway-specific. The utility of this approach in antimicrobial drug discovery is exemplified by the identification of platensimycin for use against Gram-positive bacteria (Wang et al., 2006).
Previous studies have demonstrated that Mtb strains harboring lesions in the panC, lysA and icl1 loci are impaired in their ability to infect mice (McKinney et al., 2000; Pavelka et al., 2003; Sambandamurthy et al., 2002). The essentiality of these genes for the growth and survival of Mtb in vivo (Hingley-Wilson et al., 2003), together with the lack of human homologs, make them potentially attractive targets for anti-tubercular drug discovery. In this study, we used a conditional expression system based on the tetracycline (Tet)-regulatable promoter element to generate mutants that conditionally express panC, lysA and icl1. We demonstrated the utility of these strains in confirming the essentiality of these targets for the growth of Mtb in vitro and showed that conditional silencing enhanced the sensitivity of Mtb to inhibitors of these targets. Furthermore, we identified compounds with greater potency against PanC-depleted than wild type (wt) Mtb cells in a proof-of-concept high-throughput screen (HTS) of a small compound library. Together, our findings illustrate the utility of conditional mutants as chemical genetic tools to confirm the mode-of-action of new compounds and to identify new inhibitors of mycobacterial growth in a target-based whole-cell format.
To generate conditional mutant strains of Mtb, we employed a two-step method in which the native promoter of the target gene was first replaced with a Tet-regulatable promoter element by single crossover (SCO) homologous recombination, and then plasmids expressing Tet repressors (TetRs) were introduced (Ehrt et al., 2005; Guo et al., 2007) (Figure S1). The suicide plasmids, pPanC-SCO, pLysA-SCO and pIcl1-SCO were introduced into wild type (wt) Mtb H37Rv, and the corresponding SCO recombinants were identified and genotypically confirmed (Figure S2). In the absence of TetR, a SCO strain is dependent on the activity of the Tet-regulated promoter for expression of its target gene. We observed no significant growth defects in any of the SCO strains under standard growth conditions. Since the extent of transcriptional silencing required to confer a growth phenotype was not known for either gene, we employed three different TetR-expressing vectors which vary in their mode and/or ability to repress expression from the Tet-promoter. The L5-based integration vectors, pMC1s and pMC2m, express wt TetR (wt-TetR) from strong (S) and intermediate (M)-strength mycobacterial promoters (Guo et al., 2007), producing conditional mutants in the Tet-ONS and Tet-ONM configurations, respectively. The higher level of tetR expression from pMC1s is expected to confer more stringent repression of the Tet-promoter than in cells harboring pMC2m. In contrast, pTEK-4S-OX is an episomal plasmid that expresses a reverse TetR (rev-TetR) from a strong promoter (Guo et al., 2007). The activity of rev-TetR is lower than wt-TetR in mycobacteria (Gandotra et al., 2007; Guo et al., 2007) resulting in less stringent repression and higher levels of induced gene expression in the Tet-OFF configuration. Another issue to be considered when generating promoter-replacement mutants is the chromosomal arrangement of the gene under investigation. Placement of the Tet-promoter upstream of a gene present in an operon will also affect expression of co-regulated genes downstream of the target. This can complicate attempts to correlate the phenotypes arising from gene silencing with specific alterations in the levels of the gene of interest. To circumvent potential off-target effects arising from promoter-replacement in the panC and lysA strains, we therefore generated complementation vectors to express the downstream genes in trans from the Tweety attB site (Pham et al., 2007). These were introduced into the corresponding SCO strains together with the TetR-expressing vector.
Pantothenate synthetase (PanC) catalyzes the final step in the biosynthesis of pantothenate (vitamin B5), an essential precursor of coenzyme A and acyl carrier protein (Webb et al., 2004). panC is considered essential in Mtb (Sassetti et al., 2003), and therefore, is predicted to play an important role in the growth and survival of Mtb in vivo, particularly during growth on lipid substrates. The importance of de novo pantothenate biosynthesis in Mtb was confirmed previously in studies showing that an Mtb pantothenate auxotroph lacking panC and panD was attenuated in mice (Sambandamurthy et al., 2002). We generated conditional panC mutants by introducing each TetR vector into panC-SCO, together with pPanC-Comp, and selecting on solid media supplemented with pantothenate. Conditional mutants harboring both the TetR and rev-TetR-expressing plasmids were obtained. Growth of the strains on agar supplemented with anhydrotetracycline (ATc) inducer was attenuated under repressed conditions in both Tet-ON and Tet-OFF configurations (0 and 200 ng/ml ATc, respectively) (Figure 1A). Consistent with the relative stringency of repression conferred by the TetR-expressing plasmids, the growth impairment was most severe in panC Tet-ONS, and only partially relieved upon addition of ATc at 200 ng/ml. In contrast, growth of panC Tet-ONM and panC Tet-OFF was highly titratable at ATc concentrations up to 200 ng/ml. Importantly, the growth defect of all strains under repressed conditions was rescued by the inclusion of pantothenate supplement in the culture medium, confirming that the observed phenotype resulted from a defect in pantothenate biosynthesis. In liquid culture, growth of panC Tet-ONM and panC Tet-OFF was severely inhibited by panC silencing (Figure 1B, C, D, E). As observed on agar, the growth rate of these strains was dependent on the concentration of ATc. In contrast, growth of panC Tet-ONS correlated poorly with ATc concentration in liquid culture (Figure S3A, B) and, paradoxically, appeared to be more robust at lower concentrations of inducer. Subsequent analysis revealed that this was due to the propensity of the panC Tet-ONS strain to lose Tet-dependent regulation during growth in liquid culture, particularly at growth-limiting concentrations of ATc, presumably by the acquisition of suppressor mutations that alleviate repression. However, such instability was not readily apparent for the panC Tet-ONM and panC Tet-OFF strains. The level of PanC protein produced in the wt, panC-SCO, panC Tet-ONM and panC Tet-OFF strains was then assessed by Western blotting. In these experiments, pantothenate was included in the culture medium to ensure equivalent growth of the strains. An immuno-reactive band corresponding to the predicted size of the PanC polypeptide (~33 kDa) was identified in cellular lysates from wt Mtb and panC-SCO when probed with antibodies raised against purified, recombinant PanC (Figure 2A). A corresponding band was identified in lysates from panC Tet-OFF and panC Tet-ONM grown under inducing conditions (0 vs. 200 ng/ml ATc, respectively). The level of PanC protein under these conditions was comparable to that observed in the wt and panC-SCO strains. However, following panC silencing, PanC levels dropped to <5% of wt in these mutants, confirming that the strains behave in a predictable and target-specific manner in response to inducer treatment (Figure 2A, B).
LysA catalyzes the final step in the lysine biosynthetic pathway, converting meso-2,6-diaminoheptanedioate to L-lysine and is essential for the growth of Mtb in vitro in the absence of lysine (Pavelka and Jacobs, 1999). An Mtb lysine auxotroph is severely attenuated in mice, confirming that the de novo biosynthesis of this amino acid is essential for growth in vivo (Pavelka et al., 2003). To generate lysA conditional mutants, the TetR-expressing plasmids were introduced into a lysA-SCO together with the complementation vector, pLysA-Comp, and transformants were selected under inducing conditions on 7H11 media supplemented with 0.05% Tween 80 (Pavelka and Jacobs, 1999). Viable lysA Tet-ONM and lysA Tet-OFF strains were obtained and genotypically confirmed (Figure S1). In contrast, we could not isolate a Tet-ONS derivative, suggesting that in this case, the level of lysA expression was below the threshold required to support bacillary replication. Growth of the conditional mutants was examined on solid media, with or without ATc. In accordance with the known essentiality of lysA, the lysA Tet-ONM strain grew very poorly in the absence of inducer (Figure 3A). The defect was, however, ameliorated by induction of lysA expression at the lowest concentration of ATc tested (1.5 ng/ml), with only a modest increase in growth observed at higher concentrations (Figure 3A). The growth defect of this strain under repressing conditions was also reversed by inclusion of lysine in the growth medium, confirming that it was attributable to disrupted lysine biosynthesis. Similarly, growth of the lysA Tet-OFF strain was not significantly affected by ATc-induced gene silencing, with only minor inhibition observed at the highest level of inducer (Figure 3A). Although the levels of lysA expression were not determined, the lack of ATc-responsiveness in this strain most likely reflects an inability of rev-TetR to regulate gene expression tightly, and rescue of growth by leaky expression of lysA. In liquid culture, growth of lysA Tet-ONM was repressed by omission of either inducer or lysine (Figure 3B) but ameliorated at higher concentrations of ATc or with lysine supplementation. Interestingly, this strain displayed greater ATc dose-responsiveness of growth in liquid culture than was apparent on solid media (Figure 3B). However, as observed on agar, the growth of lysA Tet-OFF was only marginally inhibited following lysA silencing with ATc at 200 ng/ml (Figure 3C), an effect reversed by lysine supplementation.
Isocitrate lyase (Icl) catalyzes the conversion of isocitrate to glyoxylate and succinate in the first step of the glyoxylate shunt, a carbon assimilatory pathway that allows the net synthesis of C4 dicarboxylic acids from C2 compounds such as acetate (Munoz-Elias and McKinney, 2006). In Mtb H37Rv, Icl activity is provided by icl1 (Rv0467) (Honer Zu Bentrup et al., 1999). Enzymes of the glyoxylate shunt in Mtb are up-regulated during growth on fatty acid substrates in vitro, and loss of Icl1 activity compromises the ability of Mtb to grow on such substrates. Icl1 has further been demonstrated to be essential for growth of Mtb in macrophages, and during acute and chronic infection in mice (Blumenthal et al., 2010; McKinney et al., 2000; Munoz-Elias and McKinney, 2005). Conditional icl1 mutants were generated by introducing the TetR-expressing plasmids into icl1-SCO and selecting under inducing conditions on media supplemented with glucose and glycerol. Mutants harboring all three TetR-expressing vectors were isolated. To examine the effects of icl1 silencing, the strains were grown in liquid 7H9 medium containing acetate or propionate as the sole carbon source, with or without inducer. None of the strains grew in propionate, even under fully induced conditions, suggesting that the inferred perturbation in the level of icl1 expression resulting from the promoter replacement rendered the strain incapable of providing sufficient 2-methylisocitrate lyase activity to sustain growth in this carbon source (Gould et al., 2006; Upton and McKinney, 2007). However, all three conditional mutants grew in acetate under non-repressing conditions, and the icl1 Tet-OFF strain displayed significant growth attenuation in response to ATc-induced icl1 silencing on solid agar containing acetate as the sole carbon source (Figure 4A), but showed no effect upon growth on glucose and glycerol (data not shown). In contrast, icl1 Tet-ONS and icl1 Tet-ONM failed to display the anticipated ATc-dependent phenotypes during growth on acetate suggesting that Tet regulation of icl1 expression was lost upon propagation of these strains. In liquid culture, the growth of icl1 Tet-OFF on acetate was significantly impaired in the presence of ATc but restored by the omission of inducer (Figure 4B). Similarly, and consistent with the phenotypes observed for icl deletion mutants of Mtb (Munoz-Elias and McKinney, 2005), icl1 Tet-OFF was not adversely affected by icl1 silencing when cultured on glucose (Figure 4C), as compared to acetate (Figure 4B).
Reducing the expression of a target gene is predicted to sensitize cells to inhibitors that act on that target. We therefore compared the susceptibility of icl1 Tet-OFF to the known Icl inhibitor, 3-nitropropionate (3-NP) (Honer Zu Bentrup et al., 1999; Munoz-Elias and McKinney, 2005), with that of wt Mtb and the icl1-SCO strain during growth in glucose- or acetate-containing medium and as a function of inducer concentration. Susceptibility testing was performed using the Alamar blue redox indicator. The indicator is blue in the oxidized state and turns pink upon reduction by metabolically active cells, thus providing a means of detecting growth colorimetrically (Yajko et al., 1995). No significant differences in the susceptibility of the three Mtb strains to 3-NP were observed during growth on glucose, irrespective of the ATc concentration used (Figure 5A). However, when grown on acetate, the susceptibility of icl1-SCO to inhibition by 3-NP was increased 16-fold relative to wt Mtb (Figure 5B). While not measured directly, these results suggest that the level of icl1 expression from the Tet-regulated promoter in icl1-SCO is significantly lower than from the native promoter in wt Mtb. The 3-NP susceptibility of wt Mtb and icl1-SCO was unaffected by ATc whereas icl1 Tet-OFF showed an ATc dose-dependent increase in susceptibility (Figure 5B). In contrast, no differential sensitivity was observed to the anti-tuberculars isoniazid, rifampicin, ethambutol, streptomycin and cycloserine, which act on other cellular targets, during growth of icl1 Tet-OFF on acetate, with or without ATc (Table S1, top panel). Together, these findings demonstrate that the whole-cell activity of 3-NP on Mtb is due to specific inhibition of Icl1, and that icl1 silencing does not confer generalized hypersensitivity to anti-tubercular drugs.
We then used the panC TetOFF strain to assess the antimicrobial activity and target specificity of inhibitors of the Mtb PanC enzyme developed using fragment-based approaches (Hung et al., 2009). The susceptibility of panC Tet-OFF to eight PanC inhibitors was measured in a broth microdilution assay. Although these inhibitors have KD values of 1.8 μM, none of the compounds was active against wt Mtb or panC-SCO at concentrations up to 250 μM. Moreover, in the absence of ATc, the growth of panC Tet-OFF was not significantly affected by the presence of inhibitor relative to the carrier (DMSO) control. However, a marked increase in sensitivity was apparent for all compounds following panC silencing with ATc (10 ng/ml), with the most potent inhibitor, Alv553 (Hung et al., 2009) (Figure 5C), displaying an MIC50 of 31–62 μM (Figure 5D). In all cases, the growth inhibition observed in the presence of ATc was reversed by exogenous pantothenate, confirming the association between growth inhibition and pantothenate biosynthesis. Under identical assay conditions, no alteration in the MICs of known anti-tubercular drugs was observed (Table S1, top panel). Moreover, upon silencing of the corresponding target genes, no increase was observed in the sensitivity of panC Tet-OFF and icl1 Tet-OFF to non-cognate inhibitors, i.e. the Icl- and PanC-specific inhibitors, respectively. Together, these findings provide compelling evidence that the PanC enzyme inhibitors are acting on-target to inhibit the activity of PanC in Mtb, and demonstrate the power of combining fragment-based approaches and TB-WCS using sensitized Mtb strains.
To generate conditional mutants of Mtb that can be used to monitor the activity of small-molecule inhibitors and are also compatible with HTS applications, we assessed the suitability of a fluorescence-based reporter to quantify growth. We cloned the gfp gene encoding green fluorescent protein (GFP) under control of the constitutive M. marinum msp12 promoter (Chan et al., 2002) in the rev-TetR expressing vector used to generate the Tet-OFF strains. The resulting plasmid was introduced into icl1-SCO and panC-SCO to produce the corresponding Tet-OFFGFP derivatives. panC Tet-OFFGFP was indistinguishable from panC Tet-OFF in terms of growth, ATc-induced gene silencing and rescue by pantothenate under repressing conditions on agar (Figure S4A) and in liquid culture (Figure S4B, C). However, low background florescence was observed for panC Tet-OFF (Figure S4D), whereas panC Tet-OFFGFP emitted a strong fluorescence signal (Figure S4E), which correlated well with growth, as measured by turbidity (Figure S4B). Similarly, no significant differences in growth of icl1 Tet-OFF and icl1 Tet-OFFGFP were observed in response to ATc-induced silencing on 7H10 agar containing glucose or acetate as carbon source. The correlation between growth and fluorescence of icl1 Tet-OFFGFP was assayed during growth in 7H9 medium containing acetate as the sole carbon source (Figure S5A). Like icl1 Tet-OFF, growth of icl1 Tet-OFFGFP was attenuated in the presence of ATc in this medium, but reversed by the omission of inducer (Figure S5A) or by replacing acetate with a glycolytic carbon source (Figure S5B). In addition, a good correlation was observed between the fluorescence emitted by icl1 Tet-OFFGFP (Figure S5C, D) and growth, as measured by turbidity (Figure S5A, C). For both reporter strains, the inhibitory activities of known anti-tubercular drugs, as measured by fluorescence-based readout, corresponded well with MIC values determined visually (Table S1, bottom panel). Furthermore, using fluorescence-based readout, the sensitivities of panC Tet-OFFGFP to a PanC-specific inhibitor and of acetate-grown icl1 Tet-OFFGFP to 3-NP were shown to increase, as expected, in the presence of ATc (Figure 6A and 6B, respectively).
The fluorescence-based whole-cell assay for PanC was adapted for use in 384-well microplate format for HTS. To validate the screen, a 600-compound subset of the NCGC library was screened in quantitative HTS (qHTS) format (Inglese et al., 2006), against PanC Tet-OFFGFP grown on butyrate, either with ATc (50 ng/ml), or without inducer. To increase the sensitivity of the screen, butyrate was selected as a fatty acid carbon source as it supports growth of M. tuberculosis via coenzyme-A-dependent β-oxidation to acetyl-CoA. The PanC inhibitor, Alv553, included in the validation subset as a control, was selectively identified as a hit in the +ATc screen. Based on these results, we proceeded to screen a larger compound library for activity against panC Tet-OFFGFP grown on butyrate with ATc in a qHTS format (Inglese et al., 2006) in which each compound was tested at 7 different concentrations. To identify compounds to which panC Tet-OFFGFP was specifically hypersensitized, the same library was screened against wt Mtb grown on butyrate (without ATc) and the results compared. Compounds possessing full concentration-response curves (Figure 7A) (Inglese et al., 2006) and displaying a 10-fold or greater potency (MIC50) against panC Tet-OFFGFP relative to the wt strain were classified as screen-selective hits. Of the 37 hits selected according to this criterion, attention was focused on compounds for which multiple hits were identified within a class, i.e. 6 flavones and 4 piperazines. The differential activity of these compounds against PanC Tet-OFFGFP in the presence and absence of ATc was assessed using the broth microdilution assay. The piperazines failed to re-confirm in this assay (Figure 7B) and were thus excluded from further analysis. However, 4 of the commercially available flavones, 3,5 dihydroxyflavone, 3,7 dihydroxyflavone, 6-methoxy-apigenin, and apigenin-7-O-neohesperidoside, did re-confirm, with the most active of these displaying an ~16-fold increase in potency against panC Tet-OFFGFP in the presence of ATc compared to the ATc-free control (Figure 7B). The growth inhibitory effects of these compounds in the presence of ATc were also reversed by exogenous pantothenate suggesting that the bacilli were sensitized to disruption of some cellular process for which pantothenate is required. To determine if the compounds act directly on PanC, they were tested in biochemical assays against the purified recombinant enzyme. However, none showed inhibitory activity against the PanC enzyme itself (data not shown).
The ability to conditionally regulate the expression of essential genes has formed the basis of cell-based assays designed to evaluate the specificity of inhibitors of a given drug target or, alternatively, to discover new antimicrobial compounds in TB-WCS (Singh et al., 2007; Wang et al., 2006). This approach is predicated on the assumption that a reduction in the levels of an essential target should increase the sensitivity of the host strain to inhibitors that act on that target or indirectly through downstream effects of that target. This approach also assays the target in its native biological environment and samples whole biochemical pathways as well as related pathways and regulatory systems that impinge on that target. In this study, we evaluated the utility of a Tet-based conditional expression system for application in cell-based hypersensitivity assays for three pathways in Mtb. PanC, LysA and Icl1 were selected as test candidates on the basis of their previously demonstrated (conditional) essentiality for growth and survival of Mtb in vitro and in vivo.
Promoter-replacement mutants were generated by SCO homologous recombination. Since the extent of target gene knockdown required to confer a growth phenotype cannot be predicted a priori and is likely to vary from one gene to the next, we employed three TetR-expressing plasmids that differ in regulator mode (forward vs. reverse TetR) and expression level (moderate vs. strong, and from an integrative (single copy) vs. episomal (multicopy) plasmid) to examine the effects of varying both the basal and fully-induced levels of target gene expression on the growth of each SCO strain. This strategy allowed the identification of conditional mutants displaying phenotypes corresponding to those obtained by conventional gene disruption methods. Notably, an icl1 conditional mutant with a predictable and tightly regulated phenotype was only generated in a Tet-OFF configuration and only one conditional lysA mutant was inducer-dependent for growth (lysA Tet-ONM). Furthermore, while a lysA Tet-ONS mutant could not be isolated at all, panC Tet-ONS was viable, albeit unstable in culture. The failure to isolate viable conditional mutants in specific genetic backgrounds, and the propensity of certain strains to lose Tet regulation when propagated in vitro most likely reflects an inability to de-repress target gene expression sufficiently to sustain growth. As such, the approach described herein is unlikely to be equally applicable to all essential targets, and in order to establish suitability for use in TB-WCS, the behavior of conditional mutant strains – especially non-auxotrophic strains that cannot be stabilized by supplementation – will need to be individually assessed.
A key consideration in drug target selection is the extent of target inhibition required before bacterial growth or survival is impaired (Wei et al., 2011). The genetic approach employed herein, which permits the phenotypic effects of target gene expression on bacterial growth to be analyzed at varying levels of stringency, provides a means to identify those targets whose activity need be only partially inhibited by a small molecule before growth and/or survival is compromised (Carroll et al., 2011; Woong Park et al., 2011). In the case of PanC, Mtb was able to grow – albeit at a significantly reduced rate – when the level of the protein was reduced by >95% which suggests that this target may be relatively invulnerable to chemical inhibition. Drug discovery efforts that target enzymes such as PanC are therefore likely to require the development of inhibitors sufficiently potent to exert near complete inhibition of target activity for growth inhibitory effects to be observed (Wei et al., 2011).
The growth of one icl1 and two panC conditional mutants were found to be highly amenable to ATc-dose-dependent gene silencing suggesting that they might serve as useful tools to identify leads and confirm the mode-of-action of drugs that act on these targets. This was demonstrated by the increase in sensitivity of icl1 Tet-OFF to 3-NP when cultured on acetate at growth-limiting concentrations of ATc. The sensitization of this strain was, moreover, limited to a known Icl1 inhibitor during growth on acetate, consistent with the notion that Icl1 is the sole or primary target of this compound in Mtb. Similarly, a specific increase in sensitivity of panC Tet-OFF to PanC inhibitors was observed following ATc-induced panC silencing, and this effect was mitigated by the inclusion of pantothenate in the growth medium. These findings confirm that the PanC inhibitors designed by fragment-based approaches (Hung et al., 2009) retain their target-specificity in whole cells. In a related study, conditional depletion of LepB was recently shown to hypersensitize M. tuberculosis to a type I signal peptidase inhibitor (Ollinger et al., 2012); however, the sensitization (two-fold) observed in this case was markedly lower than for the target/inhibitor couples described in the present study.
By introducing gfp into the panC and icl1 mutants, we generated fluorescent reporter strains that could be used to monitor the activity of antimicrobial compounds in a rapid and quantitative manner. Expression of gfp had no deleterious effects on the growth kinetics, phenotypes, stability or susceptibility to known anti-tuberculars of the conditional mutants used in this study suggesting that they have the potential to be used in HTS applications. A proof-of-concept screen against a small compound library led to the identification of a series of flavones that displayed increased potency against PanC-depleted cells. These compounds did not inhibit purified PanC directly, suggesting that they act on another target enzyme in the pathway or on a target in a related pathway dependent on PanC levels. These results illustrate the strength of TB-WCS compared to traditional in vitro enzyme based screening. Since depletion of an enzyme target is expected to cause a corresponding decrease in the cellular level of its product, conditional mutants may also be sensitized to inhibitors of enzymes in the same biosynthetic pathway – for example, those steps that utilize the target’s product as substrate. TB-WCS also identifies complex regulatory and allosteric molecules that may not be effectively sampled by simple enzyme screening. Potential targets of the flavones thus include non-catalytic sites of PanC, enzymes downstream of PanC in the coenzyme A biosynthetic pathway, and regulators of pantothenate biosynthesis. In this specific case, coenzyme A-dependent enzymes may also be rendered hypersensitive to inhibition through coenzyme A deficiency resulting from PanC depletion. These possibilities, all of which are consistent with the rescue from inhibition conferred by exogenous pantothenate, are currently under investigation. In this regard, a widely used method for target identification is whole-genome sequencing of resistant mutants. The feasibility of using this approach to identify the target(s) of the flavones has yet to be established but is expected to depend on the relative frequencies of mutation of the conditional mutant to flavone resistance vs. loss of ATc-dependent regulation of panC expression.
In summary, the Tet-based promoter replacement has proven to be useful for both confirming and detecting the activity of target/pathway-specific inhibitors in whole cells. Our results demonstrate the power of combining fragment-based approaches and whole-cell screens using sensitized Mtb strains to drive TB drug discovery. The approach described here and by others (Ollinger et al., 2012) should be applicable to a wide range of mycobacterial targets, irrespective of whether biochemical assays are available or the physiological role(s) of the targets are known, and may also find broad application in detecting the activity of compounds that perturb both targets and pathways.
Conditional mutant strains that under-express essential genes have been successfully applied in TB-WCS to identify antimicrobial compounds that possess activity against biochemical pathways. In this study, we demonstrate the feasibility of performing TB-WCS in Mtb by using promoter replacement to generate tetracycline-responsive mutants that conditionally express panC, lysA and icl1 which encode pantothenate synthetase, diaminopimelate decarboxylase and isocitrate lyase, respectively. Using this methodology, we confirmed the essentiality of panC and lysA for growth of Mtb H37Rv in vitro and as well as the conditional essentiality of icl1 for growth on fatty acid substrates. Depletion of Icl1 or PanC consequent on anhydrotetracycline treatment rendered the conditional mutants hypersensitive to target-specific inhibitors without conferring generalized hypersensitivity to drugs that act on other targets or pathways. The results in the case of PanC illustrate the power of combining fragment-based approaches to inhibitor design with whole-cell screening using strains sensitized to target-specific inhibitors by conditional gene silencing. By introducing green fluorescent protein into the conditional mutants, stable reporter strains were generated for use in high-throughput screening. Their utility was demonstrated by identifying compounds that display greater potency against a PanC-depleted conditional mutant than wild type Mtb in a proof-of-concept screen of a small compound library. Rescue from the inhibitory effects of these compounds by exogenous pantothenate suggests that PanC depletion sensitized Mtb to disruption of some pantothenate-dependent cellular process. Together, our findings validate the power of this approach in confirming the mode-of-action of antimicrobial agents directly in Mtb cells, and suggest the potential of TB-WCS to identify small molecule inhibitors against defined biochemical targets or pathways. TB-WCS bridges traditional target-based approaches that engage fully the medicinal chemistry tools of modern drug discovery, and agnostic whole-cell screens that produce the desired phenotype upfront.
Bacterial growth conditions are described in detail in the Supplemental Information.
The procedures used to construct and genotypically characterize the promoter replacement mutants are described in the Supplemental Information.
Western blot analysis of the PanC protein in H37RvMA and the panC conditional mutants was performed as described in the Supplemental Information.
Drug susceptibility testing was performed using broth microdilution assays, as described in the Supplemental Information.
Cells were grown to OD600 ~ 0.2 in 7H9 medium supplemented with glucose or acetate as the sole carbon source, and used as an inoculum following 500-fold dilution in the same medium. 50 μL 7H9 media supplemented with 50 μg/ml Hyg, 25 μg/ml Kan and ATc, as specified, was added to the remaining experimental wells. Using a starting concentration of 500 μM, 2-fold serial dilutions were performed for each of the inhibitors. 50 μL cells were added to each experimental well, to yield a final volume of 100 μL per well. The plates were incubated at 37°C for 7 to 14 days, and the extent of growth determined by the reduction of Alamar Blue (Invitrogen), as specified by the manufacturer. MIC determinations were performed in triplicate.
Cells were grown to an OD600 ~ 0.2 in 7H9 medium, and used as an inoculum following 500-fold dilution in unsupplemented 7H9 medium. Fifty μL of 7H9 media supplemented with 25 μg/ml Hyg, 12.5 μg/ml Km and 2.5 μg/ml Gm in either the absence or presence of 20 ng/ml ATc, or 20 ng/ml ATc and 50 μg/ml pantothenate, was added to the remaining experimental wells. Using a starting concentration of 500 μM, 2-fold serial dilutions were performed for each of the inhibitors. Fifty μL of cells were added to each experimental well, to yield a final volume of 100 μL per well. The plates were incubated at 37°C for 7 to 14 days, and growth determined by the visible growth present in each well. MIC determinations were performed in triplicate for each of the inhibitors.
The NCGC screening library comprises a diverse collection of 13,549 small molecules designed by the NIH Chemical Genomics Consortium using commercially available compounds consisting of targeted library chemotypes focused on proteases, kinases, GPCRs, ion channels and nuclear receptors; a diversity collection of ‘drug like’ compounds with MW < 500, LogP < 5; a small collection of natural product derived compounds; and a known actives collection consisting of Sigma’s 1280-member LOPAC library, a Microsource collection, a Prestwick library and a collection from Tocris. For screening of the library in 384-well format, 13 μL of Middlebrook 7H9 media supplemented with butyrate as the sole carbon source (H. Boshoff & C.E. Barry III, details to be described elsewhere), was dispensed into each of the wells of 384-well, black, clear-bottomed micro-titer plates (Greiner Bio One, Monroe, NC) using a Biomek FX liquid handler with 384-well dispensing head (Beckman Coulter, Brea, CA). Rifampicin (50 μg/ml) and DMSO served as positive and negative controls, respectively, and were included in the first two columns of each screening plate. Using a 384-well pin tool (V&P Scientific, San Diego, CA), 130 nl of each screening compound was dispensed into the remaining wells of the micro-titer plate. Mtb cultures were grown to an OD600 of ~0.4 and diluted 10-fold in 7H9 media in either the presence of ATc (100 ng/ml), or absence of inducer, as required, immediately prior to use. Thirteen μL of the diluted culture was aspirated into the wells of each plate using the Biomek FX liquid handler. Plates were sealed using the Abgene ALP-300 Sealer (ABgene, Surrey, UK) and stored in a plate hotel prior to transfer to 37°C. The plates were centrifuged at 3000 rpm and the seals removed and replaced with a Kalypsys stainless steel lid with rubberized gasket (Wako Automation, Richmond, VA). The lidded micro-titer plates were then placed in a 37°C incubator (VWR, Radnor, PA) in stacks of six; no more than 36 plates were placed in a single incubator. After 72 hours, the plates were removed from the incubator and the stainless steel lids replaced with plate seals using the Abgene ALP-300 sealer. The plates were centrifuged briefly (3000 rpm for 3 seconds) and read using an EnVision plate reader (Perkin Elmer, Wellesley, MA), using excitation and emission wavelengths of 485 and 528 nm, respectively. Analysis of compound concentration-responses was performed as previously described (Inglese et al., 2006). Data for each titration point were normalized relative to the rifampicin (100% inhibition) and DMSO controls (0% inhibition). Concentration-response titration points for each compound were fitted to the Hill equation to calculate half-maximal activity (IC50) values, and those compounds possessing full concentration-curves (Class 1.1, 1.2, 2.1, or 2.2) (Inglese, et al., 2006) were selected for further analysis. Confirmatory screens using the broth micro-dilution assay were performed using compounds purchased from Fisher Scientific, Sigma-Aldrich or Tocris.
Recombinant PanC was isolated and purified as described in the Supplemental Information.
This work was funded by grants from the Bill and Melinda Gates Foundation and, in part, by the intramural research program of the NIAID, NIH. We thank Paul Shinn, Adam Yasgar, Carleen Klumpp and Ajit Jadhav for construction of the NCGC library, Dirk Schnappinger and Sabine Ehrt for providing pSE100, pMC1s, pMC2m and pTEK-4S-OX, Courtney Aldrich for the PanC over-expressor, Jim Sacchettini for the anti-PanC antibodies, Christopher Sassetti for Mtb H37RvMA, Deborah Hung for pMSP12::GFP and Graham Hatfull for pTT1B. We also thank Ben Winterroth for technical assistance, Sir Tom Blundell, Leonardo Silvestre, Alessio Ciulli, Joshua Odingo, Bavesh Kana, Bhavna Gordhan and Kristi Guinn for advice and assistance, and Digby Warner for critically reviewing the manuscript.
Supplemental Information includes six figures, three tables, and experimental procedures.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.