|Home | About | Journals | Submit | Contact Us | Français|
The type II secretion (T2S) system in Gram-negative bacteria is comprised of the Sec and Tat pathways for translocating proteins into the periplasm and an outer membrane secretin for transporting proteins into the extracellular space. To discover Sec/Tat/T2S pathway inhibitors as potential new therapeutics, we used a Pseudomonas aeruginosa bioluminescent reporter strain responsive to SecA depletion and inhibition to screen compound libraries and characterize the hits. The reporter strain placed a luxCDABE operon under regulation of a SecA depletion-responsive up-regulated promoter in a secA deletion background complemented with an ectopic lac-regulated secA copy. Bioluminescence was indirectly proportional to the IPTG concentration and stimulated by azide, a known SecA ATPase inhibitor. A total of 96 compounds (0.1% of 73,000) were detected as primary hits due to stimulation of luminescence with a z-score ≥5. Direct secretion assays of the 9 most potent hits, representing 5 chemical scaffolds, revealed that they do not inhibit SecA-mediated secretion of β-lactamase into the periplasm, but do inhibit T2S-mediated extracellular secretion of elastase with IC50 values from 5 – 25 μM. In addition, 7 of the 9 compounds also inhibited the T2S-mediated extracellular secretion of phospholipases C by P. aeruginosa and of protease activity by Burkholderia pseudomallei.
The prevalence of multidrug resistance (MDR) is increasing among bacterial pathogens, especially among Gram-negative nonfermenters such as Pseudomonas aeruginosa.1 As a last resort, polymyxins such as colistin have been revived for use against these infections, but resistance to this agent is also becoming more common and some species such as Burkholderia spp. are intrinsically polymyxin B resistant.2 The development of new antibacterials with novel mechanisms of action or that inhibit novel targets is considered an important strategy for combating drug resistance since no pre-existing target-based resistance mechanisms would exist for such therapeutics.
The type II secretion (T2S) system in Gram-negative bacteria represents an important under-exploited potential target for the treatment of bacterial infections. Proteins are first translocated across the inner membrane by either the Sec 3 or Tat 4 machinery, and then, a subset of those proteins are transported into the extracellular milieu by an outer membrane secretin.5 The Sec-mediated portion of the pathway is essential to bacterial growth,6 and the secretin-mediated portion is important for bacterial virulence.5 Consequently, inhibitors of this set of secretion pathways are expected to be useful therapeutically to block bacterial cell growth or viability or to reduce virulence in infections.
Unfortunately, the T2S secretory pathways are complex molecular machines that are not amenable to monitoring with simple biochemical assays. While inhibition of the SecA ATPase can be measured readily,7 such assays can only detect inhibitors of a single component of the complex pathway. In addition, ATPase inhibitors may not be sufficiently selective for development as antibacterials. The difficulties in developing suitable biochemical screens together with the obvious utility of whole-cell screens for discovering inhibitors that escape efflux and are capable of cell growth inhibition led us to develop a cellular reporter screen for T2S inhibitors. The strategy employed the cellular transcriptional regulatory network to transduce target inhibition into increased bioluminescence. Suitable target gene depletion-responsive promoters were identified by expression profiling of cells under conditions in which the target gene was down-regulated, resulting in reduced cell growth.8 Promoter regions from responding genes were fused to the Photorhabdus luminescens luxCDABE operon and integrated into the reporter strain chromosome. The resulting whole-cell, bioluminescent, gain-of-signal, target-biased reporter strains for T2S inhibitors provide sensitivity to compounds capable of acting on whole cells, but are insensitive to compounds that eliminate luminescence by inhibiting luciferase or damaging the cellular energy-generating capability.
We report here the results of a screen for compounds that stimulate luminescence by a SecA-depletion-responsive reporter strain. Screening hits were selected and tested in secondary assays directly for inhibition of secretion of proteins utilizing different aspects of the T2S pathway, including secretion of β-lactamase (Sec-mediated) into the periplasmic, and extracellular secretion of elastase (Sec-mediated) and phospholipases C (Tat-mediated). The nine most potent inhibitors did not affect Sec-mediated translocation of β-lactamase into the periplasm, but did inhibit the extracellular secretion of elastase, and 7 of the 9 also inhibited the extracellular secretion of phospholipases C. In addition, most of the inhibitors blocked the extracellular secretion of proteases from the related species Burkholderia pseudomallei.
Bacterial strains and plasmids used for assays are described in Table 1. E. coli TOP10 (Invitrogen®), E. coli DB3.1 (Gateway® host, Invitrogen®), E. coli SM10,8 and E. coli S17-1 (ATCC 47055) were used as hosts for molecular cloning. Luria-Bertani (LB) medium (liquid and agar) was purchased from Difco. LB was supplemented with 200 μg/ml spectinomycin (50 μg/ml for efflux-deficient PAO397-derived strains) with or without 10 μg/ml gentamicin or isopropyl-β-D-thiogalactopyranoside (IPTG; 0.05 mM unless indicated otherwise) to create media LBS, LBSG and LBSGI, respectively. Trypic soy dialysate treated with Chelex-100 (TSBDC) was used for growth of B. pseudomallei for protease secretion.9
A gene encoding TEM1 β-lactamase (βLA) was cloned behind the tac promoter in plasmid pVLT35 10 as follows. The blaM gene was amplified from plasmid pBR322 DNA by PCR with primers BLA-F+Eco (5′-ATCCTACGTAgagtatgagtattcaacatttccgtgt) and BLA-R+Hind (5′-TCCCCAAGCTT-ttggtctgacagttaccaatgc), carrying EcoRI and HindIII sequence tails, respectively (denoted in upper case). Resulting PCR products were gel-purified, cleaved with EcoRI and HindIII, ligated directionally into pVLT35 which had been cut with the same two restriction endonucleases, and used to transform E. coli TOP10 cells to ampicillin and spectinomycin-resistance. Plasmid pMDM1350 was confirmed by PCR and sequencing to contain the TEM1 βLA gene downstream from the tac promoter in pVLT35. Plasmid pMDM1350 was introduced into P. aeruginosa PAO-LAC by electroporation,11 selecting for spectinomycin resistance on LBSG medium to generate strain MDM1368.
Deletions of the T2S secretin-encoding xcpQ gene marked with a tetracycline-resistance element were generated in strains PAO1 and PA14 of P. aeruginosa by a standard Gateway™ and PCR splicing by overlap extension technology as described.12 The following primers were used (attB1 and attB2 tails in all caps): Pae-xcpQ-UP-F+GWL (TACAAAAAAGCAGGCTggtaatggggctgtcatcat), Pae-xcpQ-UP-R+Tc (TCCTGCGTTATCCCCTGATTCTGTGGATAAgactggaaggggcaaacag), Pae-xcpQ-DWN-F+Tc (GCTAACGGATTCACCACTCCAAGAATTGGAcaaccagttgttcgatggac), and Pae-xcpQ-DWN-R+GWR (TACAAGAAAGCTGGGTgaagacctcgacgacctcaa).
Compounds screened in this study were purchased from Chembridge (San Diego, CA) and Timtec (Newark, DE), diluted in 96-well master plates at 2.5 mM in DMSO, and stored at −20°C.
For inhibitor screening, compound master plates were thawed at room temperature, and 2 μl of compound (2.5 mM stock) was added to 384-well opaque white screening plates using a Sciclone ALH 3000 liquid handling robot (Caliper, Inc.) and a Twister II Microplate Handler (Caliper, Inc.). Compounds were screened at 50 μM in 2% DMSO. Positive control wells contained 25 μM 5,7-dichloro-hydroxyquinoline (DHQ), and the negative control wells contained 2% DMSO only. Construction of the SecA-depletion responsive, bioluminescent reporter strain MDM1145 has been described previously.8 Reporter strain MDM1145 cells were inoculated into 10 ml LBSGI and grown overnight at 37°C. The following morning, the culture was diluted 125-fold into LBSG and grown for ~ 2h until it reached an OD600 = 0.1. The screen was initiated by the addition of 100 μl culture into each well. Plates were covered with a translucent gas-permeable seal (Abgene, Inc., Cat. No. AB-0718) and incubated at room temperature for 180 min. Then, luminescence was read in an Envision Multilabel microplate reader (PerkinElmer). The Z′-factor 13 was calculated for each screening plate, and the z-score (number of standard deviations above or below the negative control) was derived for each sample well. Screening results and secondary assay data were stored in a ChemBioOffice database (v. 12; CambridgeSoft, Inc.). Validated hits were reordered from the vendor and confirmed to be >95% pure and to be of the expected mass by LC-MS analysis.
The effects of sodium azide on reporter strains MDM1143 (PA2403-luxCDABE), MDM1144 (PA2408-luxCDABE), MDM1145 (PA1365-luxCDABE), and MDM1167 (PA2404-luxCDABE)(Table 1) was measured in white opaque 96-well microplates as follows. Cells were grown as described for the screen (above). Then, 100 μl of cells were added to wells containing either 2 μl DMSO or sodium azide in DMSO (25 mM stock; final azide is 0.5 mM). Luminescence was measured at 1.5, 2.0, 2.5, 3.0 and 3.5 h in an Envision Multilabel microplate reader (PerkinElmer).
Hit compounds were profiled for their ability to stimulate luminescence of a panel of efflux-deficient reporter strains designed to respond to inhibition of Sec-related and unrelated targets (Table 1). Each reporter strain in the panel contains a transcriptional fusion to P. luminescens luxCDABE in the efflux-deficient strain PAO397 generated by single cross-over insertion with the pGSV3 constructs described previously.8 However, the fatty acid biosynthesis pathway reporters were generated by mini-Tn7 insertion of the promoter elements from the genes indicated in Table 1.14 Cells were grown as described above, and 200 μl of culture was added to wells of a white opaque 96-well microplate containing 2 μl of DMSO or test compound (2.5 mM stock; 25 μM final). Luminescence was measured at 2 h in an Envision Multilabel microplate reader (PerkinElmer).
After growth of strain MDM1368 (carrying tac-promoted blaM gene on plasmid pVLT35, see Table 1) in LBS for 4h in the presence and absence of test compounds, IPTG was added to induce βLA production for 1.5h. At a culture density of A600~0.5, cells were processed as described,15 and 100 μl of osmotic shock fluid representing the contents of the periplasm was added to wells of clear 96-well microplates. Slopes from the kinetics of nitrocefin (Calbiochem) hydrolysis (A490 in a Victor3V 1420 Multilabel HTS Counter (PerkinElmer)) were normalized to the cell density (OD600) of harvested cells as a relative measure of the quantity of the βLA secreted. The positive control (100% inhibition) consisted of shock fluid from cells subcultured into LBS without IPTG. Typical signal:background ratios were 10–15.
The effect of test compounds on T2S-mediated extracellular secretion of elastase from P. aeruginosa was determined by a previously described method,16 which measures the proteolysis of elastin conjugated with congo red. Briefly, P. aeruginosa PA14 cells were cultured for 16 h to saturation in LB in the presence or absence of test compounds in a 2-fold dilution series. Cells were removed by centrifugation, and the quantity of elastase in cleared supernatants was determined by the digestion of elastin-Congo Red (Sigma). Readings (A495) were normalized to the harvested cell density (A600), and % inhibition of elastase secretion was determined relative to values from untreated PA14 (no inhibition control) and from untreated T2S secretion defective PA14 ΔxcpQ (strain MDM1485, Table 1) (100% inhibition control). Typical signal:background ratios ranged from 6 to 9.
The effect of test compounds on the type II-mediated secretion of phospholipases C (PlcH/N) from P. aeruginosa was determined as described.15 Briefly, P. aeruginosa PAO1 cells were grown for 16h to saturation in low-phosphate TY medium in the presence or absence of test compounds in a 2-fold dilution series. Cells were removed by centrifugation, and the quantity of PlcH/N in the clear supernatants was assayed using the chromogenic substrate p-nitrophenylphosphoryl choline (NCCP, Toronto Research Chemicals, Inc.). Readings (A405) were normalized to the cell density (A600) of harvested cells, and % inhibition of PlcH/N secretion was determined relative to the values from untreated PAO1 (no inhibition control) and from T2S-defective PAO1 ΔxcpQ (strain MDM1484, Table 1)(100% inhibition control). Typical signal:background ratios were 10–15.
The effect of test compounds on the T2S-mediated secretion of proteases from Burkholderia pseudomallei was determined using a published assay.17 Cells of B. pseudomallei strains DD50318 (T2S-competent) and DD21319 (T2S-defective xcpQ::Tn5-OT182 mutant) were grown in TSBDC medium for 16h in the presence or absence of test compounds at indicated concentrations. Cells were removed by centrifugation, and the quantity of secreted proteolytic activity in cleared supernatants was determined using the chromogenic substrate azocasein. Readings (A440) were normalized to the cell density (A600) of harvested cultures, and the % inhibition of protease secretion was determined relative to values from untreated DD503 cells (no inhibition control) and from DD213 cells (100% inhibiton control). Typical signal:background ratios were about 2.5.
The construction and preliminary evaluation of four bioluminescent, gain-of-signal, P. aeruginosa reporter strains for inhibitors of Sec/T2S mediated secretion have been described previously.8 Briefly, down-regulation of a lac-regulated complementing plasmid copy of either the P. aeruginosa or B. pseudomallei secA gene in a P. aeruginosa secA deletion strain resulted in complete inhibition of growth and the up-regulation of several genes as assessed by expression profiling. Internal coding regions from each of the four most selectively up-regulated genes were fused to the P. luminescens luxCDABE operon and used to direct integration by single cross-over recombination into the P. aeruginosa chromosome to produce transcriptional fusion strains that respond to depletion of SecA by increased bioluminescence (Table 1). The four reporter strains are isogenic except for the chromosomal location of the lux operon and appear to represent only two transcriptional regulatory elements. In strain MDM1145, the lux operon is under control of the PA1365 regulatory element. In the other three strains, MDM1143, MDM1167, and MDM1144, the lux operon is fused to three different members of an apparent operon, PA2403, PA2404, and PA2408, likely placing the lux operon under control of the same regulatory element in all three cases. Interestingly, none of the predicted functions of these loci bears any obvious relationship to the Sec, Tat or T2S pathway-encoding genes. Nevertheless, expression of these genes is elevated selectively in response to depletion of SecA vs. depletion of GyrA, CoaD, GlmU or a variety of stress conditions.8 Therefore, we reasoned that they could be used as the basis of a screen for SecA inhibitors, provided that suitable secondary assays were employed to eliminate compounds acting by any non-selective mechanisms.
We characterized the four potential SecA-responsive reporter strains further in order to evaluate their utility for screening. Bacterial mutations conferring resistance to millimolar levels of sodium azide map to the secA gene,20 and biochemical studies have demonstrated that azide inhibits the SecA ATPase and traps SecA in the membrane inserted state, thus blocking protein translocation.21 Therefore, we anticipated that sodium azide addition would produce bioluminescence from an appropriate reporter strain that is sensitive to SecA inhibition or depletion and capable of generating a luminescent response prior to cell death or ATP limitation. Accordingly, bioluminescence from the four reporter strains was examined in the presence of sodium azide. As shown in Fig. 1, only the strain carrying luxCDABE transcribed from the PA1365 promoter responded appropriately, generating luminescence at least 4 standard deviations above that of the untreated control by 2.5h after azide addition. Based on its bioluminescent response to azide addition, strain MDM1145 carrying the PA1365-luxCDABE construct was selected for further evaluation.
To ensure that the induction of PA1365-luxCDABE and the resulting bioluminescent report are related to depletion of SecA and not a general consequence of growth inhibition or stress response, we compared the growth and luminescence of a secA complemented deletion and a glmU complemented deletion, both carrying the same PA1365-luxCDABE reporter construct, upon withdrawal of the IPTG inducer. Both strains ceased growth within 3 hr of IPTG removal (Fig. 2AB), indicating that cells are quite vulnerable to the loss of SecA or GlmU and suggesting that these are useful targets for screening. Depletion of SecA but not GlmU produced a luminescent signal within 3h of IPTG withdrawal, with intensity peaking at 5h (Fig. 2C). The luminescent response during the measurement phase was inversely proportional to the concentration of IPTG inducer present during the overnight growth phase, indicating that the report is proportional to the extent of SecA depletion. Since growth of reporter cells in the presence of IPTG (0.05 mM) did not result in a significant production of luminescence after withdrawal of the inducer, these conditions were used for screening. Clearly, the bioluminescent report from MDM1145 is not a stress response to growth limitation since it responds to SecA depletion but not to GlmU depletion.
In order to assess the optimal conditions for the strain MDM1145 bioluminescent screen, we performed a pilot screen of a library consisting of 2,000 compounds from a biologically active and structurally diverse set of known drugs, experimental bioactives, and pure natural products (“Spectrum” library; Microsource Discovery, Inc.) in 384-well microplates. A total of 21 compounds generated luminescence values (RLU) ≥4 standard deviations (z-score ≥4) above the average of the negative control wells (DMSO only) for a hit rate of about 1%. The 21 hits were picked from the master plates and retested in quadruplicate in the standard screening assay with strain MDM1145, resulting in confirmation of 14 hits and a confirmed hit frequency of 0.7%. The three most potent (highest z-scores) hits were characterized further, demonstrating concentration-dependent RLU generation (data not shown) and establishing the kinetics of RLU generation (Fig. 3). The two most potent hits are both hydroxyquinolines (8-hydroxyquinoline and 5,7-dichlorohydroxyquinoline). A tetracycline analog, demeclocycline, generated more modest but significant luminescence, confirming the need for selective secondary assays to validate screening hits. Based on the fact that these pilot screen hits generated maximal, stable luminescence between 2h and 3h, which is similar to the timeframe for azide stimulation of bioluminescence (Fig. 1), we selected 3h of incubation in the presence of compounds for the high throughput screening protocol.
Nearly 73,000 discrete chemical compounds were screened at 50 μM for stimulation of luminescence from reporter strain MDM1145. Screening results are shown graphically for 11 representative 384-well assay plates (3,520 compounds) in Fig. 4. The signal-to-background ratio for this set of screening plates is 2.8, and the calculated Z′-factor is 0.38.13 These values were typical for the entire screen, and they reinforce the need for discriminating secondary assays to identify true secretion inhibitors from among the primary hits (see below). While the Z′-factor is sub-optimal for HTS, it is important to note that the maximum possible signal in this gain-of-signal reporter screen is undefined since a different positive control or a higher concentration of the control could generate more luminescence. Therefore, the variation of the negative controls (DMSO only) is a more critical factor for screening quality, and the 7.7% variation (standard deviation/average) of the negative controls in Fig. 4 indicates that stimulation of luminescence can be detected sensitively. Consequently, only the negative controls were used in hit calculations, and highly significant up-regulation of luminescence was required for each hit (i.e., z-score ≥5 above the average of the negative control wells).
A total of 96 compounds (0.13% of the library) were detected as primary hits (z-score ≥5, represented by the solid line in Fig. 4), and about 35% of them met two confirmation requirements: (a) confirmation as inhibitors when re-tested in the same assay in triplicate and (b) demonstration of concentration-dependent stimulation of luminescence. Further evaluation for specificity of luminescence response and direct measurements of inhibition of secretion (see below) reduced the number of validated hits to nine, which are shown in Table 2.
As noted above, the bioluminescence of MDM1145 is stimulated by reduced expression of secA (Fig. 2) and by the presence of azide (Fig. 3). However, since the promoter driving the lux operon in the reporter strain is derived from a gene of unknown relationship to secretion, it is also possible that compounds could stimulate luminescence by alternate unknown mechanisms. To investigate the specificity of the cellular luminescent response to the nine hits, we tested them for stimulation of bioluminescence of a panel of efflux-deficient reporter strains carrying promoters that were demonstrated to be responsive to depletion of a variety of secretion-related as well as unrelated targets. We reasoned that compounds that only stimulate the SecA-depletion-responsive promoters were most likely to act on the secretion pathway. The results in Fig. 5 demonstrate that the nine confirmed hits stimulate luminescence driven by the same promoter (PA1365) as used in the screening strain and one or more additional SecA reporter strains, but do not stimulate luminescence (insignificant or negative z-scores) from 7 other reporters designed to respond to inhibitors of GyrA, GlmU, and CoaD8 or to inhibitors of the fatty acid biosynthesis pathway.
The nine specific, confirmed hits were examined in secondary assays for direct inhibition of protein secretion. We reasoned that the reporter screen may detect inhibitors of any step in the T2S pathway. Therefore, effects on the secretion of β-lactamase into the periplasm and of elastase and phospholipases C into the extracellular medium were investigated in order to validate the hits as actual inhibitors of secretion and to distinguish between inhibition of Sec-mediated transport across the inner membrane vs. secretin-mediated translocation across the outer membrane. None of the 9 compounds inhibited Sec-mediated secretion of β-lactamase into the periplasm, but all of them inhibited T2S-mediated secretion of elastase into the extracellular medium, with IC50 values ranging from 5 to 32 μM (Table 2). Most of the compounds also inhibited T2S-mediated secretion of phospholipases C with potencies similar to those observed for inhibition of elastase secretion, except for compounds 3 and 7, which failed to demonstrate significant inhibition of phospholipases C secretion. In all cases, control experiments in which compounds were added after the incubation period for secretion and just before the addition of the secreted enzyme substrate demonstrated clearly that the compounds did not inhibit elastase or phospholipases C themselves, but only the secretion of those proteins (data not shown). In addition, none of the compounds inhibited growth of the P. aeruginosa cells (data not shown), consistent with their failure to inhibit the essential Sec pathway.
Since B. pseudomallei and P. aeruginosa are evolutionarily related organisms whose proteins share considerable sequence similarity and cross-complementation capability,8 we tested the T2S inhibitors for inhibition of T2S-mediated secretion by B. pseudomallei. All but two of the 9 compounds displayed significant inhibition of B. pseudomallei protease secretion (Table 2). Compound 3 failed to exhibit significant inhibition, and it also failed to inhibit phospholipases C secretion from P. aeruginosa. Compound 9 also failed to inhibit protease secretion from B. pseudomallei, but was a potent inhibitor of phospholipases C secretion in P. aeruginosa. Thus, the compounds generally functioned as T2S inhibitors in B. pseudomallei, but displayed some species-specificity in their potency.
In this report, we describe the application of a cellular, bioluminescent, gain-of-signal, reporter screen to identify inhibitors of T2S in P. aeruginosa and B. pseudomallei. The Sec/T2S secretory pathways have long been considered potential targets for the discovery of new antibacterial therapeutics. However, these pathways are complex, and most aspects are not readily probed by biochemical assays. In addition, biochemical screens of large numbers of antibacterial targets have not yielded inhibitors with growth inhibitory properties in acceptable frequency.22
An interesting alternative approach to biochemical screens, with the capability of solving both the problem of targeting complex molecular machines and the difficulty of identifying compounds that gain access to the cell interior, is the use of whole cell reporter screens. In fact, Alksne et al. 23 took advantage of the observation that translation of SecA is up-regulated when secretion is blocked in E. coli to construct and apply a translational reporter secA-lacZ fusion screen for secretion inhibitors. Unfortunately, all of the compounds identified in the screen exhibited non-specific effects on bacterial and mammalian membrane integrity,23 possibly related to the fact that the Sec pathway involves several membrane-localized proteins. This result illustrates one of the limitations of cell-based reporter screens, namely, that signal reports are not always due to specific target-inhibitor interactions. The translational or transcriptional regulation used as the basis for generating signals is often multi-factorial and incompletely understood, allowing for both target specific and non-specific generation of signals. Nevertheless, these limitations in assay specificity appear to be a reasonable trade-off for gaining high throughput capability, provided that suitable secondary assays can be applied rapidly to the small subset of the screening library appearing as hits. Others have used secA anti-sense expression to sensitize cells to SecA inhibitors and identified a natural product with weak potency against several Gram-positive bacterial species24. However, such screens are cumbersome to perform, requiring comparison of zones of inhibition on agar media.
Considerable effort has also been expended on developing reporter strains based on transcriptional regulation, and we took this approach to developing secretion inhibitor screens. Bioluminescent reporter screens for a variety of targets have been described in Bacillus subtilis,25; 26 Escherichia coli,27 and Pseudomonas aeruginosa.14 In these cases, transcriptional responses to down-regulation or inhibition of targets are captured by fusions of the responsive promoter regions to genes encoding luciferase. Luminescent assays provide several benefits over absorbancy or fluorescence read-out assays. Primarily, they suffer from fewer false positives and negatives due to compound absorption, fluorescence, or quenching. But, in addition, they are extremely convenient and inexpensive if the entire lux operon is fused to the appropriate promoter because no substrate addition is required for generation of luminescence.
Construction of luminescent reporter screens requires the identification of suitable promoters to drive the luminescent response. We previously used an artificially imposed SecA limitation to probe the transcriptional regulatory system of P. aeruginosa cells to find genes that are up-regulated in response to SecA depletion,8 and the strains constructed in that study were examined here for their suitability for inhibitor screening. Surprisingly, none of the highly responsive genes exhibited any clear relationship to secretion. Genes encoding obviously mechanistically related products were unperturbed by the down-regulation of secA. For example, expression of Sec component genes secB, secD, secE, secF, secG, secY and yjaC was unaltered in response to SecA depletion, as were the secretin-encoding xcp and hxc loci 28; 29 and the tat loci.4 Expression of only a few genes appeared to be reproducibly highly up- or down-regulated in a selective manner in response to SecA depletion. Since we preferred to build a gain-of-signal reporter strain, we selected several highly responsive up-regulated genes that did not respond significantly to depletion of GyrA, CoaD, GlmU, or to stressors such as heat shock, H2O2, or nitrosative stress,8 and we built reporter strains from four of them for evaluation in this study.
The predicted product of gene PA1365 is annotated as a probable siderophore receptor localized to the outer membrane, but has no obvious involvement with Sec, Tat, or T2S secretion pathways. Nevertheless, several characteristics validated it as the most appropriate source of a promoter to drive the P. luminescens luxCDABE operon in reporter strains for identifying secretion inhibitors. First, down-regulation of secA, resulting in depletion of SecA to the point of growth rate reduction, stimulated the production of luminescence from the PA1365-luxCDABE transcriptional fusion in P. aeruginosa cells. Second, the PA1365-luxCDABE fusion, but not the PA2403, PA2404, or PA2408 fusions to luxCDABE, produced significant luminescence when cells were treated with azide, a known SecA ATPase inhibitor. Third, the PA1365-luxCDABE fusion responded selectively to SecA depletion and not to GlmU depletion, even though reduction in expression of both genes resulted in growth inhibition. These features, together with the availability of secondary assays to measure direct inhibition of secretion of several specific proteins, argued for applying the reporter screen to a set of compounds in order to verify its utility.
The screening results indicate that the reporter strain is capable of detecting inhibitors of T2S. Indeed, nine inhibitors were demonstrated to inhibit the outer membrane translocation of elastase (LasB) from P. aeruginosa cells, and seven of the nine also inhibited outer membrane translocation of phospholipases C (PlcH/N). Since LasB is transported across the cytoplasmic membrane by Sec and PlcH/N by Tat, the inhibitors must be acting on a target distal to the Sec and Tat machinery. This is consistent with the failure of all nine compounds to inhibit the Sec-mediated secretion of β-lactamase into the periplasm.
In this study, we also examined the effectiveness of a panel of different target-selective bioluminescent reporter strains as a secondary assay to profile confirmed primary hits. The nine T2S inhibitors produced significant luminescent signals from strains carrying one or more Sec-depletion responsive promoter-lux fusions, but failed to generate luminescence from promoter-lux fusions designed to respond to depletion of GlmU, CoaD, GyrA, or enzymes of the fatty acid biosynthetic pathway (Fig. 5). These results are consistent with previous reports using B. subtilis reporter strains to characterize known antibacterial compounds,26 and indicate that this approach is helpful for early profiling of hits.
The structures of the inhibitors may be classified in five different scaffold series, with multiple members of two of the series (Table 2). The presence of the 8-hydroxyquinoline motif in compounds 4–8 suggests metal co-ordination capability. However, addition of excess zinc or iron ions failed to modulate the inhibition of PlcH/N secretion by the compounds (D.T.M. and M.D., unpublished observations). Others have shown that 8-hydroxyquinolines may display specific inhibition of targets despite their general ability to co-ordinate metals.30 The failure of compound 7 to inhibit PlcH/N secretion is surprising since it is clearly related structurally to compounds 4, 5, and 6, all of which inhibit PlcH/N secretion. However, compound 7 lacks the aromatic substituent on the piperazine moiety, which may be important for the potency of inhibition. Compound 3 failed to inhibit secretion of either PlcH/N from P. aeruginosa or protease from B. pseudomallei. This is consistent with its poor z-score in the reporter screen confirmation and its relatively weak inhibition of LasB secretion. However, the failure of compound 9 to inhibit protease secretion from B. pseudomallei despite its reasonably good potency for inhibition of LasB and PlcH/N secretion from P. aeruginosa is puzzling. Interestingly, the 8-hydroxy moiety is acylated in compound 9 but not in compounds 4–8. Testing of additional analogs will be required to understand whether this difference explains the poor activity in B. pseudomallei.
It is of interest that none of the validated inhibitors target SecA or any step in the essential Sec-mediated cytoplasmic translocation process. Consequently, these confirmed hits could lead to useful adjunctive anti-virulence therapeutics, but none have the potential to be optimized as antibiotics. This limitation in the types of inhibitors identified could be due to the reporter promoter used, or it could be due to the fact that the screen was carried out using an efflux-competent P. aeruginosa strain. This was an intentional choice since we wished to identify compounds that could penetrate and accumulate in wild-type cells. However, it is widely recognized that few compounds are capable of this,22 and the easier access to the outer membrane components of the T2S pathway may have biased the screening results. Future studies will focus on construction and application of the reporter screen in an efflux-deficient strain of P. aeruginosa, which may enable the discovery of inhibitors of the inner membrane Sec and Tat machines. Nevertheless, the results of this study demonstrate the utility of cellular, gain-of-signal, bioluminescent reporter strains for detecting P. aeruginosa and B. pseudomallei secretion inhibitors.
We thank Dr. Stephen Lory (Harvard U. Medical School) for advice. We thank Dr. Timothy Opperman for advice and for a critical reading of the manuscript. We thank Daniel Aiello for assistance with data formats and database administration.
This work was supported in part by DHHS/NIH grant AI-056644 from the National Institute of Allergy and Infectious Diseases (NIAID).