ABSTRACT
LcrF (VirF), a transcription factor in the multiple adaptational response (MAR) family, regulates expression of the Yersinia type III secretion system (T3SS). Yersinia pseudotuberculosis lcrF-null mutants showed attenuated virulence in tissue culture and animal models of infection. Targeting of LcrF offers a novel, antivirulence strategy for preventing Yersinia infection. A small molecule library was screened for inhibition of LcrF-DNA binding in an in vitro assay. All of the compounds lacked intrinsic antibacterial activity and did not demonstrate toxicity against mammalian cells. A subset of these compounds inhibited T3SS-dependent cytotoxicity of Y. pseudotuberculosis toward macrophages in vitro. In a murine model of Y. pseudotuberculosis pneumonia, two compounds significantly reduced the bacterial burden in the lungs and afforded a dramatic survival advantage. The MAR family of transcription factors is well conserved, with members playing central roles in pathogenesis across bacterial genera; thus, the inhibitors could have broad applicability.
Three species of the genus Yersinia are pathogenic to mammals, including humans: Yersinia enterocolitica, Yersinia pseudotuberculosis, and Yersinia pestis, the causative agent of plague. Treatment and prophylaxis for Y. pestis infection relies on antibiotics (22), yet single- and multiple-drug resistance (MDR) has been described for clinical isolates (17, 41). For this reason, a novel approach for preventing and/or controlling infections caused by Yersinia spp. (and other Gram-negative bacterial species), with a low propensity to select for resistance, is sought. Such an opportunity exists with the use of therapeutics that inhibit pathogen virulence but not growth outside the host. To this end, we have targeted a family of virulence-regulating transcription factors called multiple adaptational response (MAR) proteins (6).
MAR proteins are members of the AraC transcription factor family and regulate the expression of virulence factors in many clinically important bacterial species, including Yersinia spp., Pseudomonas aeruginosa, Escherichia coli (including enteroaggregative, enterotoxigenic, and enteropathogenic strains), Klebsiella spp., Shigella spp., Salmonella spp., Vibrio cholerae, Staphylococcus aureus, and Streptococcus pneumoniae (18). MAR proteins are established virulence factors; inactivation of these proteins through genetic mutation attenuates the virulence of bacterial pathogens in animal models of infection (7, 9, 16) and in humans (5). Although MAR-null mutants are avirulent, they remain competent for growth outside the host. We therefore reasoned that MAR proteins could be inactivated with small molecule nonantibacterial inhibitors which curtail virulence without inhibition of growth and therefore without selective pressure for resistance, as do antibiotics.
Our initial efforts to develop small molecule inhibitors of MAR proteins began with a structure-based drug design approach. Members of the MAR (AraC) family of bacterial transcription regulators contain two conserved helix-turn-helix DNA-binding domains (14). Published crystal structures of the DNA-binding domains from E. coli MAR proteins MarA and Rob (26, 35) were used as “active-site” templates in computer-aided small molecule docking studies. These experiments identified compounds of the benzimidazole class which “docked” at the DNA-binding domain. These were chosen for subsequent structure-activity relationship (SAR)-directed medicinal chemistry efforts. Prior studies had demonstrated that the MarA, Rob, and SoxS proteins, which have overlapping roles in the regulation of resistance to multiple antibiotics, oxidative stress agents, and organic solvents (2), are also required for full E. coli virulence in a murine model of ascending pyelonephritis (7). MAR inhibitors targeting MarA, Rob, and SoxS in vitro dramatically reduced infection of the kidney in the ascending pyelonephritis model (6). Based on the above-mentioned findings, we sought to extend the development of MAR protein inhibitors to infections caused by Yersinia spp.
The Y. pestis and Y. pseudotuberculosis MAR protein LcrF (called VirF for Y. enterocolitica) regulates expression of a well-characterized virulence determinant, the type III secretion system (T3SS) (10). The T3SS consists of a needle-like apparatus that delivers a group of effector proteins (Yersinia outer proteins, or Yops) directly into the cytoplasm of host cells. The Yops disrupt host cell signal transduction pathways, including the regulation of phagocytosis, immune system signaling, and apoptosis. Expression of the T3SS is required for full Yersinia virulence in in vitro and in vivo infection models (10, 27, 28, 30). Flashner and colleagues have shown that disruption of the Y. pestis lcrF gene by transposon insertion causes severe attenuation of the organism in a mouse model of septic infection (16). The MAR protein LcrF plays a critical role in Y. pestis virulence and therefore could be a valid target for antivirulence therapeutics. The amino acid sequences of the Y. pestis and Y. pseudotuberculosis LcrF proteins are identical (GenBank). A mouse model of Y. pseudotuberculosis pneumonia which reproduces several aspects of the lethal pneumonia caused by Y. pestis, including the requirement of the T3SS for full virulence, has been developed (15).
Recently, we reported the chemical synthesis and initial structure-activity relationship studies of small molecule inhibitors of the Yersinia MAR protein LcrF (25). Here, we describe the screening of additional compounds and characterization of LcrF inhibitors in in vitro and in vivo infection models. Two LcrF inhibitors identified in this study demonstrated significant inhibition of Y. pseudotuberculosis virulence both in cell infection assays in vitro and in a mouse model of Y. pseudotuberculosis pneumonia.
MATERIALS AND METHODS
Bacterial strains and strain construction. Y. pseudotuberculosis wild-type (WT) strains YPIIIpIB1 (28) and IP2666pIB1 (34) and mutant strain YPIIIpIB1− (4) were used in this study. Bacteria were cultured in 2× YT broth (Invitrogen). To generate the ΔlcrF strains, the following primers were used to amplify DNA flanking the lcrF gene by PCR: LcrF11, 5′-GTGTGAGTCGACATGCCAGCTCAGCC-3′; LcrF22, 5′-GACAGTGCATGCAGTTGGTGAGTTAT-3′; LcrF3, 5′-CCAACTGCATGCACTGTCACAGGCTA-3′; and LcrF4, 5′-CTGTGAGAGCTCCACCTTGTTTATCGGCAACA-3′. The PCR products from the primer pair comprising LcrF11 and LcrF22 and the pair comprising LcrF3 and LcrF4 were used as template DNA in a second round of PCR with primers LcrF11 and LcrF4. The PCR product was cloned into the suicide vector pCVD442 (13) at the SacI and SalI restriction sites. The resulting pCVD442-ΔlcrF plasmid was isolated in E. coli strain SM10 λpir and introduced into Y. pseudotuberculosis strains YPIIIpIB1 and IP2666pIB1 by conjugation as previously described (27). The genotypes of the ΔlcrF strains were confirmed by PCR.
To complement the YPIIIpIB1ΔlcrF strain, the lcrF gene was amplified by PCR from strain YPIIIpIB1 using primers LcrFF2 (5′-CGAGCAGGTACCATGGCATCACTAGAGATTATTAAAT-3′) and LcrFR1 (5′-CGAGTCGGATCCTTAGCCTGTGGTTGCTATTTTAGTAAG-3′). The PCR product was cloned into the KpnI and BamHI sites of pTrc99A (Amersham), adding a sequence encoding 7 additional amino acids to the N terminus of the lcrF open reading frame. The resulting pTrc99A-LcrF plasmid was used to transform the YPIIIpIB1ΔlcrF strain by electroporation. The YPIIIpIB1 and YPIIIpIB1ΔlcrF strains were also transformed with the empty pTrc99A plasmid as controls.
LcrF protein expression and purification.The lcrF gene was amplified by PCR from Y. pseudotuberculosis strain YPIIIpIB1 by use of primers LcrFF1 (5′-CGAGCAGGATCCGATGGCATCACTAGAGATTATTAAAT-3′) and LcrFR1. The PCR product was cloned into the BamHI site of pET-15b (Novagen), adding a sequence encoding an amino-terminal 6-histidine tag to the lcrF open reading frame. The resulting plasmid was used to transform E. coli strain Rosetta/BL21(DE3) (Novagen), and protein overexpression was induced from a lac promoter by growth in 0.1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). Pelleted cells were resuspended on ice in 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA with Complete Mini EDTA-free protease inhibitor cocktail (Roche) and 5 mM β-mercaptoethanol. N-Lauroylsarcosine was added at 1.5%, and the suspension was sonicated on ice for 5 cycles of 30 s at 40 W. Triton X-100 was added at 3.0%, and the sample was rocked for 15 min. The sample was centrifuged at 34,500 ×g for 20 min at 4°C, passed through a 0.2-μm filter, and loaded onto a NiSO4 preloaded Hi-Trap HP chelating column (GE Healthcare). The 6His-LcrF protein was eluted with a 0 to 500 mM imidazole gradient with 10% glycerol and transferred to 12,000- to 14,000-molecular-weight-cutoff (MWCO) dialysis tubing (Spectrum Laboratories). The sample was dialyzed overnight at 4°C in 20 mM Tris-HCl (pH 8.0), 200 mM NaCl, 10% glycerol, and 1 mM dithiothreitol (DTT). The 6His-LcrF protein was concentrated using a Vivaspin column (Sigma-Aldrich), and the final protein concentration was determined using a standard Bradford assay.
Cell-free DNA-binding assay.Compounds were tested for inhibition of 6His-LcrF protein binding to virC promoter DNA in a cell-free assay, as previously described (6, 25). Briefly, 6His-LcrF protein was incubated with titrated concentrations of test compounds or vehicle and the annealed DNA oligonucleotides VirCpF (5′-biotin-CTAAAATAGCAACCACAGGCTAAAATTATCTGTTTTTT-3′) and VirCpR (5′-AAAAAACAGATAATTTTAGCCTGTGGTTGCTATTTTAG-3′) in a streptavidin-coated microtiter plate. Following a washing step, bound 6His-LcrF protein was detected by labeling with a primary anti-5His antibody and a secondary horseradish peroxidase (HRP)-conjugated antibody. LumiGLO (Cell Signaling Technology) HRP substrate was added, and luminescence was measured with a Victor V plate reader. LcrF-DNA-binding inhibition data were plotted with Microsoft Excel, and the XLfit (IDBS) program was used to fit the inhibition curves and determine the compound concentrations that inhibited LcrF-DNA binding by 50% (IC50s).
Western blot analyses. Y. pseudotuberculosis bacteria were cultured under T3SS-inducing conditions (25) with 100 μg/ml ampicillin to maintain the plasmid and up to 100 μM IPTG. Cultures were centrifuged and separated into pellet and supernatant fractions. Samples were combined with sample buffer, separated by SDS-PAGE, transferred to polyvinylidene fluoride (PVDF) membranes, and probed with antisera to YopE and S2. Rabbit antisera to the Y. pseudotuberculosis YopE protein and rabbit antisera to the E. coli ribosomal subunit S2 protein were used at a 1:10,000 dilution (12). A goat anti-rabbit horseradish peroxidase secondary antibody (Cell Signaling Technology) was used at a 1:2,000 dilution. LumiGLO chemiluminescence reagent (Cell Signaling Technology) was used per the manufacturer's instructions.
Yersinia cytotoxicity assay.Compounds were tested for inhibition of Y. pseudotuberculosis (strain YPIIIpIB1) cytotoxicity toward J774A.1 (ATCC strain TIB-67) macrophage cells as previously described (25, 29). Briefly, YPIIIpIB1 and YPIIIpIB1ΔlcrF bacteria were cultured under LcrF-inducing conditions, washed, and added to J774A.1 cell wells at a multiplicity of infection (MOI) of 5 to 10 bacteria per J774A.1 cell. Test compounds or equal volumes of vehicle (dimethyl sulfoxide [DMSO] with 0.4% ethanolamine) were added at 50 μg/ml to the LcrF induction YPIIIpIB1 cultures and also to the J774A.1 cell wells during infections. Infected plates were centrifuged at 1,000 rpm (148 × g) for 5 min at room temperature and then incubated at 37°C with 5% CO2 for 2 h. Extracellular bacteria were eliminated by overnight incubation (∼18 h) with 50 μg/ml gentamicin. Plates were centrifuged at 1,000 rpm (148 × g) for 5 min at room temperature and supernatants collected. The Cytotox 96 (Promega, Madison, WI) colorimetric assay was used to measure the lactate dehydrogenase (LDH) activity in the supernatants. To express the results as percent wild-type cytotoxicity, the LDH activity in test compound wells was compared to that in wells infected with YPIIIpIB1 treated with vehicle. In additional control wells included in each assay, 50 μg/ml test compound was incubated with J774A.l cells in the absence of bacteria.
Mouse infection model.All animal experiments were conducted at Tufts University according to protocols approved by the Tufts University Institutional Animal Care and Use Committee.
A murine model of Y. pseudotuberculosis pneumonia (15) was used to evaluate the virulence of the IP2666pIB1 (WT) and IP2666pIB1ΔlcrF (ΔLcrF) strains. BALB/c mice (6- to 8-week old females; Taconic Labs) were anesthetized with isoflurane and infected intranasally (i.n.) with titrated numbers of CFU of WT or ΔLcrF Y. pseudotuberculosis bacteria in 40 μl phosphate-buffered saline (PBS). Each inoculum was serially diluted and plated to determine the number of CFU delivered for each experiment. Survival was monitored daily for 30 days. Moribund mice were sacrificed by CO2 asphyxiation and counted as dead.
BALB/c mice showed an intolerance to the drug formulation vehicle in preliminary drug tolerance experiments. CD1 mice were susceptible to Y. pseudotuberculosis infection at infectious doses similar to those for BALB/c mice and tolerated drug dosing without sensitivity. CD1 mice were therefore used for compound efficacy experiments.
CD1 mice (Charles River Laboratories) were dosed subcutaneously (s.c.) with 25 mg/kg of body weight of test compound or vehicle in a volume of 100 μl once at 18 h prior to infection (−18 h), at the time of infection (0 h), and at 6, 24, 48, and 72 h following i.n. infection with 420 to 440 CFU of WT Y. pseudotuberculosis. A control group of mice was treated with vehicle and infected i.n. with similar numbers of ΔLcrF Y. pseudotuberculosis. Four days after infection, the mice were sacrificed by CO2 asphyxiation, and lung tissues were aseptically removed, weighed, and homogenized in PBS with 15% glycerol as previously described (15). Serial dilutions of lung tissue homogenates were plated on L agar with 0.5 μg/ml irgasan to determine bacterial burden in terms of number of CFU per gram of lung tissue.
In compound efficacy experiments monitoring survival, 7- to 8-week-old male CD1 mice were dosed s.c. with 25 mg/kg of test compound or vehicle in a volume of 100 μl once on the day prior to infection (−18 h), twice on the day of infection (0 h and 6 h), and once daily for 8 days following i.n. infection with 115 to 215 CFU of WT Y. pseudotuberculosis. A control group of mice was dosed with vehicle and infected with similar numbers of ΔLcrF Y. pseudotuberculosis. Survival was monitored for 26 days.
Statistical analysis.Differences in survival of infected animals were calculated using a Kaplan-Meier survival analysis log rank test (JMP 5.0.1.2; SAS). Differences in bacterial burden in the lungs of infected animals were calculated using a Kruskal-Wallis one-way analysis of variance (ANOVA) and a chi-square statistic. A P value of <0.05 was considered statistically significant.
RESULTS
LcrF mutant Y. pseudotuberculosis strains are attenuated in tissue culture and animal infection models. Y. pseudotuberculosis lcrF deletion mutants were constructed in Y. pseudotuberculosis wild-type strains YPIIIpIB1 (28) and IP2666pIB1 (34). The lcrF gene was cloned from YPIIIpIB1, and sequencing confirmed that the predicted amino acid sequence of the LcrF protein was identical to that of the C092 and KIM 10 strains of Y. pestis (GenBank accession numbers NP_395183 and NP_857733, respectively). LcrF is known to positively regulate the expression of the T3SS. Expression of a representative type III-secreted Yop protein, YopE, was evaluated for the ΔlcrF strains. Western blot analysis of culture supernatants and pelleted whole bacteria confirmed the lack of YopE protein expression in the YPIIIpIB1ΔlcrF strain (Fig. 1 A, lanes 1 and 2). Induced expression of the lcrF gene on a plasmid complemented the YPIIIpIB1ΔlcrF mutant for YopE expression and secretion (Fig. 1A, lanes 3 to 7).
Expression of LcrF in trans restores Yop secretion and cytotoxicity to the ΔLcrF mutant. (A) Western blot of Y. pseudotuberculosis culture supernatants (S) or whole bacteria pelleted by centrifugation (P) probed with antisera to YopE or S2. The strains are as follows: lanes 1 and 2, YPIIIpIB1ΔlcrF plus pTrc99A; lanes 3 to 7, YPIIIpIB1ΔlcrF plus pTrc99A-LcrF; lane 8, blank; lanes 9 and 10, YPIIIpIB1 plus pTrc99A. The culture IPTG concentrations are as indicated, plus the following: lane 4, 12.5 μM; lane 5, 25 μM; lane 6, 50 μM. (B) Y. pseudotuberculosis YPIIIpIB1ΔlcrF (ΔLcrF) shows reduced cytotoxicity to J774A.1 macrophage cells compared to the level for Y. pseudotuberculosis YPIIIpIB1 (WT), as measured by released lactate dehydrogenase (LDH) activity. Also shown is YPIIIpIB1−, which lacks the entire pIB1 plasmid. Percent cytotoxicity represents the mean LDH activity compared to the mean total LDH activity from uninfected J774A.1 cells lysed with detergent (100% cytotoxicity). Error bars indicate standard deviations. For lysed and uninfected cells, n = 4 wells; for the WT, ΔLcrF, and pIB1− strains, n = 16 wells. (C) Cytotoxicity of the YPIIIpIB1-plus-pTrc99A (WT), YPIIIpIB1ΔlcrF-plus-pTrc99A, and YPIIIpIB1ΔlcrF-plus-pTrc99A-LcrF strains to J774.1 cells ± 100 μM IPTG. Percent cytotoxicity represents the mean LDH activity compared to the mean total LDH activity from uninfected J774A.1 cells lysed with detergent (100% cytotoxicity). Error bars indicate standard deviations. For lysed and uninfected cells, n = 4 wells; for the WT-plus-pTrc99A, ΔLcrF-plus-pTrc99A, and ΔLcrF-plus-pTrc99A-LcrF strains, n = 12 wells.
Wild-type Y. pseudotuberculosis bacteria are cytotoxic to macrophage cells in culture, and this cytotoxicity is dependent on a functional T3SS (30). Infection of J774A.1 macrophage cells with the ΔlcrF mutant caused no detectable difference over the background level seen in uninfected control wells and was similar to what was observed for a mutant lacking the entire pIB1 virulence plasmid which encodes the T3SS (Fig. 1B). Induced expression of the lcrF gene on a plasmid complemented the YPIIIpIB1ΔlcrF mutant for cytotoxicity toward J774A.1 cells (Fig. 1C).
The virulence of the IP2666pIB1ΔlcrF mutant was then tested in a mouse model of Y. pseudotuberculosis lung infection (15). Mice were infected intranasally with increasing numbers of wild-type IP2666pIB1 or IP2666pIB1ΔlcrF bacteria to determine the lethal dose. Mice were monitored daily for survival and signs of morbidity (labored breathing, scruffiness, lethargy, hunched appearance, etc.). Moribund mice were sacrificed immediately, and all remaining mice were sacrificed at 30 days postinfection. The absence of LcrF had a profound effect on the lethality of Y. pseudotuberculosis in the lung infection model. The apparent 50% lethal dose (LD50) of 390 CFU for the ΔlcrF mutant was more than 55-fold higher than the wild-type LD50 of 7 CFU over the 30-day experiment (Fig. 2). These results demonstrate the validity of LcrF as a virulence determinant.
Deletion of the lcrF gene attenuates Y. pseudotuberculosis virulence. (A) Survival of BALB/c mice infected i.n. with the indicated numbers of CFU of wild-type Y. pseudotuberculosis IP2666pIB1 (WT) (n = 4 mice per group). **, P < 0.01 by the Kaplan-Meier survival analysis log rank test for 90 or 400 CFU compared to 1 CFU. (B) Survival of BALB/c mice infected i.n. with the indicated numbers of CFU of Y. pseudotuberculosis IP2666pIB1ΔlcrF (ΔLcrF) (n = 4 mice per group). **, P < 0.01 by the Kaplan-Meier survival analysis log rank test for 2,200 CFU compared to 2 CFU. Curves with open symbols indicate the apparent LD50s.
Identification of small molecule inhibitors of LcrF.We have reported the details of the chemical synthesis of the benzimidazole compounds and the initial exploration of the structure-activity relationship for LcrF inhibition in vitro (25). Compound A and compound B (called compound 14 in reference 25) emerged as leads from in vitro screening assays as follows. In previous work, we established a cell-free assay to evaluate the binding of purified E. coli MAR proteins MarA, Rob, and SoxS to DNA containing promoter sequences from genes known to be regulated by these proteins (6). In a similar manner, the lcrF gene was cloned from Y. pseudotuberculosis and the LcrF protein with an amino-terminal 6-histidine tag (6His-LcrF) was overexpressed in E. coli and purified (see Materials and Methods). The 6His-LcrF protein was used in a cell-free DNA-binding assay with DNA containing LcrF binding site sequences from the virC operon promoter. Two hundred five compounds from our benzimidazole collection were screened for inhibition of LcrF-DNA binding in vitro. The concentration of compound necessary to inhibit LcrF-DNA binding by 50% (IC50) was calculated from the results of testing with serial dilutions of the compound. One hundred nine of the tested compounds, including compounds A and B (results for single, representative assays are shown in Fig. 3), showed a dose-dependent inhibition of LcrF-DNA binding. In repeated assays, testing inhibition of LcrF-DNA binding, compound A showed an IC50 of 18.2 ± 11.2 μM (mean ± standard deviation; n = 3 assays) and compound B showed an IC50 of 5.2 ± 2.2 μM (mean ± standard deviation; n = 16 assays). All compounds that inhibited LcrF-DNA binding were additionally screened for inhibition of SlyA-DNA binding at a concentration of 25 μg/ml in similar cell-free assays (25). SlyA is a non-MAR family transcription factor and serves as a specificity control (37). Compounds that showed inhibition of LcrF-DNA binding and no inhibition of SlyA-DNA binding were studied further. Compounds A and B each showed a SlyA IC50 of >53.8 μM, which indicated no inhibitory activity at the highest compound concentration tested (25 μg/ml). Compounds were next tested for antibacterial activity against a panel of antibiotic susceptible Gram-negative and Gram-positive bacteria in MIC assays using the Clinical and Laboratory Standards Institute (CLSI) methodology. Compounds were also tested for cytotoxicity toward mammalian cells in vitro using African green monkey kidney (COS-1) and Chinese hamster ovary (CHO-K1) cell lines according to standard methods (43). LcrF inhibitory compounds which did not inhibit SlyA-DNA binding and were nonantibacterial and noncytotoxic were pursued further.
Compounds A and B inhibit LcrF binding to the virC promoter in vitro. Percent inhibition of binding of purified 6His-LcrF protein to virC promoter DNA in a cell-free assay. The chemical structure and titration of LcrF inhibition by compound A (A) and compound B (B) are shown. Solid lines represent inhibition curve fits, and dotted lines represent the IC50s calculated by the software program XLfit. Data are from single representative experiments.
LcrF inhibitors reduce Y. pseudotuberculosis virulence in a whole-cell cytotoxicity assay.As demonstrated above, Y. pseudotuberculosis T3SS-mediated cytotoxicity is LcrF dependent and was therefore used as a measure of LcrF activity in live bacteria. Compounds with measurable IC50s in the LcrF-DNA-binding assay were tested for inhibition of Y. pseudotuberculosis cytotoxicity toward J774A.1 macrophages in vitro. Twenty of 104 compounds tested at the screening concentration of 50 μg/ml in the whole-cell cytotoxicity assay inhibited cytotoxicity by 50% or more. The Y. pseudotuberculosis cytotoxicity in the presence of compound A was 32.7% ± 19.4% (mean ± standard deviation; n = 6 assays), and the Y. pseudotuberculosis cytotoxicity in the presence of compound B was 34.6% ± 11.6% (mean ± standard deviation; n = 19 assays) of the cytotoxicity of Y. pseudotuberculosis in the presence of vehicle. These compounds were tested in titration experiments with a maximum concentration of 100 μM (46.5 μg/ml for both compounds), which was slightly lower than the screening concentration. Both compound A and compound B showed a dose-dependent inhibition of Y. pseudotuberculosis cytotoxicity (Fig. 4 A). No cytotoxicity was observed when J774A.1 cells were incubated with compounds A and B in the absence of bacteria (Fig. 4C). The addition of compounds A and B to J774A.1 wells at the ends of Y. pseudotuberculosis infections or to J774A.1 cell lysates showed that compounds A and B did not inhibit the LDH enzymatic activity measured in the assay (Fig. 4B and C). Compounds A and B were tested in CLSI standard assays to determine the MIC for bacterial growth against Y. pseudotuberculosis YPIIIpIB1 as well as antibiotic-susceptible E. coli ATCC 25922 and antibiotic-susceptible S. aureus RN450. Compounds A and B both showed no antibacterial activity, with MICs of >64 μg/ml for all bacterial strains tested. Thus, the decrease in Y. pseudotuberculosis cytotoxicity seen with compounds A and B was not linked to effects on bacterial growth or survival.
Compounds A and B inhibit LcrF-dependent Y. pseudotuberculosis cytotoxicity. (A) Compounds A and B were added at the indicated concentrations to J774A.1 cells during infection with wild-type Y. pseudotuberculosis YPIIIpIB1 (WT). The percent WT cytotoxicity indicates the amount of LDH activity released compared to the level for J774A.1 cells infected with WT Y. pseudotuberculosis and receiving vehicle (Fig. 1). Bars indicate mean percentages of WT cytotoxicity, and error bars indicate standard deviations. For WT- and YPIIIpIB1ΔlcrF (ΔLcrF)-infected wells with vehicle, n = 8. For WT-infected wells with compound A or B, n = 4. (B) J774A.1 cells were infected with WT Y. pseudotuberculosis, and compound A, compound B, or vehicle was added to infected wells 5 min prior to sample collection at the end of the assay (n = 8). (C) Compound A, compound B, or vehicle was incubated with J774A.1 cells in the absence of bacteria for the same duration as that for an infection experiment. Half the wells were then lysed with detergent, and the LDH activity was compared to that of lysed, untreated cells (100% cytotoxicity) (n = 8).
LcrF inhibitors attenuate Y. pseudotuberculosis virulence in vivo.As an initial screen for efficacy in vivo, eight compounds were tested for effects on the bacterial burden in the lungs. Mice were infected intranasally with either the wild-type Y. pseudotuberculosis strain or the ΔlcrF mutant. Mice infected with the wild-type strain were dosed subcutaneously (s.c.) with the test compound or the vehicle control. Animals were sacrificed 4 days following infection, and the bacterial burden in the lungs was determined. There were significant reductions in the numbers of bacteria in the lungs of ΔlcrF mutant-infected, compound A-dosed, and compound B-dosed mice compared to the level for wild type-infected, vehicle-dosed controls (P < 0.05) (Fig. 5). These results suggest that nonantibacterial LcrF inhibitors A and B prevent Y. pseudotuberculosis colonization or survival within the lungs of infected mice.
Compounds A and B reduce the bacterial burden in the lungs in a murine model of Y. pseudotuberculosis pneumonia. Groups of 5 CD1 mice were dosed s.c. with vehicle or 25 mg/kg of compound A (A) or compound B (B) prior to and following i.n. infection with wild-type Y. pseudotuberculosis IP2666pIB1 (WT). Control groups of mice were dosed with vehicle and infected with Y. pseudotuberculosis IP2666pIB1ΔlcrF (ΔLcrF). Mice were sacrificed 4 days following infection, and the bacterial burden was determined in terms of number of CFU/gram lung tissue. Symbols represent output for individual animals, and bars represent the geometric mean for the group. P values were determined by Kruskal-Wallis one-way analysis of variance and a chi-square approximation. **, P < 0.01 for comparison to the WT; *, P < 0.05 for comparison to the WT. In panel B, for ΔLcrF, n = 4, and for all other groups, n = 5. Data represent the outcome from one of three experiments conducted for each compound.
The efficacy of compounds A and B was further evaluated in experiments which monitored the survival of Y. pseudotuberculosis-infected mice. Mice were dosed s.c. with compound A, compound B, or vehicle starting 1 day prior to infection. Subsequently, the mice were infected intranasally with wild-type Y. pseudotuberculosis (strain IP2666pIB1) or Y. pseudotuberculosis IP2666pIB1ΔlcrF. The morbidity and survival of the mice were monitored for 26 days. All mice infected with wild-type bacteria and dosed with vehicle succumbed to infection (Fig. 6). In contrast, all mice infected with the ΔlcrF strain and dosed with vehicle survived and appeared healthy at the end of the experiment. Three out of four mice dosed with compound A or B survived and appeared healthy at the end of the 26-day experiment. There was no statistically significant difference in survival between mice infected with the ΔlcrF mutant and those infected with the wild-type organism and receiving compound A or B. In this model, nonantibacterial LcrF inhibitors A and B significantly (P < 0.05) protected mice from lethal Y. pseudotuberculosis lung infection.
Compounds A and B improve survival in a murine model of Y. pseudotuberculosis pneumonia. Symbols represent the survival of groups of 4 CD1 mice dosed s.c. with 25 mg/kg of the indicated compound or vehicle prior to and following i.n. infection with wild-type Y. pseudotuberculosis IP2666pIB1 (WT). A control group of mice was dosed with vehicle as described above and infected with Y. pseudotuberculosis IP2666pIB1ΔlcrF (ΔLcrF). P values were determined by the Kaplan-Meier survival analysis log rank test. **, P < 0.01 for comparison to the WT; *, P < 0.05 for comparison to the WT. Data represent the outcome from one of two experiments conducted for each compound.
DISCUSSION
LcrF is required for full virulence of Y. pseudotuberculosis in tissue culture and animal models of infection and is therefore a valid target for antivirulence agents. Small molecule compounds that inhibit LcrF-DNA binding at low micromolar concentrations have been identified. At least two of these compounds, compounds A and B, protected infected mammalian cells from Y. pseudotuberculosis-mediated cytotoxicity. Compound A or B reduced bacterial burden in the lungs and protected animals from lethal Y. pseudotuberculosis lung infections. These LcrF inhibitors are nonantibacterial, demonstrating a novel and effective new antivirulence approach for preventing and/or treating Yersinia infections and potentially other infections caused by Gram-negative bacteria. Because of the similarity of Y. pseudotuberculosis and Y. pestis (1, 8), their identical LcrF protein sequences, and the role that LcrF plays in the virulence of these two pathogens, we believe that the LcrF inhibitors identified in this study have the potential to be developed as antiplague agents.
Inhibitors of LcrF or other MAR proteins could be used as prophylactic agents to prevent infection or in conjunction with antibiotics to treat infection. In the case of Yersinia pestis infections, one principal utility would be for biodefense as a prophylactic countermeasure for individuals at risk for exposure. Weaponized Y. pestis strains could potentially be multidrug resistant (MDR), rendering current antibiotics ineffective. Importantly, since MAR inhibitors act independently of antibiotic mechanisms, MAR inhibitors should be effective against current or future antibiotic-resistant bacterial strains. Furthermore, prophylaxis with a MAR inhibitor could spare the use of antibiotics as prophylactic agents.
Although the lead compounds were selected for inhibition of LcrF in vitro and produce an LcrF-null-like phenotype, they may have affected host factors or inhibited other Y. pseudotuberculosis virulence factors. Additional studies will be required to establish that the compounds act uniquely through inhibition of LcrF. It is possible that the compounds inhibited Y. pseudotuberculosis MAR proteins other than LcrF. At least one other Y. pseudotuberculosis MAR protein, YbtA, is known to play a role in virulence through regulation of the expression of the yersiniabactin siderophore (19). For Y. pestis, an agent that could inhibit virulence through LcrF, but also multiple-drug resistance and virulence regulated by MAR proteins MarA47 (39) and RobA (38), could be envisioned. In a previous report, we showed that individual benzimidazole compounds can inhibit DNA binding by MAR proteins from multiple bacterial pathogens (6). Compound B from this study showed inhibition of DNA binding of the P. aeruginosa MAR protein ExsA (25). These results suggest that a broad-spectrum agent could be developed.
Targeting the conserved DNA-binding domain of LcrF was intended not only to increase the likelihood of identifying a broad-spectrum agent but also to minimize the potential for resistance development. In our approach, mutations in LcrF that disrupt inhibitor binding might also be expected to disrupt DNA binding and thus prevent the protein from functioning as a transcription factor, effectively making the strain an LcrF-null mutant. Because the LcrF protein is not required for survival outside a mammalian host, the selection pressure to develop resistance to an LcrF inhibitor should be far less than that for an antibacterial agent.
Small molecule inhibitors of MAR proteins represent a novel strategy for preventing infections, but there are other antivirulence approaches being developed for Yersinia (3). Most prominent among these is the development of small molecule inhibitors of type III secretion (T3S) (11, 23, 24, 32, 33). These inhibitors have been identified using whole-cell screens to detect functional inhibition of T3S. Some of the T3S inhibitors block the expression of the T3SS, possibly by inhibition of LcrF, while others appear to interfere with the secretion mechanism directly. The broad-spectrum potential of the “direct” T3S inhibitors is being established, with some inhibitors showing cross-activity against Salmonella enterica serovar Typhimurium, Shigella flexneri, and Chlamydia trachomatis (20, 31, 40, 42). Other antivirulence approaches include inhibition of individual Yop cytotoxins delivered by the T3SS or production of iron-scavenging siderophores (3). In the pathogen Vibrio cholerae, a small molecule inhibitor of the MAR protein ToxT which targets the dimerization domain of the protein has been identified (21, 36). One advantage of the MAR protein inhibition approach is the potential to inhibit the expression of multiple virulence factors in multiple bacterial species with a single agent.
Coupled with our previous work on MAR inhibitors in E. coli (6), these studies provide further insights into a new paradigm for preventing infections caused by Gram-negative bacteria.
ACKNOWLEDGMENTS
We thank P. Casaz, J. Edmonds, P. J. Donovan, A. Macone, and J. Donatelli for technical assistance. We thank R. Gay Leahy for statistical analysis and critical reading of the manuscript.
This work was funded in part by NIH grant 5R43AI058627-2 (Paratek) (M.N.A.). Support from the NIH was also provided to J.M. and W.S. (AI056068), M.L.F. (AI007422), and C.C. (R25GM066567). This work was funded in part by Paratek Pharmaceuticals, Inc. L.K.G. -R., O.K.K., S.K.T., and S.B.L. are employees of Paratek Pharmaceuticals, Inc. V.J.B., A.K.V., and M.N.A. are former employees of Paratek Pharmaceuticals, Inc. Employees and former employees of Paratek Pharmaceuticals, Inc., may own stock options in the company.
FOOTNOTES
- Received 24 November 2009.
- Returned for modification 20 December 2009.
- Accepted 25 August 2010.
- Copyright © 2010 American Society for Microbiology
