We constructed a plasmid-based expression library of the Yp genome comprised of 15,648 Ec clones and screened the library for strains with reduced OFX susceptibility. A strain (Ec pGEM-OFXr1) selected in the screen exhibiting reduced susceptibility that was confirmed to be plasmid-mediated and transferable to Yp was chosen for further characterization (Figure 1). The plasmid (pGEM-OFXr1) carried by this strain was isolated and the restriction digestion pattern and sequence of its genomic insert were examined. This analysis revealed a 4,158-bp fragment (Yp KIM chromosome coordinates 4,137,482 to 4,141,639) (Figure 2). The 5' and 3' ends of the fragment included the 5' end of y3722 (creA) and the 3' end of y3727 (slt), respectively. The products of creA and slt are annotated as a conserved hypothetical protein and a putative soluble lytic murein transglycosylase, respectively, in the Yp genome database. The center of the fragment encompassed four genes: y3723 (robA, herein referred to as robA_Yp); y3724 (gpmB); y3725; and y3726 (trpR). The products of gpmB and trpR are annotated as a putative phosphoglyceromutase and a putative regulator of tryptophan metabolism genes, respectively. The product of y3725 is annotated as a conserved hypothetical protein. Our in silico search for conserved domains (via CD-Search; please see Availability & requirements for more details) revealed the presence of an NTPase (PRK05074) domain in this protein. The NTPase domain is characteristic of proteins with pyrophosphatase activity 1718. This suggested that y3725 may be involved in nucleoside triphosphate metabolism. Lastly, the predicted product of robA_Yp (RobA_Yp) is annotated as an orthologue of Ec RobA (RobA_Ec), a transcriptional regulator of unclear physiological function and member of the AraC/XylS family 19. Importantly, overexpression of robA_Ec and Enterobacter cloacae robA confers low-level resistance in Ec and E. cloacae, respectively, to a number of unrelated antibiotics 202122. Thus, the analysis of the insert in pGEM-OFXr1 suggested that robA_Yp is responsible for the reduced OFX susceptibility observed in Ec pGEM-OFXr1 and Yp pGEM-OFXr1 (Figure 1). These results validate the utility of our library and suppressor screen approach as a means to identify antibiotic resistance determinant candidates in Yp.

We investigated whether overexpression of robA_Yp alone would reduce OFX susceptibility in Ec and, more importantly, in Yp. To this end, we evaluated the antibiotic susceptibility of Ec pGEM-Rob_Yp and Yp pGEM-Rob_Yp. These test strains carried pGEM-Rob_Yp, a plasmid constructed by inserting the fragment encompassing robA_Yp and its promoter region (identified by using the robA_Ec promoter as reference 23) into the vector pGEM-4Z. The antibiotic susceptibilities of these test strains were compared to that of the corresponding Ec pGEM-4Z and Yp pGEM-4Z control strains. These control and test strains were isogenic, except for the lack of the plasmid-borne robA_Yp, and their growth in ampicillin (AMP)-containing liquid media was indistinguishable from that of their cognate test strains (not shown). A first examination of Ec pGEM-Rob_Yp and Yp pGEM-Rob_Yp indicated that these strains retained the reduced OFX susceptibility phenotype seen in Ec pGEM-OFXr1 and Yp pGEM-OFXr1 on solid media (Figure 1), thus indicating that robA_Yp alone was sufficient to reduce OFX susceptibility. In view of this, we conducted further OFX susceptibility testing in liquid media. In addition, we compared the susceptibility of the test and control strains to two other fluoroquinolones [ciprofloxacin (CIP) and levofloxacin (LVX)], a quinolone (NAL), and antibiotics of other classes, including two tetracyclines [tetracycline (TET) and DOX], four aminoglycosides [STR, GEN, kanamycin (KAN), and apramycin (APR)], and CHL.

The IC₅₀ and MIC₉₉ values determined for the aforementioned antibiotics are shown in Table 1. Comparison of the OFX IC₅₀ and OFX MIC₉₉ values of the test strains and their respective control strains revealed that overexpression of robA_Yp reduced OFX susceptibility in both Yp and Ec. In Yp, robA_Yp overexpression increased OFX IC₅₀ and OFX MIC₉₉ values by 5-fold and 23-fold, respectively. The IC₅₀ and MIC₉₉ values of CIP, LVX, and NAL also increased significantly (3- to 5-fold change range) in Yp pGEM-Rob_Yp compared with Yp pGEM-4Z. The reduced OFX and LVX susceptibility of Yp pGEM-Rob_Yp was also revealed by time-kill experiments described below. In Ec, robA_Yp overexpression produced an increase in the IC₅₀ and MIC₉₉ of the fluoroquinolone antibiotics (2- to 4-fold change range), but had no significant effect (<2-fold change) on NAL susceptibility.

In both Yp and Ec, robA_Yp overexpression also correlated with an increase in the IC₅₀ and MIC₉₉ of the two tetracyclines tested and CHL (2- to 7-fold change range). No substantial impact (<2-fold change) on the MIC₉₉ values of four aminoglycosides tested was detected in Yp upon overexpression of robA_Yp. Interestingly, however, the IC₅₀ values of these aminoglycosides were reproducibly and consistently lower (2- to 3-fold reduction range) in Yp pGEM-Rob_Yp compared with Yp pGEM-4Z. These results indicated that robA_Yp overexpression increased the susceptibility of Yp to aminoglycosides. The hypersensitivity of Yp pGEM-Rob_Yp to aminoglycosides was also observed in time-kill experiments described below. Aminoglycoside hypersensitivity was not observed in Ec pGEM-Rob_Yp. On the contrary, the strain had a modest decrease in the susceptibility to GEN, STR, and KAN (2- to 3-fold change range) relative to Ec pGEM-4Z.

Overall, the phenotypic comparison of the antibiotic susceptibility of pGEM-Rob_Yp-bearing strains and pGEM-4Z-bearing strains clearly demonstrates that robA_Yp overexpression affects antibiotic susceptibility in both Yp and Ec, yet in a noticeably species-specific manner. As discussed below, the effects on antibiotic susceptibility induced by robA_Yp overexpression are likely due to an upregulation of efflux pumps. Thus, the species-specific differences in antibiotic susceptibility are probably produced by species-specific differences in efflux pump upregulation.

The comparative analysis of IC₅₀ and MIC₉₉ values described above indicated that Yp pGEM-Rob_Yp has reduced fluoroquinolone susceptibility and increased aminoglycoside susceptibility compared with Yp pGEM-4Z. To further probe these phenotypes, we examined the killing kinetics of these two strains when exposed to OFX, CIP, STR, and GEN (Figure 3). The profiles of the time-kill curves for OFX and CIP demonstrated that the Yp pGEM-4Z control was more rapidly killed by the fluoroquinolones than Yp pGEM-Rob_Yp. Conversely, the profiles of the time-kill curves for STR and GEN revealed that Yp pGEM-Rob_Yp was more rapidly killed by the aminoglycosides than the Yp pGEM-4Z control. Both strains had comparable growth in the absence of fluoroquinolone or aminoglycoside antibiotics during the time frame of the time-kill assays. The contrasting effects of robA_Yp overexpression on fluoroquinolone- and aminoglycoside-mediated killing are consistent with the results of the comparative analysis of IC₅₀ and MIC₉₉ values (Table 1). The observed aminoglycoside hypersensitivity is somewhat unexpected and contrasts with both the reduced susceptibility observed for all other antibiotics tested and the increased tolerance to organic solvents described below.

We investigated whether robA_Yp overexpression affected the susceptibility of Yp and Ec to O₂^.--generating compounds (paraquat, menadione, and plumbagin), heavy metals (zinc, cobalt, and copper), and organic solvents (n-pentane, n-hexane, cyclohexane, p-xylene, and diphenyl ether). No effect on the susceptibility to O₂^.--generating compounds, cobalt, and copper was observed (not shown). Conversely, overexpression of robA_Yp drastically increased organic solvent tolerance in both Yp and Ec (Figure 4) and reduced the susceptibility of Yp to zinc (not shown). Overexpression of robA_Yp increased the tolerance of Yp to n-hexane and cyclohexane and the tolerance of Ec to cyclohexane and n-pentane. All the Yp and Ec strains were resistant to diphenyl ether and sensitive to p-xylene. Ec was also resistant to n-hexane, a result that is in agreement with previous reports 24. These findings parallel the reduction of organic solvent susceptibility induced by robA_Ec overexpression in Ec 21.

In Ec, the RobA_Ec-induced multidrug resistance and solvent tolerance phenotypes have been shown to be largely dependent on the upregulation of the multidrug efflux pump AcrAB, which belongs to the resistance-nodulation-cell division (RND) superfamily 2526. With this precedent in mind, we compared the expression of genes belonging to 34 drug efflux pumps between Yp pGEM-Rob_Yp and Yp pGEM-4Z using quantitative real-time PCR (qRT-PCR). These pumps were identified using a variety of bioinformatic approaches (see Methods) to compile an extensive list that included most, if not all, putative drug efflux pump systems encoded in the genome of Yp. It is worth mentioning as a reference that there are 37 drug efflux pumps annotated in the Ec genome 2728. Our expression analysis detected transcripts for 33 of the 34 genes investigated and revealed that four efflux pumps were significantly upregulated (≥ 5-fold change) in Yp pGEM-Rob_Yp compared with Yp pGEM-4Z (Table 2). Interestingly, two of these upregulated pumps (y3392-y3393 and y1050-y1049) are Ec AcrAB homologs. The other two upregulated pumps (y2173 and y0010) belong to the major facilitator superfamily (MFS). The transcript level of hasF (y3516), encoding the ortholog of Ec TolC, which is the outer membrane protein channel that partners with Ec AcrAB and other RND and MFS pumps 2930, was drastically upregulated as well (5.8-fold change; not shown).

Inspection of the promoter regions upstream of the upregulated genes in Yp pGEM-Rob_Yp revealed the presence of a putative RobA_Ec binding site in each of these regions (Figure 5). These results suggest that RobA_Yp may act as a positive regulator for the y0010, y1050-y1049, y2173 and y3392-y3393 systems. This possible regulatory scenario is consistent with the upregulation in the expression levels of these pumps induced by robA_Yp overexpression in Yp.

It is likely that the multidrug resistance and solvent tolerance phenotypes induced by robA_Yp overexpression are due, at least in part, to increases in compound extrusion by one or more of the upregulated pumps mentioned above. This idea is supported by the reported observation that the RobA_Ec-induced multidrug resistance and solvent tolerance in Ec is largely dependent on the AcrAB-TolC efflux pump system 2526.

Unless otherwise indicated, all molecular biology and microbiological manipulations were conducted using standard procedures 31 and reagents acquired from New England Biolabs or Sigma-Aldrich. Genomic DNA from the avirulent Yp strain KIM6+ 32 was used for the library. This strain lacks the Lcr virulence plasmid 5 and is excluded from the Select Agent Program (please see Availability & requirements for more details). Genomic DNA was prepared using AquaPure™ Genomic DNA Isolation Kit (Bio-Rad Laboratories) and partially digested with BsrFI, which, on average, cleaves the genome of Yp every ~760 bp. Independent partial digestions were resolved by agarose gel electrophoresis and the fragment populations in the 4,000-bp to 8,000-bp range were purified using QIAquick Gel Extraction Kit (Qiagen). The fragments were ligated to the multicopy plasmid vector pGEM-4Z (Promega) linearized with XmaI and dephosphorylated with calf intestine alkaline phosphatase. Genes inserted into pGEM-4Z can be transcribed from the gpt-lac hybrid promoter located at the 5'-end of the cloning site and, potentially, from their native promoters. Ligations were transformed into Ec DH5α (Invitrogen) and transformants were selected in Luria-Bertani (LB) agar plates containing AMP (100 μg/ml) and 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (40 μg/ml) for blue/white colony screening 31. White colonies were streaked onto the same medium to verify their white phenotype and 15,648 confirmed white clones were independently grown in AMP-containing LB broth in 96-well plates. After plate incubation for culture growth to early stationary phase, aliquots from each of the 96 cultures of each plate were pooled, cells from each pool were harvested, and plasmids from each pool were purified using QIAprep Spin Miniprep Kit (Qiagen). The cultures in the plates were supplemented with glycerol (25%) and this master library was stored at -70°C. Agarose gel electrophoresis analysis confirmed plasmid population heterogeneity and restriction digestion analysis of plasmids from several clones verified insert diversity (not shown). The library provides a theoretical ~760-fold genome coverage. Our in silico BsrFI restriction analysis of the Yp genome revealed the presence of four BsrFI fragments of ≥ 8,000 bp. These fragments, which add up to 39,118 bp, are unlikely to be represented in the library.

The library was replicated using a 96-pin inoculator (Clonemaster™; Immusine Laboratories, Inc.) to inoculate 96-well plates loaded with fresh culture medium (150 μl/well). After plate incubation for culture growth (9 h, 37°C, 200 rpm), the 15,648 cultures were pooled and the cells of the pool were harvested. The pooled cells were resuspended in fresh medium (1/10 × pool's volume) containing AMP (100 μg/ml) and glycerol (25%), and the suspension was aliquoted (1 ml library stock aliquots) and stored at -70°C. Multicopy suppressor gene-containing clones were screened for by plating a 1/100 dilution of a library stock aliquot on LB agar plates containing AMP (100 μg/ml) and OFX at the MIC (0.35 μg/ml). The OFX MIC was determined by plating Ec carrying pGEM-4Z (Ec pGEM-4Z) on LB agar plates containing AMP (100 μg/ml) and OFX at increasing concentrations and defined as the concentration for which no colonies were obseved after plate incubation (37°C, 48 h). Clones identified in the screen were streaked on plates containing AMP (100 μg/ml) and OFX (0.35 μg/ml) to confirm their resistant phenotype. The plasmid from each confirmed clone was isolated and transformed into Ec and Yp. Each transformant was streaked on AMP and OFX-containing plates [LB agar for Ec and tryptose blood agar base (TBA; Difco Laboratories) for Yp] to ascertain whether the resistance was plasmid mediated. The insert of each plasmid conferring resistance was sequenced using M13 forward and reverse universal primers (Invitrogen). The sequences obtained were used as queries in sequence similarity searches against the Yp KIM genome using BLAST (please see Availability & requirements for more details) to determine the genome fragment carried by the plasmid.

The fragment encompassing Yp robA (herein referred to as robA_Yp) and its promoter region was PCR-amplified from plasmid pGEM-OFXr1 (see results) with primers Robfor1 and Robrev1 (Table 3). The PCR product (1030 bp) was cloned into pCR2.1-TOPO (TOPO TA Cloning Kit, Invitrogen) and the fidelity of the insert was verified by DNA sequencing. The insert was recovered from the pCR2.1-TOPO clone as an EcoRI fragment and sub-cloned into the EcoRI site of pGEM-4Z using Ec DH5α as host. A clone with robA_Yp in the same orientation as the lacZ gene of pGEM-4Z was designated pGEM-Rob_Yp. pGEM-Rob_Yp and pGEM-4Z were introduced into avirulent Yp by electroporation as reported earlier 33 to create strains Yp pGEM-Rob_Yp and Yp pGEM-4Z, respectively. Yp strains were grown in heart infusion broth (HIB; Difco Laboratories) and on TBA plates without or with antibiotics as appropriate.

Dose-response experiments were done in triplicate and using 96 well plate-based microdilution assays as reported 1214. Briefly, wells contained 200 μl of broth (LB for Ec, HIB for Yp) inoculated with 10⁴ cfu/ml and supplemented with AMP (100 μg/ml) and a second antimicrobial compound at the concentration indicated below. Antimicrobial compounds were added from stock solutions in water, ethanol, or DMSO. Control cultures lacking the antimicrobial compounds contained water (2%), ethanol (1%), or DMSO (0.5%). After incubation (37°C, 200 rpm, 24 h for Yp and 16 h for Ec), growth was measured as optical density (A₆₂₀) using a Spectra Max Plus spectrophotometer plate reader (Molecular Dynamics). IC₅₀ values were calculated from sigmoidal curves fitted to triplicate sets of dose-response data using KaleidaGraph (Synergy Software). MIC₉₉ values were calculated as the lowest concentration tested that inhibited growth by ≥ 99%. The range of concentrations tested were: OFX (Sigma), 2.5-0.001 μg/ml; KAN (Shelton Scientific), 50-0.024 μg/ml; CHL (Calbiochem), 10-0.005 μg/ml for Yp and 25-0.012 μg/ml for Ec; TET (Sigma), 40-0.020 μg/ml; APR (Sigma), 50-0.024 μg/ml; NAL (Sigma), 25-0.012 μg/ml for Yp and 400-0.195 μg/ml for Ec; STR (Sigma), 50-0.024 μg/ml for Yp and 100-0.049 μg/ml for Ec; GEN (EM Science), 25-0.012 μg/ml for Yp and 50-0.024 μg/ml for Ec; DOX (Sigma), 10-0.005 μg/ml; CIP (Fluka), 0.75-0.0004 μg/ml for Yp and 1.25-0.0006 μg/ml for Ec; LVX (Fluka), 0.75-0.0004 μg/ml for Yp and 1.25-0.0006 μg/ml for Ec; plumbagin (Sigma), 50-0.024 μg/ml for Yp and 200-0.098 μg/ml for Ec; menadione (Sigma), 50-0.024 μg/ml for Yp and 400-0.195 μg/ml for Ec; paraquat (Sigma), 200-0.098 μg/ml for Yp and 400-0.195 μg/ml for Ec; CoCl₂ (Sigma), 1-0.0005 mg/ml; CuSO₄ (Sigma), 2-0.001 mg/ml for Yp and 4-0.002 mg/ml for Ec; ZnCl₂ (Sigma), 1-0.0005 mg/ml.

The test for solvent tolerance was conducted essentially as reported previously 21. Overnight cultures of Ec and Yp strains grown in LB broth and HIB, respectively, were inoculated (1%) into fresh media and allowed to grow to A₆₂₀ = 0.4. Then, 5 μl of each culture were spotted on solid medium (LB agar for Ec, TBA for Yp) with 100 μg/ml AMP for transformants carrying pGEM plasmids or without antibiotic for other strains. The surface of the medium was then overlaid with the organic solvent (7 ml) to a thickness of ~3 mm. The plates were sealed and incubated for 24 h for Ec strains and 48 h for Yp strains before naked-eye examination for bacterial growth.

Yp pGEM-Rob_Yp and Yp pGEM-4Z (control) were treated with STR, GEN, OFX, or CIP at 5 × MIC₉₉. The MIC₉₉ values were those determined using Yp pGEM-4Z in the dose-response experiments above (5 × MIC₉₉ values: STR, 15.6 μg/ml; GEN, 5.2 μg/ml; OFX, 0.17 μg/ml; CIP, 0.12 μg/ml). For each antibiotic tested, three tubes with 10 ml of preheated (37°C) HIB containing AMP (100 μg/ml) were inoculated with 10 μl of an overnight culture of the corresponding Yp strain and incubated at 37°C with shaking at 200 rpm for 2 h. After incubation, a sample of each culture was taken and cfu/ml were determined by plating serial dilutions on TBA plates containing AMP (100 μg/ml) and enumerating colonies after plate incubation. Immediately after culture sampling, the test antibiotic was added (from stock solutions in water for GEN and STR or stock solutions in DMSO for OFX and CIP) and the cultures were returned to incubation (37°C, 200 rpm). Samples from these cultures were then taken at time points 0 (immediately after antibiotic addition), 30, 60, 90, 120, 180, 240, 300, 360, and 420 min for cfu/ml determination as above. Triplicate control cultures where water or DMSO was added in place of the antibiotic solution were included in the experiments and treated and analyzed in the same way as the antibiotic-treated cultures. The time-kill data were plotted using Kaleidagraph (Synergy software).

Yp and Ec were cultured in HIB and LB broth, respectively. AMP (100 μg/ml) was added to the medium for strains carrying pGEM-Rob_Yp or pGEM-4Z. Cultures were incubated (37°C, 200 rpm) until they reached A₆₂₀ of ~0.5 before RNA was isolated using the RiboPure-Bacteria Kit (Ambion) according to the manufacturer's instructions. RNA was isolated from triplicate cultures and treated with DNase I (Ambion) (4 units, 37°C, 30 min) in DNase I Buffer (Ambion). After the treatment, DNase I was inactivated by adding DNase Inactivation Reagent (Ambion) at 20% of the final volume of RNA treated. The inactivation was allowed to proceed at room temperature for 2 min. The RNA sample was then centrifuged at maximum speed in a microcentrifuge for 1 min to pellet the inactivation reagent. The RNA was then transferred to a new RNase-free microcentrifuge tube. cDNA was prepared from each RNA sample using TaqMan^® Reverse Transcription Reagents Kit (Applied Biosystems) according to the manufacturer's instructions. Each cDNA sample was analyzed in triplicate by qRT-PCR using SYBR^® Green Master Mix (Applied Biosystems) according to the manufacturer's instructions. cDNA was kept undiluted for qRT-PCR analysis of robA cDNA and diluted 1:5 for analysis of other cDNAs. qRT-PCR and target sequence relative quantification were carried out using a 384-multiwell platform with an ABI-PRISM 7900 HT Sequence Detection System (Applied Biosystems) as described previously 3435. The thermocycling program included 1 cycle of 95°C for 5 min followed by 40 cycles of 95°C for 30 sec, 55°C for 30 sec, and 72°C for 1 min. Relative quantification was conducted using the standard equation 2^-ΔΔCT {i.e. 2^{- [(CT of target cDNA in sample1-CT of 16S rRNA cDNA in sample 1)-(CT of target cDNA in sample 2-CT of 16S rRNA cDNA in sample 2)]}} 36. The equation expresses n-fold difference of the target cDNA in the sample from the strain carrying pGEM-Rob_Yp (sample 1) relative to the target cDNA in the sample from the strain carrying pGEM-4Z (sample 2, control) with normalization to an endogenous control (16S rRNA cDNA). The cycle threshold (CT) values utilized in the equation were the average of three independent cultures of the same strain, each analyzed in triplicate by qRT-PCR.

Searches for efflux pumps were conducted in the sequenced genomes of Yp KIM and and Yp CO92 (please see Availability & requirements for more details). The search included the following strategies. First, the tables of functional classes in the genome websites were examined for annotated pumps. Second, the navigator function in the Artemis genome viewer software (please see Availability & requirements for more details) was used to search for the terms multidrug, efflux, transport, translocase, pump, and drug resistance as qualifiers in the annotated genome sequences. Third, the names of annotated Ec multidrug efflux pumps were used as search keywords using the navigator function in Artemis to find potential pumps not yet identified with the other strategies. Fourth, all putative pumps identified in Yp KIM were used as queries in BLASTP-based searches against the Yp CO92 genome and vice versa. Fifth, annotated Ec multidrug efflux pumps were used as queries in BLASTP-based searches against the Yp KIM and Yp CO92 genomes. Potential RobA_Ec binding sites were searched for using the navigator function in Artemis and the naked eye.