A multicopy suppressor screening approach as a means to identify antibiotic resistance determinant candidates in Yersinia pestis

Background Yersinia pestis is the causative agent of plague and a potential agent of bioterrorism and biowarfare. The plague biothreat and the emergence of multidrug-resistant plague underscore the need to increase our understanding of the intrinsic potential of Y. pestis for developing antimicrobial resistance and to anticipate the mechanisms of resistance that may emerge in Y. pestis. Identification of Y. pestis genes that, when overexpressed, are capable of reducing antibiotic susceptibility is a useful strategy to expose genes that this pathogen may rely upon to evolve antibiotic resistance via a vertical modality. In this study, we explored the use of a multicopy suppressor, Escherichia coli host-based screening approach as a means to expose antibiotic resistance determinant candidates in Y. pestis. Results We constructed a multicopy plasmid-based, Y. pestis genome-wide expression library of nearly 16,000 clones in E. coli and screened the library for suppressors of the antimicrobial activity of ofloxacin, a fluoroquinolone antibiotic. The screen permitted the identification of a transcriptional regulator-encoding gene (robAYp) that increased the MIC99 of ofloxacin by 23-fold when overexpressed from a multicopy plasmid in Y. pestis. Additionally, we found that robAYp overexpression in Y. pestis conferred low-level resistance to many other antibiotics and increased organic solvent tolerance. Overexpression of robAYp also upregulated the expression of several efflux pumps in Y. pestis. Conclusion Our study provides proof of principle for the use of multicopy suppressor screening based on the tractable and easy-to-manipulate E. coli host as a means to identify antibiotic resistance determinant candidates of Y. pestis.


Background
Yersinia pestis (Yp) is one of the most virulent known bacteria [1] and a potential agent of bioterrorism and biowarfare [2,3] included in the Category A of biological agents for public health preparedness against bioterrorism [4]. Yp is the etiologic agent of plague, a disease responsible for millions of human deaths during the history of civilization [5,6]. Cases are reported every year in many parts of the world [7] and the increasing number of worldwide cases has placed plague in the category of re-emerging diseases [8].
Patients with plague need prompt antibiotic treatment or else death may be unavoidable. The aminoglycosides streptomycin (STR) and gentamicin (GEN) are the preferred antibiotics for treatment, but a number of other drugs are also effective [9,10]. Tetracyclines [such as doxycycline (DOX)], chloramphenicol (CHL), or selected sulfonamides are the recommended antibiotics for prophylactic therapy in the event of exposure or high risk of exposure to Yp [2,9,10]. Fluoroquinolones have also been suggested for treatment and prophylaxis and are noted as a chemotherapeutic alternative against strains resistant to the first line anti-plague drugs [2,10].
The threat of bioterrorism-generated plague outbreaks with engineered (multi)drug-resistant Yp strains [2,3] and the documented outbreak of multidrug-resistant plague [11] underscore the need to develop alternative chemotherapeutic solutions to this disease. In line with this view, we are exploring the development of anti-infectives that target the high-affinity iron acquisition system of Yp [12][13][14] and may offer novel therapeutic possibilities [15]. The plague biothreat also underscores the need to increase our understanding of the intrinsic potential of Yp for developing antimicrobial resistance and to anticipate the mechanisms of resistance that may emerge in Yp clinical isolates in the future. With this consideration in mind, we explored herein the use of a multicopy suppressor screening approach as a means to expose antibiotic resistance determinant candidates in Yp. Multicopy suppressor screening has been useful to study potential drug targets or mechanisms of antibiotic resistance in other species [16]. We constructed a multicopy plasmid-based, Yp genome-wide expression library of nearly 15,000 clones in E. coli (Ec), a tractable and easy-to-manipulate surrogate bacterial host, and screened the library for suppressors of the antimicrobial activity of the fluoroquinolone antibiotic ofloxacin (OFX). Noteworthy, fluoroquinolones have been suggested by the Working Group on Civilian Biodefense as alternative drugs in the event of the use of aerosolized Yp as a bioweapon against a civilian population [2]. The screen permitted the identification of a gene that reduced the susceptibility of Yp to fluoroquinolones and other antibiotic classes when overexpressed from a multicopy plasmid. Our study provides proof of principle for the utilization of multicopy suppressor screening using an Ec host as a means to identify antibiotic resistance determinant candidates in Yp.

A multicopy suppressor screen led to the isolation of a Y. pestis genomic fragment involved in ofloxacin resistance
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 plasmidmediated 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 [17,18]. 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 [20][21][22]. 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.

Overexpression of robA Yp affects susceptibility to multiple antibiotics
We investigated whether overexpression of robA Yp alone would reduce OFX susceptibility in Ec and, more impor-tantly, 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 con-trol 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. 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 IC 50 and MIC 99 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 IC 50 and MIC 99 of the two tetracyclines tested and CHL (2-to 7-fold change range). No substantial impact (<2-fold change) on the MIC 99 values of four aminoglycosides tested was detected in Yp upon overexpression of robA Yp . Interestingly, however, the IC 50 val-Reduction of ofloxacin susceptibility conferred by plasmids pGEM-OFXr1 and pGEM-Rob Yp Figure 1 Reduction of ofloxacin susceptibility conferred by plasmids pGEM-OFXr1 and pGEM-Rob Yp . E. coli (Ec) and Y. pestis (Yp) strains were streaked on solid media without or with ofloxacin: 0.35 μg/ml (the concentration used in the screen) for E. coli and 0.15 μg/ml for Y. pestis. Ampicillin (100 μg/ml) was also added to the media for plasmid-carrying strains.
Genetic map of the robA Yp -containing region of the Y. pestis KIM chromosome and inserts of pGEM-OFXr1 and pGEM-Rob Yp ues 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-4Zbearing 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 speciesspecific differences in efflux pump upregulation.

Overexpression of robA Yp in Y. pestis reduces killing by fluoroquinolones but enhances killing by aminoglycosides
The comparative analysis of IC 50 and MIC 99 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 IC 50 and MIC 99 values (Table 1). The observed aminoglycoside hypersensitivity is some- what unexpected and contrasts with both the reduced susceptibility observed for all other antibiotics tested and the increased tolerance to organic solvents described below.

Overexpression of robA Yp increases tolerance to organic solvents
We investigated whether robA Yp overexpression affected the susceptibility of Yp and Ec to O 2 .--generating compounds (paraquat, menadione, and plumbagin), heavy metals (zinc, cobalt, and copper), and organic solvents (npentane, n-hexane, cyclohexane, p-xylene, and diphenyl ether). No effect on the susceptibility to O 2 .--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 agree-ment with previous reports [24]. These findings parallel the reduction of organic solvent susceptibility induced by robA Ec overexpression in Ec [21].

Overexpression of robA Yp in Y. pestis induces changes in efflux pump gene expression
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-nodulationcell division (RND) superfamily [25,26]. 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 [27,28]. Our expression analysis detected transcripts for 33 of the 34 genes investigated and revealed that four efflux pumps Effect of robA Yp overexpression on the rate of Y. pestis killing by fluoroquinolones and aminoglycosides 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 [29,30], 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 [25,26].

Conclusion
The identification of Yp genes that, when overexpressed, are capable of reducing antibiotic susceptibility is a useful strategy to expose genes that this pathogen may rely upon to evolve resistance via a vertical modality. In this study, we explored the use of a multicopy suppressor, Ec hostbased screening approach as a means to identify antibiotic resistance determinant candidates in Yp. To seek proof of principle for this approach, we constructed a multicopy plasmid-based, Yp genome-wide expression library of nearly 16,000 clones in Ec and screened this library for suppressors of the antimicrobial activity of the fluoroquinolone antibiotic OFX. The screen permitted the identification of a gene that, when overexpressed, reduces the susceptibility of Yp not only to OFX, but also to other (fluoro)quinolones, tetracyclines and CHL. This gene (robA Yp ) encodes a putative transcriptional regulator, and our results clearly demonstrate that its overexpression in Yp and Ec confers low-level resistance to multiple antibiotics. Overexpression of robA Yp also increases organic solvent tolerance in both Yp and Ec and reduces the susceptibility of Yp to zinc.
The molecular mechanism by which overexpression of robA Yp leads to a reduction in the susceptibility to antibiotics and other compounds remains to be determined.
Our results indicate that overexpression of robA Yp induces a drastic upregulation in the transcript levels of four of the 34 predicted efflux pump gene systems and of hasF (tolC) in Yp. Increased expression of chromosomally encoded efflux pumps is a known cause of multidrug resistance in many bacteria [30]. Thus, it is likely that the reduction in the susceptibility to antibiotics and other compounds induced by robA Yp overexpression is due, at least in part, to multidrug efflux pump-mediated increases in compound extrusion.
Overall, our findings provide proof of principle for the utilization of an Ec host-based suppressor screen to identify antibiotic resistance determinant candidates in Yp. This methodology will be useful in the identification of genetic determinants involved in target-dependent and target-independent resistance to antimicrobials with known and unknown mechanisms of action. Identification of such genetic determinants will provide first insights to guide further studies to obtain mechanistic information on novel modes of antimicrobial activity and antimicrobial resistance.

Construction of genomic library
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 everỹ 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-3indolyl-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 Potential RobA binding sites in the promoter regions of genes upregulated in robA Yp -overexpressing Y. pestis Figure 5 Potential RobA binding sites in the promoter regions of genes upregulated in robA Yp -overexpressing Y. pestis. The consensus shown is the 20-bp asymmetric marbox consensus sequence determined by Martin et al., 1999 [37]. R = A or G, Y = C or T, W = A or T, and N = A, T, G, C. Column C is the number of bp's in agreement with the 20-bp consensus sequence. The location of each RobA binding site with respect to the first codon of its cognate gene is indicated by the numbers flanking the putative binding site.
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.

Multicopy suppressor screening
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.

Construction of pGEM-Rob Yp and transformation of Y. pestis
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.

MIC 99 and IC 50 determinations
Dose-response experiments were done in triplicate and using 96 well plate-based microdilution assays as reported [12,14].

Organic solvent tolerance assay
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 A 620 = 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.

Time-kill experiments
Yp pGEM-Rob Yp and Yp pGEM-4Z (control) were treated with STR, GEN, OFX, or CIP at 5 × MIC 99 . The MIC 99 values were those determined using Yp pGEM-4Z in the doseresponse experiments above (5 × MIC 99 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).

Isolation of total RNA and qRT-PCR
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 incu-