The bacterial strains in this study are listed in Table 1. Y. pestis strain 201 (F1⁺, VW⁺, Pst⁺, and Pgm⁺) isolated from Microtus brandti in Inner Mongolia, China, belongs to biovar Microtus. This strain is avirulent to humans but highly lethal to mice (Zhou et al., 2004). Strain 201 has gene content that is almost identical to that of Y. pestis strain 91001, which possesses four plasmids, namely, pCD1, pMT1, pPCP1, and pCRY1 (Song et al., 2004). All the combinations of plasmid(s)-cured strains derived from strain 201 were constructed by Bin et al. based on the plasmid incompatibility in our laboratory (Ni et al., 2008). E. coli and Y. pestis were grown to exponential phase on LB (Luria–Bertani) agar with 0.1% arabinose at 37°C or 26°C. Approximate concentrations of antibiotics (100 μg/mL ampicillin and 34 μg/mL chloramphenicol) were added to LB agar throughout this study.

Plasmid pXG10-SF, which is an improved gfp-based translational fusion vector with the constitutive PLtetO promoter (Corcoran et al., 2012), was used to construct the GFP reporter fusion to ompF. An amplicon containing the 5′ UTR of the AUG start codon (92 nt) and 16 amino acids of OmpF was obtained by using the primer pair ompF-F/R. The resulting fragment was fused to sfGFP by inserting it into the NsiI/NheI-digested pXG10-SF, which yielded the plasmid pXG-ompF::gfp.

QuikChange® Lightning Site-Directed Mutagenesis Kit (Stratagene) was used to construct an inducible transcriptional fusion vector by modifying an expression vector, pBAD/HisA. According to the manufacturer's instructions, the pBAD/HisA plasmid was amplified with pBAD-F/R primer pair by PCR followed by DpnI treatment. The resulting plasmid pBAD-TF removed the 327-nt fragment containing several elements (RBS, AUG, and polyclonal sites) and introduced EcoRI and HindIII restriction sites upstream of the rrnB terminator sequence. The modified vector pBAD-TF was used as pBAD control. To construct the MicF overexpression vector, a DNA fragment spanning full-length (85 nt) and 58 nt downstream of MicF was ligated to modified pBAD vector and pBAD-MicF was obtained.

The similar protocol was also used to construct a control plasmid designated as pXG-1 with sfGFP expression controlled by PtetO by modifying pXG10-SF. The pXG10-SF plasmid was amplified with primer pair pXG-1-F/R. The resulting plasmid removed the 725-nt lacZ-containing fragment between AatII and BheI and introduced RBS and AUG sequence (GAGGGGAAAUCUGAUG) upstream of lacZ from the transcriptional fusion vector pRW50 into the same site.

Images of bacteria expressing plasmid-borne gfp fusions and grown overnight on LB plates were taken using a CCD camera in Gel Doc XR⁺ image analyzer (Bio-Rad) under the SYBR Green mode. The aliquots of each strain were scraped from the plates and resuspended into PBS buffer. Cell densities were adjusted to OD_620nm ≈ 1.5 and 200 μL of bacterial suspensions were placed into 96-well microtiter plates. Two independent cultures as biological replicates and three aliquots as technical replicates were used throughout the study for each strain. Green fluorescence images were captured by SpectraMax M2 MicroplateReaders (Molecular Devices) with an excitation/emission wavelength of 485/525 nm. Fold changes in the MicF-mediated OmpF expression were calculated by dividing the specific fluorescence of strains with MicF overexpression by that of strains with negative control plasmid. The data were analyzed using One-Way analysis of variance (ANOVA) and Sidak's multiple comparisons test, where P-values of < 0.01 were considered significant.

Y. pestis were grown in LB agar plate at 26°C for 36 h. Equal amounts of bacterial aliquots were collected and lysed by ultrasonication. Total cellular proteins were separated on SDS-PAGE and immunoblotted with anti-OmpF multiclonal antibody and DyLight 680–labeled goat anti-rabbits antibody followed by detection using the Odyssey Infrared Imaging System. The abundance values were calculated as the expression level of the derivatives divided by that of strain 201 using the Quantity One software. The GroEL protein was detected in parallel as control.

Total RNA was then extracted from various bacterial strains grown in LB agar plate in the presence of 0.1% arabinose at 26°C for 36 h using the TRIzol Reagent (Invitrogen). Total RNA samples (1 μg) were denatured at 70°C for 5 min, separated on 6% polyacrylamide 7 M urea gel, and transferred onto Hybond N+ membranes (GE) via electroblot. The membranes were UV-crosslinked and pre-hybridized for 1 h. Northern hybridization was performed by adding the DIG-labeled MicF-specific RNA probe synthesized by in vitro transcription using T7 RNA polymerase. The RNAs were immunologically detected according to the instructions on the DIG Northern Starter Kit (Roche). Band intensities on the Northern blots were quantified by Quantity One software. The 5s rRNA species was monitored in parallel as control.

Y. pestis-specific ompF-gfp fusion and sRNA MicF were cloned into low- and high-copy vectors, respectively. Both plasmids were transformed into E. coli and Y. pestis. All the plasmids were checked throughout all the tested strains by PCR. The results showed that all the plasmids (pMT1, pCD1, pPCP1, and MicF-expressing plasmid pBAD-MicF) were present as expected (Figure S1). The reporter GFP fluorescence activity was monitored to evaluate the regulatory roles in various bacterial strains.

The inhibitory translation of ompF-gfp fusion upon MicF overexpression (approximately five-fold repression) was observed in E. coli strain MG1655. Only the 1.8-fold repression was observed in the Y. pestis strain 201 that is cured of all the endogenous plasmids (Figure 1). This finding is consistent with the results previously reported on E. coli (Mizuno et al., 1984) and also confirms that MicF-mediated ompF repression occurs in the 5′-UTR. Interestingly, the inhibitory effect was not found in the Y. pestis WT strain 201. Instead, more than three-fold upregulation of OmpF-GFP expression was observed under the same conditions (Figure 1). The relative quantification of fluorescence value was also measured in Y. pestis strains with different combinatorial plasmids grown to exponential phase in LB medium followed by arabinose induction for 1 h. Similar tendency was found in Y. pestis strains grown in liquid medium as that found in solid medium (Figures 1, 2 and Figure S2). To exclude the possibility that the effect was caused by the translational reporter system, we construct plasmid pXG-1 which sfGFP expression is constitutively P_LtetO-controlled. No changes in fluorescence intensity were found between pBAD-MicF and pBAD-vector groups in Y. pestis strain 201 and 201-null carrying pXG-1 (Figure S3). The different regulatory consequences in Y. pestis strains carrying or cured of plasmids indicated that the virulence-associated plasmids might have been recruited to sRNA-mediated regulatory networks in the chromosome during evolution.

Y. pestis strains carrying different combinations of the three plasmids (pMT1, pPCP1, and pCD1) were used to determine the probable plasmid(s) responsible for the abolished MicF-mediated regulation of ompF. Upon MicF overexpression, ompF translation remained repressed in the pCD1- or pMT1-containing strains, as observed in the plasmid-null strain. However, the expression level of OmpF-GFP fusion protein was even slightly upregulated about 1.6-fold in the Y. pestis strain carrying plasmid pPCP1. Such activation was also observed in the strain carrying both plasmids pPCP1 and pCD1 (Figure 2). Plasmid pPCP1 alone could overwhelm the translational repression in the plasmid-cured strain, thereby suggesting that pPCP1 may mainly contribute to the abolishment of sRNA-mediated regulation. Paradoxically, translational repression was still found in the strain with pPCP1 and pMT1. We speculated that addition of pMT1 might interfere with the effect of pPCP1 on sRNA-mediated regulation because of the potential interactions between these two plasmids.

We also monitored MicF in the different bacterial strains under the same conditions as those shown in Figures 1, 2. No MicF were found expressed in the pBAD control of various Y. pestis strains by using Northern Blot, which might be mainly due to huge disparity in copy number of MicF between plasmid pBAD and chromosome and/or low expression levels of endogeneous MicF under our experimental conditions. Additionally, no significant differences in MicF overexpression were found among Y. pestis strains (Figure 3).

To test whether the intrinsic plasmids interfere with the MicF-mediated regulation of the chromosome-encoded target gene ompF, the abundances of endogenous ompF transcript and OmpF protein were validated in various strains of Y. pestis by Northern Blot and Western Blot, respectively (Figure 4). In agreement with the results of gene fusion reporter systems, the abundance of ompF transcript or OmpF protein were found decreased 2.0–5.0 fold in four pPCP1-cured strains (122-null, 122-pCD1⁺pMT1⁺, 201-pCD1⁺, and 201-pMT1⁺) upon MicF overexpression relative to that in the control strains. Strikingly, the ompF transcript was stable, but approximate three-fold decrease was found in the expression level of OmpF in the strain 122-pMT1⁺pPCP1⁺, which is also consistent with the findings presented in Figure 2. However, only the comparable levels of ompF and OmpF were found among three groups of pPCP-containing strains (122, 122-pPCP1, and 122-pCD1⁺pPCP1⁺). The downregulation phenomenon is roughly consistent with that of plasmid-borne gfp fusion. The discrepancy is likely due to the different sensitivity between translational fusion assay and Western blot. However, no obvious upregulation was found in strain 201. Maybe the actual activation was magnified by translational fusion assay or veiled by detection threshold of Western blot or Northern Blot. Taken together, this observation further confirmed the conclusion that intrinsic plasmids have the potential impacts on abolishment of MicF-mediated ompF regulation in Y. pestis.

Our study demonstrated that the mobile elements affect sRNA-mediated regulation in Y. pestis. Distinct conclusions may be drawn if various strains carrying different plasmids are used to investigate the sRNA-mediated regulation. For example, Hfq reportedly represses the biofilm formation in Y. pestis KIM6+ (an avirulent derivative of the fully virulent strain KIM, which was cured of the pCD1 plasmid) grown in BHI medium (Bellows et al., 2012). Opposite effects on biofilm formation were observed in the hfq mutant of Y. pestis wild-type strain 201 and its derivative strain lacked the plasmid pCD1 (unpublished data). Therefore, the background bacteria to be used as control strain should be carefully selected. The possible roles of plasmids in gene regulation should be considered even if the exact mechanism is not fully understood. Although a relationship exists between the intrinsic plasmids and sRNA-mediated regulation, the presence or absence of any plasmid did not cause the clear-cut effects on MicF-mediated ompF regulation. This phenomenon may indicate that mutual interactions exist among intrinsic plasmids in Y. pestis, and such interactions further influence sRNA-mediated regulation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.