Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Nov 1.
Published in final edited form as: Mol Microbiol. 2012 Sep 7;86(3):661–674. doi: 10.1111/mmi.12011

Hfq-dependent, coordinate control of cyclic diguanylate synthesis and catabolism in the plague pathogen Yersinia pestis

Lauren E Bellows 1, Benjamin J Koestler 2, Sara M Karaba 1, Christopher M Waters 2, Wyndham W Lathem 1,*
PMCID: PMC3480973  NIHMSID: NIHMS403221  PMID: 22924957

Summary

Yersinia pestis, the cause of the disease plague, forms biofilms to enhance flea-to-mammal transmission. Biofilm formation is dependent on exopolysaccharide synthesis and is controlled by the intracellular levels of the second messenger molecule cyclic diguanylate (c-di-GMP), but the mechanisms by which Y. pestis regulates c-di-GMP synthesis and turnover are not fully understood. Here we show that the small RNA chaperone Hfq contributes to the regulation of c-di-GMP levels and biofilm formation by modulating the abundance of both the c-di-GMP phosphodiesterase HmsP and the diguanylate cyclase HmsT. To do so, Hfq coordinately promotes hmsP mRNA accumulation while simultaneously decreasing the stability of the hmsT transcript. Hfq-dependent regulation of HmsP occurs at the transcriptional level while the regulation of HmsT is post-transcriptional and is localized to the 5' untranslated region/proximal coding sequence of the hmsT transcript. Decoupling HmsP from Hfq-based regulation is sufficient to overcome the effects of Δhfq on c-di-GMP and biofilm formation. We propose that Y. pestis utilizes Hfq to link c-di-GMP levels to environmental conditions and that the disregulation of c-di-GMP turnover in the absence of Hfq may contribute to the severe attenuation of Y. pestis lacking this RNA chaperone in animal models of plague.

Keywords: cyclic di-GMP, small noncoding RNAs, post-transcriptional regulation, flea, bubonic plague

Introduction

Yersinia pestis is the causative agent of plague and has been responsible for more than 200 million deaths over the past two millennia (Stenseth et al., 2008). Rodent populations are the natural reservoirs of Y. pestis, and transmission occurs through infected insect vectors, typically fleas (Perry and Fetherston, 1997). Humans are accidental hosts for Y. pestis and can become infected via the bite from an infected flea, contact with infected tissues, or inhalation of respiratory droplets from another infected individual (Huang et al., 2006). Currently, there is no licensed vaccine for plague available in the United States.

When a flea takes a blood meal from an infected mammal, Y. pestis enters the proventriculus, a valve-like chamber located between the esophagus and the midgut (Darby, 2008; Bobrov et al., 2007). Y. pestis colonizes this chamber as a biofilm, blocking the proventriculus and preventing the passage of blood to the midgut. Biofilm formation by Y. pestis within the flea enhances the regurgitation of bacteria into the dermis of the mammalian host during attempts at feeding and plays a significant role in the long-term maintenance of the plague bacillus in the wild (Wimsatt and Biggins, 2009; Hinnebusch and Erickson, 2008; Hinnebusch et al., 1996; Eisen et al., 2006). The extracellular matrix of Y. pestis biofilms consists of poly-β-1,6-N-acetyl-d-glucosamine exopolysaccharides (EPS) that are produced by the proteins encoded by the structural hmsHFRS genes, found on the 102 kb chromosomal pigmentation (pgm) locus. These proteins are 50–83% similar to the PgaABCD proteins from Escherichia coli and produce the same EPS as E. coli (Bobrov et al., 2008; Perry et al., 1990; Itoh et al., 2005; Wang et al., 2004).

In many bacteria, the second messenger molecule cyclic diguanylate (c-di-GMP) controls the switch between motility and biofilm formation/extracellular matrix production (Hengge, 2009). Although the plague bacillus is non-motile due to mutations in several flagellar genes (Wren, 2003), biofilm formation by Y. pestis is dependent on the intracellular levels of c-di-GMP. Diguanylate cyclases (DGCs), defined by the presence of a GGDEF domain in the primary amino acid sequence of the protein, synthesize c-di-GMP from two molecules of GTP (Yan and Chen, 2010). The genome of Y. pestis encodes 6 GGDEF-containing proteins, however only 2 of these are functional, including HmsT and the protein encoded by the open reading frame y3730 in the KIM strain (YPO0449 in the CO92 isolate) (Bobrov et al., 2011; Sun et al., 2011).

Conversely, c-di-GMP is degraded by c-di-GMP-specific phosphodiesterase (PDE) proteins, characterized by EAL or HD-GYP domains (Yan and Chen, 2010). While Y. pestis carries the genes for 6 EAL-containing proteins, only the PDE HmsP is active (Kirillina et al., 2004; Bobrov et al., 2011; Bobrov et al., 2005; Sun et al., 2011). In vitro, it appears that HmsT serves as the primary DGC required for biofilm formation, while during flea colonization, Y3730 in the KIM6+ strain predominates (Sun et al., 2011). Biofilm formation is also controlled post-translationally by temperature (due to the decreased stability of HmsH, HmsR, and HmsT at temperatures above 30°C), and requires the production of polyamines mediated by SpeA, an arginine decarboxylase, and SpeC, an ornithine decarboxylase (Perry et al., 2004; Patel et al., 2006). Interestingly, HmsP (but not HmsT or Y3730) is required for virulence in mouse models of infection, suggesting that excess c-di-GMP is deleterious during bubonic plague (Sun et al., 2011; Bobrov et al., 2011).

While the balance of c-di-GMP synthesis and catabolism modulates the extent of EPS production, the mechanisms by which HmsP, HmsT, and Y3730/YPO0449 are regulated to control biofilm formation are not well established. Although it was recently reported that the Rcs phosphorelay system acts as a negative regulator of hmsT at the transcriptional level (Sun et al., 2012), it is becoming clear that the processes controlling c-di-GMP levels and biofilm formation involve multiple transcriptional, post-transcriptional, and post-translational events (Sondermann et al., 2012).

In this report we show that the small RNA (sRNA) chaperone Hfq contributes to the repression of c-di-GMP production and biofilm formation in vitro through both the positive transcriptional regulation of hmsP and the negative post-transcriptional control of hmsT. Small, noncoding RNAs typically base-pair with target mRNAs, the interaction of which can result in translation inhibition, RNase E-mediated mRNA degradation, translation stimulation, and transcript stabilization (Waters and Storz, 2009; Gottesman and Storz, 2011). The outcomes of sRNA activity in a bacterial cell are guided by various environmental conditions and changes in growth. sRNAs modulate a range of cellular circuits, from maintaining homeostasis and quorum sensing to the induction of virulence determinants. As an RNA chaperone, Hfq is often required for the activity of sRNAs in that it binds both the sRNA and mRNA, likely stabilizing the RNARNA interaction. In addition, Hfq frequently contributes to proper sRNA expression, protects sRNAs from RNase degradation, and accelerates RNA strand exchange and annealing. In so doing, Hfq may interact with the majority of sRNAs in the cell (Brennan and Link, 2007; Gottesman and Storz, 2011). The loss of Hfq results in a defect in the virulence of multiple pathogens, demonstrating its importance in maintaining proper bacterial cell physiology (Chao and Vogel, 2010). Indeed, our laboratory and others have shown the contribution of Hfq to the virulence of Yersinia species during mammalian infection (Schiano et al., 2010; Geng et al., 2009). Through the disregulation of HmsP and HmsT synthesis, the data presented here suggest that increased levels of c-di-GMP may provide at least a partial explanation for the decreased virulence of Y. pestis during bubonic plague in the absence of Hfq.

Results

Hfq-dependent control of biofilm formation by Y. pestis

Our laboratory has shown that Hfq is required for the full virulence of the gastrointestinal pathogen Y. pseudotuberculosis (Schiano et al., 2010). In studies examining how Hfq affects the physiology of the closely related species Y. pestis, we observed that in overnight cultures of bacteria grown at 26°C in brain-heart infusion (BHI) broth, the Δhfq strain of Y. pestis forms visible aggregates and settles to the bottom of the culture tube to a greater extent than the wild-type. We found this phenomenon to be pgm-dependent, which suggests that the loss of Hfq leads to increased biofilm formation by Y. pestis (data not shown).

To quantitate this observation, we performed a crystal violet binding assay on bacteria cultured in polystyrene dishes at 26°C. Although there is a growth defect of the Hfq mutant at 26°C in broth culture (Fig. S1), we found a statistically significant increase in crystal violet bound to adherent Y. pestis Δhfq cells compared to wild-type (Fig. 1A). To confirm that this phenomenon is due to the absence of Hfq, we complemented the mutation with the promoter and coding sequence (CDS) for hfq in single copy on the chromosome at a neutral site (Δhfq + hfq) and found that this restored crystal violet binding (and growth in broth) to wild-type levels. Although crystal violet binding by the wild-type strain cultured under our conditions is modest, the loss of the pgm locus, even in the absence of Hfq, resulted in minimal bound crystal violet, indicating that the Hfq-mediated increase in biofilm formation in vitro is pgm (and therefore likely HmsHFRS)-dependent (Fig. 1A). As biofilm formation by Y. pestis is regulated in part by temperature, we also performed the same crystal violet binding assays at 21°C and 37°C. We found that at all temperatures, Hfq significantly represses biofilm formation by Y. pestis in a pgm-dependent manner (Fig. S2A and C). These data show that Hfq contributes to the negative regulation of biofilm formation in Y. pestis when cultured in BHI broth, regardless of physiological temperature.

Fig. 1.

Fig. 1

Open in a new tab

The loss of Hfq alters pgm-dependent biofilm formation by Y. pestis in BHI broth. Wild-type, Δhfq, Δhfq + hfq, pgm−, or pgm− Δhfq Y.pestis were cultured overnight in BHI broth (A) or in TMH broth + 0.2% galactose (B) with shaking at 26°C in polystyrene dishes and a crystal violet binding assay was performed on adherent bacteria. *p < 0.001, NS = not significant. Error bars indicate standard deviation of the mean. Data are representative of at least 3 independent experiments.

In contrast to these results, however, Rempe et al. recently reported that the wild-type KIM6+ isolate of Y. pestis forms significantly elevated biofilms compared to the equivalent Hfq mutant when cultured in the defined medium TMH containing 0.2% galactose (Rempe et al., 2012). To address this discrepancy, we performed the same crystal violet assay on Y. pestis CO92 strains cultured in TMH + galactose at 21°C, 26°C, and 37°C. We found that, unlike in BHI, overall crystal violet binding by pgm+ bacteria was elevated and biofilm formation by the Δhfq strain in TMH broth was equivalent to (or just slightly below that of) wild-type bacteria (Fig. 1B and Fig. S2B and D). Considering that the growth of Y. pestis CO92 in TMH eliminates the Hfq-dependent repression of biofilm formation, our data indicate that environmental and/or nutrient conditions significantly affects the role of Hfq in controlling biofilm formation. As BHI is our standard media for cultivation of Y. pestis, all further experiments presented in this report were performed with this media at 26°C as described in the Experimental Procedures.

Increased c-di-GMP levels in the absence of Hfq

Biofilm formation by Y. pestis is regulated by the intracellular concentrations of the second messenger c-di-GMP. This molecule is predicted to interact with the putative glycosyltransferase HmsR to alter EPS synthesis (Forman et al., 2006; Bobrov et al., 2007). As we observed that Hfq influences the extent of biofilm formation by Y. pestis, we hypothesized that the loss of Hfq results in elevated levels of c-di-GMP when bacteria are cultured in BHI. To test this, we cultured wild-type, Δhfq, and Δhfq + hfq Y. pestis CO92 for 6 hours in BHI broth at 26°C and analyzed cell extracts by LC-MS/MS to measure c-di-GMP concentrations as described previously (Bobrov et al., 2011). C-di-GMP levels were compared to those found in ΔhmsP and ΔhmsT Y. pestis mutants grown under the same conditions. Consistent with the crystal violet binding data, under our conditions total c-di-GMP levels in the wild-type strain are generally low (Fig. 2). We determined, however, that the absence of Hfq results in a significant increase in total c-di-GMP levels compared to wild-type bacteria that approach those found in the ΔhmsP strain; complementation of Hfq restores c-di-GMP to wild-type levels (Fig. 2). These data indicate that Hfq participates in the regulation of c-di-GMP levels in a manner that correlates with the extent of biofilm formation.

Fig. 2.

Fig. 2

Open in a new tab

The loss of Hfq results in significantly increased intracellular c-di-GMP levels. Wild-type, Δhfq, Δhfq + hfq, ΔhmsP, and ΔhmsT Y. pestis were cultured for 6 hours in BHI broth at 26°C. Intracellular c-di-GMP was then extracted and measured as described in the Experimental Procedures. *p <0.001. Error bars indicate standard deviation of the mean. Data are representative of 3 independent experiments.

Hfq affects the levels of both HmsP and HmsT proteins

During in vitro growth, 75–80% of intracellular c-di-GMP in Y. pestis is synthesized by the DGC HmsT and is exclusively degraded by the PDE HmsP (Bobrov et al., 2011). As we observed that Hfq influences the concentration of c-di-GMP, we hypothesized that the loss of Hfq results in elevated HmsT and/or decreased HmsP levels to affect c-di-GMP production and/or turnover. To examine this, we replaced the hmsP and hmsT genes with HA-tagged versions by allelic exchange in the wild-type, Δhfq, and Δhfq + hfq strains of Y. pestis CO92. We then cultured bacteria for 6 hours in BHI at 26°C before performing immunoblot analyses on cell lysates with an anti-HA antibody. We found that the absence of Hfq results in 4.0-fold decreased HmsP protein levels compared to wild-type bacteria (Fig. 3, left panel). Conversely, the loss of Hfq leads to 7.1-fold increased HmsT levels compared to wild-type bacteria (Fig. 3, right panel). In both cases, wild-type levels of protein are restored in the Δhfq strain by complementing the mutation. These results implicate Hfq in controlling biofilm formation in Y. pestis by coordinately repressing the production of the DGC responsible for the majority of in vitro c-di-GMP synthesis and increasing the abundance of the PDE required to metabolize c-di-GMP.

Fig. 3.

Fig. 3

Open in a new tab

Hfq coordinately controls HmsP and HmsT protein levels. Wild-type, Δhfq, and Δhfq + hfq Y. pestis carrying HA-tagged versions of the hmsP and hmsT genes were cultured for 6 hours in BHI broth at 26°C and whole cell lysates were analyzed by immunoblot with an anti-HA antibody. RpoA (lower panels) is shown as a loading control. Numbers to the left of the HmsPHA panel indicate molecular weight in kDa. The relative density of each HmsP-HA or HmsT-HA band as compared to wild-type is indicated beneath the RpoA panels. Data are representative of at least 3 independent experiments.

Effects of Hfq on hmsP and hmsT mRNAs

Hfq influences both HmsP and HmsT protein abundance in a manner that is consistent with cdi-GMP concentrations. Therefore, to determine if the loss of Hfq alters the steady-state levels of either hmsP or hmsT mRNA, we cultured wild-type, Δhfq, and Δhfq + hfq Y. pestis for 6 hours in BHI at 26°C, extracted total RNA, and performed quantitative reverse transcription PCR (qRT-PCR) on these transcripts. Relative to wild-type Y. pestis, the loss of Hfq results in significantly increased hmsT transcript levels and significantly decreased hmsP mRNA under these conditions (Fig. 4A).

Fig. 4.

Fig. 4

Open in a new tab

(A) Altered hmsP and hmsT transcript levels in the absence of Hfq. Wild-type, Δhfq, and Δhfq + hfq Y. pestis were cultured with shaking for 6 hours at 26°C and the relative transcript levels of the indicated hms genes were determined by qRT-PCR. Fold change was compared to wild-type bacteria (set at 1) and determined by the ΔΔCt method. *p <0.001. Dashed lines indicate 2-fold increased or decreased change compared to wild-type. (B, C) Half-life of hmsP (B) and hmsT (C) mRNA. Wild-type and Δhfq Y. pestis were cultured as above and at time 0 rifampicin was added to prevent de novo RNA synthesis. Percent remaining mRNA at each time-point was measured by qRT-PCR and compared to time 0. hmsP half-life: 1.66 +/− 0.004 min for wild-type, 1.76 +/− 0.105 min for Δhfq, difference not significant. hmsT half-life: 1.075 +/− 0.045 min for wild-type, 1.65 +/− 0.1 min for Δhfq, p <0.005. Error bars indicate standard deviation of the mean. Data are representative of 3 independent experiments.

One mechanism by which Hfq and sRNAs can affect mRNA levels is by altering the stability of target transcripts, changing the rate by which RNase E gains access to and cleaves the message. To determine if the reduction in hmsP and accumulation of hmsT transcripts are due to changes in mRNA stability, we measured the half-life of these mRNAs. Wild-type and Δhfq bacteria were cultured for 6 hours at 26°C, after which rifampicin was added to inhibit initiation of de novo transcription. Aliquots of the cultures were removed at time 0, after 30 seconds, and then every minute for 5 minutes, RNA was extracted, and relative hmsP or hmsT mRNA levels normalized to the gyrB transcript were determined by qRT-PCR. We found no significant difference in the half-life of the hmsP transcript between the wild-type and Δhfq strains (1.66 +/− 0.004 min for wild-type, 1.76 +/− 0.105 min for Δhfq, Fig. 4B), suggesting that Hfq does not influence hmsP mRNA stability at the post-transcriptional level. On the other hand, we found that the half-life of the hmsT transcript in the wild-type strain is 1.075 +/− 0.045 min, while in the Hfq-deficient strain the half-life of the mRNA is significantly increased to 1.65 +/− 0.1 min (Fig. 4C). In this case, Hfq acts to decrease the stability of the hmsT mRNA, resulting in lowered HmsT protein levels and therefore reduced c-di-GMP production. Notably, the threshold cycle (Ct) values for the gyrB transcript remained unchanged in both the wild-type and Δhfq strains at each time-point in this assay, indicating that Hfq does not affect the stability of the gyrB mRNA under these conditions (data not shown).

Hfq contributes to the regulation of hmsP transcription

To further characterize the mechanism by which Hfq affects HmsP abundance, we asked if the regulatory effects of Hfq are at the transcriptional or post-transcriptional level. Post-transcriptional regulation mediated by Hfq and sRNAs most often occurs within the 5' UTR of the targeted mRNA (Gottesman et al., 2006). Therefore, we isolated total RNA from wild-type Y. pestis cultured in BHI broth at 26°C to map the 5' end of the hmsP mRNA by the method of rapid amplification of cDNA ends (RACE). Sequence analysis of the 5' cloned cDNA ends determined the transcription start site of hmsP is 224 nt upstream of the translation start site (ATG) of the protein (Fig. 5A).

Fig. 5.

Fig. 5

Open in a new tab

Effects of Hfq at the hmsP locus. (A) Coding strand of DNA for the hmsP locus. The transcriptional start site (indicated by the arrow) was determined by 5′ RACE. Predicted −35 and −10 sites are boxed; the start codon is underlined. (B) Wild-type or Δhfq Y. pestis strains with the chromosomally integrated PhmsP-gfp or PtetO-hmsP 5′ UTR-gfp reporter constructs were cultured at 26°C for 6 hours and fold change in fluorescence compared to wild-type (set at 1), normalized to the optical density of the culture, was determined. For the PtetO-hmsP 5′ UTR-gfp reporters, ATc was added at time 0. *p <0.005, NS = not significant. Error bars indicate standard deviation of the mean. Data are representative of 3 independent experiments.

With this information, we designed two green fluorescence protein (GFP)-based reporter constructs to measure Hfq-dependent post-transcriptional or transcriptional + post-transcriptional activity at the hmsP locus. To determine Hfq-dependent post-transcriptional effects, the 5' UTR + 27 nt of the CDS of hmsP were cloned immediately downstream of a modified PtetO promoter that lacks the PtetO 5' UTR, followed by the CDS for gfp (PtetO-hmsP 5' UTR-gfp). This construct was integrated in single copy onto the chromosome of wild-type or Δhfq Y. pestis at the attTn7 site containing a constitutively active tetR gene. This fusion allows for exogenously controlled promoter activity through the addition of anhydrotetracycline (ATc) to the bacteria; therefore any differences in GFP fluorescence can be attributable to post-transcriptional effects mediated by the 5' UTR of hmsP. To determine transcriptional + post-transcriptional effects, 700 nt upstream of the hmsP translational start site + 27 nt of the hmsP CDS were cloned immediately adjacent to the CDS for gfp (PhmsP-gfp) and integrated onto the chromosome of either wild-type or Δhfq Y. pestis at the attTn7 site. In this case, both endogenous promoter activity and any post-transcriptional effects can be measured by changes in fluorescence.

Bacteria were cultured at 26°C for 6 hours, and in the case of the PtetO-hmsP 5' UTR-gfp reporter strains, ATc was added at time 0. GFP fluorescence produced by the strains was measured in a fluorescence microplate reader and normalized to the optical densities of the cultures. We found that the fluorescence signal produced by the PhmsP-gfp reporter was significantly reduced in the Δhfq strain compared to wild-type (Fig. 5B, left). On the other hand, there was no significant difference in fluorescence measured from the PtetO-hmsP 5' UTR-gfp reporter between the strains (Fig. 5B, right). These data show that the differences in Hfq-dependent effects on the hmsP locus are specific to the promoter region, as opposed to the 5' UTR and proximal CDS, and suggest a transcriptional, as opposed to post-transcriptional, regulatory role for the sRNA chaperone.

Hfq contributes to the post-transcriptional regulation of hmsT

We also sought to identify the mechanism by which Hfq contributes to the regulation of hmsT. By using the same RNA extracted for 5' RACE analysis of the hmsP mRNA, we determined the transcription start site of hmsT to be 21 nt upstream of the HmsT translation start site (ATG) of the protein (Fig. 6A). Recently, Sun et al. determined the transcriptional start site of hmsT in the KIM6+ strain of Y. pestis to be 128 nt upstream of the ATG (Sun et al., 2012). To determine if the CO92 isolate also produces hmsT mRNAs that include a 5' UTR of this length, we used the same cDNAs generated for 5' RACE in RT-PCR reactions with primers designed to anneal at the predicted −128, −77 (as an intermediate-length product), or −21 positions of the hmsT transcript. While this reaction generated a product when primed from the −21 position, we were unable to detect PCR products that encompass either the −77 or −128 nucleotides of the predicted transcript (Fig. 6B). Therefore, we conclude that the primary transcriptional start site of hmsT in Y. pestis strain CO92 cultured under our conditions is at the −21 position.

Fig. 6.

Fig. 6

Open in a new tab

Effects of Hfq at the hmsT locus. (A) Coding strand of DNA for the hmsT locus. The transcriptional start site (indicated by the arrow) was determined by 5′ RACE. Predicted −35 and −10 sites are boxed; the start codon is underlined. (B) hmsT 5′ UTR transcript analysis. PCR using genomic DNA (g) or cDNA (c) used for 5′ RACE from wild-type Y. pestis as templates was performed with 5′ primers 128, 77, or 21 nt upstream from the start codon. Transcript is detected only from the −21 site. Numbers to the left of the image indicate nucleotide size markers. (C) Wild-type or Δhfq Y. pestis strains with the chromosomally integrated PhmsT-gfp or PtetO-hmsT 5′ UTR-gfp reporter constructs were cultured at 26°C for 6 hours and fold change in fluorescence compared to wild-type (set at 1), normalized to the optical density of the culture, was determined. For the PtetO-hmsT 5′ UTR-gfp reporters, ATc was added at time 0. *p <0.01, **p <0.005. Error bars indicate standard deviation of the mean. Data are representative of 3 independent experiments.

To ascertain the point at which Hfq contributes to the regulation of hmsT, we generated post-transcriptional (PtetO-hmsT 5' UTR-gfp) and transcriptional + post-transcriptional GFP (PhmsT-gfp) reporter fusion constructs similar to those described for hmsP above. For the PhmsT-gfp reporter, 500 nt upstream of the hmsT translation start site + 27 nt of the hmsT CDS were cloned in front of the gfp CDS. Both reporters were integrated in single copy onto the chromosomes of wild-type or Δhfq Y. pestis, bacteria were cultured for 6 hours at 26°C (ATc was added at time 0 to the strains carrying the PtetO-hmsT 5' UTR-gfp fusion), and fluorescence was measured and normalized to the optical densities of the cultures. For both the PtetO-hmsT 5' UTR-gfp and PhmsT-gfp reporters, fluorescence intensity was significantly higher in the absence of Hfq compared to wild-type (Fig. 6C). These data demonstrate that Hfq post-transcriptionally represses HmsT synthesis through regulatory effects at the hmsT 5' UTR and/or within the first 27 nt of the CDS, and combined with the data presented in Fig. 4C, suggest that this occurs in part through the alteration of hmsT mRNA half-life.

Effects of Hfq on YPO0449-dependent biofilm formation and c-di-GMP synthesis

In addition to HmsT, the DGC encoded by the gene y3730 (YPO0449 in strain CO92) is also active in Y. pestis. To determine if Hfq influences biofilm formation via regulatory effects on YPO0449, we generated mutants of Y. pestis deleted for hmsT or YPO0449 in the wild-type and Δhfq backgrounds and then performed a crystal violet binding assay. As has been reported previously, we found that YPO0449 also contributes to biofilm formation by Y. pestis, although to a lesser extent than does HmsT (Fig. 7A). In the absence of Hfq, however, we observed a significant increase in crystal violet bound by not only ΔYPO0449 bacteria (when HmsT is the only DGC present), but also by ΔhmsT Y. pestis compared to their Hfq+ counterparts. To test if the Hfq-dependent repression of biofilm formation in the absence of HmsT acts via effects on YPO0449, we measured c-di-GMP levels in these same strains. We found that, in contrast to ΔYPO0449 bacteria, the loss of Hfq does not significantly alter c-di-GMP levels in ΔhmsT bacteria compared to Hfq+ Y. pestis (Fig. 7B), suggesting that any Hfq-dependent effects on biofilm formation instead occur independently of YPO0449. Consistent with these data, the loss of Hfq does not affect the steady-state transcript levels of YPO0449 (data not shown).

Fig. 7.

Fig. 7

Open in a new tab

Effects of Hfq on YPO0449-dependent biofilm formation and c-di-GMP synthesis. (A) Wild-type, ΔYPO0449, or ΔhmsT Y. pestis in either the Hfq+ or Hfq− background strains were cultured overnight in BHI broth with shaking at 26°C in polystyrene dishes and a crystal violet binding assay was performed on adherent bacteria. (B) Intracellular c-di-GMP levels were measured as in Fig. 2 in the same strains. *p < 0.01, **p < 0.001, ***p < 0.0005, NS = not significant. Error bars indicate standard deviation of the mean. Data are representative of 2 independent experiments.

Decoupling hmsP expression from Hfq-based regulation reduces c-di-GMP levels and biofilm formation

We have shown that the loss of Hfq leads to increased c-di-GMP levels and biofilm formation by Y. pestis. Consistent with these observations, we have demonstrated that levels of the PDE HmsP are repressed in the absence of Hfq. If these phenomena are linked, then we hypothesized that uncoupling hmsP expression from Hfq-dependent regulation should reverse the regulatory defect and suppress biofilm formation in the Δhfq strain. To test this, we integrated onto the chromosome of either wild-type or the Y. pestis hfq mutant a copy of the hmsP gene and including a 2xHA-tag on the 3' end whose expression is driven by the PtetO promoter via the addition of ATc. We then cultured wild-type, wild-type + PtetO-hmsP-HA, Δhfq, and Δhfq + PtetO-hmsP-HA Y. pestis in the presence or absence of ATc at 26°C and performed a crystal violet binding assay. While the addition of ATc had no effect on biofilm formation in either the wild-type or Δhfq strains lacking the PtetO-hmsP-HA construct, we found that the induction of hmsP expression in the Δhfq + PtetO-hmsP-HA strain significantly reduced the level of crystal violet binding to that observed in wild-type Y. pestis (Fig. 8A). This effect was specific to the addition of ATc, as PtetO-hmsP-HA strains cultured in the absence of ATc formed biofilms to the same extent as their cognate partner strains (Fig. 8A). To confirm that HmsP-HA was produced by these bacteria in the presence of ATc, we generated cell lysates from the same cultures and performed immunoblot analyses with an anti-HA antibody. Consistent with the crystal violet binding assay results, we observed that the addition of ATc induced HmsP synthesis, demonstrating that overproduction of this PDE is sufficient to rescue the effects of the loss of Hfq on biofilm formation (Fig. 8B). To show that this phenomenon is linked to c-di-GMP levels, we determined intracellular c-di-GMP concentrations in these same strains as described above. Induction of hmsP expression results in significantly reduced c-di-GMP levels (Fig. 8C), demonstrating that the regulation of c-di-GMP synthesis and breakdown via Hfq is a major mechanism by which Y. pestis controls biofilm formation.

Fig. 8.

Fig. 8

Open in a new tab

Hfq-decoupled expression of hmsP reverses the impact of the hfq mutant on biofilm formation and c-di-GMP levels. (A) Wild-type, Δhfq, wild-type + PtetO-hmsP-HA, or Δhfq + PtetO-hmsP-HA Y. pestis strains were cultured overnight with shaking at 26°C in polystyrene dishes in the presence or absence of ATc and a crystal violet binding assay was performed on adherent bacteria. (B) Lysates from the wild-type + PtetO-hmsP-HA and Δhfq + PtetO-hmsP-HA strains were analyzed by immunoblot with an anti-HA antibody to determine the abundance of HmsP-HA in the presence or absence of ATc. RpoA is shown as a loading control. (C) Intracellular c-di-GMP levels were measured as in Fig. 2 in the same strains in the presence or absence of ATc. *p <0.05, **p = 0.0005, ***p <0.0001, NS = not significant. Error bars indicate standard deviation of the mean. Data are representative of 2 independent experiments.

Discussion

It is now well established that small, noncoding RNAs are important regulators of gene expression in bacteria, directly influencing protein synthesis at the post-transcriptional level (Gottesman and Storz, 2011). In most cases, trans-encoded sRNAs rely on the protein chaperone Hfq for their expression, stability, and/or function. Thus, the loss of Hfq often leads to pleiotropic effects on the cell, as multiple sRNA-regulated pathways are disrupted, including those required for virulence. In this report we show that Hfq influences intracellular c-di-GMP levels in the plague pathogen Y. pestis, with consequent impact on cellular aggregation and biofilm formation. This effect is primarily due to the coordinate transcriptional activation of the gene encoding the PDE HmsP and post-transcriptional repression of the DGC HmsT (Figs. 5 & 6), resulting in reduced c-di-GMP (Fig. 2) and biofilm synthesis in an HmsHFRS-dependent manner (Fig. 1). Consistent with our data, deletion of hmsP leads to overproduction of biofilms in Y. pestis, while mutation of hmsT abrogates the ability of the plague pathogen to form biofilms (Kirillina et al., 2004).

Unlike the role we observe for Hfq in Y. pestis, Hfq acts to increase biofilm formation in other pathogens such as Salmonella enterica Typhimurium, uropathogenic E. coli, Vibrio cholerae, and Pseudomonas fluorescens (Wu et al., 2010; Kulesus et al., 2008; Hammer and Bassler, 2007; Kint et al., 2010). For instance, in V. cholerae Hfq mediates Qrr sRNA repression of the transcriptional regulator HapR. In the absence of Hfq, HapR levels increase, which results in repression of the vps genes required for biofilm formation (Lenz et al., 2004; Hammer and Bassler, 2003; Jobling and Holmes, 1997). In contrast, our data suggest an alternate and divergent Hfq-dependent mechanism of biofilm formation in the plague bacillus when cultured in BHI. In general, we found that crystal violet binding and c-di-GMP levels in the wild-type CO92 strain of Y. pestis are low compared to that observed by others (Bobrov et al., 2011; Rempe et al., 2012). This effect may be dependent on environmental niche and/or nutrient status, however, as growth of Y. pestis in the defined medium TMH results in relatively higher levels of crystal violet binding by the bacteria compared to BHI and also eliminates the Hfq-dependent repression of biofilm formation (Fig. 1B). While it is not yet known what factor(s) in TMH vs. BHI mediate this switch, one possible explanation may lie in glucose levels, as it has been shown that the replacement of gluconate with glucose in TMH alters bacterial aggregation (Staggs and Perry, 1991). While temperature does not seem to influence the Hfq-dependent repression of biofilm formation in our assays, the differences we observed between BHI and TMH suggest that Hfq instead links environmental sensing with biofilms through nutrient availability, possibly by changes in the expression of one or more sRNAs that mediate the regulatory effect.

Nevertheless, our results indicate that this effect on biofilm formation is a result of the influence of Hfq on c-di-GMP synthesis and breakdown through the modulation of both HmsP and HmsT levels. Indeed, decoupling the synthesis of the PDE HmsP from Hfq is sufficient to reduce biofilm formation and suppress c-di-GMP levels (Fig. 8), demonstrating the role of the sRNA chaperone in manipulating c-di-GMP concentrations when bacteria are cultured in BHI. How, then, does Hfq contribute to the control of both HmsP and HmsT synthesis? In the case of the DGC, Hfq (and presumably one or more sRNAs) acts post-transcriptionally at the 5' UTR/proximal 27 nt within the CDS of the hmsT mRNA and alters the stability of this transcript, decreasing its half-life by approximately 50%. This suggests that the binding of Hfq and sRNAs to the hmsT 5' UTR affects RNase E accessibility, potentially by exposing RNase E cleavage sites and leading to increased mRNA turnover (Aiba, 2007; Brennan and Link, 2007). Our data, however, do not preclude the possibility of additional post-transcriptional effects, for example by decreasing ribosomal access to the transcript. Furthermore, Hfq may also influence the transcription of hmsT indirectly. For instance, in E. coli the synthesis of RpoS is dependent on the Hfq-binding sRNAs RprA and DsrA when the Rcs two-component system is activated (Jones et al., 2006). As Sun et al. recently demonstrated that the RcsCDB phosphorelay positively regulates hmsT transcription (Sun et al., 2011), it is possible that sRNAs could also contribute to the full induction of Rcs-mediated hmsT expression in a similar manner. Although Y. pestis lacks DsrA (Koo et al., 2011), if this regulatory pathway is otherwise conserved in Yersinia, other sRNAs may serve as functional equivalents.

While the effect of Hfq on hmsT may be direct, the control of hmsP by Hfq appears to be indirect, as the absence of Hfq does not result in changes in transcript stability or in regulatory effects specifically within the 5' UTR/proximal CDS region. This suggests that an upstream regulator or other factor that affects the transcription of hmsP is in fact the target of Hfq and sRNAs. For example, the E. coli sRNAs OmrA/B control the translation of the CsgD transcriptional activator, which then regulates the expression of genes involved in curli synthesis. Curli proteins, in turn, participate in biofilm formation, thereby indirectly linking sRNAs to EPS synthesis (Holmqvist et al., 2010). It will be of interest to determine if the same sRNA that regulates hmsT post-transcriptionally also indirectly affects hmsP, or if Y. pestis uses independent but coordinately expressed sRNAs to fine-tune c-di-GMP synthesis. In addition, there may be environmental conditions (e.g., in the flea proventriculus or during mammalian infection) that decouple the Hfq- and sRNA-based regulation of HmsP and HmsT to modulate the synthesis of c-di-GMP to more appropriately match the needs of the cell.

These environmental signals that result in the Hfq-dependent regulation of biofilm formation may also be multi-factorial. Again using V. cholerae as an example, in addition to Qrr-mediated regulation of biofilm formation, the catabolite repressor protein (CRP) negatively regulates vps genes, DGCs, and PDEs in the presence of cAMP, but in a HapR-independent manner (Fong and Yildiz, 2008). Thus, multiple independent regulatory pathways that control the biofilms of Y. pestis likely exist, and include (but are clearly not limited to) Hfq- and sRNA-dependent systems. That HmsT is the DGE responsible for in vitro biofilm formation but Y3730 is predominant in the flea is a striking example of this (Sun et al., 2011). Although we found that Hfq also represses biofilm formation when YPO0449 is the only active DGC present in Y. pestis (Fig. 7), our data indicate that this is not due to direct effects on c-di-GMP levels and thus YPO0449 itself may not be regulated via Hfq. Instead, we postulate that Hfq may contribute to the regulation of the hmsHFRS operon or other factors involved in biofilm production, or alternatively this may be due to a pleiotropic effect of the Hfq deletion on an autoaggregation factor such as an adhesin or other cell surface structure. Nevertheless, as recent data demonstrate that Hfq is required for gut blockage of the flea (Rempe et al., 2012), it will be of interest to examine if the Hfq-dependent regulatory processes we have determined here are also active during flea colonization.

As Hfq typically acts in conjunction with sRNAs, our data suggest that one or more of these non-coding RNA molecules participate in the regulation of HmsP and HmsT. While Yersinia species express at least 216 putative noncoding RNAs at either 26°C or 37°C in broth culture (Koo et al., 2011; Koo et al., 2012), it is not yet known whether the sRNAs involved in controlling c-di-GMP levels are unique to Y. pestis, shared with other Yersinia species, or broadly distributed in other bacterial genera. Nevertheless, disruption of the pathways that lead to the appropriate balance of c-di-GMP levels in an Hfq-dependent manner may partly explain the attenuation of Y. pestis Δhfq in animal models of disease, as both Sun et al. and Bobrov et al. have demonstrated that excess c-di-GMP during plague during Y. pestis colonization attenuates the infection (Bobrov et al., 2011; Sun et al., 2011). This attenuation may occur upon the initial introduction of Y. pestis Δhfq into the mammal (when c-di-GMP levels are higher), or later during the infection, if the loss of Hfq is unable to relay signals from the in vivo environment to modulate DGC and PDE levels. Similar effects have been observed for the essential Yersinia virulence regulator RovA, which is maximally synthesized in vitro at lower temperatures unless the pH of growth medium is lowered to match that found in the small intestine (Ellison et al., 2004; Heroven and Dersch, 2006). In addition, the loss of Hfq has also been shown to affect the flea phase of the Y. pestis lifecycle, interfering with the efficiency of transmission to the mammalian host (Rempe et al., 2012). The identification of the sRNAs involved in these processes will therefore help delineate the pathways and regulatory mechanisms that control this important Y. pestis virulence determinant.

Experimental Procedures

Reagents, bacterial strains, plasmids, and growth conditions

All reagents were obtained from Sigma-Aldrich (St. Louis, MO) unless otherwise indicated. Bacterial strains used in this study are listed in Table S1, plasmids are described in Table S2, and oligonucleotides are listed in Table S3. All Y. pestis strains described herein lack the pCD1 virulence plasmid unless otherwise stated; some strains lack the chromosomal pgm locus and are indicated as such in the text and in Table S1. Y. pestis was routinely cultured in liquid BHI medium (Difco), or on BHI agar at 26°C unless indicated otherwise. For experiments performed in the defined medium TMH (Rempe et al., 2012), Y. pestis strains were passaged once in BHI and then subcultured once in TMH to acclimate the bacteria prior to the initiation of the experiment. Media were supplemented with ampicillin (100 μg ml-1) or kanamycin (50 μg ml-1), as appropriate.

Deletion of hmsP, hmsT, and YPO0449

The genes for hmsP, hmsT, and YPO0449 were deleted from Y. pestis strain CO92 and derivatives (listed in Table S1) by lambda red recombination as previously described (Lathem et al., 2007; Koo et al., 2011). Regions of homology upstream and downstream of the genes were amplified by PCR using the primer sets listed in Table S3. The kanamycin resistance cassette used for the selection of recombinants was excised as described earlier (Lawrenz et al., 2009). When applicable, deletions were generated successively and the kanamycin resistance cassette was excised between each deletion.

Complementation and Tn7-based chromosomal integration

Complementation of the hfq gene deletion in the pCD1- derivative of Y. pestis strain CO92 (referred to as Y. pestis Δhfq in this report) (Koo et al., 2011) was performed by restoring the gene + 492 bases upstream of the coding sequence for hfq onto the chromosome of Y. pestis Δhfq at the glmS-pstS intergenic region via the Tn7 site-specific transposon (Choi et al., 2005). The hfq locus from Y. pestis strain CO92 was amplified by PCR and cloned into the plasmid pUC18R6K-mini-Tn7T-kan. The Tn7 transposon carrying the hfq locus was introduced into Y. pestis Δhfq and the kanamycin cassette was resolved as described previously (Lathem et al., 2007). Construction of ATc-inducible hmsP-HA strains was accomplished as follows: the coding sequence for hmsP was PCR-amplified from the Y. pestis CO92 genome, and the PtetO promoter + 5' UTR was PCR-amplified from the plasmid pWL213 (Lathem et al., 2007). The resulting products were then joined by PCR using the technique of splicing by overlap extension (SOE) and cloned into pWL212, a Tn7-based integration plasmid that contains a constitutively expressed tetR gene (Lathem et al., 2007). This construct was integrated onto the chromosome of Y. pestis and Y. pestis Δhfq as described above and the kanamycin cassette resolved.

Crystal violet assays

Strains of Y. pestis were cultured in triplicate at the indicated temperatures for 16–18 hours in one ml of BHI broth or TMH broth containing hemin (1 μg ml-1) and galactose (0.2%) in a 24-well polystyrene plate at 200 RPM. Planktonic bacteria were removed and crystal violet dye (0.01% in water) was added for 20 min, after which the adherent bacteria were washed twice with distilled water. Bound crystal violet was solubilized with 80% ethanol/20% acetone and absorbance was measured in a spectrophotometer at 570 nm. For experiments measuring the effects of hmsP overexpression on crystal violet binding, ATc (1 μg ml-1) or vehicle was added to the cultures at time 0 before proceeding. Statistical analyses were performed using the students' t test.

Quantitative RT-PCR

Strains of Y. pestis were cultured in triplicate overnight in BHI broth at 26°C in a roller drum before being diluted to an OD620 of 0.1 in 25 ml BHI broth in a 125 ml Erlenmeyer flask. Bacteria were then cultured at 26°C for 6 hours in a shaker set at 250 RPM before aliquots were removed and immediately added to 2 volumes of the RNA-Protect Bacteria reagent (Qiagen). Total RNA was extracted using the RiboPure-Bacteria kit (Life Technologies), treated with DNase (Life Technologies), and cDNA was synthesized with the Superscript II reverse transcriptase (Invitrogen) using random primers (Invitrogen), all according to the manufacturers' instructions. qRT-PCR for the target genes was performed in triplicate with the SYBR green dye in a MyiQ iCycler thermocycler (Bio-Rad) using the primer sets listed in Table S3. The calculated Ct was normalized to the Ct of the gyrB transcript from the same cDNA sample before calculation of fold change using the ΔΔCt method as described (Lathem et al., 2007). Statistical analyses were performed using the students' t test.

Measurement of c-di-GMP levels

Y. pestis strains were cultured in triplicate as described for qRT-PCR in the absence or presence of ATc (1 μg ml-1) or vehicle as necessary. Bacteria (11.25 OD equivalents) were centrifuged at 5,000 × g for 15 min, and the pelleted cells were then mixed with 250 μl of extraction buffer (40% methanol, 40% acetonitrile, 0.1 N formic acid). This slurry was incubated for 30 min at −20°C, and the insoluble material was precipitated by centrifugation at 14,000 × g for 5 min. To measure the intracellular concentrations of c-di-GMP from these lysates, 10 μl of each sample was analyzed using liquid chromatography tandem mass spectrometry (LCMS/MS) as described previously (Bobrov et al., 2011). Concentrations are reported as nmols of c-di-GMP per mg of wet bacterial cell pellet. Statistical analyses were performed using the students' t test.

HA-tagging and immunoblot analysis

The genes encoding HmsP and HmsT were replaced by allelic exchange with variants carrying the sequence for the hemagglutinin (HA) tag on the 3' end immediately preceding the stop codon. Five hundred bases upstream of the stop codon were PCR-amplified with a 2x-HA sequence on the 3' end followed by the stop codon; likewise, 500 or 551 bases downstream of the stop codon were PCR-amplified with a 2x-HA sequence on the 5' end and preceded with the stop codon. The two fragments were then joined by SOE-PCR, cloned into the R6K-dependent, sacB-containing vector pSR47S (Merriam et al., 1997), and the sequences were confirmed. The resulting plasmids were introduced into Y. pestis strains by electroporation and merodiploids were selected by plating on BHI agar containing kanamycin. Merodiploids were then passaged on BHI agar containing 5% sucrose, kanamycin-sensitive colonies were identified, and incorporation of the HA-tag at the appropriate locus was confirmed by PCR. Strains containing the HA-tagged hmsP or hmsT genes were cultured as described for qRT-PCR. Aliquots of the cultures were removed, centrifuged at 5,000 × g for 10 min at 4°C and washed once with PBS, and cells were incubated for 30 min on ice with lysozyme (50 μg ml-1) followed by sonication using 3 × 30 sec pulses. Lysates were centrifuged at 10,000 × g for 10 min at 4°C and protein concentrations were measured by the Bradford assay (Bio-Rad). Equal concentrations of lysates were separated by SDS-PAGE, transferred to nitrocellulose, and analyzed by immunoblot with antibodies to the HA-tag (Roche) or RpoA (as a loading control). For analyses of HmsP-HA levels when produced from the PtetO system, the same cultures used in the crystal violet assays were processed and immunoblotted for with antibodies to HA or RpoA as above.

mRNA half-life determination

Y. pestis strains were cultured in triplicate as described for qRT-PCR. At time 0 (6 hours following sub-culture), rifampicin (50 μg ml−1) was added to the cultures to prevent initiation of transcription. Aliquots of bacteria were then removed at time 0, 30 sec, 1 min, and every min thereafter for 5 min, and immediately mixed with 2 volumes of the RNA-Protect Bacteria reagent. RNA was extracted, cDNA synthesized, and qRT-PCR was performed to determine the relative fold change of the hmsP and hmsT transcripts as described above. mRNA half-life was determined by plotting the relative fold change compared to time 0 for each strain on a semi-log plot. The first 4 data points were then fit with a linear curve and the equation t1/2=0.693/k was used, where k=slope of the line. Data are represented as a percentage of mRNA transcript remaining over time. Statistical analysis was performed using the students' t test.

Transcriptional start site determination

Y. pestis was cultured and RNA was extracted as described for qRT-PCR. The transcriptional start sites of hmsP and hmsT were determined on two independent cultures using the 5' RACE System for Rapid Amplification of cDNA Ends, Version 2.0 kit (Invitrogen) according to the manufacturers' instructions using the primers listed in Table S3. The transcriptional start site of hmsT was confirmed by PCR as follows: genomic DNA isolated from Y. pestis or cDNA generated as for 5' RACE from the same strain was subjected to PCR (30 cycles of 94°C/30 sec - 47°C/30 sec - 72°C/30 sec) with the primers hmsT 5' -128, hmsT 5' -77, or hmsT 5' -21 paired with the primer hmsT 3' 346 and the resulting products were analyzed by agarose gel electrophoresis and stained with ethidium bromide before visualization.

GFP assays

To construct PhmsP-gfp and PhmsT-gfp reporters, the promoter regions of hmsP and hmsT (729 bp upstream of the ATG of hmsP and including the proximal 27 bp of the hmsP CDS, or 500 bp upstream of the ATG of hmsT and including the proximal 27 bp of the hmsT CDS) were amplified by PCR. The gene for GFP was amplified by PCR and the resulting products were subsequently joined by SOE-PCR before being cloned into pUC18R6K-mini-Tn7T-kan. To construct the PtetO-hmsP 5' UTR-gfp and PtetO-hmsT 5' UTR-gfp reporters, the PtetO sequence lacking the 5' UTR (Lutz and Bujard, 1997) was PCR-amplified from pWL213, and the 5' UTR of hmsP or hmsT as determined by 5' RACE was amplified from the genome of Y. pestis (224 bp upstream of the ATG of hmsP and including the proximal 27 bp of the hmsP CDS, or 21 bp upstream of the ATG of hmsT and including the proximal 27 bp of the hmsT CDS). These products were subsequently joined to the CDS of gfp by SOE-PCR and cloned into pWL212. All sequences were confirmed and the reporter constructs were then introduced onto the chromosome of Y. pestis or Y. pestis Δhfq via Tn7-based integration as described above.

Y. pestis strains carrying the GFP reporters were cultured in triplicate overnight in BHI broth at 26°C in a roller drum before being diluted to an OD620 of 0.1 in 25 ml BHI broth in a 125 ml Erlenmeyer flask. ATc (0.5 μg ml−1) was added where appropriate. Bacteria were then cultured at 26°C in a shaker at 250 RPM for 18 hours before aliquots were removed to measure relative GFP fluorescence. Briefly, the OD620 of the cultures were determined and then the fluorescence of the same samples were measured using a Tecan Safire2 microplate reader with a gain of 60 at an excitation wavelength of 395 nm and an emission wavelength of 509 nm. The background fluorescence of the BHI broth followed by fluorescence produced from equivalent strains of Y. pestis carrying a promoterless gfp gene integrated at the attTn7 site were sequentially subtracted. Relative fluorescence units (RFU) were then normalized to the OD of the bacterial culture, and the RFU of each sample was measured in duplicate. Statistical differences were determined by the students' t test.

Supplementary Material

Supp Table S1-S3&Supp Fig S1-S2

Acknowledgements

We thank members of the Lathem lab and Drs. Robert Perry, Paul Price, Viveka Vadyvaloo, and Fitnat Yildiz for helpful discussions. We also thank Chelsea Schiano for technical support and Dr. Melanie Marketon for the kind gift of the RpoA antibody. This work was supported by the Northwestern University Searle Leadership Fund and NIH grant K22 AI-073781 to WWL, NIH grant K22 AI-080937 to CMW, and the NIH/NIAID Regional Center of Excellence for Bio-defense and Emerging Infectious Diseases Research (RCE) Program. The authors wish to acknowledge membership within and support from the Region V `Great Lakes' RCE (NIH award U54 AI-057153).

Footnotes

The authors have no conflicts of interest to declare.

References

  1. Aiba H. Mechanism of RNA silencing by Hfq-binding small RNAs. Curr Opin Microbiol. 2007;10:134–139. doi: 10.1016/j.mib.2007.03.010. [DOI] [PubMed] [Google Scholar]
  2. Bobrov AG, Kirillina O, Perry RD. The phosphodiesterase activity of the HmsP EAL domain is required for negative regulation of biofilm formation in Yersinia pestis. FEMS Microbiol Lett. 2005;247:123–130. doi: 10.1016/j.femsle.2005.04.036. [DOI] [PubMed] [Google Scholar]
  3. Bobrov AG, Kirillina O, Perry RD. Regulation of biofilm formation in Yersinia pestis. Adv Exp Med Biol. 2007;603:201–210. doi: 10.1007/978-0-387-72124-8_17. [DOI] [PubMed] [Google Scholar]
  4. Bobrov AG, Kirillina O, Ryjenkov DA, Waters CM, Price PA, Fetherston JD, Mack D, et al. Systematic analysis of cyclic di-GMP signalling enzymes and their role in biofilm formation and virulence in Yersinia pestis. Mol Microbiol. 2011;79:533–551. doi: 10.1111/j.1365-2958.2010.07470.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bobrov AG, Kirillina O, Forman S, Mack D, Perry RD. Insights into Yersinia pestis biofilm development: Topology and co-interaction of Hms inner membrane proteins involved in exopolysaccharide production. Env Microbiol. 2008;10:1419–1432. doi: 10.1111/j.1462-2920.2007.01554.x. [DOI] [PubMed] [Google Scholar]
  6. Brennan RG, Link TM. Hfq structure, function and ligand binding. Curr Opin Microbiol. 2007;10:125–133. doi: 10.1016/j.mib.2007.03.015. [DOI] [PubMed] [Google Scholar]
  7. Chao Y, Vogel J. The role of Hfq in bacterial pathogens. Curr Opin Microbiol. 2010;13:24–33. doi: 10.1016/j.mib.2010.01.001. [DOI] [PubMed] [Google Scholar]
  8. Choi K-H, Gaynor JB, White KG, Lopez C, Bosio CM, Karkhoff-Schweizer RR, Schweizer HP. A Tn7-based broad-range bacterial cloning and expression system. Nat Methods. 2005;2:443–448. doi: 10.1038/nmeth765. [DOI] [PubMed] [Google Scholar]
  9. Darby C. Uniquely insidious: Yersinia pestis biofilms. Trends Microbiol. 2008;16:158–164. doi: 10.1016/j.tim.2008.01.005. [DOI] [PubMed] [Google Scholar]
  10. Eisen RJ, Bearden SW, Wilder AP, Montenieri JA, Antolin MF, Gage KL. Early-phase transmission of Yersinia pestis by unblocked fleas as a mechanism explaining rapidly spreading plague epizootics. Proc Natl Acad Sci USA. 2006;103:15380–15385. doi: 10.1073/pnas.0606831103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Ellison DW, Lawrenz MB, Miller VL. Invasin and beyond: Regulation of Yersinia virulence by RovA. Trends Microbiol. 2004;12:296–300. doi: 10.1016/j.tim.2004.04.006. [DOI] [PubMed] [Google Scholar]
  12. Fong JCN, Yildiz FH. Interplay between cyclic AMP-cyclic AMP receptor protein and cyclic di-GMP signaling in Vibrio cholerae biofilm formation. J Bacteriol. 2008;190:6646–6659. doi: 10.1128/JB.00466-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Forman S, Bobrov AG, Kirillina O, Craig SK, Abney J, Fetherston JD, Perry RD. Identification of critical amino acid residues in the plague biofilm Hms proteins. Microbiol (Reading, Engl) 2006;152:3399–3410. doi: 10.1099/mic.0.29224-0. [DOI] [PubMed] [Google Scholar]
  14. Geng J, Song Y, Yang L, Feng Y, Qiu Y, Li G, Guo J, et al. Involvement of the post-transcriptional regulator Hfq in Yersinia pestis virulence. PLoS ONE. 2009;4:e6213. doi: 10.1371/journal.pone.0006213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gottesman S, Storz G. Bacterial small RNA regulators: Versatile roles and rapidly evolving variations. Cold Spring Harbor Perspect Biol. 2011;3 doi: 10.1101/cshperspect.a003798. 10.1101/cshperspect.a003798. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gottesman S, McCullen CA, Guillier M, Vanderpool CK, Majdalani N, Benhammou J, Thompson KM, et al. Small RNA regulators and the bacterial response to stress. Cold Spring Harbor Symp Quant Biol. 2006;71:1–11. doi: 10.1101/sqb.2006.71.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hammer BK, Bassler BL. Quorum sensing controls biofilm formation in Vibrio cholerae. Mol Microbiol. 2003;50:101–104. doi: 10.1046/j.1365-2958.2003.03688.x. [DOI] [PubMed] [Google Scholar]
  18. Hammer BK, Bassler BL. Regulatory small RNAs circumvent the conventional quorum sensing pathway in pandemic Vibrio cholerae. Proc Natl Acad Sci USA. 2007;104:11145–11149. doi: 10.1073/pnas.0703860104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hengge R. Principles of c-di-GMP signalling in bacteria. Nat Rev Microbiol. 2(009;7:263–273. doi: 10.1038/nrmicro2109. [DOI] [PubMed] [Google Scholar]
  20. Heroven AK, Dersch P. RovM, a novel LysR-type regulator of the virulence activator gene rovA, controls cell invasion, virulence and motility of Yersinia pseudotuberculosis. Mol Microbiol. 2006;62:1469–1483. doi: 10.1111/j.1365-2958.2006.05458.x. [DOI] [PubMed] [Google Scholar]
  21. Hinnebusch BJ, Erickson DL. Yersinia pestis biofilm in the flea vector and its role in the transmission of plague. Curr Top Microbiol Immunol. 2008;322:229–248. doi: 10.1007/978-3-540-75418-3_11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hinnebusch BJ, Perry RD, Schwan TD. Role of the Yersinia pestis hemin storage (Hms) locus in the transmission of plague by fleas. Science. 1996;273:367–370. doi: 10.1126/science.273.5273.367. [DOI] [PubMed] [Google Scholar]
  23. Holmqvist E, Reimegård J, Sterk M, Grantcharova N, Römling U, Wagner EGH. Two antisense RNAs target the transcriptional regulator CsgD to inhibit curli synthesis. EMBO J. 2010;29:1840–1850. doi: 10.1038/emboj.2010.73. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Huang XZ, Nikolich MP, Mullarky LE. Current trends in plague research: From genomics to virulence. Clin Med & Res. 2006;4:189–199. doi: 10.3121/cmr.4.3.189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Itoh Y, Wang X, Hinnebusch BJ, Preston JF, Romeo T. Depolymerization of beta-1,6-N-acetyl-D-glucosamine disrupts the integrity of diverse bacterial biofilms. J Bacteriol. 2005;187:382–387. doi: 10.1128/JB.187.1.382-387.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Jobling MG, Holmes RK. Characterization of hapR, a positive regulator of the Vibrio cholerae HA/protease gene Hap, and its identification as a functional homologue of the Vibrio harveyi luxR gene. Mol Microbiol. 1997;26:1023–1034. doi: 10.1046/j.1365-2958.1997.6402011.x. [DOI] [PubMed] [Google Scholar]
  27. Jones AM, Goodwill A, Elliott T. Limited role for the DsrA and RprA regulatory RNAs in rpoS regulation in Salmonella enterica. J Bacteriol. 2006;188:5077–5088. doi: 10.1128/JB.00206-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kint G, De Coster D, Marchal K, Vanderleyden J, De Keersmaecker SCJ. The small regulatory RNA molecule MicA is involved in Salmonella enterica serovar Typhimurium biofilm formation. BMC Microbiol. 2010;10:276. doi: 10.1186/1471-2180-10-276. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kirillina O, Fetherston JD, Bobrov AG, Abney J, Perry RD. HmsP, a putative phosphodiesterase, and HmsT, a putative diguanylate cyclase, control Hms-dependent biofilm formation in Yersinia pestis. Mol Microbiol. 2004;54:75–88. doi: 10.1111/j.1365-2958.2004.04253.x. [DOI] [PubMed] [Google Scholar]
  30. Koo JT, Alleyne TM, Schiano CA, Jafari N, Lathem WW. Global discovery of small RNAs in Yersinia pseudotuberculosis identifies Yersinia-specific small, noncoding RNAs required for virulence. Proc Natl Acad Sci USA. 2011;108:E709–E717. doi: 10.1073/pnas.1101655108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Koo JT, Lathem WW. Global identification of small noncoding RNAs in pathogenic Yersinia species. Adv Exp Med Biol. 2012;954:305–314. doi: 10.1007/978-1-4614-3561-7_38. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Kulesus RR, Diaz-Perez K, Slechta ES, Eto DS, Mulvey MA. Impact of the RNA chaperone Hfq on the fitness and virulence potential of uropathogenic. Escherichia coli. Infect Immun. 2008;76:3019–3026. doi: 10.1128/IAI.00022-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Lathem WW, Price PA, Miller VL, Goldman WE. A plasminogen-activating protease specifically controls the development of primary pneumonic plague. Science. 2007;315:509–513. doi: 10.1126/science.1137195. [DOI] [PubMed] [Google Scholar]
  34. Lawrenz MB, Lenz JD, Miller VL. A novel autotransporter adhesion is required for efficient colonization during bubonic plague. Infect Immun. 2009;77:317–326. doi: 10.1128/IAI.01206-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Lenz DH, Mok KC, Lilley BN, Kulkarni RV, Wingreen NS, Bassler BL. The small RNA chaperone Hfq and multiple small RNAs control quorum sensing in Vibrio harveyi and Vibrio cholerae. Cell. 2004;118:69–82. doi: 10.1016/j.cell.2004.06.009. [DOI] [PubMed] [Google Scholar]
  36. Lutz R, Bujard H. Independent and tight regulation of transcriptional units in Escherichia coli via the LacR/O, the TetR/O and AraC/I1–I2 regulatory elements. Nucleic Acids Res. 1997;25:1203–1210. doi: 10.1093/nar/25.6.1203. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Merriam JJ, Mathur R, Maxfield-Boumil R, Isberg RR. Analysis of the Legionella pneumophila fliI gene: Intracellular growth of a defined mutant defective for flagellum biosynthesis. Infect Immun. 1997;65:2497–2501. doi: 10.1128/iai.65.6.2497-2501.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Patel CN, Wortham BW, Lines JL, Fetherston JD, Perry JD, Oliveira MA. Polyamines are essential for the formation of plague biofilm. J Bacteriol. 2006;188:2355–2363. doi: 10.1128/JB.188.7.2355-2363.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Perry RD, Bobrov AG, Kirillina O, Jones HA, Pedersen L, Abney J, Fetherston JD. Temperature regulation of the hemin storage (Hms+) phenotype of Yersinia pestis Is posttranscriptional. J Bacteriol. 2004;186:1638–1647. doi: 10.1128/JB.186.6.1638-1647.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Perry RD, Fetherston JD. Yersinia pestis--etiologic agent of plague. Clin Microbiol Rev. 1997;10:35–66. doi: 10.1128/cmr.10.1.35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Perry RD, Pendrak ML, Schuetze P. Identification and cloning of a hemin storage locus involved in the pigmentation phenotype of Yersinia pestis. J Bacteriol. 1990;172:5929–5937. doi: 10.1128/jb.172.10.5929-5937.1990. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Rempe KA, Hinz AK, Vadyvaloo V. Hfq regulates biofilm gut blockage that facilitates flea-borne transmission of Yersinia pestis. J Bacteriol. 2012;194:2036–2040. doi: 10.1128/JB.06568-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Schiano CA, Bellows LE, Lathem WW. The small RNA chaperone Hfq is required for the virulence of Yersinia pseudotuberculosis. Infect Immun. 2010;78:2034–2044. doi: 10.1128/IAI.01046-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Sondermann H, Shikuma NJ, Yildiz FH. You've come a long way: c-di-GMP signaling. Curr Opin Microbiol. 2012;15:140–146. doi: 10.1016/j.mib.2011.12.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Staggs TM, Perry RD. Identification and cloning of a Fur regulatory gene in Yersinia pestis. J Bacteriol. 1991;173:417–425. doi: 10.1128/jb.173.2.417-425.1991. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Stenseth NC, Atshabar BB, Begon M, Belmain SR, Bertherat E, Carniel E, Gage KL, et al. Plague: Past, present, and future. PLoS Med. 2008;5:e3. doi: 10.1371/journal.pmed.0050003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sun Y-C, Koumoutsi A, Jarrett C, Lawrence K, Gherardini FC, Darby C, et al. Differential control of Yersinia pestis biofilm formation in vitro and in the flea vector by two cdi-GMP diguanylate cyclases. PLoS ONE. 2011;6:e19267. doi: 10.1371/journal.pone.0019267. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Sun Y-C, Guo X-P, Hinnebusch BJ, Darby C. The Yersinia pestis Rcs phosphorelay inhibits biofilm formation by repressing transcription of the diguanylate cyclase gene hmsT. J Bacteriol. 2012;194:2020–2026. doi: 10.1128/JB.06243-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Wang X, Preston JF, Romeo T. The pgaABCD locus of Escherichia coli promotes the synthesis of a polysaccharide adhesin required for biofilm formation. J Bacteriol. 2004;186:2724–2734. doi: 10.1128/JB.186.9.2724-2734.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Waters LS, Storz G. Regulatory RNAs in bacteria. Cell. 2009;136:615–628. doi: 10.1016/j.cell.2009.01.043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Wimsatt J, Biggins DE. A review of plague persistence with special emphasis on fleas. J Vector Dis. 2009;46:85–99. [PubMed] [Google Scholar]
  52. Wren BW. The Yersiniae-a model genus to study the rapid evolution of bacterial pathogens. Nat Rev Microbiol. 2003;1:55–64. doi: 10.1038/nrmicro730. [DOI] [PubMed] [Google Scholar]
  53. Wu X-G, Duan H-M, Tian T, Yao N, Zhou H-Y, Zhang L-Q. Effect of the Hfq gene on 2,4-diacetylphloroglucinol production and the PcoI/PcoR quorum-sensing system in Pseudomonas fluorescens 2p24. FEMS Microbiol Lett. 2010;309:16–24. doi: 10.1111/j.1574-6968.2010.02009.x. [DOI] [PubMed] [Google Scholar]
  54. Yan H, Chen W. 3′,5′-cyclic diguanylic acid: a small nucleotide that makes big impacts. Chem Soc Rev. 2010;39:2914–2924. doi: 10.1039/b914942m. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp Table S1-S3&Supp Fig S1-S2
NIHMS403221-supplement-Supp_Table_S1-S3_Supp_Fig_S1-S2.pdf (855.7KB, pdf)

RESOURCES