A Trimeric Autotransporter Enhances Biofilm Cohesiveness in Yersinia pseudotuberculosis but Not in Yersinia pestis

The evolution of Yersinia pestis from its Y. pseudotuberculosis ancestor involved gene acquisition and gene losses, leading to differences in biofilm production. Characterizing the unique biofilm features of both species may provide better understanding of how each adapts to its specific niches. This study identifies a trimeric autotransporter, YadE, that promotes biofilm stability of Y. pseudotuberculosis but which has been inactivated in Y. pestis, perhaps because it is not compatible with the Hms polysaccharide that is crucial for biofilms inside fleas. We also reveal that the Rcs signaling cascade, which represses Hms expression, activates YadE in Y. pseudotuberculosis. The ability of Y. pseudotuberculosis to use polysaccharide or YadE protein for cell-cell adhesion may help it produce biofilms in different environments.

ABSTRACT

Cohesion of biofilms made by Yersinia pestis and Yersinia pseudotuberculosis has been attributed solely to an extracellular polysaccharide matrix encoded by the hms genes (Hms-dependent extracellular matrix [Hms-ECM]). However, mutations in the Y. pseudotuberculosis BarA/UvrY/CsrB regulatory cascade enhance biofilm stability without dramatically increasing Hms-ECM production. We found that treatment with proteinase K enzyme effectively destabilized Y. pseudotuberculosis csrB mutant biofilms, suggesting that cell-cell interactions might be mediated by protein adhesins or extracellular matrix proteins. We identified an uncharacterized trimeric autotransporter lipoprotein (YPTB2394), repressed by csrB, which has been referred to as YadE. Biofilms made by a ΔyadE mutant strain were extremely sensitive to mechanical disruption. Overexpression of yadE in wild-type Y. pseudotuberculosis increased biofilm cohesion, similar to biofilms made by csrB or uvrY mutants. We found that the Rcs signaling cascade, which represses Hms-ECM production, activated expression of yadE. The yadE gene appears to be functional in Y. pseudotuberculosis but is a pseudogene in modern Y. pestis strains. Expression of functional yadE in Y. pestis KIM6+ weakened biofilms made by these bacteria. This suggests that although the YadE autotransporter protein increases Y. pseudotuberculosis biofilm stability, it may be incompatible with the Hms-ECM production that is essential for Y. pestis biofilm production in fleas. Inactivation of yadE in Y. pestis may be another instance of selective gene loss in the evolution of flea-borne transmission by this species.

IMPORTANCE The evolution of Yersinia pestis from its Y. pseudotuberculosis ancestor involved gene acquisition and gene losses, leading to differences in biofilm production. Characterizing the unique biofilm features of both species may provide better understanding of how each adapts to its specific niches. This study identifies a trimeric autotransporter, YadE, that promotes biofilm stability of Y. pseudotuberculosis but which has been inactivated in Y. pestis, perhaps because it is not compatible with the Hms polysaccharide that is crucial for biofilms inside fleas. We also reveal that the Rcs signaling cascade, which represses Hms expression, activates YadE in Y. pseudotuberculosis. The ability of Y. pseudotuberculosis to use polysaccharide or YadE protein for cell-cell adhesion may help it produce biofilms in different environments.

INTRODUCTION

Environmental persistence, host interaction, and transmission of Yersinia depend on biofilms, which are tightly regulated by both transcriptional and posttranscriptional control mechanisms (1, 2). Arguably, the best-studied Yersinia biofilms are those made by Yersinia pestis while in the flea digestive tract that block the proventriculus and increase transmission to new hosts during flea feeding. These biofilms require the HmsHFRS proteins to produce and export a polysaccharide extracellular matrix of poly-β-1,6-N-acetylglucosamine that is crucial in forming and maintaining bacterial cell-cell attachments (3, 4). Identical polysaccharide produced by several other species promotes their adherence to abiotic surfaces (5, 6) and enhances virulence of some pathogens (7–9) while reducing that of others (10, 11). Without high levels of Hms-dependent extracellular matrix (Hms-ECM), the biofilms formed by Y. pestis while in fleas are not sufficiently cohesive to cause proventricular blockage. Adaptation to flea-borne transmission was precipitated in part by mutations that led to high levels of Hms-ECM compared to those of its Yersinia pseudotuberculosis ancestral lineage (12–14).

Among these mutations, modification of the Rcs regulatory system was especially important in enhancing Hms-ECM production. The Rcs signaling system includes the inner membrane kinase RcsC and the phosphorelay protein RcsD. RcsC phosphorylates itself when an inducing signal is present, and that phosphate is passed to RcsD and then to RcsB. Phosphorylated RcsB is a transcriptional regulator, binding to target promoters either as homodimers or as heterodimers with the auxiliary protein RcsA or other proteins (15). Normally, the system is kept in the “off” state by an inner membrane protein IgaA that blocks signaling through RcsD. When an appropriate activating signal is present, the RcsF lipoprotein sensor interacts with IgaA and relieves the inhibition (16). RcsAB heterodimers negatively regulate Yersinia biofilms by binding to the promoters of the diguanylate cyclase genes hmsT and hmsD, as well as the hmsHFRS operon itself (17–19). The rcsA gene is inactive in Y. pestis due to an insertion in the open reading frame, and restoring the function of this gene prevents biofilm formation and flea blockage (20). In Y. pestis, RcsB homodimers bind to the hmsD promoter and increase transcription of this gene (17).

As in many bacteria, exopolysaccharide production is positively regulated in Yersinia by cyclic di-GMP made by diguanylate cyclases (21, 22). When RcsAB repression of hmsT and hmsD transcription is eliminated, the resultant high levels of the second messenger enhance Hms-ECM production, likely by activating the HmsRS inner membrane proteins (18, 23, 24). Two phosphodiesterases that degrade cyclic di-GMP are functional in Y. pseudotuberculosis but not in Y. pestis. A Y. pseudotuberculosis mutant strain wherein these genes (rcsA and both phosphodiesterases) are replaced with Y. pestis nonfunctional alleles can block fleas but not to the same extent as Y. pestis (12). This indicates that additional biofilm-related differences exist between the two species.

Several other regulatory influences on biofilm production in both Y. pseudotuberculosis and Y. pestis have been identified. We recently reported that the BarA/UvrY two-component system represses biofilms in Y. pseudotuberculosis by activating the CsrB small RNA (25). Although Y. pseudotuberculosis mutants lacking csrB or uvrY make more cohesive biofilms than the wild-type strain, their production of Hms-ECM does not approach that of Y. pestis. This suggested that Y. pseudotuberculosis biofilms may contain additional extracellular matrix components that are responsible for their cohesion. In this study, we investigated the basis for the increased cohesiveness of csrB mutant biofilms in an effort to identify novel Y. pseudotuberculosis-specific biofilm features. We found that Y. pseudotuberculosis biofilms have a significant protein component that is not present in Y. pestis. We focused on the uncharacterized YPTB2394 protein (YadE), which is repressed by csrB. This predicted lipoprotein is a part of the trimeric autotransporter family of proteins that function in bacterial adhesion to host surfaces or bacterial cell-cell attachments in biofilms (26–28). Here, we show that production of YadE leads to Y. pseudotuberculosis biofilms that strongly resist disruption. Expression of yadE is activated by the Rcs system in contrast to repression of Hms-ECM by RcsAB. Conversely, yadE is a pseudogene in Y. pestis, and we demonstrate that yadE expression changes Hms-ECM-related phenotypes and weakens Y. pestis biofilms.

RESULTS

Y. pseudotuberculosis biofilm cohesion requires proteins.

We previously demonstrated that mutations in the BarA/UvrY two-component regulatory system or in the CsrB small RNA activated by UvrY increase cohesion of Y. pseudotuberculosis strain IP32953 biofilms (25). In that study, lectin staining and Congo red binding of the csrB mutant strain indicated only moderately increased Hms-ECM on the surface of these bacteria. This suggested the possibility of additional structural components that are increased in the csrB mutant strain, which assist in holding Y. pseudotuberculosis biofilms together. Protein adhesins and extracellular DNA are present in the biofilm extracellular matrices of numerous other bacteria. Filter biofilms formed by the Y. pseudotuberculosis wild type or the csrB mutant strain were treated with DNase or proteinase K enzymes prior to testing their dispersal in mechanical disruption assays (Fig. 1). As expected, the csrB mutant formed more cohesive biofilms than the wild-type strain. DNase did not significantly increase disruption of either wild-type or csrB mutant biofilms compared with that of the control saline treatment. In contrast, proteinase K markedly increased the proportion of the csrB mutant strain biofilms that were dispersed. Y. pestis biofilms were very stable in these assays and were not affected by proteinase K or DNase treatment.

FIG 1.

Increased stability of Y. pseudotuberculosis csrB::Tn5 mutant biofilms may be mediated by proteins. Biofilms were grown on polycarbonate filters for 72 h. Filter biofilms were then treated with proteinase K or DNase for 30 min prior to agitation (200 rpm) for 15 min in PBS. The absorbance of the PBS was measured after the 15-min interval and measured again after vortexing the filter to dislodge the entire biofilm. The proportion of the biofilms that were dislodged by agitation alone is reported. Pretreatment of Y. pseudotuberculosis csrB::Tn5 mutant biofilms with proteinase K enhances their dispersal relative to the PBS control (****, P < 0.001 by Student's t test) while DNase had no effect. Y. pestis biofilm stability was not affected by pretreatment with either enzyme.

YPTB2394 putative autotransporter expression confers biofilm cohesion.

Since Y. pseudotuberculosis csrB mutant biofilms are likely more cohesive in part due to a protein component of the extracellular matrix, we searched for proteins that were expressed more abundantly in the csrB mutant strain. Comparison of the wild-type and csrB mutant proteomes had been performed previously, which showed a large number of proteins that were expressed at lower levels in the csrB mutant, which is consistent with the role of CsrA mainly as a translational repressor (22). YPTB2394 was among the few proteins that were expressed more abundantly (approximately 30-fold higher) by the csrB mutant. YPTB2394 is annotated as a predicted lipoprotein with homology to type Vc autotransporter proteins. Trimeric autotransporter adhesins are membrane-anchored proteins known to mediate cell-cell attachments in other Gram-negative bacteria and promote greatly enhanced biofilm production in diverse species (29–31). YPTB2394 is predicted to have a C-terminal YadA-like anchor and several YadA-like stalk domains. The orthologous gene in Y. pseudotuberculosis strain YPIII (YPK_0761) has been referred to as yadE (32), although its function has not been investigated.

To examine the role of yadE in biofilm cohesion, we first deleted the gene from the wild-type Y. pseudotuberculosis IP32953 strain. The total accumulations of the wild-type and mutant strains on the biofilm filters were similar (data not shown), but the biofilms formed by the mutant strain were extremely fragile compared to those of the wild-type strain (Fig. 2). The mutant strain was complemented by inserting yadE on a low-copy-number plasmid with a constitutive promoter. This resulted in biofilms that were highly resistant to disruption, even more so than csrB::Tn5 mutant biofilms. This strongly suggests that YadE helps maintain intracellular contacts and promotes Y. pseudotuberculosis biofilm cohesion.

FIG 2.

YPTB2394 (yadE) affects biofilm cohesion. Biofilms formed by the wild-type IP32953, YPTB2394 (ΔyadE) mutant, and strain overexpressing yadE were each agitated and their dispersal was measured at 5-min intervals. Results were analyzed using repeated measures analysis of variance (ANOVA) with Tukey’s multiple comparison test (****, significantly different from wild type; P < 0.0001).

yadE gene expression is regulated by CsrB and by the Rcs regulatory cascade.

YadE (YPTB2394) protein levels are approximately 30-fold greater in csrB mutant bacteria than in wild-type cells (25). CsrB is a regulatory RNA that sequesters CsrA protein (33, 34). Typically, CsrA binds to target sequences found near the Shine-Dalgarno region of target transcripts and represses their translation. CsrB accumulation frees mRNA targets of CsrA to be translated more efficiently. However, translation of some CsrA-regulated proteins is enhanced by CsrA binding, and it is possible that CsrA is a direct translational activator of YadE. Alternatively, CsrA could repress translation of another transcriptional regulator, making its effect on yadE expression indirect. To further investigate regulation of yadE, a reporter plasmid containing the 356-bp region upstream of the start codon fused to a promoterless green fluorescent protein (gfp) gene was created (Fig. 3A). Transformation of wild-type cells with this plasmid resulted in low but detectable levels of fluorescence. Expression was significantly enhanced in csrB::Tn5 mutant bacteria (Fig. 3B), confirming the negative effect on YadE expression by CsrB.

FIG 3.

Expression of yadE is repressed by CsrB. (A) The upstream region of yadE (−356 to +3 relative to the ATG start codon) was fused to gfp. The predicted transcriptional start site is located at −138 based on RNA sequencing of the YPIII strain (32). A putative RcsAB binding site centered at −313 was identified based on the consensus RcsAB box of enteric bacteria (39). The promoter region, Shine-Dalgarno sequence, and start codon of the reporter gene were derived from yadE. (B) The Y. pseudotuberculosis IP32953 wild-type and csrB::Tn5 mutant strains were transformed with the yadE::gfp reporter plasmid. After growth for 48 h on solid medium, the fluorescence of the population was measured by flow cytometry. ***, a significant difference from the wild-type strain (P = 0.0002) by unpaired t test.

In order to identify possible transcriptional repressors of yadE, we created a transposon mutant library in a wild-type IP32953 strain carrying the yadE::gfp reporter plasmid. Fluorescence-activated cell sorting was then used to enrich for mutants with higher yadE expression than the wild-type strain. After sorting, we verified enhanced yadE::gfp activity in isolated single colonies of 48 mutants by flow cytometry and determined their transposon insertion sites. We identified 23 distinct insertion sites among these mutants (see Table S1 in the supplemental material), which included 9 separate transposon mutants with insertions in the igaA gene that exhibited ∼8-fold higher yadE::gfp than the wild-type strain. IgaA is a periplasmic protein that prevents overactivation of the Rcs signaling cascade (15). Multiple mutants with insertions in genes encoding adenylate cyclase cya (2.5-fold increase) and the stringent starvation protein sspA (8.5-fold increase) were also identified. We tested biofilms made by individual cya, sspA, and igaA transposon mutants and found that they were more cohesive than the wild-type strain but less so than the wild-type strain carrying the pACYC-yadE plasmid (Fig. 4). This may be due to other changes in gene expression besides high levels of yadE in these mutants that influence their biofilm stability.

FIG 4.

Mutants that overexpress yadE produce more cohesive biofilms. Transposon mutants derived from IP32953 with high yadE::gfp reporter activity were selected by FACS. Individual mutants with insertions in igaA, sspA, and cya genes were tested in biofilm disruption assays in comparison with the wild-type IP32953 strain and the wild-type strain overexpressing yadE (pACYC-yadE). One-way ANOVA was performed with Tukey's correction for multiple comparisons. Columns with the same letter are not significantly different from each other (95% confidence interval).

In Escherichia coli and Salmonella, IgaA limits the phosphorylation of RcsC and thereby prevents activation of RcsB. In some Enterobacteriaceae, igaA is an essential gene, as overactivation of the Rcs cascade is lethal (35, 36). However, Y. pestis mutants with transposon insertions in igaA have been reported (37, 38), and our results confirm that this gene is also not essential in Y. pseudotuberculosis despite its fully functioning Rcs cascade. We verified that plasmid complementation of the igaA mutation could restore lower yadE expression similar to that of the wild-type strain (Fig. 5). As IgaA represses RcsC and RcsB activation, we considered that increased RcsB activity could be responsible for the enhanced yadE expression in igaA mutants. We compared yadE promoter activities in an rcsB::Tn5 mutant background and in bacteria overexpressing rcsB via a multicopy plasmid. Lack of rcsB did not measurably decrease yadE expression compared to that of the wild-type strain, but overexpression resulted in much higher fluorescence, similar to that of the igaA mutant (Fig. 5). These results suggest that the Rcs signaling cascade positively regulates yadE expression, opposite to the Rcs repression of Hms-ECM production (18). This is supported by the presence of a putative RcsAB binding site (39) in the region upstream of yadE (Fig. 3A).

FIG 5.

Expression of yadE is controlled by the Rcs phosphorelay. Inactivation of igaA or overexpression of rcsB increased yadE::gfp reporter activity. Fluorescence was measured by flow cytometry and compared with the wild-type strain (**, significant difference by unpaired t test; P < 0.01).

yadE is a pseudogene in Y. pestis, and expression of functional yadE prevents Hms-ECM production.

Selective gene loss in Y. pestis during its divergence from Y. pseudotuberculosis has contributed to the emergence of flea-borne transmission (12, 13, 20, 40). All publicly available genome sequences contain a yadE gene fragment that is identical among Y. pestis strains. Pairwise alignment shows four small deletions in the N-terminal region of the Y. pestis sequences relative to the Y. pseudotuberculosis IP32953 yadE gene. These are predicted to maintain the reading frame with 98.5% amino acid identity across the first 706 residues. However, a single insertion in a poly(G) tract starting at nucleotide 2118 alters the reading frame of the Y. pestis sequence and results in five premature stop codons (Fig. 6). These are predicted to result in a nonfunctional protein missing the C-terminal YadA-like membrane-anchoring domain. To investigate the possible consequences of yadE loss on Y. pestis biofilm stability, Y. pestis KIM6+ was transformed with the same plasmid conferring constitutive expression of the functional yadE gene. In contrast to its effect on Y. pseudotuberculosis, expression of yadE in Y. pestis resulted in biofilms that were more easily disrupted (Fig. 7A). This strain also produced colonies that were less pigmented on Congo red agar plates, which is suggestive of reduced Hms-ECM production (Fig. 7B). Production of Hms-ECM was also tested using lectin staining and flow cytometry. Wild-type Y. pestis binds strongly to wheat germ agglutinin (WGA) specific for N-acetylglucosamine polysaccharides. Conversely, Y. pestis expressing yadE exhibited a bipolar WGA-lectin staining pattern, wherein a significant portion of the cells did not bind to the lectin while another population bound at higher levels than the wild-type strain (Fig. 7C). This result is not entirely consistent with the Congo red binding phenotype of the yadE⁺ Y. pestis strain, but it is suggestive of alterations to the cell envelope or extracellular matrix including, but not limited to, Hms-ECM that affect biofilm cohesion. Thus, expression of functional yadE in Y. pestis weakens biofilm cohesiveness, perhaps by altering Hms-ECM production, stability, or distribution.

FIG 6.

The YPTB2394 and y1486 orthologues of Y. pseudotuberculosis and Y. pestis. Domains predicted in the YPTB2394 protein sequence typical of trimeric autotransporters (73) include a transmembrane helix containing a secretion signal peptide (30–51), a HANS domain that typically connects α-to-β regions of proximal to head regions, COG5295 comprising the β-strand head domains with repetitive motifs, and the YadA-like anchor consisting of a 12-stranded outer membrane β-barrel. The asterisk denotes the region containing the premature stop codons found in the Y. pestis sequences. Pairwise alignment of the Y. pseudotuberculosis IP32953 yadE (YPTB2394) and Y. pestis KIM pseudogene (y1486) nucleotide sequences shows that the sequences are highly similar across the 2,643-bp Y. pseudotuberculosis sequence. All Y. pestis sequences have 14-, 7-, 4-, and 5-bp deletions between nucleotides 185 and 209 and 326 and 340, which maintain the reading frame (not shown). A guanine insertion at nucleotide 2118 changes the reading frame and results in 5 predicted stop codons in the C-terminal region beginning at nucleotide 2230 that are predicted to eliminate the C-terminal YadA-like anchor.

FIG 7.

Expression of functional yadE in Y. pestis reduces biofilm stability and alters Hms-ECM production. (A) Disruption of biofilms formed by Y. pestis KIM6+ wild type or strains transformed with pACYC-yadE or the pACYC184 empty vector. Results were analyzed using repeated measures ANOVA with Tukey’s multiple comparison test (**, P < 0.01; ***, P < 0.001; ****, P < 0.0001). (B) Pigmentation of Y. pestis (yadE⁺ or empty vector) grown on Congo red agar plates. (C) Flow cytometry analysis of Y. pestis stained with GFP-labeled WGA lectin that binds to HMS-ECM. The wild-type strain (orange) is uniformly labeled compared to the unstained control (red), while the strain expressing functional yadE (blue) exhibits a bipolar staining pattern with either no detectable lectin bound to the surface or with very high staining.

DISCUSSION

Biofilms may be held together by extracellular polysaccharides, DNA, or proteins. Among Y. pseudotuberculosis strains, carriage of the hms genes encoding for polysaccharide extracellular matrix is ubiquitous, although they vary greatly in their expression of these genes (41). The Hms-ECM is required for Y. pseudotuberculosis biofilm production in some conditions, including on the mouthparts of predatory nematodes (42, 43). Extracellular DNA has also been detected in at least some Y. pseudotuberculosis nematode-associated biofilms, although the source and functional significance of the DNA remain unclear (44). We did not detect a role for DNA in Y. pseudotuberculosis biofilm cohesion in the limited conditions that we tested. The extracellular matrix of biofilms formed by Y. pestis, while they are in the flea digestive tract, incorporates material derived from blood digestion (4), but the bacteria are not known to produce any other biofilm extracellular matrix material other than Hms-ECM. Our study is the first to demonstrate a role for a protein in mediating cohesion of Yersinia biofilms. The YPTB2394 (yadE) gene encodes a trimeric autotransporter protein vital to the stability of Y. pseudotuberculosis biofilms, as its inactivation results in a fragile biofilm and its overexpression significantly strengthens cohesiveness of biofilms formed on filters. The role of yadE in biofilms made in other environments or experimental systems remains to be investigated. Further mechanistic studies should focus on how YadE contributes to attachment, maturation, or dispersal of Y. pseudotuberculosis biofilms in diverse conditions.

Multiple autotransporter proteins belonging to the type Va, Vc, or Ve families are present in Y. pestis and/or Y. pseudotuberculosis (45). Type Va Yersinia autotransporters have been mainly investigated in the context of Y. pestis infection. YapE, YapJ, YapK, and YapV promote development of bubonic and/or pneumonic plague in mice (46–48), likely by enhancing adherence to epithelial cells. YapC can induce aggregation when expressed in E. coli, but no effects on Y. pestis biofilms following deletion of its gene were observed (49). Y. pseudotuberculosis strains also possess an additional type Va autotransporter gene, yapX, whose function is unknown and which is also a pseudogene in all Y. pestis strains (46).

YadE is predicted to be a member of the type Vc trimeric autotransporter family. Numerous species use trimeric autotransporters to adhere to other bacteria within biofilms, host extracellular matrix proteins, or inanimate surfaces (28). The prototype of the Vc family is YadA, which is made by Y. pseudotuberculosis and Yersinia enterocolitica but has been inactivated in Y. pestis. YadA promotes autoagglutination and tight adherence to the eukaryotic cells necessary for proper injection of effector proteins via type III secretion (50). Restoring yadA gene function in Y. pestis has been reported to decrease its virulence in mouse infections (51). It is not known whether YadA expression also interferes with Hms-ECM production by Y. pestis. However, it is coexpressed along with the type III secretion effectors encoded on the virulence plasmid at 37°C rather than at lower temperatures when Hms-ECM is produced. Y. pseudotuberculosis and Y. pestis both express two additional type Vc proteins, YadB and YadC, also at 37°C, that promote invasion of host cells and bacterial survival in skin (52).

Versatile biofilm production strategies are probably most helpful to bacteria, such as Y. pseudotuberculosis, that are found in many different environments (free-living or within amoeba in soil or water, in plants, or in the digestive tracts of multiple animals) (53). Conversely, niche specialization might be promoted by the selection of one biofilm pathway at the expense of others. Many Staphylococcus strains produce biofilms that are dependent on polysaccharide intracellular adhesin, which is identical to Hms-ECM (54). However, expression of this polymer tends to be suppressed in methicillin-resistant Staphylococcus aureus strains in favor of fibronectin binding proteins (55). Staphylococci that predominantly inhabit environments with high shear stress or grow on medical devices produce primarily polysaccharide-dependent biofilms (56, 57), whereas those that interact more directly with host tissues may benefit from protein-based biofilms that allow them to incorporate fibrin or other host proteins into a protective shield (58). It is tempting to speculate that Y. pseudotuberculosis strains retain multiple biofilm strategies that provide flexibility according to changing environments, whereas Y. pestis may have jettisoned alternatives to Hms-ECM biofilms as it adapted to its restricted lifestyle of flea-rodent transmission.

Although yadE (y1486 in the Y. pestis KIM sequence) is a pseudogene in Y. pestis, it is one of the 100 most highly transcribed genes by this strain during infection of fleas (59). Furthermore, even greater expression was measured in a Y. pseudotuberculosis mutant strain that is able to infect and block fleas (60). Thus, the regulatory influences necessary for strong induction of this gene exist during flea infections. The regulatory controls on yadE expression may include the Csr system, the Rcs phosphorelay, and the catabolite activator protein/cyclic AMP receptor protein (CRP). We found that Tn5 insertions in the adenylate cyclase responsible for producing cyclic AMP greatly enhance yadE expression and biofilm stability. These results are consistent with previously reported transcriptome comparisons of a CRP mutant to the wild-type YPIII strain (32). This is significant because CRP is known to activate biofilms in Y. pestis (61, 62), but its role in Y. pseudotuberculosis biofilm regulation was not known. Our results suggest that Y. pseudotuberculosis yadE-dependent biofilms might be repressed by CRP-cAMP since the cya mutation enhances cohesiveness and yadE expression.

Our results also demonstrate that induction of Rcs signaling, either due to inactivation of igaA or overexpression of rcsB, dramatically increases expression of yadE. The control of Yersinia biofilms by the Rcs system includes many different targets, as illustrated in Fig. 8. RcsAB dimers directly repress transcription of the diguanylate cyclase genes hmsT and hmsD as well as the polysaccharide synthesis and export hmsHFRS operon itself. The phosphodiesterase gene hmsP is also indirectly activated by RcsAB, but the precise mechanism is not known. Thus, the effect of RcsAB activation is to repress cyclic di-GMP production and Hms-ECM. The activation of YadE by Rcs signaling might involve direct binding of RcsAB to the yadE promoter. Alternatively, RcsB homodimers could also control expression, which might explain why Y. pestis transcribes this gene at high levels in fleas. Understanding the mechanistic details of yadE regulation by Rcs requires further investigation.

FIG 8.

Regulation of Hms-ECM and YadE biofilms in Y. pseudotuberculosis. During normal growth in noninducing conditions, IgaA maintains low activation of the RcsC and RcsD proteins, favoring dephosphorylation of RcsB. The diguanylate cyclase genes hmsT and hmsD are expressed and cyclic di-GMP is produced, which activates the HmsHFRS proteins and production of Hms-ECM. Expression of the trimeric autotransporter YadE is low when Rcs signaling is suppressed, and the Csr system may also repress yadE, perhaps through RcsB or other components of the Rcs cascade. Upon cell envelope perturbations, osmotic changes, or other environmental signals, RcsF interacts with IgaA, which favors phosphorylation of RcsD and RcsB. Phosphorylated RcsB forms dimers with accessory proteins, such as RcsA and binds to target promoters. Transcription of diguanylate cyclases is repressed while the phosphodiesterase hmsP is enhanced indirectly, reducing cyclic di-GMP and lowering Hms-ECM production. Transcription of yadE is increased, perhaps through direct binding to the RcsAB box in the yadE upstream region. *, in Y. pestis, rcsA and yadE pseudogenes ensure that when Rcs signaling is active, RcsAB homodimers do not form, but RcsB homodimers can activate hmsD transcription and Hms-ECM expression is maintained. Absence of YadE eliminates putative interference with Hms-ECM production or stability.

The Rcs cascade is very complex and can respond to many different signals, including lipopolysaccharide and peptidoglycan perturbations (63, 64). The Csr system is also important in responding to outer membrane stresses, potentially indicative of cross talk with the Rcs function (65). We previously reported an 8-fold decrease in RcsB protein abundance in a CsrB mutant strain (25). Although the details of Csr control of yadE are not clear at this time, loss of CsrB clearly enhances its expression. Recently, activation of RcsB by the Csr system has been demonstrated in Salmonella. It will be important to determine whether CsrB control of yadE occurs through the Rcs cascade.

Innate immune defenses, osmotic changes, or other factors present in the flea digestive tract induce global regulatory adaptations in Y. pestis (59). The flea gut environment supports upregulation of surface protein genes yadB and yapL as well as high levels of transcription of yadE. Some of these genes are required for flea gut blockage or resistance to innate immune cells in mammals (59, 66). At the same time, high levels of yadE expression may be disruptive to production, accumulation, or stability of the Hms-ECM. Hms-ECM is essential for proper biofilm formation and proventricular blockage, the major transmission mode of Y. pestis by rat fleas. Therefore, loss of yadE gene function may have been an additional key step in the divergence of Y. pestis from Y. pseudotuberculosis.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

Y. pestis KIM6+ and Y. pseudotuberculosis IP32953 were grown at 21°C in Luria-Bertani (LB) or heart infusion broth containing 0.2% galactose (HIG) (25). E. coli strains were grown in LB agar or broth at 37°C. Where required, kanamycin (30 μg/ml), chloramphenicol (10 μg/ml), or ampicillin (100 μg/ml) was added to the medium.

Deletion of yadE from Y. pseudotuberculosis IP32953 was accomplished through allelic exchange. The PCR primers used for creating the strains and plasmids for these experiments are listed in Table S2 in the supplemental material. Upstream and downstream regions adjacent to YPTB2394 were cloned by overlapping extension PCR (67) and inserted into plasmid pRE112 (68) to create pREΔyadE. The ampicillin resistance (bla) gene from pHSG415 (69) was then inserted into this plasmid, replacing the yadE coding region to create pREΔyadE::bla. After biparental mating with E. coli strain MFDλpir (pREΔyadE::bla) (70) as donor and Y. pseudotuberculosis IP32953 as recipient, double-crossover mutations were selected on medium containing 10% sucrose and ampicillin. The mutation was verified using PCR and Sanger sequencing (Eton Biosciences, Inc.).

To overexpress yadE, the YPTB2394 gene as well as the upstream 200-bp region were placed under the control of the tetracycline resistance promoter on plasmid pACYC184. PCR products for the yadE gene and the pACYC184 backbone were combined in an overlapping extension PCR mixture and transformed into E. coli DH5α to create plasmid pACYC-yadE. In a similar way, the pACYC-yadE::gfp reporter plasmid was created by amplifying the gfp-coding region from pUC18R6K-miniTn7T sig70c35_GFP (71), the YPTB2394 promoter region, and the pACYC184 backbone. These PCR products were purified and combined in an overlapping extension PCR mixture and transformed into E. coli DH5α. Plasmids containing igaA and rcsB (pJET-igaA and pJET-rcsB) were created by PCR amplification of the respective genes and cloning into the pJET1.2 plasmid (Thermo Fisher). All plasmids were sequence verified and transformed into Yersinia strains by electroporation.

Biofilm disruption assay.

Biofilm stability was measured as previously described (25). Overnight broth cultures of strains to be tested were transferred to polycarbonate track etched (PCTE) membrane filters (n = 4 or 5) on HIG agar plates. After 72 h of growth at 21°C, individual filters containing the biofilm samples were placed in 10 ml of phosphate-buffered saline (PBS). Tubes were shaken vertically at 200 rpm, and the optical density of dislodged cells was measured (A₆₀₀) at specific time points. The filters were then vortexed at high speed until biofilms were completely disrupted and the A₆₀₀ measured to determine the total biomass on the filters. The proportion of the total biofilm that had become dislodged at each time point was calculated. Enzymatic treatment of biofilms prior to disruption tests were performed with proteinase K (Sigma) or DNase (Ambion) at 5 mg/ml. Solutions of enzyme in PBS (100 μl) were placed directly on top of the biofilms and incubated at 21°C for 60 min.

Transposon mutagenesis and fluorescence-activated cell sorting.

A Tn5 transposon mutant library in strain IP32953 (pACYC-yadE::gfp) was created using the pRL27 donor plasmid as previously described (72). After plating the mating mixture on HIG agar containing kanamycin and chloramphenicol, ∼100,000 individual mutant colonies were suspended and washed in PBS and diluted to 10⁶ CFU/ml for fluorescence-activated cell sorting (FACS). A total of ∼4 × 10⁶ individual Y. pseudotuberculosis mutant bacteria were sorted using FACSAria fusion cell sorter (BD Biosciences) at the BYU Cell Sorting/Bio-Mass Spectrometry core facility. Cells exhibiting one standard deviation greater than the wild-type IP32953 strain in the GFP channel were collected, diluted, and plated on HIG agar containing kanamycin and chloramphenicol to grow single colonies. The plates were then incubated for 24 h. Approximately 1,000 colonies from these plates were examined individually using an LED MiniBlue transilluminator (IO Rodeo). From this pool, 48 colonies were confirmed visually to express high levels of GFP after regrowth and were selected for further verification using flow cytometry. The Tn5 insertion sites in these mutants were determined by arbitrary PCR and sequencing as previously described (58). Of the 48 colonies, 23 were found to have unique insertion sites as reported in Table S1 in the supplemental material.

Flow cytometry.

To measure yadE::gfp expression, bacteria containing the reporter plasmid were grown on HIG agar plates at 21°C for 24 h. For each strain, bacteria from individual colonies were suspended in PBS and the GFP fluorescence measured by flow cytometry. For Hms-ECM measurement on the surface of bacteria, fixed bacterial cells were incubated with fluorescein isothiocyanate (FITC)-labeled wheat germ agglutinin (Sigma) as previously described (25). GFP expression and lectin binding of individual cells were measured using a BD Accuri C6 flow cytometer and analyzed using FACSDiva software (BD Biosciences).

Statistical analysis.

Statistical analysis was performed using GraphPad Prism 6.0. The details for each test are provided in the relevant figure legend.