A starvation-induced regulator, RovM, acts as a switch for planktonic/biofilm state transition in Yersinia pseudotuberculosis

The transition between the planktonic state and the biofilm-associated state is a key developmental decision for pathogenic bacteria. Biofilm formation by Yersinia pestis is regulated by hmsHFRS genes (β-1, 6-N-acetyl-D-glucosamine synthesis operon) in its flea vector and in vitro. However, the mechanism of biofilm formation in Yersinia pseudotuberculosis remains elusive. In this study, we demonstrate that the LysR-type regulator RovM inversely regulates biofilm formation and motility in Y. pseudotuberculosis by acting as a transcriptional regulator of these two functions. RovM is strongly induced during growth in minimal media but strongly repressed in complex media. On one hand, RovM enhances bacterial motility by activating the expression of FlhDC, the master regulator of flagellar genes, via the recognition of an operator upstream of the flhDC promoter. On the other hand, RovM represses β-GlcNAc production under nutrition-limited conditions, negatively regulating hmsHFRS expression by directly binding to the −35 element of its promoter. Compared to wild-type bacteria, the rovM mutant established denser biofilms and caused more extensive mortality in mice and silkworm larvae. These results indicate that RovM acts as a molecular switch to coordinate the expression of genes involved in biofilm formation and motility in response to the availability of nutrients.

During their life cycles, bacterial pathogens must often transition rapidly between planktonic and sessile states to adapt to changing environmental conditions [1][2][3][4] . Sessile bacteria may form biofilms for protection from diverse environmental stressors, including conventional antimicrobial agents and immune reactions. These biofilms have significant impacts on bacterial virulence in chronic infections 5,6 . Understanding the molecular mechanism underlying biofilm regulation in pathogens is therefore essential to the development of innovative treatment strategies. The process of biofilm formation in response to environmental signals is dynamic and complex. Molecular switches including sigma factors, transcription factors, small regulatory RNAs, and secondary messengers control the transition between motile and biofilm-associated states by modulating mutually exclusive motility and adhesion-related biofilm matrix components such as amyloid curli fimbriae, cellulose, and β-(1 → 6)-poly-N-acetyl-D-glucosamine (β-GlcNAc) 4, 7-10 .
Among a remarkable range of pathogens, the transition from the planktonic state to a biofilm is mediated by one central switch, (3′-5′)-cyclic-diguanosine monophosphate (c-di-GMP), which reduces motility and promotes biofilm formation at high concentrations [11][12][13][14] . Additionally, CsgD is reported as an important switch, as it induces expression of the csgBAC operon, required for the production of curli fimbriae and cellulose, as well as the production of c-di-GMP 15 . Most regulators of CsgD and c-di-GMP concentration have also been identified as planktonic/biofilm switches, including McaS 16 , RpoS [17][18][19] , and Hha 20 . On the other hand, switches that activate dispersal of planktonic cells from a biofilm are also essential, as they permit bacteria to escape the confines of the biofilm and colonize new locations 5 . Unlike other molecular switches, CsrA, a small RNA-binding protein, activates biofilm dispersal by inhibiting the synthesis of β-GlcNAc via direct repression of pgaA translation 21 , while also enhancing motility by protecting the transcript of the flagellar master regulator FlhDC from degradation 22 .

RovM enhances motility via transcriptional activation of flhDC expression. Previously, RovM
(ypk_1559) was reported to enhance bacterial motility and flagellar synthesis in Y. pseudotuberculosis 24,32 , however, the underlying mechanism has not been identified. Since FlhDC is the master regulator of flagellar production, we sought to determine whether RovM enhances bacterial motility by altering the expression of the flagellar master regulator flhDC (ypk_1745-1746) in Y. pseudotuberculosis. To this end, we quantified the expression of flhD and flhC in the wild-type, the ΔrovM mutant, and the complemented strain during the late exponential phase. As shown in Fig. 1a, the expression of flhD and flhC in the ΔrovM mutant was significantly lower than in the wild-type and complemented strains. Similarly, expression of the P flhDC ::lacZ transcriptional fusion reporter, which monitors the expression of flhDC, was considerably lower in the ΔrovM mutant compared with the wildtype strain, whereas expression of the rovM gene in ΔrovM resulted in a marked increase in the expression of the P flhDC ::lacZ reporter (Fig. 1b).
To determine whether RovM directly regulates flhDC expression, we examined the interaction between RovM and the flhDC promoter using an electrophoretic mobility shift assay (EMSA). Incubation of a probe harboring the flhDC promoter sequence [−1 to −652] relative to the ATG start codon of the first open reading frame (ORF) of the flhDC operon with purified His 6 -RovM led to the formation of protein-DNA complexes, and the abundance of such complexes depended on the amount of RovM present (Fig. 1c). The interactions between His 6 -RovM and the flhDC promoter are specific because excessive unlabeled probe abolished the formation of the protein-DNA complex. However, an unrelated fragment amplified from the coding region of the flhDC operon could not disrupt the formation of such complexes (Fig. 1c). Next, we identified a protected region of DNA with high affinity to RovM extending from −242 to −268 bp upstream of the start codon of the flhD ORF using DNase I footprinting analysis (Fig. 1d,e). Our data demonstrate that RovM plays a crucial role in flagellar synthesis and motility by directly regulating the expression of flhDC. This conclusion was further supported by the finding that complementation of rovM eliminated the motility defects of the ΔrovM mutant but failed to restore motility of the ΔrovMΔflhDC double mutant (Fig. 1f).
RovM represses β-GlcNAc production under nutrition-limited conditions. We observed that liquid suspensions of the ΔrovM mutant grown in nutrition-limited M9 medium, but not those grown in nutrient-rich YLB medium, formed large aggregates that settled quickly when left standing (Fig. 2a). However, the wild-type strain formed fewer aggregates, which did not settle out of suspension when grown under the same conditions. Moreover, the formation of aggregates in the ΔrovM mutant could be rescued by providing the rovM gene in trans (Fig. 2a). To investigate this phenomenon in more detail, we used scanning electron microscopy (SEM) to visualize aggregates in late exponential phase cultures of the wild-type strain, ΔrovM mutant, and the complemented strain. As shown in Fig. 2b, the ΔrovM mutant formed large aggregates in which bacteria appear to be embedded within a web-like matrix. However, this matrix was not observed in the wild-type or complemented strains (Fig. 2b).
As polysaccharides are known to be a major component of bacterial aggregates, we reasoned that the ΔrovM mutant formed large bacterial aggregates due to over-production of polysaccharides. To test this hypothesis, we monitored the presence of polysaccharides in aggregates of the wild-type, ΔrovM mutant, and complemented strains using a Congo red (CR) staining assay. As predicted, the ΔrovM mutant exhibited a stronger CR-positive phenotype than the wild-type and complemented strains (Fig. 2c). Furthermore, treatment of the ΔrovM aggregates with metaperiodate, a chemical known to degrade polysaccharides by oxidizing the carbon atoms (3 and 4) bearing vicinal hydroxyl groups and cleaving their C-C bonds 36 , resulted in near-complete disruption of the aggregates (Fig. 2d). These results suggest that RovM might be involved in the repression of polysaccharide production. Consistent with the report that β-GlcNAc is the major polysaccharide in the extracellular matrix of Y. pseudotuberculosis 27 , the ΔrovM aggregates stained positively using a wheat germ agglutinin-rhodamine (WGA-R) conjugate (Fig. 2e), which is known to bind specifically to β-GlcNAc and its oligomers. To more Scientific RepoRts | 7: 639 | DOI:10.1038/s41598-017-00534-9 quantitatively assess the production of polysaccharides, we determined the total amount of β-GlcNAc produced by each strain using a 3-methyl-2-benzothiazoninone hydrazone (MBTH) assay 37 . The ΔrovM mutant produced twice as much β-GlcNAc as the wild-type strain, while the complemented strain produced the least β-GlcNAc (Fig. 2f). These results suggest that RovM suppresses β-GlcNAc production in Y. pseudotuberculosis through regulating the hms operon.

RovM directly represses hmsHFRS expression.
To verify the role of RovM in the repression of β-GlcNAc production, we investigated the effect of RovM on the expression of hmsHFRS (ypk_2241-2238), the gene locus known to be responsible for synthesis and translocation of β-GlcNAc in Y. pseudotuberculosis. To this end, we introduced a single copy of the P hmsHFRS ::lacZ transcriptional reporter fusion into the chromosome of the Y. pseudotuberculosis wild-type strain, the ΔrovM mutant, and the complemented strain. We then quantitatively assessed the LacZ activity of the resulting strains. As shown in Fig. 3a, P hmsHFRS ::lacZ promoter activity increased dramatically in the ΔrovM mutant grown in nutrition-limited M9 medium, and this increase was absent in the complemented strain. However, increased P hmsHFRS ::lacZ promoter activity was not observed in the ΔrovM mutant grown in nutrient-rich YLB medium. We also confirmed the negative regulation of hmsHFRS by RovM in M9 medium through qRT-PCR analysis, which revealed that expression of the hmsH and hmsR genes was enhanced approximately 4-to 8-fold in the ΔrovM mutant relative to the wild-type strain and the complemented strain (Fig. 3b). These data suggest that RovM represses β-GlcNAc production by negatively regulating hmsHFRS expression under nutrient-limited conditions.
To further investigate whether expression of hmsHFRS is regulated directly by RovM, we performed an EMSA assay. Incubation of His 6 -RovM with a 361-bp hmsHFRS promoter inhibited the mobility of the probe (Fig. 3c), which indicates direct binding of this protein to the hmsHFRS promoter. Furthermore, the amount of the protein-DNA complexes increased in response to increased levels of His 6 -RovM. The interactions between His 6 -RovM and the hmsHFRS promoter are specific since excessive unlabeled probe abolished the formation of the protein-DNA complex; similarly, an unrelated fragment amplified from the coding region of the hmsHFRS operon could not disrupt the formation of such complexes (Fig. 3c). DNase I footprinting analysis revealed a region protected from DNase I digestion extending from −245 to −279 bp upstream of the start codon of the hmsH gene (Fig. 3d). This region overlapped partially with the putative conserved −35 element identified by the program for prediction of bacterial promoters, BPROM (Fig. 3e). Collectively, these results indicate that RovM represses hmsHFRS expression by binding directly to its promoter. Consistent with these conclusions, deletion of rovM from the wild-type strain resulted in the formation of bacterial aggregates in M9 medium due to the over-production of β-GlcNAc. However, deletion of rovM from the ΔhmsHFR mutant caused bacterial aggregates not to form under the same conditions (Fig. 3f), further confirming the role of RovM in inducing bacterial auto-aggregation by directly repressing production of β-GlcNAc, encoded by hmsHFRS.

RovM inhibits biofilm formation by suppressing β-GlcNAc production.
Since the Y. pseudotuberculosis biofilm matrix is primarily composed of β-GlcNAc exopolysaccharide and the hmsHFRS gene locus is essential for biofilm formation in Y. pseudotuberculosis 27 , we hypothesized that RovM plays a role in Y. pseudotuberculosis biofilm formation. To test this hypothesis, we used the nematode Caenorhabditis elegans as a biotic surface on which to study biofilm formation by Y. pseudotuberculosis. Biofilm assays were performed using the Y. pseudotuberculosis wild-type, the ΔrovM mutant and the complemented strain labeled with the constitutive GFP-plasmid pKEN-GFP mutant3*. Biofilm severity indices were calculated 24 h post-infection. Each nematode was assigned a score between 0 and 3, with 0 representing the lowest level of biofilm formation and 3 representing the highest level of biofilm formation (Fig. S1a-d) 38 . These results revealed that ΔrovM formed more vigorous biofilms than the wild-type strain in this C. elegans model. Strikingly, there was no biofilm formation observed on nematodes infected with the complemented strain (Fig. 4a). This result was further confirmed by examining biofilm formation on abiotic surfaces (Fig. 4b). Consistent with the report that the hmsHFRS gene locus is essential for biofilm formation in Y. pseudotuberculosis, there was no biofilm formation observed on the nematodes infected with the ΔrovMΔhmsHFR double mutant (Fig. 4). Together, these findings indicate that RovM plays a negative role in biofilm formation via suppressing the production of β-GlcNAc exopolysaccharide. RovM attenuates bacterial virulence by repressing hms expression. RovM was previously reported to attenuate virulence by repressing the expression of rovA, which controls expression of virulence genes in Y. pseudotuberculosis 32 . Since biofilm formation is crucial for bacterial virulence, we hypothesized that the effect of RovM on Y. pseudotuberculosis virulence was also mediated by regulation of hms. We thus tested bacterial virulence by injecting larval silkworms with the wild-type strain, ΔrovM, ΔrovMΔhms, and ΔrovMΔflhDC. The ΔrovM mutant caused more than 60% mortality within 72 h of inoculation. Larvae infected with mutants lacking rovM and flhDC survived at a similar rate, while the wild-type bacteria were less virulent. Notably, mutations in both hms and rovM caused near-complete loss of the virulence to larvae (Fig. 5a), implying that regulation of hms is essential to the virulence of ΔrovM. This result was confirmed by orogastrically inoculating relevant bacterial strains into C57BL/6 mice (Fig. 5b). Together, these results indicate that in addition to repressing the expression of rovA, RovM attenuates bacterial virulence by repressing hms expression and biofilm formation.
RovM coordinates the planktonic/biofilm state transition. Because motility and adhesion are mutually exclusive, one can imagine that the production of extracellular polysaccharides might be hampered  during the dispersal stage. In support of this hypothesis, deletion of hmsHFR, which is required for the production of β-GlcNAc, leads to enhanced motility (Fig. 6a). To further test this hypothesis, we constructed a promoter-replacement mutant ΔP hms (P flhDC ) in which the hmsHFRS promoter was replaced by the flhDC promoter. As expected, the promoter-replaced mutant that allowed both flhDC and hmsHFRS to be controlled by the flhDC promoter exhibited dramatically reduced motility compared with the wild-type strain (Fig. 6a). In contrast, the biofilm developed on C. elegans by the ΔP hms (P flhDC ) strain was enhanced compared to the wild-type strain (Fig. 6b). These data suggest that coordinated regulation of flhDC and hmsHFRS is crucial for the transition between the planktonic and biofilm states.

Discussion
In this study, we revealed a new motility/biofilm switch, RovM, a LysR family transcriptional regulator that facilitates transition from biofilm to motility via enhancing the expression of flhDC and repressing the expression of hmsHFRS directly. RovM was recently shown to act as both an activator and repressor in fine-tuning the expression of the T6SS4-and AR3-dependent acid survival system 33 . Herein we demonstrated that RovM repressed the expression of the β-GlcNAc synthesis operon hmsHFRS by directly binding to a region overlapping the −35 element in the hmsHFRS promoter but activated the expression of the flagella master regulator operon flhDC by binding a region 61 bp upstream of the −35 element in the flhDC promoter. RovM has been shown to repress RovA expression by recognizing a binding region within the rovA promoter, which includes two palindromic sequences (−63 to −53: ATcaTTT-N 5 -AAAgaAT; −62 to −55: TCaTT-N 6 -AAaGA) with similarity to the conserved T-N 11 -A core motif of LysR-type binding sites 39,40 . Similar palindromic sequences were also identified in the RovM binding sites on the flhDC promoter (−186 to −199: TAAgT-N 3 -AaTTA) (Fig. 1e) and the hmsHFRS promoter (−38 to −23: ATAT-N 7 -ATAT) (Fig. 3e). Although a number of LysR-type regulators have been shown to act as both activators and repressors in other bacteria 41 , this is the first demonstration that a LysR-type regulator plays a role in inversely regulating the motility-and biofilm formation-related genes, depending on the localization of the binding site on the target promoter.
For pathogenic bacteria, biofilm formation was found to play pivotal roles in the pathogenicity of important pathogens Staphylococcus epidermidis 42,43 , Pseudomonas aeruginosa 44 and Salmonella enterica serovarTyphimurium 45 . RovM has been shown to repress the RovA-dependent expression of internalization factor invasion 32 . Accordingly, we found that deletion of rovM gene leads to hyper-biofilm formation and increase the virulence of Y. pseudotuberculosis to both silkworm larvae and mice. Intriguingly, ΔrovMΔhmsHFR double mutant failed to form biofilm and became almost completely avirulent in both silkworm larvae and mice infection models. The loss of virulence of ΔrovMΔhmsHFR can result from lacking the ability to form biofilm, suggesting that this decreased virulence was directly related to biofilm accumulation, and RovM affects Y. pseudotuberculosis virulence is partially dependent on regulation of hms genes. Therefore, this finding provided a new perspective for revealing the mechanisms of regulation of bacterial virulence by RovM in pathogenesis.
The expression of rovM is very high in minimal medium but strongly inhibited during growth in complex media 24,32 . Indeed, as a nutrient-sensing regulator, the expression of rovM was growth phase-dependent. Expression was observed in the exponential phase of growth, reached a maximum in the post-exponential-phase (Fig. S2a,b). And this nutrient-dependent expression has been shown to be controlled by CsrA. In minimal media, CsrA activates RovM expression, leading to repression of RovA 46 . The CsrA-RovM-RovA regulatory cascade was further shown to be regulated by the cAMP-Crp complex, which links nutrient availability to CsrA activity via activation of CsrC. Deletion of crp strongly affects the levels of CsrC and results in the strong upregulation of RovM and repression of RovA 46 . The regulatory activity of CsrA is antagonized by two small noncoding RNAs, CsrB and CsrC, which contain multiple CsrA binding sites that sequester CsrA dimers away from target mRNAs [47][48][49] . Thus, RovM forms a unique switch that inversely regulates motility and biofilm formation at the transcriptional level by sensing nutrient availability mediated by Crp.
Based on our results, we proposed a model in which the RovM acting as a switch controls Y. pseudotuberculosis state transition by inversely regulating motility and biofilm formation in response to nutrient status (Fig. 7). As nutrient availability is limited, CsrA is activated by the cAMP-Crp complex through reducing the amount of CsrB/C. RovM regulator activated by CsrA enhances bacterial motility by directly activating flhDC expression while inhibiting β-GlcNAc accumulation by directly repressing the expression of the hmsHFRS operon. Thus, RovM formed a switch modulates motility/biofilm transition on transcriptional level. In the meanwhile, RovM represses bacterial virulence through negative regulation of RovA and inhibition of biofilm formation as observed in Fig. 7.
However, in Y. pestis, cellular RovM level was found to change following a temperature shift from 37 °C (warm-blooded host temperature) and 26 °C (flea gut temperature) 35 , whereas RovM expression was not shown to be temperature-dependent in Y. pseudotuberculosis 32 . The plague bacillus Y. pestis is evolved from Y. pseudotuberculosis 1,500-6,400 years ago and transmitted by fleas 50,51 . During being transmitted by flea, Y. pestis is not toxic to fleas, whereas Y. pseudotuberculosis exhibits significant oral toxicity to the flea vectors of plague 52 . The highly induced expression of rovM in Y. pestis by sensing temperature signals plays an important signal in abolishing the bacterial toxicity through inhibiting the expression of the major virulence transcriptional regulator rovA 35 . Thus, the development of temperature-dependent rovM expression in Y. pestis facilitates its adaptation to the flea-borne transmission route.
Since the regulation of biofilm formation by Y. pestis is an important characteristic in its flea-borne transmission, developments have to get evolved changes to acquire efficient adhere to flea proventriculus during its evolution from Y. pseudotuberculosis. RcsA and NghA that strongly repress biofilm formation in Y. pseudotuberculosis, are thought to represent for anti-transmission factors due to loss of function in Y. pesis 53,54 . The acquisition of pCD1 virulence plasmid during evolution also results in the opposite phenotypes of β-GlcNAc production, even in different Y. pestis species 55 . The chaperone RNA-binding protein Hfq was reported to inhibit biofilm development in pCD1 deficient Y. pestis strain CO92 56 while enhances biofilm formation in pCD1-cured Y. pestis strain 201 35 . Similarly, RovM was found to activate biofilm accumulation in Y. pestis strain 201 35 , whereas the deletion of rovM gene in pCD1 deficient Y. pestis strain KIM6 + does not affect biofilm formation 57 . Notably, different from the behaviour of RovM from Y. pestis strain 201 35 which positively regulates biofilm formation through regulating hmsHFRS, Y. pseudotuberculosis YPIII represses biofilm formation through blocking hmsH-FRS transcription in a directly manner. This difference may be caused by genetic backgrounds and the presence of pCD1 plasmid. Furthermore, the RovM protein from Y. pestis strain 201 is found to possess four more Arginine residues in 35 th to 39 th position comparing with the RovM from YPIII (Fig. S3). The highly conserved "A 6 " motif in YPIII turns into "A 10 " in Y. pestis strain 201 58 , and this may also lead to a reversely functional alteration. During the evolution of Y. pestis, differential regulation of RovM by environmental signals, as well as functional mutation of RovM on biofilm formation were likely subject to strong Darwinian (positive) selection during the early adaptation of Y. pestis to the new transmission route. Altogether, this implies that the development on regulation behave of RovM regulator plays a key role during the evolution of Y. pestis into a flea-borne pathogen from Y. pseudotuberculosis.
In conclusion, as a distinct motility/biofilm switch, the RovM represses biofilm formation enhances motility directly depending on the localization of the binding site on their promoters. The regulation is subjected to precisely control of bacteria lifecycle in response to nutrient levels, and in turn modulates Yersinia pathogenicity more rigorously. During evolution from the progenitor Y. pseudotuberculosis to the deadly Y. pestis, the evolved change on RovM seems to be one of the key steps that are important for flea-borne transmission of Y. pestis. Totally, regulation of RovM allows immediately adaptation of bacteria to ever-changing environments and is crucial for efficiently regulating of bacterial virulence. Bacterial strains and growth conditions. Bacterial strains and plasmids used in this study are listed in Supplementary Table S1. E. coli strains were cultured in Luria-Bertani (LB) and Y. pseudotuberculosis strains were cultured in Yersinia-LB (YLB) broth or M9 medium as previous reported 33 . In-frame deletions were generated by means of the method described by Wang et al. 59 . Table 2. To construct the lacZ fusion reporter vector pDM4-P hmsHFRS ::lacZ, primers Phms-3F/Phms-3R were used to amplify the 503 bp hmsHFRS promoter fragment from Y. pseudotuberculosis genomic DNA. The PCR product was digested with SalI/XbaI and inserted into similarly digested pDM4-lacZ to produce pDM4-P hmsHFRS ::lacZ. pDM4-P rovM ::lacZ was constructed in a similar manner using primers P rovM -F/P rovM -R.

Plasmid construction. Primers used in this study are listed in Supplementary
The ΔrovM and ΔflhDC in-frame deletion mutant of Y. pseudotuberculosis were made in our previous study 33,38 . The plasmid pDM4-ΔhmsHFR (ypk_2241-2239) was used to construct the ΔhmsHFR in-frame deletion mutant of YPIII. A 761-bp upstream fragment and a 702-bp downstream fragment of hmsHFR operon were amplified using the primer pair hms-F/hms-MR and hms-MF/hms-R, respectively. The upstream and downstream PCR fragments were ligated by overlapping PCR. The resulting PCR products were digested with SalI and BglII and inserted into the SalI/BglII site of pDM4 to produce pDM4-ΔhmsHFR.
To replace the hmsHFRS promoter with flhDC promoter in Y. pseudotuberculosis, the plasmid pDM4-ΔP hmsHFRS (P flhDC ) was construct. A 957-bp upstream fragment and a 924-bp downstream fragment flanking hmsHFRS promoter were amplified with primer pairs P hms -1F/P hms -1MR and P hms -1MF/P hms -1R, and the flhDC promoter was amplified with primer pairs P flhDC -1F/P flhDC -1R. The upstream PCR fragment of hmsHFRS promoter and flhDC promoter fragment were ligated by overlap PCR to generate fragment PFP hms -P flhDC . The fragment PFP hms -P flhDC and the downstream fragment of hmsHFRS promoter were ligated by overlap PCR with primer pairs P hms -1F/P hms -1R to generate fragment PFP hms -P flhDC -PRP hms . The PFP hms -P flhDC -PRP hms fragment was digested with SalI/BglII and inserted into the SalI/BglII site of pDM4 to produce pDM4-ΔP hmsHFRS (P flhDC ).
To complement the ΔrovM mutant, primers rovM-F/rovM-R were used to amplify the rovM gene fragment including its native promoter from the YPIII genome. The PCR product was digested with BamHI/SalI and was inserted into similarly digested pKT100. For complementation and overexpression, plasmids pKT100-rovM was introduced into respective strains by electroporation. The integrity of the insert in all constructs was confirmed by DNA sequencing.
Scientific RepoRts | 7: 639 | DOI:10.1038/s41598-017-00534-9 Overexpression and purification of recombinant protein. To express and purify His 6 -RovM, plasmid pET15b-rovM was transformed into the E. coli transB(DE3) competent cells. For protein production, bacteria were grown at 37 °C in LB medium to an OD 600 of 0.5. The strains were then induced with 0.2-0.4 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) and cultivated for an additional 16 h at 22 °C. Harvested cells were disrupted by sonication and purified with the His·Bind Ni-NTA resin (Novagen, Madison, WI, USA), according to the manufacturer's instructions. Purified recombinant proteins were dialyzed in phosphate-buffered saline (PBS) overnight at 4 °C and stored at −80 °C until use. The purity of the purified protein was verified as >95% homogeneity based on SDS-PAGE analysis. Protein concentrations were determined using the Bradford assay 61 .
Construction of chromosomal fusion reporter strains and β-galactosidase assays. The lacZ fusion reporter vectors pDM4-P hmsHFRS ::lacZ, pDM4-P flhDC ::lacZ and pDM4-P rovM ::lacZ were transformed into E. coli S17-1λpir and mated with Y. pseudotuberculosis strains according to the procedure described previously 59 . The lacZ fusion reporter strains were grown in YLB or M9 broth and β-galactosidase activities were assayed with o-nitrophenyl-β-galactoside (ONPG) as substrate. The assays were performed in triplicate at least three times, and error bars represent standard deviation. Statistical analysis was carried out with Student's t-test.
Biofilm assay. Biofilm formation of Y. pseudotuberculosis strains (labelled with plasmid pKEN-GFP mutant3*) on C. elegans was assayed as described 62 . Biofilm accumulation was classed as level 0-3. The level of biofilm accumulation on C. elegans was denoted as the biofilm severity incidence and was calculated as previously described 63 : Biofilm severity incidence = {[∑(level X number of samples in this level)]/(highest level X total sample numbers)} × 100%. Biofilm formation on abiotic surface was assayed in 96-well polystyrene microtiter plates as previously described 64 . Congo red assay, EPS quantification and WGA-R staining assay. Two OD of an overnight culture grown at 26 °C in M9 medium was collected and washed using ddH 2 O 3 times. Subsequently, sediments were suspended with 0.4% congo red solution and incubated for 30 min at 37 °C. Nonspecifically bound congo red was removed by washing with 1 M NaCl for 20 min, and the bacteria were washed with ddH 2 O three times. Finally, the bacteria were suspended in 1 ml ddH 2 O. Extracellular polysaccharide was quantified by 3-methyl-2-benzothiazolone hydrazone hydrochloride (MBTH) method as previously described 65 . The presence of the β-GlcNAc in the extracellular of Y. pseudotuberculosis aggregates was demonstrated using a WGA-R conjugate as previously reported 27 . Electron microscopy. For field emission scanning electron microscopy, glass coverslips were coated with a poly-L-lysine solution, and the samples were fixed in 2% glutaraldehyde in cacodylate buffer. Dehydration was performed in a graded series of acetone concentrations (10%, 30%, 50%, 70%, 90%, 100%) on ice for 15 min for each step. Samples were then critical point dried with liquid CO 2 and covered with a gold film by sputter coating. Examination was performed with a field emission scanning electron microscope (Hitachi S4800).
Western blot analysis. Western blot analysis was performed as described previously 38 . Samples were resolved by SDS-PAGE and transferred onto polyvinylidene fluoride membranes (Millipore). The membrane was blocked in 5% (w/v) non-fat milk for 4 h at room temperature and incubated with primary antibodies at 4 °C overnight: anti-VSVG (Santa Cruz Biotechnology), 1:500; anti-RNA pol β (Santa Cruz Biotechnology). The membrane was washed three times in TBST buffer (50 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.4) and incubated with a 1:5000 dilution of horseradish peroxidase-conjugated secondary antibodies (Shanghai Genomics) for 1 h. Signals were detected using the ECL plus kit (GE Healthcare) following the manufacturer's protocol.
Mouse infections. All mice were maintained and handled in accordance with the animal welfare assurance policy issued by Northwest A&F University. Post-exponential phase Y. pseudotuberculosis strains grown in M9 medium at 26 °C, washed twice in sterilized PBS and used for orogastric infection of 6-8 weeks old female C57BL/6 mice using a ball-tipped feeding needle. For survival assays 3 × 10 9 bacteria of each strain were applied to different groups of mice (10/group) for one time, and the survival rate of the mice was determined by monitoring the survival everyday for 21 days 59 .
Scientific RepoRts | 7: 639 | DOI:10.1038/s41598-017-00534-9 Silkworm rearing and infection. The silkworm Bombyx mori (Nistari strain) was reared on mulberry leaves at 27 °C in 70% RH and a photo period of 13:11 (light:dark). Y. pseudotuberculosis strains were grown to post-exponential phase in M9 medium at 26 °C. The cells were collected by centrifugation at 8000× g. The pellet was washed twice in sterilized PBS and 5 × 10 8 bacteria were injected into each hemocoel of day-3 fifth-instar silkworm for one time. Each group contains 25 silkworms. And the survival rate of the silkworm was determined by monitoring the survival every 12 hours for 6 days 67 .
Statistical analysis. Statistical analyses were performed using paired two-tailed Student's t-test. Survival times were analyzed using Kaplan-Meyer curves and comparisons were performed using the Log-Rank test. Statistical analyses were performed using GraphPad Prism Software (GraphPad Software, San Diego California USA).