Study on the Promotion of Bacterial Biofilm Formation by a Salmonella Conjugative Plasmid and the Underlying Mechanism

To investigate the effect of the pRST98 plasmid, originally isolated from Salmonella enterica serovar Typhi (S. Typhi), on biofilm (BF) formation, we carried out in vitro experiments using S. Typhi, Salmonella enterica serovar Typhimurium (S. Typhimurium) and Escherichia coli (E. coli). We further explored the effects of pRST98 in vivo by establishing two animal models, a tumor-bearing mouse model and a mouse urethral catheter model. Moreover, we examined the relationship between the quorum-sensing (QS) system and pRST98-mediated BF formation. These studies showed that pRST98 enhanced BF formation in different bacteria in vitro. In both animal models, pRST98 promoted BF formation and caused more severe pathological changes. It was previously reported that Salmonella senses exogenous N-acylhomoserine lactones (AHLs) through the regulatory protein SdiA and regulates the expression of genes including the virulence gene rck, which is located on the virulence plasmid of some serotypes of Salmonella. In this study, we confirmed the locus of the rck gene on pRST98 and found that AHLs increased rck expression in pRST98-carrying strains, thereby enhancing bacterial adherence, serum resistance and bacterial BF formation. In conclusion, the Salmonella conjugative plasmid pRST98 promotes bacterial BF formation both in vitro and in vivo, and the mechanism may relate to the AHL-SdiA-Rck signaling pathway.


Introduction
Salmonella, a facultative anaerobic bacterium that has a broad range of hosts including humans, farm animals and plants, causes serious infection and thousands of deaths each year, posing a significant threat to humans.
A large outbreak of Salmonella enterica serovar Typhi (S. Typhi) infection occurred in the 1980s. Five hundred ninety-one strains were isolated from the blood of patients who had acute and severe clinical symptoms. It was shown that more than 80% of isolates were multi-drug resistant, which was attributed to a large plasmid (R plasmid) with a size of 159 kb, designated as pR ST98 , belonging to the IncC group ( Fig. 1) [1]. Our previous study showed that pR ST98 is a chimerical plasmid carrying genes responsible for drug resistance and virulence. The strains harboring pR ST98 were found resistant to trimethoprim, streptomycin, kanamycin, sulfonamide, neomycin, gentamicin, chloramphenicol, tetracycline, carbenicillin, ampicillin, and cephalosporin. It was confirmed in our previous studies that pR ST98 contains a DNA sequence homologous to the Salmonella plasmid virulence gene (spv), which was found in all pathogenic Salmonella spp. except S. Typhi. The sequence of the ORF (open reading frame) of spvR and spvB on pR ST98 shared more than 99% similarity with that of spvR and spvB on the virulence plasmid in Salmonella enterica serovar Typhimurium (S. Typhimurium) [2], indicating the presence and distribution of spv in Salmonella. Later studies demonstrated that pR ST98 increased the serum resistance of Salmonella, promoted S. Typhi survival in macrophages in vitro and decreased the LD 50 (50% lethal dose) values of S. Typhimurium in infected mice [3]. Recent studies in our laboratory found that pR ST98 had inhibitory effects on autophagy in macrophages, thus weakening the innate immunity of host cells [4][5]. In addition, pR ST98 is a conjugative plasmid that spreads easily among S. Typhi, S. Typhimurium, Escherichia coli (E. coli) and Shigella flexneri (S. flexneri) in vitro, and it was very easily transferred from S. Typhimurium to E. coli in mice [6]. Given these characteristics of pR ST98 , it is expected that this plasmid plays important roles in bacterial resistance against hostile immune factors and in causing aggravated infection.
Due to their significance in the food industry and in public health, bacterial biofilms (BFs) have become the focus of studies since their first description in 1978. A biofilm is a structured community of bacterial cells enclosed in a self-produced polymeric matrix adherent to abiotic or living surfaces. Bacterial BF formation is described in three phases: initial attachment, proliferation and maturation, and detachment [7]. It was reported that approximately 80% of bacterial infections are related to BFs [8]. In the transition to BF status, some characteristics of bacteria change, including their adherence, invasion, virulence, and resistance. Therefore, it is extremely difficult to eradicate BFrelated contamination using routine methods such as disinfectants [9][10]. Taking Salmonella BF as an example, Barker and Bloomfield found even when treated with cleaning products, Salmonella BF that developed in toilets could live up to four weeks after patients were cured of salmonellosis infections [11]. Bacterial BF formation during food processing has caused severe consequences in public health. The resistance against multiple antibiotics is greatly increased when Salmonella is enclosed in a BF [12], which makes BF-related diseases more difficult to treat or cure. The persistence of bacterial BFs on the surface of teeth damages the tooth enamel and induces an inflammatory reaction in the surrounding gums [13]. S. Typhi BFs formed on the gallbladder were reported to be associated with the occurrence of liver cancer [14]. In addition, bacterial BFs in medical implants such as indwelling catheters could led to severe consequences. Therefore, the effects of BFs on causing endocarditis and intraabdominal, pelvic, and urinary tract infections (UTIs) have been extensively studied [12].
It has been suggested that a conjugative plasmid could promote BF formation in E. coli and other bacteria. This phenomenon could be attributed to conjugative-plasmid related factors. It has been proposed that the conjugative pili act as adhesion factors at the early stage of BF formation [15]. Colanic acid, curli, and adhesion dynamics in E. coli all contribute to conjugative plasmidmediated BF formation [16][17]. Alvise et al. suggested that extracellular DNA (eDNA) is responsible for increased BF formation mediated by the conjugative plasmid TOL in Pseudomonas putida KT2440 [18]. Furthermore, type 3 fimbriae, encoded by the conjugative plasmid pOLA52, were implicated in conjugative plasmid-enhanced BF production in E. Coli [19][20]. However, very few factors conducted by conjugative plasmid were reported in Salmonella. Because the pR ST98 plasmid has the ability to conjugate, we explored the relationship between pR ST98 and BF formation in different Salmonella and E. coli strains by multiple methods in vitro, including violet dye staining, scanning electron microscopy (SEM) and confocal laser scanning microscopy (CLSM). Furthermore, two animal models were established to investigate the effects of pR ST98 on BF formation in vivo. One was a tumor-bearing mouse intravenously infected by S. Typhimurium x3337lux and x3337lux/pR ST98 (by the conjugal transfer of pR ST98 to x3337lux) [21]. Here S. Typhimurium was used as a surrogate for S. Typhi because S. Typhi only causes human infections, and no suitable model has been established for investigation of S. Typhi pathogenesis. S. Typhimurium is a facultative anaerobic bacterium that can survive both in tumor active areas and necrosis areas. In addition, S. Typhimurium is driven toward tumors through chemoattraction in infections. Three important receptors, the aspartate receptor, the serine receptor, and the ribose/galactose receptor, bind to compounds released by tumor and specifically attract S. Typhimurium to preferentially migrate to the tumor [22]. The other animal model was a mouse with a urethral catheter infected by E. coli K 12 W 1485 and E. coli K 12 W 1485 /pR ST98 (by the conjugal transfer of pR ST98 to E. coli K 12 W 1485 ) because E. coli is one of the most common microbes in nosocomial infections.
N-acylhomoserine lactones (AHLs) are signaling molecules of the quorum sensing (QS) system, which responds to bacterial population density and triggers some gene expressions. AHLs play an important role in BF formation. Though Salmonella does not produce AHLs, it synthesizes the signal molecule receptor SdiA, which responds to AHLs released by other bacteria [12]. Lee found that SdiA binds extracellular signals and affects BF formation in E. coli; however, no direct link has been found between AHLs and BF formation in Salmonella [23]. Encoding an outer membrane protein, the rck gene on the virulence plasmid of some serotypes of Salmonella was regulated by SdiA. It was found that the rck operon affects the expression of plasmid-encoded fimbriae, which were shown to be vital components of the extracellular matrix and to promote BF formation [24][25]. In this study, we investigate the effects of pR ST98 on BF formation and its interactions with the AHLs-SdiA-Rck pathway.

Bacteria and culture conditions
The bacteria used in our study were listed in Table 1. Bioluminescent strains of S. Typhi and S. Typhimurium were constructed by electroporation of the pBEN276 plasmid containing a constitutive lux expression cassette, and the lux expression cassette recombined within the bacterial chromosome according to reference [26]. The use of bioluminescent bacteria provides an effective tool in the detection of S. Typhimurium BF formation in vivo. These strains were grown to mid-logarithmic phase in Luria-Bertani (LB) medium at 37uC, with a shaking speed of 200 r.p.m. Ampicillin was added into the medium at a concentration of 100 mg/ml to maintain the stability of the pR ST98 plasmid in some strains. The bacterial population density was determined by measuring OD 600 values with a spectrophotometer.  Typhi used as representative strains that naturally harbored pR ST98 and were resistant to chloramp henicol, streptomycin, trimethoprim and sulphonamide, gentamicin, neomycin, kanamycin, cephalosporin ampicillin, carbenicillin and tetracycline; Lane 4, antibiotic-sensitive S. Typhi, which were plasmid free, and used as the negative control. doi:10.1371/journal.pone.0109808.g001 monolayer in RPMI1640 Medium (Sigma, America) supplemented with 10% (v/v) heat-inactivated fetal calf serum (Thermo Scientific, America). Six-to seven-week-old female BALB/c mice were purchased from the Experimental Animal Center of Soochow University.

Ethics statement
All animal experiments were approved by the Animal Experimental Committee of the Soochow University (Grant 2111270) and were in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals (NIH Guidelines).

Comparison of BF by crystal violet staining
Bacteria cultured overnight in LB medium were diluted to OD 600 0.4. BF formation in polystyrene microtiter plates was assayed as described by O9Toole & Kolter [27] with modification. Briefly, cells were grown in the wells of the microtiter plates in 200 ml of LB medium supplemented with 1% glucose for 72 h at 30uC. The medium was then removed and replaced by 200 ml of a 1% (w/v) solution of crystal violet. After incubation at room temperature for 15 min, the dye was removed, and the wells were washed thoroughly with phosphate buffered saline (PBS). Following drying, BFs were observed with inverted microscopy and imaged. To quantify the attached bacteria, the crystal violet was solubilized with 200 ml of 30% (v/v) acetic acid solution, and the absorbance was measured at 570 nm (i.e., OD 570 ) in an ELISA reader (Biotek). The experiment was repeated three times with each sample in 4 wells.

Observation of BF structure with CLSM
Bacteria were cultured in the 24-well polystyrene plates at 30uC for 72 h. The pellicles collected from the air-broth interface were placed on the microscope slides and stained with 0.01% Acridine Orange (AO). After sealed with 40% glycerine, the samples were observed with a Leica TCS-SP2 CLSM. Imaging was performed using the 40*/1.3 objective, and simulated three-dimensional images were generated with COMATAT software. The experiment was repeated three times with duplicate samples.

Detection of BF using SEM
The cultured pellicles were transferred to cover slips pre-coated with lysine, followed by fixation with 4% glutaraldehyde and postfixation with 1% osmic acid before dehydration with a graded series of tert-butyl alcohol dilutions (30 to 100%). After the critical point in drying, the samples were observed with an xL-20 scanning electron microscope (Philip, Netherlands).

BF formation in two different animal models in vivo
For the tumor-bearing BALB/c mouse model, each group of six was subcutaneously inoculated with 1610 6 CT26 cells at the preabdomen site. When the tumor reached a diameter of 5-8 mm, the tumor-bearing mice were injected intravenously with 1610 7 CFU of S. Typhimurium x3337lux or x3337lux/pR ST98 in PBS. In-vivo imaging was performed at 1 d, 2 d and 3 d postinfection (p.i.) using an FX Pro in-vivo imaging system (IVIS, DXS4000pro) to observe the injected bacteria in mice. Mice were sacrificed at 3 d p.i., and tumors, livers, and spleens were collected for SEM and colony forming unit (CFU) analysis.
For the urethral catheter model, polyethylene tubes (PE10 with inside and outside diameter of 0.28 mm and 0.6 mm, respectively) pretreated with 75% ethanol and UV sterilized for 12 h, were incubated with E. coli K 12 W 1485 or with E. coli K 12 W 1485 /pR ST98 for 1 d. Female mice in each group of six were anesthetized by injecting 10% chloral hydrate in the enterocoelia. The periurethral area was sterilized with 75% ethanol, and the pre-incubated PE10 tubes were gently inserted transurethrally. PE10 tubes, livers and kidneys were aseptically collected from sacrificed mice on 5 d and 8 d p.i., and washed with PBS. PE10 tubes were fixed in glutaraldehyde for SEM or stained with 0.01% AO staining Analysis of the mechanism of pR ST98 promoted BF formation by adherence assay HeLa cells were seeded in 24-well tissue culture plates at 10 5 cells per well and incubated at 37uC and 5% CO 2 for 12 h. Cells were infected with ST8, ST 8 -c-pR ST98 or ST 8 -DpR ST98 with an MOI of 100:1 in the presence of 1 mM C8-AHLs dissolved by DMSO (Sigma, America) or saline. The plates were incubated at 37uC with 5% CO 2 for 60 min, and the cells were washed three times with PBS before lysing with 200 ml 0.2% Triton X-100 for 30 min at 37uC. The supernatant was collected for CFU counting. Each bacterial strain was assayed in triplicate, and experiments were repeated twice.

Serum resistance
Serum collected from 5 healthy rabbits and guinea pigs was filter-sterilized. S. Typhi were cultured in LB for 16 h at 37uC, gradual diluted OD 600 value to 1610 4 CFU/ml. Then, 20 ml bacterial cultures were incubated with 200 ml serum plus 1 mM C8-AHLs or saline 2 h at 37uC. CFUs were enumerated to count the surviving bacteria. The experiment was repeated twice with triplicate samples.

PCR and sequencing of rck gene
Genomic DNA was extracted from ST8, ST 8 -c-pR ST98 and ST 8 -DpR ST98 by boiling. PCR was performed using primers rck-F: 59-GTTGTATCCCGGCATGCTGA-39 and rck-R: 59-ATATTGCCCAGAGCCGGATAGAG-39 [28]. to detect the rck gene located on pR ST98 . Then, the gene was linked to the pEJT1.2 plasmid and transduced into E. coli TOP10. The rck gene was sequenced.

RT-PCR of rck gene
Total RNA extraction was performed using the Total RNA kit I (OMEGA bio-tek, America). The samples were centrifuged at 4000 r.p.m. for 10 min, and the supernatant was discarded. The pellet was resuspended in 100 ml lysis buffer (50 mg/ml lysozyme in Tris-EDTA buffer) and incubated at room temperature for 7 min. The subsequent steps of the RNA purification were performed according to the manufacturers' instructions. The quality of the isolated RNA was assessed via gel electrophoresis (PowerPac Basic, America). RNA concentrations were determined using the NanoDrop System (Thermo Scientific, America). The expression of the rck gene was determined by SuperScript TM III platinum One-Step Quantitative RT-PCR System (Invitrogen, America) according to the manufacturers' instructions. The reaction solution contained 25 ml of 26 reaction mixes, 1 ml of TaqMix, 0.2 ml of specific primers, 2 ml of mRNA, and 21.6 ml of DEPC water. Reactions were performed on a PCR system (MJ Research, America). cDNA was first produced in the RT step with 50uC for 15 min, followed by a DNA amplification step at 95uC, 5 min for denaturing, and 35 cycles (95uC for 40 s, 55uC for 30 s and 72uC for 115 s). The DNA product was observed and analyzed by gel electrophoresis and an automatic gel imaging analysis system (Syngene, UK). The primers used in this experiment were rck-F and rck-R.

C8-AHLs on BF formation
ST 8 lux, ST 8 -DpR ST98 lux and ST 8 -c-pR ST98 lux were cultured in the 24-well polystyrene plates at 30uC for 24 h adding 1mM C8-AHLs in the experimental group and 1mM saline in the control group. The media were then removed and washed thoroughly with PBS for 3 times. BFs were observed with IVIS.

Statistical methods
Data among groups were compared by three independent analyses, using an unpaired two-tailed Student t test, a one-way ANOVA, and a SNK-q (Student-Newman-Keuls) analysis. Among all the analyses, a p value ,0.05 was considered statistically significant. All the experiments were repeated three times with duplicate samples.

The promotion effects of pR ST98 on BF formation in different bacteria in vitro
To study the effect of the plasmid pR ST98 on BF formation in different strains, several methods were employed, including crystal violet staining, CLSM, and SEM. Including S. Typhi ST 8 , S. Typhimurium x3306 (the bioluminescent S. Typhimurium strains were also studied), E. coli K 12 W 1485 and their derivatives, three groups of bacteria were used in the crystal violet staining method to compare their ability to form BFs. For the intra-group comparison in the ST 8 group, ST 8 and ST 8 -c-pR ST98 were found to develop thicker BFs than ST 8 -DpR ST98 (p ,0.05) (Fig. 2A).
Similarly, E. coli K 12 W 1485 /pR ST98 had a stronger ability to form BFs than E. coli K 12 W 1485 (p ,0.05). These results indicate that pR ST98 plays an important role in promoting BF formation. For the inter-group comparison, Salmonella developed thicker BFs than E. coli did, and the difference was even more significant when both Salmonella and E. coli harbored pR ST98 , suggesting that pR ST98 might enhance BF formation in Salmonella more strongly than in E. coli. Meanwhile, the lux gene was shown to have no effect on BF formation (data not shown), and there was no difference observed between x3306 and x3337 ( Fig. 2A to C).
Bacteria harboring pR ST98 developed slimy and viscous pellicles, while pR ST98 -free bacteria formed loose and less coherent BFs [13]. Tomography and three-dimensional reconstruction by CLSM showed that BFs in S. typhi ST 8 and ST 8 -c-pR ST98 were developed with 43.23 mm and 47.62 mm thicknesses, respectively, which were much thicker than that in ST 8 x3337 and x3337lux (Fig. 3).
SEM provides a detailed view of the connections in a bacterial community. Bacteria harboring pR ST98 significantly promoted BF formation as indicated by SEM, which showed that bacteria forming three-dimensional BF structures were embedded within denser matrices. However, the BFs of bacteria that did not harbor pR ST98 were discontinuous and discretely patchy (Fig. 4). These results corroborate those from violet staining and CLSM, suggesting that pR ST98 promotes BF formation in all of the tested bacteria, including S. typhi, S. Typhimurium, and E. coli.

pR ST98 promotes BF formation in different bacteria in vivo
To study the effect of the pR ST98 plasmid on bacterial proliferation and BF formation in vivo, we established two animal models, a tumor-bearing mouse model and a mouse urethral catheter model. Electrotransforming the bacteria with the lux gene made it possible to detect dissemination in tumor-bearing mice by a non-invasive method, and lux was shown to have no effect on bacterial growth. After intravenously infecting mice, S. Typhimurium quickly circulated within the blood in the bodies of the mice. It was found that x3337lux and x3337lux/pR ST98 accumulated preferentially in tumors detected by IVIS at 3 d p.i., and x3337lux/pR ST98 in tumor emitted stronger bioluminescence signals than x3337lux did, indicating that x3337lux/pR ST98 formed thicker BFs. The same load of x3337lux/pR ST98 was used to infect normal mice as a control, but no signal was observed at the desired sites (Fig. 5A), most likely due to the quick dissemination in the blood that was beyond the detection limit of IVIS. To further analyze the histological changes in infected mice and bacterial load, the tumor, livers and spleens were sterilely recovered based on the IVIS images at 3 d p.i. for SEM and CFU counting. Metastasis in livers and spleens by tumor cells, along with swelling organs, were found. The inflammation was more severe in the x3337lux/pR ST98 -infected group. Consistent with the results from IVIS, SEM showed that more x3337lux/pR ST98 was accumulated in tumor. The livers and spleens from mice infected with x3337lux/pR ST98 were loaded with more bacteria as well, indicating that the pR ST98 plasmid promoted bacterial spread and proliferation as well as enhancing virulence (Fig. 5B and C).
PE10 tubes pre-incubated with E. coli were inserted into the mouse urethras. The mice were still active at 5 d post-insertion.
Stable BFs of E. coli K 12 W 1485 /pR ST98 or E. coli K 12 W 1485 developed on the surface of PE10 tubes were detected by CLSM after 5 d post-insertion under bright light. SEM, CLSM and CFU counting showed that the BFs formed by E. coli K 12 W 1485 /pR ST98 were thicker and had denser extracellular matrices compared with those in the control strain E. coli K 12 W 1485 (p ,0.05) (Fig. 6A to  C). However, the livers and kidneys recovered from mice showed no pathological changes at 5 d post-insertion. When the insertion was extended to 8 d, sluggish behavior appeared in all mice, and more severe symptoms were observed in the E. coli K 12 W 1485 / pR ST98 group demonstrated by abdominal dropsy, swelling in livers and kidneys, and punctate lesions. The symptoms induced by E. coli K 12 W 1485 /pR ST98 BFs showed further histological changes in livers and kidneys by H&E staining, including inflammatory cell infiltration and severe damage in the hepatic lobule and the glomerular structure (Fig. 6D). At 12 d postinsertion, most of the mice infected with E. coli K 12 W 1485 /pR ST98 died, while the mice with E. coli K 12 W 1485 infection survived longer than 17 d after insertion.
3. C8-AHLs enhances bacterial adherence, resistance, rck locus and transcription and bacterial BF formation AHLs, signaling molecule of the QS system, were shown to effect the BF formation in E. coli and the bacterial adherence [16]. To determine whether AHLs have similar effects on the BF formation in Salmonella, bacterial adherence assays were performed in the ST 8 group treated with C8-AHLs. It was found that ST 8 and ST 8 -c-pR ST98 displayed higher adherence rate than ST 8 -DpR ST98 (p ,0.05), while no difference was observed for the adherence rate between ST 8 and ST 8 -c-pR ST98 (p. 0.05). As for the control group (treated with saline), the adherence of the three strains to HeLa cells was similar (p. 0.05). ST 8 and ST 8 -c-pR ST98 incubated with C8-AHLs showed more adherence than with saline (Fig. 7A). This result indicates that AHLs promoted bacterial adherence, on which pR ST98 may have an effect.
AHLs promote BF formation in E. coli, which subsequently increases bacterial resistance against hostile factors including serum. To investigate whether AHLs enhanced Salmonella resistance, a complement-mediated killing assay was performed. When incubating with rabbit and guinea pig serum, the survival rate of ST 8 and ST 8 -c-pR ST98 treated with C8-AHLs significantly increased compared to the survival in the control group treated with saline, suggesting that AHLs enhanced Salmonella resistance. Furthermore, pR ST98 was indicated to participate in this process because ST 8 and ST 8 -c-pR ST98 showed more resistance against killing by serum than ST 8 -DpR ST98 (p ,0.05). Meanwhile, no significant difference was observed between ST 8 -DpR ST98 treated with or without C8-AHLs for their survival in serum (p. 0.05) (Figs 7B and C).
It was reported that rck located on the virulence plasmid of some serotypes of Salmonella, whose expression is regulated by the AHL receptor, effects the expression of plasmid-encoded fimbriae. In this study, it was proven that the rck gene was located on pR ST98 (Fig. 8A). To measure rck expression in the presence of AHLs and its relationship with pR ST98 , the transcription of rck was measured in the ST 8 group treated with C8-AHLs or saline. RT-PCR results showed that rck was only expressed in ST 8 and ST 8 -c-pR ST98 strains treated C8-AHLs but not in strains treated with saline. rck was not detected in ST 8 -DpR ST98 treated with C8-AHLs or saline (Fig. 8B). These results indicated that the expression of rck was stimulated by C8-AHLs.
To further determine whether AHLs have effects on BF formation in Salmonella, BF formation assays were performed in the ST 8 lux group treated with C8-AHLs. Compared with the control group, C8-AHLs significantly enhanced bacterial BF formation. In the C8-AHLs group, ST 8 lux and ST 8 -c-pR ST98 lux emitted brighter fluorescence signals than ST 8 -DpR ST98 lux did, indicating that C8-AHLs promoted BF formation in ST 8 and ST 8c-pR ST98 (Fig. 9).

Discussion
In response to limited nutrients and stressful conditions, many microorganisms form BFs by secreting polymeric matrices to interweave individual cells and build structural communities on abiotic or living surfaces. Due to the significance of BF formation in increasing the resistance of bacteria against hostile environments, BFs have become a significant research interest in the medical, food and environmental fields.
Jean-Marc Ghigo first found that natural conjugative plasmids have the capability of promoting BF formation in E. Coli [15]. In addition, bacteria harboring conjugative plasmids developed thicker BFs than those not harboring such plasmids. However, the relationship between the conjugative plasmids in Salmonella and BF formation has not been studied.
The effects of pR ST98 on BF formation were explored in this study. Crystal violet staining, SEM and CLSM revealed that S. Typhi, S. Typhimurium and E. coli harboring pR ST98 formed thicker BF in vitro, compared with the isogenic strains not carrying pR ST98 . It was also observed that S. Typhimurium x3306 and x3337 had similar abilities to form BFs, which is inconsistent with the study of Teodósio JS et al [29]. We speculated the different plasmids and BF-producing systems may contribute to this inconsistency. We noticed that E. coli K 12 W 1485 /pR ST98 had a weak ability to form BFs compared with Salmonella strains harboring pR ST98 . This heterogeneity in BF formation may arise because the synthesis of extracellular polymeric substances (EPS) in Salmonella outcompetes that in E. coli in medium, as reported by Rong Wang et al. Regarding the heterogeneity in the promotion of BFs by conjugative plasmids, Røder HL et al. proposed that the different genetic backgrounds of the plasmidharboring hosts may account for different BF formation when the same plasmid was used [30]. Our previous study demonstrated that in different genera, the conjugal transfer conditions of the pR ST98 plasmid were different in vitro or in mice, and the resistance markers encoded by the same plasmid varied in different strains, which showed the diversity and complexity of the gene expression from the plasmid. Thus, the effects of BF formation by different plasmids in various hosts may demand specific analysis.
In animal experiments, a tumor bearing mouse model was used to study the effects of pR ST98 on BF formation in S. Typhimurium, which was used as a surrogate of S. Typhi because no animal model is available for S. Typhi infection. In the tumor-bearing mouse model, x3337lux/pR ST98 was found preferentially in tumors with a considerably larger amount than x3337lux. The observation that solid tumors are treatable via bacterial infection  was made previously [31][32]. Colonization of bacteria on solid tumors could cause growth retardation or even the complete elimination of the tumors [33]. pR ST98 promoting host bacterial BF formation may have a therapeutic potential in fighting against tumors. Furthermore, our invasion study in vitro proved that bacteria in BFs showed a lower invasion ability compared with the corresponding planktonic form (data not shown), which is consistent with the finding by Katja Crull et al. that BF-forming bacteria did not invade intracellularly in vivo after they established BFs. The intracellular invasion by Salmonella may be due to the differential expression of invasive genes on Salmonella pathogenicity island 1 (SPI-1) induced by BF formation [34].
Another animal model, a mouse urethral catheter model, was established to study the effects of pR ST98 in E. coli on BF formation in vivo. E. coli K 12 W 1485 /pR ST98 was found to form only discrete patchy BFs at 3 d post-implantation, while E. coli K 12 W 1485 was not detected in tubes until 5 d post-implantation (data not shown). E. coli K 12 W 1485 /pR ST98 developed denser BFs at 5 d post-implantation, in line with bacterial titers recovered from established BFs on tubes. No histological changes were observed in the livers and kidneys of either group. When the implantation with tubes pre-incubated with E. coli was extended to 8 d or beyond, more severe inflammation was observed. Significantly, S. Typhimurium x3337lux/pR ST98 caused more severe inflammation in organs than x3337lux did. A similar phenomenon was observed for E. coli K 12 W 1485 /pR ST98 and K 12 W 1485 . These results indicate that pR ST98 aggravates the infection by promoting BF formation. Recently Rong Wang and Victoria J. Savage et al. demonstrated that the BF increases horizontal transfer of multi-resistant conjugative plasmids to plasmid-free bacteria compared to planktonic bacteria [35][36]. Therefore, it seems that conjugative plasmids facilitate BF formation, and vice versa. Therefore, given the intestinal origin and the conjugative transfer of pR ST98 , interaction between pR ST98 and BF may make Salmonella infections worsen.
QS, a bacterial communication system, has been implicated in BF formation. To date, three types of Salmonella-associated QS signals have been described as AHLs, autoinducer-2 (AI-2) and autoinducer-3 (AI-3). However, the study on AI-2 and AI-3 revealed their minor roles in Salmonella BF formation in some conditions. While Salmonella does not produce AHLs, the AHL receptor SdiA was found in Salmonella to sense exogenous AHL signals to influence BF formation. A recent study revealed that the presence of SdiA enhances E. coli O157:H7 (O157) colonization and persistence in fecal shedding of the bovine large intestine, the prerequisites for developing a BF. Rck is a 17-kDa outermembrane protein encoded by the rck gene located on the virulence plasmid of Salmonella enterica serovars Enteritidis and Typhimurium. The expression of rck in both E. coli and S. Typhimurium confers bacterial resistance against complementmediated killing [37]. Rck is homologous to Yersinia enterocolitica Ail, which is capable of influencing bacterial adherence to epithelial cell lines [38]. We hypothesized that rck may influence BF formation. In the present study, it was proven that the rck gene was located on pR ST98 , and rck-containing pR ST98 and C8-AHLs enhanced the cellular adherence of bacteria harboring pR ST98 and increased bacterial resistance against serum by activating transcription of rck. In addition, C8-AHLs promoted BF formation in bacteria containing pR ST98 . These results partially explained the pR ST98 -mediated BF promotion.
The mechanism of the effects of conjugative plasmids on BF formation is certainly complex and reciprocal. It is not clear whether the reported explanations could be applied to this study, although the studies on the mechanism may provide some clues. Further investigations will be focused on the factors that contribute to pR ST98 -mediated BF formation and the mechanisms associated with the heterogeneity in BF formation.
Taken together, we demonstrated that the conjugative plasmid pR ST98 , which was isolated from S. typhi, can promote BF formation in intestinal bacteria such as S. Typhi, S. Typhimurium, and E. coli. Animal models showed that pR ST98 promotes BF formation in S. Typhimurium and E. coli. In attempting to investigate the underlying mechanism, we found that the transcription of rck located on pR ST98 is activated by C8-AHLs. Therefore, it is reasonable to conclude that pR ST98 promotes BF formation in its host bacteria through the AHLs-SdiA-Rck pathway. The relationship between the conjugative plasmid pR ST98 and BF formation could provide insights into the prevention and treatment of Salmonella BF-related disease and intestinal infection.