Global transcriptional regulation by cell-free sup of S. Typhimurium peptide transporter mutant leads to inhibition of intraspecies biofilm initiation


 Salmonella is a genus of widely spread Gram negative, facultative anaerobic bacteria, which is known to cause ¼th of the diarrhoeal morbidity and mortality globally. It causes typhoid fever and gastroenteritis by gaining access to the host gut through contaminated food and water. Salmonella utilizes its biofilm lifestyle to strongly resist antibiotics and persist in the host. Although the biofilm removal or dispersal have been studied widely, the inhibition of initiation Salmonella biofilm has not been studied much. This study was conducted to determine the anti-biofilm property of the cell-free supernatant obtained from a carbon-starvation inducible proline peptide transporter mutant (ΔyjiY) strain. Our study shows that Salmonella ΔyjiY culture supernatant inhibits biofilm initiation by regulating biofilm associated transcriptional network. This work demonstrates that highly abundant proteases like ecotin, HslV and GrpE cleave the protein aggregates, whereas global transcription regulators H-NS, FlgM regulate expression of SPIs and flagellar genes. Relatively low abundance of flavoredoxin, glutaredoxin, thiol peroxidase etc. leads to accumulation of ROS within the biofilm, and subsequent toxicity. This work further suggests that targeting these oxidative stress relieving proteins might be a good druggable choice to reduce Salmonella biofilm.

colonization by pathogenic Staphylococcus aureus 15 . In another study probiotic Escherichia coli Nissle (EcN) inhibited bio lm formation in pathogenic EHEC, by a secreted bifunctional (protease and chaperone) protein DegP 16 . There are various other modes of interspecies bio lm regulation such as bio lm dispersal proteins 17 , indole-mediated bio lm regulation 18 , quorum sensing molecules, and antimicrobial peptides 19 . The inhibition can also be inter-kingdom, such as Aspergillus bio lm inhibition by Pseudomonas culture supernatants 20 . While there are a few strategies to remove preformed bio lms 21 , the inhibition of bio lm initiation is comparatively less explored in human pathogens. Interestingly we found that Salmonella WT could not form bio lm when co-cultured with ΔyjiY strain. This study was designed to investigate the novel bio lm inhibitory activity of bio lm de cient strain Salmonella Typhimurium ΔyjiY.
In this study, we are reporting a novel mechanism of bio lm initiation inhibition by complex transcriptional regulatory network. This complex network causes adhesion impairment, agellar motility inhibition, cleavage of protein aggregates and quorum sensing inhibition. Further we have shown that STM ΔyjiY supernatant can also inhibit bio lm formation by E. coli. We have also shown that the supernatant treatment can impair the invasion of the pathogen in C. elegans gut, and thus reducing its virulence.

Salmonella ΔyjiY culture supernatant inhibits WT bio lm formation
In order to understand the bio lm inhibitory property of STM ΔyjiY culture supernatant, we inoculated WT bacteria in bio lm inducing low salinity media with or without the culture supernatant from different strains. The ΔyjiY bacterial culture supernatant exhibited the bio lm inhibitory property (Fig. 1A), which was absent in the supernatant of the strain where yjiY was trans-complemented in a plasmid (STM ΔyjiY:pQE60-yjiY). To check if the bio lm inhibitory property is speci c to ΔyjiY supernatant, we inoculated WT strain with bio lm de cient strain ΔcsgD (Fig. S1A). We observed the inhibition only in case of ΔyjiY supernatant indicating that the inhibitory molecules are unique to the ΔyjiY secretome, and are independent of the activity of CsgD. To further validate that the inhibition is a cell-free phenomenon, we inoculated WT bacteria along with either live cells of other strains along with spent culture media (coculture), supernatant-free cell pellets, cell free supernatant, or whole cell lysate, and quanti ed the bio lm on solid-liquid-air interface. Although in all the setups ΔyjiY showed signi cant inhibition on bio lm formation, the maximum inhibition was observed with ΔyjiY culture supernatant (Fig. 1B). The supernatants were concentrated using Amicon ultra lter device and total protein was quanti ed. We found that minimum bio lm inhibitory concentration (MBIC; minimum concentration of total protein required to inhibit bio lm formation by WT strain) falls between 15-20 ng protein/ml (Fig. 1C), therefore we used supernatant containing 20ng protein/ml of sup for further experiments. We did not nd any difference in the growth (Fig. S1B), suggesting the supernatant lacks any bactericidal or bacteriostatic property. To determine if the secretion of the inhibitory component(s), is dependent on the culture media, we grew the bacteria in a minimal media (M9 media supplemented with 0.5% glucose) that exerts nutrition stress and used the supernatant to treat WT bacteria. We observed that the bio lm inhibitory molecule(s) were active even in minimal media (Fig. 1D), suggesting that the production of the inhibitory molecule(s) is not dependent on the nutritional condition and it is an intrinsic property of the ΔyjiY strain. We have also checked the temporal expression or accumulation of the inhibitory component(s) by inoculating WT strain with culture supernatant harvested from 2, 3, 4 and 5 day old ΔyjiY culture. Our results suggest that the optimum concentration of the inhibitory component(s) is/are reached after 3 days of growth (Fig. S1C), which remains unchanged on 4-5 th days of growth. Therefore, we have used ltered culture supernatant from 3 days old ΔyjiY culture for further experiments.
ΔyjiY culture supernatant weakens the WT bio lm and interferes with cell structure Characteristics EPS components are cellulose (produced by bcsA encoded cellulose synthase), curli mbriae (encoded by csgAB), BapA and LPS 22 . We observed a signi cant reduction of the EPS bound congo red uorescence intensity ( Fig. 2A, S2A) and bio lm thickness (Fig. S2B, S2C) upon ΔyjiY sup treatment. We also found that ΔyjiY sup treatment signi cantly reduced the strength of the bio lm pellicle (Fig. S2D, S2E, 2B) than that of untreated or WT supernatant treated samples. SEM and AFM analysis of the bio lm surface showed that the ΔyjiY supernatant treated bio lms lack the characteristic dome shape of a proper bio lm (Fig. 2C, S2F). We also noticed that the median cell length increased upon ΔyjiY supernatant treatment (1.58+0.30 µm) as compared to the untreated WT cells (1.38+0.34 µm) (Fig. 2D, 2E, S2G), hinting towards the presence of multiple regulatory components in the ΔyjiY supernatant that can modulate multiple phenotypic effects.
The active molecule(s) is/are protein(s) Previous studies have shown that secreted components from some bacteria can inhibit bio lm formation by the wild type strain or closely related species 15,16,20 . In order to delineate the chemical nature of the inhibitory molecule(s) present in ΔyjiY supernatant, we treated the supernatant with chemical agents, such as a divalent cation chelator (EDTA), protease (Proteinase K), protease inhibitor (PMSF), RNase and DNase and quanti ed the bio lm inhibition. We found that upon pre-treatment with different concentrations of EDTA, the inhibitory property remained intact. We found that 10mM EDTA enhanced the bio lm formation with both treated and untreated WT strain (Fig. 3A). Since EDTA is known to chelate divalent cations and inhibit a few proteases at higher concentrations 23,24,25,26 , our data signi es the requirement of divalent cations and/or active proteases for inhibition of bio lm. Upon treating the supernatants with 20mg/ml Proteinase K, the ability of ΔyjiY supernatant to inhibit bio lm formation was signi cantly reduced, suggesting that the inhibitory molecule(s) are protein(s) (Fig. 3B). Since activity of many proteins are sensitive to even small changes in pH and temperature, we checked the activity of ΔyjiY supernatant at different pH and temperatures. Surprisingly, we found that the active component(s) is/are heat stable at both the temperatures (Fig. 3C) and stable over a wide range of acidic and alkaline pH (Fig. 3D), although there was a small reduction in bio lm inhibition at pH 9.0. As recent studies show that several small non-coding RNA regulate bio lm formation and other virulence traits in Vibrio cholerae and Pseudomonas aeruginosa 27, 28 , we tested the stability of the component(s) after treating the supernatant with RNase. Although the inhibition was lost upon RNase treatment and incubation at 37 o C, we found a similar loss of inhibition with only heating the supernatant at 37 o C (Fig. S3A), suggestive of the heat, rather RNase treatment, to be the reason for the loss of inhibition. We also fractionated the supernatant using Amicon 3k MWCO ultra lter device, and we found that the active component(s) of ΔyjiY supernatant is/are of >3kDa molecular weight (Fig. S3B), quashing the role of small molecules and ions in the bio lm inhibition by ΔyjiY supernatant. Since the inhibitory activity was abolished upon proteinase treatment, we attributed the inhibition to protein components and quanti ed the total protein for further experiments.
Active molecules inhibit bio lm only during the initial phases, and cannot disrupt mature bio lm In order to check the effect of the active molecule on mature bio lm, we treated matured bio lm pellicle with the supernatants and checked for dispersion. Our results showed that the ΔyjiY supernatant could not disrupt pre-formed bio lm, hinting towards the effect of the active molecule(s) at the bio lm initiation (Fig. 4A). To answer if the supernatant treated WT cells remain bio lm defective in the absence of the inhibitory molecules, the ΔyjiY supernatant treated WT cells were re-inoculated in bio lm medium without the supernatant, and monitored for pellicle formation. We observed that the inhibition is diminished in the absence of the supernatant (Fig. 4B), although the defect reappeared when these cells were re-treated with the inhibitory supernatant. Together our data suggests that the supernatant mediated initial molecular reprogramming is required for bio lm inhibition, and the inhibitory molecules do not alter the inherent bio lm forming ability of the WT cells.
The inhibitory molecules impair agella-mediated initial attachment to abiotic surface Salmonella bio lm formation can be divided into 4 distinct steps, (i) initial attachment to the abiotic surface, (ii) secretion of adhesive components, (iii) maturation and (iv) dispersion of bio lm upon relief of the stress 29 . Since the bio lm initiation was inhibited and retained beyond 72 hours, we checked the expression of bio lm-associated genes (csgD, bcsA, iC, invF, sopD) and virulence factors (phoP, sodA, mgtC) from untreated or treated WT cells, 72 hours post inoculation. We found csgD, bcsA, iC and invF expression was signi cantly downregulated in ΔyjiY supernatant treated WT cells (Fig. 4C). Since, agella-mediated initial attachment initiates bio lm formation 30, 31 , we investigated the effect of ΔyjiY supernatant on agellar motility. We observed that ΔyjiY supernatant treatment signi cantly reduced both swimming and swarming motility (Fig. 4D, 4E and S4A), indicating a agella-mediated motility defect. We further observed the downregulation of iC sets in as early as 4-6 hours post inoculation with the ΔyjiY supernatant (Fig. S4B). Interestingly, we observed an increase in the iC expression at 12 hour post inoculation. We reasoned the adhesion de ciency arises from the sequential effect of initial downregulation of iC, followed by an increase in planktonic population. SEM analysis showed that ΔyjiY supernatant treated bacteria had signi cantly less number of agella (Fig. 4F, S4C). Altogether these results indicate a defect in agella-mediated initial attachment of the bacteria, putting them at risk of bio lm de ciency.
ΔyjiY supernatant reduces adhesion and virulence of Salmonella both in vitro and in C. elegans gut Initial attachment of Salmonella on the host gut epithelial cells requires swimming through the mucus layer 32 . SPI1 encoded Invs, Sops and Sips are important in the initial invasion 33 . Since ΔyjiY supernatant treatment downregulated invF and sopD expression, we checked the infectivity of the supernatant treated cells in mammalian intestinal epithelial cells, Int407. Invasion assay shows that the ΔyjiY supernatant treated cells are defective in initial invasion, although these cells show signi cantly higher intracellular proliferation as compared to the untreated cells ( Fig. 5A and 5B), suggesting that the ΔyjiY supernatant only makes the cells invasion defective by inhibiting agella-mediated adhesion. In order to check the infectivity of the ΔyjiY supernatant treated cells in a systemic condition, we fed young adult C. elegans N2 worms with RFP-STM-WT bacteria and quanti ed the gut colonization. The micrograph images and CFU analysis show that ΔyjiY supernatant treated RFP-STM-WT cells were able to colonize the gut lumen when fed continuously, but not persist (Fig. 5C, 5D). Therefore, our data suggest that the ΔyjiY supernatant treatment impairs in vitro invasion and in vivo colonization.

The inhibitory molecules inhibit bio lm formation in closely related species
We next checked the effectivity of STM ΔyjiY supernatant on the bio lms of common human pathogens. We inoculated E. coli DH5α, Pseudomonas aeruginosa PA01, Klebsiella pneomoniae, and Staphylococcus aureus wild-type strains with STM WT, STM ΔyjiY supernatants, and quanti ed the bio lm. We found that the STM ΔyjiY supernatant signi cantly inhibited only E. coli DH5α bio lm (Fig. 5E). Therefore, we concluded that the ΔyjiY supernatant can cross-react with closely related enterobacteriaceae, E. coli, but not effective against the distant members like Klebsiella pneumoniae or a different family (Pseudomonas aeruginosa) or phylum (Staphylococcus aureus).

The active components are majorly global transcription factors that regulates multiple cellular processes
To further identify the bio lm inhibitory molecule(s), concentrated supernatants from WT and ΔyjiY culture were resolved on 10% SDS-PAGE. After colloidal CBB staining, we clearly visualized 14 differential bands in ΔyjiY supernatant as compared to WT supernatant (Fig. 6A). We next analysed the secretome samples in LC Q-TOF MS/MS. Among the total of 244 proteins, 188 proteins were present in both the supernatants in differential level, 38 proteins were detected only in the WT supernatant, and 55 proteins in ΔyjiY supernatant (Fig. 6B). These proteins may have resulted from cell lysis during growth or they were secreted in the supernatant via active secretion system. After careful data-mining, among the proteins found only in ΔyjiY supernatant (listed in supplementary table 1), probable transcriptional regulatory protein YebC, anti-sigma28 factor FlgM, a serine protease ecotin and transcription termination/antitermination protein NusG were selected for further analysis. Many transcriptional regulator (H-NS, Rnk, StpA), cold shock proteins (CspE, CspC), chaperones (GrpE), ATP-dependent protease (HslV), proteins related to oxidative stress and iron homeostasis (FldA, SodB, YdhD, Tph and Ftn, Bcp, Bfr, respectively) were differentially present in both the supernatants (listed in supplementary table 2). Since H-NS is a global transcription regulator involved in multiple cellular pathways, we further performed protein interaction analysis of H-NS by STRING. The network suggests that H-NS is involved in agellar motility, replication regulation, transport of outer membrane lipoproteins and assembly of LPS (Fig. 6C), many of which were altered upon ΔyjiY supernatant treatment. We also constructed the interaction map among the protein hits found in LC QTOF MS/MS analysis by STRING (Fig. 6D).

Discussion
Salmonella utilizes its bio lm formation ability to evade host defence, colonize in host, persist in asymptomatic host and for its transmission to a new host 34 . The transition from planktonic to bio lm mode of life depends upon various stress factors. The master regulator CsgD controls the production of EPS components such as cellulose, ta , curli mbriae etc. by regulating AdrA. BapA is important both for bio lm production and attachment to intestinal epithelium 35 . In this study we have shown how metabolic stress decides the fate of the infectious WT strain. From preliminary data, we found that co-culture of STM WT and ΔyjiY led to bio lm defect in WT strain. Although various studies have shown that interspecies and intra-species bio lm inhibitions exist in nature 15,20,36 , in Salmonella, this phenotype in novel. Therefore, we carried out experiments to understand the underlying mechanism.
In this study we found that the bio lm inhibition was unique to ΔyjiY sup and the inhibition was reversed when WT cells were treated with culture supernatant from complement strain (STM ΔyjiY:pQE60-yjiY). We also observed that the ΔyjiY sup had only bio lm inhibitory effect and not bactericidal property as observed from the growth analysis of the WT strain. Further, the exopolysaccharide cellulose was found to be sparse in the bio lm of ΔyjiY sup treated WT cells. Since only cellulose mutant Salmonella Typhimurium was pro cient in adhering to tumour cells 37 , our data imply the presence of multiple inhibitory factors involved in the inhibition by the supernatant. While incubation of the ΔyjiY sup at 37˚C was found to reduce the inhibition, heating at 65˚C and 95˚C enhanced the inhibition, indicating that the inhibitory molecules might be heat-shock proteins or high temperature inducible stress proteins. Fascinatingly we observed that the effect of the inhibitory molecules is temporary, and they do not cause any genetic change to alter the inherent bio lm forming competency of the WT cells. The ΔyjiY sup was also found to exert its anti-bio lm activity only when administered at the beginning of bio lm formation, suggesting a modi cation in the complex network through which bio lm develops. In the ΔyjiY sup treated WT cells, the expression of bio lm genes and virulence genes was reduced signi cantly. sodA downregulation suggests that these cells are prone to ROS assault. Previous study has shown that YjiY depletion upregulates mgtC leading to bio lm defect 14 . Interestingly, we observed downregulation of mgtC, implying that the bio lm inhibition by ΔyjiY sup and the inherent bio lm defect of ΔyjiY strain follow different mechanisms. The ΔyjiY sup treated WT cells showed signi cantly less invasion in Int407 cells. It was recently proposed that Salmonella persist in C. elegans gut by forming bio lm 38 . ΔyjiY supernatant treated WT cells showed a concomitant reduction in colonization in worms gut, suggesting an in vivo bio lm defect. We also observed absence of agella and other protein aggregates upon ΔyjiY sup treatment, leading to a defect in cell aggregate formation and bio lm initiation. While there was a temporal increase in iC in ΔyjiY sup treated WT cells, these cells exhibited defective motility after 72 hours of treatment, further validating the importance of agella-mediated motility in the initial attachment of the bacteria to the substratum. Interestingly, we found that the STM ΔyjiY sup effectively inhibited E. coli bio lm. Although Salmonella had diverged from E. coli by acquiring virulence-associated genes, they share many evolutionarily conserved cellular pathways 39,40,41 . Therefore this observation reiterates that the inhibitory effects are more profound among closely related species. In the proteomics analysis of the ΔyjiY supernatant, we speci cally detected three potential inhibitory candidates: proteinase K sensitive ecotin 42 , anti-sigma28 factor FlgM (a negative transcriptional regulator of class III agellar genes 43,44 ) and YebC (negatively regulates quorum sensing in Pseudomonas aeruginosa PA01 45 ). Abundance of FlgM correlates with the absence of agella in the ΔyjiY supernatant treated WT cells. Since, NusG works synergistically with the global transcriptional regulator H-NS that binds speci cally to AT-rich SPIs in Salmonella genome and represses the genes 46,47 . Interestingly, in E. coli, H-NS has been linked to cell cycle, since the cells attempt to optimize H-NS concentration by maintaining a constant ratio of H-NS to chromosomal DNA in the cell 48 , which might explain the increase in cell length after ΔyjiY supernatant treatment. HslV, a heat-inducible ATP-dependent protease subunit of a proteasome-like degradation complex 49 and GrpE, which is involved in removal of protein aggregates leading to unsuccessful bio lm formation 50 were found in higher abundance in ΔyjiY supernatant. Doyle et al. 51 showed that excess GrpE inhibits the interaction between DnaK and regulatory protein, RepA. Since RepA helps maintaining copy number of plasmids in E. coli 52 , inactivation of RepA might explain the cell elongation upon ΔyjiY supernatant treatment.
Oxidative stress proteins such as avoredoxin, SodB, glutaredoxin, thiol peroxidase (Tph), thioredoxindependent thiol peroxidase (Bcp) were less abundant in ΔyjiY supernatant, among which Tph and SodB help reducing oxidative stress in STEC bio lm 53 . In E. faecalis, Tph is required for in vitro oxidative stress response and survival inside murine macrophages 54 . Iron availability has been shown to both positively and negatively regulate bio lm formation through complex network systems 55 . Although high Fe can lead to ROS production, Kang and Kirienko suggested that iron uptake and homeostasis is essential for successful bio lm formation in Pseudomonas aeruginosa 56 . Fe-storage proteins bacterioferritin (Bfr) and ferritin A (Ftn), required to prevent ROS generation via Fenton reaction and DNA damage, are functionally very large proteins with a core to accommodate 3000 Fe atoms. Fe-rich conditions induces E. coli FtnA and Bfr due to loss of repression by small RNA RyhB 57 . Similarly, Salmonella Typhimurium Bfr is involved in reducing intracellular Fe toxicity and absence of Bfr causes increased intracellular free Fe 2+ ion and oxidative stress 58 .
Therefore, we conclude that ΔyjiY supernatant treatment inhibits the bio lm formation by WT in three major ways-(i) NusG-HNS mediated transcriptional repression of AT-rich SPI-encoded genes, making the WT bacteria less virulent; (ii) FlgM mediated downregulation of class III agellar genes, impairing agellamediated initial attachment to abiotic surfaces, and (iii) high abundance of proteases and proteolytic molecules, hindering cell-cell adhesion, cellular aggregate formation in the EPS matrix. We reasoned the lower abundance of redox homeostasis proteins and ferritin-like molecules in ΔyjiY supernatant might facilitate accumulation of toxic oxidative species, causing cellular stress and toxicity. In this light, these oxidative stress relieving proteins can be exploited as potential druggable targets to inhibit Salmonella bio lm initiation.

Bacterial strains
All Salmonella Typhimurium strains used in this study are listed in Table 1 with their genetic description. Salmonella enterica serovar Typhimurium strain 14028S was used as the wild type strain, and was also the parental background for all the mutant strains used in this study, i.e. ΔyjiY, ΔcsgD, Δ iC and Δ iC Δ jB. All strains were grown and maintained in Lennox broth (LB; 0.5% NaCl, 1% casein enzyme hydrolysate and 0.5% yeast extract) at 37˚C under shaking conditions. STM GFP, STM mCherry (RFP-STM-WT) and STM ΔyjiY:yjiY were cultured in Lennox broth with 50µg/ml Ampicillin at 37˚C in shaking condition. E.coli DH5α, Staphylococcus aureus, Pseudomonas aeruginosa and Klebsiella pneumoniae were grown in Lennox Broth at 37˚C in shaking condition.
List of strains used in this study.

Preparation of cell free supernatant
To 2 mL of bio lm media, 1:100 dilution of overnight grown culture was added in at-bottom 24-well polystyrene plate (Tarsons). The plate was incubated at 28˚C without shaking for 72 hours. After 72 hours, the media from each well was collected, centrifuged twice at 10000 rpm for 15 minutes, the supernatant was collected and ltered using a 0.2µm lter and stored at -20˚C. To collect supernatant from M9 (0.5% glucose) media, the cultures were grown at 37˚C in shaking conditions. After 72 hours, the supernatant was collected as described above.

Bacterial Bio lm Inoculation with culture supernatant
To

Crystal Violet Staining
To quantify the bio lm at the solid-liquid interface, crystal violet (CV) staining was carried out. Brie y, the protocol followed for bio lm formation was the same as mentioned above. After 72 hours, the media in each well was discarded and the plates were thoroughly washed with RO water to remove all planktonic cells. The plates were then dried and 2 mL of 1% CV was added into each well. After 15 minutes, the CV was removed, and the excessive stain was thoroughly rinsed with RO water. The stained bio lm was destained with 70% ethanol and the intensity of color of the destained solution was quanti ed at OD595 in Tecan plate reader (In nite Pro 200). The absorbance was plotted in GraphPad Prism 6 and signi cance values determined using Student's t-test or two-way ANOVA.

Confocal microscopy
Sterile square coverslips (18 mm) were placed in at-bottom 12-well polystyrene plate (Tarsons). Cultures were inoculated for bio lm formation as mentioned previously. After 72 hours, bio lm appeared on the coverslip at the liquid-air interface, in the form of a thin line spanning the width of the coverslip (18 mm). The coverslip was washed thoroughly with water to remove planktonic cells and stained with Congo red (20 mg/ml in water) for 20 min at room temperature. After washing with water, the coverslip was mounted on a slide and imaged for bio lm distribution, with a laser scanning confocal microscope (Zeiss LSM 710) using a 40x objective. Z stacks were taken to generate a three-dimensional image. The MFI of the images were calculated using the ImageJ software. The MFI and thickness of the bio lm were plotted using GraphPad Prism 6. Single layer of cells were imaged and cell length of ~1000 cells from each coverslips was measured using ImageJ.

Glass bead assay
To test the strength of the pellicle at the air-liquid interface, medium sized (0.5mm to 1mm) glass beads were added one by one onto the pellicle. The initial weight of the glass beads was noted, and the number of glass beads added until the pellicle just collapsed was noted down and plotted using GraphPad Prism 6.

Scanning electron microscopy
Bio lm was allowed to form on coverslips as mentioned in the previous section. After thorough washing with water, the sample was xed in 2.5% gluteraldehyde for 48 hours at room temperature. Excess gluteraldehyde was removed by washing with water and the sample was dehydrated by gradient washes in increasing concentrations of 30%, 50%, 75%, 85% and 95% ethanol. The coverslips were then air dried under vacuum before coating with gold (JOEL-JFC-1100E ion sputtering device) for imaging by scanning electron microscope. For checking agellar morphology, 20 µl of STM WT overnight culture and 20 µl of supernatant treated STM WT were smeared on autoclaved 18mm coverslips, air dried, and processed using the abovementioned protocol. Flagellar structure was imaged using eld emission-SEM (FEI Sirion, Eindhoven, The Netherlands) scanning electron microscope.

Atomic Force Microscopy
Sterile 18 mm square coverslips were placed in at-bottom 12-well polystyrene plates (Tarsons) and bio lm inoculation was done. After 72 hours, the coverslips were removed and washed with sterile MilliQ water. The coverslips were dried and AFM analysis was done using XEISS AFM systems and was analyzed using XEI software.

Supernatant conditioning
The supernatant was treated with different concentrations of EDTA (2.5 mM, 5 mM, 7.5 mM and 10 mM) and incubated at 65˚C for 1 hour. To 200 µl of supernatant, 10 µl of proteinase-K (NEB, stock 20 mg/ml) was added and incubated at 37˚C for 1 hour. The proteinase was inactivated with 0.5 mM PMSF and incubated at 28˚C for 1 hour. The supernatant was heated to 37˚C, 65˚C or 95˚C for 15 minutes and immediately frozen at -20˚C. The pH of the bio lm inducing media as well as that of the supernatant were adjusted using concentrated HCl and 10N NaOH to obtain pH of 4, 7.4, and 9. The supernatants were also treated with 10 µl RNase (stock 1 mg/ml) for 1 hour at 37˚C. Post treatment, the supernatants were used to inoculate bio lm in order to check the activity of the inhibitory molecule(s).
Preformed bio lm disruption STM WT was allowed to form bio lm as described earlier. 72 hours post inoculation, the formed bio lms were treated with 20, 50 or 100 µl (1% (v/v), 2.5% (v/v) or 5% (v/v), respectively) supernatants and was incubated at 28˚C for 72 hours and bio lm was quanti ed using CV staining.
Recovery of bio lm formation STM WT was allowed to form bio lm in the presence of supernatants as described earlier. 72 hours post inoculation, the cells were harvested by centrifuging the culture at 10000 rpm for 15min. 20 µl of these treated or untreated cells were inoculated in fresh bio lm media (with and without supernatants for retreatment or recovery, respectively) and incubated at 28˚C for 72 hours and bio lm was quanti ed using CV staining.

Supernatant concentrating and separation based on molecular weight
The supernatants from bio lm culture were harvested and lter sterilized as previously mentioned. 4 ml of the sups were transferred to the Amicon ultra lter device (Amicon® Ultra-4 centrifugal lter device, 3k MWCO, UFC800324) and centrifuged at 4000g for 30 min in a swing bucket rotor. The concentrated solute from the bottom of the lter device was collected by inserting a pipette.
Quantitative RT PCR STM WT was allowed to form bio lm in the presence or absence of supernatants as described earlier.
After 72 hours, the bio lm population as well as the planktonic cells were harvested by thorough pipetting and centrifugation. From these cells, RNA was isolated by the TRIzol method (Takara). cDNA was synthesized with reverse transcriptase (GCC Biotech). Quantitative PCR was carried out using SYBR Green Q-PCR kit (Takara).
List of oligonucleotides used in this study. In vitro motility assay 2 µl of bacterial samples (treated or untreated) were spotted onto the 0.3% agar plates supplemented with 0.5% yeast extract, 1% casein enzyme hydrolysate, 0.5% NaCl and 0.5% glucose (swim agar plates) or 0.5% agar plates supplemented with 0.5% yeast extract, 1% casein enzyme hydrolysate, 0.5% NaCl and 0.5% glucose (swarm agar plates). The plates were incubated at 37˚C and images were taken every 2 hours using a digital camera (Olympus). The diameters of the motility halos were measured using ImageJ. At least ve replicate plates were used for each condition, and statistical signi cance was calculated using Student's t-test.   Salmonella ΔyjiY cell free supernatant signi cantly reduces the bio lm biomass by reducing cell-cell adhesion. A. Representative images of bio lm formed on coverslips that were stained with Congo red and imaged using a confocal microscope to generate 3D images and quantify cellulose biomass. Scale is shown on the X-and Y-axes. B. The tensile strength of the bio lm was measured by glass bead assay. Weight of glass beads required to just sink the bio lm to bottom, was plotted (Data are presented as mean + SEM of 5 independent experiments). C. Representative scanning electron micrograph of bio lm formed on a coverslip. Scale bar is 10 µm. D. Cell length of the bio lm inoculated treated or untreated STM WT cells was measured using ImageJ, and plotted (Data are presented as mean + SEM of 1200 cells were measured from 3 independent experiments). E. Representative confocal images of bio lm cells showing a difference in cell length. Scale bar is 5 µm (1000-1200 cells were measured from 3 independent experiments for each treatment). Student's t-test was used to analyze the data.

Figure 3
The physicochemical property of the inhibitory molecules is proteinaceous. A. The cell-free supernatants were treated with different concentrations of EDTA and checked for their ability to inhibit bio lm formation by STM WT (Data are presented as mean + SEM of 3 independent experiments). B. The supernatants were treated with proteinase K, and heated to 37 ˚C for 1 hour, and checked for the bio lm inhibitory activity. PMSF was used to inactivate proteinase K after 1 hour, and was also used as a control (Data are presented as mean + SEM of 3 independent experiments). C. The supernatants were heated to 65 ˚C and 95 ˚C to check the thermostability of the active component(s), followed by bio lm inoculation with the treated or untreated sups (Data are presented as mean + SEM of 3 independent experiments). D.
The pH sensitivity of bio lm inhibitory action of the supernatants were checked. The pH of the supernatants as well as the bio lm media was made acidic (pH 4.0) or alkaline (pH 9.0) with concentrated HCl or NaOH, respectively, and checked for the activity of the active molecule(s) (Data are presented as mean + SEM of 3 independent experiments). Two-way ANOVA and Student's t-test were used to analyze the data.   showing presence of multiple differential bands in the supernatants (* represents differential bands, # represents bands of equal density). B. Venn diagram showing proteins present in the supernatants detected in LC QTOF MS/MS analysis, blue and red circles represent number of proteins found in STM