Extracellular DNA, cell surface proteins and c-di-GMP promote biofilm formation in Clostridioides difficile

Clostridioides difficile is the leading cause of nosocomial antibiotic-associated diarrhoea worldwide, yet there is little insight into intestinal tract colonisation and relapse. In many bacterial species, the secondary messenger cyclic-di-GMP mediates switching between planktonic phase, sessile growth and biofilm formation. We demonstrate that c-di-GMP promotes early biofilm formation in C. difficile and that four cell surface proteins contribute to biofilm formation, including two c-di-GMP regulated; CD2831 and CD3246, and two c-di-GMP-independent; CD3392 and CD0183. We demonstrate that C. difficile biofilms are composed of extracellular DNA (eDNA), cell surface and intracellular proteins, which form a protective matrix around C. difficile vegetative cells and spores, as shown by a protective effect against the antibiotic vancomycin. We demonstrate a positive correlation between biofilm biomass, sporulation frequency and eDNA abundance in all five C. difficile lineages. Strains 630 (RT012), CD305 (RT023) and M120 (RT078) contain significantly more eDNA in their biofilm matrix than strains R20291 (RT027) and M68 (RT017). DNase has a profound effect on biofilm integrity, resulting in complete disassembly of the biofilm matrix, inhibition of biofilm formation and reduced spore germination. The addition of exogenous DNase could be exploited in treatment of C. difficile infection and relapse, to improve antibiotic efficacy.


Results
Visualization and composition of C. difficile biofilms. Clostridioides difficile strain 630, a ribotype 012 (RT012) was grown on thermanox coverslips in BHIS broth under conditions that allowed biofilm formation. Using scanning electron microscopy (SEM), we show three distinct stages of biofilm formation: initial attachment of bacteria to an abiotic surface via bacterial appendages, through to early biofilm formation, where we observe formation of a matrix, culminating at late biofilm formation, in which the bacteria appear to be encased within an extracellular matrix ( Fig. 1 a and Supplementary Fig. S1 online). Late biofilms formed in static TC flasks, were detached after a 72 h incubation from the abiotic surface (bottom) by gentle agitation, fixed and transferred onto coverslips for SEM imaging (Fig. 1c). The morphology and structure of biofilms formed in static TC flasks (Fig. 1c) is similar to biofilms formed directly on coverslips in 24-well plates (Fig. 1b).
Structure of the biofilm matrix. Visualization by confocal microscopy of C. difficile biofilms produced by strains 630 and R20291 with a FilmTracer Live-Dead stain, revealed the presence of viable and non-viable cells, encased within an extracellular matrix ( Supplementary Fig. S2a online). Nucleic acid and less abundant protein components within the extracellular matrix were positively stained with Acridine Orange, DAPI and SYPRO-Ruby Matrix stain respectively ( Supplementary Fig. S2a online). Late biofilms were disrupted with the addition of DNase and visualized by DAPI staining (Supplementary Fig. S2b online). The concentration of eDNA alone and total DNA (intracellular and eDNA) was significantly less in biofilms formed by R20291 (12.7 ± 0.7 mg/biofilm) compared to 630 (18.5 ± 1.6 mg/biofilm) ( Supplementary Fig. S2c online).
The role of cell surface proteins in early biofilm formation. Clostridioides difficile encodes a number of cell surface proteins, either non-covalently anchored to the cell wall or covalently anchored to peptidoglycan via sortase SrtB [48][49][50][51] . A number of cell wall associated proteins are regulated by the secondary messenger c-di-GMP. We therefore investigated the impact of c-di-GMP and sortase substrates of C. difficile on biofilm formation.
The effect of the secondary messenger c-di-GMP on attachment and biofilm formation. C-di-GMP has been shown to promote biofilm formation 42,44,53 . In order to investigate whether elevated levels of c-di-GMP altered attachment, early biofilm (24 h) or late biofilm formation (72 h), we used C. difficile strain 630 harbouring a plasmid encoding the diguanylate cyclase gene dccA under the transcriptional control of an anhydrotetracycline (ATc) inducible promoter (P tet ) 49 . In line with Purcell et al. 44,57 , we observed a significant increase in early biofilm formation ( Fig. 3) with induction of dccA (p < 0.001 at 25 ng/mL and p < 0.01 at100 ng/mL ATc), which elevates www.nature.com/scientificreports/ c-di-GMP. There was no significant difference in the effect of c-di-GMP on late biofilm formation or initial attachment and no significant modulatory effect of exogenous anhydrotetracycline added to control cells, at any stage of biofilm formation (Fig. 3). This indicates that the significant increases in biofilm biomass observed during early biofilm formation were due to c-di-GMP induction.
The role of PPEP-1 cleaved cell surface proteins in early biofilm formation. Surface exposure of the sorted proteins CD2831 and CD3246 is tightly controlled by a c-di-GMP dependent pathway, via a type II riboswitch, summarized in Fig. 4A. There is an inverse transcriptional regulation of CD2831 and CD3246 with the metalloprotease PPEP-1, which is under the control of a c-di-GMP dependent type I riboswitch (Fig. 4A). Elevated c-di-GMP increases expression of surface localized CD2831 49 and CD3246 proteins 54,55 . Gene inactivation of CD2831 ( Supplementary Fig. S4 online) significantly reduced early biofilm formation (p < 0.05, Fig. 5B), which is complemented in trans with the addition of CD2831 under the control of an anhydrotetracycline (ATc) inducible promoter (Fig. 4B). In-frame deletion of PPEP-1 does not significantly alter early biofilm formation (Fig. 4C), however, over-expression of CD2831 or CD3246 in a PPEP-1 mutant significantly enhances early biofilm formation (p < 0.01, p < 0.05 respectively, Fig. 4C). This suggests that CD2831 and CD3246 are important in early (e) Duplicate late biofilm cultures were grown in tissue culture flasks, then detached by gentle agitation from the bottom of the tissue culture flask. One was left untreated and the other treated with 100 µg/mL DNase for 15 min to disrupt the biofilm matrix. Images of the tissue culture flasks were taken with a Cannon 600D SLR (mounted on a copy stand with lighting unit (Kaiser RS2) with a 50 mm prime lens). This was undertaken for C. difficile strains from all five lineages: 630 (RT012), R20291 (RT027), M120 (RT078), M68 (RT017) and CD305 (RT023). All error bars are SD. Statistical differences were assessed using linear regression analysis and significant differences are indicated *p < 0.05, **p < 0.01, ***p < 0.001. The role of additional cell surface proteins in early biofilm formation. Using gene inactivation mutagenesis, we studied the roles of other cell wall proteins in biofilm formation. CD3392 is a sortase substrate and is covalently anchored to the cell wall 52,54,55 . CD0183 contains the canonical (S/P)PXTG motif, but is not processed by sortase and is partially bound to the cell wall, presumably by non-covalent interactions 49,50 . CD3145 (CbpA) carries an LPXTG-like cell wall anchoring domain, but is not processed by sortase 52 , however it is expressed on the cell surface and binds collagens I and V with high affinity 58 . Mutation of CD3392 or CD0183 significantly reduces early biofilm formation (p < 0.01, Fig. 5a), which is complemented by CD3392 or CD0183 expression, in trans (Fig. 5a). Mutation of CD3145 (CbpA), also significantly reduced biofilm formation, however, this phenotype was not successfully complemented with expression of cbpA in trans (Fig. 5a).
We then asked whether CD3392 could be detected within the biofilm matrix (Fig. 5b). We separated the biofilm matrix from the planktonic fraction (culture supernatant) of biofilms grown in TC flasks (Fig. 6a). The biofilm matrix fraction was disrupted by DNase treatment and both the matrix and supernatant fractions were filtered to remove C. difficile cells prior to SDS-PAGE and western blotting. We identified CD3392 predominantly in the matrix fraction of C. difficile cultures, with none detectable in the planktonic fraction (Fig. 5b). CD3392 was not detected in the matrix from a mutant strain, but was present in large amounts in the matrix from a complemented strain, suggesting CD3392 may form part of the biofilm matrix.
Detection of matrix proteins. To identify further proteins in the matrix, biofilms were produced in TC flasks and the biofilm matrix fraction was separated from the planktonic fraction (culture supernatant) (Fig. 6a). The biofilm matrix fraction was disrupted by (1) vortexing prior to quantification by BCA assay, or (2) by DNase treatment prior to SDS-PAGE analysis. After disruption of the biofilm matrix, aliquots from both the matrix and supernatant fractions were filtered to remove C. difficile cells before analysis was undertaken. We performed LC-MS/MS on the four main bands identified by SDS-PAGE as unique to the filtered biofilm matrix from C. difficile strain 630∆erm (Fig. 6b). These four bands analysed were also present in the biofilm matrix of the CD3392 mutant and complement strains, indicating their potential importance in biofilm formation (Fig. 6). Analysis of these bands indicate the presence of multiple C. difficile proteins: 12 proteins in band 1, 5 proteins in band 2, 13 proteins in band 3 and 10 proteins in band 4 (Fig. 6c). The predominant band present in the matrix fraction (1) was a combination of the low molecular weight (LMW) portion of the major surface layer protein SlpA, with intracellular proteins, which include those involved in butyrate fermentation (Hbd, Buk and ThlA1), oxygen stress tolerance (Rbr), amino acid degradation (GluD) and the utilization of glutamate and proline by Stickland fermentation (PrdA) ( Fig. 6; Supplemenatry Table S1). Alongside these C. difficile proteins, bovine DNase used to disrupt the biofilm was also detected in Band 1. The other three bands unique to the biofilm matrix (2, 3 and 4) were composed of a combination of C. difficile proteins, including; cell surface proteins (FbpA, Cwp19, Cwp6, CD1131, SlpA), pathogenicity and adaption proteins (PotA and ModA (ABC transporters)), Rbr (Rubrerythrin) and DnaK (chaperone), a phage protein, and proteins involved in amino acid metabolism (proline), fermenta- In-vitro attachment, early biofilm and late biofilm formation in 24 well plates were quantified using crystal violet assays with strains of C. difficile 630 containing an empty plasmid (630 vector), or with a plasmid-based diguanylate cyclase (dccA) under control of a P tet (inducible) promoter. These were compared to a 630 control strain containing a plasmid in the presence of 100 ng/mL anhydrotetracycline (630 vector + ATc). Biofilm and attachment assays were performed with a minimum of biological quadruplicate. The concentration of ATc used to induce dccA is listed in the graphs (0, 25 or 100 ng/mL), strain labelled 630P tet dccA 0 is an uninduced vector control. Error bars are SD. Significant differences are calculated using linear regression analysis compared to the uninduced control strain 630P tet dccA 0, **p < 0.01, ***p < 0.001. www.nature.com/scientificreports/ tion (butyrate) and phosphoenolpyruvate-dependent sugar phosphotransferase system (PTS) (Supplemenatry Table S1; Fig. 6).

Scientific Reports
Biofilm disassembly and resistance to vancomycin. DNase both inhibited biofilm production and disassembled preformed biofilm in TC flasks and on coverslips, whereas Proteinase K partially disrupted preformed biofilms. In contrast, bovine serum albumin had no measurable effect on biofilm initiation and RNase did not influence biofilm integrity ( Supplementary Fig. S5 online). Therefore, we assessed the effects of the biofilm dispersal agents, DNase and Proteinase K on the vulnerability of vegetative cells and spores to the antibiotic vancomycin ( Fig. 7), by comparing the viable counts of vegetative cells and spores within intact and disrupted biofilms in the presence of vancomycin. We observed that vancomycin alone significantly reduced the viable vegetative cell count in intact biofilms (p < 0.01 COV = − 6.243) to 7.7% of the untreated level, whereas disruption with DNase (p < 0.01 COV = − 10.805) combined with vancomycin treatment reduced the vegetative cell count to 0.68% of the untreated control. However, treatment with proteinase K in combination with vancomycin  Table 2), which was less effective than the addition of vancomycin alone, suggesting that proteinase K may affect the structural integrity of the glycopeptide antibiotic vancomycin, thus reducing its potency. This indicates that the combined effect of DNase and vancomycin produced the most significant reduction in viable vegetative cell counts. As expected, vancomycin had no effect on spore viability irrespective of biofilm disruption, compared to the untreated biofilms ( Fig. 7a; Table 2). The addition of DNase alone had no effect on vegetative cells (p = 0.989, COV = 0.010) viability, however DNase had a small but significant effect on spore viability in the biofilm matrix (p < 0.05 COV = − 0.568) (Fig. 7b) and the planktonic fraction p < 0.01 ( Supplementary Fig. S6 online). The addition of proteinase K alone had no effect on the viability of vegetative cells (p = 0.897 COV = − 0.086) or spores (p = 0.109, COV = 0.437) compared to the untreated biofilms (Fig. 7b). However, proteinase K decreased the relative proportion of spores (p < 0.05, COV = − 0.494) in the presence of vancomycin ( Fig. 7a; Table 2) compared to the vancomycin treated biofilms, suggesting that proteinase K reduced the spores' viability only in the presence of vancomycin. This adds additional evidence to the protective nature of the biofilm and the efficacy of DNase to disrupt the biofilm and reduce spore viability. Additionally, when used in combination with a current first line therapy DNase promotes the efficacy of vancomycin to reduce the viability of vegetative cells.

Discussion
In this study, we identify several factors involved in biofilm formation in all five C. difficile lineages, as well as exogenous factors involved in biofilm disassembly, which enhance the efficacy of vancomycin, a first line anti C. difficile treatment (summarized in Supplementary Fig. S7 online). A greater understanding of biofilm formation, dispersal and inhibition in other pathogenic bacteria has reduced disease burden and prevented relapse of infections 59 . Many patients suffer from long lasting chronic infections, driven by bacteria persisting in biofilms or other protected niches, impenetrable or intrinsically resistant to antibiotic therapy 7,60 . The current crisis in widespread antimicrobial resistance has raised fears over an era of untreatable infections 61 . The purpose of a biofilm matrix is aligned within and between bacterial species to protect the cells encased therein from environmental stresses, antibiotics and the immune response 6 . Interestingly, several bacterial species modulate their biofilm matrix composition between (1) exopolysaccharide, (2) extracellular and cell surface proteins and (3) extracellular DNA (eDNA) 13 : The human pathogens S. aureus and P. aeruginosa produce multiple distinct biofilm matrix compositions 19,62 . In S. aureus, biofilm composition can be proteinaceous (cell surface proteins and teichoic acid) combined with eDNA or alternatively a specifically synthesized exopolysaccharide [β-1,6-linked N-acetylglucosamine (PIA/PNAG)] 45,63 , yet the clinical importance of these two variants is not completely understood 19 . Combinations of eDNA, proteins and polysaccharides have been identified surrounding adherent colonies of C. difficile, although the extent to which they contribute to biofilm formation and integrity remains to be determined 24,25 . In this study, we show that clinical strains representing each of the five C. difficile lineages RT012 (630), RT027 (R20291), RT023 (CD305), RT017 (M68) and RT078 (M120) produce biofilms on Figure 5. Cell wall proteins involved in biofilm formation. (a) Insertional inactivation mutants of non-c-di-GMP regulated proteins CD0183, CD3145 and CD3392 were created in C. difficile strain 630Δerm. Mutants were complemented by introduction of plasmids expressing the wild type gene. These mutants and their complements were assessed for their ability to form early biofilms (24 h) on an abiotic surface in vitro compared to a wild-type control. Crystal violet assays were performed with a minimum of 6 biological replicates. All error bars are SD. Statistical differences were assessed using linear regression to compared to all strains against their respective parent strain 630Δerm, *p < 0.05, **p < 0.01, ***p < 0.001. (b) Western blot using anti CD3392 antibodies of the matrix (M) and planktonic fractions (P) of a late biofilm, from strains 630Δerm, CD3392::CT and CD3392 complement (CD3392::CT (pCD3392)). The biofilm matrix was detached from the bottom of a TC flask, disrupted by the addition of DNase (100 µg/mL), then loaded onto an SDS-PAGE gel alongside a protein ladder (L) for analysis by Western blot. www.nature.com/scientificreports/ an abiotic surface, comprised of vegetative cells and spores encased within an EPS matrix comprised mainly of eDNA combined with intracellular and cell surface associated proteins. In P. aeruginosa, temporal and genetic control of eDNA release by cell lysis is essential in initial development of biofilm formation 18,20 and for many pathogenic bacteria, including P. aeruginosa, Enterococcus faecalis and L. monocytogenes DNA is a major component of the biofilm matrix 18, 21,64,65 . In this study, we show that biofilm biomass positively correlates with eDNA concentrations within the biofilm matrix. Representatives of three C. difficile lineages, RT012(630), RT023(CD305) and RT078(M120) form larger more robust biofilms with a higher eDNA concentration and an increased sporulation frequency within the biofilm matrix, compared to representatives of the other two C. difficile lineages, RT027 (R20291) and RT017(M68). We show that without DNA, a major component of the biofilm matrix in C. difficile, initial biofilm formation is severely inhibited, and late biofilms are disassembled in all five C. difficile lineages. Our data complements a study from Dubois et al., who identified eDNA as a component of C. difficile biofilms formed by strain 630∆erm in the presence of the secondary bile salt deoxycholate (DOC), which is present in the human gut and toxic to vegetative cells 29 . The structural importance of eDNA should not be underestimated, in other species eDNA aids bacterial adhesion, stabilizes the biofilm structure and contributes to antimicrobial resistance, by chelating cations and restricting diffusion of cationic antimicrobials 66 . Environmental conditions can trigger alterations in the abundance of exopolysaccharides and eDNA within S. aureus biofilms, however, even small quantities of eDNA are essential for biofilm integrity 67 . Herein, we demonstrate that although there are variations in eDNA concentration within www.nature.com/scientificreports/ the biofilm matrix of different C. difficile lineages, eDNA is a key factor in development and structural integrity of C. difficile biofilms across all five lineages. We detected a significantly higher spore titre in the biofilm matrix compared to the planktonic phase for four of the five lineages. C. difficile sporulation is linked to persistence of CDI 68,69 and is mediated by the sporulation master regulator spo0A, inactivation of which results in a significant reduction in biofilm biomass 23,25,29 , indicating a key role for Spo0A in biofilm formation. Biofilms enhance persistence for many chronic bacterial infections, providing protection from environmental stresses and antibiotics 60 . Therefore, we hypothesise that Spo0A may support C. difficile biofilm formation and persistence, by inducing sporulation and consequently cell lysis during stationary phase, resulting in a release of eDNA, which can be incorporated into the biofilm matrix. Vegetative cells and spore encased within these protective biofilms could be released to re-seed the gut and facilitate relapse of CDI.
Biofilm architecture is a key factor of recalcitrance (resistance to antibiotics), which affects drug diffusion, as well as initiation of the starvation and SOS response, which all contribute to biofilm stability 7 . Biofilms produced by different bacterial species can be artificially disrupted by the addition of exogenous polyamines, DNase, d-amino acids or alginate lyase depending of the structure and composition of the biofilm matrix 18,71-73 . To eradicate chronic infections combinatorial approaches that induce biofilm dispersal via enzymatic degradation (DNase, Dispersin A or glucanhydrolases) with current antibiotic therapy 7 enhances eradication of pathogens such as Streptococcus mutans and P. aeruginosa in cystic fibrosis patients 75 . Herein, we show that DNase successfully disassembles biofilms produced by all five C. difficile lineages and that degradation of eDNA within the biofilm by DNase, increases the effectiveness of vancomycin from 92.3 to 99.3% at reducing vegetative cell titre. We show DNase reduced the viability of C. difficile spores in both the biofilm matrix and planktonic fraction, measured by germination, which aligns with a study in B. subtilis, showing a combination of lysozyme and DNase damaged spore coat proteins affecting viability and germination 76 . Dapa et al. observed a decrease in biofilm integrity in preformed biofilms with both proteinase K and DNase, with extended incubation 25 and Dubois et al., found both DNase and proteinase K dispersed C. difficile biofilms induced in the presence of the secondary bile salt deoxycholate 29 . However, in our system, proteinase K only marginally reduced biofilm biomass, this potentially results from media differences, as Dubois et al., used BHISG growth media, known to preferentially promote exopolysaccharide production and reduce eDNA within a S. aureus biofilm 67 . More importantly, we show that proteinase K reduced the bactericidal effect of vancomycin on C. difficile, indicating that proteinase K should not be used synergistically with vancomycin. This highlights the merits of a dual approach for reducing C. difficile biofilm recalcitrance, in which DNase reduces both spore germination and enhances the efficacy of vancomycin.
In many organisms, particularly Gram-negative bacteria, elevated levels of the secondary messenger c-di-GMP promote biofilm formation by regulating many cellular systems, including biosynthesis of matrix polysaccharides in P. aeruginosa 59,62 , E. coli 77 and S. aureus 45 , whereas low level c-di-GMP inhibits biofilm formation 54,78 . Experiments were undertaken with a minimum of 3 biological replicates. The data was analysed in Excel and GraphPad Prism 7.0, and error bars represent SD. Statistical analysis to determine the effect of vancomycin on disrupted and intact biofilm was performed using linear regression, **p < 0.01 (grey) and the effect of vancomycin and vancomycin combined with DNase and Proteinase K compared to untreated biofilms *p < 0.05, **p < 0.01, ***p < 0.001 (black). www.nature.com/scientificreports/ Sub-inhibitory concentrations of certain antibiotics trigger an increase of c-di-GMP to promote biofilm formation in E. coli and P. aeruginosa 62,77 . C. difficile encodes multiple enzymes involved in the synthesis (diguanylate cyclases) and degradation (phosphodiesterases) of c-di-GMP 79 . In C. difficile, c-di-GMP inversely regulates aggregation and cell motility 57 , by modulating expression of the flgB flagella operon 43 and the type IV pili cluster 80 .
Herein, we show that c-di-GMP enhances early biofilm formation, corroborating observations from Soutourina et al. 53 and Purcell et al. 44,57 , however, c-di-GMP has little effect on attachment and late biofilm formation. Some cell wall anchored proteins in C. difficile, including CD2831 and CD3246 are regulated via a c-di-GMP dependent pathway, which increases transcription via a type II dependent riboswitch 54 . Under elevated c-di-GMP, CD2381 and CD3246 are covalently attached to peptidoglycan in the cell wall of C. difficile via a sortase enzyme (SrtB) 49 .
Conversely, under low level c-di-GMP, PPEP-1, a metalloprotease (CD2830/ZmpI) controlled by type I dependent riboswitch is upregulated and cleaves CD2831 and CD3246 from the cell wall 54 . We observed that the cell wall protein CD2831 enhances early biofilm formation in C. difficile and overexpression of CD2831 or CD3246 in a PPEP-1 deletion background mimics an elevation of c-di-GMP levels, which results in significantly increased biofilm formation. Similar to our observations, surface associated cell wall proteins regulated by sortase in S. aureus are linked to biofilm formation 47 . We speculate that increased biofilm formation under elevated c-di-GMP (in trans) in early biofilm formation could result from reduced cleavage of CD2831 and CD3246 from the cell wall, as well as promoting expression of c-di-GMP dependent genes, these include cell wall genes, PTS systems and ABC transporters, that were identified by McKee et al. as directly or indirectly regulated by c-di-GMP riboswitches 54 .
In line with our results, Arato et al. showed a contribution of CD2831 in biofilm formation 81 , whereas Poquet et al. 28 did not observe a difference in biofilm formation upon deletion of CD2831. However, Poquet et al., used BHISG media for biofilm formation, which as discussed, triggers environmental biofilm variants in S. aureus, with increased exopolysaccharide formation and reduced eDNA 67 . In some species, environmental conditions trigger a biofilm matrix preferentially composed of cell surface proteins alongside eDNA 47,82,83 . In addition to media, differences in biofilm composition in C. difficile observed between laboratories may be a result of variation in intrinsic c-di-GMP levels, which regulate the cleavage of CD2831 from the peptidoglycan via PPEP-1. To address this, we overexpressed CD2831 and CD3246 in a PPEP-1 deletion strain and showed a significant increase in biofilm formation. We suggest that temporal production of c-di-GMP positively modulates early biofilm formation, rather than increasing biomass of an established biofilm by modulating attachment of cell wall proteins to peptidoglycan, alongside upregulating PTS systems and ABC transporters identified by McKee et al. to modulate metabolism within a biofilm and downregulate motility, which favors a sessile lifestyle ( Supplementary Fig. S7  online). Alongside the c-di-GMP regulated cell surface proteins, we also identified non-c-di-GMP regulated cell surface proteins which contributed to early biofilm formation in C. difficile, some of which are covalently anchored to peptidoglycan via the sortase (StrB) 49,50,52 . We identified that deletion of the cell surface proteins CD0183; a putative cell wall hydrolase, CbpA; a collagen binding protein and CD3392; a predicted collagen binding protein, significantly reduced the biomass of early biofilms. Several sortase substrates of S. aureus, including FnbpA, FnbpB, ClfA and CflB, are known to promote biofilm formation 84 . Interestingly, methicillin resistant S. aureus (MRSA) forms biofilms with predominantly cell surface proteins and eDNA, whereas methicillin sensitive S. aureus (MSSA) forms biofilms with the PIA exopolysaccharide 47 . This highlights the role for a combination of c-di-GMP dependent and c-di-GMP independent cell surface proteins and eDNA within a biofilm matrix. We, like Semenyuk et al. 85 were unable to identify a specific highly produced exopolysaccharide in the C. difficile biofilm matrix. We observed the presence of c-di-GMP dependent and c-di-GMP independent cell wall proteins, PTS and ABC transporters, and intracellular proteins within the biofilm matrix of C. difficile, which contribute to the biofilm architecture alongside eDNA. There are many parallels identified between the proteins we detected in the bands unique to the biofilm matrix compared to those identified by Dubois et al. in response to DOC 29 and genes identified as upregulated in biofilm cultures by Poquet et al. 28 , including the cell surface proteins, amino acid metabolism and degradation, and fermentation, particularly butyrate. In S. aureus, positively charged cytoplasmic proteins 86 , eDNA-binding proteins and membrane-attached lipoproteins released into the extracellular environment by cell lysis make electrostatic interactions with the negatively charged cell surface proteins and eDNA to produce an electrostatic net, enhancing attachment and forming a stable biofilm matrix 13,82 . We identified the low molecular weight (LMW) cell wall protein SlpA as the most abundant protein within the biofilm matrix unique bands, which has been shown to enhance adhesion of vegetative C. difficile cells to host epithelial cells 87 and potentially promote biofilm formation. Other proteins identified within the biofilm architecture were involved in amino acid fermentation, particularly Stickland fermentation including PrdA, which was also upregulated in C. difficile biofilms formed in response to DOC 29 . Other matrix proteins were involved in nitrogen and glutamate metabolism, including GluD, which facilitates resistance to H 2 O 2 and found in the stool of CDI patients 88 and Buk, involved in butyrate fermentation. Interestingly, we identified proteins involved in detoxification, most notably the rubrerythrin protein Rbr, involved in oxidative stress response, which complements data from Dubois et al. showing an upregulation of stress associated genes in C. difficile biofilms formed in the presence of DOC 29 . Other biofilm matrix proteins identified in the four unique bands include the chaperone DnaK, which is important in biofilm formation in E. coli 89 and C. difficile 38 . This provides additional evidence of the presence of both cell wall and intracellular proteins within the C. difficile biofilm matrix.
There are a number of sources of eDNA and intracellular proteins within biofilms, including autolysis (E. faecalis, S. aureus, P. aeruginosa, S. epidermidis) 64 , phage induced lysis (S. pneumoniae, Shewanella oneidensis) 90,91 , active secretion 66 and extracellular vesicles (OMVs), although the contribution of OMVs to biofilm formation is unclear 92 . Recent data suggests a role for the autolysin Cwp19 in biofilm formation in C. difficile in the presence of DOC 29 . Cwp19 has been shown to be involved in autolysis of C. difficile in stationary phase in BHIS media 93 . We observed Cwp19 and a phage protein (phiCD24-1) by LC-MS analysis of the matrix specific proteins, indicating that an autolysin and/or phage mediated cell lysis potentially contributes to the release of eDNA, in combination with lysis of sporulating cells driven by Spo0A. Other species such as Cryptococcus neoformans www.nature.com/scientificreports/ enhance biofilm formation by actively shedding their capsular polysaccharide glucuronoxylomannan 94 , therefore we could hypothesise that some of the cell surface proteins identified in the biofilm matrix maybe actively shed by C. difficile. We could also hypothesize that in a similar way to S. aureus, cytoplasmic proteins released during stationary phase may be recycled 86 to form the architecture of the biofilm matrix in combination with eDNA and cell surface associated proteins, and that eDNA stabilizes the biofilm structure and contributes to antimicrobial resistance. We show that eDNA is essential for the formation and structural integrity of C. difficile biofilms in all five C. difficile lineages, and that there is a positive correlation between biofilm biomass, eDNA and sporulation frequency. We hypothesize that the biofilm scaffold, comprised of eDNA, cell wall and intracellular proteins, surrounding vegetative cells and spores is most likely a result of cell lysis at stationary phase when C. difficile undergoes sporulation driven by the sporulation master regulator Spo0A, in combination with lysis of a subpopulation of cells via Cwp19 or phage mediated autolysis. Biofilms provide a potential reservoir of bacteria and spores to re-colonise an individual post-treatment, therefore we present a strategy using DNase to interfere with both biofilm initiation, maturation and recalcitrance for C. difficile, which reduces spore viability and promotes the efficacy of a current antibiotic therapy, such as vancomycin: which may prove valuable in the treatment of C. difficile infection or relapse.
Gene inactivation in C. difficile. Isogenic mutants were constructed using either the ClosTron or CodA system (Table 1; Supplementary Fig. S4 online) 49,51,58,95,[97][98][99] in C. difficile strain 630Δerm 95 or 630, respectively. These mutants were used alongside the previously published CD3145 Clostron mutant and complement 58 . For the Clostron mutant, the group II Ll.LtrB intron was retargeted to the gene of interest by SOE-PCR as previously described 100 , with oligonucleotides (listed in Table 1) designed using the Sigma TargetTron website (http://www. sigma -genos ys.com/targe tron/websi te). PCR products were cloned into HindIII and BsrGI sites of pMTL007C-E2 to create the plasmids pCD0183, pCD2831, and pCD3392 (Table 1). Retargeted pMTL007C-E2 plasmids were transformed into the E. coli conjugation donor strain CA434 97 and transferred into C. difficile strain 630Δerm by conjugation then screened for successful mutants as outlined previously 100,101 . DNA extracted from potential mutants (lincomycin resistant, thiamphenicol sensitive colonies), were screened by PCR and sequenced across the insertion/deletion site using gene specific primers or a gene specific primer with EBS universal primer, to verify insertion site. Mutants CD3392 and CD2831 were screened by Southern blot. Southern blot analyses were performed using AlkPhosDirect Labelling and detection kit (GE Healthcare) and detection reagents, in accordance with the manufacturer's guidelines and visualised using CDP star (GE Healthcare). Genomic DNA from wild type and potential mutants was digested with BbaB1 for the CD2831 screen (Southern band size for the mutant 3.8 Kb) or BtgI and Pcil for the CD3392 screen (Southern band size for the mutant 3.5 Kb). The probe was produced by PCR using RAM F and R primers (Table 1), from within the group II intron sequence. Mutations were complemented on a plasmid as outlined previously 100 (Table 1).
Biofilm formation. Attachment. This assay was performed in 24-well plates pre-equilibrated with 2 mL BHIS medium. The plates were incubated under anaerobic conditions for 6 h before inoculation at a 1/10 dilution with standardized primary cultures (overnight cultures in BHIS broth back diluted to OD 595nm 0.5) of C. difficile to optical density readings OD 595 = 0.05 ± 0.01. Three control wells were left blank, without addition of bacterial cultures. Thiamphenicol (15 µg/mL) and anhydrotetracycline (25, 50, 100 or 250 ng/mL) were added as required. Plates were incubated for 16 h. The plates were then inverted to remove any cells that were not tightly attached to the abiotic surface, meaning the majority of the early stage biofilm is removed, leaving only the attached cells. Each well was then washed twice with 1 mL PBS. The liquid was removed by inversion and tapping between each step. The plates were allowed to air-dry in the anaerobe cabinet for 10 min before the attached cells were stained with 800 µL of 0.1% crystal violet for 30 min. Plates were washed four times with 1 mL PBS, inverting plates between each step, and the crystal violet was solubilized with 1 mL methanol. Attachment was determined using OD 595 readings, with a subtraction of the mean of three blanks present on each individual plate. Assays were performed with a minimum of six independent replicates. The data was analysed in Excel and GraphPad prism 7.0. Statistical analysis was performed (see below).
Early biofilm formation. Biofilms were produced by inoculating overnight cultures of C. difficile into 24-well plates containing 2 mL pre-equilibrated BHIS (+ 0.1% l-Cysteine) per well at OD 595 = 0.05 ± 0.01, with the exception of three control wells, left un-inoculated. All strains were compared to C. difficile control strain 630, carrying a plasmid without dccA. These cultures were incubated statically for 24  www.nature.com/scientificreports/ removed with a pipette to not disturb the early biofilm attached to the plate. These biofilms were washed twice with 800 µL 1X PBS, before the addition of 800 µL of 0.1% crystal violet. The plates were washed four times with 1X PBS and the crystal violet was solubilized with 1 mL methanol. The OD 595 readings were obtained and the mean of three blank wells present on each individual plate was subtracted. As required, cultures were supplemented with thiamphenicol and/or anhydrotetracycline (ATc). Biofilm biomass was determined using OD 595 readings of the methanol solubilized crystal violet. A minimum of six independent replicates were performed. The data was analysed in Excel and GraphPad prism 7.0. Statistical analysis was performed (see below).
Late biofilm formation. Biofilms were produced as outlined above but were incubated statically for 72 h to enable maturation. As required, media were supplemented with thiamphenicol and/or ATc. Biofilm biomass was determined using OD 595 readings of the methanol solubilized crystal violet. The mean OD 595 reading of three blank wells present on each individual plate was subtracted from each sample well. A minimum of six independent replicates were performed. The data was analysed in Excel and GraphPad Prism 7.0. Statistical analysis was performed (see below).
Statistical analysis of biofilm biomass. Biofilm formation measured by crystal violet assays outlined above was assessed using a Students t test or Linear Regression Analysis in Stata IC 15 23 . For linear regression, the data was transformed using log 10 to approximate a normal distribution before a Linear Regression and partial F tests were performed to answer the following questions: a) Is there strong evidence that the concentration of eDNA or total DNA contained within a biofilm matrix varies between strains? Linear regression was used to determine if there is strong evidence that there are differences in eDNA concentration in biofilm biomass linked to strain, where p < 0.05 indicates strong evidence a biofilm contains less eDNA or total DNA within the biofilm matrix compared to the control strain (630). The coefficient of variance (COV) (used throughout) determines whether this difference is higher (positive number) or lower (negative number) than the reference (630) (COV = 0) ( Table 2). b) Is there strong evidence that cell surface proteins play a role in biofilm formation, measured by biofilm biomass. Linear regression was used to determine if there is strong evidence that there are differences in biofilm biomass taking individual strains into account. A p < 0.05 indicates strong evidence that deletion of cell surface genes or overexpression of cell surface genes, affect biofilm biomass, relative to their specific wild-type strains (630∆erm or 630). (Table 2). c) Is there strong evidence that the addition of c-di-GMP influences biofilm biomass in attachment, early biofilm or late biofilm formation, compared to the un-induced control strain (630 ptet-dccA NI). A partial F test was performed to determine if c-di-GMP had an effect on biofilm biomass (p < 0.05). Linear regression was used to determine if there is strong evidence that differences in biofilm biomass are linked to induction of dccA, where a p < 0.05 indicates strong evidence that c-di-GMP levels, driven by induction of dccA affect biofilm biomass compared to the control (Table 2). d) Is there strong evidence that the strain of C. difficile influences biofilm biomass? A partial F test and linear regression were performed to determine if there were strain specific differences in biofilm formation using representatives of the five C. difficile clades. p < 0.05 indicates strain specific differences (Table 2). e) Is there strong evidence that there are strain specific differences in vegetative cells and spores between different C. difficile lineages compared to strain 630. Is there strong evidence that percentage sporulation within a biofilm differs between ribotypes of the five main lineages. Where a p < 0.05 indicates strong evidence of strain specific differences in viable cell counts or sporulation frequency.

Structural components of the biofilm matrix.. SDS-PAGE and Western blot.
Primary overnight cultures of 630Δerm, CD3392::CT and the complement CD3392:: (pCD3392) were diluted to OD 595nm 0.5, then inoculated 1/10 to give a final OD 595 0.05 ± 0.01 into TC flasks (Falcon), and static biofilm cultures were grown for three days (late biofilms) under anaerobic conditions. Carefully without disturbing the biofilm matrix, 1 mL of the supernatant (planktonic fraction) was removed from the static biofilm culture and transferred to a 1.5 mL microcentrifuge tube, then samples were filter sterilised (0.2 µM filter) to remove viable cells and spores. The biofilm matrix was detached from the base of the TC flask by gentle agitation 23 and removed intact using a 1 mL pipette tip before being transferred to a fresh 1.5 mL microcentrifuge tube (as outlined in Fig. 6a). The intact biofilm matrix was disrupted by digestion with DNase for 15 min at room temperature (R/T) with 10 µg/mL DNase I. The planktonic and matrix fractions (obtained using DNase I digestion of the biofilm matrix) were visualised on an SDS-PAGE gel (data not shown). In parallel CFU counts were obtained from the planktonic and matrix fractions at early biofilm (24 h-planktonic fraction (2.48 × 10 8 vegetative cells, 1.91 × 10 4 spores) The iBlot 2 Dry Blotting System (Thermofisher) was used for transfer of samples from the SDS-PAGE gel to a nitrocellulose membrane (Thermofisher). Transfer stacks were assembled and loaded onto the iBlot according to manufacturer's instructions and transferred for 7 min at 20 V. Following transfer, membranes were washed for 5 min in 1 × PBS + 0.1% Tween20 then incubated with blocking buffer (1 × PBS + 0.1% Tween20 and 5% skimmed milk) for 1 h, at room temperature on a rocking platform (Stuart). Primary antibody (anti-CD3392 antibody) was added at 1:250 dilution into fresh blocking buffer and incubated for 1 h. Membranes were washed 3 × 5 min with 1 × PBS + 0.1% Tween20 before incubation with the fluorescently label goat anti mouse secondary antibody 1/5000 at the appropriate dilution, in the dark. The membrane was washed 3 × in 1 × PBS + 0.1% Tween20 before visualisation on a LI-COR. Images were scanned at 680 nm and labelled in Photoshop elements to retain image resolution.
Protein sample preparation and analysis. Primary overnight cultures of C. difficile strains were diluted to OD 595nm 0.5, then inoculated 1/10 to give a final OD 595 0.05 ± 0.01 into TC flasks (Falcon), and static biofilm cultures were grown for three days under anaerobic conditions. 1 mL of the supernatant (planktonic fraction) was transferred to a 1.5 mL microcentrifuge tube for processing, then the biofilm matrix was detached from the base of the TC flask by gentle agitation 23 and removed intact using a 1 mL pipette tip before being transferred to a fresh 1.5 mL microcentrifuge tube for processing (Fig. 6a). The biofilm matrix was disrupted by digestion with DNase for 15 min at room temperature (R/T) with 10 µg/mL DNase I. The planktonic and matrix samples were filter sterilised (0.2 µM filter) to remove viable cells and spores. A duplicate late biofilm (72 h) was processed for protein quantification, herein, the supernatant and matrix were separated as outlined above, however, for protein quantification the matrix samples were vortexed for 2 min to disrupt the biofilm (without the addition of DNase I). The total protein concentration in the planktonic fraction (supernatant) (765 µg/mL) and matrix (793 µg/ mL) fractions were determined using a BCA assay (Pierce BCA assay kit) in accordance with the manufacturer's instructions. The planktonic and matrix fractions (obtained using DNase I digestion of the biofilm matrix) were visualised on an SDS-PAGE gel, with a DNase I only control lane (Fig. 6). The four bands unique to the matrix fraction were excised and sent for LC-MS/MS analysis (King's College London, CEMS Proteomics facility) (Fig. 6). In-gel reduction, alkylation and digestion with trypsin were performed on the excised gel bands prior to subsequent analysis by mass spectrometry. Cysteine residues were reduced with dithiothreitol and derivatized by treatment with iodoacetamide to form stable carbamidomethyl derivatives. Trypsin digestion was carried out overnight at R/T after initial incubation at 37 °C for 2 h. Peptides were extracted from the gel pieces by a series of acetonitrile and aqueous washes. The extract was pooled with the initial supernatant and lyophilised. The sample was then re-suspended in 10 μL of resuspension buffer (2% ACN in 0.05% FA) and analysed by LC-MS/ MS. Chromatographic separation was performed using a U3000 UHPLC NanoLC system (ThermoFisherScientific, UK). Peptides were resolved by reversed phase chromatography on a 75 μM C18 column (15 cm length) using a three-step linear gradient of 80% acetonitrile in 0.1% formic acid. The gradient was delivered to elute the peptides at a flow rate of 250 nL/min over 60 min. The eluate was ionised by electrospray ionisation using an Orbitrap Fusion Lumos (ThermoFisherScientific, UK) operating under Xcalibur v4.1.5. The instrument was programmed to acquire in automated data-dependent switching mode, selecting precursor ions based on their intensity for sequencing by collision-induced fragmentation using a TopN CID method. The MS/MS analyses were conducted using collision energy profiles that were chosen based on the mass-to-charge ratio (m/z) and the charge state of the peptide. Raw mass spectrometry data were processed into peak list files using Proteome Discoverer (ThermoScientific; v2.2). The raw data file was processed and searched using the Mascot search algorithm (v2.6.0; www.matri xscie nce.com) and the Sequest search algorithm (Eng et al.; PMID 24226387) against the current All Taxonomy and C. difficile databases from Uniprot, at a stringency threshold of 5% false discovery rate (FDR) for protein and peptide and a minimum of one peptide per protein. The Database generated file was uploaded into Scaffold 4 (v4.8.7) software (www.prote omeso ftwar e.com) for analysis. This was also repeated for Bovine DNase I. The stringency threshold parameters were therefore set to 95% protein, minimum 1 peptide and 95% peptide. The data from each band was then aligned with a protein classification (Riley classification) and processed in Excel (Supplemenatry Table S1). Pie charts were produced to show the percentage representation of proteins in the matrix. www.nature.com/scientificreports/ DNA quantification. Primary overnight cultures of C. difficile strains 630, R20291, M120, M68 and CD305 were back diluted in pre-reduced BHIS + Cysteine to OD 595 0.5 and used to inoculate 1/10 into two matched vented tissue culture flasks (Falcon) containing 10 mL fresh BHIS + cysteine, or duplicate 24-well plates (Corning) with each well containing 2 mL fresh BHIS + cysteine at an inoculum density of OD 595 = 0.05 ± 0.01. These cultures were incubated statically for three days. Biofilms were detached from the surface of tissue culture flask by gentle agitation (Fig. 6a), after which the intact biofilm was removed with a pipette within a volume of 1 mL and transferred to a 2 mL screw cap tube. For the 24-well plates, 1.5 mL of excess liquid was removed, then the biofilm was detached from the bottom of the well using a pipette and was re-suspended in 500 μL BHIS broth. Biofilms were vortexed to disrupt the biofilm matrix. Matched biofilms were processed in parallel to (1) extract matrix eDNA by removal of vegetative cells and spores by passing the disrupted biofilm through a 0.22 µM filter or (2) extraction of total DNA (matrix eDNA and intracellular DNA), disrupted biofilm matrix samples were transferred to a Lysis Tube B (MP Biomedical) before ribolysing at 6.0 m/s for 45 s using a FastPrep-24 classic (MP Biomedicals). Both Matrix eDNA and total DNA samples were then processed and extracted using the DNeasy Blood and Tissue Kit (Qiagen), following the manufacturers guidelines for the Purification of Total DNA from Animal Blood or Cells. In short, 500 µL of the sample from tissue culture flasks or 200 µL from 24-well plates were processed with the AL buffer in accordance with the manufacturer's guidelines. Samples were eluted in 100 µL of elution buffer. DNA concentrations were then measured by spectrophotometry using a DS-11 FX+ (DeNovix) according to manufacturer's instructions, and concentrations were normalized to per biofilm (mg/biofilm).  Iodide). Acridine orange was scanned at both the green (488 nm/> 560 nm) and red (543 nm/> 650 nm), as Acridine orange stains double stranded DNA green and ssDNA/RNA red. Z-stacks were performed to measure the biofilm depth. The assays were performed with a minimum of quadruplicate independent replicates, imaging a minimum of three fields of view per slide and images were analysed using Zeiss LSM image browser and Volocity imaging software.
Scanning electron microscopy. Triplicate independent 2 mL liquid cultures of 630 (in BHIS + cysteine) were inoculated into pre-equilibrated low evaporation lid 24-well plates (Nunc) containing Thermanox coverslips (Fisher), including two blank media controls. The plates were sealed with Nescofilm and incubated statically at 37 °C as outlined above. Coverslips were processed at 16 h, 24 h and 72 h for attachment, early biofilm formation and late biofilm formation, respectively. Coverslips were washed in 1X PBS before fixation. Alternatively, late biofilms were detached from the bottom of a TC flask, the biofilm was carefully extracted using a pipette (Fig. 6a), then added directly into 1 mL fixative before being transferred to a coverslip and stub for SEM imaging. The fixative was applied to coverslips or late biofilms, 1 mL fixative (2.5% Paraformaldehyde/2.5% Glutaraldehyde/0.1 M Sodium cacodylate pH 7.4) was added to each sample, including the media blank controls. The samples were washed in 0.1 M Na cacodylate, post-fixed in 1% aqueous/0.1 M Na cacodylate, and stored in MilliQ water overnight at 4 °C. These were then air dried and mounted onto an aluminium stub prior to sputter coating with gold. Samples were imaged using a JSM35 Scanning Electron Microscope. A minimum of three fields of view were imaged per slide at 1× K, 3.5× K, 5× K, 7× K and 10× K magnification for the biofilms grown on coverslips in 24-well plates. For the biofilms grown in TC flasks the samples were images at 150×, 1× K and 3× K and 10× K on a Scanning Electron Microscope.

Inhibition and degradation of biofilms. Degradation of biofilms.
Liquid cultures of C. difficile strains 630∆erm, 630, R20291, CD305, M120 and M68 were incubated in TC flasks under anaerobic conditions for 72 h as outlined above, then were removed from the anaerobe hood. The biofilms were detached from the bottom of the flask by gentle agitation. These intact biofilms were treated with 1 mg/mL and 100 µg/mL Recombinant DNase (Sigma) (10 mg stock dilutions were made in 0.15 M NaCl 2 ), 1 mg/mL and 100 µg/mL RNase A (Sigma) (dilutions were made in nuclease free dH 2 0) or 1 mg/mL Proteinase K (Qiagen) (stock 20 mg/mL). Samples were incubated at R/T for 15 min. Before and after addition of the DNase, RNase or Proteinase K, images were taken using a Canon 600D SLR (50 mm prime lens), mounted on a Kaiser RS2 copy stand with lighting unit. Late biofilms produced on Thermanox coverslips as outlined above were incubated for 15 min with DNase (1, 10, 100 µg/mL and 1 mg/mL). The slides were processed for confocal microscopy using FilmTracer LIVE/DEAD biofilm viability kit as outlined above. www.nature.com/scientificreports/ Inhibition of biofilm formation. Cultures of C. difficile strain 630∆erm were inoculated at OD 595 0.05 ± 0.01 as outlined above into TC flasks (Falcon) containing BHIS media with recombinant DNase (500 µg/mL or 1 mg/ mL final concentration) as outlined above. Images were taken of the 72-h cultures flasks as outlined above.
The effect of vancomycin on CFU counts in disrupted and intact biofilms. The effect of vancomycin treatment on both vegetative cells and spores within an intact and degraded biofilm was undertaken in 3-day old biofilms. The biofilms were detached from the bottom of the flask by gentle agitation. These intact biofilms were treated with either 1 mg/mL recombinant DNase (Sigma), 1 mg/mL Proteinase K (PK) (Qiagen), 12.5 mg/mL Vancomycin, or a combination of these, compared to an untreated intact biofilm. Samples were incubated for 24 h at 37 °C in an anaerobic atmosphere. Samples were then pelleted by centrifugation at 4500×g for 10 min. Pellets were washed twice in 10 mL sterile PBS, then pelleted and re-suspended in 1 mL of 1× PBS and vortexed to break down any remaining intact biofilm. Total counts were obtained from 1 mL of each disrupted biofilm culture, serially diluted in 1× PBS and plated in triplicate onto BHIS plates supplemented with 0.1% taurocholate. Spores were isolated from total cells by heat inactivating 1 mL of each sample (65 °C for 25 min), killing the vegetative cells, and were plated in triplicate onto BHIS agar plus 0.1% sodium taurocholate. The data was analysed in Excel and GraphPad Prism 7.0. Regression analysis was performed to identify significant differences in total cell counts and spore counts caused by exposure to vancomycin in disrupted versus intact biofilms (see below).
Statistical analysis of the effect of vancomycin on the CFU counts. The CFU data were transformed using log 10 to approximate a normal distribution, after which a linear regression analysis was performed to determine whether there was strong evidence that disruption of biofilms increased the vancomycin susceptibility of the vegetative cells and spores encased within compared to the intact biofilm. Analysis was undertaken to answer if there: were differences in viability of the vegetative cells (1) or spores (2) in the presence of vancomycin in intact and DNase or PK disrupted biofilms compared to the untreated biofilm control? These analyses were repeated to see if DNase and PK treatment of the biofilms made the (3) vegetative cells or (4) spores more susceptible to vancomycin than the non-disrupted vancomycin treated control. For all the analyses p < 0.05 indicates a significant difference and the coefficient (C) determines whether this difference is higher (positive number) or lower (negative number) than the reference (indicated in Table 2) (C = 0). www.nature.com/scientificreports/