SaeRS-Dependent Inhibition of Biofilm Formation in Staphylococcus aureus Newman

The SaeRS two-component regulatory system of Staphylococcus aureus is known to affect the expression of many genes. The SaeS protein is the histidine kinase responsible for phosphorylation of the response regulator SaeR. In S. aureus Newman, the sae system is constitutively expressed due to a point mutation in saeS, relative to other S. aureus strains, which results in substitution of proline for leucine at amino acid 18. Strain Newman is unable to form a robust biofilm and we report here that the biofilm-deficient phenotype is due to the saeSP allele. Replacement of the Newman saeSP with saeSL, or deletion of saeRS, resulted in a biofilm-proficient phenotype. Newman culture supernatants were observed to inhibit biofilm formation by other S. aureus strains, but did not affect biofilm formation by S. epidermidis. Culture supernatants of Newman saeSL or Newman ΔsaeRS had no significant effect on biofilm formation. The inhibitory factor was inactivated by incubation with proteinase K, but survived heating, indicating that the inhibitory protein is heat-stable. The inhibitory protein was found to affect the attachment step in biofilm formation, but had no effect on preformed biofilms. Replacement of saeSL with saeSP in the biofilm-proficient S. aureus USA300 FPR3757 resulted in the loss of biofilm formation. Culture supernatants of USA300 FPR3757 saeSP, did not inhibit biofilm formation by other staphylococci, suggesting that the inhibitory factor is produced but not secreted in the mutant strain. A number of biochemical methods were utilized to isolate the inhibitory protein. Although a number of candidate proteins were identified, none were found to be the actual inhibitor. In an effort to reduce the number of potential inhibitory genes, RNA-Seq analyses were done with wild-type strain Newman and the saeSL and ΔsaeRS mutants. RNA-Seq results indicated that sae regulates many genes that may affect biofilm formation by Newman.

Introduction unphosphorylated SaeR in cells, with the ratio being higher in strains with the saeS P allele [20,24]. Mainiero et al. [24] suggested that SaeS may possess a phosphatase activity that is defective in the SaeS P protein. In support of this, Jeong et al. [20] found that SaeS L does have phosphatase activity, which is activated by the interaction of SaeP and SaeQ with SaeS. The constitutive activity SaeS P in Newman may be due to loss of catalytic activity or the inability of SaeS P to interact with SaePQ. Either of these possibilities is consistent with the finding that SaeS L is dominant over SaeS P , when saeS L is cloned into wild-type Newman [24]. Because SaeRS is activated by hydrogen peroxide and alpha-defensins and because many toxins are SaeRS-regulated, it has been hypothesized that this system could promote escape from PMNs after phagocytosis [22]. Indeed, a saeRS mutant strain demonstrated an impaired ability to survive in human neutrophils after phagocytosis [17]. We report here that strain Newman, with the saeS P allele, secretes a protein capable of inhibiting biofilm formation by other S. aureus strains. Moreover, deletion of saeRS P , or replacement of SaeS P with saeS L , blocked production of the inhibitory protein by strain Newman. Additionally we performed RNA-Seq analysis to compare gene expression in strain Newman, Newman with a repaired saeS gene (saeS L ) and Newman deleted for saeRS.

A SaeRS-regulated protein prevents biofilm formation
The capacity of S. aureus Newman to form a biofilm has been addressed by several laboratories. Some have reported that Newman can form a biofilm whereas others, including our laboratory, have concluded that strain Newman is a poor biofilm producer [12,15,26,27]. As stated above, strain Newman constitutively expresses the SaeRS two-component system. To determine if constitutive activation of SaeRS in Newman contributes to the biofilm-deficient phenotype, strains Newman, Newman saeS L , Newman deleted for sae (ΔsaePQRS::kan or ΔsaeRS) and the ΔsaeRS complemented with either saePQRS P or saePQRS L , were assessed in a static microtiter plate biofilm assay. Strikingly, the saeS L , ΔsaePQRS::kan and ΔsaeRS strains formed robust biofilms compared to Newman wild-type ( Fig 1A). Introduction of the saePQRS P , but not saePQRS L , bearing plasmid into the ΔsaeRS mutant resulted in restoration of the Newman wild-type, biofilm-deficient phenotype ( Fig 1A).
We next tested the possibility that strain Newman may secrete a factor that inhibits biofilm formation. Supernatants from cultures of strain Newman and its derivatives were harvested and added to the biofilm medium at 25% (v/v). S. aureus strain UAMS-1 was subsequently inoculated into the medium, and biofilm assays were performed ( Fig 1B). Supernatants from the ΔsaeRS or ΔsaePQRS::kan strains did not inhibit biofilm formation by UAMS-1, while the supernatant from strain Newman did. In fact all of the biofilm forming derivatives of Newman were unable to inhibit biofilm formation by UAMS-1 whereas the biofilm deficient derivatives were able to inhibit (Fig 1B). These results demonstrated that Newman derivatives with the saeS P allele secrete a factor that inhibits biofilm formation.
To determine if the inhibitor is a protein, supernatants were treated with proteinase K (New England Biolabs) and the proteinase was subsequently heat inactivated. As shown in Fig 2A, proteinase K-treated supernatant from strain Newman did not inhibit biofilm formation by strain UAMS-1, suggesting the inhibitory factor is a protein. To further characterize the protein, supernatants were subjected to heating at 75°C for 60 min or 95°C for 15 minutes. Heat treated supernatants were then added to the biofilm media at 25% and strain UAMS-1 was used for the biofilm assay. None of the treatments inactivated the protein (Fig 2B), suggesting that the protein is heat stable.

Regulation of attachment and dispersal by SaeRS
To test whether the inhibitory protein affects the attachment step of biofilm formation, attachment assays were performed by diluting mid-log phase bacteria into biofilm media and inoculating plasma-coated 12-well plates. Bacteria were incubated for 1 h at 37°C, then were washed, fixed, and stained with crystal violet. Newman wild-type attached as distinct clumps, while the ΔsaePQRS::kan strain attached as a uniform layer of cells ( Fig 3A). To confirm this effect, a strain UAMS-1 culture was supplemented with supernatants from either strain Newman or Newman ΔsaePQRS::kan and inoculated as stated above. Strain UAMS-1 cultures supplemented with supernatant from strain Newman exhibited the same clumping phenotype as strains were diluted to an OD 660 of 0.05 and inoculated into wells of a microtiter plate. The microtiter plate wells had been precoated with human plasma. After 16 h, biofilms were washed, fixed, and stained with crystal violet. The wells labeled "medium" indicates wells that contained sterile biofilm medium. (B) Inhibition of UAMS-1 biofilm formation by culture supernatants. Stationary phase culture supernatants of strain Newman and its derivatives were harvested, filter sterilized and added to microtiter plate wells preinoculated with S. aureus UAMS-1 suspended in biofilm medium. The source of each culture supernatant is listed to the right of the picture. The wells labeled "medium" indicates sterile TSB-0G medium was added in place of an exogenous culture supernatant. Plates were incubated and processed as described for A.  The biofilm inhibitory factor is a heat stable protein. Stationary phase culture supernatants of strain Newman (wt) and Newman ΔsaeRS were harvested, filter sterilized, and diluted into biofilm medium. Antibiofilm activity of supernatants was tested against S. aureus strain UAMS-1. (A) Proteinase K treatment of culture supernatants. Supernatants from the indicated strains were incubated with proteinase K and Tris buffer, or buffer only (mock prtK), at 42°C for 1 hr and then 95°C for 10 minutes, before adding to biofilm medium. "None" indicates wells containing UAMS-1 but no culture supernatant. Wells labeled "medium" contained sterile biofilm medium. (B) Supernatants were incubated at the indicated temperatures for the indicated time periods before adding to wells containing UAMS-1 suspended in biofilm medium.
strain Newman, whereas UAMS-1 incubated with the ΔsaePQRS::kan mutant supernatant attached as a uniform layer of cells ( Fig 3B). These results suggested the SaeRS-regulated protein affects the attachment step of biofilm formation. To test a possible role of SaeRS in biofilm dispersal, strain UAMS-1 was inoculated into a plasma-coated microtiter plate at time 0. Every two hours for 10 hours post inoculation of strain UAMS-1, the media was changed to one of the following: 1) fresh biofilm media, 2) fresh biofilm media supplemented with Newman supernatant, or 3) fresh biofilm media supplemented with the ΔsaePQRS::kan mutant supernatant. At 24 h post inoculation, biofilms were washed, fixed, and stained with crystal violet. Media supplemented with strain Newman supernatant did not affect mature biofilms (Fig 3C), suggesting that the putative inhibitory protein does not have the capacity to disrupt a mature biofilm. Collectively, these data suggest that a SaeRS P -regulated protein inhibits the attachment step in biofilm formation.

Strain-dependent expression of the biofilm inhibitory protein
As stated previously, strain Newman has a point mutation in saeS that results in constitutive activation of the SaeRS two-component system. Our data suggested that a consequence of this constitutive activation was the inability to form a biofilm. To determine if the saeS P allele would affect other S. aureus strains, a SaeRS constitutively active strain was constructed in the USA300 FPR 3757 (USA300) background. First, the ΔsaePQRS::kan mutation was introduced into wildtype strain USA300 FPR 3757 via phage transduction. Next, either the saePQRS operon from strain Newman (saePQRS P ) or from strain 8325-4 (saePQRS L ) was introduced into USA300 ΔsaePQRS::kan, using pCWSAE51 or pCWSAE50, respectively. The vector carrying the sae operons is a single copy integration vector that integrates within the hlb gene at the phage Ø13 attachment site [24]. The benefit of using an integration vector is stable maintenance of the construct in the chromosome allowing biological assays to be performed without the use of antibiotics to maintain the plasmid. Resulting strains were tested for biofilm production and culture supernatants from planktonic cultures were subsequently tested for anti-biofilm activity against strain UAMS-1. As shown in Fig 4A, wild-type strain USA300 and the ΔsaePQRS::kan mutant formed a biofilm, and complementation of ΔsaePQRS::kan with saePQRS from strain 8325-4 (denoted saePQRS L ) resulted in no change in the biofilm phenotype. Interestingly, complementation of ΔsaePQRS::kan with saePQRS from strain Newman (denoted saePQRS P ) resulted in an apparent reduction in biofilm formation, further confirming our hypothesis that constitutive activation of SaeRS limits biofilm formation. To assess production of the secreted biofilm inhibitory protein, USA300 and derivatives were grown to stationary phase and culture supernatants were harvested, filter sterilized and added to biofilm assays of strain UAMS-1. Culture supernatants from strain USA300 expressing saePQRS P did not contain any apparent anti-biofilm activity (Fig 4B), which suggests the inhibitory factor is not secreted by USA300. It is possible that the inhibitory protein is associated with the cell surface in USA300, but is released into the environment by Newman. Newman does encode some truncated versions of cell-wall anchored proteins that, due to the truncation, would no longer be anchored [25,28]. Identification of the inhibitory protein should aid in understanding this result. It is worth noting that another USA300 isolate (USA300 LAC) does secrete a DNA nuclease that can inhibit biofilm of other staphylococci [29]. However, as discussed below, the nuclease is distinct from the inhibitory factor produced by Newman.

Inhibition of biofilm formation by other S. aureus and S. epidermidis isolates
The results above indicate that a secreted protein produced by strain Newman inhibited biofilm formation by strain UAMS-1 when added to biofilm medium at the time of inoculation. To test anti-biofilm activity of the putative secreted protein against other staphylococci, including S. aureus strains USA100, USA300, three isolates of USA500, and S. epidermidis strains RP12 and RP62A were inoculated into biofilm media containing supernatant from strain Newman. The putative protein effectively inhibited biofilm formation by all S. aureus strains tested but did not inhibit biofilm formation by either S. epidermidis strain. Supernatant from Newman ΔsaePQRS::kan did not affect biofilm formation by any strain (Fig 5). The lack of an inhibitory effect on S. epidermidis could be due to species-dependent differences in biofilm formation. S. epidermidis relies more heavily on exopolysaccharides than exoproteins for biofilm formation. Some S. aureus strains on the other hand, rely primarily on exoproteins for biofilm formation, at least in the presence of plasma proteins [4,6]. Alternatively, the actual target of the inhibitor may not be expressed by S. epidermidis.

Screens to identify the biofilm inhibitor
A variety of biochemical methods were used in attempts to identify the biofilm inhibitory factor produced by strain Newman. These included fractionation of culture supernatants using Ultra centrifugal filters, NH 4 SO 4 precipitation, size exclusion and ion exchange chromatography and gel electrophoresis or a combination of these methods. Using these methods a number of candidate proteins were identified and these were subjected to trypsin digestion and tandem mass spectrometry sequencing and identification. S1 Table lists the candidate genes identified as possibly encoding the inhibitor protein. Each of the candidate proteins listed in S1 Table was tested for anti-biofilm activity using several genetic approaches, including insertional inactivation, gene deletion, and complementation. However, despite our efforts, the inhibitory protein was not identified.
In summary, our studies show that strain Newman expresses a heat stable extracellular protein, approximately 50 kDa in size, that can inhibit biofilm formation by the producing organism, as well as other strains of S. aureus. The protein is regulated by saePQRS and is dependent on the presence of saeS P allele for an effective level of expression. Preparations of the inhibitory factor seem to have high activity, suggesting the inhibitor is an enzyme or a signaling factor. Although we tested a number of candidate genes, the gene encoding the inhibitor was not identified. It is possible that the inhibitory factor is actually more than one protein and inactivation of a single gene is insufficient for complete loss of activity.

RNA-Seq analysis of the SaeRS transcriptome
Based on the data summarized above, we anticipated that the gene encoding the biofilm inhibitory factor would be expressed at a higher level in strain Newman than in either the saeS L or Δsae strains. In an effort to identify the Newman encoded biofilm inhibitory protein, we carried out RNA-seq analysis using RNA from CYL5876 (Newman SaeRS P ), CYL11481 (Newman saeS L ), and CYL11771 (Newman ΔsaeRS). Although several previous studies have identified SaeRS-regulated genes [17,24,30,31], no studies to date have compared three strains with varying levels of SaeRS activity. Furthermore, previous transcriptional profiling studies employed microarray-based approaches. Microarrays are based on annotated genes and will not detect every transcriptionally active region in the genome of all S. aureus strains. RNA-Seq does not rely on gene probes and directly sequences and identifies all transcriptionally active regions in a genome rather than detecting only annotated genes [32,33]. Our data confirm previous reports that SaeRS regulates many genes, and also confirmed previous findings regarding straindependent differences in SaeRS activity. As expected, our analysis is more comprehensive and identified far more genes in the SaeRS region than earlier reports [17,24,30,31].
The RNA-Seq results indicated that expression of 225 genes was at least 2-fold lower in the ΔsaeRS strain than in strain Newman (S2 Table) and 80 genes were expressed at a lower level in the ΔsaeRS strain than by saeS L strain (S3 Table). Expression of 80 genes was lower in the saeS L strain relative to the wild-type (S4 Table). These genes represent those that are under positive regulation by sae. A smaller set of genes, 64, are up regulated in the ΔsaeRS strain relative to wild type Newman (S5 Table). Many of the latter group are up regulated in the SaeRS L strain as well, thus these genes are negatively regulated by sae (S6 and S7 Tables). It seems probable that the gene(s) encoding the biofilm inhibitory protein would be listed in both S2 and S4 Tables.
It has been reported that SaeR must be phosphorylated to bind DNA and that a mutation of the phosphorylation site in SaeR results in a complete loss of sae activity [21,24]. Mainiero et al. (2010) [24] argued that transcription of some saeRS-regulated genes requires a high level of SaeR phosphorylation whereas transcription of other genes can occur with a lower level of SaeR phosphorylation. They also proposed that low level phosphorylation of SaeR can occur in the absence of SaeS. The constitutive activity of SaeS P results in a high level of SaeS kinase activity, or possibly a low level of phosphatase activity, resulting in a high ratio of phosphorylated to unphosphorylated SaeR [18,20,24]. Our RNA-Seq results identified a number of genes where there is an apparent dose-dependent response to SaeRS levels (S8 Table), meaning that expression of a particular gene is highest in strain Newman (with constitutively active SaeRS) and lowest in the ΔsaeRS strain. Expression of these genes is intermediate in the saeS L strain. It seems likely that these results reflect different levels of phosphorylated SaeR in the three strains. There are a number of genes that appear to be expressed nearly equally well in the wild type and the saeS L strains, but poorly expressed in the deletion strain. The most striking example of this is the hla gene (NWMN_1073) which is expressed at a much higher level in the saeS P and saeS L strains than in the deletion strain. (S2 and S3 Tables) The saeS L strain apparently has a level of phosphorylated SaeR sufficient for full expression of these genes.

Confirmation of RNA-Seq by real-time RT-PCR
To confirm the RNA-Seq results, we performed real-time RT-PCR of some select genes. Cultures of strains CYL5876, CYL11481, and CYL11771 were grown as for the RNA-Seq analysis. RNA was isolated from these cultures as described for RNA-Seq analysis except that rRNA depletion was not performed. As we are interested in genes that might influence biofilm formation, we decided to focus on genes predicted to affect the level of extracellular DNA (eDNA). We also verified expression of saeR in the three test strains (saeR in the saeS L strain was 4.2±1.2% of the wild type and undetectable in the saeRS strain). We measured expression of 6 saeRS-regulated genes, lytS, lrgA, atlR, atlA, arlR and aaa, that could potentially affect autolysis and eDNA release by strain Newman. The results shown in Fig 6 are in general agreement with the RNA-Seq results, although differences in gene expression between the strains were generally greater in the RNA-Seq analysis. As anticipated, expression levels of lytS, lrgA, arlR and atlR were highest in Newman and lowest in the deletion strain with expression in the saeS L strain being intermediate. The reverse pattern was observed for atlA and aaa; where the expression levels were greatest in ΔsaeRSsaeS L >Newman. Most of these genes are probably indirectly regulated by saeRS. For example, lytS and atlR regulate expression of lrgAB and atlA, respectively, but neither lytS nor atlR has a predicted SaeR binding site in their promoter region [34]. The aaa gene does not appear to have a SaeR binding site either and may not be directly regulated by sae. SaeR has been shown to bind to the arlR promoter, suggesting direct regulation of this gene by saeRS. The arlRS 2-component system is involved in the negative regulation of the LytN autolysin [35] but an effect of sae on lytN was not apparent in our analysis.

Regulation of autolysis and eDNA accumulation by saeRS
Autolysis could result in the release of bacterial DNA into the environment and extracellular or eDNA has been proposed to play a positive role in biofilm formation, therefore, we were interested in how sae would affect cell lysis and release of bacterial DNA. LytSR is a 2-component regulatory system that activates expression of lrgAB [34]. LrgA is an antiholin and an antagonist of CidA, a holin-like protein that has been proposed to disrupt the cell membrane leading to cell lysis [36][37][38]. Because saeS P upregulates lytSR and lrgAB, saeS P would be predicted to decrease the release of cellular DNA into the environment, relative to the saeS L and ΔsaeRS strains [37]. Thus the overexpression of lytS in wild-type Newman would have a negative impact on biofilm formation. The impact of sae on biofilm would be lessened in strains that express the saeS L allele. The negative effect of saeS P on expression of atlA, which encodes the major S. aureus autolysin, would also be predicted to decrease autolysis and DNA release [39,40]. In this instance, however, the inhibitory effect would be manifest through sae upregulation of atlR which encodes a repressor of altA [40]. It is important to note that atlA has been shown to increase biofilm in S. aureus by promoting attachment and early biofilm formation [40,41]. Increased SaeRS does decrease expression of aaa, which encodes a N-acetylmuramoyl-L-alanine amidase precursor [42]. The amidase may well contribute to the cell lysis. Phages ØNM1 and ØNM4 encode a holin (NWMN_1770) and an amidase (NWMN_0313), respectively. However, these genes are increased slightly in the saeS L but not in the ΔsaeRS strain.
The transcriptional analysis of murein hydrolase genes summarized above, suggested that the ΔsaeRS and saeS L strains may produce higher levels of these enzymes than does wild type Newman. To test this possibility, zymographic analyses of the three strains were performed (Fig 7). Extracts of these strains were prepared and fractionated on gels containing either heatkilled S. aureus strain RN4220 cells or cells of Micrococcus luteus. As shown in Fig 7, increased murein hydrolase activity, relative to Newman, was associated with both the saeS L and ΔsaeRS strains. The differences were most prominent for protein bands of approximately 60 and 90 KDa, and somewhat less prominent for a protein migrating at approximately 30 kDa. Additionally, the ΔsaeRS strain was deficient in expression of 2 small hydrolases that migrated at the 17 to 20 kDa range.
The atlA encoded protein is a bifunctional N-acetyl-glucosaminindase (GL) and N-acetylmuramyl-L-alanine amidase (AM), expressed as a 137 kDa precursor that is processed to 53.6 kDa and 63 kDa AM proteins [41]. It seems likely that the approximate 60 kDa protein in the RN4220 zymogram (Fig 7, leftmost panel) represents the AtlA protein. The aaa gene encodes a 35.8 kDa precursor AM [42]. It is possible that the approximate 30 kDa hydrolase, which is expressed at a somewhat higher level in the ΔsaeRS and saeS L strains, represents the aaa gene product. The LytM and LytH autolysins are of approximately the same size, but neither appeared to be regulated by sae.
Increased expression of at least 2 autolysins is apparent in the M. luteus zymogram (Fig 7,  middle panel). Both the saeRS and saeS L strains show increased expression of 50 and 68 kDa enzymes. These could represent the AtlA GL and AM enzymes, respectively. While we have not verified the identities of the various autolysins, these results do verify that constitutive expression of saeRS in Newman does decrease autolytic activity associated with S. aureus cells.
In order to determine if SaeRS does impact eDNA accumulation, we quantified the amount of DNA present in culture supernatants (Fig 8). A mutant (Δnuc) strain that does not express the secreted nuclease was included in our analysis as a positive control for eDNA accumulation. The amount of eDNA was over 10-fold higher in cultures of the ΔsaeRS strain than the saeS L strain and eDNA was essentially undetectable for wild type Newman. These results are consistent with the proposal that saeS P is a negative regulator of autolysis and eDNA production. The fact that eDNA was higher in the ΔsaeRS strain than in the Δnuc strain suggests that factors other than nuclease affect eDNA accumulation. The positive activation of nuc and lrgAB and the negative repression of autolysis genes by SaeRS P , would suggest that eDNA degradation may play a role in the biofilm deficient phenotype of Newman. Additionally, Nuc is a thermostable extracellular protein which is highly expressed in Newman. Nuc was previously shown to inhibit biofilm formation by S. aureus USA300 LAC and culture supernatants of USA300 LAC were found to inhibit biofilm formation by other strains [29]. Therefore, it seemed likely that Nuc could be the biofilm inhibitor produced by Newman. Indeed, Nuc was identified during our earlier screening by biochemical methods (S1 Table). However, Newman nuc::ltrB was still biofilm negative (S1 Fig), indicating that it is not the inhibitory factor that we were looking for. To test whether lrgAB plays a role in biofilm formation in Newman, a lrgAB::erm single mutant of Newman was generated and tested. The mutation did not affect the biofilm negative phenotype of Newman (S1 Fig). These results suggest that neither nuc nor lrgAB are involved in biofilm formation in Newman. Although Newman produces a significant amount of nuclease (Fig 7) any inhibitory effect appears to be masked by the yet to be identified inhibitory protein produced by Newman.

Regulation of bacteriophage genes by saeRS
Strain Newman is known to carry 4 prophages, ØNM1, ØNM2, ØNM3 and ØNM4 that contribute to pathogenesis [25,43,44]. We found that each of these prophages has genes that are subject to sae regulation (S9 Table). ØNM3 is a defective prophage that encodes multiple virulence factors including sea (enterotoxin A), sak (staphylokinase), chp (chemotaxis-inhibitor protein), and scn (staphylococcal complement inhibitor). Of these, sak, chp and scn have predicted SaeR binding sites in their promoter regions [21] and the RNA-Seq analysis indicated all are regulated by sae. RNA-Seq also revealed that at least 15 other ØNM3 genes are regulated by sae including genes encoding head proteins and DNA replication and transcription factors. Because ØNM3 is defective, it seems unlikely that the prophage would excise from the chromosome and undergo replication. It is possible that ØNM3 may function as a helper phage for other bacteriophages or for packaging and transfer of other S. aureus mobile genetic elements. The ØNM3 genome does encode an HNH endonuclease of the type reported to be involved in phage-mediated horizontal gene transfer in S. aureus and other bacteria [45]. Our results indicate that saeRS could possibly play a role in horizontal gene transfer in S. aureus via the activation of prophage genes.
Interestingly, numerous ØNM4 genes are expressed at a higher level in strain Newman than in the saeS L strain. These genes seem to represent most of the NM4 genome including genes involved in prohead and tail assembly, DNA packaging and host cell lysis [43]. It may be that the ØNM4 prophage is induced to undergo lytic growth at a higher frequency in Newman than in the saeS L strain. The differences in expression level are small, but could well be significant as several ØNM4 genes are known to contribute to pathogenesis [43,44]. It is anticipated that prophage induction would ultimately result in lysis of bacterial cells and release of cell constituents into the environment.
To determine if the biofilm inhibitory protein might be encoded and/or regulated by a strain Newman prophage, we tested S. aureus strain TB3 for biofilm formation. Strain TB3 is a derivative of Newman which has been cured of prophages NM1, NM2, and NM4 [43]. We also tested strain TB1(pCWsae51). Strain TB1 is an NM3-deleted strain but also has a mutation in the sae locus (T. Bae, personal communication). We therefore constructed TB1(pCWsae51) to provide a complemented copy of saeS P . None of the prophage deleted strains produced a biofilm (S2 Fig) suggesting that the sae regulated biofilm inhibitor is not encoded by any of the Newman prophage.

Proteases are not involved in biofilm inhibition
It has been reported that proteases play a key role in biofilm inhibition and the transcriptional regulator, SarA, is the major repressor of proteinase production. Ten proteases were recognized in S. aureus. In strain Newman, SaeRS act synergistically with SarA to repress proteases, but the SaeRS effect is less than SarA [12]. Two proteinases, SspB and Aur were found in Newman culture supernatants, but mutations of the sspB or aur genes did not affect biofilm inhibition (S1 Table). In our RNA-seq analysis, we also found that the six serine proteinases encoded by the splABCDEF operon were under positive regulation by saeRS (S2 Table), thus it seemed possible that one or more of the spl-encoded enzymes could be the biofilm inhibitory factor. However, deletion of the spl operon had no effect on biofilm (S3 Fig). Since the other two proteases were not affected by SaeRS, these results indicate that the inhibitor per se is unlikely a protease or a group of proteases.
Because SaeS P is a negative regulator of extracellular proteinases [12,15] (S2 Table), it seemed possible that the putative inhibitory protein may be degraded by a proteinase in the saePQRS::kan strain such that the supernatant is no longer capable of inhibiting biofilm formation. To test this possibility, experiments were performed using mixed culture supernatants following the rationale that a proteinase in the saePQRS::kan mutant could degrade the inhibitory protein in wild-type strain Newman supernatants. Supernatants were combined in ratios of Newman: saePQRS::kan at 4:1, 1:1 and 1:4 and incubated overnight at 37°C (S4 Fig). Supernatant mixtures were then TCA precipitated and protein pellets re-suspended in the initial volume of strain Newman supernatant. Samples were dialyzed overnight to remove residual TCA and tested for anti-biofilm activity against strain UAMS-1. Every sample that contained strain Newman supernatant exhibited anti-biofilm activity, regardless of the amount of saePQRS::kan supernatant included in the incubation (S4 Fig) suggesting the inhibitory factor is not sensitive to saeRS-repressed proteinases, and expression is likely transcriptionally regulated by saeRS.

Conclusion
In summary, RNA-Seq showed that the saeS P allele of strain Newman affects expression of nearly 125 genes, relative to strains with the saeS L allele. The RNA-Seq analyses confirmed that the saeS P allele has a profound effect on gene expression, including numerous genes that may play a role in biofilm formation. Genes affected by SaeS P include those that encode autolysins and extracellular nuclease. In accordance, extracellular DNA can be found in culture supernatants of sae mutants, but is essentially undetectable in Newman cultures. Inactivation of the Newman nuc gene also increased the level of extracellular DNA, but the mutation was insufficient to permit biofilm formation. Expression of the ica genes and production of PIA is associated with biofilm formation by some staphylococci but we found no evidence that sae affects ica expression in strain Newman at least under the conditions we used. Wild type Newman (saeS P ) produces a factor that inhibits biofilm by the producing strain, as well as biofilm formation by other strains of S. aureus. The inhibitory factor is not produced at a detectable level by a Newman sae deleted strain or a Newman mutant expressing SaeS L . The inhibitory factor was determined to be a high-molecular weight, thermostable, extracellular protein, but extensive biochemical and genetic analyses have not revealed the identity of the inhibitor. The inhibitory protein appears to be highly potent as we have detected activity in preparations of low protein content. Our results also indicate that the inhibitor is not likely a protease or a group of proteases. We are currently using genetic methods to identify Newman mutants that either do not express or exhibit reduced expression of the inhibitory protein. Without knowing its identity, it is difficult to speculate on the inhibitor's mechanism of action. However, it is possible that it alters the bacterial surface or inhibits biofilm promoting surface proteins thereby preventing biofilm formation.

Bacterial strains and culture conditions
The bacterial strains and plasmids used in this study are listed in Table 1. S. aureus strains were cultivated in Trypticase soy broth without added glucose (TSB-0g) (17 g/L tryptone; 3 g/L soytone; 5 g/L NaCl; 2.5 g/L K 2 HPO 4 3H 2 O), TSB (Difco) or tryptic soy agar (TSA) (Difco). Antibiotics were added to S. aureus culture media, as appropriate, at final concentrations of 10 μg per ml chloramphenicol (Cm), 3 μg per ml tetracycline (Tc), or 50 μg per ml kanamycin (Kn). Escherichia coli strain DH5α was used for plasmid construction and maintenance. E. coli was cultivated in Luria-Bertani broth or agar (Difco) supplemented with, as appropriate, 100 μg per ml penicillin (Pen), 34 μg per ml Cm or 50 μg per ml Kn. The mutants used for screening

Biofilm Assay
Biofilm assays were performed as described [11,46] with a few modifications. The medium used was TSB-0g supplemented with 1% NaCl and 0.5% glucose (biofilm medium). Prior to inoculation, microtiter plates were pre-coated with 20% human plasma proteins (Sigma) and incubated for at least one hour at 4°C. After inoculation, plates were incubated at 37°C for 16 h. For experiments that tested anti-biofilm activity of culture supernatants, supernatants were added to fresh biofilm media at 25% vol/vol. Cultures used to harvest supernatants were grown to stationary phase in TSB-0g. To test fractionated culture supernatants for anti-biofilm activity, 20 μl of each fraction was added to each well of a microtiter plate and 180 μl of biofilm medium, pre-inoculated with strain UAMS-1, was added to each well.

Attachment assays
Stationary phase cultures of strain Newman and the saePQRS::kan mutant were diluted to an OD 660 of 0.05, grown to an OD 660 of 1 and diluted to OD 660 0.1 in biofilm medium. The cells were then inoculated into a 24-well plate pre-coated with plasma proteins. Bacteria were allowed to attach for 1 h at 37°C and then were washed, fixed, and stained with crystal violet.

Protein fractionation
For ammonium sulfate fractionation of culture supernatants, solid ammonium sulfate was added to filter sterilized culture supernatants at concentrations of 30%, 45% or 60% (w/v) and incubated at 4°C for at least one hour. Precipitated proteins were collected by centrifugation at 10,000 x g for 30 min. Protein pellets were resuspended in 1X PBS and dialyzed against PBS to remove residual ammonium sulfate.

Size exclusion and ion exchange chromatography
For chromatography experiments, stationary phase cultures of strain CYL12640 were harvested and concentrated 100-fold using Amicon Ultra centrifugal filters with a 30 kDa molecular weight cut off. For size exclusion chromatography, this sample was separated using a HiPrep 16/60 Sephacryl S-300 HR column (GE Healthcare) without dialysis. For ion exchange chromatography, the sample was separated using a HiTrap Q XL column (GE Healthcare) after dialysis against 50 mM Tris, pH 7.6. For parallel sae mutant comparisons, strain CYL12688 was fractionated in the same manner. In-gel trypsin digestion and tandem mass spectrometry for protein identification was performed by the Proteomics core in the Translational Research Institute at the University of Arkansas for Medical Sciences.

RNA Isolation and rRNA removal
S. aureus strains were grown overnight in TSB-0g, diluted 1:100 in fresh TSB-0g and grown for 5 h (OD 660~2 .0) in a 10:1 flask:volume ratio. The cultures were mixed with an equal volume of an ice-cold 1:1 mixture of ethanol-acetone and kept frozen at −80°C for at least 2 hours. RNA was isolated as described previously [47]. Isolated RNA was then enriched for mRNA using the MicrobExpress kit (Ambion). mRNA samples were examined for rRNA depletion using a 2100

Real-time RT-PCR
To confirm RNA-Seq data, we selected genes that were regulated by sae and assessed their relative expression levels by real-time RT-PCR as described previously [48] using the primer pairs listed in S10 Table. Cycling conditions were as previously described [48].

Zymographic analysis
For profiling S. aureus autolysis, cultures were grown in TSB for 3 h (OD660 of~1.7) and harvested by centrifugation. Surface-associated proteins were prepared with 4% SDS as described [47]. A modified method of Sugai et al. [49] was used for bacteriolytic enzyme profiling analysis as previously described [47]. Samples (15 μl) were mixed with 4X SDS sample buffer (final DTT concentration of 12.5 μM), incubated at room temperature for 30 min, and loaded on to 10% SDS-PAGE gels containing heat-killed S. aureus RN4220 (about 2 mg per ml of gel solution) or heat killed Micrococcus luteus (Sigma). After electrophoresis, gels were first washed 3 times each with 250 ml deionized water for 30 min with slow shaking at room temperature and then washed with 250 ml buffer A [50 mM Tris-HCl (pH 7.6), 200 mM NaCl, 5 mM CaCl 2 at room temperature for 30 min. The gels were incubated overnight with fresh buffer A at 37°C without shaking and then scanned with dark background.  We thank the Genomics and Bioinformatics division of the Laboratory for Molecular Biology and Cytometry Research at OUHSC which provided RNAseq and assistance with NGS data analysis.