Factors Contributing to the Biofilm-Deficient Phenotype of Staphylococcus aureus sarA Mutants

Mutation of sarA in Staphylococcus aureus results in a reduced capacity to form a biofilm, but the mechanistic basis for this remains unknown. Previous transcriptional profiling experiments identified a number of genes that are differentially expressed both in a biofilm and in a sarA mutant. This included genes involved in acid tolerance and the production of nucleolytic and proteolytic exoenzymes. Based on this we generated mutations in alsSD, nuc and sspA in the S. aureus clinical isolate UAMS-1 and its isogenic sarA mutant and assessed the impact on biofilm formation. Because expression of alsSD was increased in a biofilm but decreased in a sarA mutant, we also generated a plasmid construct that allowed expression of alsSD in a sarA mutant. Mutation of alsSD limited biofilm formation, but not to the degree observed with the corresponding sarA mutant, and restoration of alsSD expression did not restore the ability to form a biofilm. In contrast, concomitant mutation of sarA and nuc significantly enhanced biofilm formation by comparison to the sarA mutant. Although mutation of sspA had no significant impact on the ability of a sarA mutant to form a biofilm, a combination of protease inhibitors (E-64, 1-10-phenanthroline, and dichloroisocoumarin) that was shown to inhibit the production of multiple extracellular proteases without inhibiting growth was also shown to enhance the ability of a sarA mutant to form a biofilm. This effect was evident only when all three inhibitors were used concurrently. This suggests that the reduced capacity of a sarA mutant to form a biofilm involves extracellular proteases of all three classes (serine, cysteine and metalloproteases). Inclusion of protease inhibitors also enhanced biofilm formation in a sarA/nuc mutant, with the combined effect of mutating nuc and adding protease inhibitors resulting in a level of biofilm formation with the sarA mutant that approached that of the UAMS-1 parent strain. These results demonstrate that the inability of a sarA mutant to repress production of extracellular nuclease and multiple proteases have independent but cumulative effects that make a significant contribution to the biofilm-deficient phenotype of an S. aureus sarA mutant.


Introduction
Staphylococcus aureus is an opportunistic pathogen capable of causing diverse forms of infection. Treatment of these infections is complicated not only by the continued emergence of antibioticresistant strains but also by the fact that many S. aureus infections are associated with formation of a biofilm, which limits the efficacy of antimicrobial therapy even in cases caused by strains that are not clinically defined as resistant to the relevant antibiotics [1,2]. For this reason, the effective treatment of biofilm-associated staphylococcal infections often requires surgical debridement to remove infected tissues and/or devices [3,4].
Previous reports have implicated many S. aureus genes in biofilm formation. These include agr [5,6], arlRS [7], bap [8], hla [9], ica [10], rbf [11], sarA [12,13], sigB [14,15], tcaR [16], and traP [17,18]. However, in almost all cases, there are conflicting reports with respect to the overall contribution of specific loci. For example, some reports have concluded that ica and sigB are required for S. aureus biofilm formation [10,14,15], while others have found that mutation of these loci has little impact [13,19,20]. Similarly, there is a report concluding that alpha hemolysin is required for biofilm formation [9], but S. aureus isolates unable to produce alpha toxin owing to a nonsense mutation in the corresponding gene (hla) are capable of forming a biofilm and causing biofilm-associated infection [19,21]. In fact, a recent report concluded that nonhemolytic variants arise spontaneously within a biofilm and ultimately become the dominant subpopulation [22].
Such conflicting reports may be due to strain-dependent differences among S. aureus isolates. For instance, bap encodes a surface-associated protein (Bap) that promotes biofilm formation, but to date it has been found only in bovine mastitis isolates and even then only rarely [23,24]. Many studies focusing on biofilm formation have also utilized strains derived from NCTC 8325. This includes RN6390, which is an 8325-4 strain in which three prophage were cured from NCTC 8325 [25], and SA113, which is a mutagenized, restriction-modification deficient derivative of 8325 [10,26]. All 8325-derived strains carry natural mutations in rsbU, which renders them functionally sigB deficient [27][28][29]. They also carry a mutation in tcaR [16], a regulatory locus that has also been implicated in biofilm formation [30]. Perhaps owing to these mutations, the 8325-4 strain RN6390 has a reduced capacity to form a biofilm by comparison to clinical isolates of S. aureus [19,21].
One exception to such conflicting reports is the staphylococcal accessory regulator (sarA), mutation of which has been consistently correlated with a reduced capacity to form a biofilm in both S. aureus and S. epidermidis [12,13,21]. Indeed, with the exceptions of RN6390 and Newman, the latter also being a poor biofilm former [12] that has specific characteristics (e.g. production of truncated fibronectin-binding proteins) that distinguish it from primary clinical isolates [31], mutation of sarA has resulted in a reduced capacity to form a biofilm in vitro in every S. aureus strain we have examined [12]. Mutation of sarA in the clinical isolate UAMS-1 was also shown to result in a significant decrease in biofilm formation in vivo as defined using a catheter-based murine model [19].
The sarA locus encodes a DNA-binding protein (SarA) that has a global impact on gene expression in S. aureus [21,32], and it remains unclear which components of this global response are most relevant to biofilm formation. Mutation of sarA does result in reduced expression of the ica operon and consequently reduced production of the polysaccharide intercellular adhesion (PIA), but our direct comparison of sarA and ica mutants generated in the same strain used in the experiments reported here (UAMS-1) demonstrate that this cannot account for the biofilm defect in a sarA mutant [19].
As a first step toward defining the role of sarA in biofilm formation, we compared the regulons defined by growth within a biofilm and by mutation of sarA. We found that a large number of genes were differentially expressed in a mature biofilm by comparison to both exponential and post-exponential phase planktonic cultures [19] and that many of these were also in the sarA regulon [33]. Included among these genes was the bicistronic operon alsSD, which encodes the enzymes (acetolactate synthase and acetolactate decarboxylase) required for the conversion of pyruvate to acetoin rather than the more acidic products of glucose metabolism. Specifically, expression of alsSD was increased in a biofilm by comparison to both exponential and postexponential planktonic growth [12]. Conversely, alsSD expression was decreased in a sarA mutant [33]. This suggests that the inability to express alsSD at adequate levels in a sarA mutant may contribute to its inability to form a biofilm.
Also included in both the biofilm and sarA regulons was nuc, which encodes the S. aureus thermostable nuclease. The expression pattern of nuc was opposite that of alsSD in that it was decreased in a biofilm but increased in a sarA mutant [19,33]. This observation, together with recent reports demonstrating that extracellular DNA (eDNA) contributes to S. aureus biofilm formation [34,35], suggest that the inability to repress production of extracellular nuclease may also contribute to the biofilm-deficient phenotype of a sarA mutant.
Expression of the genes encoding extracellular proteases was also altered both in a biofilm and a sarA mutant, but in this case the scenario is more complex in that expression of some protease genes (e.g. scpA) was decreased in a biofilm while expression of others (e.g. sspAB) was increased [19]. However, expression of all of these genes, as well as the gene encoding aureolysin (aur) was increased in a sarA mutant [33]. This is consistent with the observation that the overall production of extracellular proteases is increased in a sarA mutant [36][37][38]. This is potentially important in that biofilm formation in clinical isolates of S. aureus is facilitated by coating the substrate with plasma proteins [12], and the increased protease production observed in sarA mutants has been correlated with a reduced capacity to bind host proteins including fibronectin [36,39]. Moreover, a recent report concluded that the agr-mediated induction of protease production limits biofilm formation and may play a functional role with respect to the dispersal of S. aureus from an established biofilm [40]. Such results suggest that the reduced capacity of a sarA mutant to form a biofilm could also be related to the increased production of one or more extracellular proteases.
Based on these considerations, we investigated the role of alsSD, nuc, and extracellular proteases in S. aureus biofilm formation with a specific emphasis on whether any or all of these factors contribute to the biofilm-deficient phenotype of a sarA mutant.

Results And Discussion
Impact of sarA on alsSD expression, acetoin production, stationary phase survival and biofilm formation The alsS and alsD genes constitute a bicistronic operon and encode acetolactate synthase and acetolactate decarboxylase respectively. These enzymes function sequentially to convert pyruvate to 2acetolactate and then acetoin, the latter ultimately being converted by acetoin reductase to 2,3-butanediol. Production of these neutral products rather than the more acidic products of carbohydrate metabolism is important for acid tolerance in a number of bacterial species [41]. This is consistent with the observation that alsSD is expressed at elevated levels in response to mild acid treatment of S. aureus [42]. Our previous demonstration that expression of alsSD is also increased in a biofilm by comparison to planktonic cultures is therefore consistent with the hypothesis that a central theme behind the adaptation of S. aureus to persistence within a biofilm is survival within the acidic environment associated with carbohydrate metabolism, particularly under low oxygen conditions [19]. It is also consistent with a recent report demonstrating that Streptococcus mutans is more acid tolerant when grown in a biofilm than when grown in planktonic culture [43]. Moreover, production of the enzymes required for acetoin production in Vibrio cholerae is coregulated with expression of genes directly involved in the switch between motility and biofilm formation [41]. This co-regulation provides further support for the hypothesis that acetoin production and acid tolerance are important for biofilm formation in diverse bacterial species.
A previous report described construction of a UAMS-1 alsSD mutant (KB1097) and concluded that mutation of alsSD eliminated acetoin production and resulted in reduced murein hydrolase activity and reduced stationary phase survival [44]. We subsequently demonstrated that KB1097 also has a reduced capacity to form a biofilm that is comparable to that observed with a UAMS-1 sarA mutant [33]. However, it was later discovered that KB1097 was a contaminant that ultimately proved to be S. hominis rather than S. aureus (Dr. Ken Bayles, personal communication). Based on this, we independently generated an alsSD mutant (UAMS-1489) using the pKOR1 mutagenesis system [45]. Characterization of this mutant confirmed the absence of alsSD transcription (Fig. 1), the inability to produce acetoin and 2,3-butanediol as defined by the Voges-Proskauer assay (Fig. 2), and reduced stationary-phase survival (Fig. 3). The results also confirmed that mutation of sarA impacts all three phenotypes in a similar although less definitive manner (Figs. 1-3). Additionally, the capacity to form a biofilm was significantly reduced in UAMS-1489 by comparison to the UAMS-1 parent strain (p,0.001) (Fig. 4). Transcription of alsSD, the Voges-Proskauer phenotype, stationary-phase survival, and biofilm formation were all complemented by introducing a functional alsSD operon into UAMS-1489 to generate UAMS-1551 ( Figs. 1-4). These results are consistent with the hypothesis that alsSD plays a role in biofilm formation in S. aureus.
While UAMS-1489 had a reduced capacity to form a biofilm, it remained significantly greater than that observed with the UAMS-929 sarA mutant (p,0.001) (Fig. 4). This was true despite the fact that mutation of alsSD had a greater impact than mutation of sarA on alsSD transcription, the Voges-Proskauer assay, and stationaryphase survival ( Fig. 1-4). This demonstrates that the impact of sarA on biofilm formation extends beyond its impact on transcription of alsSD. Confirmation of this was obtained by analysis of the sarA/ alsSD double mutant (UAMS-1300), which demonstrated that concomitant mutation of alsSD and sarA had no further impact on the production of acetoin (data not shown) but did reduce biofilm formation even by comparison to an alsSD mutant (p,0.001) (Fig. 4).
Although these results demonstrate that the impact of sarA on alsSD transcription cannot fully account for the biofilm-deficient phenotype of a sarA mutant, mutation of alsSD did have a significant impact on biofilm formation by comparison to the parent strain (Fig. 4), and this leaves open the possibility that the impact of sarA on alsSD transcription makes an important contribution in that regard. Based on this, we also generated derivatives of UAMS-929 (UAMS-1729 and UAMS-1730) in which alsSD expression was enhanced in a sarA mutant. Although this did not fully restore alsSD transcription to wild-type levels  . Production of acetoin and 2,3-butanediols in alsSD and sarA mutants. 18-hour culture supernatants were assayed for acetoin production using the Voges-Proskauer assay. Strain designations are the same as those cited in Fig. 1 Fig. 5A), which might be expected given the positive impact of sarA on alsSD transcription [33], it did restore transcription to levels sufficient to fully restore the Voges-Proskauer phenotype (Fig. 5B). However, it had no impact on the ability of a sarA mutant to form a biofilm (Fig. 5C). This provides further support for the conclusion that mutation of sarA results in a defect in biofilm formation that extends beyond its impact on expression of alsSD.
One possible explanation for these results is that alsSD is part of a pathway that includes acetoin reductase, which is required to convert acetoin to 2,3-butanediol, and expression of the corresponding gene (SA0239 in the N315 genome) was also reduced in a sarA mutant [33]. Thus, restoration of acetoin production in a sarA mutant may not fully restore the ability to convert pyruvate to 2,3-butanediol. While such an effect might not be evident in the Voges-Proskauer assay, it could nevertheless compromise other aspects of metabolism including the ability to maintain pH homeostasis. However, subsequent analysis of all relevant strains failed to reveal any significant difference in the pH of culture supernatants from overnight cultures grown in biofilm medium, which includes exogenous glucose (data not shown). The alternative explanation is that sarA has an impact on biofilm formation that involves genes unrelated to the acetoin/2,3butanediol pathway.
Contribution of thermostable nuclease to the biofilmdeficient phenotype of a sarA mutant Previous genome-scale transcriptional profiling experiments focusing on UAMS-1 identified a number of other sarA-regulated genes that were also differentially expressed in a biofilm by comparison to planktonic cultures [19,33]. Included among these genes was nuc, which encodes the S. aureus thermostable nuclease. Specifically, expression of nuc was reduced in a biofilm and increased in a sarA mutant [12,33]. Recent reports demonstrating that extracellular DNA contributes to biofilm formation in S. aureus  [34,35] are consistent with the hypothesis that this may also contribute to the biofilm-deficient phenotype of a sarA mutant.
To address this hypothesis, we also generated UAMS-1 nuc and sarA/nuc mutants and evaluated the impact on biofilm formation. Phenotypic assays confirmed that mutation of sarA increased nuclease production and that it was completely abolished by mutation of nuc even in a sarA mutant (Fig. 6). Mutation of nuc in UAMS-1 (UAMS-1471) had no impact on biofilm formation, but complementation of the nuc mutation with a plasmid-borne version of the nuc gene (UAMS-1552) did limit biofilm formation to a significant degree (p = 0.001) (Fig. 7), presumably because it resulted in increased production of nuclease by comparison to the parent strain (Fig. 6). More importantly, concomitant mutation of nuc in a sarA mutant (UAMS-1477) enhanced biofilm formation by comparison to the sarA mutant (p,0.001) (Fig. 7). This effect was completely reversed by complementation of the sarA/nuc double mutant with a plasmid-borne version of the nuc gene (UAMS-1725) (Fig. 7). At the same time, biofilm formation in the sarA/nuc mutant (UAMS-1477) was not restored to wild-type levels. In contrast, complementation of the sarA/nuc mutant with sarA did fully restore biofilm formation (Fig. 7). These results indicate that increased production of extracellular nuclease contributes to the biofilm-deficient phenotype of a UAMS-1 sarA mutant but that additional factors must also be involved.
Contribution of extracellular proteases to the biofilmdeficient phenotype of a sarA mutant Also included in both the biofilm and sarA regulons were most of the S. aureus genes encoding extracellular proteases, the only exception being the spl operon (splABCDEF), expression of which was largely unaltered both in a biofilm and in a sarA mutant. For example, expression of scpA was decreased in a biofilm but elevated in a sarA mutant. Expression of the gene encoding aureolysin (aur) was not altered in a biofilm but was increased in a sarA mutant [19,33]. Expression of the sspABC operon was elevated in a biofilm and a sarA mutant, but because no direct comparison has been made, it is not known whether expression of sspABC is elevated in a sarA mutant even by comparison to a biofilm.
It is clear that overall extracellular protease activity is elevated in sarA mutants [36][37][38], but to date studies that have attempted to address the issue have concluded that this has relatively little impact on the biofilm-deficient phenotype of a sarA mutant. For instance, mutation of sspA, which encodes the classic ''V8'' serine protease (SspA), had no impact on biofilm formation in either UAMS-1 or its sarA mutant [33]. Similarly, Valle et al. (2003) concluded that the increased production of proteases in sarA mutants was unlikely to explain their biofilm-deficient phenotype based on the observations that 1) the capacity of a sarA mutant to form a biofilm was not enhanced in the presence of the protease inhibitors a 2 -macroglobulin or E64, 2) the capacity of the wildtype strains (ISP479C and 15981) to form a biofilm was not reduced by incubation in concentrated supernatants from sarA mutants, and 3) concomitant mutation of sarA and sspA or sarA and the aureolysin gene (aur) failed to enhance biofilm formation.
In contrast to such studies, Boles and Horswill (2008) recently concluded that elevated protease production can be correlated with reduced biofilm formation and that the increased production of extracellular proteases plays an important role in agr-mediated dispersal from an established biofilm. These studies were limited to derivatives of SH1000, which is an rsbU-repaired derivative of 8325-4, and they did not address the potential contribution of extracellular proteases to the biofilm-deficient phenotype of sarA mutants, but    such results nevertheless suggest that previous studies may not have fully defined the impact of protease production in this context. Based on this, we employed a combination of three protease inhibitors in an attempt to collectively limit the activity of all recognized S. aureus extracellular proteases. Specifically, E-64, 1-10-phenanthroline, and dichloroisocoumarin (DIC) inhibit the activity of cysteine proteases (e.g. ScpA and SspB), metalloproteases (e.g. aureolysin), and serine proteases (e.g. SspA and the Spl proteases) respectively.
Each inhibitor was added to the medium used in our biofilm assays at a concentration of 1 mM or, in the case of phenanthroline, the highest concentration (10 mM) that did not limit growth (Fig. 8).
In fact, addition of these inhibitors both alone or in combination with each other increased growth of the sarA mutant to a limited degree at least as defined by the maximum optical density of post-exponential phase cultures (Fig. 8). This suggests that the increased production of proteases may actually limit the growth of an S. aureus sarA mutant. Subsequent analysis using skim milk-agar plates confirmed that this combination of protease inhibitors also decreased overall proteolytic activity by comparison to the sarA mutant (data not shown). More importantly, at least in the context of this report, inclusion of these inhibitors also enhanced biofilm formation in a sarA mutant (p,0.001) (Fig. 9A). Although a slight increase in biofilm formation was also observed with the parent strain, analysis of the relative effects confirmed that the impact in a sarA mutant significantly exceeded that observed in UAMS-1 (Fig. 9B). These results indicate that the increased production of extracellular proteases also makes an important contribution to the biofilm-deficient phenotype of a UAMS-1 sarA mutant.
One possible explanation is that the presence of protease inhibitors increased growth of the sarA mutant, particularly with respect to the maximum density observed with post-exponential phase cultures (Fig. 8). However, we observed a similar degree of growth enhancement when each inhibitor was included alone as well as in combination with each other, and this was not the case with respect to biofilm formation. Specifically, E-64 was the only individual inhibitor in which a significant increase in biofilm formation was observed either alone or in combination with DIC or phenanthroline (p,0.001) (Fig 10). This suggests that the effect of the inhibitor cocktail on biofilm formation cannot be explained by its impact on growth of the sarA mutant in biofilm medium. It also indicates that the cysteine proteases ScpA and/or SspB play an important role. This is in contrast to the results of Valle et al. (2003), who concluded that E-64 has no impact on the biofilmdeficient phenotype of a sarA mutant. However, the concentration of E-64 employed in these earlier experiments (10 mM) was considerably lower than that used in our experiments (1 mM). At the same time, the impact of E-64 even at the higher concentration was limited by comparison to both the wild-type strain (data not shown) and the sarA mutant grown in the presence of all three inhibitors (p = 0.012) (Fig. 10), and this clearly suggests that sarAregulated proteases other than ScpA or SspB are also involved.
In an effort to further define the role of specific proteases, we also examined the impact of mutating sspA in both UAMS-1 (UAMS-960) and its sarA mutant (UAMS-962). In both cases, mutation of sspA had no significant impact on biofilm formation (Fig. 9A). Additionally, the enhanced biofilm formation observed with a UAMS-1 sarA mutant grown in the presence of all three inhibitors was also observed with a sarA/sspA double mutant (p,0.001) (Fig. 9B). This indicates that the impact of the protease inhibitors on biofilm formation was independent of SspA. This is consistent with the results of Valle et al. (2003), who also found that mutation of sspA had no impact on biofilm formation in a sarA mutant.
To further examine this issue, we carried out zymogram analysis using both casein and gelatin gels. This analysis confirmed that both the inhibitor cocktail and E-64 alone or in combination with either DIC or phenanthroline reduced the activity of cysteine proteases to undetectable levels as defined by the sensitivity of our zymogram assays (Fig. 11). In contrast, we could not demonstrate significant inhibition of serine or metalloproteases with DIC or phenanthroline. Moreover, production of SspA was unaffected by the inclusion of DIC in the growth medium (Fig. 11). To the extent that DIC is a specific inhibitor of serine proteases, including SspA, this suggests that the concentration of DIC used in our experiments was below the level required to have a phenotypic effect. We could not determine whether the same was true for the spl-encoded serine proteases because we could not detect the activity of these enzymes in either of our zymograms. This is perhaps not surprising since the activity of these proteases was recently shown to be dependent on their binding of specific substrates in a manner analogous to the exfoliative toxins [46,47]. Similarly, we could not demonstrate inhibition of the metalloprotease aureolysin with phenanthroline because we could not detect the activity of this protease even in a sarA mutant (Fig. 11). Nevertheless, the fact that the combination of all three proteases inhibitors enhanced biofilm formation to an extent that exceeded that observed with any single or even dual combination of protease inhibitors (Fig. 10) suggests that the concentrations of both DIC and phenanthroline used in these experiments were in fact inhibitory. It is also consistent with the hypothesis that extracellular proteases other than the cysteine proteases ScpA and SspB also make an important contribution to the biofilm-deficient phenotype of a sarA mutant. To the extent that this effect was independent of sspA (see above), this suggests that the spl-encoded serine proteases and/or aureolysin may be particularly important in this regard. Although confirmation of this hypothesis will require examination of UAMS-1 sarA/aur and sarA/spl mutants, this is consistent with the results of Boles and Horswill (2008), who detected increased levels of serine proteases in the effluent from S. aureus biofilms and demonstrated that concomitant mutation of aur and spl resulted in an enhanced capacity to form a biofilm at least in the 8325-4 strain SH1000.
While the combination of protease inhibitors enhanced biofilm formation in a sarA mutant, it did not restore it to wild-type levels (Fig. 9A). One explanation for this is that the concentration of inhibitors used in our experiments was sufficient to limit protease activity in a sarA mutant but not sufficient to restore it to wild-type levels. As discussed above, this is particularly true with respect to DIC and phenanthroline. It was not possible to use either of these inhibitors at a higher concentration either due to issues related to limited solubility in biofilm medium (DIC) or inhibition of growth (phenanthroline), so addressing this possibility will also require detailed analysis of relevant protease mutants. Nevertheless, the results discussed above suggest that the increased production of both extracellular nuclease and multiple extracellular proteases contribute to the biofilm-deficient phenotype of an S. aureus sarA mutant. However, they also demonstrate that the biofilm-defect in a sarA mutant cannot be explained by its impact on either nuclease or proteases alone. To determine whether these effects might be cumulative, we also examined whether inclusion of protease inhibitors had an impact on bioflm formation with a sarA/nuc mutant. The results confirmed that the effect was cumulative in that biofilm formation was enhanced in a sarA/nuc mutant in the presence of protease inhibitors by comparison to both a sarA mutant in the presence of protease inhibitors (p,0.001) and a sarA/nuc mutant in the absence of inhibitors (p,0.001) (Fig. 9A). In fact, in the presence of the inhibitor cocktail, the sarA/nuc mutant exhibited a level of biofilm formation approaching that of the UAMS-1 parent strain.

Conclusions
Taken together, our results provide important clues with respect to explaining the biofilm-deficient phenotype of a S. aureus sarA mutant, and there may in fact be a common theme that ties them all together. For instance, a recent report demonstrated that mutation of cidA, which encodes a regulator of murein hydrolase activity, results in reduced release of extracellular DNA and a reduced capacity to form a biofilm [35]. The cidA gene is part of the cidR regulon, which also includes alsSD [44]. Our results confirmed that mutation of alsSD in UAMS-1 results in a stationary-phase survival defect manifested as cell death in the absence of cell lysis. This suggests that mutation of alsSD may limit the release of extracellular DNA as well as limit the capacity to produce acetoin and maintain pH homeostasis. In fact, one explanation for our results is that mutation of alsSD limits biofilm formation because it limits the availability of extracellular DNA but that this effect is modest by comparison to the impact of sarA on the production of extracellular nuclease. Such a scenario might also provide an alternative explanation for why restoration of alsSD expression in a sarA mutant failed to restore biofilm formation since what little extracellular DNA would be available would presumably be degraded due to the increased production of nuclease.
How the production of proteases might fit into this scenario remains undefined. The production of active extracellular nuclease does require protease processing [48], but if this were a defining characteristic of the biofilm-deficient phenotype of a sarA mutant, then mutation of nuc and inclusion of protease inhibitors would not be expected to have a cumulative effect since mutation of nuc eliminates the production of nuclease in any form. This suggests that nuclease and protease have independent effects that remain to be defined but are perhaps related to the attachment vs. accumulation phases of biofilm formation. Given our experimental approach using a cocktail of protease inhibitors, it also remains to be determined exactly what proteases are involved, how they interact with each other, what S. aureus proteins are the most relevant targets, and what role these targets play in biofilm formation. In this respect it should be noted that biofilm formation in UAMS-1 and other clinical isolates is enhanced by coating the substrate with plasma proteins [12] and that, at least in UAMS-1, it is not dependent on PIA production [19]. Moreover, we have demonstrated that the increased production of proteases in sarA mutants results in a decreased capacity to bind host proteins including fibronectin [36]. Taken together, these results suggest that the negative impact of proteases on biofilm formation may be multifactorial and involve both attachment and accumulation. A recent report also demonstrated that induction of agr expression leading to increased production of multiple proteases may also serve as a specific means of dispersal from an established biofilm [40].
While our experimental focus is on the role of sarA in biofilm formation, our results may also have broader implications for S. aureus regulatory circuits. For instance, sarA represses production of Figure 9. Effect of protease inhibitors on biofilm formation. Panel A: The microtiter plate assay of in vitro biofilm formation was performed with bacteria grown in biofilm medium (gray) or biofilm medium containing the inhibitor cocktail (black). Results represent the mean6standard deviation of 24 replicates. Panel B: Results of the biofilm assay shown in panel A represented as the ratio of biofilm formation in the presence of the inhibitor cocktail/biofilm formation in the absence of inhibitors. doi:10.1371/journal.pone.0003361.g009 many S. aureus exoproteins including nuclease and multiple proteases [33,36] while the accessory gene regulator (agr) has the opposite effect [25,40]. While there are reports concluding that reduced expression of agr can be correlated with a reduced capacity to form a biofilm [18,[49][50][51], most studies have concluded that agr expression limits rather than enhances biofilm formation [5,12,52]. In fact, it has been proposed that induction of agr expression may serve as a specific means of dispersal from an established biofilm [5,12,22,40].
One report suggested that the negative impact of agr on biofilm formation may be due to the increased production of delta-toxin, which is a phenol-soluble modulin (PSM) proposed to act as a surfactant to limit cellular accumulation [5,53]. The alternative explanation is that expression of agr results in the increased production of both extracellular nuclease and protease(s) [25]. This would presumably promote detachment and release from a mature biofilm based on degradation of the relevant S. aureus adhesins and/or extracellular matrix components including extracellular DNA. Indeed, Boles and Horswill (2008) recently concluded that agr-mediated protease production plays a primary role in this regard. Together with our results, this would suggest that sarA and agr have independent and opposite effects on biofilm formation, with the ultimate impact being dependent on the relative contribution of each with respect to the other.
In this scenario, S. aureus strains that express agr at high levels would presumably have a reduced capacity to form a biofilm. This is consistent with our observation that the 8325-4 strain RN6390, which by comparison to most clinical isolates including UAMS-1 expresses agr at high levels and produces elevated amounts of nuclease and extracellular proteases [36], has a limited capacity to form a biofilm [12]. It is also consistent with the observation that mutation of agr enhances biofilm formation in RN6390 [12]. This effect is reversed by concomitant mutation of sarA [12], which suggests that the impact of sarA is epistatic to agr at least in this context.
It is unclear whether this has any relevance with isolates other than RN6390, which as noted above has specific characteristics that distinguish it from clinical isolates of S. aureus. However, a recent report concluded that most community-acquired MRSA (CA-MRSA) isolates express agr at higher levels than their healthcare-associated MRSA (HA-MRSA) counterparts [53]. Our studies done with a USA300 CA-MRSA isolate suggest that this does not preclude biofilm formation at least in this strain [52], but there are reports demonstrating a general inverse relationship between the level of agr expression and biofilm formation [5]. Together with the fact that agr expression is also correlated with the increased production of S. aureus extracellular proteins and PSMs, the latter having both surfactant-like and anti-phagocytic properties [53], this could perhaps explain why CA-MRSA isolates often cause acute infections while HA-MRSA tend to cause chronic infections that are more likely to have a biofilm-associated component [54].
Finally, our previous studies characterizing the sarA and biofilm regulons in UAMS-1 [19,33] identified 43 genes that were differentially expressed both in a biofilm and in a sarA mutant [12,33]. Of the four possible scenarios, 17 genes were expressed at lower levels in both a biofilm and a sarA mutant, 6 genes were expressed at higher levels in both a biofilm and a sarA mutant, 17 genes were expressed at lower levels in a biofilm and higher levels in a sarA mutant, and 3 genes were expressed at higher levels in a biofilm and lower levels in a sarA mutant. Given the complex and interactive nature of regulatory circuits in S. aureus, it is difficult to predict which of these scenarios would be most important, and we certainly do not preclude the need to investigate additional genes in the sarA/biofilm regulon. However, our results suggest that the inability of a sarA mutant to repress production of specific extracellular proteins, including nuclease and multiple proteases, play a particularly important role in that regard. A detailed understanding of these processes is important given the role of biofilms not only in the development of many forms of S. aureus infection but also with respect to their impact on the ability to effectively treat these infections.

Bacterial strains and growth conditions
The strains utilized in this study are listed in Table 1. All strains were maintained as stock cultures at 280uC in tryptic soy broth (TSB) containing 25% (v/v) glycerol. For each experiment, the relevant strains were retrieved from cold storage by plating on tryptic soy agar (TSA) with appropriate antibiotic selection. Antibiotics were used at the following concentrations: kanamycin (Kan; 50 mg per ml), neomycin (Neo; 50 mg per ml), chloramphenicol (Cm; 10 mg per ml), and erythromycin (Erm 5 mg per ml). To ensure that the results of phenotypic assays were consistent, all assays other than nuclease production (see below) were done using TSB supplemented with 0.5% glucose and 3.0% sodium chloride (biofilm medium) without antibiotic selection as previously described [12].

Mutagenesis
Mutagenesis of alsSD and nuc was done as previously described [52] using the pKOR1 mutagenesis system [45]. According to the annotation for the MRSA252 genome, which was previously shown to be the most closely related of the sequenced strains to UAMS-1 [21], the specific open-reading frames (ORFs) targeted in these experiments were SAR2297 (alsS), SAR2296 (alsD), and SAR0947 (nuc). The corresponding ORFs in the N315 genome are SA2008/ SA2007 (alsSD) and SA0746 (nuc). The oligonucleotide primers used for mutagenesis are listed in Table 2. In UAMS-1489, the deleted region started 650 bp downstream of the alsS start codon and ended 182 bp downstream of the alsD stop codon (primers alsSDMut1-FattB1, alsSDMut1RSacII, alsSD2FSacII, alsSD2RattB2).
In the case of nuclease, the deleted region starts 83 bp downstream of the nuc start codon and ends 180 bp upstream of the nuc stop codon (primers nucMut1FattB1, nucMut1RSacII, nucMut2FSacII, nuc-Mut2RattB2). Genotypic confirmation of all mutations was obtained by PCR using primers flanking the deleted region (data not shown). Because pKOR1-generated mutations are not marked by an antibiotic-resistance gene and cannot be transduced, mutagenesis of alsSD and nuc was done in both UAMS-1 and its corresponding sarA mutant (UAMS-929). The UAMS-1 sspA (SAR1022/SA0901) and sarA/sspA mutants were generated by W11-mediated transduction of the ssp::tet mutation from an existing 8325-4 mutant (kindly provided by Simon Foster, University of Sheffield) into UAMS-1 and UAMS-929 respectively. Generation of UAMS-929 is described elsewhere (Blevins et al., 2002).

Complementation of alsSD, nuc and sarA mutations
Complementation of the alsSD mutation was done by cloning the region spanning 431 bp upstream and 118 bp downstream of the alsSD operon (primers alsSDproF and alsSD downstream KpnI) into the E. coli/S. aureus shuttle vector pLI50 [55]. Complementation of the nuc mutation was done by cloning the region spanning 453 bp upstream of 277 bp downstream of the nuc gene (primers nuc comp- Figure 11. Zymogram analysis of protease production in the presence of protease inhibitors. Analysis of the production of specific proteases was assessed using casein (upper panel) and gelatin gels (lower panel) in UAMS-1 (WT) and its UAMS-929 sarA mutant. Analysis was done after growth in biofilm medium in the absence of any inhibitor (W), the presence of the inhibitor cocktail (CT), or the presence of each inhibitor alone and in combination with each other. Inhibitor designations are the same as in the legend for Fig. 10. The presumed identity of individual proteases is indicated to the right. In the case of SspA, this was confirmed by analysis of the corresponding sarA/sspA mutant (data not shown). In the case of ScpA and SspB, identification is presumptive based on their identity as cysteine proteases, inhibition of both by E-64, and the fact that ScpA has greater activity than SspB on casein while the opposite is true on gelatin (Dr. Jan Potempa, personal communication). doi:10.1371/journal.pone.0003361.g011 F and nuc comp-R) into pLI50. Complementation of the sarA mutation was done as previously described [36]. In all cases, complementing plasmids were first introduced into the S. aureus strain RN4220 before transduction into the corresponding UAMS-1 mutants using W11 as previously described [56].

Phenotypic assays
The Voges-Proskauer assay was used to assess the production of acetoin and 2,3-butanediol as previously described [44]. Briefly, cultures were grown for 18 hrs in biofilm medium (TSB supplemented with 0.5% glucose and 3.0% NaCl) with constant shaking before harvesting cell-free supernatants by centrifugation. The assay was done in 96-well microtiter plates by adding 50 ml of 0.3% creatine and 60 ml of 5% alpha-naphthol freshly prepared in 100% ethanol to 120 ml of fresh culture supernatant. After gentle mixing, 30 ml of 40% KOH was added to the reaction mixture, which was then incubated at room temperature for 30-45 min with occasional mixing. Results were assessed by measuring absorbance at 540 nm.
Biofilm formation was assessed in vitro using the static, microtiter plate biofilm assay as previously described [12]. To assess stationary-phase survival, overnight cultures of each S. aureus strain were grown in TSB with appropriate antibiotic selection and then used to inoculate 21 ml of NZY broth (Fisher Scientific, St. Louis, MO) containing 35 mM glucose to an OD 600 of 0.05. Flasks were loosely capped and grown at 37uC with constant shaking. After 8, 24, 48, and 72 hrs, the optical density (OD 600 ) of each culture was determined and an aliquot was removed to determine viable count by plating on TSA.
Nucleolytic activity was assessed using D'NASE Test Agar (REMEL, Lenexa, KS). Briefly, overnight cultures were standardized to an equal optical density before spotting 10 ml aliquots onto nuclease test agar. Plates were incubated overnight at 37uC. Nuclease activity was then assessed by overlaying the agar with 1N HCl to precipitate undigested DNA and define the zone of clearance around each strain.
Overall protease activity was assessed using skim milk agar as previously described [40], the only difference being that we analyzed standardized culture supernatants after 15-fold concentration using Centricon YM-3 filter units (Millipore, Bedford, MA). The activity of specific proteases was assessed by zymography using Ready Gel Zymogram Gels containing gelatin or casein (BioRad Laboratories, Hercules, CA). For casein gels, supernatants were analyzed without further processing. For gelatin gels, supernatants were first concentrated as discussed above. Samples in both cases were loaded Table 1. Strains used in this study.

Strain
Description Reference onto gels in a buffer containing DTT but without b-mercaptoethanol. After electrophoresis, gels were first incubated for 30 min at room temperature (RT) in renaturing buffer (2.5% TritonX-100) and then overnight at 37uC in developing buffer (0.2 M Tris, 5 mM CaCl 2 , 1 mM DTT). To visualize protease bands, gels were then stained with SimplyBlue SafeStain (Invitrogen, Carlsbad, CA) at RT for 2 hrs before destaining overnight with water.

Protease inhibitors
In experiments employing E-64 (Fisher Scientific, St. Louis, MO), and dichloroisocoumarin (DIC) protease inhibitors (Sigma Chemical Co., St. Louis, MO), each inhibitor was dissolved in biofilm medium at a 1 mM concentration. For experiments employing 1-10-phenanthroline (Fisher Scientific, St. Louis, MO), the concentration was reduced to 10 mM because higher concentrations inhibited growth. Subsequent experiments confirmed that these concentrations did not inhibit growth either alone or in combination with each other.

RNA isolation and qRT-PCR analysis
To assess relative levels of alsSD expression, total bacterial RNA was isolated using the Qiagen RNeasy Mini Kit as previously described [52]. Quantitative, real-time RT-PCR (qRT-PCR) was then performed [33] using alsSD-specific primers and a corre-sponding TaqMan probe (Table 2). Results were standardized by comparison to the results obtained with the same samples using primers and a TaqMan probe corresponding to the gyrB gene (Table 2).

Statistical analysis
Statistical comparisons were done using the Student's t-test or, where appropriate, the Mann Whitney Rank Sum Test as formatted in SigmaStat Statistical Software Version 2 (SPSS Inc., Chicago, IL). Because multiple comparisons were made within each data set, statistical significance, along with the corresponding p value, is noted in the text rather than within each figure.