Mapping the Global Network of Extracellular Protease Regulation in Staphylococcus aureus

The complex regulatory role of the proteases necessitates very tight coordination and control of their expression. While this process has been well studied, a major oversight has been the consideration of proteases as a single entity rather than as 10 enzymes produced from four different promoters. As such, in this study, we comprehensively characterized the regulation of each protease promoter, discovering vast differences in the way each protease operon is controlled. Additionally, we broaden the picture of protease regulation using a global screen to identify novel loci controlling protease activity, uncovering a cadre of new effectors of protease expression. The impact of these elements on the activity of proteases and known regulators was characterized by producing a comprehensive regulatory circuit that emphasizes the complexity of protease regulation in Staphylococcus aureus.

Given the complex regulatory role of S. aureus proteases during infection, it follows that there must be, and indeed is, tight control of their expression mediated by a collection of different factors. This is evidenced by the number of elements that have been identified thus far as influencing protease production, including RNAIII SarS, SarR, SarA, SarV, SarX, SarZ, ArlRS, CodY, Rot, MgrA, and SaeRS (21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33). Of these factors, SarS, SarR, CodY, Rot, MgrA, SaeR, and SarA are considered the primary regulators, with each being shown to directly influence protease transcription (21)(22)(23)(24)(25)(26)(27). A major oversight when studying the control of protease production in S. aureus, however, has been the consideration of these factors as a single entity rather than as 10 enzymes produced from four different promoters. Of the seven major regulators, only SarA and Rot have been explored in the context of all four protease promoters (9,10), with SarA shown to specifically repress the transcription of aur, scpA, and ssp but not spl (9,10), while Rot has been described as a direct negative regulator of all secreted protease operons (23). For the other primary regulators, CodY has been shown to directly repress ssp transcription (22), while SarS and SarR have been explored only in the context of aur and ssp promoter binding (21). Finally, MgrA has been shown to activate aur, ssp, and spl transcription (25,34), while SaeR has been described as an activator for spl but a repressor for aur (24).
Consequently, the overarching goal of this study was to explore and further our understanding of the regulation of secreted proteases by known regulatory factors in S. aureus while concurrently uncovering new effectors of protease transcription. Accordingly, we present a comprehensive mapping of protease regulation by all known S. aureus transcription factors in community-acquired methicillin-resistant S. aureus (CA-MRSA) strain USA300.

RESULTS AND DISCUSSION
Exploring the differential regulation of protease expression by primary regulators. To date, seven different transcriptional regulators (Rot, CodY, SarA, SarS, MgrA, SarR, and SaeR) (21)(22)(23)(24)(25)(26)(27) have been identified as being the primary modulators of secreted protease expression. An oversight, however, is the consideration of S. aureus proteases as a single entity rather than as 10 enzymes produced from four distinct loci (Fig. 1). Thus, although these elements do indeed have the capacity to regulate the expression of one or more proteases, only a few have been explored in the context of all four operons. Therefore, our initial goal was to fill in missing gaps using quantitative real-time PCR (qRT-PCR). To assess this, wild-type and regulator mutant strains were grown to postexponential phase (5 h), which is a known window of peak protease expression (9), and assessed for the expression of each protease operon.
We began with the best-studied regulator, SarA, whose ability to repress the transcription of aur, scpA, and ssp but not spl has been well established (9,10). Here, our analysis provided the expected results: in the absence of SarA, there was a 275-fold increase in aur, 10.9-fold increase in sspA, and a 23.7-fold increase in scpA transcript levels, with no changes in spl expression ( Fig. 2A).
Next, we investigated CodY, whose ability to influence protease expression was identified by microarray analysis in UAMS-1 (22). There, Majerczyk et al. (22) found that in the absence of CodY, sspA had increased transcript levels. Additionally, in the same study, CodY was shown to bind the spl, sspA, and aur promoters; however, the binding to aur and spl was deemed biologically irrelevant, as changes in their expression were not observed upon codY deletion. As such, the ability of CodY to modulate expression of aur, scpA, and spl has not been previously described. Herein, in the absence of CodY, we observed a significant 324-fold increase in aur, 12.8-fold increase in sspA, 3.3-fold increase in scpA, and 6.2-fold increase in spl transcript levels (Fig. 2B). Collectively, these data suggest that CodY is a negative regulator of secreted protease expression that rivals SarA in its potency.
We next considered Rot, which was first shown to negatively regulate sspA and spl transcription in a RN6390 microarray (35). In another study assessing aur and sspA regulation in strain 8325-4, Rot functioned as a direct repressor of both loci (25). In support of these studies, others have demonstrated that Rot represses aur and sspA while also directly repressing spl through promoter binding in strain LAC (23). Additionally, in the same study, Rot was shown for the first time to directly repress scpA transcription. In our study, upon rot inactivation, there were significant increases of 6.2-fold for aur and 4.5-fold for sspA transcript levels, which is in line with previous research (23). Additionally, a significant 2.1-fold decrease in scpA expression along with no change for spl was observed, contradicting previous studies, where increased transcription for both was observed upon rot deletion (Fig. 2C). We note, however, that previous studies regarding Rot regulation differ from ours through the use of medium supplemented with different nutrients. Specifically, in work by Mootz et al. (23), growth medium was supplemented with glucose, which has been documented as repressing the agr quorum sensing system via the decreased pH produced from carbon metabolism (23,36,37). As such, this decrease in agr activity could alter the expression of downstream factors also capable of regulating the secreted proteases. Similarly, Said-Salim et al. (35) used Casamino Acids-yeast extract-glycerol phosphate broth for their FIG 2 Individual protease loci are differentially controlled by major regulators of S. aureus. qRT-PCR was used to determine transcript levels for aur, scp, ssp, and spl in regulator mutants after 5 h of growth. The strains used were wild type (WT) USA300 Houston (HOU) and mutants for sarA (A), codY (B), rot (C), sarS (D), saeR (E), mgrA (F), and sarR (G). RNA was isolated from three independent cultures. The 16S rRNA gene was used as an internal control. Fold change from WT was determined using the 2 ϪΔΔCT method. Student's t tests were used to determine statistical significance. *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001; ****, P Ͻ 0.0001 relative to the wild-type strain. Error bars show the standard deviations (SDs). studies. Here, the addition of glycerol, as well as the use of an entirely different complex medium, altered the activity of other transcriptional regulators such as CodY, CcpE, CcpA, and RpiRC, which are known to sense the carbon status of the cell (38). Therefore, while Rot has the potential to regulate all four protease loci, our data suggest that Rot primarily controls expression of aur and sspAB, likely in an agr-dependent manner.
SarS was formerly shown to have no significant effect on aur and ssp transcription during investigation in strain 8325-4 (27). Oscarsson et al., however, established that when sarS is overexpressed in 8325-4, aur and sspA transcription is suppressed (25). In support of a role in sspA regulation, another study showed that SarS could bind the sspA promoter (27). To date, the effects of SarS on scpA and spl transcription have not yet been investigated. Our analysis of protease transcription in the absence of SarS revealed significant increases for aur (6.9-fold), sspA (2.9-fold), and spl (1.6-fold) but a 1.7-fold decrease in scpA transcript levels (Fig. 2D). These data thus support a role for SarS as a repressor of aur and sspA expression and identify the spl operon as a new target of negative regulation by this factor. Conversely, we reveal scpA as a being activated by SarS, demonstrating, as with our data for Rot, that each of the four proteases are often subject to differential and opposing regulation by the same element.
The ability of SaeR to influence protease expression was previously described by microarray analysis, where, in the absence of SaeR/S in strain LAC, there was a decrease in spl transcription (24). Furthermore, in that same study, it was observed that this effect was direct, as SaeR was shown to bind to the spl promoter. Additionally, in the same background, Cassat et al. showed a decrease in SplA-F protein levels following sae inactivation (39). In support of this, we observed a striking 671-fold decrease in spl transcript levels upon saeR deletion, which is the most pronounced alteration in expression for any protease observed in this study (Fig. 2E). With regard to aur, the previously referenced studies revealed an increase in aur transcription (24) as well as an increase in Aur protein levels (39) in the absence of saeRS. In our study, however, no change in transcription was observed, which is in line with Oscarsson et al., who derived similar findings in strain RN6390 (25). Of note, the changes observed during microarray and proteomic analyses were during stationary phase rather than postexponential phase. Therefore, the disagreement regarding aur regulation could be a product of different time points used for assessment. This is supported by our observation that, when analyzed throughout growth, SaeRS is the only major regulator in S. aureus to demonstrate a rebound in transcriptional activity during stationary phase (our unpublished observation). This suggests that SaeRS may have various or biphasic functions with regard to virulence factor regulation during S. aureus growth. Regarding scpA, the effect of SaeR on transcription has not until now been investigated. Herein, we observed a 2.5-fold decrease in scpA transcription in the absence of SaeR, indicating that, similarly to the spls, it is activated by this factor. Lastly, no change in sspA transcription was observed, which, while in line with Oscarsson et al. (25), contradicts Cassat et al. (39), who observed an increase in SspA and SspB protein levels during stationary phase. As previously suggested, this conflict is likely explained by the varying impact of SaeRS during different growth phases. As such, our data support a role for SaeR during postexponential growth in the activation of spl and identify scpA as a new target for SaeR upregulation.
We next investigated MgrA, which was previously shown to activate aur and sspA transcription in 8325-4 (25). Using RNA sequencing in LAC, others have shown that the absence of MgrA decreased aur and spl transcript levels (34). Herein, in agreement with previous studies, we observed a significant 7.6-fold decrease in aur, 3.2-fold decrease in sspA, and 26.7-fold decrease in spl transcript levels (Fig. 2F). Lastly, until now, the effect of MgrA on scpA had not been investigated. In our study, no changes in scpA transcript levels were identified, which again demonstrates differential regulation of the various protease loci. This is particularly interesting, as it is an additional example of the two staphopain enzymes (SspB and ScpA), which share strong homology (40-42) although quite different substrate specificities (42), as being regulated in opposing fashions.
Finally, we investigated SarR, which was formerly shown to positively affect aur and sspA transcript levels in 8325-4 (21). In contrast, in another study, it was shown to negatively affect aur when overexpressed in an 8325-4 agr sarA double mutant (25). Interestingly, however, in our study, no change in aur transcript levels was detected in the absence of sarR. When considering ssp expression, we observed a significant 1.6-fold decrease in transcript levels ( Fig. 2G) in the sarR mutant, which is in agreement with Gustafsson et al. (21). With regard to scpA and spl, SarR was not previously investigated as controlling their transcription. Herein, we observed a significant 29.1fold decrease in scpA and 48.8-fold decrease in spl transcript levels. Our data therefore support a role for SarR in upregulating the ssp operon to a minor extent while serving as one of the strongest activators of scpA and spl expression identified thus far.
Defining the pathway of control for secreted protease expression by known major regulators. Collectively, our findings confirm 14 regulatory pathways for secreted protease transcription while identifying eight new nodes of expression (Fig. 3). For aur, we found it was regulated by CodY in addition to SarA, Rot, SarS, and MgrA. Interestingly, with the exception of MgrA, each of these factors engenders repression of aur expression, with some (SarA and CodY) exerting profound influence. This is perhaps explained by the observation that aureolysin sits atop the protease activation cascade, which flows from Aur to V8 and then staphopain B (11,(43)(44)(45). As such, repressing aureolysin would allow the S. aureus cell to keep the majority of proteases' activity restrained by the single act of limiting expression from P aur . This would be to the cells advantage as, although proteases are undoubtedly valuable enzymes with important roles, they are also destructive in nature. Thus, limiting their activity until it is absolutely required is a major goal of living cells from all kingdoms (46,47). This would be particularly true of aureolysin, given that it has among the broadest substrate specificities of any S. aureus protease (48). In the context of enzymes from the ssp operon, we did not identify new regulatory nodes but confirmed their broad regulation, albeit at modest levels, in a fashion that closely resembles that of aur control. This finding is again logical, given that the enzymes produced from these loci are part of the protease activation cascade referenced above.
Interestingly, much of the new knowledge generated herein involves the regulation of the more underappreciated proteases, staphopain A and the Spls. While the importance of scpA in virulence has been shown through in vivo studies, as well as by its ability to cleave specific host proteins (13,49,50), its transcriptional regulation has been underexplored. While it has been shown previously that scpA is regulated by Rot and SarA, we identified herein that SarS, CodY, SaeR, and SarR also control its expression. While much of this regulation is at modest levels, scpA expression is profoundly influenced in opposing fashions by SarA (repressor) and SarR (activator). This presents a scenario whereby the presence of this enzyme during infection could be discretely titrated, with high SarA activity resulting in decreased staphopain A, while elevated SarR levels would engender significant production of this enzyme. This could then provide rapid niche-specific control of the pathogenic process through staphopain A activity (or lack thereof) toward self-and host-derived proteins. The need for such a network of opposing and stringent control is supported by the observation that staphopain A is one of only two S. aureus secreted proteases with a broad and promiscuous substrate specificity (aureolysin being the other) (51); thus, tightly modulating its influence is a necessity for a coordinated and controlled infectious process.
When exploring control of spl expression, we note that this operon is subject to some of the strongest regulation observed for any protease loci in this study. Specifically, MgrA, SarR, and SaeR each bring about profound upregulation of the spl operon, to levels that rival and, in the case of SaeR, exceed, that of SarA and CodY for protease control. This is of interest because the Spls are well known for their narrow substrate specificity (52)(53)(54). Indeed, these enzymes share strong homology and many enzymatic characteristics with the exfoliative toxins of S. aureus. In the case of these latter proteases, they have only a single known target, desmoglein-1 in the skin of humans, the cleavage of which results in scalded skin syndrome (55). The Spl enzymes are projected to have a similarly narrow range of substrates (56); thus, it is logical that the cell would limit the production of these enzymes until it finds itself in an environment where their activity would prove beneficial. As such, it is logical that the presence and activity of the Spl enzymes can be selectively and rapidly stimulated by these regulatory factors in response to environmental cues within the host to facilitate infection.
Identification of a cadre of new effectors of protease activity. Given the complex regulatory function of S. aureus secreted proteases, tight modulation of their expression is required. As such, we set out to more deeply characterize their network of control by uncovering novel effectors of their activity. This was achieved by screening all 108 available transcriptional regulator mutants within the Nebraska Transposon Mutant Library (NTML) (57) for alterations in proteolytic capacity. Culture supernatants from all strains grown for 15 h (a window of peak accumulation for secreted proteases) were prepared and subjected to zymography using gelatin as a substrate, as described by us previously (9). Of the 108 mutants screened, five of the seven primary regulators (sarS, saeR, rot, sarA, and codY mutants) were included as controls (sarR and mgrA mutants are not present in the NTML), along with two other major regulators of protease production: agrA and sigB. As expected, an increase in proteolytic activity was observed with sarS, rot, sarA, codY, and sigB mutants, while a decrease was observed for saeR and agrA mutants, in comparison to that in the wild type (Fig. 4). For all strains, the intensities of proteolytic banding resulting from gelatin degradation were assessed visually and by densitometry using ImageJ software (Fig. 5).
Excluding the known major regulators, a total of 45 mutants were identified as having notable alterations in proteolytic activity from our screen, with 26 found to have decreased proteolysis (see Table S1 in the supplemental material), while 19 had an increase (Table S2). When assessing mutants that showed increased proteolysis, we identified SarX and NsaR, which were both previously identified as regulating proteases. SarX has been shown to repress sspA transcription in strain RN6390 (31), while NsaR was shown to be a repressor of scpA, sspA, and splA-F in strain SH1000 (58). When considering mutants that had decreased proteolysis, we noted SarV and CcpE, both of which have been implicated in modulating protease activity. Specifically, sarV disruption in RN6390 led to a decrease in transcription for aur, scpA, and splA (32), while loss of ccpE in strain Newman results in impaired expression of all protease loci (59).
Beyond these known factors, we identified a number of intriguing regulators which have yet to be implicated in protease regulation. Of these, several displayed a prominent decrease in protease activity, including SarU. This regulator is an understudied transcription factor belonging to the Sar family, with many of its counterparts already known to have a role in regulating protease production (60). In addition, a notable decrease in protease activity was also observed for mutants of rbf and atlR, which encode regulators known to control biofilm formation (61)(62)(63). Further, Rex and MntR, both of which regulate different aspects of cellular metabolism, also caused pronounced decreases in protease activity upon ablation. We also observed a decrease in protease activity upon disruption of argR2, which is located within the arginine catabolism metabolic element (ACME) found in USA300 strains (64). Finally, XdrA/xdrA, which has a role in immune evasion via its involvement in the production of protein A (65), was found to produce a notable increase in protease activity upon disruption.
Exploring protease control via a secondary network of regulation. To more deeply explore the new protease regulatory factors identified herein, the seven referenced above were chosen for more detailed study. First, each mutation was transduced into a clean USA300 HOU background, and protease activity was continuously monitored throughout growth (see Fig. S1). In agreement with results from our zymography screen, a decrease in protease activity was observed at all time points for mutants of argR2, mntR, atlR, rbf, sarU, and rex, while the xdrA mutant demonstrated a minor decrease in protease activity at early times points, but produced the expected increase in proteolysis thereafter. To ensure that the changes observed were not the result of a simple growth defect, growth curves were performed for all strains, revealing no notable alterations compared to the growth of the wild type (see Fig. S2).
Our next step was to determine if the changes observed in the novel regulatory mutants were driven by changes at the level of transcription. Thus, qRT-PCR analysis for each protease loci was performed for the wild-type and regulator mutant strains during postexponential phase, with the exception of the argR2 mutant, which appears to most notably alter proteolysis at 3 h of growth; thus, this time point was used for this strain. When studying changes in the argR2 mutant, a 1.6-fold decrease in aur, 1.8-fold increase in sspA, and 1.7-fold increase in spl transcripts were observed (Fig. 6A), along with no change in scpA transcription. Next, with the mutant of atlR, we observed a significant 2-fold decrease in aur and a 2.2-fold decrease in spl transcripts (Fig. 6B), whereas with scpA and sspA, no changes in transcript levels were noted. For the mntR mutant, we observed a significant 2.5-fold decrease in sspA and 1.7-fold decrease in spl transcript levels (Fig. 6C), with no changes detected for aur and scpA. In the context of rex, a significant 3.3-fold decrease was seen with sspA transcript levels, while there were no changes in transcription for the other protease loci in this mutant (Fig. 6D). Following this, we investigated the xdrA mutant, in which we observed a significant 1.9-fold FIG 4 Impact of primary regulator mutation on protease activity. Gelatin zymography was performed to visualize protease activity on 15-h culture supernatants obtained from USA300 JE2 and mutants of sarS, saeR, codY, sigB, agrA, rot, and sarA. All strains were adjusted to equal optical densities prior to analysis. increase for aur and 4.2-fold increase in scpA transcript levels (Fig. 6E); however, with spl, we observed a significant 2.4-fold decrease in expression. When studying the rbf mutant, there was a significant 1.7-fold decrease for aur, 2-fold decrease for sspA, and 1.8-fold decrease for spl transcript levels (Fig. 6F), along with no changes for scpA transcription. Lastly, for the sarU mutant, we observed a significant 2.3-fold decrease for sspA and 1.7-fold decrease for aur transcript levels (Fig. 6G), while no changes were noted for scpA and spl transcription. Collectively, almost all of the regulators solely activate protease transcription, with the exception of XdrA, which differentially regulates protease loci in opposing fashions, akin to that observed with Rot and SarS.
Determining the pathway of control for the novel protease regulators. In the work described above, we identified 14 new regulatory pathways for secreted protease transcription. These data allow us to construct a map of protease regulation for these factors, detailing specific effects on individual protease loci (Fig. 7). To delineate the pathway by which these regulators exert their effects, we next assessed their impact on the primary regulators of protease expression considered previously. As such, qRT-PCR analysis was performed on the seven novel protease regulator mutants for sarA, codY, rot, sarS, saeR, mgrA, and sarR at the respective time points in which their protease transcripts were previously assessed. SarA, SarR, MgrA, and CodY are able to regulate protease production by direct action, but can also act via control of the Agr quorum sensing system (26,(66)(67)(68)(69)(70)(71)(72). Agr in turn activates secreted protease production during FIG 5 Quantitative profiling of protease activity for all available regulator mutants of S. aureus. Zymogram band intensities from all 108 regulator mutants contained within the NTML were measured using densitometry (ImageJ). Depicted is fold change of band intensity relative to that of the USA300 JE2 wild-type strain. postexponential phase by inhibiting translation of the negative regulator Rot (73)(74)(75). As such, for completeness, we also included analysis of the agr operon in these studies.
When data for the argR2 mutant were analyzed, we found no significant changes in expression for any of the primary protease regulators (Fig. 8A). As such, the changes in ssp transcript levels in the argR2 mutant are either the result of direct action by ArgR2 qRT-qPCR was performed to determine transcript levels for aur, ssp, scp, and spl in the regulator mutants. The strains used were WT USA300 HOU and mutants of argR2 (A), atlR (B), mntR (C), rex (D), xdrA (E), rbf (F), and sarU (G). RNA was isolated from three independent cultures. The 16S rRNA gene was used as an internal control. Fold change from WT was determined using the 2 ϪΔΔCT method. Student's t tests were used to determine statistical significance. *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001 relative to the wild-type strain. Error bars are SDs. or are mediated by an as yet unknown circuit. When assessing the atlR mutant, a significant 1.4-fold decrease in saeR and a 1.5-fold increase in sarS transcripts were observed (Fig. 8B). The decrease in saeR could explain the observed decrease in spl expression, as SaeR was shown by ourselves and others to activate spl transcription (24,39). In addition, the increase in sarS expression could explain the decrease in both aur and spl transcripts, as SarS was shown in this study to repress transcription of spl and was shown here and elsewhere to repress aur expression (25).
Next, with the mntR mutant, we observed a significant 1.4-fold decrease in mgrA, 1.5-fold decrease in codY, 1.5-fold decrease in saeR, and 1.8-fold decrease in sarR transcript levels (Fig. 8C). With regard to the decrease in ssp and spl transcripts, these changes cannot be explained by the decrease in transcription for codY, as we show that CodY represses both of these loci. The decrease in the saeR transcript, however, could result in a decrease in spl transcription, as it has been shown by ourselves and others to be an activator of this operon (24,39). Furthermore, the decrease in mgrA and sarR transcripts could lead to a decrease in ssp and spl expression, as we confirm the work FIG 8 Determining the pathway of control for the novel protease regulators. qRT-PCR was performed to determine transcript levels for agrB, sarA, mgrA, rot, codY, saeR, sarR, and sarS in the regulator mutants. The strains used were WT USA300 HOU and mutants of argR2 (A), atlR (B), mntR (C), rex (D), xdrA (E), rbf (F), and sarU (G). RNA was isolated from three independent cultures. The 16S rRNA gene was used as an internal control. Fold change from the WT was determined using the 2 ϪΔΔCT method. Student's t tests were used to determine statistical significance. *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001 relative to the wild-type strain. Error bars are SDs. of others demonstrating that MgrA activates expression for both proteases (25,34) while newly identifying SarR as acting in a similar fashion.
When exploring the influence of Rex, we observed a significant 1.4-fold decrease in agrB, 1.3-fold decrease in sarA, 1.3-fold decrease in mgrA, 1.5-fold decrease in saeR, 2-fold decrease in sarR, and 1.5-fold decrease in sarS transcript levels (Fig. 8D). The changes in sarA, saeR, and sarS cannot explain the decrease we observed for the ssp transcript, because as shown by ourselves and others, both are repressors of ssp (9,10,25). However, as we and others have shown that MgrA, SarR, and Agr are activators of ssp transcription (9,21,25), decreases in their expression could explain our data. When assessing the xdrA mutant, a significant 2.1-fold decrease in agrB and 1.5-fold decrease in codY transcript levels were observed (Fig. 8E). Additionally, a significant 1.8-fold increase in mgrA, 4-fold increase in saeR, 1.7-fold increase in sarR, and 3-fold increase in sarS transcripts were observed. The increase in mgrA transcript could explain the increase in aur expression as MgrA has been shown here and by others to activate its transcription (25,34). Next, as we showed SaeR, SarR, and SarS are activators of scp expression, increases in the transcription of each could result in enhanced scp transcript abundance. Additionally, the decrease in codY expression could explain the increase in transcript for aur and scp, as we showed CodY is a repressor of both. Lastly, the decrease in spl transcript levels in the xdrA mutant could be explained by either the increase in sarS or by the decrease in agrB expression, as we show that SarS is a repressor of this locus, while it is well known that Agr is an activator of spl transcription (10). Next, with the rbf mutant, we observed a significant 1.3-fold increase in rot transcription as well as a 2.1-fold increase for sarS (Fig. 8F). The decrease in aur and ssp transcript levels observed in the rbf mutant could be explained by the increase in sarS expression, as it was shown by ourselves and others to be a repressor for both loci (25). Furthermore, we show SarS is a repressor of spl, and as such, the increase in sarS could have resulted in the decrease in the spl transcript. In addition, Rot was shown herein, and by others, to be a repressor for aur and ssp; therefore, the increase in rot transcription could result in the decrease of aur and ssp expression (23). Lastly, with the sarU mutant, we observed a significant 1.6-fold increase in rot transcription (Fig. 8G). In the sarU mutant, the decrease in aur and ssp transcription could be explained by the increase in rot transcription, as it has been shown by ourselves and others to be a repressor of both (23).
Integrating the novel secondary protease regulators into the global picture of protease control. Using the findings from this study, along with existing knowledge, we put forward a comprehensive map of secreted protease regulation (Fig. 9). With this knowledge, we are able to identify specific regulatory pathways connecting our novel protease effectors with the major protease regulators. Specifically, with regard to Rbf, it is possible that its repressive effect on sarS transcription is through Rot, as it was previously shown to activate sarS transcription (35,71) and rot transcription is increased in the rbf mutant. Next, with MntR, its positive effect on sarR transcription is likely occurring through MgrA, as it was previously shown that MgrA activates sarR transcription (34) and mgrA expression is decreased in the mntR mutant. As for Rex, its activation of sarR transcription could be occurring through MgrA, as it has been shown that MgrA activates sarR (35,71) and mgrA transcription is decreased in the absence of rex. Lastly, with XrdA, it is possible that its represses saeR via CodY, as it has been shown that CodY represses saeR transcription (76,77) and codY transcription is decreased in the xdrA mutant. Additionally, the negative effect of XdrA on sarR and sarS transcription could be occurring via MgrA, as it was previously shown that MgrA activates sarR and sarS transcription (34,71) and mgrA transcription is increased in the xdrA mutant. Finally, the activation of agr by XdrA could by occurring via the MgrA-SarR pathway, as SarR has been shown to repress agr transcription (68) and, as already noted, sarR transcription is increased in the xdrA mutant.
Concluding remarks. In this study, we set out to completely characterize the locus-specific effects of regulatory factors on secreted protease expression. In so doing, we have identified an abundance of novel regulatory nodes controlling their production and present a comprehensive regulatory circuit that emphasizes the complexity of protease regulation (Fig. 9). When one compares this regulatory overview with the literature on virulence factor control in S. aureus, it becomes clear that the expansive and complex regulatory circuits that exist to oversee secreted protease expression rivals that of alpha-toxin and protein A, which are arguably some of the most important virulence-affecting entities produced by this organism (35,65,(78)(79)(80)(81)(82)(83)(84)(85). Indeed, we suggest that the existence of such a broad network of control speaks to the importance of the secreted proteases to S. aureus physiology and pathogenic potential. We also contend that there is a clear and obvious need for such a network, so as to limit or enhance the abundance (and thus activity) of these enzymes. The rationale for this is that a primary function of these enzymes is to control the progression of infection by selectively modulating the stability of individual virulence factors produced by the cell (19). Thus, in this context, it makes sense that a network of control exists to selectively titrate in or out a given protease (and thus its activity), so as to specifically influence the abundance (or lack thereof) of an individual virulence factor(s). This would then facilitate the selective and niche-specific pathogenic behaviors of S. aureus and provide a basis for control of the broad and varied infection types that is the hallmark of this organism's disease-causing nature. In addition to this, there is abundant evidence in the literature implicating the secreted proteases as facilitating the infectious process by attacking the host and cleaving host proteins. It is thus in line with the above hypothesis that tightly controlling protease activity, by selectively limiting or enhancing their activity in specific niches, is to the advantage of S. aureus and its highly effective and efficient infectious process.

MATERIALS AND METHODS
Media and growth conditions. All cultures were grown overnight at 37°C with shaking at 250 rpm in 5 ml of either tryptic soy broth (TSB) or lysogeny broth (LB). When required, antibiotics were added at the following concentrations: for Escherichia coli, 100 g ml Ϫ1 ampicillin, 12.5 g ml Ϫ1 tetracycline; for S. aureus, 5 g ml Ϫ1 tetracycline, 5 g ml Ϫ1 erythromycin, 25 g ml Ϫ1 lincomycin, and 2.5 g ml Ϫ1 chloramphenicol. To obtain synchronous cultures, overnight S. aureus cultures were diluted 1:100 into 5 ml of fresh medium and grown for 3 h before being standardized to an optical density at 600 nm (OD 600 ) of 0.05 in 100 ml of fresh TSB. When assessing growth, OD 600 was measured hourly using a Synergy 2 plate reader (Bio-Tek).
Bacterial strains. All bacterial strains and plasmids used in this study are listed in Table 1. Transposon mutants for all available transcriptional regulators in S. aureus USA300 JE2 were obtained from the Nebraska Transposon Mutant Library (NTML). Those subjected to further study were transduced into USA300 Houston, as described by us previously (86), using 11. The construction of an mgrA mutant in S. aureus Becker was previously described (87). This mutation was transduced into USA300 Houston using 85.
Construction of a sarR mutant strain. A tetracycline-marked disruption of sarR was generated using pJB38, as described by Bose et al. (88). Regions up-and downstream of sarR, including portions of the 5= and 3= ends of the coding gene, were amplified via PCR using primers OL4208/OL4209 and OL4210/ OL4211. A tetracycline resistance cassette was amplified using OL4299/OL4300 from a SH1000 sigS::tet mutant (89). Using MluI sites, the tetracycline cassette was ligated between the upstream and downstream fragments of sarR and ligated directly into pJB38 using EcoRI and KpnI sites. Using the established protocol, the majority of sarR was deleted in USA300 Houston using allelic replacement (88). Strains were confirmed by PCR and sequencing (Eurofins Genomics) using primers OL4577/OL4578, which amplify across the deleted region where the tetracycline cassette was inserted.
Quantitative real-time PCR analysis. To quantify expression changes for target genes (primers are listed in Table S3 in the supplemental material), quantitative real-time PCR (qRT-PCR) was performed, as described by us previously (90). All targets were normalized using 16S rRNA expression, and fold change from the wild-type was determined using the threshold cycle (2 ϪΔΔCT ) method (91). All graphical representations of fold changes are relative to the wild-type, Ϯ1.
Zymography. Strains grown for 15 h overnight were adjusted to equal optical densities and pelleted. When assessing proteolytic activity over time, synchronized cultures were grown to exponential phase and standardized to an OD 600 of 0.05 in 100 ml of TSB. At the desired time points, cells were pelleted. Thereafter, for all samples, 2 ml of supernatant was processed through an Amicon Ultra 3K centrifugal filter for 60 min at 4,000 ϫ g. Concentrated supernatants were recovered by removing filtrate collection tubes, inverting filter devices, and spinning again for 2 min at 1,000 ϫ g. Equal volumes of Laemmli loading buffer were added to the concentrated supernatants and incubated for 30 min at 37°C. Next, 20 l of each sample was loaded onto preprepared SDS-PAGE gels containing 0.1% gelatin and run until the dye front reached the edge of the plates. Gels were washed twice using 2.5% Triton X-100 at room temperature. Following a rinse with distilled water (dH 2 O), developing buffer (0.2 M Tris, 5 mM CaCl 2 , 1 mM dithiothreitol [DTT], pH 7.6) was added and gels were incubated overnight at 37°C static. After incubation, gels were rinsed with dH 2 O and covered with 0.1% amido black for 1 h. Once gels were stained, destain 1 (30% methanol, 10% acetic acid) was added for 5 to 10 min, replaced with destain 2 (10% acetic acid) until bands became clear, and then replaced with destain 3 (1% acetic acid) for storage. Changes in band intensity were quantified using ImageJ software.

ACKNOWLEDGMENT
This study was supported by grant AI124458 (to L.N.S.) from the National Institute of Allergy and Infectious Diseases.