Diflunisal Attenuates Virulence Factor Gene Regulation and Phenotypes in Staphylococcus aureus

Virulence factor expression is integral to pathogenicity of Staphylococcus aureus. We previously demonstrated that aspirin, through its major metabolite, salicylic acid (SAL), modulates S. aureus virulence phenotypes in vitro and in vivo. We compared salicylate metabolites and a structural analogue for their ability to modulate S. aureus virulence factor expression and phenotypes: (i) acetylsalicylic acid (ASA, aspirin); (ii) ASA metabolites, salicylic acid (SAL), gentisic acid (GTA) and salicyluric acid (SUA); or (iii) diflunisal (DIF), a SAL structural analogue. None of these compounds altered the growth rate of any strain tested. ASA and its metabolites SAL, GTA and SUA moderately impaired hemolysis and proteolysis phenotypes in multiple S. aureus strain backgrounds and their respective deletion mutants. Only DIF significantly inhibited these virulence phenotypes in all strains. The kinetic profiles of ASA, SAL or DIF on expression of hla (alpha hemolysin), sspA (V8 protease) and their regulators (sigB, sarA, agr (RNAIII)) were assessed in two prototypic strain backgrounds: SH1000 (methicillin-sensitive S. aureus; MSSA) and LAC-USA300 (methicillin-resistant S. aureus; MRSA). DIF induced sigB expression which is coincident with the significant inhibition of RNAIII expression in both strains and precedes significant reductions in hla and sspA expression. The inhibited expression of these genes within 2 h resulted in the durable suppression of hemolysis and proteolysis phenotypes. These results indicate that DIF modulates the expression of key virulence factors in S. aureus via a coordinated impact on their relevant regulons and target effector genes. This strategy may hold opportunities to develop novel antivirulence strategies to address the ongoing challenge of antibiotic-resistant S. aureus.


Introduction
Staphylococcus aureus is an important human pathogen responsible for a broad range of infections causing significant morbidity and mortality worldwide [1][2][3][4][5]. The ability of S. aureus to cause myriad disease manifestations is mediated by the coordinated expression of an extensive repertoire of virulence factors, including exotoxins and proteolytic enzymes. Moreover, the rapid emergence of strains exhibiting multidrug resistance phenotypes (e.g., methicillin-resistant S. aureus (MRSA); vancomycin-intermediate S. aureus (VISA)) has accelerated the search for novel strategies to prevent or mitigate S. aureus infections. Therefore, the identification of molecules that interfere with virulence factor regulation and/or expression represents a potentially viable therapeutic strategy.
In the current study, ASA and its major metabolites, SAL, gentisic acid (GTA) and salicyluric acid (SUA), as well as structural analogue, diflunisal (DIF) (Figure 1) were each assessed for their ability to inhibit virulence regulation and associated phenotypes in well-characterized S. aureus strains. Each of these metabolites is physiologically relevant, as nearly all ASAs are rapidly converted to SAL, GTA and SUA in vivo [25,26]. These compounds have been shown to exert beneficial anti-infective and anti-inflammatory properties in humans and experimental models of disease [27,28]. Similar to ASA, DIF is a non-steroidal anti-inflammatory drug (NSAID) that is frequently prescribed in clinical settings for the treatment of cardiac amyloidosis and arthritis [29,30]. In addition to its antivirulence effects on S. aureus in vitro, as well as in cutaneous and endovascular models of infection [10,11,[18][19][20][21][22]24], DIF has been shown to reduce bone destruction during experimental S. aureus osteomyelitis [16]. Specifically, the impact of the above compounds on hemolysin and protease phenotypes, as well as on the kinetics of their respective regulatory (sigB, agr, sarA) and effector genes (hla (α-hemolysin), sspA (V8 protease)) were compared using a panel of strategic S. aureus strains (Table 1).
In the current study, ASA and its major metabolites, SAL, gentisic acid (GTA salicyluric acid (SUA), as well as structural analogue, diflunisal (DIF) (Figure 1) were assessed for their ability to inhibit virulence regulation and associated phenotypes in characterized S. aureus strains. Each of these metabolites is physiologically releva nearly all ASAs are rapidly converted to SAL, GTA and SUA in vivo [25,26]. These pounds have been shown to exert beneficial anti-infective and anti-inflammatory pr ties in humans and experimental models of disease [27,28]. Similar to ASA, DIF is a steroidal anti-inflammatory drug (NSAID) that is frequently prescribed in clinical se for the treatment of cardiac amyloidosis and arthritis [29,30]. In addition to its ant lence effects on S. aureus in vitro, as well as in cutaneous and endovascular mod infection [10,11,[18][19][20][21][22]24], DIF has been shown to reduce bone destruction during e mental S. aureus osteomyelitis [16]. Specifically, the impact of the above compoun hemolysin and protease phenotypes, as well as on the kinetics of their respective re tory (sigB, agr, sarA) and effector genes (hla (α-hemolysin), sspA (V8 protease)) were pared using a panel of strategic S. aureus strains (Table 1).   Figure 1. Chemical structure of study compounds. Study compounds included the parent compound ASA (aspirin; acetylsalicylic acid) (A), SAL (salicylic acid) (B), GTA (gentisic acid) (C) and SUA (salicyluric acid) (D) and the salicylate analogue DIF (diflunisal) (E).

Study Compounds Differentially Modulated Hemolysin Activity in S. aureus
The impact of the study compounds on hemolytic phenotypes is summarized ures 3 and 4. Overall, the study compounds exerted differential effects on hemo the S. aureus study strain set. DIF exposures resulted in marked reductions in the p age of hemolysis relative to control for all study strains as compared to ASA or an metabolites. This outcome was true regardless of genetic background (e.g., USA100 USA300 (LAC), USA400 (MW2)) or classical laboratory (RN6390) or reference (ATCC29213). As SAL is the primary biometabolite of ASA, compounds were com to SAL in reducing hemolysis. Log OD 600 Figure 2. Study compounds do not affect growth of S. aureus strains. Growth curve analyses were performed to assess the direct anti-staphylococcal activity of test compounds. Log-phase organisms were inoculated into fresh BHI broth (OD 600 = 0.05; inoculum 5 × 10 7 CFU/mL) containing a given study compound (range: 0, 10, 25, 50, 100 µg/mL) and incubated at 37 • C with shaking. Bacterial growth was analyzed by spectrophotometry (OD 600 ) at 1-8 and 24 h timepoints. ( Control; • ASA 25 µg/mL; SAL 25 µg/mL; GTA 25 µg/mL; SUA 25 µg/mL; DIF 25 µg/mL).

Study Compounds Differentially Modulated Hemolysin Activity in S. aureus
The impact of the study compounds on hemolytic phenotypes is summarized in Figures 3 and 4. Overall, the study compounds exerted differential effects on hemolysis in the S. aureus study strain set. DIF exposures resulted in marked reductions in the percentage of hemolysis relative to control for all study strains as compared to ASA or any of its metabolites. This outcome was true regardless of genetic background (e.g., USA100 (COL), USA300 (LAC), USA400 (MW2)) or classical laboratory (RN6390) or reference strain (ATCC29213). As SAL is the primary biometabolite of ASA, compounds were compared to SAL in reducing hemolysis.      Figure 4. Relative hemolysis of S. aureus strains exposed to study compounds. Quantitative analyses of zones of clearing were measured and normalized to no compound control. Mean (bold) values and standard deviations (brackets) for each strain and compound combination are presented. Statistics were performed using two-way ANOVA and presented as: * p < 0.05, ** p < 0.01 and *** p < 0.001 for To explore the role of virulence factor regulation relative to compound efficacy, a panel of strategic mutants was evaluated. When comparing mutants against WT parents, only ASA yielded a greater inhibition of hemolysis in the SH1000 agr-deficient mutant (78.82 vs. 98.79; p = 0.05). However, this mutant had a significantly greater hemolysis in the presence of SAL or GTA (98.34 vs. 89.35; p = 0.02 and 100.22 vs. 109.81; p = 0.01, respectively) as compared to ASA (Figure 4). To explore the impact of the sigB regulon on hemolysis modulation by study compounds, rsbU-, rsbVand rsbW-deficient mutants were studied in the FDA486 MSSA background (Figures 3 and 4). Only GTA exposure revealed a significantly greater hemolysis in the FDA486 rsbU-deficient mutant as compared to its parent (116.29 vs. 88.94; p = 0.0017; Figure 4). No significant differences in the inhibitory effects of other study compounds were observed with respect to rsb-deficient mutants as compared to the parent. Likewise, SAL, GTA SUA and DIF exerted a greater modulation of hemolysis than SAL in RN6390 and ISP479C, but only DIF did so in the ISP479 snoD mutant. Overall, DIF exhibited a significantly greater inhibition of hemolysis relative to all other study compounds in the diverse panel of strains tested.

Study Compounds Differentially Inhibited Proteolysis Activity in S. aureus
The impacts of ASA, its metabolites or DIF on proteolytic phenotypes are summarized in Figures 5 and 6. Only DIF exerted significant reductions in the percentage of proteolysis relative to control for all study strains. Moreover, DIF achieved a significantly greater inhibition of proteolysis than ASA in all study strains. Relative to SAL, only DIF exhibited a significantly greater inhibition of proteolysis in most strains studied. No significant differences were observed in terms of DIF inhibition of proteolysis in any parent vs. mutant strain pairs. To explore the role of virulence factor regulation relative to compound efficacy, a panel of strategic mutants was evaluated. When comparing mutants against WT parents, only ASA yielded a greater inhibition of hemolysis in the SH1000 agr-deficient mutant (78.82 vs. 98.79; p = 0.05). However, this mutant had a significantly greater hemolysis in the presence of SAL or GTA (98.34 vs. 89.35; p = 0.02 and 100.22 vs. 109.81; p = 0.01, respectively) as compared to ASA (Figure 4). To explore the impact of the sigB regulon on hemolysis modulation by study compounds, rsbU-, rsbV-and rsbW-deficient mutants were studied in the FDA486 MSSA background (Figures 3 and 4). Only GTA exposure revealed a significantly greater hemolysis in the FDA486 rsbU-deficient mutant as compared to its parent (116.29 vs. 88.94; p = 0.0017; Figure 4). No significant differences in the inhibitory effects of other study compounds were observed with respect to rsb-deficient mutants as compared to the parent. Likewise, SAL, GTA SUA and DIF exerted a greater modulation of hemolysis than SAL in RN6390 and ISP479C, but only DIF did so in the ISP479 snoD mutant. Overall, DIF exhibited a significantly greater inhibition of hemolysis relative to all other study compounds in the diverse panel of strains tested.

Study Compounds Differentially Inhibited Proteolysis Activity in S. aureus
The impacts of ASA, its metabolites or DIF on proteolytic phenotypes are summarized in Figures 5 and 6. Only DIF exerted significant reductions in the percentage of proteolysis relative to control for all study strains. Moreover, DIF achieved a significantly greater inhibition of proteolysis than ASA in all study strains. Relative to SAL, only DIF exhibited a significantly greater inhibition of proteolysis in most strains studied. No significant differences were observed in terms of DIF inhibition of proteolysis in any parent vs. mutant strain pairs. GTA, SUA or DIF, or no compound. Plates were incubated for 24 h for bacterial growth followed by cold shock (4 °C, 4 h) for hemolysin activity. Zones of clearing were measured and normalized to no compound control. Closed circled colony (top left) is SH1000. Dotted circled colony (third row, second position) is LAC. Data are presented in Figure 6 as percent of control. Top row (left to right): SH1000 (1), SH1000 Δagr (2), RN6390 (3); second row (left to right) FDA486(wt) (4), FDA486 ΔrsbU (ALC2128) (5), FDA486 ΔrsbV (ALC2129) (6), FDA486 ΔrsbW (ALC2130) (7); third row (left to right) MW2 (8), LAC (9), ISP479R (10), ISP479C (11); fourth row (left to right) ATCC29213 (12), COL (13), COL Δagr (14).

Figure 6.
Relative proteolysis of S. aureus strains exposed to study compounds. Quantitative analyses of zones of clearing were measured and normalized to no compound control. Mean (bold) values and standard deviations (brackets) for each strain and compound combination are presented. Statistics were performed using two-way ANOVA and presented as: * p < 0.05, ** p < 0.01 and *** p < 0.001 for compound vs. ASA; ^ p < 0.05, ^^ p < 0.01 and ^^^ p < 0.001 for DIF vs. SAL. Significantly decreased values are presented in blue while increased values are presented in red. * See statistical outcomes key in the Materials and Methods section for more details.
In comparison to hemolysis, the components of the rsb regulon had significant differences relative to their impact on DIF efficacy in suppressing proteolysis (Figures 5 and 6). For example, as compared to the parent, proteolysis in rsbU-and rsbW-deletion mutants was significantly less inhibited by DIF (19.90 vs. ≤ 5; p = 0.006 and 32.69 vs. ≤ 5; p = 0.001, respectively). However, relative to the rsbW-deletion mutant, DIF exerted a significantly greater inhibition of proteolysis in the rsbU-and rsbV-deletion mutants (19.90 vs. 32.69; p Figure 6. Relative proteolysis of S. aureus strains exposed to study compounds. Quantitative analyses of zones of clearing were measured and normalized to no compound control. Mean (bold) values and standard deviations (brackets) for each strain and compound combination are presented. Statistics were performed using two-way ANOVA and presented as: * p < 0.05, ** p < 0.01 and *** p < 0.001 for compound vs. ASA;ˆp < 0.05,ˆˆp < 0.01 andˆˆˆp < 0.001 for DIF vs. SAL. Significantly decreased values are presented in blue while increased values are presented in red. * See statistical outcomes key in the Materials and Methods section for more details.
In comparison to hemolysis, the components of the rsb regulon had significant differences relative to their impact on DIF efficacy in suppressing proteolysis (Figures 5 and 6). For example, as compared to the parent, proteolysis in rsbUand rsbW-deletion mutants was significantly less inhibited by DIF (19.90 vs. ≤ 5; p = 0.006 and 32.69 vs. ≤ 5; p = 0.001, respectively). However, relative to the rsbW-deletion mutant, DIF exerted a significantly greater inhibition of proteolysis in the rsbUand rsbV-deletion mutants (19.90 vs. 32.69; p = 0.029 and 9.64 vs. 32.69; p = 0.03, respectively). Together, these data suggest that the rsb regulon contributes to DIF efficacy in modulating proteolysis inhibition.

ASA, SAL and DIF Exhibited Differential Kinetics of Virulence Gene Inhibition
To explore the mechanistic basis of study compounds which inhibited virulence phenotypes, we assessed the transcriptional kinetics of target structural and regulatory genes involved in hemolysis and proteolysis. Transcriptional profiles for two prototype strains (SH1000; and LAC-USA300) are summarized in Figure 7.

ASA, SAL and DIF Exhibited Differential Kinetics of Virulence Gene Inhibition
To explore the mechanistic basis of study compounds which inhibited virulence notypes, we assessed the transcriptional kinetics of target structural and regulatory g involved in hemolysis and proteolysis. Transcriptional profiles for two prototype st (SH1000; and LAC-USA300) are summarized in Figure 7. Figure 7. DIF inhibited virulence factor mRNA expression. Expression of regulatory genes R (agr), sarA, sigB as well as the predominant hemolysin hla (α-hemolysin) and protease sspA (V8 tease) genes were quantified over the course of 6 h, following exposure to ASA, SAL or DI µg/mL) as compared to control. DIF inhibited mRNA expression of RNAIII, hla and sspA as pared to ASA (* = p < 0.05; *** = p < 0.001) or SAL ( + = p < 0.05; +++ = p < 0.001) in SH1000 (MSS LAC (MRSA USA300).
The compounds ASA, SAL and DIF were the focus of transcriptomic studies as t agents were phenotypically the most impactful on hemolysis and proteolysis. These pounds had differential effects on the quantity and kinetics of regulatory gene trans tion in these two S. aureus strains (Figure 7; Supplementary Figure S1). DIF significantly reduced RNAIII transcription within 2 h in both strains. The inhibition of RNAIII transcription occurred at 4 h for SH1000 (15-fold reduction; p = vs. 0.5-h) and at 2 h for LAC (40-fold reduction; p < 0.0001 vs. 0.5-h). Consistent with effect, the RNAIII counter-regulatory genes sarA or sigB increased in their expressio SH1000 or LAC, respectively. The sarA peak transcription occurred at 4 h post-expo to ASA in SH1000 (p = 0.06 vs. 0.5-h). The sigB peak transcription occurred by 2 h exposure to DIF in LAC (p = 0.15 vs. 0.5 h). Study compounds did not significantly d in their impact on sarA or sigB expression. Interestingly, sarA transcription did not ap ciably increase in LAC, and sigB transcription did not appreciably increase in SH regardless of time or compound. Notably, regulatory gene transcription essentiall turned to baseline in both strains by 6 h post-exposure regardless of the study compo   Figure 7. DIF inhibited virulence factor mRNA expression. Expression of regulatory genes RNAIII (agr), sarA, sigB as well as the predominant hemolysin hla (α-hemolysin) and protease sspA (V8 protease) genes were quantified over the course of 6 h, following exposure to ASA, SAL or DIF (25 µg/mL) as compared to control. DIF inhibited mRNA expression of RNAIII, hla and sspA as compared to ASA (* = p < 0.05; *** = p < 0.001) or SAL ( + = p < 0.05; +++ = p < 0.001) in SH1000 (MSSA) or LAC (MRSA USA300).
The compounds ASA, SAL and DIF were the focus of transcriptomic studies as these agents were phenotypically the most impactful on hemolysis and proteolysis. These compounds had differential effects on the quantity and kinetics of regulatory gene transcription in these two S. aureus strains (Figure 7; Supplementary Figure S1). DIF significantly reduced RNAIII transcription within 2 h in both strains. The peak inhibition of RNAIII transcription occurred at 4 h for SH1000 (15-fold reduction; p = 0.001 vs. 0.5-h) and at 2 h for LAC (40-fold reduction; p < 0.0001 vs. 0.5-h). Consistent with this effect, the RNAIII counter-regulatory genes sarA or sigB increased in their expression in SH1000 or LAC, respectively. The sarA peak transcription occurred at 4 h post-exposure to ASA in SH1000 (p = 0.06 vs. 0.5-h). The sigB peak transcription occurred by 2 h post-exposure to DIF in LAC (p = 0.15 vs. 0.5 h). Study compounds did not significantly differ in their impact on sarA or sigB expression. Interestingly, sarA transcription did not appreciably increase in LAC, and sigB transcription did not appreciably increase in SH1000, regardless of time or compound. Notably, regulatory gene transcription essentially returned to baseline in both strains by 6 h post-exposure regardless of the study compound.
Study compounds also had differential effects on the quantity and kinetics of virulence gene transcription in comparative S. aureus strains (Figure 7). Consistent with downregulated RNAIII transcription, in SH1000, DIF caused a significant reduction in hla (~140-fold) transcription at 4 h as compared to baseline; this effect was significantly greater than ASA or SAL (Figure 7). The reduction in hla transcription by DIF was observed at every time point in LAC, but did not reach significance relative to baseline or other compounds. In both strains, DIF inhibition of sspA expression was significant at 4 h post-exposure as compared to ASA or SAL (Figure 7). In SH1000, hla expression remained highly suppressed by DIF at 6 h, but did not achieve significance as compared to baseline or other study compounds. Likewise, DIF inhibition of sspA expression trended to return to baseline by 6 h in SH1000.
In contrast, DIF inhibition of sspA expression in LAC continued to remain significant over the 6 h study period as compared to baseline (p = 0.01 vs. 0.5-h), but was only significant at 4 h as compared to ASA (Figure 7).

Discussion
Recent serendipitous clinical observations revealed that ASA exerts beneficial antiinfective efficacy in multiple human infectious diseases. For example, low-dose ASA and its de-acetylated metabolite SAL significantly decrease the risk of S. aureus bacteremia in patients with hemodialysis tunneled catheters and with infected prosthetic joints [12,13]. Experimental models also show that these compounds reduce the severity and progression of infective endocarditis [10,11,17]. These observations have prompted increasing interest in the potential to repurpose ASA or other nonsteroidal anti-inflammatory drugs (NSAIDs) as novel adjunctive anti-infective therapies.
In the current study, the impact of ASA, its salicylate metabolites (SAL, GTA, SUA) and the fluorinated structural analogue, DIF, were studied for their modulation of virulence phenotypes and transcriptional regulation using a panel of defined S. aureus strains in vitro. ASA, SAL and DIF exerted inhibitory effects on hemolysis and proteolysis capabilities. However, only DIF inhibited both virulence phenotypes in all study strains, and did so significantly greater than ASA and SAL (Figures 4 and 6). Interestingly, in SH1000 lacking agr, a greater inhibition of hemolysis was observed by ASA as compared to the wild-type parent strain. In contrast, agr did not significantly impact DIF efficacy in either SH1000 or COL backgrounds (Figure 4). Similarly, the presence or absence of agr did not affect proteolysis inhibition by DIF or other study compounds in all genetic backgrounds ( Figure 6). These findings suggest agr alone is not a primary mechanism of DIF inhibition of hemolysis or proteolysis activity in S. aureus. The impact of genes within the sigB locus (rsbU, rsbV, rsbW) on study compound efficacy was also investigated. The deletion of rsbV or rsbW had no significant effect on hemolysis or proteolysis as compared to wild-type FDA486. However, the absence of rsbU (considered to be the "sensor" of this operon) [33] promoted hemolysis in the presence of GTA as compared to the parent. By comparison, the deletion of any of the rsb genes consistently reduced the efficacy of DIF proteolysis inhibition; none of these differences achieved statistical significance as compared to control. These findings suggest that sigB and its rsb components are involved directly or indirectly in the inhibitory mechanisms of DIF against proteolysis ( Figure 6).
Next, we explored the transcriptional kinetics of target regulatory and effector genes in response to compounds ASA, SAL or DIF that were shown to inhibit virulence phenotypes. Gene expression was monitored over a 6 h period in prototypic MRSA and MSSA strains to enable the temporal assessment of transcriptional profiles in response to compound exposure. In prototypic MSSA and MRSA strains, distinct kinetic patterns of transcription were identified in response to specific study compounds. In SH1000 (MSSA), an increased expression of sarA by 4 h post-DIF exposure coincided with a significant suppression of RNAIII. By comparison, in LAC (MRSA), an increased expression of sigB by 2 or 4 h post-DIF exposure paralleled with the significant inhibition of RNAIII. These findings are consistent with sigB as a strong counter regulator of RNAIII [38,39]. Regardless of sarA or sigB regulation, suppression of RNAIII was strongly correlated with the inhibition of hla and sspA. These transcriptional profiles were concordant with the significant inhibition of hemolysis and proteolysis by DIF. It is notable that rapid transcriptional inhibition of hla and sspA by 2-4 h resulted in a durable suppression of their respective virulence phenotypes even at 24 h. Given that gene expression had largely returned to baseline by 6 h, early and/or temporary interference in virulence gene expression can have lasting effects that may benefit antivirulence efficacy. Neither ASA nor SAL significantly altered virulence gene expression in these strains over 6 h.
We recognize several potential limitations with our investigation. First, these in vitro outcomes may not fully represent S. aureus virulence regulation within the host. Preliminary data not presented here, support the efficacy of DIF in antibiotic therapy of MRSA infection in vivo [24]. Second, study compounds appeared to have relatively different degrees of activity against different S. aureus strains. However, the fact that DIF strongly inhibited RNAIII, hla and sspA expression, as well as hemolysis and proteolysis phenotypes in every studied strain is promising as a therapeutic strategy regardless of S. aureus genetic background. Third, the observation that hemolysis was not abolished in agr-deleted SH1000 and COL backgrounds suggests agr-independent regulation of hla and sspA. This unexpected finding has also been reported by Liu et al. [23,40] and suggests that novel anti-infective targets of virulence inhibition are yet to be explored.
The currents studies further substantiate our original findings regarding the antivirulence properties of ASA, its metabolites and the structural analogue DIF [20][21][22]24]. A hypothetical model integrating the putative mechanisms of DIF is illustrated in Figure 8. Our original observations have also been supported by work from several other laboratories [16,[41][42][43]. In light of the burgeoning threat of MRSA resistance to multiple antibiotic classes, novel approaches to attenuate virulence without inducing resistance are attractive strategies. For example, the complementary inhibition of virulence factor expression and targets of conventional antibiotics may translate to a greater microbicidal impact, reduced emergence of resistance and enhanced immune-mediated efficacy in MRSA infection. The current findings provide further proof of concept that DIF and the structural analogues of SAL attenuate prototypic virulence factor expression in S. aureus. The translation of these strategies is currently under investigation in our laboratories.
phenotypes even at 24 h. Given that gene expression had largely returned to baseline by 6 h, early and/or temporary interference in virulence gene expression can have lasting effects that may benefit antivirulence efficacy. Neither ASA nor SAL significantly altered virulence gene expression in these strains over 6 h.
We recognize several potential limitations with our investigation. First, these in vitro outcomes may not fully represent S. aureus virulence regulation within the host. Preliminary data not presented here, support the efficacy of DIF in antibiotic therapy of MRSA infection in vivo [24]. Second, study compounds appeared to have relatively different degrees of activity against different S. aureus strains. However, the fact that DIF strongly inhibited RNAIII, hla and sspA expression, as well as hemolysis and proteolysis phenotypes in every studied strain is promising as a therapeutic strategy regardless of S. aureus genetic background. Third, the observation that hemolysis was not abolished in agr-deleted SH1000 and COL backgrounds suggests agr-independent regulation of hla and sspA. This unexpected finding has also been reported by Liu et al. [23,40] and suggests that novel anti-infective targets of virulence inhibition are yet to be explored.
The currents studies further substantiate our original findings regarding the antivirulence properties of ASA, its metabolites and the structural analogue DIF [20][21][22]24]. A hypothetical model integrating the putative mechanisms of DIF is illustrated in Figure 8. Our original observations have also been supported by work from several other laboratories [16,[41][42][43]. In light of the burgeoning threat of MRSA resistance to multiple antibiotic classes, novel approaches to attenuate virulence without inducing resistance are attractive strategies. For example, the complementary inhibition of virulence factor expression and targets of conventional antibiotics may translate to a greater microbicidal impact, reduced emergence of resistance and enhanced immune-mediated efficacy in MRSA infection. The current findings provide further proof of concept that DIF and the structural analogues of SAL attenuate prototypic virulence factor expression in S. aureus. The translation of these strategies is currently under investigation in our laboratories.

Materials and Methods
In this study, the potential for ASA and its relevant metabolites and DIF to modulate virulence factor expression in S. aureus was investigated. Study compounds included the parent ASA and metabolites, SAL, SUA and GTA; the salicylate analogue DIF was tested in parallel (diflunisal) (Figure 1).

Compounds
Unless otherwise noted, the following compounds were obtained from Sigma-Aldrich, Inc. (St. Louis, MO, USA) and prepared as stock solutions from powder: ASA; SAL; SUA (Acros Organics, NJ, USA); GTA; and DIF ( Figure 1). Stock solutions were prepared in ethanol and stored at 4 • C until use.

Bacterial Strains
Staphylococcus aureus strains (MSSA; and MRSA) used in this study are described in Table 1. Strains included both prototypic laboratory and well-characterized clinical isolates with known genotypes and phenotypes. For experiments detailed below, all strains were cultured to mid-logarithmic or stationary phase in brain-heart infusion broth (BHI; Difco Laboratories, Detroit, MI, USA), at 37 • C with agitation, washed, and resuspended in phosphate-buffered saline (PBS; Irvine Scientific, Santa Ana, CA, USA; pH 7.2). Experimental inocula were determined by spectrophotometry and validated by quantitative culture.

Anti-Staphylococcal Activity of Compounds
Growth curve analyses were performed to assess the direct anti-staphylococcal activity of all test compounds. To do so, log-phase organisms were inoculated into fresh BHI broth (OD 600 = 0.05; inoculum 5 × 10 7 CFU/mL) containing a given study compound (range: 0, 10, 25, 50, 100 µg/mL) and incubated at 37 • C with shaking. At selected time-points (1-8 h and 24 h), cultures were analyzed by spectrophotometry (OD 600 ) and quantitative culture in comparison to respective untreated controls ( Figure 1).

Influence of Compounds on Hemolysin or Protease Expression
Two pivotal phenotypes in S. aureus that govern many of its virulence capacities involve secretion of hemolysins and proteases [44]. The effects of our study compounds on expression of secreted hemolysins and proteases were compared in the panel of S. aureus strains summarized in Table 1. In these assays, blood agar-tryptic soy agar plates (TSA; Beckton Dickinson, CA) contained either 5% sheep or 5% rabbit blood (Hardy Diagnostics, CA) and 25 µg/mL of ASA, SAL, GTA, SUA or DIF, or no compound. These concentrations encompassed the known human blood levels for ASA and DIF after standard dose regimens [45]. Likewise, for protease assays, standard-method caseinate agar (SMCA) plates containing 25 µg/mL of each study compound were compared to control plates without these compounds. For these assays, strain inocula were prepared as above. To ensure that maximal hemolysin activity was detected, hemolysin assay plates were coldshocked (4 • C for 4 h) prior to reading zones of hemolysis (α-hemolysin activity facilitated by temperature shock [46]). Zones of hemolytic or proteolytic activity were measured by quantitative imaging (AlphaEaseFC imager and software; Alpha Innotec, Kasendorf, Bayern, Germany). Statistics were performed using two-way ANOVA. Data are presented as: * p < 0.05, ** p < 0.01 and *** p < 0.001 for compound vs. ASA.ˆp < 0.05,ˆˆp < 0.01 andˆˆp < 0.001 for DIF vs. SAL. + p < 0.05 and ++ p < 0.01 for mutant vs. parent.

Influence of Compounds on Gene Expression
To identify the influence of the study compounds on transcriptional correlates of the above two virulence phenotypes, expression of selected virulence regulon or effector genes were compared in prototypic MSSA (SH1000) and MRSA (LAC) strains. Expressions of regulatory genes RNAIII (agr), sarA and sigB, as well as the predominant hemolysin gene, hla (producing α-hemolysin) and the protease gene, sspA (producing V8 protease) were quantified over a 6 h time course, following exposure to ASA, or its analogues (vs. untreated controls). In brief, 10 9 CFU of log phase cells were isolated and exposed for 1 h to ASA, SAL, GTA, SUA or DIF (concentration, 25 µg/mL). Control samples were exposed to buffer alone in the absence of the compound. In parallel experiments, organisms were cultured for 2 h, 4 h or 6 h in a fresh medium, and then exposed to these compounds as above. The expression of target genes-of-interest (Supplemental Table 1) as assessed by quantitative real-time PCR (qRT-PCR). In brief, mRNA was extracted from cells treated as above using standard methods, and purified using RNeasy (QIAGEN Inc., Germantown, MD, USA) and Turbo DNA-free (Ambion, Austin, TX, USA) kits per manufacturer's instructions. Resulting mRNA was converted to cDNA using the RETROscript kit for reverse transcriptase PCR (RT-PCR; Ambion) per manufacturer's instructions. Each cDNA template was then used for qRT-PCR based on target gene primers and optimized for ABI 7000 system implementing a SYBR green PCR master mix (Applied Biosystems, Foster City, CA, USA). In all cases, the threshold cycle (Ct) values were normalized to 16S rRNA. Relative fold changes in gene expression were determined using the 2 −∆∆Ct method.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/antibiotics12050902/s1, Table S1: Quantitative heatmap of SAL and DIF impact on MRSA and MSSA virulence gene mRNA expression. Expression of regulatory (REG) genes RNAIII (agr), sarA, sigB as well as the predominant virulence factors (VIR) hemolysin hla (α-hemolysin) and protease sspA (V8 protease) genes were quantified over the course of 6 h following exposure to SAL or DIF (25 µg/mL) as compared to control. DIF inhibited the expression of RNAIII, hla and sspA as compared to SAL in SH1000 (MSSA) and USA300 (MRSA) backgrounds.   (Figures 2, 4, 6 and 7). Additional datasets are available upon request.

Conflicts of Interest:
MRY holds patents on novel anti-infective therapy, immunotherapy and vaccines, and is a consultant for Genentech-Roche, Alexion/AstraZeneca and Horizon Pharmaceuticals, which are involved in the development of immunotherapeutic agents and strategies. ASB holds patents on novel anti-infective therapy. RAP is a consultant for IBT, which is producing a multivalent antitoxin vaccine and is a member of the review board for the University of Rochester, which is involved in an anti-Gmd vaccine effort. Other authors report no disclosures.