Therapeutic Inhibition of Staphylococcus aureus ArlRS Two-Component Regulatory System Blocks Virulence

ABSTRACT

Staphylococcus aureus is a common cause of severe infections, and its widespread antibiotic resistance necessitates search for alternative therapies, such as inhibition of virulence. As S. aureus produces multiple individual virulence factors, inhibition of an entire regulatory system might provide better effects than targeting each virulence factor separately. Herein, we describe two novel inhibitors of S. aureus two-component regulatory system ArlRS: 3,4’-dimethoxyflavone and homopterocarpin. Unlike other putative ArlRS inhibitors previously identified, these two compounds were effective and specific. In vitro kinase assays indicated that 3,4’-dimethoxyflavone directly inhibits ArlS autophosphorylation, while homopterocarpin did not exhibit such effect, suggesting that two inhibitors work through distinct mechanisms. Application of the inhibitors to methicillin-resistant S. aureus (MRSA) in vitro blocked ArlRS signaling, inducing an abnormal gene expression pattern that was reflected in changes at the protein level, enhanced sensitivity to oxacillin, and led to the loss of numerous cellular virulence traits, including the ability to clump, adhere to host ligands, and evade innate immunity. The pleiotropic antivirulence effect of inhibiting a single regulatory system resulted in a marked therapeutic potential, demonstrated by the ability of inhibitors to decrease severity of MRSA infection in mice. Altogether, this study demonstrated the feasibility of ArlRS inhibition as anti-S. aureus treatment, and identified new lead compounds for therapeutic development.

INTRODUCTION

Staphylococcus aureus is one of the leading human pathogens, and a common cause of infections, which range from benign to acutely life-threatening (1). Treatment of staphylococcal infections, already challenging in itself, is complicated by the emergence of antibiotic-resistant strains, especially methicillin-resistant Staphylococcus aureus (MRSA) that led the CDC to label MRSA as an “urgent threat” and WHO to include it in their Global Pathogen Priority List (2, 3). This has prompted ongoing efforts to design alternatives to antibiotics, such as therapeutics targeting bacterial virulence (4–7). As bacteria rely on virulence factors to establish infection and to cause damage to the host, inhibition of these factors would probably render invading pathogens less harmful to the patient (4, 5). A major challenge has been identifying the best targets for treatment, especially in pathogens like S. aureus that possesses a vast array of virulence factors.

S. aureus can successfully infect nearly every tissue and organ in the body and cause multiple infections. The infection sites often include skin and soft tissues, bones and joints, or prosthetic devices, contributing to infections such as pneumonia, sepsis, and infective endocarditis (1). This versatility and ability to adapt to such different conditions requires S. aureus to precisely control the production of its virulence factors in response to the changing environment. S. aureus achieves such temporal and spatial control of virulence expression through two-component regulatory systems (TCS) (8, 9). TCS are composed of a membrane-bound sensory histidine kinase, that senses an input signal, which results in autophosphorylation of the kinase. This phosphoryl group is subsequently transferred to a response regulator protein, the second component of the TCS. Upon phosphorylation, the response regulator generally binds to DNA and modulates expression of target genes. When the input signal disappears, the sensory kinase returns to the basal state of phosphatase activity, which dephosphorylates the response regulator. The S. aureus genome encodes 16 such TCS (8). Except for WalKR, S. aureus TCS are nonessential for the growth, but are necessary for the correct response to environment during the infection, and thus for the staphylococcal virulence (10). This makes staphylococcal TCS an especially fitting target for antivirulence therapies (11). Instead of targeting an individual virulence factor, disruption of TCS could lead to simultaneous dysregulation of multiple virulence factors and to a much more pronounced reduction in virulence than could be achieved with inactivation of only a single virulence factor.

The ArlRS TCS has been shown to be necessary for virulence in multiple models of infection, including localized skin or muscle infections, biofilm on implants, and systemic bacteremia associated with peritonitis, endocarditis, or sepsis (12–16). ArlRS was responsible for virulence both in vertebrate (mouse, rabbit) and in invertebrate (silkworm larvae, honeybee) animal models, demonstrating its importance across vastly different infectious environments (12–18). The ArlRS TCS is composed of a sensor kinase ArlS and a response regulator ArlR, and it was initially described as a regulator of autolysis (“Autolysis related locus”), biofilm formation, multidrug efflux pumps, and exoprotein secretion (19–21). Since these initial studies, ArlRS was revealed to control expression of S. aureus surface adhesin genes, synthesis of capsule and other polysaccharides, secretion of diverse virulence factors, cell wall integrity, and metabolic adaptations to urea, metal, and oxygen availability (22–25). Importantly, S. aureus mutants lacking ArlRS are unable to attach to host cells and extracellular matrix, do not clump in plasma, cannot adapt to manganese starvation imposed by a host, fail to evade innate immunity, and have decreased resistance to some antibiotics (14–16, 26–30). Together, these studies identified ArlRS as an especially fitting target for antivirulence therapies. Although two compounds with anti-S. aureus activity, biochanin A and oritavancin, have been identified to also weakly target ArlRS (26, 27), to date there are no known specific and potent ArlRS inhibitors. Such improved inhibitors are necessary to test the effects of ArlRS-targeting therapies on S. aureus virulence and study the exact mechanism(s) of ArlRS activation and signaling.

In this work, we report identification of two compounds that show some specificity of inhibiting ArlRS signaling: 3,4’-dimethoxyflavone and homopterocarpin. They likely inhibit ArlRS through distinct mechanisms, including direct inhibition of autophosphorylation. We demonstrate that their use decreases S. aureus virulence, both in vitro and in a mouse model of infection.

RESULTS

3,4’-DMF and HPC inhibit ArlRS signaling.

To date, the literature suggests that ArlRS does not directly regulate expression of its effector genes. Instead, ArlR drives expression of two additional global regulators: MgrA and Spx, which then control expression of the target genes (Fig. 1A) (22, 24, 26, 31). While involvement of Spx is limited to the ArlRS-mediated adaptation to cell-wall targeting antibiotics like oxacillin, MgrA controls most of the ArlRS-regulated responses, including the regulation of virulence genes. Therefore, we used a plasmid-based reporter system with the mgrA P2 promoter (a promoter exclusively dependent on ArlRS) fused to the green fluorescent protein (GFP) gene as a reporter tool to measure ArlRS activity (31). S. aureus LAC, a representative community-acquired MRSA isolate from the widespread USA300 lineage (32), was used for the assays and hereafter referred to as MRSA WT.

FIG 1.

Novel inhibitors 3,4’-DMF and HPC decrease activity of ArlRS. Overview of the ArlRS regulatory cascade (A). Two novel ArlRS inhibitors: 3,4’-DMF and HPC, and their structures (B). When added to growth medium at increasing doses, the putative ArlRS inhibitors did not affect the growth of MRSA (C, D), but they caused a dose-depended decrease in ArlRS activity, measured through a fluorescent sGFP fusion with ArlRS-dependent P2 mgrA promoter (E, F). An sGFP fusion with ArlRS-dependent P2 spx promoter was used as a secondary readout and a dose-dependent inhibitory effect was observed again (G). All assays included a strain with ArlRS deletion (ΔarlRS) for comparison. ANOVA with a posttest for linear trend was used to determine statistical significances; ****, P < 0.0001. Data shown as mean ± SEM (C to F, n = 3 to 4) or as individual values of area under curve (AUC) of fluorescent signal, normalized by the OD₆₀₀ growth, with means indicated by horizontal lines (G).

After extensive analysis of a previous screen for ArlRS inhibitors (27), we noticed two overlooked compounds with possible anti-ArlRS activity: 3,4’-dimethoxyflavone (3,4’-DMF) and homopterocarpin (HPC) (Fig. 1B). 3,4’-DMF and HPC are relatively small plant-derived isoflavonoid derivatives previously studied for their anticancer properties (33, 34). Neither 3,4’-DMF nor HPC inhibited growth of MRSA when tested at the concentrations up to 100 μM (Fig. 1C and D). However, both compounds caused a pronounced, statistically significant dose-dependent decrease of ArlRS activity, as assessed via the mgrA P2-GFP reporter (Fig. 1E and F).

To ensure that the observed effect was not limited to mgrA, the expression of spx, the other gene directly dependent on the ArlRS activity (Fig. 1A), was analyzed using a reporter plasmid with GFP expression under the control of an ArlRS-dependent spx P2 promoter. The presence of either 3,4’-DMF or HPC in S. aureus culture caused a dose-dependent decrease of spx expression (Fig. 1G). This further confirmed that both compounds are inhibiting ArlRS activity.

Control experiments using a GFP reporter fused to the arlRS promoter indicated that neither 3,4’-DMF nor HPC decreased the expression of ArlRS itself, and even caused a minimal increase in ArlRS expression (Fig. S1A). This demonstrates that the observed inhibitor effects were due to the direct inhibition of ArlRS activity, rather than a side effect of decreased ArlRS expression. Overall, our results demonstrate 3,4’-DMF and HPC to be bona fide inhibitors of ArlRS activity.

3,4’-DMF and HPC are effective inhibitors with at least partial specificity.

Previous efforts identified two putative small-molecule ArlRS inhibitors: biochanin A (BchA) and oritavancin (ORI) (26, 27). However, both molecules also displayed anti-MRSA bactericidal or bacteriostatic activity. At subinhibitory concentrations, less than half of BchA antivirulence activity was attributable to ArlRS inhibition, suggesting low specificity and the presence of alternate targets (27). The observed ArlRS inhibition by ORI was weak and occurred only at a growth-inhibitory concentration (26). A direct comparison of the four known ArlRS inhibitors (3,4’-DMF, HPC, BchA, and ORI) using the mgrA P2-GFP fusion reporter at the highest possible concentrations that do not cause growth inhibition (6.25 μM for BchA; 0.05 μM for ORI) or surpass the solubility limit (100 μM for 3,4’-DMF and HPC) showed no significant decrease of ArlRS activity for ORI (Fig. 2A), suggesting that the previously observed anti-ArlRS activity was likely an artifact of its direct anti-MRSA bactericidal activity.

FIG 2.

The compounds 3,4’-DMF and HPC inhibit ArlRS more than other TCS systems. ArlRS inhibitors 3,4’-DMF and HPC were used in comparison with biochanin A (BchA) and oritavancin (ORI), added to MRSA cultures at the highest concentration being soluble or not causing growth inhibition (100 μM for 3,4’-DMF and HPC; 6.25 μM for BchA; 0.05 μM for ORI). When the ArlRS inhibition (measured through a fluorescent sGFP fusion with ArlRS-dependent P2 mgrA promoter), (A) Agr inhibition (measured through a fluorescent YFP fusion with Agr-dependent P3 RNAIII promoter) (B), or SrrAB inhibition (measured through a fluorescent sGFP fusion with SrrAB-dependent Pplc promoter) (C) were assayed, only the 3,4’-DMF and HPC demonstrated both high inhibitory activity and at least partial specificity for the ArlRS. Data shown as areas under curve (AUC) of fluorescent signal, normalized by the OD₆₀₀ growth measurements. Individual values are shown with means indicated by horizontal lines. ANOVA with a Dunnett’s multiple comparison posttest was used to determine statistical significances; *, P < 0.05; ***, P < 0.001; ****, P < 0.0001.

Two of the most studied S. aureus TCS are Agr and SrrAB. They are both heavily involved in regulation of S. aureus virulence, and Agr is being often explored as a drug target, although it appears to be very susceptible to nonspecific inhibition by structurally diverse compounds (35, 36). Therefore, we tested if any of the compounds could inhibit these two important TCS. Agr activity was measured using a plasmid carrying an agr P3-YFP promoter fusion (37), and SrrAB activity was measured using a plasmid carrying an Pplc-GFP promoter fusion (plc requiring active SrrAB for its expression [38]). Neither 3,4’-DMF, HPC, nor ORI had any effect on the Agr activity, but BchA strongly inhibited Agr activity (Fig. 2B). This indicates that BchA is not a specific ArlRS inhibitor, and explains why previously the majority of the BchA antivirulence effect was described as ArlRS-independent (27). In case of SrrAB, we observed a slight but statistically significant 10% reduction of SrrAB activity caused by 3,4’-DMF, and an increase in SrrAB activity caused by BchA. Neither HPC nor ORI had any effect on SrrAB activity (Fig. 2C). Overall, 3,4’-DMF and HPC appear to be more effective and somewhat more specific than any of the previously identified ArlRS inhibitors.

3,4’-DMF and HPC inhibit ArlRS in diverse S. aureus strains.

Additional experiments were conducted to ensure that 3,4’-DMF and HPC are active against various S. aureus strains beyond the MRSA strain LAC used for their initial identification. ArlRS inhibition was measured in S. aureus strain MW2, belonging to an MRSA USA400 lineage, and in methicillin-sensitive strains 502A, MN8 (USA200), and Newman. Irrespective of the strain used, both 3,4’-DMF and HPC caused significant inhibition of the ArlRS activity (Fig. S1B).

3,4’-DMF and HPC have distinct effects on ArlRS enzymatic activity.

Cell-based assays demonstrated strong evidence that the 3,4’-DMF and HPC inhibitors target ArlRS; thus, we sought to determine which domains in the TCS are being targeted. ArlS contains an extracellular sensing domain (Cache domain), a histidine kinase, adenylyl cyclase, methyl-accepting chemotaxis protein, and phosphatase (HAMP) domain, a dimerization and histidine phosphotransfer (DHp) domain and a catalytic ATP binding (CA) domain. ArlR is predicted to be an OmpR-like response regulator with receiver domain and winged helix-turn-helix DNA-binding domain (Fig. 1A) (22). To probe direct inhibition of ArlRS, we took advantage of established autophosphorylation and phosphotransfer assays (22, 36). Radioactivity-based ([γ-³²P]-ATP) autophosphorylation assays with the ArlS DHpCA catalytic region were performed in the presence of solvent (dimethyl sulfoxide, DMSO) or inhibitor (3,4’-DMF or HPC) in the final reaction mixture. At both 50 μM and 100 μM 3,4’-DMF, the percent phosphorylation of the ArlS DHpCA was consistently below the control levels, reaching statistical significance throughout the experiment at 100 μM (Fig. 3A). In contrast, addition of HPC caused only a slight decline in the phosphorylation, and this was not statistically significant (Fig. 3B). Control experiments with the cytoplasmic region of another S. aureus TCS, the SrrB histidine kinase (HAMP-PAS-DHpCA region) (36) showed no change in percent phosphorylation over 60 min in the presence of up to 100 μM 3,4’-DMF or HPC (Fig. S2A and B). We conclude that 3,4’-DMF inhibits the ArlRS TCS, in part, by targeting the autophosphorylation activity of the histidine kinase, while HPC inhibits ArlRS by some other mechanism.

FIG 3.

ArlS DHpCA autophosphorylation and phosphotransfer in the presence of 3,4’-DMF and HPC. Autophosphorylation of ArlS DHpCA in the presence of 3,4’-DMF (A) or HPC (B) is plotted. Phosphotransfer of ArlS DHpCA to ArlR in the presence of 3,4’-DMF (C) or HPC (D) is plotted. Unpaired two-tailed t test was used to determine statistically significant differences between compounds and the control; *, P < 0.05. Data shown as mean ± SEM (n = 3).

In addition, we tested how the inhibitors affect phosphoryl transfer from ArlS to its response regulator, ArlR. Autophosphorylation of ArlS proceeded for 1 h prior to adding full-length ArlR and incubating over ~15 min. In the absence of inhibitor, the percent of phosphorylated ArlR increased dramatically within 15 s and then over the next 3 min it began to disappear due to ArlS phosphatase activity (Fig. 3C and D). There was no statistical difference in the kinetics of ArlR phosphorylation/dephosphorylation in the presence of 3,4’-DMF or HPC. Similarly, control phosphotransfer experiments with SrrB HAMP-PAS-DHpCA and full-length SrrA showed the inhibitors did not affect the kinetics of phosphotransfer (Fig. S2C and D). We conclude that neither 3,4’-DMF nor HPC affect the phosphotransfer activity of the ArlRS TCS.

Overall, our data demonstrate that while both 3,4’-DMF and HPC inhibit activity of ArlRS (as evident in decreased expression of its targets mgrA and spx), their mechanisms of action must be different. In the case of 3,4’-DMF, its mechanism could possibly involve interference with autophosphorylation of ArlS, but in the case of HPC the inhibitory mechanism requires further investigation.

3,4’-DMF and HPC affect expression of ArlRS-regulated genes.

As the regulon of the ArlRS has been identified in the LAC strain (22), testing effects of the ArlRS compounds on expression of ArlRS-regulated genes was possible. The ΔarlRS strain, completely lacking ArlRS activity, is notable for a decreased expression of genes coding secreted virulence factors, such as nuclease (Nuc) and leukotoxins (e.g., Panton-Valentine Leukocidin [PVL], consisting of LukS and LukF subunits). On the other hand, expression of genes coding some unusual surface proteins, such as Ebh and SdrD, is increased in the ΔarlRS mutant (22). When expression of these genes was measured with GFP transcriptional fusions of respective genes’ promoters, a characteristic pattern emerged. In the presence of 3,4’-DMF and HPC, the expression of nuc and lukS decreased, while expression of ebh and sdrD increased, in a dose-response manner (Fig. 4A to D). The direction of these changes agreed with what was expected for agents inhibiting ArlRS activity.

FIG 4.

The inhibitors 3,4’-DMF and HPC modulate expression of genes regulated by ArlRS. The ArlRS inhibitors were added at increasing doses to growth medium of MRSA, and expression of various genes regulated by ArlRS was measured through fluorescent sGFP fusions with promoters of the respective genes: nuclease (nuc, A); LukS subunit of PVL (lukS, B), and surface proteins Ebh (ebh, C), and SdrD (sdrD, D). Data shown as areas under curve (AUC) of fluorescent signal, normalized by the OD₆₀₀ growth measurements. Individual values are shown with means indicated by horizontal lines. All assays included an ArlRS deletion (ΔarlRS) strain for comparison. All expression changes in the presence of inhibitors paralleled direction of changes observed in the ΔarlRS strain. ANOVA with a posttest for linear trend was used to determine statistical significances; ****, P < 0.0001. Effect of ArlRS inhibitors on level of selected proteins (LukS and Ebh) paralleled changes observed at the gene expression level, when it was visualized through a Western blot of proteins precipitated from an overnight culture (E).

The observed changes in gene expression also correlated with changes in protein levels of secreted LukS and sheared Ebh measured in culture supernatants by Western blotting. Strains treated with 3,4’-DMF and HPC had decreased amount of LukS and increased amount of Ebh, as seen in the ΔarlRS strain (Fig. 4E).

3,4’-DMF and HPC block S. aureus clumping, adhesion, and immune evasion.

One of the virulence mechanisms controlled by the ArlRS is S. aureus clumping in plasma. In the absence of ArlRS, a subset of giant surface proteins (e.g., Ebh) become derepressed, and their presence on bacterial surface interferes with clumping (16, 31). When S. aureus was grown in the presence of 3,4’-DMF or HPC, it lost its ability to clump in human plasma, like the ΔarlRS mutant (Fig. 5A). As ArlRS regulates production of the giant surface proteins through MgrA, complementation with MgrA expressed from a plasmid restores a wild-type phenotype in the ΔarlRS mutant (31). As expected, neither 3,4’-DMF nor HPC inhibitors had any effect on clumping in a strain constitutively expressing MgrA from a plasmid (Fig. 5A).

FIG 5.

The ArlRS inhibitors 3,4’-DMF and HPC block virulent properties of S. aureus in an MgrA- and Spx-dependent manner. When MRSA was grown in the presence of the ArlRS inhibitors at 100 μM, a marked decrease in its ability to clump in the presence of human plasma (A), to attach to fibrinogen-coated surface (B), and to kill human polymorphonuclear leukocytes (neutrophils) with its culture supernatant (C) was observed. These changes paralleled changes observed in the ΔarlRS strain. The ArlRS inhibitors had no effect on clumping, adhesion, and killing in strains with MgrA provided from a complementing plasmid pCM28::mgrA (A to C). Additionally, when S. aureus LAC MRSA was grown for 5 h in the presence of 0.5 μg/mL of oxacillin, addition of the ArlRS inhibitors at 100 μM caused a significant decrease in its growth, comparable to the level of the ΔarlRS strain (D). This effect did not occur in strains with Spx provided from a complementing plasmid pCM28::spx (D). An empty pCM28 backbone was used as a control for the experiments. ANOVA with a Dunnett’s multiple comparison posttest was used to determine statistical significances; *, P < 0.05; **, P < 0.01; ***, P < 0.001. Individual values from different replicates are shown with means indicated by horizontal lines. A total of 3 different donors were used for the (C).

In addition to inhibition of clumping, giant surface proteins derepressed in strains lacking arlRS additionally block S. aureus adhesion to host cells and extracellular matrix proteins (14, 27). The ArlRS inhibitors had the same effect—S. aureus grown in the presence of 3,4’-DMF or HPC showed decreased adhesion to human fibrinogen (Fig. 5B). Once again, the ArlRS inhibitors had no effect on a strain overexpressing MgrA from a plasmid (Fig. 5B).

In the absence of ArlRS, the secretion of leukocidins (LukAB and PVL/LukSF) is decreased, and this effect too is mediated by decreased mgrA expression (14, 22). Lack of leukocidins, which would normally lyse the incoming host phagocytes, renders S. aureus ΔarlRS susceptible to host innate immunity (14). This same phenotype was observed when ArlRS inhibitors were used. Supernatants from S. aureus grown in the presence of 3,4’-DMF or HPC had decreased ability to kill human neutrophils. This effect could be reversed by the forced production of MgrA from an expressing plasmid (Fig. 5C).

Altogether, the observed changes in gene expression and protein production caused by the two ArlRS inhibitors translated to the loss of S. aureus virulence in in vitro assays. These effects were mediated by the decreased mgrA expression, resulting from the ArlRS inhibition.

3,4’-DMF and HPC increase S. aureus oxacillin susceptibility.

The global regulator Spx is the second system, in addition to MgrA, which has its expression regulated by ArlRS and could be responsible for the effects of ArlRS inhibition. Thus far, Spx was identified to mediate adaptation of S. aureus to oxacillin, explaining reduced growth of ΔarlRS mutants in the presence of this antibiotic (26). Indeed, when oxacillin was present in the growth medium, addition of 3,4’-DMF or HPC led to reduced growth, comparable with reduction observed in the ΔarlRS mutant strain (Fig. 5D). As expected, the effect of inhibition disappeared when lack of Spx was complemented by expression of spx from a plasmid (Fig. 5D). This demonstrates that inhibition of ArlRS activity by 3,4’-DMF and HPC blocks both of its downstream effectors: the MgrA and the Spx regulators.

3,4’-DMF and HPC prevent tissue damage in MRSA skin infection in vivo.

Injection of MRSA into mouse skin causes localized infection, noticeable as an initial swelling and white discoloration of the skin, followed by the development of dark-colored dermonecrotic lesions. In the presence of 3,4’-DMF or HPC, the severity of an MRSA skin infection was significantly decreased, noticeable as a smaller total lesion area and a reduced dermonecrosis (Fig. 6A to C). This result was especially pronounced in the case of HPC, which almost completely prevented formation of dermonecrosis (Fig. 6B). At the same time, there were no statistically significant differences in the number of viable bacteria recovered from the skin lesions on day 3 between animals with and without inhibitors (Fig. S3), indicating that the observed changes in clinical outcome are due to a reduced virulence, and not due to any direct killing of the pathogen. Overall, the mouse experiments confirmed the potential of ArlRS inhibitors to decrease S. aureus virulence during a real-life infection.

FIG 6.

The inhibitors 3,4’-DMF and HPC decrease the severity of a mouse MRSA skin infection. ArlRS inhibitors were injected with MRSA (4 × 10⁶ CFU in 50 μL of 5 mM inhibitors or DMSO control) subcutaneously into the shaved abdomen of BALB/c mice. The total affected area (A) and the area of skin necrosis (B) was measured. Total lesion area of white discoloration is outlined with a dashed line, and dark-colored necrosis area in control group is indicated with an asterisk in representative photographs of skin lesion on day 1 (C). Data shown as mean ± SEM, n = 6 to 7. ANOVA with a Dunnett’s multiple comparison posttest was used to determine statistical significances; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

DISCUSSION

As concerns rise about bacteria becoming resistant to antibiotics and about antibiotics killing beneficial microbiota, directly targeting signaling pathways that regulate pathogen virulence emerges as a promising alternative approach. Antivirulence drugs, either used alone or as an adjuvant to antibiotic therapy, are likely the future of anti-infective therapy. Even though typical drugs targeting bacterial virulence are not yet available in the clinic, the success of vaccines, antisera, and engineered antibodies targeting virulence factors show the potential of this approach (6, 39). In the case of S. aureus, with its arsenal of redundant virulence factors, the challenge with antivirulence therapeutic strategies lies not only in designing the drugs but also in identifying optimal targets.

In this work we described two novel inhibitors (3,4’-DMF and HPC) of S. aureus regulatory system ArlRS, one of the main TCS controlling S. aureus gene expression (22). We also demonstrated that their use blocks virulence of S. aureus both in in vitro assays and in in vivo infection model. Targeting an entire regulatory system, instead of individual virulence factors, prevents bacteria from regulating gene expression in response to changing environment in the host, making bacteria functionally avirulent. We observed this phenomenon with the ArlRS inhibitors, as they blocked S. aureus ability to secrete virulence factors normally used to evade host immunity, and induced production of normally repressed surface proteins that prevent S. aureus from binding to host tissues and ligands. This result illustrates the advantage of targeting virulence regulation instead of individual virulence factors for therapeutic gain.

Ideal antivirulence drugs should not be directly toxic to bacteria to avoid strong selective pressure for resistance development. Therapeutics should also be specific, especially in the case of S. aureus infections, where correct timing of intervention, and precise targeting of regulatory system to match the type of the infection and the patient’s condition, are of utmost importance (40). For instance, nonspecific inhibition of S. aureus Agr signaling during an incorrect phase of infection might aggravate the disease. In line with these expectations for potential therapeutics, the two ArlRS inhibitors identified here (3,4’-DMF and HPC) indeed have no direct S. aureus killing activity. Moreover, the use of these two candidate inhibitors led to changes in gene expression, protein production, and virulence properties in perfect agreement with changes seen after ArlRS inactivation through genetic deletion in an ΔarlRS strain (22). To the extent that we were able to determine, they also display at least partial specificity toward ArlRS. Importantly, the inhibitors did not affect the Agr system, which is another TCS often targeted by small molecule compounds. In the case of 3,4’-DMF, a weak inhibitory effect toward the SrrAB TCS was observed, but HPC displayed no off-target effects. The lack of strong inhibitory action toward the Agr and SrrAB TCS is a promising sign that 3,4’-DMF and HPC are at least relatively specific to ArlRS. Thus, both compounds appear to be good examples of ArlRS inhibitor leads to improve upon for future therapeutic use.

The mechanisms of action of 3,4’-DMF and HPC remain unclear. Our autophosphorylation data suggest that 3,4’-DMF likely inhibits ArlS by affecting its kinase activity. Inhibition of some histidine kinases by flavonoid compounds has been previously shown (41–43). For one such flavonoid, luteolin, structural studies showed that it binds to the nucleotide pocket of HK853 from Thermotoga maritima (43). Therefore, we suspect that the likely target of 3,4’-DMF is also the ArlS nucleotide binding pocket. In contrast, HPC has a distinct structure compared with luteolin and 3,4’-DMF that presumably is not able to bind in this pocket. Thus, additional mechanisms of action most likely exist. For example, ArlS contains a Cache domain, and the DHpCA construct used here did not contain this domain. Cache domains are known to bind small molecule ligands and binding can regulate HK activity (44) and therefore binding to the Cache domain may be another mode of inhibition. Further work involving full-length recombinant ArlS and ArlR and improved compounds based on the scaffolds of 3,4’-DMF and HPC will be necessary to clarify the mechanisms of inhibition.

Overall, 3,4’-DMF and HPC are the two most effective and at least somewhat specific ArlRS inhibitors identified to date. While their potency (in the micromolar range, compared with nanomolar and picomolar ranges of reported antivirulence drug candidates [45]) does not make them good candidates for direct clinical application, they provide a scaffold for the development of improved inhibitors. Importantly, testing of 3,4-DMF and HPC activity in a mouse infection model demonstrated that they can alleviate the severity of an MRSA skin infection. This study therefore established for the first time that specifically targeting ArlRS during infection can be a successful treatment strategy. The central relevance of the ArlRS system in many different experimental models provides justification for cautious optimism that pharmacological inhibition of this TCS might be an effective treatment for a wide range of S. aureus infections. However, as this study investigated ArlRS modulation in only a skin infection model, additional studies in other type of models will be needed. As many of the ArlRS-regulated immune evasion factors are human-specific and are inactive in mouse models (14), we predict the effect of ArlRS inhibition to be even more pronounced during human infection. While our studies only assessed prophylactic treatment of MRSA infection, improved derivatives 3,4-DMF and HPC will be necessary to demonstrate a curative effect in animal models of infection. Overall, our study adds ArlRS to the already established Agr quorum-sensing system as two of the most clinically relevant TCS targets for anti-S. aureus drug development (35).

In summary, we have identified the first, at least partially specific, ArlRS inhibitors and demonstrated that inhibition of ArlRS can be an efficient treatment strategy for S. aureus infections. While there has been marked progress in deciphering mechanistic aspects of the ArlRS signaling (22) and ArlR DNA binding specificity (46, 47), there still remain gaps in understanding ArlS function, such as identifying the input signal for ArlS and solving the structure of full-length ArlS (22, 46, 47). This constantly growing body of knowledge will eventually allow for a more detailed understanding of ArlRS regulation and increased opportunities to develop novel and effective anti-S. aureus treatments.

MATERIALS AND METHODS

Strains, reagents, and growth conditions.

Bacterial strains and plasmids used are listed in the Table 1. For experiments, S. aureus was grown in tryptic soy broth (TSB), and Escherichia coli was grown in lysogeny broth (LB), both at 37°C with shaking. When needed for plasmid maintenance, antibiotics were added to the media: chloramphenicol (Cm, 10 μg/mL), erythromycin (Erm, 10 μg/mL), or ampicillin (Amp, 100 μg/mL). Unless indicated, all reagents were from Millipore-Sigma, Thermo Fisher Scientific, or Research Products International.

TABLE 1.

Strains and plasmids used

Strain/plasmid	Description	Source/reference
S. aureus
AH1263	USA300 MRSA, ermS (= LAC)	50
AH1975	LAC ΔarlRS	16
AH3052	LAC Δspa	51
AH843	MW2	52
AH3060	MW2 ΔarlRS::tetM	31
AH1178	Newman	53
AH3062	Newman ΔarlRS::tetM	31
AH2413	MN8	54
AH3063	MN8 ΔarlRS::tetM	31
AH3610	502A	55
AH3624	502A ΔarlRS::tetM	31
E. coli
DC10B	Cloning strain	58
Plasmids
pDB59	agr P3-YFP fusion reporter, CmR	37
pHC68	mgrA P2-sGFP fusion reporter, ErmR	31
pHC177	spx P2-sGFP fusion reporter, CmR	22
pHC148	PlukS-sGFP reporter, CmR	22
pHC149	Pebh-sGFP reporter, CmR	22
pHC151	PsdrD-sGFP reporter, CmR	22
pCM38	Pnuc-sGFP fusion reporter, CmR	56
pCM29	sGFP expression vector, CmR	57
pJK16	ParlRS-sGFP reporter, CmR	This paper
pJK21	Pplc-sGFP reporter, CmR	This paper
pCM28	Empty backbone for expression vectors, CmR	57
pHC66	mgrA complementing plasmid (pCM28::mgrA), CmR	31
pJK15	spx complementing plasmid (pCM28::spx), CmR	This paper

For construction of new plasmids, all enzymes used were from New England Biolabs, all cloning was performed in E. coli DC10B, the final sequences were confirmed by automated sequencing at Molecular Biology Core Facility at University of Colorado School of Medicine, and the created plasmids were introduced into S. aureus LAC through electroporation.

Inhibitors and antibiotics.

The inhibitors 3,4’-dimethoxyflavone (3,4’-DMF) and homopterocarpin (HPC) were from MicroSource Discovery Systems. Biochanin A and oritavancin diphosphate were from Millipore-Sigma. Oxacillin sodium was from Acros Organics. All compounds were prepared as 10 mM stocks in DMSO. In all experiments, DMSO was added to growth medium to give a final concentration of 1% DMSO (adjusted for different amount of added compound dissolved in DMSO). Samples with 1% DMSO and no compound were used as controls.

Human plasma and neutrophils.

Human plasma diluted 1:1 with heparin/dextran sulfate to prevent clotting (this mixture referred to as 100% plasma) was obtained from a plasma bank at the University of Iowa Inflammation Program. Healthy human adult polymorphonuclear leukocytes were purchased from the Division of Pulmonary, Critical Care and Sleep Medicine at the National Jewish Health at Denver, where they were isolated by the plasma-Percoll method (48), and used within 4 h from isolation.

Construction of arlRS expression reporter.

To generate an arlRS promoter-sGFP transcriptional reporter, first the promoter of arlRS was amplified with Q5 polymerase from S. aureus LAC genomic DNA using the primers JK129 (TGCTCTAGATAGTGAAAAGTCAGTATAT) and JK130 (TCAGGTACCTACGACTTTTTCTAATAAG). This DNA fragment was digested with XbaI and KpnI, and ligated into pCM29, digested with the same enzymes, upstream of an optimized ribosome binding site and codon-optimized sGFP, to create the reporter plasmid pJK16.

Construction of plc expression reporter.

To generate an plc promoter-sGFP transcriptional reporter, first the area of 179 bp between the previous gene and the plc transcription start, which include the promoter of plc (38), was amplified with Q5 polymerase from S. aureus LAC genomic DNA using the primers JK167 (TAGGTACCTACATTAATTATACATCTTTTTTAAATAAAAATATGTGTAAAATTTT) and JK168 (CTGGCTAGCGATAAGGAGCTGGCGATTA). This DNA fragment was digested with NheI and KpnI, and ligated into pCM29, digested with the same enzymes, upstream of an optimized ribosome binding site and codon-optimized sGFP, to create the reporter plasmid pJK21.

Construction of spx complementing plasmid.

To generate a spx complementing plasmid, first the spx and its promoter were amplified with Q5 polymerase from S. aureus LAC genomic DNA using the primers JK125 (ACAGGATCCGTCTCCATTTAAATGCCTACTTTCTTAG) and JK126 (GATGTCGACTTAGTCAACCATACGTTGTGCTTC). This DNA fragment was digested with BamHI and SalI, and ligated into pCM28 digested with the same enzymes, producing plasmid pJK15.

Expression reporter assays and growth curves.

To assess the expression of promoter-sGFP fusions, overnight cultures of S. aureus carrying appropriate plasmid were diluted 1:1,000 in 200 μL of TSB with antibiotic for plasmid maintenance in 96-well black-walled plates. Plates were incubated at 37°C with shaking (1,000 rpm) in an SI505 humidified plate shaker (Stuart), and growth (OD₆₀₀) and fluorescence (excitation 495 nm, emission 515 nm) were measured at regular intervals with an Infinite 200 plate reader (Tecan). For easier visualization of the results, some graphs had the fluorescent signal normalized to the OD₆₀₀ and displayed as the calculated area under the curve (AUC).

For measuring growth in the presence of oxacillin, cultures were prepared as for the reporter assays, and oxacillin was added to the medium to the final concentration of 0.5 μg/mL. After 5 h of incubation (37°C, 1,000 rpm shaking, SI505 humidified plate shaker), the S. aureus growth was measured as OD₆₀₀ with an Infinite 200 plate reader.

Western blotting.

S. aureus LAC strain lacking protein A (Δspa), which interferes with Western blotting, was used to detect LukS and Ebh. Proteins from overnight TSB culture supernatants (containing secreted LukS, and Ebh sheared off from bacterial surfaces), cultured with inhibitors or DMSO, were precipitated with trichloroacetic acid, separated through an SDS-PAGE gel electrophoresis, and transferred to a Trans-Blot Turbo Mini 0.2 μm nitrocellulose membrane (Bio-Rad). Proteins were detected with rabbit polyclonal anti-LukS antibodies (IDT BioServices) or rabbit anti-Ebh serum (16), followed by IRDye 680RD goat anti-rabbit IgG secondary antibody (Li-Cor) and visualized with an Odyssey CLx Imaging System (Li-Cor) (22).

Clumping assay.

S. aureus from exponential culture in TSB (OD₆₀₀ = 1.5) was washed and resuspended in PBS. Human plasma was added to 1.5 mL of bacterial suspension in a microcentrifuge tube to a 2.5% vol/vol, vortexed, and left for 2 h at room temperature. The percent of clumping (visible as sedimentation of the formed clumps at the bottom of the tube, resulting in a loss of optical density in the upper part of the tube) was determined by removing the uppermost 100 μL of the suspension, measuring its OD₆₀₀, and comparing it with the OD₆₀₀ of the suspension at time zero (16, 27, 31).

Adhesion to fibrinogen assay.

Cell culture 96-well plates were coated with 20 μg/mL of human fibrinogen in PBS overnight at 4°C. Afterwards the wells were washed, blocked with 5% bovine serum albumin (BSA) at 37°C for 2 h, and filled with 100 μL of S. aureus suspension in PBS at OD₆₀₀ = 1.0, prepared from exponential TSB culture as for the clumping assay. After 1 h of incubation at 37°C, the bacterial suspensions were aspirated, the wells washed, dried, and stained with 0.1% crystal violet to visualize the adherent bacteria. Bound stain was solubilized with 33% acetic acid and the absorbance at OD₅₇₀ was measured. The adhesion was standardized to the average untreated control (27).

Neutrophil killing assay.

Human polymorphonuclear leukocytes were seeded at 1 × 10⁵ cells per well into 96-well plates in 95 μL of RPMI 1640 medium with 10% fetal bovine serum, and with 5 μL of filter-sterilized supernatants from overnight S. aureus cultures (final concentration of 5%). After 3 h of incubation at 37°C in 5% CO₂, the plates were centrifuged at 250 g for 10 min. The resulting supernatants were used to measure lactate dehydrogenase (LDH) leakage from damaged cells as a marker for neutrophil lysis using an LDH Cytotoxicity Detection Kit (Roche). The percent of neutrophil lysis was calculated using neutrophils incubated with 5% of pure TSB+DMSO as “0% lysis,” and incubated with 0.2% Triton X-100 as “100% lysis” references (14).

Mouse skin infection.

Mouse experiments were approved by the University of Colorado Institutional Animal Care and Use Committee, protocol #00486. S. aureus LAC from mid-log growth phase in TSB was washed with PBS and resuspended in sterile saline. Bacterial suspensions were mixed 1:1 with DMSO or inhibitors suspended in DMSO for a final inhibitor concentration of 5 mM immediately before injection. Abdomens of 7- to 8-week-old BALB/c female mice (Jackson Laboratories) were depilated using an electric razor followed by a hair removal cream (Nair). Abdomens were wiped with alcohol pads and 50 μL of bacteria suspension containing 4 × 10⁶ CFU and DMSO or inhibitors was injected subcutaneously using an insulin syringe. Mice were weighed and photographed daily, and the lesion area and dermonecrosis was documented using FIJI image analysis software (49). On day 3 of infection, mice were euthanized and 8-mm diameter punch biopsies of the infected area were obtained, homogenized, and plated on tryptic soy agar for CFU determination.

Protein production and purification.

Previously described expression constructs were used (22). Briefly, the ArlS DHpCA construct (residues 231 to 415) was cloned into a modified pET21a vector (Novagen) containing a N-terminal 6×His-tag followed by a recombinant Tobacco Etch Virus (rTEV) protease cleavage sequence (ENFLQG). The full-length ArlR response regulator (residues 1–to 219) DNA was cloned into a modified pET31b vector (Novagen) containing a 6×His-tag followed by a rTEV protease cleavage sequence (ENFLQG). The vectors were transformed into Rosetta(DE3) E. coli cells (Novagen) for protein production. Two 1 L cultures per construct were grown to an OD₆₀₀ between 0.6 and 0.8 and induced with 0.5 mM isopropyl β-d-1-thiogalactopyranoside overnight with shaking at 18°C. The cell pellet was resuspended in 50 mL of buffer A (50 mM Tris pH 7.5, 250 mM NaCl, 10 mM imidazole, 5% glycerol, 10 mg/mL NaN₃) with an EDTA-free protease inhibitor tablet (Roche). The cells were lysed and purified by nickel affinity chromatography (Nickel Sepharose 6 fast flow media, GE Healthcare) using an increasing concentration of buffer B (5% glycerol, 50 mM Tris pH 7.5, 250 mM NaCl, 300 mM imidazole, 10 mg/mL NaN₃) to elute bound protein. The fractions containing the desired protein were pooled and dialyzed in 50 mM Tris pH 7.5, 150 mM NaCl, 1 mM dithiothreitol (DTT) for 36 h at 4°C in the presence of rTEV (0.75 mg) to cleave the His-tag from the fusion protein. The cleaved proteins were further purified by size exclusion chromatography (SEC) on an S200 column (Superdex 200, GE Healthcare) with SEC buffer (25 mM Tris pH 7.5, 150 mM NaCl, 10% glycerol, 0.5 mM EDTA, 1 mM DTT). The final fractions containing the protein of interest were concentrated to 50 μM and dialyzed into storage buffer (25 mM Tris pH 7.6, 125 mM NaCl, 0.5 mM EDTA, 1 mM DTT, 1% Triton X-100, 50% glycerol) and stored at −20°C.

Autophosphorylation and phosphotransfer assays.

The autophosphorylation activity of the ArlS DHpCA construct and inhibitors was observed overtime by monitoring incorporation of ³²P from [γ-³²P]-ATP present in the reaction mixture (22, 36). Phosphorylation assays were performed in a 50 μL reaction volume in kinase buffer (25 mM Tris pH 7.6, 1 mM MgCl₂, 1 mM MnCl₂, 1 mM CaCl₂, 50 mM KCl, 1 mM DTT), in the presence of the indicated amount of inhibitor (or DMSO control) and 5 μM ArlS DHpCA. The reaction was equilibrated for 5 min at room temperature. ATP mix (0.25 μM ATP-[γ-32P] [Perkin Elmer] and 250 μM “cold” ATP [Sigma]) in kinase buffer was added to start the reaction. The reaction was stopped at desired time points by adding SDS-PAGE loading buffer. Samples were resolved by SDS-PAGE gel electrophoresis and radioactive bands were visualized using a Phosphor imager (GE Healthcare Typhoon FLA 9500) and quantified with FIJI image analysis software (49).

Phosphotransfer between phosphorylated ArlS DHpCA and full-length ArlR response regulator was determined in kinase buffer. The ArlS DHpCA was first subjected to an autophosphorylation reaction for 1 h (as described above) to obtain phosphorylated DHpCA. A response regulator mix containing ArlR response regulator and inhibitor was assembled. The phosphotransfer assay was initiated by combining an aliquot of phosphorylated DHpCA with the ArlR response regulating regulator mix (5 μM final concentration of ArlS DHpCA and ArlR). Aliquots (5 μM) of the reaction were removed at desired time points and quenched with SDS-PAGE buffer. Samples were loaded on 4% to 20% or 12% SDS-PAGE gradient gels and electrophoresed. Radioactive bands were visualized using a Phosphor imager (GE Healthcare Typhoon FLA 9500) and quantified with FIJI image analysis software (49).

Statistics.

For all biological assays, data were pooled from two to three independent experiments. Differences between experimental groups and controls were analyzed by ANOVA with a Dunnett’s multiple comparison posttest, by ANOVA with a posttest for linear trend (for dose-response experiments), or by Kruskal-Wallis test with a Dunn’s multiple comparison posttest (for CFU).

For autophosphorylation and phosphotransfer assays, normality was confirmed for all data sets with a Shapiro-Wilk normality test (P > 0.05), and a t test was performed to detect differences between experimental (compound) and control (DMSO) samples.

For all assays, two-tailed P values were used, and Prism 7 (Graph Pad Software) was used for calculations.