Repression of Virulence in Staphylococcus aureus by 2‐Aminothiol‐Mediated Quorum Quenching

ABSTRACT

We present a novel strategy for the inhibition of quorum sensing in Staphylococcus bacteria. Previous work has focused on the development of peptides that can compete with native autoinducing peptides (AIPs), to antagonize the receptor histidine kinase responsible for activation of quorum sensing and thereby repress the expression of virulence genes. Here we report simple analogues of cysteine that can eliminate the native AIPs extracellularly, before they activate their receptor, and validate that this alternative strategy can also be applied to turn off quorum sensing in different Staphylococcus aureus reporter strain assays. Further, we show that two selected compounds can reduce skin infection caused by MRSA in a murine model. This work provides impetus for the development of simple small molecules for the topical treatment of staphylococcal skin infections. We anticipate that further studies, including delivery strategies, will help illuminate this potential.

Simple analogues of cysteine that can eliminate the native autoinducing peptides (AIPs) of staphylococci extracellularly, before they activate their receptor, and turn off quorum sensing in different Staphylococcus aureus. The elimination of AIPs occurs through a mechanism similar to native chemical ligation (NCL). Further, this strategy showed promise for reducing skin infection caused by MRSA in a murine model.

The Gram‐positive bacterium Staphylococcus aureus is an opportunistic human pathogen, causing skin and soft tissue infections, among others, and alarming rates of emerging methicillin‐resistant S. aureus (MRSA) strains are challenging the global health [1, 2, 3]. Thus, novel treatment options are needed. An alternative to conventional antibiotics that has received significant attention, is reducing virulence rather than eliminating the bacteria [4, 5, 6]. Expression of bacterial virulence factors is often regulated through quorum sensing (QS), which is the ability of the bacteria to change gene expression in response to cell density [7, 8]. In staphylococci, the QS machinery is encoded by the accessory gene regulator (agr) locus and relies on the secretion and detection of cyclic thiodepsipeptides, called autoinducing peptides (AIPs) that auto‐activate their AgrC receptor histidine kinase (Figure 1), which phosphorylates the response regulator (AgrA), in turn, leading to global changes in gene expression downstream, including expression of virulence factors [9]. However, crosstalk has also been observed, where AIPs of one species are inhibitory towards agr systems of other species and even between strains of the four different S. aureus agr types (agr‐I–IV) [5, 10]. Such QS inhibition can repress virulence gene expression in S. aureus, which has been demonstrated to attenuate MRSA infection in murine skin models [11, 12, 13]. A recent pioneering study even investigated the disease outcome of S. aureus infection upon co‐colonization with the AIP producing commensal bacterium Staphylococcus hominis in a human clinical trial [14].

FIGURE 1.

Cartoon representation of the quorum quenching strategy, relying on AIP destruction by native chemical ligation (NCL)‐like reaction.

Our laboratory has developed a platform for mapping the chemical messenger molecules (AIPs) responsible for QS, which is based on the covalent trapping of AIPs in a highly chemo‐selective manner, using solid supported cysteine [15]. In extension of this work, we hypothesized that cysteine in solution could likewise react with excreted AIPs through a native chemical ligation [16] type reaction to linearize the AIP and, in turn, inhibit QS (Figure 1). To test this hypothesis, we first incubated S. aureus AIP‐I with cysteine (Cys) at different concentrations and evaluated the destruction of the thiolactone ring over time in an HPLC‐based assay (Figure S1A,B). At 40 equivalents of added cysteine, the AIP was rapidly destroyed with close‐to‐full conversion at the 30‐min time point, and we therefore tested a series of additional aminothiol compounds from commercial sources.

The most efficacious compounds were cysteine (Cys) and cysteamine (CA), while homocysteine (hCys) was less effective and no elimination was observed for penicillamine (PA) and N‐acetylcysteine (NAC) after 30 min incubation (Figure 2A,B). In addition, we tested the destruction of AIP‐I in a differently designed UPLC assay, where the products were quenched with acid, resulting in appearance of potential transient thioester species (Figure S2). This assay revealed the presence of a minor ring‐opened species resulting from the NAC treatment, which agrees with the presence of an equilibrium between the AIP and its C‐terminal thioester species as suggested in Figure S2H. Because both species would give the same quenching product in the former UPLC assay format, these data are also in agreement with complete lack of conversion shown in said assay (Figure 2B). We further tested the potential effect of aminolysis by treatment with ethylenediamine or 1‐aminobutane. Not surprisingly, based on recent investigation of the reactivity of thioesters toward amine nucleophiles at neutral pH [17], only minor degrees of aminolysis were observed (Figure S2). Finally, we tested whether the AIP destruction would also occur in the more complex environment of a growth medium. We therefore spiked fresh medium with AIP and treated the solution with either Cys or NAC for 30 min, before applying our previously developed AIP trapping workflow (Figure S3A). The mass spectrometry, following enrichment of residual AIP showed the presence of AIP after treatment with NAC but not after treatment with cysteine, supporting that occurrence of AIP destruction by the 1,2‐aminothiol in the growth medium and showing that high concentrations of thiol‐containing species (i.e., 10 mM NAC in addition to cysteine‐containing peptides present in the growth medium) are not sufficient for destroying the AIP (Figure S3).

FIGURE 2.

Compound collection. (A) Commercially obtained 2‐aminothiols. (B) Bar graph representation of the ability of the commercial compound collection to destroy the thiolactone moiety of S. aurues AIP‐I as measured in the UPLC assay described in Figure S1. (C) Synthesized compounds (see, Schemes S1–S4). (D) Inhibition of S. aureus agr‐I in fluorescent reporter assay. Note (*): Compound 3 visibly precipitated upon dilution in medium.

We then tested the effect of Cys and NAC in a reporter strain where the agr‐regulated P3 promoter of the central virulence regulator, RNAIII is fused to yfp, encoding the yellow fluorescent protein (YFP) [18]. These assays showed that the free 2‐aminothiol (Cys) indeed was able to inhibit QS as expected, because they can engage in the NCL‐like mechanism, while NAC exhibited no effect in agreement with its inability to cause the S‐to‐N acyl shift in the NCL reaction [16] (Figure S4). To address whether the rather high concentrations of compound added had any effect on the bacteria's fitness, which could indirectly affect QS, we performed independent CFU count experiments. No difference in CFU count were observed for Cys (5 mM) or NAC (5 mM), compared to vehicle control cultures after 24 h (Figure S5).

Next, we extended our series of 2‐aminothiol‐containing small molecules through chemical synthesis (compounds 1–12) and screened all compounds for quorum quenching activity against the S. aureus agr‐I specificity group in the fluorescent reporter strain assay. The compound collection was designed to incorporate both esters (1–3), an amino alcohol (4), and amides (5–12) with various electronic properties and polarity to cover a span of pKa values of the thiol and amino groups as well as varying polarity (Figure 2C and Schemes S1–S4).

The compound screening against the S. aureus agr‐I YFP reporter strain revealed that most compounds could quench QS at 5 mM concentration; however, few exhibited >50% inhibition at the 2.5 mM concentration (1, 5, 7, 8, and 11), and only 1, 5, and 11 exhibited >40% inhibition at 1.25 mM (Figure 2D). We envisioned that the amide‐containing compounds would be less prone to hydrolysis than the ester‐containing (1) and therefore selected compounds 5 and 11 for further evaluation in dose–response experiments with all four S. aureus agr group I–IV reporter strains. We also included the parent compound cysteine (Cys) and the minimal analogue cysteamine (CA) for comparison. Across all four specificity groups, cysteine (Cys) was least potent and compounds 5 and 11 were equipotent, showing the highest activity (Figure 3). Interestingly, we observed the lowest degree of repression of agr‐IV, which is in line with the hyperactivity of this strain and the lowest inhibitory potential seen in our previous studies of inhibition with AIPs.

FIGURE 3.

Dose‐response curves for quorum sensing inhibition in S. aureus agr‐I–IV P3 reporter strain assays. At least two individual assays were performed in technical triplicate. Each curve represents the two biological replicates with IC₅₀ curves fitted by nonlinear regression (variable slope) in GraphPad Prism; data points and error bars represent the mean ± SD.

With the promising data relying on activation/inhibition of RNAIII expression in these reporter strains, we were interested in gaining further proof that the observed effect on QS was indeed due to destruction of native AIP. We therefore designed an experiment where we pre‐incubated activating AIP‐I with serial dilutions of either NAC or Cys for 30 min before performing β‐lactamase reporter assays, which showed no inhibition of QS for NAC and a dose‐dependent reduction in QS for Cys (Figure S6A,B). Further, we tested synthetic linearized AIP that contains an additional cysteine residue at the C‐terminus (i.e., the product of AIP‐I–Cys NCL reaction), which produced a minor level of activation (<10%) at µM concentrations, but no activation at the common activator concentration of 100 nM (Figure S6C). These assays provide further support of a mechanism of action involving NCL‐like destruction of the activating AIP to inhibit QS.

We also developed a reporter strain where the green fluorescent protein (GFP) gene is fused to the spa promoter, which is repressed during activated QS [10]. Thus, an AIP‐I activated culture is nonfluorescent, and fluorescence increases when QS is inhibited. In this newly developed spa::gfp plasmid‐based reporter assay [10], the potent inhibitor of agr‐I (AIP‐III_D4A ) [19] was able to inhibit an activated culture, while NAC showed no activity against the AIP‐I activated culture (Figure 4A), which is in agreement with its inability to engage in the NCL‐like AIP destruction as shown in our HPLC assays. We then selected cysteamine (CA) and compound 11 due to its high potency and higher aqueous solubility compared to compound 5, for further testing in this assay. Both 2‐aminothiol compounds exhibited inhibition of QS in the spa‐gfp reporter assay, with the effect of CA at 20 mM concentration reaching the quenching level of the potent inhibitor AIP‐III_D4A (Figure 4A). Thus, these data provide additional proof, supporting our hypothesis that the excreted native AIP is indeed eliminated by our 2‐aminothiol compounds to furnish the quorum quenching effect. Further, we performed dose–response experiments with CA and compound 11, but experienced challenges with the solubility of compound 11 at concentrations above 5 mM in this assay, which prohibited determination of IC₅₀ values for this otherwise potent compound (Figure 4B). The CA exhibited similar IC₅₀ values regardless of whether the culture had been activated by AIP‐I at 50 nM (IC₅₀ 6.9 ± 0.57 mM) or 100 nM (IC₅₀ 7.0 ± 0.44 mM) concentration. This is different from the behavior of inhibitory AIPs in this assay but may reflect that these quenchers are not competing for binding to the receptor.

FIGURE 4.

Evaluation of the ability of selected 2‐aminothiols to quench already turned on quorum sensing in a spa‐gfp reporter assay [10]. (A) Continuous assay with all cultures induced by AIP‐I (50 nM) and challenged with the selected compounds as noted. (B) Dose–response experiments for CA and compound 11 against cultures activated by AIP‐I at either 50 nM or 100 nM concentration. Each curve represents the two biological replicates with IC₅₀ curves fitted by nonlinear regression (variable slope) in GraphPad Prism; data points and error bars represent the mean ± SD.

Finally, we tested the ability of compound 11 to treat a skin infection caused by MRSA in a previously reported murine skin infection model [10, 20]. We also included cysteamine (CA), because it is an approved drug substance for human clinical use in the treatment of cystinosis [21], with a well‐documented toxicity profile [22]. Gratifyingly, a significant reduction in the skin lesion size, measured on day 2, was observed after a single treatment with CA or 11 (10 mM), similar to that achieved by daily application of Fucidin (39 mM). On day 4, however, the effect was not statistically significant for any of the treatments, due to the reduced skin lesion area also observed for the vehicle control mice (Figure 5).

FIGURE 5.

Murine skin infection model performed with vehicle control and fusidic acid groups. All groups were of eight animals and half were sacrificed on day two, ending up with four animals per group on day four. (A) Skin lesion size measured at days 2 and 4. (B) Body weight measurements. Data are presented as mean and error bars represent the standard deviation (SD) of the mean. Statistical significance between treatment and vehicle groups was assessed by one‐way analysis of variance (ANOVA) with multiple comparisons (Dunnett's test), p > 0.05 (ns), p < 0.05 (*), p < 0.01 (**). The murine model was performed under contract at Statens Serum Institut (DK) as previously described [20]. All animal experiments were approved by the National Committee of Animal Ethics, Denmark.

In conclusion, we provide evidence to support that elimination of AIPs secreted by S. aureus (and MRSA), using simple small molecules with a 2‐aminothiol motif, should be considered as a novel anti‐virulence strategy. We show that the thiolactone moiety of AIPs is destroyed by a native chemical ligation mechanism. Further, we show that this approach can be applied for QS inhibition in bacterial culture, using two fundamentally different reporter strain assays. Finally, promising data were obtained in a preliminary in vivo study, in a murine MRSA skin infection model.

It is our hope that these data will inspire further investigations of the in vivo potential of quorum quenching by AIP destruction, including additional dosing and delivery experiments, which may provide new avenues for the treatment of bacterial infections.

Author Contributions

Bengt H. Gless: conceptualization, investigation, writing – review and editing, formal analysis, methodology, visualization, data curation. Riccardo Milazzo: investigation, writing – review and editing, formal analysis. Martin S. Bojer: investigation, writing – review and editing, methodology, formal analysis. Benjamin S. Sereika‐bejder: investigation, writing – review and editing. Hanne Ingmer: funding acquisition, writing – review and editing, resources, supervision. Christian A. Olsen: funding acquisition, writing – original draft, visualization, writing – review and editing, project administration, supervision, resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: The authors have cited additional references within the Supporting Information [23, 24, 25].

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.