Introduction

Though the undeniable efficiency of anti-infective measures like vaccines and antimicrobials have made us believe that infectious diseases are nowadays under control, nothing could be further from the truth. An increasing number of studies alert that unless actions are taken, infections caused by antibiotic-resistant bacteria will kill an extra 10 million people a year worldwide by 20501.

Staphylococcus aureus is one of the bacterial species whose manage is especially challenging due to the emergence of methicillin-resistant (MRSA) strains. With a population-weighted mean of invasive MRSA strains of about 17% in terms of European prevalence2, the major health care concern related to MRSA incidence lies in the limitations of currently approved treatments, which, in turn, leads to high rates of morbidity and mortality even in industrialized nations.

Despite of being an inoffensive colonizer of the nasal epithelium of one-third of the general population3, Staphylococcus aureus might become a dangerous life-threatening pathogen when it defeats host immune system, crosses the epithelial barrier and get access to deeper tissues like blood, dermis, gastrointestinal tract, heart valves or bones3,4. This biological versatility is based on a highly orchestrated regulation of circuits that sense a plethora of environmental signals and modulate gene expression for fine tuning crucial traits like cell-wall structure, biofilm formation or resistance to antibiotics. The core feature of such circuits is the two-component-signaling transduction system (TCS)5, which is actually one of the most conserved and effective mechanisms in nature for coupling external stimuli and gene expression. In its most basic form, a canonical TCS normally consists of a membrane-bound histidine kinase and a cytosolic response regulator that, once phosphorylated, elicit appropriate changes in the cell by regulating gene expression, protein interactions, or enzymatic activity6.

Over the last decade, the scientific community has gained in-depth knowledge of the genes affected by specific staphylococcal TCSs, giving rise to a vast body of bibliography and information about mutants in the respective sensor kinases, response regulators and auxiliary genes. Thus, the pivotal role of AgrCA and SaeRS TCSs on virulence gene expression or the involvement of BraRS and GraXRS in antibiotic resistance has been studied in great detail7,8,9,10. Furthermore, since TCSs are a matter of life and death to bacteria and they are not present in host’s cells, these regulatory pathways have always been listed as promising antibacterial targets. Precisely based on the assumption that solution to present-day therapeutic limitations might somehow lie on impairing the way Staphylococcus senses and integrates environmental stimuli, significant effort in the form of ambitious High Throughput Screenings (HTSs) and Structure-Based Virtual Screenings (SBVSs) has been done to find new molecules with inhibitory effects on staphylococcal sensor kinases11. However, with few exceptions such as the molecule named walkmycin B12, biochemical screens normally identify a high number of compounds acting through nonspecific inhibitory mechanisms and thus render non-viable drugs in terms of clinical application11,13,14.

With the aim of designing a whole-cell drug discovery tool that could complement in silico docking and crystallographic analysis of kinase-ligands structure in TCS-targeting approaches, we decided to explore the potential of a recently developed staphylococcal strain that lacks the whole non-essential TCS machinery (∆XV strain)15. Among all the TCSs whose individual contribution to a specific S. aureus phenotype has been defined, and based on its overall responsibility for resistance to host defenses like polymorphonuclear cells (PMNs) or cationic antimicrobial peptides, we selected GraXRS as a candidate of therapeutic target9,16,17,18,19. Additional evidence supporting our choice was given by a recent work showing that S. aureus uses this regulatory system to sense and adapt to the acidified phagolysosome in macrophages20, but also by several previous studies unveiling the potential of GraXRS to impact the bacterial capacity to colonize and survive on aortic valves in a rabbit endocarditis infection model18 or to play a crucial function in a murine model of systemic infection16. Most of these preceding articles conclude that mechanisms underlying GraXRS activity are related with changes in bacterial surface charge via its target downstream gene mprF and the operon dltABCD9,17,18,21.

In this study, we have used the S. aureus strain deprived of fifteen TCS15 and isogenic derivatives containing exclusively the GraXRS TCS as a whole-cell platform to identify drugs that specifically target this signaling pathway. Upon evaluating the GraXRS-blocking activity of 1280 FDA-approved off-patent drugs, we found that molecules with antioxidant activity as acetylsalicylic acid, ascorbic acid, the porphyrin derivative verteporfin, or the flavonoid hesperidin, are capable of inhibiting the activity of GraXRS-dependent promoters. Among all the compounds, only verteporfin made a significant contribution to the susceptibility of S. aureus to human PMNs-mediated killing and rendered lower levels of bacterial colonization when its effect was assessed using an in vivo murine model. Though further analysis is needed to fully understand the precise molecular targets of verteporfin, data presented in this work suggest that the redox-active cysteine of GraS is required for this molecule to exert its inhibiting effect. Altogether, our results enlighten the potential of verteporfin as a supplement and(or) alternative antimicrobial therapy and provide evidence that this compound could be included in a recently described category of drugs known as “Potential Drugs for Repurposing against Infectious Agents”.

Results

Design and validation of a GraXRS-focused screening platform

With the aim of designing a highly specific GraXRS-targeting screening assay, we first restored the GraXRS TCS into the chromosome of S. aureus ΔXV strain, which only contains the essential WalKS TCS system in its genome. The resulting S. aureus ΔXV Gra-RES strain, together with the corresponding S. aureus MW2 wild type strain and the single GraXRS mutant derivative, were transformed with two different reporter plasmids in which lacZ expression depends on the GraXRS-regulated promoters of mprF and dltX9. As shown in Fig. 1, transcriptional activity of reporter genes was barely detectable in the GraXRS deficient strains (single and multiple ΔXV mutants), whereas mprF and dltX-based reporter constructs were highly induced in the GraXRS containing strains (wild type and ΔXV GraRES). As an additional test for evaluating the behavior of the reporter strain-set, positive response of the GraXRS-dependent promoters to the presence of sublethal concentrations of colistin was also analyzed. Noticeably, and confirming the concept of TCS as self-sufficient modules previously envisioned15, lacZ expression driven by dltX promoter was essentially equal in the wild type and graXRS restored genetic backgrounds. This result validates both the use of the GraXRS restored strain and dltXP::lacZ transcriptional fusion for high-throughput-screening designs aimed at the discovery of GraXRS-blocking drugs and molecules.

Figure 1
figure 1

Validation of GraXRS-dependent transcriptional fusions. Graphic illustration of GraXRS-dependent activity of dltX (A) and mprF (B) promoters in S. aureus MW2 wild type, ∆XV, ∆XV Gra-RES and ∆graRS strains. Transcriptional activity in TSB medium (light gray bars) and TSB supplemented with colistin 50 μg/ml (dark grey bars) is presented. Means and standard deviations values are shown from at least three independent experiments. A visual example showing different degrees of GraXRS activity [∆XV (1), ∆XV Gra-RES (2) and ∆XV Gra-RES in the presence of colistin (3)] in the 96-well format is also shown.

Full size image

Primary and secondary screenings

From the initial screening of the 1280 drugs included in the Prestwick library (https://www.prestwickchemical.com/libraries-screening-lib-pcl.html), we selected 77 compounds that led to reduced activity of dltX promoter by more than 55% as determined by beta-galactosidase activity (see table in the supplemental material). Because the intention of this work was to repurpose those FDA-approved drugs that specifically targeted GraXRS-mediated signaling pathway, and this TCS has been shown to be crucial for bacterial growth under acidic conditions15, the capacity of the selected compounds to affect OD600 values at pH 5.5 was tested, expecting a significant growth arrest in the presence of GraXRS-blocking compounds. Following a similar approach, the dose–response behavior of selected compounds was evaluated. To do so, both bacterial growth and dltxP transcriptional activity were quantitatively assessed in the presence of variable concentrations, ranging from 0 to 20 μM, of selected drugs (Fig. 2C).

Figure 2
figure 2

Screening results. (A) Graphic illustration of GraXRS-dependent activity of dltX promoter in the presence of selected drugs (10 μM) in S. aureus ∆XV Gra-RES (light gray bars) and MW2 wild type strains (dark grey bars). Data are presented as relative M.U. values obtained in the absence of any compound for S. aureus ∆XV Gra-RES and MW2 wild type strains respectively. Values exceeding 100% correspond to those cases where the observed transcriptional activity is lower to that showed by ∆XV and/or bacterial growth is inhibited. Standard deviations values correspond to at least three independent experiments. Results regarding additional selection criteria including the repression of the GraXRS-dependent alternative mprF promoter, significative inhibition of wild type strain under acidic growth conditions (pH 5.5) and transcriptional co-inhibition of the SaeRS-dependent sec4 promoter are presented. (B) Structure of selected drugs included in drug bank profiles (https://www.drugbank.ca) is shown. (C) Graphic illustration of Dose Dependency profiles in terms of dlxP inhibition exhibited by selected compounds. Data are presented as relative M.U. values obtained in the absence of any compound for S. aureus ∆XV Gra-RES (dark grey bars). Relative percentage of survival in the presence of increasing doses of compounds is also included (light grey bars).

Full size image

In order to reinforce the involvement of GraXRS TCS in the phenotypic outcomes rendered by the drugs to be chosen, their ability to down regulate the GraXRS-dependent alternative mprF promoter was considered as an additional selective criterion.

To verify the specificity of GraXRS for the selected compounds, we restored a different TCS, saeRS, into the ∆XV genome. In this case, the reporter lacZ gene was transcriptionally fused to sec4 promoter, which has previously described as part of SaeRS regulon22. No shift in the transcription levels of sec4P::lacZ fusion appeared when ∆XV Sae-RES were incubated in the presence of each selected compound.

Finally, and though equivalency between wild type and GraXRS restored strains in terms of GraXRS-mediated sensing had been previously validated, the possibility that other TCSs or derived circuits absent on ∆XV Gra-RES could somehow affect dltX transcriptional activity was considered. In order to address this issue, the MW2 wild type strain was transformed with the dltX-derived reporter plasmid and beta-galactosidase activity was measured in the absence and presence of mentioned compounds.

As shown in Fig. 2, the whole consecutive screening process significantly restricted the number of active compounds from almost 80 to 5. Acetylsalicylic acid (ASA), hesperidin (HES), ascorbic acid (VITC), verteporfin (VER) and troglitazone (TGZ) were capable of inhibiting bacterial growth under acidic conditions, affected mprFP activity negatively and exerted a suppressing effect on dltXP in the wild type genetic background. Among this final group of drugs, TGZ displayed an additional negative impact on SaeRS, and thus was firstly considered as a potential multi-target drug. However, since this molecule was withdrawn in 2000 due to high risk of hepatotoxicity, only ASA, HES, VITC and VER were contemplated as real candidates for therapeutic reposition. Curiously, though these four compounds show no common (known) pharmacological features, they are all chemically classified as redox-active drugs.

In vitro effect of selected compounds: phagocytosis and killing of S. aureus by human PMNs

Deepening into the anti-virulence potential of selected drugs, we next proceeded with an in vitro assay in which the bacterial susceptibility to phagocytosis and killing by human polymorphonuclear cells was assessed. Considering that GraXRS has been shown to have a pivotal role for S. aureus to resist PMN attack20, we isolated this cellular fraction from peripheral human blood and the effect of ASA, HES, VITC and VER on MW2 susceptibility to these immune cells was assessed. In all cases the process was boosted by opsonic antibodies naturally present in human sera due to unavoidable exposure to S. aureus. After incubating PMNs-S. aureus MW2 suspensions for 30 min in the absence or presence of the four selected compounds (5 μM each), removal of extracellular bacteria via gentamicin exposure, and subsequent lysis of eukaryotic cells at basic pH, bacterial viability was estimated via plate counting. Data shown in Fig. 3 prompted us to conclude that the presence of ASA, HES and VITC caused a slight increase in the sensitivity to PMN attack, while the effect of VER on reducing the number of surviving intracellular bacteria was substantially obvious and statistically significant (Fig. 3). As a result, from this in vitro approach, we chose VER as the sole candidate to proceed with the next assays.

Figure 3
figure 3

Effect of selected drugs in the susceptibility to phagocytosis and killing by human PMNs. Graphical scheme of bacterial counts expressed as log values after incubating PMNs and S. aureus MW2 wild type suspensions for 30 min in the presence or absence of four selected compounds (5 μM each), subsequent gentamicin treatments and lysis of eukaryotic cells (see “Materials and methods” section). Means and standard deviations are presented, illustrating the bactericidal effect of PMNs and the additional contribution of ASA, VITC, VER, TIR, DIA and HES. In the case of VER, two asterisks denote an associated p value of 0.006 when ANOVA and Tukey’s pairwise post hoc tests were applied according to prior normality and homoscedasticity tests.

Full size image

Effect of verteporfin in a murine model of wound infection

Following on from the in vitro testing, we next evaluated VER using the murine model of wound (or surgical site) infection. In this in vivo approach, a silk suture contaminated with S. aureus MW2 or MW2ΔgraXRS strain (4.5 × 105 cfus cm−1) was used for sewing up a previous incision on the back of the mice. VER was topically applied 2 and 24 h after suture implantation using hydrogel-based formulations that contained either no active ingredient or the porphyrin under study at a low (0.125 mg/kg) or a high (2.5 mg/kg) dose. Assessment of the infection was performed by counting viable bacteria in tissue homogenates that were obtained upon animal euthanasia, 24 h after the last treatment.

As shown in Fig. 4, the significantly lower quantity of viable bacteria present in tissue samples that had been infected with the GraXRS deficient strain unveiled the critical role of this TCS in surgical wound infections. When the effect of VER was assessed, data showed that topical administration of this drug significantly reduced the bacterial load in a dose-dependent manner. Noticeably, application of VER at a low dose led to a similar degree of bacterial colonization to that followed by the implantation of sutures that had been contaminated with the GraXRS lacking strain. While such a result reinforces the potential of VER as a GraXRS inhibitory drug, its specificity is suggested by the fact that the observed outcome after treatment of wounds infected with the wild type strain is certainly evident but it becomes almost imperceptible in the case of ∆graXRS-associated infection. Equally certain, however, is that high doses of VER could exert a GraXRS-independent and/or antibiotic effect.

Figure 4
figure 4

Effect of VER in a murine model of surgical wound infection. BOX-PLOT illustration (PAST statistical package) of bacterial counts expressed as log values per gram of wound tissue homogenates. Different treatments include the topical administration of hydrogel-based formulations with two doses (0.125 and 2.5 mg kg−1) of VER, 2 and 24 h after making an incision that was immediately infected with a surgical suture contaminated with 4 × 105 CFUs of S. aureus MW2 wild type or ∆graXRS strain. Hydrogel without any active pharmaceutical ingredient was applied to control mice. All the animals were euthanized 24 h after the last treatment. Apart from visual interpretation in accordance to IQR overlapping, data were statistically analyzed via Kruskal–Wallis and post hoc Mann–Whitney test. Corrected p-values corresponding to relevant post hoc statistical comparisons are also shown.

Full size image

Exploring the mechanistic basis of verteporfin

Because VER is a tetradentate chelating porphyrin23 that might be involved in redox sensing, just like heme complex, and such molecules are normally sensed through thiol-based switches, we also analyzed the contribution of the single redox-active cysteine present in GraS, C227, to VER effect24. To do so, C227 was mutated to S or A in ∆XV Gra-RES background and transcriptional activity of dltXP in the resulting ∆XV Gra-RES S(C227-S) and ∆XV Gra-RES S(C227-A) strains was measured. An additional strain in which GraS H129 amino acid, the residue that undergoes phosphorylation upon activation of the kinase, had been mutated to Q was also constructed and included as a reference of complete GraXRS inactivation. As shown in Fig. 5A, data verified that replacement of cysteine by another residue had a negative impact on GraXRS activity, being such an outcome dependent on the polarity of the substituted amino acid. Thus, C227-A (non-polar) GraS isoform led to a lower degree of transcriptional activity of dltXP in comparison to that showed by the isoform in which C227 had been mutated to the polar amino acid serine. In accordance with this observation, bacterial growth arrest under acidic conditions, a phenotype that strictly reflects GraXRS status, also showed dependence on the mutation polarity (Fig. 5B). When the effect of VER was assessed, transcriptional data revealed that GraXRS repression led by mutations was far from being as drastic as the one achieved by VER (Fig. 5A), discarding the possibility of considering the C227 redox switch as the exclusive mechanism underlying VER effect. However, impairment of the signaling via C227 resulted in the insensitivity to VER, fact that suggests that intermolecular cysteine-disulfide-bond formation is required, though not entirely, for VER to have a blocking GraXRS-dependent effect.

Figure 5
figure 5

C227-mediated redox switch involvement in the sensitizing effect exerted by VER. (A) Transcriptional activity of GraXRS-dependent dltX promoter in S. aureus ∆XV Gra-RES, ∆XV Gra-RES S(C227-S), ∆XV Gra-RES S(C227-A) and ∆XV Gra-RES S(H129-Q) strains is illustrated both in the presence (dark grey bars) or absence (light gray bars) of VER (10 μM). Means and standard deviations values are shown form at least three independent experiments. (B) Pictures illustrating the capacity of serially-diluted suspensions of bacterial strains S. aureus MW2 wild type, ∆XV, ∆graRS, ∆XV Gra-RES, ∆XV Gra-RES S(C227-S), ∆XV Gra-RES S(C227-A) and ∆XV Gra-RES S(H129-Q) to grow in neutral (left) and acidified (right) TSA medium are also included in this figure.

Full size image

Discussion

Current strategies aimed at antimicrobial discovery prioritize innovative concepts like alternative molecular mechanisms of action, new natural product sources, pro-drugs, or even approved compounds that were originally intended for other therapeutic indications, as it is the case of the present work (Statement of Antimicrobial drug discovery, EASAC-201425). Particularly speaking of S. aureus, and discarding the essential TCS WalKR/S as the antibiotic target per excelence, continuous efforts are being made toward the discovery of inhibitors of TCS involved in virulence and biofilm formation (for review26,27,28). Impairment of these non-essential biological pathways has the advantage of requiring a lower plasmatic dose compared with a MIC, reducing the tendency to resistance and minimizing side effects on neutral and beneficial microbiota that colonize treated human or animal hosts29. Up to date, several novel and previously approved drugs with the capacity to exert an inhibitory effect on Agr, SaeRS, and ArlRS TCSs have been described (for review26,27,28) but, to our knowledge, GraXRS had never been considered as a target for disarming S. aureus. The involvement of this TCS in the resistance to antimicrobial peptides and macrophages had already been envisioned in several occasions9,30,31, but it has just been recently proven that GraXRS is entirely responsible for the response to pH inside acidified macrophage phagolysosomes15,20. These premises led us to consider GraXRS as a clear target to counteract S. aureus response to innate host immunity and impede replication of the pathogen in the acute stage of systemic infection.

Taking advantage of the S. aureus strain that lacks its complete sensorial TCS network15, we developed a series of reporter strains that could be helpful for selecting compounds capable of blocking transcriptional activity of the GraXRS-dependent dltX promoter. Working with ∆XV Gra-RES strain gave us the opportunity to perform a bioassay where bacterial sensing entirely depends on GraXRS (and WalkRS), thus reducing the probability of selecting off-target drug candidates. Proof of this last claim is the fact that among selected drugs, only one of them (Troglitazone) displayed a GraXRS-SaeRS multi target effect.

At the time of deciding the type of compound to be tested for the identification of GraXRS inhibitors, we, as many other researchers, opted for the drug-repurposing approach. This strategy is based on the identification of “off” antimicrobial targets for drugs that were approved for other clinical diseases29, hence bypassing the financial and regulatory barriers that have to be overcome to bring a drug to market. At present day, this concept of repurposing has gained renewed interest and a novel category of drugs known as “Potential Drugs for Repurposing against Infectious Agents” is exponentially thriving32 By way of example, the old antimalarial drug chloroquine is being tested as SARS-CoV-2 inhibitor33. Though the precise mechanistic basis of their effect remains to be completely elucidated, current candidate PDRIAs targeting S. aureus SaeRS and/or AgrTCSs are floxuridine, streptozotocin and diflunisal28,34,35.

The screening methodology applied here consisted in the analysis of changes in dltXP transcriptional activity, followed by several consecutive steps where additional criteria like the effect on an alternative GraXRS-dependent promoter, TCS-selectivity, or the determination of bacterial growth in the presence of selected drugs under acidic conditions were applied. The overall process ended up in the selection of five candidate drugs: acetylsalicylic acid (ASA), hesperidin (HES), ascorbic acid (VITC), verteporfin (VER) and troglitazone (TGZ). Curiously, all compounds are classified as redox-active drugs. While ASA, HES, VITC and TGZ are commonly sorted as antioxidant molecules, VER can induce oxidative stress through the production of free radicals or be alternatively combined with soluble metals and display a redox potential similar to that showed by the heme complex23. These observations are in agreement with the previously unveiled connection between GraXRS and oxidative stress, evidenced by the deciphering of the GraXRS regulon and the proved essentiality of this TCS in staphylococcal resistance to redox compounds like paraquat or H2029. Furthermore, a recent RNA-seq transcriptomic approach has just corroborated the involvement of GraXRS, collectively with VraSR, SaeRS, MgrA, SigB or Fur, in the cell response to thiol-oxidative stress36,37.

After proving that VER was the only compound really capable of sensitizing bacteria against the effect of human PMNs, this compound was further examined using a murine model of surgical wound infection. This model has been previously used to assess the effect of systemic and topical antimicrobial agents, finding a close correlation with efficacy in clinical trials with human subjects. Noticeably, our results provided further evidence concerning the critical role of GraXRS in skin and wound infections and insinuated a pharmaceutical potential of VER as a novel local treatment for S. aureus infections. Since bacterial count after the infection with the wild type strain and subsequent treatment with 0.125 mg kg−1 of VER was quite similar to that proceeding from the infection with a GraXRS negative strain, the effect exerted by the porphyrin derivative seems to be highly dependent on the activity of this TCS.

When the possible path(s) of how VER inhibits the activity of GraXRS was envisaged, we thought of VER as a heme-like porphyrin capable of binding iron in different oxidation states23. Recent transcriptomic studies conducted with the constitutively-active forms of staphylococcal kinases have unveiled the involvement of GraXRS in the regulation of heme synthase A (MW_RS05355;38), fact that led us to attach importance to the chemical mimicry between both molecules. Since porphyrins are normally involved in thiol-based molecular switches24, the unique potentially redox-active residue in GraS, C227, was considered a potential molecular target of VER. To address this question, the impact of C227 mutations to S and A on GraXRS activity and VER sensitivity was assessed. To our knowledge, the results achieved in this work by punctual mutations of C227 have evidenced for the first time that this cytosol-located redox-active residue actually participates in GraS kinase activity. This line of reasoning, which has support from studies that have characterized other bacterial kinases like AcrB39, showed that GraS degree of silencing depended on the polar nature of the amino acid that substituted C227 and suggested the involvement of this redox-switch as a potential molecular target concerning VER effect. However, and though C227 substitutions led to insensitivity to VER, the inhibitory outcome yielded by mutations was not as drastic as the one achieved by exposure to VER, suggesting that additional molecular paths must be involved in this process. An additional candidate that might be considered is Stk1, the unique serine-threonine kinase which cross-phosphorylates GraR40 and shows homology with OXR1, one of the recently discovered mammalian target of VER that, curiously, is also related to oxidative stress41. However, in accordance with Fridman et al.40, we have verified that both Stk1 and GraS-mediated phosphorylation on GraR as T128, T130, T149 and D51 respectively are equally required for full dltX expression (data not shown), fact that seriously complicates the use of our reporter systems when it comes to holding Stk1 accountable for intervening on VER effect.

Could VER be considered a viable antimicrobial candidate? Though we are fully aware that further in vitro and in silico studies helping to understand the whole molecular scenario underlaying VER effect and alternative in vivo approaches or definition of strategic dosages are some pending issues to claim a novel anti-virulence pharmaceutical indication for VER, we are convinced that this drug presents some strengths, apart from those inherent to anti-virulence drugs, that might be worth considering. In terms of pharmacology, for instance, VER excretion is dependent on hepatic function, while many antibiotics primarily undergo elimination via kidney filtration. Thus, VER could be administered to patients suffering from kidney disease, or to those who were being concomitantly treated with antibiotics that are prone to cause nephrotoxicity (e.g. vancomycin)42. On the other hand, though side effects associated to parenteral administration of VER include hypersensitivity reactions or blood pressure alteration, these symptoms have lower mean severity ratings in comparison to those showed by many antibiotics (ema.europa.eu/Find medicine/Human medicines/European public assessment reports). Finally, taking into account that we have also observed that VER-containing topical formulations are effective and that further studies might confirm that chemical mimicry between porphyrin derivatives and heme group could actually be harnessed for disarming S. aureus, our results may be considered as a step forward in re-proposing VER as a plausible alternative in combating antimicrobial resistance.

Materials and methods

Bacterial strains, plasmids, oligonucleotides and culture media

Bacterial strains, plasmids and oligonucleotides (purchased to IDT) are listed in Tables 1, 2 and 3 respectively. Escherichia coli XL1blue strain was grown in LB broth and LB agar. Staphylococcus aureus strains were grown on trypticase soy broth (TSB), trypticase soy agar (TSA), trypticase soy broth with 0.2% of glucose (TSB-glu), Mueller Hinton (MH) and B2 medium [1% casein hydrolysate, 2.5% yeast extract, 2.5% NaCl, 0.1% K2HPO4, and 0.5% glucose (w/v)]. When required for growth or selection, medium was supplemented with 5-Bromo-4-Chloro-3-Indolyl β-d-Galactopyranoside (XGal) and/or the appropriate antibiotic at the following concentrations: erythromycin (Eri) 10 μg ml−1, ampicillin (Am) 100 μg ml−1, chloramphenicol (Clo) 10 μg ml−1 and 20 μg ml−1.

Table 1 Strains used in this study.
Full size table
Table 2 Plasmid used in this study.
Full size table
Table 3 Oligonucleotides used in this study.
Full size table

DNA manipulations

Plasmids were purified using NucleoSpin Plasmid kit (Macherey–Nagel) according to the protocol provided by the manufacturer. PCR fragments and enzymatic reactions were purified using GeneJET Gel Extraction and DNA Cleanup Micro Kit (Thermo Scientific). FastDigest restriction enzymes, Rapid DNA ligation kit, Dreamtaq DNA polymerase and Phusion DNA polymerase were supplied by Thermo Scientific and used according to provided instructions. Sequence verification of PCR-amplified products and plasmid constructions was performed by Stab Vida. Transformation of Staphylococcus aureus was performed following previously standardized protocols43,44.

Allelic exchange of chromosomal genes

To generate markerless deletions, two fragments of at least 500 bp that flanked upstream (primers A and B, Table 3) and downstream fragments (primers C and D, Table 3) of the region to be deleted were amplified by PCR. Amplified products were digested using the corresponding restriction enzymes (Table 3), purified and cloned by ligation into pMAD shuttle vector. To restore individual TCSs into ∆XV chromosome, a fragment containing the two flanking regions used to generate the deletion15 and the original TCS sequence were amplified by PCR using chromosomal DNA from MW2 strain as template and oligonucleotide pair gra or sae 1–3. For restoration of the TCS with single amino acid substitutions at C227, graRS was amplified using MW2 chromosomal DNA as template and a two-step PCR protocol. First, the oligonucleotides A and G or H were used for generating two overlapping PCR products, while a second amplification step with A and D oligonucleotides using both purified PCR products as templates generated graRS S(C227-A) and graRS S(C227-S) isoforms. Such DNA fragments were purified, digested with the corresponding enzymes (see Table 3) and inserted by ligation into the pMAD shuttle vector45. Homologous recombination experiments were performed as previously described46. Final plasmidless erythromycin sensitive white colonies were tested by PCR using primers E and F (Table 3).

Reporter plasmid construction

pSA14 was used for the construction of different reporter plasmids. Promoter regions of mprF, dltX and sec4 were amplified by PCR using oligonucleotides described in Table 3. PCR fragments were purified and cloned into pSA14 through restriction enzymes to generate transcriptional fusion with lacZ. GraRS-dependent reporter plasmids were transformed via electroporation into S. aureus MW2, MW2ΔgraXRS, ΔX, ΔXV Gra-RES and ΔXV Gra-RESΔgraX, while plasmid harboring Sae-dependent sec4 promoter was inserted into MW2, ΔXV and ΔXV Sae-RES strains. To analyze mprF, dltX and sec4 expression, ON cultures were chemically lysed beta-galactosidase activity was measured.

High throughput beta-galactosidase-based screening

Screening of the 1280 off-patent FDA-approved drugs in Prestwick Chemical drug library (Prestwick Chemical) was based in a method for beta-galactosidase assays in 96 well plates. Working solutions of the compounds were prepared in 50 μl of sterile distilled-deionized water at a concentration of 20 μM, and combined with 50 μl of a 1:30 dilution on 2 × TSB medium of an overnight (ON) culture of S. aureus strains, thus generating 100 μl of 1:60 cell dilution on 1 × TSB at a final concentration of 10 μM of each drug. Plates were incubated during 24 h at 37 °C and, upon incubation, OD600nm was measured (Multiskan Go; Thermo Scientific). Bacterial cells were subsequently lysed by the addition of 100 μl well−1 of Z buffer supplemented with lysostaphin (0.5 mg ml−1) during 2 h at 37 °C; Next, 30 μl well−1 of Ortho-Nitrophenyl-beta-galactoside (ONPG, 4 mg ml-1) was added and, when required, the reaction was stopped with 100 μl well-1 of 1 M Na2CO3. OD420 and OD550 values were finally recorded for Miller Units calculation. Untreated reporter ∆XV-GraRES and ∆XV strains were included in every plate as internal controls. Experiments were carried out in triplicate.

Phagocytosis and killing of S. aureus by human PMNs

Phagocytosis and killing of S. aureus by human neutrophils in presence of selected compounds was determined as described before47. Polymorphonuclear cells (PMNs) were isolated from healthy human heparinized-defibrinated blood (Seralab Logistics) using Ficoll-Plaque PREMIUM (GE-Healthcare) according to manufacturer’s protocol and resuspended at a final concentration of 1 × 107 PMNs ml−1 in HBSS supplemented with human serum. S. aureus strains were cultured to the early stationary phase and 10 ml of culture were centrifuged, washed twice with sterile PBS and resuspended in Hank’s Balanced Salt Solution (HBSS) supplemented with human serum at a final concentration of 4 × 105 bacteria ml−1. Finally, 0.2 ml of PMNs solution was mixed with 0.2 ml of S. aureus solution and 600 μl of HBSS supplemented with human serum. Compounds were added at a final concentration of 5 μM. After incubation at 37 °C for 30 min, each sample was treated with gentamicin 100 mg ml−1 and then 100 μl of each mixture were added to 1 ml of pH 11 solution. Finally, serial dilutions were plated on TSA to determine the number of colony-forming units (CFU) in presence of the different compounds. All data were referred to initial CFU number.

Mouse infection model

The experimental animal study was reviewed and approved by the “Comité de Ética, Experimentación Animal y Bioseguridad” of the Universidad de Navarra-Centro de Investigación Médica Aplicada (CIMA). Work was carried out at the CIMA animal facility under the principles and guidelines described in the “European Directive 86/609/EEC” for the protection of animals used for experimental purposes. Six-week-old female Swiss mice (20–25 g) were obtained from ENVIGO and confined in groups of 6 animals.

The model was performed as previously described50. Briefly, 10 cm fragments of commercial brailed silk (TC-15, Lorca Marín) were contaminated with 4 × 106 CFU cm−1 of S. aureus MW2 or S. aureus MW2ΔgraXRS strains by immersion during 30 min. Fragments were then blotted- dried. One day prior to the experiment, the interscapular skin was shaved using a sharp razor. On the day of the infection, superficial wounds were produced on the exposed back surface though a longitudinal midline incision of 2 cm approximately. The skin of either side of the incision was retracted, and the wound was infected by stitching it with contaminated suture and a suturing needle. Wounds were topically treated 1 h and 8 h after infection with approximately 100 μl of hydrogel formulations containing 0.125 and 2.5 mg/100 μl. The hydrogel base without any active substance was applied in the control group. Treatments were repeated 24 h after infection and mice were sacrificed 24 h after the last application. Finally, the wounded tissue was resected, homogenized in PBS, and dilution series of homogenates was plated on TSB agar for enumeration of CFU (output). After an overnight incubation at 37 °C, CFU gr of tissue-1 were calculated and expressed as log10.

Statistical analysis

Data generated by PMN-mediated killing assay were compared using ANOVA, applying Tukey’s pairwise as post hoc test. Data obtained from the bacterial counts in the murine model were treated and compared using Kruskal–Wallis test, Mann–Whitney pairwise and Dunn’s post hoc tests. All tests were two-sided, and the significance level was 5%. The statistical analysis was performed with Past and R softwares.