We screened several compounds that interact with C5aR or CD45 to attenuate PVL-induced cytotoxicity and inflammasome activation in vitro.

Of the tested compounds, avacopan, PMX205 and W-54,011 were able to mitigate the PVL-induced cytotoxicity on PMNs and inflammasome activation in monocytes. Avacopan is approved for the treatment of ANCA-associated vasculitis and is contra-indicated in patients with serious infections [10]. Thus, safety issues in patients with PVL-positive S. aureus infections need to be addressed first before this compound can be considered as a virulence-modifying adjuvant to antimicrobial therapy in future trials. In addition, the Maximum Recommended Human Dose (MRHD) of avacopan is 349 ng/ml (0.6 µM) which is less than the lowest concentration that had a significant effect on PVL attenuation (1 µM, Fig. 1; Table 1). PMX205 is an anti-inflammatory drug currently tested in a phase 1 dose escalation trial (ACTRN12619001639112 on www.anzctr.org.au); data on the plasma pharmacokinetics after a single dose of PMX205 (0.02 mg/kg), including a dose escalation to up to 0.4 mg/kg are pending. Similarly, and to the best of our knowledge, W-54,011 has not been tested yet in clinical trials. We therefore cannot assess if the concentrations of PMX205 or W-54,011 that significantly reduced the PVL-induced cytotoxicity in PMNs can be safely achieved in humans.

We observed that low concentrations (1–100 nM) of some compounds had a paradoxical effect: PVL cytotoxicity on PMNs was slightly enhanced, particularly within the range of 0.5 to 2 nM PVL, and even for competitors with a protective effect at higher concentrations (Fig. 1). It has previously been reported that low concentrations of PVL induce apoptosis rather than necrosis [11], by binding of the LukS-PV subunit to the C5a receptor [12]. It could be speculated that binding of the (ant)agonists to C5aR influences the two modes of action of PVL on neutrophils in different ways: induction of apoptosis is stimulated while pore formation and subsequent necrosis are inhibited.

Although all C5aR (ant)agonists tested in our study (avacopan, PMX205, W-54011, DF2593A, JPE-1375, BM213) have been described as potent C5a competitors [13, 14] they showed converse effects on PVL-induced cytotoxicity on PMNs and inflammatory cytokine response in monocytes. The observed differences could be related to the specific binding affinities and mechanisms exhibited by the compounds.

Binding of the LukS subunit to the C5a receptor has been reported with a dissociation constant (Kd) of 127 nM [3]. The binding affinity (reported as inhibition constant Ki) of avacopan (0.1 nM) and W54011 (2.2 nM) was comparable to that of C5a (0.57 nM), whereas PMX205 (220 nM) and JPE-1375 (111 nM) were much lower [13]. Although two different measures of binding (Kd and Ki) were used, they can be compared to some extent. In agreement with our results, avacopan and W54011 have a high binding affinity to C5aR, most likely higher than that of LukS. However, PMX205 has a relatively low affinity but was able to compete with PVL in our experiments. This could be explained by the different modes of interaction of the competitors with the C5aR.

The non-peptide antagonists avacopan and W-54,011 interact with an allosteric site on C5aR, formed by residues of the transmembrane domains and may thereby stabilize the inactive receptor state [15]. DF2593A is an additional non-peptide, allosteric inhibitor, but targets a different allosteric pocket of C5aR [15]. PMX205 is an analogue of PMX53, a cyclic peptidomimetic antagonist that mimics the C-terminal structure of C5a, allowing it to bind to the orthosteric site of C5aR and compete directly with the ligand [13, 15]. Similarly, JPE-1375 and BM213 target the orthosteric binding pocket of C5aR, with JPE-1375 being another but linear analogue of PMX53 and BM213 being a linear C5aR1-selective agonist [13, 16]. Despite targeting similar sites to exert their orthosteric action, only PMX205 and not JPE-1375 and BM213 antagonized PVL-induced cytotoxicity.

Variable modes of ligand-receptor interaction may differently compete for binding with PVL or stabilize C5aR1 in different conformations, interfering with subsequent pore-formation. In agreement, W-54,011 has been shown to decrease binding of PVL to C5aR and reduce its activity on neutrophils, while it potentiated that of a different bicomponent pore-forming toxin HlgC/HlgB [17].

Strikingly, the converse effect of the different C5aR (ant)agonists on PVL-induced PMN cytoxicity was even more evident in the monocyte immunomodulation assay. Avacopan, PMX205 and W-54,011 inhibited the release of pro-inflammatory IL-1β, while DF2593A, JPE-1375 and BM213 even synergistically enhanced the inflammasomal response (Fig. 2). Interestingly, JPE-1375 and BM213 previously showed immunomodulatory effects and their orthosteric competition with C5a was stronger than allosteric competition by the non-peptide antagonists avacopan and W-54,011 [13]. However, in a Neisseria meningitidis infection study, pharmacological blockade of C5aR by PMX205 as well as W-54,011 reduced the inflammatory cytokine release and enhanced mouse survival in a sepsis model [18].

We observed an increase in PVL toxicity when PMNs were treated with the CD45 allosteric inhibitor NQ301. It is unclear whether this is mediated by a direct effect on the interaction of CD45 with the LukF-PV subunit [4] or indirectly affecting C5aR.

In the absence of detailed structural knowledge of the binding of PVL to C5aR/CD45 at the molecular level, the reason for the differential competition with PVL displayed by various C5aR or CD45 (ant)agonists remains elusive. Nevertheless, our study demonstrates that small compounds targeting the cellular receptor of PVL provide a rationale for therapeutic intervention and the development of adjunctive treatments for PVL-related infections.

Our study has several limitations. First, the experiments were conducted with a small group of participants. Although the findings provide valuable insights, the small sample size may limit the generalizability of the results to a broader population. Given the interpersonal variance in the effect of PVL on leukocytes, we consider this limitation to be relevant. However, our approach of testing the use of C5aR antagonists against PVL-induced cytotoxicity was designed to provide a qualitative rather than a quantitative answer. Even with a small donor sample size, some compounds could be identified as beneficial or detrimental. Second, the isolated cell system used in this study may not account for potential influences of host-pathogen interactions, including other components of the immune system and conditions at sites of infection. The extent to which our in vitro results on PVL competition translate to in vivo infection with PVL-positive S. aureus needs to be evaluated in appropriate animal models, e.g. rabbits. Third, we did not test our recombinant PVL for LPS contamination. The PVL used in our experiments was produced in one batch and pooled, with aliquots stored at -80 °C to ensure comparability of experiments. The PVL-specific inhibition demonstrated in our study suggests that the observed effects must be C5aR specific. Lastly, our in vitro setting may not accurately reflect physiological concentrations of PVL [19] and recommended doses of inhibitor compounds in plasma. Further studies are needed to quantify the amount of PVL at sites of infection (e.g. abscesses) and whether compounds can be applied locally in adjuvant treatment.

In conclusion, C5aR-antagonizing compounds, particularly avacopan, are promising candidates to repurpose drugs in the treatment of PVL-related infections. Further in vivo experiments using suitable animal models (e.g. rabbit or humanized mouse model) are imperative to validate the efficacy of these compounds. Whether they can be repurposed as adjuvants in combination with standard antimicrobial therapies needs to be tested in clinical trials.