Candida albicans Augments Staphylococcus aureus Virulence by Engaging the Staphylococcal agr Quorum Sensing System

Relatively little is known about the complex interactions and signaling events that occur between microbes and even less so about how microbial “cross talk” shapes human health and disease. Candida albicans (a fungus) and Staphylococcus aureus (a bacterium) are formidable human nosocomial pathogens, causing severe morbidity and mortality. Moreover, they are frequently coisolated from central venous catheters and deep-seated infections, including intra-abdominal sepsis. In this work, we have shown that coinfection with C. albicans and S. aureus is highly lethal, leading to >80% mortality by day 1 postinfection, whereas monoinfection with C. albicans or S. aureus does not cause mortality. This infectious synergism is dependent on the expression of staphylococcal alpha-toxin, and secretion of this potent virulence factor is actually augmented by C. albicans via an agr-dependent mechanism. Moreover, prophylactic neutralization of alpha-toxin with a monoclonal antibody is sufficient to elicit protection during coinfection. Therefore, we have demonstrated that a pathogenic fungus can enhance virulence determinants of a bacterium in vivo with devastating consequences to the host. These results have important implications in the surveillance and treatment of polymicrobial disease and highlight the dynamic intersection of environment, pathogens, and host.

sepsis. In this work, we have shown that coinfection with C. albicans and S. aureus is highly lethal, leading to Ͼ80% mortality by day 1 postinfection, whereas monoinfection with C. albicans or S. aureus does not cause mortality. This infectious synergism is dependent on the expression of staphylococcal alpha-toxin, and secretion of this potent virulence factor is actually augmented by C. albicans via an agr-dependent mechanism. Moreover, prophylactic neutralization of alpha-toxin with a monoclonal antibody is sufficient to elicit protection during coinfection. Therefore, we have demonstrated that a pathogenic fungus can enhance virulence determinants of a bacterium in vivo with devastating consequences to the host. These results have important implications in the surveillance and treatment of polymicrobial disease and highlight the dynamic intersection of environment, pathogens, and host.

KEYWORDS Candida albicans, intraabdominal infection, polymicrobial
T he pathogenic fungus Candida albicans and the ubiquitous bacterial pathogen methicillin-resistant Staphylococcus aureus (MRSA) remain serious clinical threats (1,2). Together, these microorganisms rank among the most prevalent causes of nosocomial sepsis and catheter-related bloodstream infections, and recent reports have identified their coisolation with increasing frequency (3,4). While polymicrobial infection is often associated with poor patient prognosis, studies designed to mechanistically evaluate microbial community composition and fungal-bacterial interactions in the context of host immunity are still in their infancy (5).
Comprehensive epidemiological data on S. aureus has identified several distinct clades, each characterized by unique disease pathology and virulence factors. Among these are USA200 and USA300 strains, which are commonly referred to as hospitalacquired (HA) and community-acquired (CA) MRSA, respectively. Generally, HA-MRSA strains are robust formers of staphylococcal biofilm and demonstrate wider antimicrobial resistance profiles, and some clades (e.g., CC30) exhibit lower levels of secreted bacterial toxins (6). These strains are often associated with orthopedic and medical device-related infections. Conversely, CA-MRSA strains secrete comparatively higher levels of bacterial toxin, including Panton-Valentine leukocidin (PVL) and the poreforming cytolytic alpha-toxin (1). These strains have been associated with skin infection outbreaks (e.g., in prisons, care facilities, locker rooms), and it is believed that high levels of toxin and others secreted factors allow for efficient skin-skin and skin-fomite transfer. Toxin expression in S. aureus has been linked to multiple bacterial sensory regulators, but perhaps the best studied is the agr (accessory gene regulator) quorum sensing system (7).
The agr system is the product of the RNAII transcript, consisting of an operon composed of four genes (agrABCD) (8). AgrD is the signal peptide, which gets secreted across the bacterial membrane into the extracellular space and modified to its mature form, autoinducing peptide 2 (AIP-2), via translocation through AgrB. AIP-2 is sensed by the cell surface-associated AgrC receptor that induces phosphorylation of AgrA. Activated AgrA can then bind again to the P2 promoter to increase RNAII transcription, completing a positive feedback cycle that is increased in a cell-density dependent fashion (i.e., quorum sensing). However, activated AgrA can also bind to the P3 promoter to drive expression of the RNAIII transcript, which directly encodes delta-toxin and a posttranscriptional regulatory RNA. Expression of RNAIII decreases transcription of rot (repressor of toxin), thereby increasing exotoxin secretion (9). Both alpha-and delta-toxin are well-characterized staphylococcal virulence factors, mediating a variety of pathological effects, including hemolytic activity, dermonecrosis, inflammasome activation, abscess formation, leukocyte oxidative burst, and reduced macrophage phagocytic killing (10)(11)(12)(13)(14). Toxin production by S. aureus is crucial for systemic disease, as high titers of antibody to staphylococcal exotoxins (including alpha-toxin) directly correlates with improved survival rates during clinical S. aureus sepsis (15).
A series of studies by Carlson demonstrated that peritoneal coinoculation of mice with C. albicans and S. aureus resulted in synergistic increases in mortality, while mice inoculated with each of these microbes alone efficiently cleared the infection (16)(17)(18). It was also observed that not all strains of S. aureus resulted in polymicrobial infectious synergism equally, with some decreasing the staphylococcal 50% lethal dose (LD 50 ) by as much as 70,000-fold while others demonstrated only modest synergistic effects (2to 3-fold) (16). It was proposed that the expression of specific staphylococcal toxins governed increased mortality, namely, alpha-toxin and delta-toxin. Unfortunately, lack of isogenic controls complicated data interpretation. Recently, our labs recapitulated these findings using a USA200 strain of S. aureus (NRS383) and identified that the host eicosanoid prostaglandin E 2 (PGE 2 ) is associated with disease severity and that pharmacologic blockade of PGE 2 synthesis and PGE 2 receptors 1 and 3 dramatically improves survival rate (19)(20)(21). Interestingly, staphylococcal exotoxins have been implicated in the activation of phospholipase A2 and subsequent prostaglandin release (22,23). Furthermore, synergistic effects on mortality and PGE 2 generation during coinfection with S. aureus are independent of the capacity to undergo fungal morphogenesis (the major virulence attribute of C. albicans) and are not limited to C. albicans, as various other Candida species (including C. dubliniensis, C. tropicalis, and C. krusei) also enhance morbidity and mortality during intra-abdominal infection (IAI) (19,24).
Regarding the potential link to toxin expression and synergistic lethality with C. albicans, the objective of this study was to identify whether staphylococcal toxins may be required for lethal coinfection and/or whether toxin expression is elevated during polymicrobial IAI. In support of this, we show strong in vitro and in vivo evidence that staphylococcal alpha-toxin is necessary for robust infectious synergism and that C. albicans augments staphylococcal toxin production via engagement of the agr quorum sensing system. other strain-dependent differences). However, JE2 was nonlethal during monomicrobial infection. Thus, we hypothesized that coinfection with C. albicans may actually augment toxin expression.
The staphylococcal agr quorum sensing system is tightly linked to toxin regulation. As part of a two-component signal transduction system, upon its activation, AgrA is phosphorylated and drives expression of RNAIII, the major toxin effector, by binding to the P3 promoter. We constructed an agrA reporter strain in JE2 based on plasmid pDB22 [S. aureus(pDB22)] with the P3 promoter fused to green fluorescent protein (GFP) for use as a proxy of toxin production (see Fig. S1 in the supplemental material). During planktonic culture in 0.6ϫ Trypticase Soy Broth containing 0.2% glucose (TSB-g), GFP levels were absent at early time points, consistent with the fact that activation of the agr system is cell density dependent ( Fig. 2A). At ϳ8 h, reporter activities were noticeable in both monomicrobial and polymicrobial cultures harboring S. aureus-(pDB22). However, at 12 and 16 h, reporter activities of polymicrobial cultures were significantly higher (as much as doubled) than those during monomicrobial growth, indicating augmented activation of the major signaling pathway controlling toxin production ( Fig. 2A). To confirm these findings, aliquots from cultures were removed and imaged by fluorescence microscopy, revealing a qualitatively enhanced GFP signal in polymicrobial cultures in comparison to that in monomicrobial cultures (Fig. 2B). To address whether organism burden was drastically altered during polymicrobial growth at the 16 h time point, C. albicans, S. aureus, or the two combined were grown in TSB-g and plated on selective microbiological media. Enumeration revealed nearly identical C. albicans and S. aureus CFU levels between the monomicrobial and polymicrobial cultures (Fig. 2C). However, cell-free supernatants added to Trypticase Soy Agar (TSA)blood revealed strikingly higher levels of hemolysis from polymicrobial cultures, suggesting enhanced levels or activity of hemolytic toxins during growth with C. albicans, consistent with GFP reporter results. Importantly, cell-free supernatants derived from C. albicans cultures were devoid of hemolytic activity (Fig. 2D). Augmented lysis of rabbit red blood cells (RBCs) (a cell type exquisitely sensitive to alpha-toxin) from polymicrobial cultures further confirmed this phenotype quantitatively (Fig. 2E) (26).
In order to determine whether downstream agr-regulated genes were also being increased during polymicrobial growth, we assessed the transcriptional responses of S. aureus or C. albicans in combination with S. aureus by quantitative real-time PCR (qPCR). By rupturing cells with lysostaphin (an enzyme that breaks down staphylococcal cell walls), we were able to selectively isolate S. aureus RNA so that the amounts of recovered staphylococcal nucleic acid between monomicrobial and polymicrobial cultures could be made equivalent. qPCR analysis revealed increases in hld (encoding delta-toxin, 5.1-fold), hla (alpha-toxin, 80-fold), RNAII (agr operon, 2.5-fold), agrA (quorum sensing regulator, 7-fold), and spa (surface protein A, -0.5-fold) (Fig. 3A). As RNAII, hld, hla, and agrA itself are downstream targets of agr activation, these results provide further evidence of augmented agr activity during growth with C. albicans. The exception to this gene list was the downregulation of spa. However, it has been well documented that spa is negatively correlated with agr activation, and thus its downregulation is in agreement with previous observations (27). Similar regulation of these quorum sensing effectors was also conserved at the protein level, as evidenced by Western blotting of cell-free supernatants (Fig. 3B). Phenol-soluble modulins (of which delta-toxin is a member) are small, ethanol-soluble peptides that prove somewhat difficult to resolve by electrophoresis (28,29). Therefore, we utilized an ethanol precipitation step and subsequent drying to fractionate supernatants to improve toxin recovery, as described previously (28). Densitometric analysis demonstrated that alphatoxin (37 kDa) was increased by 8.2-fold, delta-toxin (3 kDa) was increased by 4.3-fold, and protein A (56 kDa) was decreased by 0.7-fold, further explaining increased in vitro hemolytic activity. The protein expression levels were in line with transcriptional results, except that for alpha-toxin, which was lower than anticipated by qPCR and may indicate involvement of other mechanisms of posttranscriptional control (30).
These findings revealed that agr activity was elevated during polymicrobial growth, although it was undetermined whether agr was required for augmented toxin secretion during coculture with C. albicans. Therefore, we assessed toxin secretion of a ΔagrA mutant during polymicrobial growth. Using the qualitative blood agar-based and quantitative RBC lysis assays, we confirmed that, indeed, enhanced toxin expression during coculture is agrA dependent, as growth with C. albicans was unable to rescue this agr defect ( Fig. 4A and B). Given the strong upregulation of alpha-toxin observed during in vitro growth with C. albicans, we hypothesized that this toxin may drive  synergistic lethality during coinfection. Use of an isogenic Δhla strain revealed loss of lytic activity against sheep RBCs, indicating that the in vitro lytic phenotype is hla dependent (Fig. 4C). However, complementation of hla on a low-copy-number plasmid (Δhla-phla) restored the lytic phenotype to wild-type (WT) levels. Importantly, both the WT and complemented strains exhibited elevated RBC lysis during coculture with C. albicans (Fig. 4C). An alpha-toxin-specific enzyme-linked immunosorbent assay (ELISA) used to assess toxin levels in culture supernatants quantitatively confirmed these results (Fig. 4D). Alpha-toxin was required for early synergistic lethality during IAI, since coinfection with the Δhla mutant was avirulent in comparison to coinfection with WT and Δhla-phla strains (Fig. 4E). As expected, coinfection with the ΔagrA mutant and monomicrobial infections were nonlethal over this same time course (data not shown). We next wished to determine whether agr was engaged early during infection, prior to presentation of severe symptomatology. Mice were inoculated with either S. aureus, S. aureus(pDB22), C. albicans and S. aureus, or C. albicans and S. aureus(pDB22), sacrificed 8 h p.i., and subjected to peritoneal lavage. Analysis of the microbial burden in the lavage fluid revealed that similar to results in in vitro culture, the staphylococcal burdens were similar in monomicrobial and polymicrobial infections (Fig. 5A). Microscopic analysis of the lavage fluid revealed the presence of C. albicans and S. aureus, as well as a robust mononuclear infiltrate. The S. aureus(pDB22) strain demonstrated agr activity, as evidenced by GFP signal, but only sparingly and mostly in association with resident immune cells (Fig. 5B). In contrast, S. aureus(pDB22) coinoculated with C. albicans demonstrated a frequent GFP signal and when located near C. albicans (Fig. 5B). While this result is qualitative, it visually demonstrates that the agr system is indeed engaged in the murine peritoneal cavity. Importantly, at this early time point, there was a nearly 3-fold increase in alpha-toxin (Fig. 5C) and roughly a 4-fold increase in PGE 2 (Fig. 5D) levels observed in the peritoneal lavage fluid of coinfected mice. Mice infected with C. albicans alone demonstrated modest PGE 2 and no measurable alphatoxin levels in the lavage fluid, demonstrating assay specificity.
Lastly, we wanted to determine whether a blockade of alpha-toxin was sufficient to elicit protection during polymicrobial IAI. Therefore, we passively immunized mice 24 h prior to coinfection with the anti-alpha-toxin-specific monoclonal IgG1 antibody (MAb) MEDI4893* at doses of 15 and 45 mg/kg or with the isotype control R347 (11). Mice receiving control antibody succumbed to coinfection within 1 day, similar to that previously observed in untreated mice (Fig. 6A). Mice receiving MEDI4893* at 15 mg/kg exhibited a slight delay in mortality (ϳ2 days p.i.). However, approximately 50% of these mice succumbed to infection by day 3 p.i., and this increased to 70% by day 8 p.i. Importantly, mice receiving MEDI4893* at a dose of 45 mg/kg exhibited no mortality until day 4 p.i. and an overall mortality rate of only 30% (Fig. 6A). Thus, passive administration of MEDI4893* improves survival during polymicrobial IAI. We next wanted to determine whether administration of alpha-toxin during infection with C. albicans was sufficient to cause synergistic lethality. First, a nonlethal i.p. dose of alpha-toxin (0.5 g) was determined (Fig. 6B). However, coinoculation of mice with C. albicans and 0.5 g alpha-toxin did not recapitulate the lethal synergy between C. albicans and S. aureus (Fig. 6C). Collectively, these results demonstrate that the staphylococcal regulator agr is robustly increased during fungal-bacterial coinfection and that alpha-toxin is necessary but not sufficient to mediate early lethal synergism with C. albicans.

DISCUSSION
In nature, microorganisms rarely grow as single species; instead, most exist as polymicrobial communities characterized by mutualistic, parasitic, commensalistic, or antagonistic interactions. With the advent of high-throughput sequencing technologies, we have now begun to realize the tremendous microbial diversity that exists, not only in the environment but also within the human biome. Although we are rapidly amassing incredible amounts of data regarding microbiome composition, functional and mechanistic relationships between microbial consortia and the human host are only now being explored in depth. Furthermore, the boundaries of Koch's postulates are being challenged, as some disease phenotypes are now realized to result from a complex set of microbial virulence mechanisms, partially governed by microbe-microbe interactions. One extreme example is "lethal synergism," in which coinfection mediates the mortality of otherwise survivable infections with each microbe independently. This phenomenon has been previously reported in a variety of biological systems but is perhaps best characterized by fungal-bacterial IAI with C. albicans and S. aureus.
Our labs have explored the interaction between C. albicans and S. aureus extensively. Previous studies revealed that, indeed, C. albicans influenced global protein expression during growth with S. aureus, including the modulation of several putative virulence factors (31). It has also been determined that S. aureus preferentially binds the invasive hyphal form of C. albicans via the candidal adhesin Als3p and that this interaction allows for enhanced mortality and systemic dissemination from the oral cavity during oropharyngeal candidiasis (32)(33)(34). Furthermore, during biofilm growth, C. albicans confers high-level vancomycin resistance to S. aureus by forming a drug permeability barrier via elaboration of a carbohydrate-dense extracellular matrix (35,36). These studies highlight the highly dynamic relationship between these microbes.
Data in this study build off recent efforts by our labs seeking to define the mechanism by which C. albicans augments S. aureus infection during polymicrobial IAI, as observed by Carlson many years earlier. Initial studies determined that host inflammation, including the eicosanoid PGE 2 and cytokine interleukin-6 (IL-6), was synergistically exacerbated in mice challenged with the two microbes simultaneously compared to that of monomicrobial infection and correlated with a lethal outcome (24). Furthermore, protection was mediated by anti-inflammatory administration and subsequent PGE 2 blockade (20,21). Using various transcriptional regulator mutants, lethal infection was surprisingly independent of C. albicans morphological transition, as other experimental C. albicans disease models are almost universally dependent on morphogenetic regulation (19). This observation was further confirmed by the fact that various other Candida species mediate synergistic lethality independent of the capacity to form true hyphae during IAI (24). Importantly, all of these studies were conducted using S. aureus strain NRS383 (USA200), which is grouped into the CC30 clonal complex (37). While synergistic lethality was achieved during coinfection with this strain, kinetics (ϳday 3 p.i.) were modest compared to that of infection with JE2 (Ͻday 1 p.i.) in this study. Interestingly, CC30 strains are documented to have very low to no production of alpha-toxin under in vitro conditions, in some cases due to a premature stop codon within hla (38). NRS383 exhibits a weak zone of hemolysis on blood agar, further emphasizing the importance of staphylococcal alpha-toxin in driving lethal synergism with C. albicans.
The major staphylococcal regulator of alpha-toxin production is indeed governed by agr-dependent signaling. However, several reports suggest contributions of agr- Candida albicans Enhances S. aureus Toxin Production ® independent mechanisms regulated via the sarA and saeRS networks (39). In support of this, clinical isolates that are agr negative have been observed to activate hla transcription in vivo. Similarly, deletion of sae causes loss of hla transcription, despite an intact agr regulon (40). Results from our previous study of polymicrobial IAI demonstrated that S. aureus disseminates to distant tissues (including the brain) quickly and, at later time points, reaches greater numbers during polymicrobial infection than during monomicrobial infection (19). Thus, it is possible that delayed activation of these alternative alpha-toxin regulators at sites of dissemination in vivo may explain lethality with strains exhibiting low hemolytic activity in vitro. Given that a relatively large fraction (15 to 60%) of hospital-acquired MRSA strains are agr defective, it will be interesting to delineate the role of these alternative alpha-toxin regulators in driving synergistic lethality and their potential activation during growth with C. albicans (41).
Experiments utilizing the isogenic Δhla strain during coinfection demonstrated a clear requirement for alpha-toxin to mediate infectious synergism. These findings were corroborated by a study by Rauch et al., in which loss of hla during high-dose monomicrobial peritoneal challenge with S. aureus strain Newman dramatically improved survival rates over that of wild-type infection (25). Moreover, blockade of alpha-toxin using the investigational neutralizing antibody MEDI4893* further confirmed the importance of this virulence determinant in driving mortality during IAI. MEDI4893* exerts its effects by not only inhibiting engagement of alpha-toxin with its host receptor ADAM10 but also by preventing oligomerization of the mature heptamer required for membrane pore formation and lytic activity (42). Aside from its clear protective effects demonstrated against polymicrobial IAI in this study, MEDI4893* has also demonstrated efficacy in both immunocompetent and immunocompromised animal models, including models of necrotizing pneumonia, dermonecrosis, and sepsis (11,(43)(44)(45). Although prophylactic treatment with MEDI4893* did not confer 100% survival in this study, it did significantly delay mortality to 2 (15 mg/kg) and 4 (45 mg/ kg) days p.i., respectively, which could allow for an extended therapeutic window for antimicrobial administration. Although beyond the scope of this work, future studies to determine whether MEDI4893* can be given therapeutically to confer protection postinfection are warranted. In any case, passive immunization against alpha-toxin is an intriguing possibility for those at elevated risk of IAI, including abdominal surgery, peritoneal dialysis, or tertiary peritonitis patients.
Cohen et al. recently demonstrated that alpha-toxin is necessary and sufficient for S. aureus lethal synergy during lung coinfection with Gram-negative opportunists (46). However, it is still unclear whether alpha-toxin is sufficient to mediate lethal synergism during C. albicans infection or if other components of S. aureus are also required. Carlson had previously observed that inoculation of C. albicans with cell-free spent culture supernatants of S. aureus could induce synergistic lethality, suggesting that staphylococcal secreted products (e.g., toxins) were perhaps implicated (18). The potency of alpha-toxin is impressive, with the LD 50 being reported as ϳ1 g via systemic injection or intranasal challenge; an intradermal dose of 10 g is sufficient to induce large skin lesions (11,13,47). Despite initial signs of morbidity, inoculation of mice with C. albicans, followed by a sublethal dose of purified alpha-toxin, was unable to recapitulate synergistic mortality (Fig. 6). There are several reasons why this may be the case. It is possible that a sublethal bolus dose of toxin is not sufficient to potentiate the effects with C. albicans. It is also equally likely that sustained expression of S. aureus toxin during infection is required to augment coinfection. Alternatively, alpha-toxin may only potentiate host-or microbe-derived signals present during infection with intact fungal or bacterial pathogens or may be required to be present with C. albicans at the immunological synapse. Additionally, Carlson's observations may be explained by the presence of multiple toxins or shed cell wall material present in the culture supernatant that was not completely inactivated by heating. These soluble factors may maintain proinflammatory activity to elicit robust PGE 2 and cytokine production during coinfection by stimulating pattern recognition receptors (PRRs) and serving as dangerassociated molecular patterns (DAMPs).
This study describes the novel finding of a human-pathogenic fungus engaging the virulence of a bacterial pathogen with dire consequences for the host. That said, cross-species and cross-domain quorum sensing communication has been observed in several other reports. For example, Bamford et al. demonstrated that the diffusible chemical signal AI-2, derived from the luxS gene of the ubiquitous oral commensal Streptococcus gordonii, enhances C. albicans hyphal formation during polymicrobial biofilm growth (48). Perhaps one of the best-studied interactions occurs between C. albicans and Pseudomonas aeruginosa, whereby the bacterial quorum sensing molecule 3-oxo-C12 homoserine lactone inhibits the fungal yeast-to-hypha transition (49). P. aeruginosa also attaches to the fungal surface, secreting a variety of fungicidal substances (phenazine, pyocyanin, etc.) (50). In response, C. albicans secretes its own quorum sensing compound, farnesol, which has antibacterial activity (51). Therefore, it is not unreasonable to speculate that secreted fungal quorum sensing molecules (or otherwise) may elicit toxigenic responses by S. aureus via quorum sensing induction. Indeed, at supraphysiologic concentrations, farnesol does exhibit antistaphylococcal effects (52). However, an apparent lack of an antagonistic relationship between these microbes during coculture, as evidenced by previous imaging studies, further clouds this hypothesis (31).
It is possible that cell-cell contact, and not necessarily a secreted factor, mediates the toxigenic response in S. aureus. Previous studies from our labs have demonstrated that the major adhesin responsible for C. albicans-S. aureus interactions is the candidal protein Als3p (33). This fact, coupled with the ability of several transcriptional regulator mutants of C. albicans incapable of hypha formation to still promote lethal synergism during IAI, suggests an alternative mechanism, as these strains would also be predicted to not express hypha-associated Als3p (19,24). Furthermore, coinfection with a Δ/Δals3 mutant of C. albicans still exhibited synergistic lethality with JE2 during experimental IAI (unpublished studies). That said, maximal engagement of a soluble factor to its potential target in vivo may occur if both microbes maintain intimate contact. We also cannot rule out the possibility that an environmental factor or host component mediates enhanced toxin secretion in vivo. It is well known that agr is sensitive to several environmental stimuli, including pH; the pH of the abdominal cavity during C. albicans peritonitis is slightly alkaline (53,54). Thus, it is possible that anatomical niches within the abdomen buffer the acidic end products of staphylococcal metabolism, thereby maintaining neutral pH levels optimal for sustained agr activation. C. albicans infection may also generate host or fungal by-products that S. aureus can use as metabolic substrates or signaling molecules to upregulate toxin expression.
Collectively, these results demonstrate that C. albicans augments S. aureus virulence via an agr-and alpha-toxin-dependent mechanism during polymicrobial IAI, resulting in devastating consequences for the host. These findings are important, given the frequency with which these microbes are coisolated from biologic (e.g., vagina, gastrointestinal tract, skin, oropharynx) and artificial (e.g., indwelling catheters) sites during colonization or infection. Furthermore, systemic disease caused by Candida species is associated with mortality rates exceeding 40% even with appropriate therapy (55). Given our observations, it is possible that these unacceptably high mortality rates may not only result from fungal infection alone but may also be engaging virulence mechanisms in otherwise commensal microbes or coinfecting pathogens (56)(57)(58). Novel strategies to inhibit fungal-bacterial quorum sensing cross talk or its downstream effectors may be warranted to better manage IAI, including the use of an antibodymediated vaccine against staphylococcal alpha-toxin in at-risk patient populations to limit the incidence of severe sepsis.

MATERIALS AND METHODS
Ethics statement. The animals used in this study were housed in AAALAC-approved facilities located at the University of Tennessee Health Sciences Center (UTHSC) in the Regional Biocontainment Laboratory. The UTHSC Animal Care and Use Committee Laboratory Animal Care Unit (LACU) approved all animal usage and protocols (protocol number 18-060). Mice were given standard rodent chow and water ad libitum. Mice were monitored daily for signs of distress, including noticeable weight loss and lethargy.
Candida albicans Enhances S. aureus Toxin Production ® UTHSC LACU uses the Public Health Policy on Humane Care and Use of Laboratory Animals (PHS) and the Guide for the Care and Use of Laboratory Animals as a basis for establishing and maintaining an institutional program for activities involving animals. To ensure high standards for animal welfare, UTHSC LACU remains compliant with all applicable provisions of the Animal Welfare Act (AWAR), guidance from the Office of Laboratory Animal Welfare (OLAW), and the American Veterinary Medical Association Guidelines on Euthanasia.
Strains and growth conditions. C. albicans strains SC5314 (wild type, reference isolate) was used for all experiments. The following S. aureus strains obtained from the Biodefense and Emerging Infectious (BEI) Research Resources repository were used: JE2 (WT; USA300 derived from strain LAC but cured of erythromycin resistance and cryptic plasmids), NE1532 (ΔagrA), and NE1354 (Δhla) (59). The S. aureus reporter strain harboring plasmid pDB22 encoding erythromycin resistance and a P3 promoter fused to GFP mut2 was a kind gift from Pete Greenberg (University of Washington) (60). Using standard techniques, plasmid pDB22 was isolated from MN8-pDB22 and transformed into strain JE2 by electroporation to yield S. aureus(pDB22) (61). Using fluorescence microscopy, it was confirmed that S. aureus(pDB22) exhibited robust GFP expression during stationary growth, consistent with agr activation (see Fig. S1 in the supplemental material). All strains were maintained as 20% glycerol stocks and stored at Ϫ80°C.
Unless specifically noted otherwise, C. albicans was streaked onto Yeast Peptone Dextrose (YPD) agar. Isolated colonies were selected and inoculated into liquid YPD medium and grown overnight at 30°C with shaking at 200 rpm. S. aureus was streaked onto Trypticase Soy Agar (TSA) with antibiotic selection as required. Isolated colonies were selected and inoculated into liquid Trypticase Soy Broth (TSB) medium and grown overnight at 37°C with shaking at 200 rpm. The following day, cultures were diluted 1:100 in fresh TSB and returned to the shaking 37°C incubator for 3 h until cultures reached the logarithmic phase of growth.
Construction of an hla-complemented strain. In order to complement alpha-toxin expression, primers (restriction enzyme sites indicated by underlined text) hlaF-HindIII (5=-GTAAAGCTTCATACGATA CTTTTTCGTTATCTATTAG) and hlaR-BamHI (5=-CGGGGATCCCAGTATAAAAATTAGCCGAAAAACATCATTT CTG) were used to PCR amplify genomic DNA isolated from strain JE2 to generate a fragment containing the entire hla open reading frame, ϳ500 bp of the 5=untranslated region (UTR), and ϳ100 bp of the 3=UTR. Vector pSK5630 (a low-copy-number plasmid, ϳ5 per cell) and the hla PCR product were digested with HindIII and BamHI and ligated to yield plasmid pSK-hla. The dcm-deficient Escherichia coli strain IM08B was made chemically competent and used as a propagation host for pSK-hla after standard heat shock transformation and selection on Luria-Bertani (LB) agar containing ampicillin (50 g/ml) to yield strain IM08B-phla (62). pSK-hla was isolated from selective overnight cultures of IM08B-phla using a miniprep procedure according to the manufacturer's protocol (GeneJet; ThermoFisher) and verified by restriction digestion with HindIII and BamHI, followed by Sanger sequencing (UTHSC Molecular Resource Center). Using a vendor-optimized protocol for staphylococcal electroporation (Gene Pulser Xcell; Bio-Rad), pSK-hla was transformed directly into strain NE1354 (Δhla background) and plated on brain heart infusion agar containing chloramphenicol (10 g/ml) to yield the Δhla-phla strain. Individual colonies were replica plated onto TSA containing 5% sheep's blood and chloramphenicol (10 g/ml) to screen for those with a hemolytic phenotype. Plasmids were isolated from hemolysis-positive colonies by a miniprep procedure using an S. aureus-specific manufacturer-supplied protocol (Qiaprep; Qiagen) and verified by PCR amplification using primers pSK5630-F (5=-ACGATGCGATTGGGATATATCAACG) and hla-DETR (5=-GTGTTGTTGTTACTGAGCTGACTATACG).
Murine model of IAI. Intraperitoneal inoculations were conducted as described previously (19,20,24). In most experiments, groups (n ϭ 4) of 6-week-old Swiss Webster mice were injected intraperitoneally (i.p.) using a 27-gauge 0.5-in needle with 1.75 ϫ 10 7 CFU of C. albicans, 8 ϫ 10 7 CFU of S. aureus, or both 1.75 ϫ 10 7 and 8 ϫ 10 7 CFU of the respective microbes simultaneously. Inocula were prepared in a final volume of 0.2 ml pyrogen-free phosphate-buffered saline (PBS). After inoculation, mice were observed up to 10 days for morbidity (hunched posture, inactivity, ruffled fur) and mortality. Mice that exhibited severe morbidity were humanely sacrificed and tallied as a lethal outcome. In some experiments, mice were sacrificed 8 h p.i. prior to severe morbidity. Peritoneal cavities were lavaged by injection of 2 ml of sterile PBS containing 1ϫ protease inhibitors (cOmplete; Roche), followed by gentle massaging of the peritoneal cavity. Peritoneal lavage fluid was then removed using a pipette inserted into a small incision in the abdominal cavity. Animal experiments were repeated in duplicate, and results were combined.
Passive immunization with anti-alpha-toxin monoclonal antibody. Groups of mice (n ϭ 4) were inoculated i.p. with 200 l of the alpha-toxin-specific IgG1 neutralizing antibody MEDI4893* (15 mg/kg, 45 mg/kg) or human IgG1 isotype control R347 (45 mg/kg) prepared in sterile PBS 24 h prior to coinfection with C. albicans and S. aureus. Mice were monitored for up to 10 days p.i. for morbidity and mortality. Animal experiments were repeated in duplicate, and results were combined.
Intraperitoneal delivery of alpha-toxin. Groups of mice (n ϭ 4) were inoculated i.p. with 0.2, 0.5, 0.75, 1, or 5 g of purified native alpha-toxin in 0.1 ml of PBS alone or sequentially with 1.75 ϫ 10 7 CFU of C. albicans in 0.1 ml PBS (13). Mice inoculated with alpha-toxin only received a sham injection of 0.1 ml PBS. Mice were monitored for morbidity and mortality up to 10 days p.i. Data are representative of at least two independent repeats. CFU analysis. Microbial burdens were enumerated by serial dilution plating of peritoneal lavage fluid and culture media onto YPD containing 20 g/ml ampicillin and 2 g/ml vancomycin (for C. albicans enumeration) and TSA containing 20 g/ml ampicillin and 2.5 g/ml amphotericin B (for S. aureus enumeration) via the drop-plate method (63). The plates were incubated overnight at 37°C, and the microbial burden was enumerated and expressed as the number of CFU per ml. CFU values are representative of at least two independent repeats (n ϭ 4 mice per group) and are represented as the median (Ϯstandard deviation [SD]).
Prostaglandin quantitation. Recovered lavage fluid was centrifuged at 500 ϫ g for 5 min, and the supernatant was transferred to clean microcentrifuge tubes. PGE 2 is rapidly converted to its 13,14dihydro-15-keto metabolite in vivo, so the colorimetric prostaglandin E metabolite competitive enzyme immunoassay (EIA) was used to measure PGE 2 as a function of its breakdown product, according to the manufacturer's directions (Cayman Chemicals, Ann Arbor, MI). This assay is highly sensitive and detects as little as 2 pg/ml of PGE metabolites.
S. aureus alpha-toxin quantitation by ELISA. Wells of a polystyrene 96-well microtiter plate were coated with 50 l of 0.1 g/ml anti-alpha-toxin antibody MEDI4893* (diluted in coating buffer [0.2 M carbonate/bicarbonate buffer]) and incubated overnight at 4°C. The plates were washed sequentially with PBS-Tween 20 (PBS-T) and blocked with SuperBlock (Pierce) for 1 h at room temperature. After washing with PBS-T, 50 l of diluted peritoneal lavage fluid or diluted filter-sterilized culture supernatants in PBS was added. Additionally, serial dilutions of native alpha-toxin were included as the standard curve. The plate was then incubated for 1 h at room temperature with shaking (600 rpm). After washing with PBS-T, 50 l of affinity-purified rabbit polyclonal anti-alpha-toxin antibody (2 g/ml) was added and incubated as described above. After washing, 50 l of a 1:10,000 dilution of AffiniPure horseradish peroxidase (HRP)-coupled goat anti-rabbit IgG detection antibody (Jackson ImmunoResearch) was added and incubated for 1 h. The plates were washed extensively with PBS-T, and then 100 l of 3,3',5,5'-Tetramethylbenzidine (TMB) substrate was added for 10 min, followed immediately by the addition of 100 l of ELISA stop solution (0.2 M H 2 SO4). Wells were read at 450 nm using a plate reader (Synergy; BioTek), and experimental values were extrapolated to the standard curve. Culture supernatants and peritoneal lavage fluid generated from the Δhla strain were blank subtracted to account for any potential nonspecific antibody binding to shed protein A. Results were repeated in duplicate (n ϭ 4 mice per group), and data were combined and expressed as the mean Ϯ standard error of the mean (SEM).
Agar plate assay for toxin activity. Colonies of S. aureus (JE2 and ΔagrA and Δhla strains) and C. albicans were grown overnight as described above, and cell densities were adjusted to 1 ϫ 10 7 CFU/ml by counting on a hemocytometer. A 1:100 dilution (50 l) was made into 5 ml of 0.6ϫ TSB containing 0.2% glucose (TSB-g) (a medium previously shown to support good growth of both C. albicans and S. aureus) to establish the following groups: S. aureus alone, C. albicans alone, or C. albicans plus S. aureus (64,65). Monomicrobial cultures received an additional 50 l of sterile PBS. Tubes were placed in a 37°C incubator with shaking at 200 rpm. Aliquots were removed at several time points, centrifuged at 5,000 rpm to remove cellular debris, and passed through a 0.2-m syringe filter. Wells were aseptically formed in TSA agar containing 5% sheep's blood using a cut sterile pipette tip. Sterile culture supernatant (20 l) was added to each well, and the plate was incubated at 37°C for up to 24 h. Plates were imaged using a digital scanner, and the results are representative of at least three independent repeats. RBC hemolytic assay. An aliquot (1 ml) of rabbit blood (Lampire Biologicals) was centrifuged at 1,000 ϫ g for 5 min to pellet red blood cells (RBCs). Packed RBCs were washed two more times by centrifugation and resuspended in 1 ml of PBS, and a 4% (vol/vol) dilution of RBCs was made. Isolated cell-free supernatants (as described above) were serially diluted in sterile TSB-g, and 200 l was added to wells of a microtiter plate. To this, 50 l of RBCs was added, and the plate was incubated at 37°C for 1 h with gentle shaking. Plates were centrifuged at 1,000 ϫ g for 5 min to pellet unlysed RBCs, supernatants were transferred to a fresh microtiter plate, and the optical density at 405 nm (OD 405 ) was collected as a measure of hemoglobin release. Positive (1% Triton X-100) and negative (sterile TSB-g) controls were included. Experiments were repeated in triplicate, and values were normalized as the percentage of results for the positive control and expressed as the mean Ϯ SEM.
agrA-GFP reporter assay. The culture setup for the agrA-GFP reporter assay was similar to that of the agar plate assay described above, except that the S. aureus(pDB22) strain was also included and 10 g/ml of erythromycin was included for plasmid maintenance. Aliquots of cultures (100 l) were removed at various time points postinoculation in TSB-g and placed into wells of a black microtiter plate in triplicate. Fluorescence (488 nm excitation, 515 nm emission) was captured on a plate reader (Synergy; BioTek). Experiments were repeated in triplicate, and results are expressed as mean arbitrary fluorescence units (AFUs) Ϯ SEM.
Fluorescence microscopy. Aliquots of cells from the reporter assay were stained with the fluorescent DNA stain Syto62 per the manufacturer's protocol (Invitrogen) and placed onto coverslipped glass slides. Images were captured on an Olympus FV1000 confocal microscope using GFP and Texas Red filter sets. Samples of peritoneal lavage fluid were placed onto glass slides and imaged using differential interference contrast (DIC) and GFP filter sets. Images are representative of at least three independent repeats.
RNA isolation for quantitative real-time PCR (qPCR). Cultures were set up as described for the agar plate assay. RNA from S. aureus was selectively isolated as described previously, with some modifications (66). Briefly, cells were removed from the incubator and an equal volume of ice-cold 50% acetone-50% ethanol was added to prevent RNA degradation. Cells were centrifuged twice at 4,000 rpm and resuspended in 1 ml of diethyl pyrocarbonate (DEPC)-treated water. Cells were transferred to Eppendorf tubes and centrifuged, and 5 l of lysostaphin was added (10 mg/ml). The pellet was vortexed vigorously, resuspended in 200 l of DEPC-treated water, and incubated in a 37°C water bath for 45 min. To this, 2.5 l of proteinase K (Qiagen) was added and incubated for an additional 15 min. After incubation, 0.2 volumes of 10% SDS was added, followed by an equal volume of a 25:24:1 phenolchloroform-isoamyl alcohol mixture at pH 4 and a small amount of 0.1-mm-diameter zirconia beads. The tubes were vortexed vigorously for 2 min, followed by 1 min of incubation on ice; this was repeated three Candida albicans Enhances S. aureus Toxin Production ® times in total. RNA was extracted by the hot phenol-chloroform method (5:1, pH 4), precipitated with 3 M sodium acetate, and washed with 70% ethanol. The pellet was air-dried and resuspended in 25 l DEPC-treated water. Concentrations were assessed spectrophotometrically (A 260 /A 280 ratio), and RNA integrity was verified by running ϳ1 g of RNA on a 1.4% Tris-borate-EDTA (TBE) agarose gel. Analysis of 16S and 23S (but lack of 18S and 28S) rRNA bands confirmed that staphylococcal RNA from polymicrobial cultures was selectively extracted by this procedure. qPCR for staphylococcal genes. Trace amounts of contaminating DNA were removed from RNA samples (1 g) by treatment with RNase-free DNase I per the manufacturer's protocol (Thermo Fisher). RNA was reverse transcribed using the RevertAid first-strand kit and primed with random hexamers per the manufacturer's protocol (Thermo Fisher). cDNA (100 ng) was amplified using gene-specific primers (Sigma) and the Maxima Sybr green 2ϫ PCR mastermix per the manufacturer's protocol (Thermo Fisher). Amplification and fluorescence measurement were conducted using the 7500 real-time PCR system platform (Applied Biosystems). Expression levels of target genes in polymicrobial cultures were compared to those in monomicrobial cultures and normalized to the staphylococcal reference gene gyrB using the ΔΔC T method as described previously (67). RNA was extracted from triplicate experiments, and ΔΔC T values were calculated, averaged, and expressed as the mean Ϯ SD.
Western blotting for staphylococcal proteins. Cell-free supernatants were separated into ethanolsoluble and -insoluble fractions by treatment with 80% (vol/vol) final concentration ice-cold ethanol for 1 h at Ϫ80°C. Samples were centrifuged at 3,000 ϫ g for 15 min. The ethanol-soluble fraction was transferred to a clean tube, and both fractions were placed in an N-EVAP nitrogen evaporator (Organomation) set at 75°C and 10-mm flow. Upon reaching near-dryness, samples were resuspended in 0.5 ml of ultrapure water, and total protein content was assessed by the bicinchoninic acid (BCA) assay (Pierce). Equivalent volumes (40 l) of concentrated supernatants were diluted in 6ϫ Laemmli sample buffer, boiled, and separated by SDS-PAGE in duplicate. We chose not to normalize results to total protein, as the polymicrobial culture would have elevated levels of protein (both C. albicans and S. aureus) and may bias interpretation when it was compared to monomicrobial cultures. Ethanol-soluble fractions were assessed using Tris-Tricine electrophoresis, and ethanol-insoluble fractions were assessed using a Trisglycine system. An equivalent protein load was confirmed by staining one set of gels with BioSafe Coomassie (data not shown), while the other gel was transferred to nitrocellulose membranes and blocked in 5% powdered milk PBS-T for antibody probing as follows. To detect alpha-toxin, a 1:10,000 dilution of primary rabbit anti-alpha-toxin (Sigma; S7531) and 1:50,000 dilution of HRP-coupled antirabbit IgG antibodies were used. To detect delta-toxin, a 1:1,000 dilution of primary rabbit anti-deltatoxin (LSBio, LS-C156026) and a 1:50,000 dilution of HRP-coupled anti-rabbit IgG antibodies were used. To detect protein A, a 1:2,000 dilution of mouse anti-protein A (Abcam; ab181627) and a 1:50,000 dilution of goat anti-mouse IgG (Fab specific) were used. Primary antibodies were incubated overnight at 4°C in blocking buffer with gentle rocking. Secondary antibodies were incubated for 1 h at room temperature with gentle rocking. Gels were washed at least three times for 5 min with PBS-T between incubation steps. Signal was detected using the SuperSignal chemiluminescent substrate kit (Pierce) per the manufacturer's protocol, and exposure images were captured on the ChemiDoc XRS system (Bio-Rad). Densitometry of the resulting bands was calculated using ImageJ software (NIH). Blots are representative of three independent repeats.
Statistical analyses. A two-tailed Mann-Whitney test was used to compare CFU values between polymicrobial and monomicrobial groups. A two-tailed unpaired Student's t test was used to compare AFU and RBC lysis values for polymicrobial versus monomicrobial groups. A Wilcoxon log rank test was used to determine the significance of mortality plotted in Kaplan-Meier curves. Statistical analyses were performed using GraphPad Prism. All graphs were constructed using GraphPad Prism. Figures were composed using MS Powerpoint and rendered for publication with Adobe Photoshop.