ABSTRACT
The Gram-positive bacterium Enterococcus faecalis and the fungus Candida albicans are both found as commensals in many of the same niches of the human body, such as the oral cavity and gastrointestinal (GI) tract. However, both are opportunistic pathogens and have frequently been found to be coconstituents of polymicrobial infections. Despite these features in common, there has been little investigation into whether these microbes affect one another in a biologically significant manner. Using a Caenorhabditis elegans model of polymicrobial infection, we discovered that E. faecalis and C. albicans negatively impact each other's virulence. Much of the negative effect of E. faecalis on C. albicans was due to the inhibition of C. albicans hyphal morphogenesis, a developmental program crucial to C. albicans pathogenicity. We discovered that the inhibition was partially dependent on the Fsr quorum-sensing system, a major regulator of virulence in E. faecalis. Specifically, two proteases regulated by Fsr, GelE and SerE, were partially required. Further characterization of the inhibitory signal revealed that it is secreted into the supernatant, is heat resistant, and is between 3 and 10 kDa. The substance was also shown to inhibit C. albicans filamentation in the context of an in vitro biofilm. Finally, a screen of an E. faecalis transposon mutant library identified other genes required for suppression of C. albicans hyphal formation. Overall, we demonstrate a biologically relevant interaction between two clinically important microbes that could affect treatment strategies as well as impact our understanding of interkingdom signaling and sensing in the human-associated microbiome.
INTRODUCTION
The literature is replete with examples of specific interkingdom interactions, ranging from highly synergistic (bacterial-fungal symbiosis in some lichens) to highly antagonistic (antibiotic production by Penicillium) and everything in between (reviewed in references 1 to 3). While the nonbacterial component of the human microbiota is poorly understood, what we do know hints at a remarkable complexity. For instance, perturbation of the mouse microbiota via the introduction of fungal species significantly alters global host immune responses (4, 5). The opportunistic bacterial pathogen Pseudomonas aeruginosa and the important human fungal pathogen Candida albicans are openly antagonistic in vitro, with the bacterial species killing fungal hyphae and C. albicans responding to prokaryotic quorum-sensing factors to revert to a resistant yeast form (6, 7). Yet, in vivo, the presence of C. albicans doubles susceptibility to ventilator-associated pneumonia caused by P. aeruginosa (8), and coinfection greatly increases mortality in a burn model (9). Mixed-species biofilms between C. albicans and either Staphylococcus aureus or Streptococcus gordonii are larger and more drug resistant than monomicrobial biofilms (10–12).
We became particularly interested in the interactions between C. albicans and Enterococcus faecalis, a Gram-positive bacterium, because they share a number of similarities beyond the obvious difference that one is a eukaryote and the other a prokaryote. Both are common, commensal organisms found in the gastrointestinal (GI) tract, oral cavity, and other nonsterile sites of mammals, including humans, and are therefore part of the same host-associated polymicrobal communities. In healthy individuals, both species comprise a relatively small proportion of the microbiota but can expand dramatically, given appropriate physiological perturbations, such as drug regimens. Both are opportunistic pathogens that can take advantage of a weakened immune system to cause systemic disease associated with very high morbidity and mortality. This is increasingly common: Enterococcus species are the third most common nosocomial infectious agent, and Candida species are the fourth (13, 14). C. albicans is the most frequently isolated species within the genus, while Enterococcus faecalis and E. faecium are the most prominent enterococci (13). Coisolation of C. albicans and enterococci from infection sites is common (15, 16), indicating that the conditions favoring infection with either are similar. A very recent study investigating the recovery of the gut microbiota in antibiotic-suppressed mice found a remarkable increase in the population of enterococci in animals into which C. albicans had been introduced (17).
Because C. albicans and E. faecalis inhabit many of the same niches and are frequently associated in polymicrobial infections, we wanted to further investigate how they affect each other in the context of an infection. The nematode Caenorhabditis elegans has been used as a model host system for pathogenesis studies with a variety of microbes, including C. albicans and E. faecalis (18–20). Upon ingestion, both these pathogens form a persistent infection in the intestine, and survival can be monitored over time (18, 19). In the case of C. albicans, the fungus forms hyphae that penetrate from the intestine outward, eventually piercing the cuticle; mutants that cannot filament are avirulent (18). C. elegans has also been used to study polymicrobial infections. The Gram-negative bacteria P. aeruginosa, Salmonella, and Acinetobacter baumannii were shown to antagonize C. albicans hyphal formation and killing of the worm, while the Gram-positive species S. aureus and E. faecium did not (21, 22). In this study, we show that when exposed to both organisms, the gut of C. elegans becomes cocolonized with E. faecalis and C. albicans. The coinfection inhibits C. albicans filamentation and killing of the worm. We show that the inhibition of filamentation is partially dependent on the major virulence regulatory system in E. faecalis, Fsr (23), and is mediated by a compound secreted from the bacterial cells. We also observed that filamentation is inhibited in the context of in vitro biofilm formation but E. faecalis has little effect on C. albicans filamentation in the context of in vitro shaking cultures. We further characterized the molecular properties of the inhibitory signal and screened for additional genes that are required for the observed inhibition of C. albicans hyphal morphogenesis by E. faecalis.
MATERIALS AND METHODS
Strains and media.C. elegans glp-4(bn2) sek-1(km4) nematodes were used for all worm liquid assay experiments (24). The glp-4 sek-1 nematodes were propagated on Escherichia coli strain OP50 on nematode growth medium (NGM) agar using standard techniques (25). The E. faecalis strains used were OG1RF (26), V583 (27), SD234 (28), TX5266 (fsrB) (29), TX5128 (gelE sprE) (30), TX5264 (gelE) (31), and TX5243 (sprE) (32). The Tn917 transposon mutant library has been described previously (33, 34). Bacterial strains were grown overnight at 37°C in brain heart infusion (BHI) broth (Difco) or M9HY (35) with 100 μM glucose.
Fungal strains were routinely propagated in yeast extract-peptone-dextrose (YPD) (36). C. albicans strains SC5314 (wild-type prototrophic strain) and CAI4-F2 (ura3/ura3) have been described previously (37, 38). Strain JRC52, which constitutively expresses mCherry, was constructed using plasmid JRC50, which is a derivative of pACT1-GFP (39). Briefly, we replaced the ACT1-green fluorescent protein (GFP) construct with an ACT1-mCherry transcriptional fusion via overlap PCR using genomic DNA to amplify the ACT1 promoter and pRS316-GAP-mCherry (40) for mCherry. This plasmid was linearized by digestion with StuI and used to transform CAI4-F2, selecting for uridine prototrophy. Integration at the RPS10 locus was confirmed by PCR.
C. elegans coinfection assay for survival.The methodology used for the C. elegans coinfection liquid assay, with a few minor modifications, was described previously (21). Briefly, synchronized, young adult nematodes were sequentially preinfected with C. albicans and E. faecalis for 2 h on BHI agar medium containing gentamicin (10 μg/ml). The nematodes were then washed four times with 2 ml of sterile M9W (63). They were collected by centrifugation at 1,000 rpm between each wash and before being pipetted (∼30 worms per well with 2 wells per condition for a total of about 60 worms assayed) into wells of a six-well microtiter dish (BD Falcon) containing 2 ml liquid medium (20% BHI and 80% M9W). Alternatively, for assays in which specific numbers of CFU of E. faecalis were used, the animals were allowed to feed on C. albicans for 2 h and then the nematodes were washed four times with sterile M9W and pipetted (∼30 worms per well with 2 wells per condition) into wells as described above. Bacteria from an overnight culture were directly inoculated into the liquid medium immediately before the preinfected worms were introduced. Plates were incubated at 25°C, and worm death was scored daily. Kaplan-Meier log rank analysis was used to compare survival curves pairwise. P values of <0.05 were considered to be statistically significant. The software GraphPad Prism (version 5.0) was used for the analyses.
Fluorescence microscopy of infected nematodes.To investigate colonization during coinfection of the intestine, an adaptation of a previously used methodology was employed (41). After the nematodes were infected as described above and allowed to incubate for 4 or 7 days, they were washed with 2 ml of sterile M9W and collected by centrifugation at 1,000 rpm four times. They were then transferred to fresh tubes and paralyzed with 1 mM levamisole for 30 min. Anesthetized nematodes were mounted on 2% agarose pads and imaged using an Olympus IX81 automated inverted microscope and Slidebook software (version 5.0). Imaging was performed using the tetramethyl rhodamine isocyanate (TRITC) and fluorescein isothiocyanate (FITC) filter sets to determine the coculture colonization. The level of colonization was scored categorically by E. faecalis only, C. albicans only, or mixed in the nematode intestine. By the chi-square test, the order of infection did not cause a significant difference in the pattern of colonization observed. All experiments were performed at least three times, and GraphPad Prism (version 5.0) was used for the analyses.
Electron microscopy of infected nematodes.Approximately 600 animals/sample were infected by sequential infection on plates, as described above. After incubation in liquid medium for 4 days, specimens were fixed in 3% glutaraldehyde overnight. They were then rinsed in Millonig's buffer for 5 min. The Millonig's buffer was drawn off, and the sample was layered with 2% OsO4 for 60 min at 4°C. The samples were rinsed for 5 min with deionized H2O and were dehydrated at room temperature using sequential incubations in 50% ethanol (5 min), 70% ethanol (10 min), 95% ethanol (10 min), 100% ethanol (10 min, repeated three times), and 100% propylene oxide (10 min, repeated three times). The samples were then permeated with 50% LX-112 resin and 50% propylene oxide for 3 h, followed by 100% LX-112 for 2 h, before embedding in 100% LX-112 in BEEM capsules for overnight polymerization at 70°C. Sections 500 nm thick were cut using a glass knife and a Leica Ultracut-R microtome. They were heat fixed to glass slides and stained for 20 s with toluidine blue to select the most appropriate areas for imaging. The blocks were trimmed, and 120-nm thin sections were cut using the same ultramicrotome and a diamond knife (Diatome US), placed on 100- and 150-mesh copper grids (EMS), and stained for 15 min with 2% uranyl acetate. They were then rinsed with deionized H2O and further stained for 5 min with Reynolds's lead citrate. The grids were imaged using a JEOL 1200 transmission electron microscope at 60 kV and captured with a 2K × 2K Gatan charge-coupled-device camera.
C. elegans coinfection assay for filamentation.Assays with coinfected animals were set up as described above for the survival assay. Plates were incubated at 25°C and examined on day 4 and day 7 for penetrative filamentation of nematodes by using a Zeiss Stemi 2000 microscope. Penetrative filamentation was defined as any breach in the worm cuticle by filamentous cells as seen at a ×50 magnification (21). All experiments were performed at least three times. Differences in percent filamentation were compared pairwise by the t test using GraphPad Prism (version 5.0), with P value of <0.05 considered statistically significant.
Supernatants of E. faecalis OG1RF and mutants.Filter-sterilized supernatants from stationary and log phases (early, middle, and late) were obtained from 5-ml cultures grown in BHI medium that were collected by centrifugation at 4,000 rpm for 10 min in sterile 15-ml conical tubes (BD Falcon). The supernatants were transferred to fresh sterile 15-ml conical tubes and then sterilized by passing through a 0.2-μm-pore-size sterile cellulose acetate membrane syringe filter. The supernatants were diluted to the indicated concentration, and 200 μl was added to the wells of a six-well microtiter plate containing 2 ml of 20% BHI liquid medium containing ∼30 nematodes preinfected with C. albicans. Two wells were set up for each experimental condition. Filter-sterilized OG1RF supernatant was also boiled for 10 min or spun in centrifugal filter units (Millipore) at 4,000 rpm for 22 min. The boiled and supernatant extracts were added at a 10% volume to 2 ml of 20% BHI liquid medium. Plates were incubated at 25°C, and penetration by filamentous cells was scored as described above. All experiments were performed at least three times.
In vitro hyphal morphogenesis assays.C. albicans strain SC5314 (grown overnight in YPD at 30°C) and E. faecalis OG1RF (grown overnight in BHI at 37°C) were both collected by centrifugation and washed once with water. Fungal cells were inoculated into RPMI (Invitrogen) or YPD plus 10% fetal bovine serum (FBS) at a final optical density (OD) at 600 nm (OD600) of 0.1 with or without OG1RF at an initial OD of 0.05. After 2 h, hyphal morphology was assessed microscopically. For the supernatant experiments, C. albicans cells were processed as described above and then inoculated into 50% YPD-FBS and 50% either fresh BHI or sterilized supernatant from OG1RF grown overnight in BHI. Hyphal morphology was assessed after 2 h incubation at 37°C. At least 100 cells were counted per condition, and the experiment was repeated three times. Differences in each morphotype under the two conditions were compared by a pairwise t test, with a P value of <0.05 considered statistically significant.
Biofilm morphogenesis assays.Biofilm assays were conducted in a manner similar to those previously described (42–44). C. albicans strains were grown overnight at 30°C in YPD and washed three times with phosphate-buffered saline (PBS), and the final OD600 was adjusted to 0.5 in YPD. Serum-coated pads (0.5 by 0.5 cm) of surgical-grade silicone (Cardiovascular Instruments, Wakefield, MA) were inoculated with 1 ml of C. albicans in a 24-well tissue culture-treated plate and incubated at 30°C for 90 min. The supernatant of E. faecalis (OG1RF and SD234) was collected as described above. Inoculated silicone pads were then gently washed with PBS and transferred to a new 24-well plate with 500 μl of buffered Spider medium supplemented with 20% fetal bovine serum and either 500 μl of filter-sterilized BHI or E. faecalis supernatant. For dual-species biofilms, E. faecalis (OG1RF and SD234) was grown overnight at 37°C. Fresh log-phase starter cultures were made by inoculating 5 ml of Todd-Hewitt broth (THB) with 25 μl of the overnight culture and incubated at 37°C until an OD600 of 0.5 was reached (∼3 h). Bacterial cultures were then washed three times with PBS and 50 μl was added to preinoculated C. albicans silicone pads in 950 μl of Spider medium supplemented with 20% serum. All biofilms were incubated for 24 h at 37°C. For microscopic imaging, C. albicans biofilms were gently washed in PBS and stained using 35 μg/ml of calcofluor white for 5 min in the dark. Biofilms were imaged using 4′,6-diamidino-2-phenylindole and differential interference contrast (DIC) filters on an Olympus IX81 automated inverted microscope and Slidebook software (version 5.0). Dual-species biofilms were also imaged using the FITC filter to capture the GFP-expressing strain E. faecalis SD234. For quantification of C. albicans morphology, treatments were done in triplicate in three separate experiments. Cell enumeration was performed using at least 10 different fields of view and at least 500 cells. Differences in each morphotype under the two conditions were compared pairwise by the t test, with a P value of <0.05 considered statistically significant.
E. faecalis Tn917 insertion screen using the C. elegans coinfection model.An E. faecalis OG1RF library consisting of approximately 581 Tn917 insertion mutants was screened to find additional factors involved in C. albicans filamentation inhibition in C. elegans (33, 45). The E. faecalis mutants were inoculated into 96-well microtiter plates (BD Falcon) containing BHI medium and allowed to grow overnight at 37°C. Nematodes were preinfected with C. albicans and then washed four times with M9W and pipetted (∼10 to 12 worms per well) into each well of a 96-well microtiter plate. The bacteria from the overnight cultures were directly inoculated into the liquid medium, immediately after the preinfected worms were added. Filamentation was defined as any breach in the worm cuticle by filamentous cells as seen at a ×50 magnification. The percentage of worms with filamentation was assessed on day 4 and day 7. All experiments were performed at least three times.
RESULTS
Coinfection with E. faecalis and C. albicans attenuates killing of C. elegans.To assess the effects of a polymicrobial infection comprised of C. albicans and E. faecalis, we used a liquid medium assay. Models of C. elegans infection by E. faecalis have previously been developed for both solid and liquid medium (19, 24), but for C. albicans, only a liquid assay has been reported in the literature (18, 21). Following a protocol similar to that previously reported (21), we allowed C. elegans to feed on a lawn of one microbe, followed by feeding on a lawn of the second, to inoculate the animals with both organisms. The animals were then placed in M9W containing 20% BHI, and survival was assessed over time. As shown in Fig. 1A, infection with C. albicans or E. faecalis significantly reduced survival compared to that for animals infected with the E. coli nonpathogen control. However, infection with one microbe followed by infection with the second microbe dramatically reduced killing. The effect was most dramatic when the animals fed on E. faecalis prior to exposure to C. albicans.
E. faecalis attenuates C. albicans killing of C. elegans. (A) The killing of C. elegans was significantly reduced when exposed to C. albicans (JRC52) and E. faecalis (SD234) on solid medium before being transferred to liquid medium compared with the killing when exposed to C. albicans alone (P < 0.0001). C. elegans worms feeding on E. coli (OP50) were used as a control. C. albicans > E. faecalis, exposure first to the fungus and subsequent exposure to the bacterium; E. faecalis > C. albicans, exposure first to the bacterium and subsequent exposure to the fungus. (B to D) Nomarski and fluorescent views of C. elegans fed on C. albicans, E. faecalis, or both on day 4. (B) C. albicans alone; (C) E. faecalis alone; (D) C. albicans and E. faecalis coculture. (E to G) Nomarski and fluorescent views of C. elegans fed on C. albicans, E. faecalis, or both on day 4. (E) C. albicans alone; (F) E. faecalis alone; (G) C. albicans and E. faecalis coculture. Following feeding on both microbes in the order indicated, the proportion of worms with either mixed or monomicrobial colonization was quantified on day 4 (H) or day 7 (I). Colonization was scored as described in Materials and Methods, and the percentage of worms in each category is indicated. n, number of worms observed.
It has previously been determined that both C. albicans and E. faecalis infect the intestine of the worm (18, 19). To determine if feeding on one pathogen clears the second or if both colonize the gut, we performed our experiments using an E. faecalis strain (SD234) that constitutively produces GFP (28) and a C. albicans strain (JRC52) that generates mCherry (a gift of John Collette). As shown in Fig. 1D and G to I, 80 to 90% of the worms displayed gut colonization with both strains following coinfection at 4 days and 7 days. The order of exposure did not affect the results in a significant manner. C. albicans and E. faecalis appeared to form distinct patches of colonization in the gut rather than a uniform mixture (Fig. 1D and G).
E. faecalis inhibits hyphal morphogenesis by C. albicans in the C. elegans model.In previous work, it was shown that C. albicans vigorously forms hyphae in the C. elegans gut and these invasive structures come to fully penetrate the animal's tissue, including the cuticle (18). The hyphal morphogenesis appeared to be a major mechanism of worm killing, as mutants unable to form hyphae are very attenuated (18, 46). When examining the coinfected animals, we observed a striking lack of hyphal formation by C. albicans (compare Fig. 1B and E to D and G).
To examine the microbial and host tissue phenotypes during the infection at higher resolution, transmission electron microscopy (TEM) was performed on animals infected with C. albicans, E. faecalis, or both microbes. To our knowledge, no study has examined worms infected with C. albicans or E. faecalis by TEM, and it was of interest to characterize the host-pathogen interactions at this resolution. After 4 days of infection, C. albicans caused severe disintegration of the brush border lining the GI tract, and in some animals, the fungus had invaded throughout the body. Many of the C. albicans cells were in the hyphal form, corroborating the data collected at lower resolution (Fig. 2A and data not shown). Infection by E. faecalis resulted in a severely engorged intestine containing a mixture of live, electron-dense E. faecalis cells and dead, non-electron-dense cells. Damage to the brush border, including effacement of the microvilli, was observed, but breaching of the intestine was not detected (Fig. 2B and data not shown), consistent with previous observations at lower resolution (19). There were some round dark features in the intestinal cells that resembled bacteria, but on closer inspection at higher magnification, they were determined not to be. They were most likely lipid droplets. We observed that the intestinal cells of worms infected with E. faecalis contained many more of these structures, some electron dense and some not, than the uninfected and coinfected animals (Fig. 2B and data not shown). When C. elegans was coinfected with both organisms, severe damage to the intestine was not apparent. There was no evidence of disintegration of the brush border or effacement of the microvilli, and the intestinal lumen was less engorged. The extensive accumulation of lipid droplets was not apparent. Additionally, all the observed C. albicans cells exhibited a round morphology indicative of the yeast morphotype (Fig. 2C and data not shown). Interestingly, all E. faecalis cells were observed to be alive (electron dense). These data confirm that coinfection with E. faecalis and C. albicans causes less tissue damage to C. elegans than the monomicrobial infections and prevents hyphal morphogenesis by C. albicans and cell death of E. faecalis.
TEM images of the C. elegans intestine when infected with C. albicans (A), E. faecalis (B), or both microbes (C). Ca, C. albicans (JRC52) cells; Ef, E. faecalis (SD234) cells; white boxes in the left panels, a region of the apical border of the intestinal cells which is magnified in the respective insets. Bars, 2 μm (right), 1 μm (left), and 0.2 μm (insets). All images are from longitudinal sections of C. elegans, except for the right of panel B, which is cross-sectional.
To further characterize the protective effects of E. faecalis on C. albicans killing of the worm, C. elegans worms that had been fed C. albicans on solid medium were introduced into culture medium inoculated with specified numbers of E. faecalis CFU. We observed a dose-dependent response, with higher numbers of CFU providing more protection from C. albicans killing (Fig. 3A). A count of 108 CFU was almost as protective as the protection observed in the standard feeding-inoculation protocol. The ability to form hyphae in the worm model was quantitatively examined (Fig. 3B). After 7 days, nearly all the worms infected only with C. albicans had visible hyphae, whereas only 3.3% of those coinfected with E. faecalis exhibited hyphae. When adding defined numbers of CFU of E. faecalis, we again observed a dose-dependent response in terms of C. albicans filamentation. When high numbers of E. faecalis, 108, were added, about 11.6% of the worms contained visible hyphae, whereas 91.6% had hyphae when 102 CFU of E. faecalis was added.
Protection of C. elegans from C. albicans killing and filamentation is dependent on the E. faecalis inoculum size. (A) C. elegans worms were exposed to C. albicans (SC5314) on solid medium and then placed in liquid medium containing E. faecalis (OG1RF) at the indicated inoculum size. Survival was scored daily, and differences in survival from the control with C. albicans alone were significant at P < 0.001 for 102 and 104 CFU and P < 0.0001 for all others. No number indicates exposure to E. faecalis on solid medium following C. albicans exposure, as was done for Fig. 1A. (B) Inhibition of C. albicans filaments on day 7 depended on the size of the E. faecalis inoculum.
The Fsr quorum-sensing system is partially required for inhibition of C. albicans killing and morphogenesis.Inhibition of hyphal morphogenesis by bacteria has previously been observed (reviewed in reference 2) and is best characterized in studies examining interactions between P. aeruginosa and C. albicans.
The inhibition was shown to require the homoserine lactone (HSL) quorum-sensing molecule (QSM) produced by this Gram-negative bacterium (7). The compound appears to mimic farnesol, a quorum-sensing compound produced by C. albicans that regulates hyphal morphogenesis by acting directly on adenylyl cyclase (47). E. faecalis does not produce HSLs but, rather, uses small peptides as QSMs, such as the gelatinase biosynthesis-activating cluster (GBAP) peptide. GBAP activates the Fsr quorum-sensing system via the FsrB transcriptional regulator (48). Because FsrB is a major regulator of virulence in E. faecalis (23), we asked whether this quorum-sensing system affects C. albicans morphology in the nematode coculture model. As shown in Fig. 4B, a previously constructed fsrB mutant (29) no longer fully inhibited C. albicans hyphal formation and did not fully suppress C. albicans virulence (Fig. 4A). We considered two explanations for this result: first, that inhibition of hyphal formation was due to the direct action of GBAP on the fungal cell, analogous to the effects of HSL molecules produced by Gram-negative bacteria. Alternatively, the inhibitory activity might derive from one or more genes of the Fsr regulon (49).
fsrB, gelE, and sprE are required for full protection from C. albicans killing and filamentation. (A) C. elegans worms were exposed to C. albicans (SC5314) on solid medium and then exposed to E. faecalis OG1RF or fsrB, gelE sprE, gelE, and sprE mutants in liquid medium and scored for survival daily. The mutants did not protect against C. albicans infection as well as wild-type OG1RF (P < 0.0001 for all mutants compared to OG1RF). (B) The mutants were scored for their ability to inhibit C. albicans filamentation on day 7. *, P < 0.05; **, P < 0.01; ***, P < 0.001; NS, not significant.
gelE and sprE are two genes located directly downstream of the fsr genes and regulated by the Fsr system. They encode proteases with major effects on virulence in a variety of infection models. Gelatinase (GelE) is a metalloprotease II with broad substrate specificity, while SprE is a serine protease. For many Fsr-related phenotypes, the effects are at least partially dependent on these proteases (23). To investigate if GelE and SprE were contributing to the ability of E. faecalis to inhibit C. albicans virulence and hyphal morphogenesis, we examined previously constructed single and double mutants (31, 32, 50). The single and double mutants of gelE and sprE were also not able to protect the worm against C. albicans killing as effectively as wild type. There were no significant differences between fsrB, gelE sprE, gelE, and sprE mutants in the survival assay (Fig. 4A). We next examined the effects of these mutants on hyphal morphogenesis in the worm infection model. gelE and sprE mutants were also not able to inhibit filamentation as well as wild type, but they were significantly more effective than the fsrB mutant (for the fsrB mutant versus the gelE sprE mutant, P = 0.0152) (Fig. 4B). We did not observe a significant difference in the ability to prevent filamentation between the double and single mutants, though there was a trend toward less filamentation with the single mutants. These data suggest that the Fsr system's effects on inhibiting C. albicans virulence and morphogenesis are largely, but not completely, dependent on these two proteases. Therefore, it is unlikely that GBAP is affecting C. albicans directly, in contrast to the QSMs produced by P. aeruginosa.
Supernatants of E. faecalis can inhibit C. albicans hyphal morphogenesis in C. elegans.To investigate further the possible involvement of a secreted signal, we asked whether or not filter-sterilized supernatants of E. faecalis cultures were capable of affecting C. albicans hyphal morphogenesis in the infected C. elegans model. Cultures of OG1RF were grown overnight at 37°C in BHI medium, resulting in stationary-phase cultures. The cultures were then filter sterilized and diluted 1:10 into the infection assay medium containing C. elegans worms that had previously been infected with C. albicans by feeding. As shown in Fig. 5A, the OG1RF supernatant added in this manner was extremely effective in inhibiting hyphal formation, with only 3.4% of the animals displaying hyphae after 7 days. In the control in which fresh BHI medium was added, greater than 93.3% of the animals infected with C. albicans contained hyphae after 7 days. These results were comparable to what was observed when wild-type E. faecalis cells were added to the assay medium (Fig. 4B). Next, we examined the effect of the culture supernatant from an fsrB mutant (Fig. 5A and C). Just as we observed when live cells were added, the supernatant from an fsrB mutant was much less effective at inhibiting hyphal formation. The effect was partially dependent on gelE and sprE, as a double mutant displayed an intermediate phenotype between fsrB and wild type. Using the supernatants, it appears that both gelE and sprE are required, as loss of one or the other resulted in wild-type levels of filamentation suppression (Fig. 5A). The result is in contrast to that obtained using cells in which some activity was still apparent in the single mutants (Fig. 4B).
Culture supernatants of E. faecalis inhibit filamentation of host-associated C. albicans. (A) Inhibition of C. albicans (SC5314) filamentation in infected worms with stationary supernatants from OG1RF or fsrB, gelE sprE, gelE, and sprE mutants on day 7. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; NS, not significant. (B) Inhibition of C. albicans filamentation in infected worms with early-, middle-, and late-log-phase supernatants from OG1RF on day 4; (C) inhibition of C. albicans filamentation with various percentages of the indicated E. faecalis supernatants on day 7; (D) inhibition of filamentation with supernatant extracts from 0 to 10 kDa and a boiled extract on day 7; (E) inhibition of C. albicans filamentation in worms infected with OG1RF grown overnight in minimal medium with 100 mM glucose; (F) inhibition of C. albicans filamentation in worms infected with E. faecalis V583 on day 7.
To investigate if the suppressive activity was associated with a certain phase of E. faecalis growth, we investigated the ability of both E. faecalis cells and supernatants to suppress C. albicans filamentation in the worm at different phases of growth (Fig. 5B). A strong growth phase dependence was not observed in supernatants added at full strength; cells and supernatants at early, middle, or late log phase effectively suppressed C. albicans filamentation. The suppression was weaker in early-log-phase and mid-log-phase supernatants when diluted, particularly early-log-phase supernatants (Fig. 5B). We attributed this effect to there being fewer numbers of CFU/ml of culture at the early stages; the results are consistent with a secreted QSM.
The experiments demonstrated that the inhibitory activity is secreted rather than cell associated. To begin preliminary characterization of this substance, we both heat treated and fractionated the supernatant. Boiled supernatant retained inhibitory activity (Fig. 5D), indicating that it is unlikely to be a protein. However, many quorum-sensing peptides, especially modified or circular ones, are heat stable. To estimate the size of the activity, we used centrifugal size-exclusion filters, showing that it is likely to be between 3 and 10 kDa (Fig. 5D). This further excludes GBAP, which has a molecular mass of 1.3 kDa.
Next, we examined whether or not the activity was still produced in cells grown in minimal medium. Using a modified minimal medium appropriate for E. faecalis growth (M9HYG [35]), we observed that OG1RF supernatants inhibited C. albicans filamentation just as effectively as supernatants from OG1RF grown in BHI (Fig. 5E). These data suggest that production of the activity is not dependent on rich medium. We also observed that the inhibitory activity was not limited to strain OG1RF. V583, a vancomycin-resistant clinical isolate frequently used in studies of E. faecalis (27), and its supernatant also inhibited C. albicans hyphal formation (Fig. 5F).
E. faecalis slightly affects C. albicans morphogenesis in vitro.A variety of extracellular conditions stimulate hyphal formation in C. albicans, including serum, CO2, neutral to alkaline pH, and temperatures over 35°C (51). Though it remains unclear what triggers filamentation within the worm, an assay which is notably performed at 25°C, we asked whether E. faecalis or culture supernatants could suppress hyphal growth in vitro. We tested two strong hypha-inducing conditions, rich medium (YPD) supplemented with 10% FBS or tissue culture medium (RPMI) at pH 7.4, both at 37°C (51). As shown in Fig. 6A, incubation of yeast-form C. albicans for 1 h under either of these conditions resulted in an almost universal switch from the yeast to the hyphal morphology. This was largely unaffected by the presence of E. faecalis cells or sterilized supernatants from overnight cultures (representative images are shown in Fig. 6C and D). There was a small shift from hyphal cells to the pseudohyphal morphology in the presence of E. faecalis, but the inhibitory activity observed in the worm assay did not notably suppress the switch to filamentous growth under these strongly inducing conditions. To ask whether the difference in temperature between the worm (25°C) and in vitro (37°C) assays explains the lack of inhibition, we embedded C. albicans cells in top agar, a condition that induces filamentation even at 25°C (51). Robust hyphal growth was observed under this condition, regardless of whether the top agar was made from fresh BHI or from expended supernatants of OG1RF cultures (data not shown). These data indicate that the lower incubation temperature was not why E. faecalis was able to inhibit C. albicans hyphal morphogenesis in the worm model.
E. faecalis does not inhibit C. albicans morphogenesis in vitro. (A) Overnight cultures of C. albicans SC5314 were diluted into RPMI or YPD–10% serum with or without E. faecalis OG1RF and incubated at 37°C for 2 h. Cultures were examined microscopically and scored for morphology (n > 200 cells in two experiments for each condition). (B) Overnight cultures of SC5314 were diluted into YPD–10% serum diluted with either fresh BHI medium or sterilized supernatants from saturated OG1RF cultures grown in BHI. Fungal morphology was assessed as described for panel A. (C) Representative images from the experiment quantified in panel A. (D) Representative images from the experiment quantified in panel B.
E. faecalis supernatants affect C. albicans morphology in a biofilm.C. albicans avidly forms biofilms composed of a high proportion of hyphal cells mixed with yeast and pseudohyphae (reviewed in reference 52). Similar to bacterial biofilms, these are a serious problem with many abiotic implanted materials used clinically, which are often fouled by polymicrobial structures (53). To address whether the presence of E. faecalis affects cellular morphology within biofilms, we grew C. albicans on squares of medical silicone, as described previously (42, 43), in buffered Spider medium at 37°C. We stained these biofilms with the chitin-binding dye calcofluor white and imaged them from the top down. The high proportion of hyphal cells in the native biofilms is clear from these images (Fig. 7A and C) and from the counting of individual cells within the biofilm for morphology (Fig. 7E). When green fluorescent E. faecalis was added to the developing biofilm, it incorporated into the structure, but its presence caused C. albicans to remain mostly in the yeast form (Fig. 7B). Suppression of hyphal growth within the biofilm could be observed in the presence of supernatants from E. faecalis cultures (Fig. 7D), as was the case in the nematode model as well. Quantitation of cellular morphology revealed that 72% of cells grown in the absence of E. faecalis supernatant were in the hyphal form, and this proportion dropped to 30% in the presence of supernatant (Fig. 7E). Thus, a secreted bacterial compound inhibits biofilm-associated hyphal growth.
E. faecalis inhibits C. albicans morphogenesis in a biofilm. Overnight cultures of C. albicans (strain SC5314) were diluted into buffered Spider medium with 20% serum with or without E. faecalis and incubated at 37°C for 24 h. (A) Representative image of a C. albicans biofilm stained with calcofluor white (blue peripheral staining); (B) representative image of a biofilm comprised of calcofluor white-stained C. albicans (blue) and GFP-expressing E. faecalis (green; strain SD234); (C) representative DIC image of C. albicans forming a biofilm; (D) representative DIC image of C. albicans forming a biofilm in the presence of E. faecalis OG1RF supernatant; (E) percentage of C. albicans cells displaying a yeast, hyphal, or pseudohyphal morphology within a biofilm with and without E. faecalis supernatant. At least 10 fields and more than 500 cells were counted in three separate experiments. ***, a significant difference (P < 0.0001) between the E. faecalis supernatant addition compared to the control.
Though C. albicans continued to form a biofilm in the presence of E. faecalis, we considered the possibility that the morphogenesis defects were due to some toxic effect on the fungus. For example, P. aeruginosa and A. baumannii kill C. albicans hyphae, contributing to the fungus remaining in the yeast form (6, 21). The mixed E. faecalis-C. albicans biofilm was stained with live/dead stain (21), but no significant death for either species was observed (data not shown). There was no apparent fungal cell death when cocultured under strong hypha-inducing conditions either (data not shown), indicating that hyphae are not specifically sensitive to the bacteria, unlike with P. aeruginosa.
A variety of E. faecalis genes contribute to inhibiting C. albicans morphogenesis.It is clear from the killing and hyphal morphogenesis assays that the ability to suppress filamentation is not completely dependent on the Fsr system (Fig. 4). The mutants still partially protected C. elegans from killing and partially inhibited hyphal formation. To identify other factors involved in this process, we carried out a screen using an ordered library of E. faecalis transposon mutants (33, 45). The mutants were grown overnight and then diluted 1:10 into the liquid culture assay containing C. elegans worms infected by feeding on plates of C. albicans.
Filamentation was scored at day 4 and day 7 of the experiment. Additionally, any worm survival at day 7 was noted (Table 1). Mutants displaying a reduced ability to inhibit C. albicans filamentation were retested twice. Mutants with significantly reduced growth rates in the liquid medium used for the assay were excluded from further analysis. In total, we report 10 mutants recovered from the screen that showed a significant loss of ability to inhibit hyphal morphogenesis but were not defective in growth in the medium used for the screen (Table 1). Note that because the ordered library was generated via transposon mutagenesis, the observed phenotypes may be due to polar effects on expression of downstream genes (45). Genes with predicted functions in metabolism, competence, and transcriptional regulation were identified in the screen and are fully discussed below.
E. faecalis transposon mutants with filament-suppressing defectsa
DISCUSSION
E. faecalis and C. albicans are found in many of the same niches of the human body, including the oral cavity and the GI tract. They are also commonly found together in clinical samples of polymicrobial infections. Despite the many descriptions of C. albicans and E. faecalis inhabiting the same places in the natural setting of the mammalian host, no study has specifically looked at how these two microbes might interact and affect one another in the host environment. The work presented herein describes for the first time a biological interaction between these two pathogens in the model host C. elegans. Specifically, we observed that the killing of C. elegans by monomicrobial infection with either pathogen is significantly attenuated during coinfection. The inhibition of killing by C. albicans could be largely attributed to the inhibition of hyphal morphogenesis, a major mechanism of worm killing by C. albicans (18). Interestingly, the presence of C. albicans appeared to protect E. faecalis from cell death. Additionally, damage to the apical intestinal membrane was not apparent in the coinfection (Fig. 2). Overall, the presence of both microbes appears to promote a peaceful interaction with the host, suggestive of them playing a commensal rather than a pathogenic role.
Further study demonstrated that the inhibitor of C. albicans hyphal morphogenesis was a heat-stable secreted product from E. faecalis of between 3 and 10 kDa, and generation of the signal was at least partially dependent on Fsr. The inhibitor of hyphal morphogenesis was also active in preventing filamentation in C. albicans biofilms but did not affect filamentation in shaking cultures of the yeast. While the manuscript was in preparation, another study reported the isolation of an anti-Candida protein (ACP) from an isolate of E. faecalis. This compound is different from the substance described here in that it kills C. albicans, is not resistant to boiling, and has a molecular mass of 43 kDa (54). To further characterize the nature of the secreted signal described in this study, a screen of a transposon insertion library (33, 45) was undertaken to isolate E. faecalis mutants incapable of inhibiting hyphal morphogenesis.
Many of the mutants unable to fully inhibit C. albicans filamentation had transposon insertions in genes with predicted functions in metabolism. For example, two genes may have products involved in amino acid transport (EF0860 and EF1594) and biosynthesis (EF1568). Some of these genes encode aspects of bacterial sugar phosphotransferase systems (PTSs), including the regulator of carbon catabolite repression, CcpA (OG1RF0141, EF0455, EF1741). Interestingly, ccpA has been identified as a major regulator of virulence in group A Streptococcus (55). Also of interest was comEA, whose gene product is predicted to form part of the complex that imports extracellular DNA during the competent state. In previous work, another QSM from Streptococcus mutans, competence-stimulating peptide (CSP), was shown to inhibit C. albicans filamentation (56). However, a naturally competent state for E. faecalis has never been reported in the literature. The strain used in this study, OG1RF, was recently reported to have a complete competence operon with similarity to the comY operon of S. mutans (33). In the first sequenced strain of E. faecalis, V583, the comY operon is disrupted due to a large phage insertion and a premature stop codon in one of the genes (33, 57). Additionally, and unlike V583, OG1RF contains orthologs of the ComDE two-component system, necessary for regulating the competence operon. No CSP with similarity to the one in S. mutans was identified, but two small open reading frames are present within the comDE operon of OG1RF, and it is possible that they encode a unique CSP (33). Because V583 lacks comDE and the complete comY locus, we postulated that it might be deficient in suppressing C. albicans hyphal formation. However, as shown in Fig. 5F, V583 was just as effective in suppressing hyphal morphogenesis as OG1RF, making it less likely that a CSP specific to E. faecalis is involved. It remains formally possible that a peptide signal like CSP, encoded elsewhere in the genome, is the cause of suppression of C. albicans morphogenesis. Identifying the compound will require further work. However, the fact that the signal is produced in minimal medium (Fig. 5E) will aid any purification attempt.
Another major question is how the signal produced by E. faecalis mechanistically acts to inhibit hyphal morphogenesis by C. albicans. Numerous signaling pathways regulate hyphal morphogenesis in C. albicans, including the cyclic AMP (cAMP)-dependent protein kinase and pheromone-responsive mitogen-activated protein kinase pathways and other pathways responding to specific inducing signals (reviewed in references 58 and 59). These pathways culminate in no fewer than a dozen transcriptional regulators, and which of these are most important for filamentation within the worn has not been addressed. The P. aeruginosa HSL acts through a direct interaction with the fungal adenylyl cyclase, reducing cAMP levels, while CO2/bicarbonate binds to the same protein to induce hyphal growth (47, 60, 61). Further genetic analysis will be required to identify the mechanism by which E. faecalis inhibits hyphal growth in worms and biofilms.
A striking finding in this work is that these two prominent and important pathogens appear to inhibit each other's virulence in this model: not only does E. faecalis attenuate killing by C. albicans, but a subsequent exposure to the fungus attenuates E. faecalis killing as well. By TEM, we observed that they appear to prevent each other's damage to the nematode's intestine as well (Fig. 2). Beyond underscoring the specific interaction between these species, it also raises the question as to what evolutionary advantage this might confer. As both these species are normally benign components of the microbiota, causing serious pathology only in immunocompromised individuals, their interaction may be driven from benefits in the gut, the primary commensal site. For instance, C. albicans has genetic mechanisms to suppress its population size in the gut (62). Though filamentation is critical for virulence, the hyphal form is also more immunogenic; thus, responding to bacterial signals to suppress hyphal morphogenesis might be a mechanism to trade virulence for long-term commensal fitness. Both species may benefit from the effects of the other on the broader microbiota, through either direct (specific interactions) or indirect (through competition for usable nutrients) means. Huffnagle and colleagues tracked the gut bacterial microbiota in mice during recovery from a transient antibiotic suppression in the presence or absence of C. albicans. Strikingly, C. albicans changed the bacterial profile such that it was dominated by Enterococcus (17). Prokaryotic-fungal interactions such as the one that we describe are clearly specific and significant but as yet poorly understood. Further investigations of these interactions in vitro and in multiple animal models, as well as increased attention to the eukaryotic microbiota, are clearly warranted.
ACKNOWLEDGMENTS
This work was supported by Public Health Service grants R01AI076406 and R56AI093699 to D.A.G. and R01AI075091 to M.C.L. from the National Institute of Allergy and Infectious Diseases.
For the TEM imaging, we thank S. Kolodziej and P. Navarro, Department of Pathology and Laboratory Medicine EM Laboratory, The University of Texas Health Science Center at Houston. We thank John Collette for strain JRC52 and Barbara Murray for strains TX5266, TX5128, TX5264, and TX5243.
FOOTNOTES
- Received 28 August 2012.
- Returned for modification 24 September 2012.
- Accepted 18 October 2012.
- Accepted manuscript posted online 31 October 2012.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.