Fungal Symbionts Produce Prostaglandin E2 to Promote Their Intestinal Colonization

Candida albicans is a ubiquitous fungal symbiont that resides on diverse human barrier surfaces. Both mammalian and fungal cells can convert arachidonic acid into the lipid mediator, prostaglandin E2 (PGE2), but the physiological significance of fungus-derived PGE2 remains elusive. Here we report that a C. albicans mutant deficient in PGE2 production suffered a loss of competitive fitness in the murine gastrointestinal (GI) tract and that PGE2 supplementation mitigated this fitness defect. Impaired fungal PGE2 production affected neither the in vitro fitness of C. albicans nor hyphal morphogenesis and virulence in either systemic or mucosal infection models. Instead, fungal production of PGE2 was associated with enhanced fungal survival within phagocytes. Consequently, ablation of colonic phagocytes abrogated the intra-GI fitness boost conferred by fungal PGE2. These observations suggest that C. albicans has evolved the capacity to produce PGE2 from arachidonic acid, a host-derived precursor, to promote its own colonization of the host gut. Analogous mechanisms might undergird host-microbe interactions of other symbiont fungi.


INTRODUCTION
Candida albicans is one of the most successful fungal symbionts in humans, colonizing 40-80% of individuals in industrialized nations and typically representing the predominant species within the fungal microbiota (Odds, 1987;MacCallum, 2010). In healthy people, C. albicans dwells on diverse barrier sites of the body, including the oral cavity, skin, female reproductive tract, and the intestines, where it does not cause symptomatic disease (Odds, 1987;MacCallum, 2010). However, C. albicans can turn pathogenic and result in mucosal or life-threatening invasive bloodstream infections under a variety of conditions that compromise host immunity, damage barrier surfaces, or disrupt the microbiota (Brown et al., 2012;Fan et al., 2015). Molecular profiling studies have identified C. albicans strains in the gastrointestinal (GI) tract as the provenance of hematogenous isolates during systemic infection (Nucci and Anaissie, 2001;Odds et al., 2006). Understanding the fungal and host factors that modulate the ability of C. albicans to thrive in different host niches as a pathogen or symbiont could therefore inform our development of new methods of combatting human fungal infections.
Screens of C. albicans mutants in mice have revealed several fungal pathways that mediate colonization of the host gut. Salient pathways include the regulation of the acquisition of iron and nutrients, in particular carbon and nitrogen, cell wall remodeling that augments fungal adhesion to intestinal mucins, and morphogenetic changes (White et al., 2007;Rosenbach et al., 2010;Chen et al., 2011;Pande et al., 2013;Perez et al., 2013;Sem et al., 2016;Bohm et al., 2017;Mamouei et al., 2017). Fungal colonization is, in turn, restricted by the host immune system.
Intestinal epithelial cells (IECs) produce antimicrobial peptides (AMPs) that kill C. albicans (Fan et al., 2015), and gut-resident CX 3 CR1 + macrophages control the growth of C. albicans and other symbiont fungi in mice (Leonardi et al., 2018). Accordingly, some of the aforementioned genetic programs in C. albicans might also promote intestinal symbiosis by calibrating the host immune response toward the fungus or vice versa. For instance, cell wall remodeling can mask the exposure of β-glucan, an essential component of fungal cell walls, to host immunological sensing via dectin-1 (Ballou et al., 2016;Pradhan et al., 2018), and recognition of surface β-glucan by dectin-1 is strongly correlated with the competitive fitness of Candida species in the murine intestine (Sem et al., 2016). In addition, C. albicans upregulates the key morphogenetic regulator, EFG1, upon colonizing the gut, and EFG1 is required for C. albicans to overcome host immunological pressures to persist in the GI tract long-term (Pierce and Kumamoto, 2012).
Of note, many fungal species, including C. albicans, also produce PGE 2 in vitro when supplied exogenously with the host-derived precursor, arachidonic acid (AA) (Noverr et al., 2002). C. albicans and most fungal species lack AA, which is derived from mammalian membrane phospholipids-but C. albicans can induce host cells to release AA (Castro et al., 1994). Moreover, fungus-derived PGE 2 exhibits immunomodulatory properties analogous to those of mammalian PGE 2 in vitro (Noverr et al., 2001) and in vivo (Kim et al., 2014). PGE 2 has also been reported to promote yeast-hyphae morphological switching (Kalo-Klein and Witkin, 1990;Noverr et al., 2001) and biofilm formation in vitro (Alem and Douglas, 2004)traits commonly associated with fungal virulence. Given that the fungus is more typically associated with the host as a benign symbiont and inter-individual transmission often occurs in the absence of overt infection (Russell and Lay, 1973;Odds et al., 2006), it seems unlikely, however, that C. albicans has evolved the cellular machinery for synthesizing PGE 2 from an exogenous substrate for the sole purpose to increase its pathogenic potential or to cause immune-related disorders in the host. We thus posit that fungus-derived PGE 2 might mediate asymptomatic colonization by C. albicans of the mammalian intestinal environment. Indeed, we found that a C. albicans mutant with impaired PGE 2 production, while not exhibiting any defects in in vitro fitness or in vivo virulence, suffered a loss of competitive fitness relative to the wildtype (WT) strain in the murine intestine. We further report evidence suggesting that fungal PGE 2 promotes intestinal symbiosis by C. albicans in part by reducing its killing by phagocytes.

Fungal Strains
Fungal strains used in this study are listed in Table S1. All fungal strains were grown in yeast peptone dextrose (YPD) media [1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) glucose] overnight at 30 • C prior to use in experiments unless otherwise stated.

Construction of Fungal Strains
Fungal mutants were generated using the SAT1-flipping strategy as previously described (Sasse and Morschhauser, 2012;Sem et al., 2016). Briefly, the plasmid pSFS2 containing the SAT1 flipper cassette was used to disrupt the open reading frame of the OLE2, FET3, and FET31 genes of C. albicans. Sequences (500-600 bp) flanking the targeted genes were amplified using the primers listed in Table S2. Gene deletion was verified by quantitative polymerase chain reaction (qPCR) using the primers listed in Table S2.

Mice and Treatments
C57BL/6J and Ccr2-KO (B6.129S4-Ccr2 tm1Ifc /J) mice were maintained under specific-pathogen-free conditions at the A * STAR Biological Resource Centre (BRC) in Biopolis, Singapore. Adult mice (6-8 weeks of age) of either sex were used in all experiments. All animal experiments were approved by the BRC Institutional Animal Care and Use Committee (IACUC) in accordance with the guidelines of the Agri-Food & Veterinary Authority (AVA) and the National Advisory Committee for Laboratory Animal Research (NACLAR), Singapore. To deplete phagocytes, mice were injected intraperitoneally (i.p.) with control or clodronate-containing liposomes (Liposoma B.V.) every 2 days starting from 2 days prior to the inoculation of mice with fungi. To perform PGE 2 supplementation, mice were injected i.p. with 16.7 µg/kg of 16,16-dimethyl-PGE 2 (Cayman Chemical) or vehicle [methyl acetate (Sigma)] daily, starting from day 0, throughout the course of fungal colonization.

Measurement of Fungal PGE 2 Production
Fungal PGE 2 production was measured by enzyme-linked immunosorbent assay (ELISA) as previously described (Yano and Fidel, 2011). Briefly, fungi were inoculated into fresh YPD to an optical density (OD) at 600 nm of 0.2 and grown at 37 • C with shaking at 250 rpm for 2-2.5 h before peroxide-free arachidonic acid (Cayman Chemical) was added to a final concentration of 0.5 mM. Fungal growth was continued for another 48 h before supernatants were harvested. Levels of PGE 2 metabolite (PGEM), a stable derivative of PGE 2 , in fungal supernatants were assayed using the PGEM EIA kit (Cayman Chemical) as per manufacturer instructions.

Quantification of Fungal Titers
Fungal titers for all experiments were determined by plating fungal cultures, tissue homogenates, or lavage fluids on YPD agar containing 50 U/ml penicillin (Gibco) and 50 µg/ml streptomycin (Gibco). Plates were incubated at 37 • C for 24-48 h and colony-forming-units (CFUs) enumerated.

Murine Infection With C. albicans
To elicit systemic candidiasis, mice were injected intravenously via the tail vein with a sublethal dose of C. albicans (2 × 10 5 cells). Oropharyngeal candidiasis (OPC) was established as previously described (Solis and Filler, 2012). Briefly, mice were injected subcutaneously (s.c.) with 100 mg/kg cortisone 21acetate (Sigma) in 0.05% (v/v) Tween R -80 (Sigma)/phosphate buffered saline (PBS) 1 day prior to infection to induce transient immunosuppression. Mice were subsequently sedated with 150 mg/kg ketamine and 10 mg/kg xylazine before sublingual inoculation with a small cotton ball saturated with C. albicans cells (2 × 10 7 cells/ml) for 75 min. Vulvovaginal candidiasis (VVC) was established as previously described (Yano and Fidel, 2011). Briefly, female mice were treated s.c. with 0.1 mg of β-estradiol 17-valerate (Sigma) in sesame oil (Sigma) 3 days prior to infection. β-estradiol injections were maintained weekly thereafter. Mice were intravaginally inoculated with 5 × 10 4 C. albicans cells in 20 µl of PBS. Infected mice were monitored daily (systemic candidiasis and OPC) for survival and weight loss. Tongues from mice with OPC were homogenized by pressing the tissues through a 40 µm cell strainer. Homogenates were plated to assess fungal burden. Mice with VVC were intravaginally lavaged with 100 µl of PBS. Vaginal lavage fluid was plated to determine fungal burden or used for neutrophil quantification by flow cytometry as described below under "Antibodies and flow cytometry."

Intestinal Colonization With C. albicans
Three to four days prior to inoculation, mice were placed on drinking water containing 1 g/l of penicillin G (Sigma) and 2 g/l of streptomycin (Sigma) (Sem et al., 2016). Antibiotics were changed twice weekly and maintained throughout the course of the experiment. Mice were gavaged with 10 7 C. albicans cells to establish intestinal colonization. To determine the intestinal fungal load, stools were collected, weighed, homogenized in PBS, and plated.

In vivo Intestinal Competition
Intestinal fungal competition experiments were performed as previously described (Sem et al., 2016). Briefly, mice were gavaged with 2 × 10 7 cells comprising the SC5314-dTomato strain and the test strain (WT or ole2/ole2) at a 1:1 ratio. The relative frequency of the test strain at each time point was determined by counting the number of fluorescent vs. non-fluorescent colonies under a fluorescence stereomicroscope (Olympus). The fitness coefficient of each test strain in each mouse was calculated as previously described (Sem et al., 2016). To control for inter-experimental variability, fitness coefficients were normalized to the mean of the coefficients for the WT strain within each treatment group (e.g., clodronate-treated group) for each experiment. Normalization was performed by calculating the difference between each fitness coefficient value and the mean for the WT strain.

In vitro Competition
In vitro competition experiments were performed as previously described (Sem et al., 2016). Briefly, 1.25 × 10 6 cells comprising the SC5314-dTomato strain and the test strain (WT or ole2/ole2) at a 1:1 ratio were inoculated into 50 ml of media [YPD or yeast nitrogen base (YNB) (BD Difco) with amino acids] containing 0.5 mM of arachidonic acid or not and grown at 37 • C. Every 8 or 16 h, fungal cultures were harvested, plated, and 1.25 × 10 6 cells re-inoculated into 50 ml of fresh media as described above. Relative proportions and fitness coefficients of each test strain were derived as described above.

Infection of Macrophages With C. albicans
Fungal resistance to macrophages at 24 h was determined as previously described (Sem et al., 2016). Briefly, confluent J774A.1 macrophages (ATCC) were infected with C. albicans cells at low multiplicities of infection (MOIs) ranging from 0.005 to 0.0003125 to ensure that essentially 100% of fungi were engulfed by the macrophages at 3 h post-infection. Infected macrophages were incubated at 37 • C and fungal colonies enumerated after 24 h. Fungi were also seeded in wells without macrophages at varying densities to assess the number of CFUs plated in each experiment. The percentage of fungal cells that escaped killing by macrophages was calculated as previously described (Noble and Johnson, 2005). For short-term fungal killing assays, J774A.1 macrophages were infected at a MOI of 0.1. Extracellular fungi were removed 1 h post-infection by gently washing the macrophages with PBS. Macrophages were lysed with 0.02% Triton X-100 after an additional 2 h of incubation at 37 • C and cell lysates plated on YPD to enumerate the intramacrophage fungal load. Where indicated, 100 nM of PGE 2 (Cayman Chemical) or the corresponding volume of vehicle (ethanol) was added to the cultures.

In vitro Cytotoxicity Assay
Confluent J774A.1 macrophages were infected with C. albicans cells at a MOI of 1 for 6 h. LDH released into the culture media was measured using the CytoTox 96 R Non-Radioactive Cytotoxicity Assay (Promega) as per kit instructions and normalized to LDH levels in uninfected cells lysed with 1% Triton X-100 (Sigma).

Measurement of LL-37 Production From IECs
Confluent human colonic epithelial cell lines HT-29 and Caco-2 were infected with C. albicans cells at a MOI of 0.5 for 24 h. LL-37 secreted into the culture media was measured by the LL-37 ELISA kit (Cusabio) as per manufacturer instructions.

Assessment of Fungal Resistance to mCRAMP
Fungal resistance to mCRAMP was evaluated as previously described (Sem et al., 2016). Briefly, overnight cultures of C. albicans were inoculated into fresh YPD in the presence of 400 nM PGE 2 (Cayman Chemical) or vehicle (ethanol) for 3 h. Cells were washed, resuspended in the appropriate concentration of mCRAMP 1−39 (Synpeptide), and incubated at 37 • C for 2 h. Cells were washed extensively after mCRAMP exposure and stained with 50 µM of propidium iodide (PI) (Thermo Fisher Scientific) for 15 min. The proportion of live PI − fungal cells was determined by flow cytometry.

Measurement of Transcript Levels by Quantitative PCR
RNA was extracted from a small segment of the distal colon harvested from C. albicans-colonized mice 7 days after inoculation. For in vitro infection experiments, J774A.1 macrophages were infected with fungi at the indicated MOIs and the macrophages lysed with TRIzol TM (Invitrogen) 6 h post-infection. Both colon and macrophage samples were homogenized in TRIzol TM using 0.5 mm glass beads (Sigma) and RNA extracted as per manufacturer instructions. cDNA was prepared from RNA using SuperScript TM III reverse transcriptase (Invitrogen) as per manufacturer instructions. Levels of transcripts of interest in cDNA were measured by qPCR using PerfeCTa SYBR R Green FastMix (Quantabio) and normalized to those of Rpl13a.

Isolation of Colonic Lamina Propria Cells
To remove the epithelial layer, colons were sliced open longitudinally to expose the luminal surface and shaken in DMEM containing 1 mM dithiothreitol, 5 mM EDTA, and 3% (v/v) FCS for 20 min at 37 • C. Remaining tissue was washed in cold DMEM, minced, and digested in DMEM containing 1.5 mg/ml collagenase D (Roche), 50 µg/ml DNase I (Roche), and 2% (v/v) FCS for 45 min at 37 • C. Digested tissue was washed and filtered twice to obtain a single-cell suspension.

PGE 2 -Defective C. albicans Mutant Exhibits No Fitness Defects in vitro
PGE 2 production has been described for C. albicans and other pathogenic fungi such as Cryptococcus neoformans and Aspergillus species (Noverr et al., 2002) as well as for some gut-dwelling symbiont Candida species (Shiraki et al., 2008;Ells et al., 2011;Mishra et al., 2014;Grozer et al., 2015). We measured in vitro PGE 2 production across an array of symbiont fungi and confirmed that all species tested, including Saccharomyces cerevisiae, secreted PGE 2 when supplied with exogenous AA (Figure 1A), implying a PGE 2 biosynthetic pathway conserved across the majority of fungal species. We next sought a C. albicans mutant with altered PGE 2 production which we could use to study the physiological function of fungus-derived PGE 2 . Although the PGE 2 biosynthetic pathway in C. albicans and other fungi remains largely uncharacterized, three genes, OLE2, FET3, and FET31, have been described to promote PGE 2 production in vitro (Erb-Downward and Noverr, 2007;Krause et al., 2015). We generated knockout (KO) mutants for each of these genes and assessed the mutants' capacities for PGE 2 secretion in vitro at 37 • C. As expected, PGE 2 production by all fungal strains was strictly dependent on the provision of exogenous AA ( Figure 1B). However, only the ole2/ole2 mutant exhibited reproducibly diminished PGE 2 secretion. Concordant with a previous report (Erb-Downward and Noverr, 2007), PGE 2 production was not completely abrogated by OLE2 ablation, implying the existence of alternative pathways to PGE 2 generation by C. albicans. We thus focused on the ole2/ole2 mutant for all subsequent experiments.
To determine whether fungal PGE 2 modifies fungal fitness in vitro, we maintained either the WT or the ole2/ole2 strain in competition with a fluorescently tagged WT strain (SC5314-dTomato) in serially passaged liquid batch cultures. The abundance of each strain relative to that of SC5314-dTomato was determined at regular intervals by plating the cultures on yeast peptone dextrose (YPD) agar followed by scoring of fluorescent and non-fluorescent colonies. The fitness of each strain can in turn be rigorously quantified by deriving a fitness coefficient from the rate of change of the relative abundance of each strain FIGURE 1 | In vitro PGE 2 production by various C. albicans strains. Levels of PGE 2 metabolite, a stable derivative of PGE 2 , in the supernatants of in vitro cultures of various fungal species (A) or C. albicans strains (B), were assayed by ELISA. Arachidonic acid (AA) was added (+) or not (-) to the cultures. Data pooled from 3 to 7 independent experiments. Mean ± s.d. *p < 0.05; One-way ANOVA with Dunnett's multiple comparisons test. (Sem et al., 2016). Even after 80 h [equivalent to ∼44 and 63 generations for growth in YNB media and YPD, respectively) of serial subculturing, we detected no differences in competitive fitness between the ole2/ole2 and WT strains in both the nutrientrich YPD broth and the nutrient-poor YNB media, whether or not exogenous AA was supplied (Figure 2, Figure S1). Therefore, fungal PGE 2 does not appear to be important for fitness in vitro.

Fungal PGE 2 Is Not Required for Virulence
Since PGE 2 was reported to increase filamentation and biofilm formation in vitro (Noverr et al., 2001;Alem and Douglas, 2004;Kim et al., 2014), we asked whether defective PGE 2 synthesis by the ole2/ole2 mutant influences its virulence. Surprisingly, the ole2/ole2 mutant was fully capable of hyphal formation in vitro, both in response to fetal calf serum (FCS) and when plated on Spider agar, regardless of the presence of AA (Figures 3A,B).
Moreover, the ability of C. albicans to damage host cellswhose cell membranes are an abundant source of endogenous AA (Martin et al., 2016)-evidenced by the release of lactate dehydrogenase from infected murine macrophages, was not significantly altered by fungal PGE 2 ( Figure 3C). Nonetheless, there exist occasional discrepancies between in vivo virulence and in vitro pathogenic traits (Noble et al., 2010). To clarify the role of fungus-derived PGE 2 during in vivo infections, we adopted murine models of hematogenously disseminated candidiasis and mucosal candidiasis of the oropharyngeal cavity and vaginal tract, all of which engage distinct anti-fungal immunological responses (Fidel, 2005;Conti et al., 2009;Kashem et al., 2015). The ole2/ole2 mutant manifested comparable virulence as the WT strain in all tissues tested, across a variety of pathological parameters, including overall survival, fungal burden, weight loss, and neutrophil recruitment (Figures 3D-F,  Figure S2). Our in vitro and in vivo infection data are hence not consistent with a role for fungal PGE 2 in mediating C. albicans virulence.

Fungal PGE 2 Promotes Intestinal Colonization
C. albicans is most commonly associated with the host as a symbiont in the gut. We therefore sought to evaluate whether fungal PGE 2 contributes to the fitness of C. albicans in the intestines. Mice are usually refractory to colonization by C. albicans but can be induced to harbor high titers of fungi for sustained durations if treated with antibiotics to deplete their indigenous microbiota (Koh et al., 2008;Fan et al., 2015). Colonization of the murine GI tract with C. albicans represents a bona fide model for fungal symbiosis with the mammalian host because it does not lead to symptomatic disease and invasive hyphae are rarely detected within the intestinal fungal population (White et al., 2007;Koh et al., 2008;Bohm et al., 2017).
To assess the intestinal fitness of the ole2/ole2 mutant, we inoculated antibiotics-treated mice with either ole2/ole2 or WT C. albicans in competition with SC5314-dTomato and tracked the relative frequency of each strain over time. The ole2/ole2 mutant was significantly compromised in its competitive fitness relative to the WT strain, starting as early as 3 days post-inoculation, as evidenced by its lower relative frequency in the fecal fungal load and its lower fitness coefficient, determined as per our in vitro competitive fitness experiments ( Figure 4A, Figure S3A). Although we were not able to test a reconstituted ole2/ole2::OLE2 mutant, colonization of mice with another independently-derived ole2/ole2 strain recapitulated the inferior competitive fitness of C. albicans in the GI tract in the absence of OLE2 (Figures S3B,C). While we consistently observed a lower fitness coefficient for the ole2/ole2 mutant relative to WT C. albicans across independent experiments, the fitness coefficient differential between the two strains varied in magnitude between experiments (Figure S3), possibly due to variation across different cohorts of mice. To facilitate a comparison of fungal fitness across pooled groups of mice, we normalized fitness coefficients to the mean coefficient of the WT strain within each experiment, thereby removing the batch effects on fungal fitness. This analysis confirmed a significant impairment in the competitive fitness of the PGE 2 -deficient ole2/ole2 mutant in the murine gut ( Figure 4B).
To determine if the loss of competitive fitness of the ole2/ole2 strain in the gut is PGE 2 -dependent, we supplemented C. albicans-colonized mice daily with 16,16-dimethyl-PGE 2 (dmPGE 2 ), a stable analog of PGE 2 . Indeed, dmPGE 2 treatment partially restored the competitive fitness of the ole2/ole2 mutant (Figures 4C,D, Figure S3D), in support of a role for fungal PGE 2 in mediating intestinal fitness of C. albicans under steady-state conditions. In contrast, dmPGE 2 provision did not further enhance the competitive fitness of the WT strain. Taken together, these data indicate that fungal PGE 2 exerts a specific function for C. albicans: promote its colonization of the host GI tract.

Fungal PGE 2 Does Not Affect AMP Production or Fungal Resistance to AMPs
AMPs such as mouse cathelicidin-related antimicrobial peptide (mCRAMP), encoded by Camp, have been demonstrated to mediate intestinal resistance to C. albicans colonization (Fan et al., 2015). We postulated that fungal PGE 2 might promote intestinal fitness of C. albicans by influencing AMP production from IECs or fungal resistance to AMPs. To address the former hypothesis, we stimulated human colonic IEC lines with WT C. albicans or the ole2/ole2 mutant but observed no difference in the secretion of LL-37, the human ortholog of mCRAMP, by the IEC lines ( Figure S4A). Indeed, infection of the IEC lines by fungi in vitro did not induce appreciable production of LL-37 above the background level. We also assessed Camp transcript levels in the colons of mice colonized with either WT or ole2/ole2 C. albicans but found no difference in the quantities of Camp mRNA induced by the two fungal strains in vivo ( Figure S4B). Finally, we exposed fungi grown in the presence or absence of PGE 2 to mCRAMP in vitro to evaluate the effects of PGE 2 on fungal resistance to AMPs. Both the WT and ole2/ole2 strains manifested comparable susceptibility to mCRAMP, and the addition of PGE 2 did not significantly alter fungal resistance to mCRAMP (Figure S4C). Hence the collective evidence does not support a role for fungus-derived PGE 2 in altering AMP production from IECs or fungal resistance to AMPs.

Fungus-Derived PGE 2 Reduces Killing of C. albicans by Intestinal Phagocytes
Since intestinal macrophages limit fungal colonization of the gut (Leonardi et al., 2018) and tissue-resident phagocytes are required for control of fungal infection in various tissues (Kashem et al., 2015;Dominguez-Andres et al., 2017;Sparber et al., 2018), we postulated that fungal PGE 2 might potentiate fungal fitness by acting on the fungi themselves and/or host tissue phagocytes to improve the ability of C. albicans to evade killing by phagocytes. To elucidate the relationship between fungal PGE 2 , tissular phagocytes, and fungal fitness in the host, we assessed the competitive fitness of the WT and ole2/ole2 strains in mice depleted of phagocytes. We first treated mice with clodronate-containing liposomes, which is toxic to a variety of phagocytic leukocytes, including macrophages and dendritic cells (DCs) (Weisser et al., 2012). In our experiments, clodronate treatment depleted colonic CD11b + dendritic cells (DCs) but not macrophages, CD11b − DCs, or non-phagocytic leukocytes (Figures S5A,B). The poor efficiency of macrophage ablation in clodronate-treated mice could possibly arise from differential accessibility and susceptibility of macrophages in the colon to intraperitoneally-administered clodronate relative to those from other tissue sites. Nonetheless, clodronate treatment was sufficient to abrogate the fitness defect of the ole2/ole2 mutant visà-vis the WT strain that was observed in control mice, thereby implicating CD11b + DCs as another tissue-resident phagocytic population that restricts C. albicans in the gut (Figures 5A,B,  Figure S6A). We further validated the preceding results by repeating the fungal competitive colonization experiment in a genetic model of phagocyte depletion. CCR2 is a chemokine receptor required for efficient reconstitution of the intestinal macrophage and CD11b + DC compartments (Bain et al., 2014;Scott et al., 2015). Consequently, Ccr2-KO mice display a paucity of both macrophages and CD11b + DCs but not CD11b − DCs or lymphocytes in the colon (Figures S5C,D). In agreement with our findings in clodronate-treated mice, Ccr2 deficiency abolished the fitness difference between the WT and ole2/ole2 strains observed in WT mice (Figures 5C,D, Figure S6B). Our results collectively support a role for fungal PGE 2 in reducing the eradication of C. albicans by host intestinal phagocytes, including macrophages and CD11b + DCs. We thus sought to compare the ex vivo fungicidal activity of host phagocytes toward WT and ole2/ole2 C. albicans. Owing to difficulties in isolating sufficient numbers of primary macrophages or DCs from the murine gut for this purpose, we turned to a well-described murine macrophage line J774A.1 and measured fungal killing by J774A.1 cells upon C. albicans infection in vitro using a previously reported method (Sem et al., 2016). Indeed, the ole2/ole2 mutant exhibited reduced resistance to macrophages relative to the WT strain at 24 h post-infection but not at 3 h ( Figure 5E,  Figures S7A,B), and this defect in resistance to host killing can be reversed by the provision of exogenous PGE 2 (Figure 5F). We reasoned that the effects of fungus-derived PGE 2 on fungal killing at 24 h post-infection might require the transcriptional reprogramming of macrophages, as previously described (Suram et al., 2013;Kim et al., 2014;Yun et al., 2016). However, we did not observe a difference in the levels of transcripts associated with pro-inflammatory (Il1b) and immunosuppressive (Arg1, Chil3, Retnla, Il10) macrophage subsets between macrophages infected with WT or ole2/ole2 C. albicans ( Figure S7C). Overall, these results suggest that fungal PGE 2 confers C. albicans with a competitive advantage in the mammalian GI tract by protecting the fungi from the microbicidal activity of intestinal phagocytes.

DISCUSSION
In this study, we sought to clarify the physiological function of PGE 2 production by C. albicans, a phenomenon long described (Noverr et al., 2001) but of unclear importance to either host or fungus. Using a PGE 2 -deficient fungal mutant, we found no contribution of fungal PGE 2 to the in vitro and in vivo virulence of C. albicans across disparate murine models of candidiasis that affect different organs and engage distinct antifungal immunological responses. Instead, we uncovered a specific role for PGE 2 production by C. albicans in potentiating fungal colonization of the murine intestine. Interestingly, the effects of fungal PGE 2 appear to be linked to protection from killing by host phagocytes. Our results raise several questions that merit further elaboration.
How has C. albicans evolved the ability to synthesize PGE 2 from AA, a substrate that it lacks? No fungal orthologs of the mammalian biosynthetic enzymes for PGE 2 -the cyclooxygenases, COX-1 and COX-2, and the PGE 2 synthase, PTGES-have been identified to date (Erb-Downward and Noverr, 2007;Erb-Downward et al., 2008). Even using a sensitive method based on profile hidden Markov models (HMMER v.3) failed to identify distant homologs of the mammalian PGE 2 biosynthetic enzymes (F. L. Sirota & S. Maurer-Stroh, unpublished observation). In addition, the three genes identified to contribute to PGE 2 synthesis in C. albicans-OLE2, FET3, and FET31-do not harbor enzymatic domains that are known to bind to or act on AA or downstream prostaglandins. Indeed, FET3 and FET31 are putative iron transporters in C. albicans and might instead modulate fungal PGE 2 production indirectly, possibly by acting as cofactors or regulating cellular levels of iron that in turn impact PGE 2 synthesis. Genetic deficiency of OLE2 in C. albicans did not completely eliminate PGE 2 production in vitro (Erb-Downward and Noverr, 2007; Figure 1), implying the existence of additional components of the fungal PGE 2 synthetic pathway. It is possible that the impact of OLE2 on fungal PGE 2 production in the gut might be more profound than that observed in vitro, though we have not been able to address this empirically due to difficulties in precisely measuring PGE 2 levels in intestinal contents, which contain high levels of substances that interfere with ELISA data linearity (data not shown). Thus, C. albicans has convergently evolved a mechanism to generate PGE 2 from a host-derived precursor, presumably in response to evolutionary pressures exerted by its lifestyle as an obligate symbiont in warm-blooded organisms.
Why does C. albicans goes through the effort of producing PGE 2 at all? Earlier studies have implicated fungus-derived PGE 2 in pathogenesis, since PGE 2 promotes virulence traits such as germ tube and biofilm formation in vitro (Noverr et al., 2001;Alem and Douglas, 2004). Moreover, fungal PGE 2 can suppress host immunity, which might in turn favor fungal growth. For instance, PGE 2 emanating from a fungal bloom in the gut induced an immunoregulatory M2 phenotype in alveolar macrophages, which in turn exacerbated allergic airway inflammation in mice (Kim et al., 2014), and exposure to fungal PGE 2 diminished the protective immune response generated in response to immunization with DCs pulsed with C. albicans yeasts (Kundu and Noverr, 2011). However, the preceding findings do not point to an overt benefit for C. albicans, for which symbiosis is the status quo and infection is adventitious and often a byproduct of sustained immunodeficiency and/or microbial dysbiosis. Accordingly, we observed no role for fungal PGE 2 in mediating the virulence of C. albicans under both in vitro and in vivo settings. Conversely, fungal PGE 2 production appeared to be important for promoting intestinal colonization by C. albicans. C. albicans is located at high titers in the gut and acquisition of the fungus often occurs early during ontogeny via transmission of maternal intestinal strains to the infant (Russell and Lay, 1973;Odds et al., 2006). It would thus be evolutionarily advantageous for C. albicans to evolve an independent means of generating PGE 2 to facilitate its own entrenchment as a symbiont in the gut.
How does fungal PGE 2 mediate intestinal colonization by C. albicans? Here we provide evidence in support of gut phagocytes, in particular macrophages and CD11b + DCs, as a target of fungal PGE 2 , since depletion of colonic phagocytes by two orthogonal approaches abrogated the fitness defect of the ole2/ole2 mutant vis-à-vis the WT strain. The ole2/ole2 mutant also sustained increased killing by a murine macrophage line in vitro, a phenomenon that was reversed by the addition of exogenous PGE 2 . PGE 2 -dependent fungal resistance to phagocytes could occur via the paracrine activity of fungal PGE 2 on proximal phagocytes, and/or an autocrine effect of fungal PGE 2 on the fungus itself. PGE 2 has wide-ranging effects on phagocytes, most of which converge on immunosuppression (Kalinski, 2012;Medeiros et al., 2012). For instance, PGE 2 reduces the phagocytosis of fungi by macrophages  and polarizes both fungi-exposed macrophages and DCs toward an immunoregulatory phenotype, characterized by an increased expression of suppressive factors such as arginase-1 (Suram et al., 2013;Kim et al., 2014;Yun et al., 2016). Hence, PGE 2 secreted by C. albicans proximal to or engulfed by intestinal phagocytes might dampen their uptake of more fungi, phagosome maturation, and/or production of inflammatory, antifungal factors beyond the ones we had tested ( Figure S7C). The effects of fungal PGE 2 can also radiate beyond those of phagocytes to T cell responses primed by fungi-experienced phagocytes (Gagliardi et al., 2010;Smeekens et al., 2010), which in turn control fungal growth in the gut. PGE 2 is labile and likely has short-range activity, possibly even at the level of individual fungicontaining phagosomes in phagocytes (Hoffmann et al., 2012). This might account for the apparent cell-autonomous effect of fungal PGE 2 on intestinal fitness of C. albicans.
In addition, PGE 2 might act directly on C. albicans itself to augment its resistance to host immunity. Mucosal microbial communities, including those of Candida, have been likened to biofilms (de Vos, 2015), and certain genes crucial for the intestinal fitness of C. albicans are also required for biofilm formation (Ganguly and Mitchell, 2011;Bohm et al., 2017;Lagree and Mitchell, 2017), indicating common mechanisms underlying fungal fitness in both biofilms and the gut. Fungus-derived PGE 2 might therefore elicit biofilm-like transcriptional programs in gut fungi that enhance fitness via various means, such as the induction of stress response pathways that improve fungal resistance to immunological assaults from the host.
Our study focused on the role of fungus-derived PGE 2 but does not rule out a role for host-derived PGE 2 or other AAderived lipid mediators in modulating intestinal symbiosis by C. albicans. Host phagocytes secrete PGE 2 when stimulated with C. albicans (Smeekens et al., 2010;Suram et al., 2013;Yun et al., 2016), though it is plausible that host-and fungus-derived PGE 2 exert distinct activities in a tissue-and cell-type-specific fashion, especially in light of the differential accessibility of immunocyte populations to fungi and the instability of PGE 2 . Nevertheless, we predict that the effects of fungal PGE 2 on the intestinal fitness of C. albicans would be amplified in the absence of host-derived PGE 2 . Notably, C. albicans, as well as other fungi and parasites, produces immunomodulatory lipid mediators from AA aside from PGE 2 in vitro, including prostaglandin D2 and leukotrienes (Szkudlinski, 2000;Herve et al., 2003;Tsitsigiannis et al., 2005;Valdez et al., 2012), some of which have been implicated in modulating microbial fitness during pathogenesis. It is tempting to speculate that non-PGE 2 lipids can, akin to PGE 2 , modulate the fitness of C. albicans and other symbiont microbes in the intestines and other barrier surfaces.
We report here a novel mechanism used by C. albicans to colonize the gut that is distinct from previously delineated pathways, such as the suppression of filamentation and changes in nutrient metabolism and iron acquisition (Nguyen et al., 2009). Since fungal colonization is a prerequisite for disseminated or mucosal diseases, this work provides new potential targets for therapeutic control of intestinal fungal populations to prevent human fungal infections.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the manuscript/Supplementary Files.

ETHICS STATEMENT
The animal study was reviewed and approved by A * STAR Biological Resource Centre (BRC) Institutional Animal Care and Use Committee (IACUC).