Candida albicans commensalism in the oral mucosa is favoured by limited virulence and metabolic adaptation

As part of the human microbiota, the fungus Candida albicans colonizes the oral cavity and other mucosal surfaces of the human body. Commensalism is tightly controlled by complex interactions of the fungus and the host to preclude fungal elimination but also fungal overgrowth and invasion, which can result in disease. As such, defects in antifungal T cell immunity render individuals susceptible to oral thrush due to interrupted immunosurveillance of the oral mucosa. The factors that promote commensalism and ensure persistence of C. albicans in a fully immunocompetent host remain less clear. Using an experimental model of C. albicans oral colonization in mice we explored fungal determinants of commensalism in the oral cavity. Transcript profiling of the oral isolate 101 in the murine tongue tissue revealed a characteristic metabolic profile tailored to the nutrient poor conditions in the stratum corneum of the epithelium where the fungus resides. Metabolic adaptation of isolate 101 was also reflected in enhanced nutrient acquisition when grown on oral mucosa substrates. Persistent colonization of the oral mucosa by C. albicans also correlated inversely with the capacity of the fungus to induce epithelial cell damage and to elicit an inflammatory response. Here we show that these immune evasive properties of isolate 101 are explained by a strong attenuation of a number of virulence genes, including those linked to filamentation. De-repression of the hyphal program by deletion or conditional repression of NRG1 abolished the commensal behaviour of isolate 101, thereby establishing a central role of this factor in the commensal lifestyle of C. albicans in the oral niche of the host.

Introduction Candida albicans is a prominent member of the human mycobiota in the oral, gastrointestinal and vaginal mucosa. As a pathobiont, C. albicans can also cause infections reaching from mild superficial manifestations to more severe and frequently life-threatening systemic candidiasis [1]. Infections occur primarily in immunocompromised individuals, with mucocutaneous candidiasis developing most frequently in patients with acquired or inherited defects in T cell and IL-17 immunity [2]. While intact host defences are critical for preventing C. albicans overgrowth and tissue invasion, the balance between commensalism and disease also depends on the virulence state of the colonizing fungal isolate. The species of C. albicans displays a large genetic diversity which translates in phenotypic differences between isolates. Although the genetic landscape of C. albicans has been described in some details [3], the functional consequences of these intraspecies variations at the host interface are less clear. Experimental infection studies in previously C. albicans-naïve mice with small sets of fungal isolates revealed diverse outcomes in terms of host survival (in the model of systemic candidiasis) or in terms of persistence of colonization (in the model of oropharyngeal candidiasis) [4][5][6][7]. The molecular basis for the commensal versus pathogenic lifestyle of genetically and phenotypically distinct natural isolates of C. albicans remains ill defined. Some pathogenicity traits of C. albicans have been identified in the gastrointestinal tract [8]. The generation of isogenic mutants of the highly virulent reference isolate SC5314 and within-host evolution experiments with SC5314 allowed the identification of genetic determinants that govern the ability of C. albicans to inhabit the gastrointestinal tract [9][10][11][12][13]. While non-filamentous morphologies are generally good colonizers, hyphae are less fit in the gastrointestinal tract but instead may be better suited for epithelial translocation and invasive infections [10,14,15]. Beyond morphology, fungal adhesion and host cell damage induction can also affect gut colonization efficiency [16]. Much less is known about the fungal determinants that enable C. albicans colonization of the oral mucosa, an important niche of the fungus [17] and a frequent site of C. albicans-mediated disease manifestation [18].
The oral cavity differs significantly from the gastrointestinal tract and other commensal niches of C. albicans with respect to physico-chemical properties such as pH and O 2 content, the nature of the lining epithelium and the composition of the microbiota [19]. During oropharyngeal candidiasis (OPC) in mice, which are not naturally colonized with C. albicans, virulent isolates of C. albicans trigger a strong inflammatory host response that results in rapid clearance of the fungus from the oral mucosa. In contrast, low virulent isolates elicit only a limited host response which coincides with fungal persistence in the stratum corneum [7]. In the latter case, commensal colonization of fully immunocompetent and non-antibiotically treated wild type mice can last for over a year [20].
While the differential response of the host to commensal versus pathogenic isolates of C. albicans was dissected in some detail, the fungal factors that determine the commensal lifestyle of C. albicans in the oral niche remain to be explored. Using the prototypical commensal isolate 101 characterized by long persistence in the OPC mouse model, the objective of this study was to determine how the fungus survives and evades the host in this specific tissue compartment. Transcript profiling of the isolate inside the host tissue revealed hallmarks of commensalism in the oral niche. High expression levels of the transcription factor NRG1 was identified as a determining factor in the commensalism of isolate 101, the relevance of which we confirmed by complementary genetic approaches.

C. albicans isolate 101 displays reduced filamentation compared to SC5314
The commensal C. albicans isolate 101 efficiently and persistently colonizes the oral epithelium of wild type C57BL/6 mice when administered sublingually. Fungal elements locate to the stratum corneum where they persist without penetrating deeper epithelial layers of the stratified epithelium [7]. This is in contrast to the highly virulent isolate SC5314, which causes damage in the oral epithelium and invades the stratified epithelium to deeper layers. To determine the characteristics of fungal commensalism in the oral niche, we first assessed the morphology of isolate 101 on PAS-stained tissue sections and after isolation of the fungus from the tongue by KOH (Fig 1A and 1B). The fungus was predominantly in the hyphal form, whereby hyphae appeared to be slightly shorter in comparison to those of the isolate SC5314. By histology we also observed some pseudohyphae.
Because quantification of filament length is hardly possible in vivo on tissue sections, we examined filamentation in vitro in serum containing medium. Filaments of isolate 101 were consistently shorter than those of SC5314, in both presence or absence of oral keratinocytes (Fig 1C-1H). The morphogenetic defect of isolate 101 was most pronounced when cultured on Spider agar (Fig 1I). The commensal phenotype of isolate 101 was thus associated an impairment in filamentation, although the isolate was able to form hyphae, especially at the interface with the host.

Isolate 101 is less invasive than SC5314
In line with the impaired filamentation phenotype, isolate 101 was also less invasive in YNB agar containing BSA when compared to SC5314 (Fig 2A). Similarly, when probing invasion of TR146 oral keratinocytes, a reduced proportion of 101 hyphae invaded host cells (Fig 2B).
Assaying fungal invasion of cultured keratinocyte monolayers monitors the invasion of C. albicans filaments into individual cells, while colonization of the stratified epithelium of the oral mucosa comprises both intra-and intercellular invasion processes. Probing the invasive capacities of isolate 101 in a fully differentiated three-dimensional skin model reproduced the situation in the oral epithelium in vivo, whereby isolate 101 remained mostly restricted to the outermost keratinized layer, while SC5314 penetrated deep into the tissue, reaching even the supporting fibroblasts that underlie the epithelium in this model (Fig 2C). This finding indicated that the localization of C. albicans inside the stratified epithelium is an intrinsic property of the isolate in interaction with epithelial cells. This localization was independent of immune cells that represent an integral component of the mucosal tissue in vivo. The less invasive phenotype of isolate 101 in reconstituted human epidermis coincided with its low capacity to induce epithelial cell damage, which was assessed by LDH release assay (Fig 2D), and production of inflammatory cytokines (Fig 2E and 2F), similarly to the situation in keratinocyte monolayer cultures (Fig 2G and 2H).

Altered interaction of isolate 101 with keratinocytes does not curb the initial infection process in the oral epithelium
To assess whether the observed phenotypic differences between isolates 101 and SC5314 affected their initial colonization of the host tissue, we tested their interaction with the oral epithelium in a competitive infection setting. To distinguish the two isolates after co-infection, we used a mCherry-expressing version of isolate 101 [21] and a GFP-expressing version of SC5314 [22], mixed in equal amounts (S1A Fig). After 1 day, both isolates were recovered from the infected tongue at a 1:1 ratio, comparable to the ratio in the infection inoculum, indicating that both isolates seeded the oral epithelium with comparable efficiency (Fig 3A),

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa confirming previous findings from single-strain infections [7]. The initial steps of the infection process seem thus not to be affected by differences in the isolates' capacity to filament or to interact with keratinocytes in vitro (invasion and damage induction).
In single infections, the cellular damage caused by high-virulent isolates such as SC5314 triggers an acute inflammatory response in the infected epithelium characterized by rapid production and release of pro-inflammatory cytokines by epithelial cells, including IL-1α and chemokines, which trigger the recruitment of neutrophils and inflammatory monocytes from the circulation to the site of infection [23,24]. A consequence of the acute host response in the oral mucosa is the rapid elimination of the fungus from the experimentally infected host. In contrast, low-damaging isolates which trigger a limited or no inflammatory response upon colonization/infection persist in the epithelium over time [7,20]. Isolate 101 is less invasive than SC5314. A. Representative images (left) and quantification (right) of the invasion depth of isolates 101 and SC5314 in YNB agar supplemented with BSA after 7 days of incubation at 37˚C and 5% CO 2 %. Agar invasion was calculated as the % of invasion depth (dotted arrow) relative to the overall thickness of the agar (solid arrow). Data were pooled from 4 independent experiments with 2-3 samples each. The mean±SEM is indicated. Similar results were obtained with YNB agar supplemented with glucose or casamino acids. B. Representative images (left) and quantification (right) of TR146 cell monolayers incubated with isolate 101 or SC5314 for 4 hours at 37˚C and 5% CO 2 . Extracellular fungal parts were stained with an anti-Candida antibody (red), full fungal elements were visualized by calcofluor white staining (blue), and keratinocytes were stained with phalloidin (green). The counted number of total hyphae (red numbers) and of invasive hyphae (underlined red numbers) are indicated on the representative images. Each symbol in the quantification is the % invasion per field analysed with an average of 8 hyphae per field (a total of 32 fields were analysed per isolate with 240 hyphae for SC5314 and 283 hyphae for 101). The mean±SEM is indicated for each group. Data were pooled from 2 independent experiments. C.-E. Reconstituted human epidermis (RHE) was infected with isolate 101 or SC5314 (bottom) for 24 hours. C shows PASstained histology sections. Scale bar = 250 μm. Cellular damage was assessed by LDH release assay in D, GM-CSF concentrations were determined in the supernatant by ELISA in E, and IL-6 and IL-8 transcript levels were quantified by RT-qPCR in F. Bars are the mean+SD of 4 (D, E) or 2 (F) replicate wells from a single experiment. G.-H. Monolayers of TR146 keratinocyte were infected with isolate 101 or SC5314 for 24 hours. Epithelial cell damage was assessed by LDH release assay in G, and IL-6 and IL-8 cytokine levels were quantified in the supernatant by ELISA in H. Bars

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa Therefore, the question arose whether in a co-infection setting, where isolate 101 is exposed to an inflammatory environment induced by SC5314, it would be cleared from the oral mucosa, similarly to SC5314 itself. We confirmed that the overall inflammatory response was indeed comparable in the co-infection setting as in a single infection with SC5314 (Figs 3B and S1B). We then assessed the fungal burden in the tongue by day 7 post-infection, when SC5314 is cleared in a single infection setting [25,26]. Tongue fungal counts in co-infected animals remained high and were comparable as in isolate 101 single-infected mice (Fig 3C). Enumeration of the colonies based on their fluorescent label revealed that all were expressing mCherry and hence identified as 101, while no SC5314 colonies were recovered (Fig 3C). Together, this indicates that, even in a co-infection setting, isolate 101 persists in the oral mucosa and resists the inflammatory environment.

Persistence of isolate 101 in the oral mucosa is associated with a reduced virulence profile
To obtain a more comprehensive and unbiased picture of the properties of isolate 101, we acquired the transcriptomes of isolates 101 and SC5314 as a reference within the oral mucosa at three consecutive time points (day 1, day 3, day 7 post-infection). However, due to its rapid clearance from the oral mucosa the analysis of isolate SC5314 transcriptome was limited to day 1 post-infection. We also included samples from in vitro-induced hyphae of both isolates, whereby we spiked this RNA into RNA isolated from naïve tongues.
Fungal RNA was enriched using a hybridization technology aimed at enriching fungal RNAs within infected host tissues where they usually account for only a small fraction of the overall RNA [27]. We verified that the baits used for RNA enrichment, which were designed on the basis of strain SC5314, or usage of SC5314 genome assembly for aligning sequencing data did not introduce a bias (see details in the methods section, S2A-S2D Fig and S1 Table). Several hundreds of genes were globally up-and downregulated under the chosen conditions (S2E Fig and S2 Table). Principal component analysis showed distinct transcript profiles of isolates 101 and SC5314, both under in vitro or in vivo conditions (Fig 4A). Moreover, the transcriptome of isolate 101 changed over time and became increasingly distinct from the one of SC5314 on day 1 post-infection (Fig 4A).
Qualitative differences of the transcriptional response of isolate 101 compared to SC5314 were reflected in a diminished expression of genes associated with filamentous growth, virulence, and immune activation, with differences increasing over time (Figs 4B and S3). Among WT mice were infected sublingually with a 1:1 mixture of isolates 101-mCherry (101-mCh) and SC5314-GFP (SC-GFP) or with either isolate alone. On day 1 (A) and on day 7 (C) the total tongue fungal burden was quantified (left panels). From the co-infected animals, the colonies were further analysed for mCherry-and GFP-expression (right panels). Quantification of tongue-infiltrating neutrophils by flow cytometry on day 1 post-infection is shown in B. Each symbol represents one animal. The mean of each group is indicated. Data in A and B are pooled from 2, data in C are pooled from 3 independent experiments. d.l., detection limit. Please see also S1

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa the most strongly downregulated genes in isolate 101 featured genes encoding adhesins and invasins, namely ALS1 and ALS3 [28,29], aspartyl proteases, namely SAP4, SAP6 and SAP8 linked to deep tissue invasion [30], and the damage-inducing factor ECE1 [24] (Fig 4D). In contrast, the increased expression of SAP1 and SAP3 by isolate 101 in the oral mucosa was consistent with the preferential residence of this isolate in the upper layers of the stratified epithelium [30]. Genes encoding transcription factors associated with filamentous growth including CPH1, HGC1, BRG1 and UME6 were also downregulated in isolate 101 as compared to SC5314. Interestingly, NRG1 that encodes a repressor of the filamentation program [31] was strongly upregulated (Fig 4D). The difference in morphogenesis between the two isolates was further associated by differential regulation of cell wall and biofilm associated genes (Fig 4D).
Among the genes displaying an increased expression in isolate 101 relative to SC5314 were stress-response genes, especially those implicated in ROS detoxification and the response to oxidative stress (Fig 4D). This observation is in support of the persistent phenotype of the strain in the mucosal tissue, even under inflammatory conditions as during co-infection with an inflammation-inducing isolate (Fig 3).
Differential expression of virulence and morphogenesis genes between isolates 101 and SC5314 in the murine oral mucosa was largely conserved under in vitro culture conditions with serum (Fig 4D). RT-qPCR assays with RNA isolated from C. albicans after exposure to TR146 keratinocytes further confirmed the RNAseq data (S4A and S4B Fig). Overall, the transcriptomic data are well in agreement with the phenotypic data indicating that isolate 101 is less filamenting, less invasive, less tissue damaging and less immune activating than SC5314 (Figs 1 and 2) [7,20]. Of note, the changes in virulence gene expression increased over time and largely accounted for the time-dependent shifts in the transcript profile of isolate 101 ( Fig  4A).

Isolate 101 is metabolically adapted to the oral niche
In addition to the filamentation-related gene expression differences described above, we identified an unanticipated, albeit prominent metabolic signature in isolate 101. A large group of genes associated with metabolic processes, including amino acid and phospholipid catabolism, and with glycolysis and pyruvate metabolism were overexpressed in isolate 101 in comparison to SC5314 (Fig 4C and 4D). Surprisingly, these differences were apparent at every time point analyzed and both under in vivo and in vitro conditions (Figs 4C, 4D and S4C). Moreover, genes encoding membrane transporters, including oligopeptide transporters and glucose, dicarboxylate and monocarboxylate transporters were upregulated in the murine tissue at at least one time point analysed (Fig 4C). Taken together, these data suggested that isolate 101 is metabolically adapted and employs an efficient strategy for nutrient acquisition in the oral epithelium of the host where nutrients are limited.
In support of this notion, isolate 101 featured enhanced phospholipase activity in comparison to SC5314 when examined experimentally on original Prices' egg yolk agar (S4D Fig). Moreover, we observed that isolate 101 displayed an enhanced growth under conditions mimicking those in the oral mucosa, including the replacement of glucose as the carbon source by N-acetlyglucosamine (GlcNAc), an amide derivative sugar abundant in the saliva and in the cell wall and biofilm extracellular matrix of oral bacteria [32,33]. The enhanced growth phenotype was also visible on casamino acids, a mixture of free amino acids and small peptides that mimics the protein-rich environment of the stratum corneum of the tongue [34], and in BSA representing a more complex nitrogen source (S4D Fig). The better adaptation of isolate 101 compared to SC5314 in sensing and using GlcNAc was further evidenced by its resistance to GlcNAc-induced cell death (S4E Fig). GlcNAc serves as a signal of nutrient availability in C.

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa albicans, inducing entry into mitotic cell cycle and rapid cell growth. However, incubation in the presence of GlcNAc without other nutrients to support growth results in cell death due to the accumulation of intracellular ROS [35]. While we confirmed this phenomenon with strain SC5314 (S4F Fig), for which it was originally demonstrated [35], isolate 101 survived unabatedly under these conditions (S4F Fig), in line with its increased expression of oxidative stress resistance genes (Fig 4D). The metabolic profile of isolate 101 was not unique: eight other isolates that we tested for growth in GlcNAc and casamino acids displayed variable growth profiles, which did not correlate with the high-or low damage inducing capacity of the isolates in epithelial cells (S4G Fig). Altogether, these findings indicated that persistent growth of isolate 101 in the oral cavity was favoured not only by restricted filamentation and virulence, but also by metabolic adaption to the nutrient-poor environment in the stratified epithelium.

Overexpression of NRG1 in SC5314 restrains its virulence potential in vitro and in vivo
Given the pronounced differences between isolates 101 and SC5314, we speculated that modulating the expression of virulence traits in isolate 101 would alter its phenotype and change its behaviour at the interface with the host. Given the link between damage induction and induction of inflammation in the host by highly virulent strains of C. albicans and because ECE1 plays a prominent role in this process, we attempted to heterologously express ECE1 of strain SC5314 in isolate 101. Despite choosing the strong TDH3 promoter, we could not achieve ECE1 expression levels that were high enough to elicit detectable levels of secreted peptides or to increase damage induction in keratinocytes (S5A Fig). The failure of achieving enhanced virulence in the commensal isolate 101 might be due to inadequate posttranslational processing of Ece1p required for generation of the peptide toxin known as Candidalysin [36] and the need for delivery of the toxin to the host cells via an invasion pocket at the hyphal tip [37]. The reduced filamentation of isolate 101 may not suffice those criteria.
We therefore turned our focus onto transcription factors that regulate C. albicans virulence more broadly and that were differentially regulated between the highly virulent SC5314 isolate and the low-virulent isolate 101. Among the most strongly regulated genes in isolate 101 was NRG1that encodes a repressor of the filamentation program. Given the filamentation phenotype of isolate 101 and the central role of filamentation C. albicans pathogenicity, we speculated that manipulating the expression of NRG1 in isolate 101 might alter its phenotype and make it more virulent. Vice versa, overexpressing NRG1 in SC5314 might have the opposite effect. Therefore, we made use of a SC5314-based isolate that conditionally overexpresses NRG1 upon doxycycline (Dox) treatment [13]. Overexpression of NRG1 (Fig 5A) did not affect the growth rate of the isolate (S5B Fig), but resulted in a defect in filamentation, both on YPD agar and in liquid medium. Dox-treated SN76 NRG1-OE cells, but not cells of a Dox-treated control isolate, displayed a yeast/pseudophyphal morphology (Fig 5B and 5C). Moreover, NRG1 overexpression was accompanied by reduced expression of diverse virulence genes such as ECE1, ALS3 and HWP1 (Fig 5D) and a reduced capacity to induce damage in TR146 keratinocytes (Fig 5E).
Based on these data, we wondered whether modulation of NRG1 levels would also affect the behaviour of C. albicans cells in vivo in the oral mucosa. Pre-treatment of the isolate and supplementation of the drinking water of the mice with doxycycline had no effect on initial steps of the infection process (Fig 5F). However, the acute inflammatory response characterized by neutrophil recruitment to the site of infection in consequence of epithelial damage, was much reduced in mice infected with C. albicans strain overexpressing NRG1 in comparison to controls as evidenced on histology sections (Fig 5G) and by enumeration of neutrophils in

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa infected tongues by flow cytometry (Fig 5H). The reduced inflammatory response of the host was accompanied by a delay in fungal clearance which manifested by elevated fungal loads on day 3 post-infection in case of the NRG1-overexpressing C. albicans strain in comparison to all

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa controls (Fig 5I). By day 7 however, mice had fully cleared the fungus and regained their initial weight (Fig 5J).
We also examined the consequences of modulating expression of another gene, CRZ2, which was also differentially expressed between isolates SC5314 and 101 and whose overexpression in SC5314 results in enhanced fitness during gastrointestinal colonization [13].

Reduced expression of NRG1 in isolate 101 increases its pathogenicity
Following from the results obtained with an C. albicans SC5314 derivative overexpressing NRG1, which revealed impaired virulence in vitro and in vivo in the OPC model, we wondered whether the opposite could be achieved in isolate 101, namely whether reducing NRG1 expression levels in this isolate would be able and sufficient to increase its pathogenicity. Deletion of one or both alleles of NRG1 in isolate 101 (101 nrg1Δ/NRG1 and 101 nrg1Δ/Δ , Fig 6A) entailed a dose-dependent increase in expression of determining virulence factors, including ALS3, ECE1 and HWP1 (Fig 6B). Expression of ALS3 in 101 nrg1Δ/Δ reached levels even higher than those in wild type SC5314 (Fig 6B). The altered gene expression profile correlated with increased filamentation. Both, the heterozygous and the homozygous deletion mutants formed longer filaments than the parental isolate 101 when exposed to human keratinocytes (Figs 6C  and S6A). Increased filamentation of the mutants was also observed on Spider agar and on YPD agar containing different carbon sources (S6B and S6C Fig) and this correlated with enhanced invasion properties (S6C Fig). Acquisition of these important virulence traits was however not sufficient for induction of detectable cellular damage in TR146 keratinocytes ( Fig  6D).
Importantly, deletion of NRG1 in isolate 101 resulted in enhanced virulence in the oral mucosa of experimentally infected mice in vivo. Deletion of a single copy of the gene was sufficient to elicit an enhanced inflammatory response characterized by increased expression of inflammatory cytokines and mediators and enhanced infiltration of neutrophils and monocytes into the colonized tongue as assessed by flow cytometry and on histology sections on day 1 post-infection, while colonization loads were comparable between 101 nrg1Δ/NRG1 and the parental isolate at this time point (Fig 6E-6H). Notably, the enhanced inflammatory response triggered by 101 nrg1Δ/NRG1 in the host impaired fungal colonization and resulted in a reduced fungal load on day 7 post-infection (Fig 6I and 6J).
Unfortunately, we were unable to extend our analysis to the homozygous deletion mutant. Isolate 101 nrg1Δ/Δ tended to form clumps in culture, which reduced the efficiency of the isolate to colonize the murine oral mucosa (Fig 6E). Clumping could not be prevented by filtering the isolate before applying it to the animals. The inequality in fungal load recovered from the tongue on day 1 post-infection (S6D Fig) prevented any further analyses of the 101 nrg1Δ/Δ mutant in vivo.
We therefore pursued a second independent approach to suppress NRG1 expression in isolate 101 very efficiently using a Dox-dependent TET-off system. Notably, one NRG1 allele was under the control of the TET promoter in the designed isolate, while the other was deleted (101 nrg1Δ/pTET-NRG1 ) (S7A and S7B Fig). As little as 1 μg/ml of Dox resulted in a strong reduction of NRG1 expression in isolate 101 nrg1Δ/pTET-NRG1 without overt clumping of the isolate (Figs 7A and S7C). Similar as with the NRG1 deletion mutants, the reduction of NRG1 transcript levels was accompanied by a dose-dependent increase in the expression of various

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa  (Figs 7B and S7C). Dox -induced suppression of NRG1 expression also resulted in enhanced filamentation, both under YPD culture conditions at 37˚C without addition of FCS or any other known hyphae-inducing agents (Fig 7C), and after exposure to TR146 keratinocytes (Fig 7D and S7D). Importantly, these phenotypic and functional changes were accompanied by enhanced pathogenicity in vivo as indicated by the strong inflammatory response induced by isolate 101 nrg1Δ/pTET-NRG1 in the oral mucosa upon Dox treatment (Fig 7F  and 7G). Together, these results demonstrated that NRG1 acts as a central regulator of C. albicans pathogenicity in the oral cavity and that suppression of high NRG1 expression levels in the commensal stain 101 is sufficient to increase its virulence and to alter the fungus-host interplay at the epithelial barrier.

Discussion
The oral cavity is an important niche for C. albicans colonization [17]. It also represents a site where infections caused by C. albicans frequently manifest [18]. Our understanding of the fungal cues that allow C. albicans to persist in this niche and to be tolerated by the host without causing disease, remains incomplete. To fill this gap in knowledge, we employed a model of oral colonization in mice and examine requirements for C. albicans commensalism using a prototypical commensal isolate as an example.
Transcript profiling of C. albicans isolate 101 in the oral tissue and functional studies with the isolate in host-involving and host-free conditions revealed a pronounced reduction in multiple virulence traits, including filamentation, invasion and damage induction. Based on this information, we attempted to increase virulence of isolate 101. Overexpression of specific virulence genes, such as ECE1, was not sufficient to induce a gain in pathogenicity in isolate 101, possibly consistent with the recent demonstration that the full damage potential of C. albicans depends not only on strong Ece1 expression but also coordinated delivery of fully processed and secreted Candidalysin in invasion pocket created by invasive hyphae [37]. Of note, we expressed the entire ECE1 gene originating from SC5314 under the strong TDH3 promoter, including the sequences flanking peptide 3 for optimal Candidalysin processing, to exclude suboptimal processing of Ece1p as recently reported [38].
Strengthening the hyphal program in isolate 101 by suppressing NRG1 expression resulted in major changes in its behaviour in the oral cavity. Forced induction of filamentation in the normally avirulent isolate 101 was sufficient to elicit an inflammatory response in the colonized host and to impair colonization, indicating that restricted filamentation is a major determinant of commensalism in the oral cavity. This is reminiscent of what was described for other commensal niches of C. albicans [15,39]. Despite the parallels in the role of filamentation in commensalism in various mucosal epithelia, there are also important differences between different sites. As such, SC5314 CRZ2-OE acquires enhanced fitness during gut colonization [13] but not during OPC suggesting that commensalism in the gastrointestinal tract and in the oral cavity integrates different fungal effectors. Morphotype switching drives C. albicans pathogenicity also during systemic infection [40]. Of interest, rapid yeast proliferation in internal organs can compensate for the lack of filamentation, as it was shown recently with an EED1 deficient mutant [41].
NRG1 is a transcriptional repressor of hyphae-specific genes and of genes required for the morphogenic switch [42]. NRG1 transcription is downregulated under conditions promoting filamentation. Forced constitutive expression of NRG1 in SC5314 established the need of morphotype switching for pathogenicity of C. albicans during disseminated candidiasis [40], while enforced filamentation due to constitutive deletion of NRG1 lowered intestinal colonization of C. albicans [43]. A limited degree of filamentation is however beneficial for commensalism in the gut: it has recently been shown to drive an IgA response targeting fungal hyphae, which in turn improves competitive fitness of C. albicans and promotes homeostasis [44]. The role of NRG1 as a driver of commensalism extends beyond the species of C. albicans: reduced virulence of C. dubliniensis was linked to the inability of this species to modulate NRG1 expression in response to environmental signals that promote filamentation in prototypic isolates of C. albicans [45].
Determining the genetic basis for the commensal phenotype of C. albicans isolate 101 represents a significant challenge due to the intrinsically high diversity in C. albicans' population genetic structure [3], consisting of numerous sequence and, higher order genomic differences [46,47]. The situation is further complicated by circuit diversification in regulatory networks that exist between isolates [48] and the fragility of such networks in the oral mucosa [49]. This

PLOS PATHOGENS
large complexity prevented us from yet identifying any specific genetic determinants underlying the commensal phenotype of isolate 101. No significant sequence differences were identified in the promoter region of NRG1 between isolate 101 and SC5314. Replacing an allele of an upstream transcriptional regulator of NRG1 (BRG1), which we found to be truncated in isolate 101, with a full-length copy was not sufficient to reduce NRG1 expression and alter the phenotype of the isolate. A genetic determinant of C. albicans commensalism was recently identified in the coding sequence of ECE1, which affects processing of the damage causing peptide toxin and may thereby contribute to differential pathogenicity among clinical isolates [38]. While this may explain, at least in part, the low virulence of isolate 529L in the OPC model [7,50], it does not so for isolate 101, which displays the SC5314 isoform at the relevant ECE1 P2-P3 junction. Establishing the genetic determinant, or rather the combination of determinants, responsible for the commensal lifestyle of isolate 101 remains a challenging task for the future.
The transcriptome of C. albicans isolate 101 in the oral tissue revealed a metabolic signature adapted for nutrient acquisition in this specific niche. In experimentally infected mice, commensal isolates are confined to the stratum corneum of the tongue [7], the outermost layer of the stratified epithelium composed of metabolically inactive corneocytes that form a tight and water impermeable barrier limiting nutrient availability [51]. The absence of tissue damage by commensal isolates further limits access to nutrients inside host cells. The high expression of genes encoding proteases, oligopeptide transporters and amino acid catabolic enzymes indicates customized nutrient acquisition mechanisms being active in this protein-rich tissue compartment. Among the amino acid catabolic enzymes whose genes are especially highly expressed in isolate 101 were those involved in catabolism of Glycine and Glutamine, which are highly enriched in structural proteins of the cornified envelope of epithelium [34].
The selective localization to the stratum corneum may thus not simply be the consequence of impaired damage induction by isolate 101 but also reflect metabolic specialization to this niche. This is further supported by the expression of specific subsets of SAP genes with SAP1 and SAP2, which mediate initial invasion of the epithelium [30], being highly expressed by isolate 101 in contrast to SAP6 and SAP8, which are associated with penetration of C. albicans to deeper epithelial layers [30]. The restricted localization of commensal C. albicans in the stratum corneum may promote fungal persistence in several ways: Exclusion from deeper layers of the epithelium limits induction and release of inflammatory mediators from metabolically active epithelial cells located in the stratum spinosum. It also protects the fungus from exposure to antifungal mechanisms of the epithelium such as antimicrobial peptides, which are lower in concentration towards the outside of the epithelium because the cornified envelope, a water-proof amalgam of crosslinked proteins and hydrophobic lipids, hardly provides access to water-soluble positively charged antimicrobial peptides. Enhanced resistance to host defences represents an additional mechanism of immune evasion that allows the fungus to survive homeostatic immunity, which the host raises to counteract fungal (over)growth [21]. C. albicans isolate 101 did even resist the strong inflammatory host response to which it was exposed during co-infection with the high-virulent SC5314, which suggests the acquisition of several factors inducing host resistance by isolate 101. Among these, resistance to oxidative stress prevailing in 101 provides an advantage during exposure to extracellular ROS, e.g. from neutrophils. An enhanced oxidative stress response also contributes to protection from GlcNAc-induced cell death by counteracting the accumulation of intracellular ROS [35]. Resistance to oxidative stress may be further enhanced by GlcNAc and amino acids [52], which are favourably metabolized by isolate 101.
That the metabolic program of C. albicans can be at least as important as its virulence for its behaviour in a specific host niche has recently been demonstrated by the Jacobsen lab. Using a filamentation-deficient mutant strain they demonstrated that metabolic adaptation, rather

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa than filamentation, was important for gut pathogenicity [41]. In case of isolate 101, the metabolic program provides this strain with an advantage for thriving in the nutrient-poor cornified layer of the tongue epithelium and may thereby favour its low pathogenicity. Here we observed that NRG1 expression can modulate host-pathogen interactions, however the precise role of NRG1 in the mode of regulation of isolate 101's metabolic profile remains still to be precisely dissected.
Isolate 101 is a prototypical commensal isolate, which we studied in its natural niche. Based on its low damage inducing capacity, it is representative of the majority of commensal isolates [7]. Similarly, with respect to its nutrient preferences, isolate 101 is not an outlier within the species of C. albicans. Importantly, however, every isolate is unique in how it combines different properties, as it becomes clear when comparing a larger number of properties across isolates [7]. While individual C. albicans isolates display multiple phenotypic differences when analysing them in isolation, a strain's lifestyle at the host interface is the result of the combination of all traits and in dependence of the niche environment. This has also been highlighted by a recent study of the Ene, Koh and Hohl labs, which in a joint effort demonstrated that multiparametric strain-intrinsic factors rather than lab-or animal facility-dependent parameters account for the commensal lifestyle of isolates [53]. Moreover, colonization of the mucosal niche by a particular strain is not readily predictable from the in vitro analysis of individual features of that strain [49].
Together, our work sheds new light on fungal attributes of C. albicans commensalism in the oral cavity. Improved understanding of how the fungus can prevent pathogenicity in various host niches paves the way for novel therapeutic approaches that aim not at eradicating the fungus but rather counteracting its virulence. This strategy has been successfully approached with antagonizing bacteria [54,55] and compounds targeting C. albicans hyphae and biofilm formation via TUP1 have been developed [56]. Based on this concept new opportunities may also open up for the NDV-3A vaccine, which can shift the hyphae/yeast balance by inducing ALS3-specific IgA [44]. Targeting fungal virulence instead of the fungus' ability to thrive on colonized surfaces will further help limiting the rise in antifungal resistance.

Ethics statement
All mouse experiments in this study were conducted in strict accordance with the guidelines of the Swiss Animals Protection Law and were performed under the protocols approved by the Veterinary office of the Canton of Zürich, Switzerland (license number 183/2015 and 166/ 2018). All efforts were made to minimize suffering and ensure the highest ethical and humane standards according to the 3R principles.

Fungal strains
Strains used in this study are listed in Table 1. All strains were maintained on YPD agar for short term and in glycerol-supplemented medium at -80˚C for long term storage. Cultures were inoculated at OD 600 = 0.1 in YPD medium (using a pre-culture to adjust the OD 600 ) and grown at 30˚C and 180 rpm for 15-18 hours. For Dox inducible strains, the cultures were supplemented with 50 μg/ml (in case of SC5314 NRG1-OE ) or 10 μg/ml Dox (in case of 101 nrg1Δ/ pTET-NRG1 , unless stated otherwise). At the end of the culture period, yeast cells were washed in PBS and their concentration was determined by spectrophotometry, whereby 1 OD 600 = 10 7 yeast cells.

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa Generation of NRG1 overexpressing mutants NRG1 was PCR amplified from C. albicans SC5314 genomic DNA using primers 23_G09_FP and 23_G09_RP. PCR product was mixed with the donor plasmid pDONR207 (Invitrogen), and subjected to a recombination reaction with Invitrogen Gateway BP Clonase, as described [60], to yield Entry vectors. The recombination mixes were transformed into E. coli Top10 and one transformant per ORF was selected for further study. The cloned ORFs were sequenced to ascertain that no mutations were introduced during PCR amplification. The Entry plasmids were used in a Gateway LR reaction together with the CIp10-P TET -GTW vector [61]. The recombination mixes were transformed into E. coli Top10 and one transformant was used for plasmid preparation. BsrGI digestion was used to verify the clones. The expression plasmids

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa were digested by StuI prior to transformation into C. albicans strain CEC4642. Transformants were selected for prototrophy and proper integration at the RPS1 locus verified by PCR using primers CIpUL and CIpUR, as described [60].

Generation of NRG1 deletion mutants
NRG1 was deleted in isolate 101 using a SAT1-recyclable system [62]. Briefly, two primers containing 70-pb of sequences flanking the NRG1 ORF and 20-bp of sequences flanking the SAT1-flipper cassette were used in a PCR with pSFS2A [62] as a template. This step was performed with Phusion DNA Polymerase (New England Biolabs, Ipswich, MA, USA) in the presence of 1M betaine. The PCR fragment was used to transform the isolate 101 by electroporation [63] and under selection with nourseothricin (400 μg/ml; Werner BioAgents, Jena, Germany). After PCR verification of the first allele deletion by primers pSFS2A Dwn Chk et NRG1_Kpn (isolate DSY5577), the SAT1 marker was recycled as described [62] to yield DSY5592. This last isolate was used for second allele inactivation resulting in DSY5599.

Generation of Tet-off mutant
To generate NRG1 downregulation in the isolate 101, the tetracycline reversed activator (CartTA) in pNIMX [61] was replaced after XbaI-MluI digestion by the tetracycline activator reported in Bijlani et al. [64] to result in pDS2136. The CatTA gene was prepared as a synthetic fragment (Eurofins Genomics GmbH, Ebersberg, Germany). The hygromycin resistance gene (HygR) was amplified with primers Hygro-TEF2-Nde et Hygro-ACT1-NdeI using pMY70 [65] (see Table 2) and replaced SAT1 as a NdeI fragment in pDS2136 to produce pDS2142. This plasmid was linearized by KpnI/SacII for integration at the ADH1 locus in isolate DSY5592 to yield DSY5617 under selection with hygromycin (600 μg/ml; Mediatech, Manassas, VA, USA).
To construct a tetracycline-dependent promoter system, a PCR fusion strategy was chosen. A first fragment containing a tetracycline operator (pTETO7-OP4) was generated with primers Tetp-5 and Tetp-3 using pNIM1 [61] as template. A second PCR fragment containing the NAT1 dominant marker was generated with overlapping ends using pJK795 [66] as template. Both fragments were used in a PCR fusion with primers of 70-bp overlapping NRG1 matching regions upstream and downstream of the ORF (NAT1-NRG1 and TetP-3-NRG1). This last step was carried out with Phusion DNA Polymerase (New England Biolabs, Ipswich, MA, USA) in the presence of 1M betaine. The final PCR fragment was precipitated with 70% EtOH and transformed in DSY5617 (NRG1/nrg1Δ::FRT ADH1::pDS2142) to yield DSY5625 which contains NRG1 under the control of pTETO7-OP4.
To test the Tet-off system, a strain with a luciferase reporter system was created by introducing CIp-PTET-SP-gLUC59 after StuI digestion in CEC161 [61], resulting in isolate DSY5622. Plasmid pDS2142 was next introduced in this isolate to yield DSY5624.

Luciferase assay
Isolate DSY5624 was cultured overnight in YEPD and grown to log phase for 4 hr in the same medium

Generation of ECE1 overexpression mutant
ECE1 was overexpressed in isolate 101 using the TDH3 promoter. The promoter was amplified by PCR using genomic DNA from SC5314 with primer pairs TDH3-XHOI/TDH3-KpnI containing KpnI and XhoI sites. After digestion of the PCR fragments by KpnI and XhoI, the promoter was cloned in the plasmid pDS2032 to yield pDS2037. Plasmid pDS2032 contains ECE1 from SC5314 under the control of the HWP1 promoter in the backbone of pDS1904 [67], a derivative of CIp10 with a SAT1 dominant marker. The HWP1 promoter and ECE1 were cloned sequentially from PCR fragments obtained with primers pairs HWP1_Kpn/ HWP1_XhoI and ECE1_XhoI/ECE1_MluI. Insertion of the KpnI-XhoI and XhoI-MluI digested PCR fragments yielded pDS2032. Plasmid pDS2037 was digested by MlscI and SmaI for introduction at the TDH3 loci of isolate 101.

Keratinocyte cell line and cell culture conditions
The oral keratinocyte cell line TR146 [68] was grown in DMEM medium supplemented with 10% FCS, Penicillin and Streptomycin at 37˚C and 5% CO 2 . For experiments, cells were seeded at 2 x 10 5 cells/well in 6-well tissue culture plates or at 2 x 10 4 cells/well in 96-well tissue culture

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa plates, respectively, and grown to confluent monolayers for 2 days prior to infection as described below for the individual assays. 1 day prior to the experiment, the DMEM medium was replaced by F12 medium (Hams's Nutrient Mixture F12 medium (Sigma) supplemented with L-Glutamine, Penicillin, Streptomycin, and 1% FCS).

Reconstituted human epidermis
Reconstructed human epidermis (RHE) models aged for 17 days were purchased from Episkin and cultured in 6-well tissue culture plates with 2 ml T-skin medium per well at 37˚C and 5% CO 2 . One day prior to infection, the models were transferred from T-skin medium to F12 medium containing 1% FCS and infected with 4 x 10 6 yeast cells. The models were then carefully swayed using sterile forceps to distribute the liquid in the insert. After incubation for 24h at 37˚C and 5% CO 2 , the supernatant was collected for cell damage quantification by LDH assay. The models were cut in half and used for RNA isolation and histology, respectively. For histology, the support of the skin reconstructs was detached from the insert with a scalpel prior to embedding the tissue in 4% paraformaldehyde.

Morphological analyses
For assessing filamentation on spider agar (10 g mannitol, 10 g nutrient broth, 2 g K 2 HPO 4 and 13.5 g agar per 1L H 2 O), 3 μl of a yeast cell suspension at 10 5 cells/ml were spotted and plates were incubated for 5 days at 30˚C. For assessing colony shape of SC5314 NRG1-OE and SC5314 crz2-OE , 3 μl of a yeast cell suspension at 10 6 cells/ml was spotted onto YPD agar containing 10% FCS and supplemented with 50 μg/ml Dox and plates were incubated for 2 days at 37˚C.
For measuring filament length, 5x10 4 yeast cells were seeded in F12 medium (see above) into 24-well tissue culture plates containing or not containing monolayers of TR146 cells. For analysis of filamentation of 101 nrg1Δ/pTET-NRG1 , 10 μg/ml Dox was added. After 3.5-4 hours of incubation at 37˚C and 5% CO 2 , cells were fixed in 4% PFA for 30 min at 4˚C and washed with PBS. Images were acquired using a Zeiss Axio Observer Z1 inverted phase contrast fluorescence microscope. 4 fields per well were imaged. The length of individual non-aggregated filaments was measured using the ImageJ software.

Invasion assays
For agar invasion, 5 μl of a yeast cell suspension at 10 7 cells/ml were spotted on YNB/BSA agar (1.7 g yeast nitrogen base w/o amino acids and ammonium sulphate, 5 g ammonium sulphate,

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa 20 g agar, 2 g BSA per 1L H 2 O) and incubated for 4 days at 37˚C and 5% CO 2 . Invasion depth was assessed after cutting the agar with a scalpel through the colonies. % invasion was calculated as the proportion of colony invasion relative to the entire agar thickness.
For assessing invasion of C. albicans into TR146 keratinocytes, monolayers of TR146 cells on sterile glass slides in 24-well tissue culture plates prepared as described above were infected with 5 x 10 4 yeast cells per well. After 3 hours of incubation at 37˚C and 5% CO 2 , cells were fixed in 4% PFA for 15-30 minutes and subjected to a two-step immunofluorescence staining protocol adapted from Wächtler et al. [69]. Extracellularly fungal parts were stained with an anti-Candida antibody (Meridian Life Science B65411R) followed by AF647-conjugated goat anti-rabbit IgG (Jackson Immunoresearch). After permeabilization with 0,5% Triton X-100, actin was stained with FITC-labelled phalloidin (Sigma) to visualize epithelial cells and fungal elements were stained with calcofluor white (Sigma). Images were acquired with a Leica SP8 inverse confocal microscope. The % of invasive hyphae was calculated relative to the total number of hyphae per imaged field. The number of total hyphae per field was determined by counting the buds (heads) from which the hyphae have emerged. Invasive hyphae were identified as those with clearly blue stretches. A total of 32 fields were quantified per sample.

Phospholipase activity
5 μl of a yeast cell suspension at 10 7 cells/ml were spotted on Prices' original egg yolk agar (4% glucose, 1% peptone, 1 M NaCl, 2,5 mM CaCl 2 , 17 g agar per 1L H 2 O and addition of 2,5% eggyolk (Sigma) after autoclaving) and incubated for 4 days at 37˚C and 5% CO 2 . Phospholipase activity was determined as the ratio of the diameter of the colony relative to the diameter of the precipitation zone forming around the colony.

Cell damage assay
Damage induction by C. albicans in TR146 cells was performed as described [70]. Briefly, monolayers of TR146 cells in 96-well tissue culture plates prepared as described above were infected with2 x 10 4 yeast cells per well and incubated for 24 hours at 37˚C and 5% CO 2 . Control wells were incubated with medium only or with 1% Triton-X-100 to determine 100% damage. LDH release into the supernatant was quantified with the LDH cytotoxicity kit (Roche) according to the manufacturer's instructions. For assessing LDH concentrations in the supernatant of reconstructed human epidermis, a standard curve was generated with recombinant LDH enzyme.

Cytokine quantification by ELISA
IL-6, IL-8, and GM-CSF levels in the supernatant of infected TR146 monolayers and RHE models were quantified by ELISA using cytokine-specific DuoSet ELISA systems (R&D Systems) according to the manufacturer's instructions.

Mice
Female WT C57BL/6j mice were purchased from Janvier Elevage and maintained at the Laboratory Animals Service Center of the University of Zurich under specific pathogen-free conditions. Mice were used at 6-15 weeks of age in age-matched groups. Infected and uninfected animals were kept separately to avoid cross-contamination.

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa

OPC infection model
Mice were infected sublingually with 2.5 x 10 6 C. albicans yeast cells as described [71], without immunosuppression. In co-infection experiments, mice were infected with 1.25 × 10 6 yeast cells of each isolate, resulting in a total of 2.5 × 10 6 yeast cells. In 7-day experiments, the body weight of the animals was assessed prior and during the course of infection. For experiments with Dox -inducible C. albicans isolates, mice were administered Dox in the drinking water (2 mg/ml) starting from 1 day prior of infection. For determination of the fungal burden, the tongue of euthanized animals was removed, homogenized in sterile 0.05% NP40 in H 2 O for 3 minutes at 25 Hz using a Tissue Lyzer (Qiagen) and serial dilutions were plated on YPD agar containing 100 μg/ml Ampicillin. For histology, tissue was fixed in 4% PBS-buffered paraformaldehyde overnight and embedded in paraffin. Sagittal sections (9 μm) were stained with Periodic-acidic Schiff (PAS) reagent and counterstained with Haematoxilin and mounted with Pertex (Biosystem) according to standard protocols. Images were acquired with a digital slide scanner (NanoZoomer 2.0-HT, Hamamatsu) and analysed with NDP.view2.

Isolation of C. albicans from infected tongue tissue
Longitudinally cut tongue halves were placed into 1.5 ml safe-lock tubes containing 800 ml sterile 20% KOH solution and incubated at 60˚C and 300 rpm for at least 30 minutes. Tubes were inverted regularly to prevent sedimentation of tissue fragments. After complete dissolution of the tissue, the suspension was centrifuged for 10 minutes at 6'000 rpm and supernatant was discarded. C. albicans cells were suspended in 495 μl Tris-HCl (0.1 M, pH 9) and 5 μl calcofluor white (5-10mg/ml in 0.1 M Tris-HCl, pH 9) to stain the cell wall. After 10 minutes of incubation in the dark, C. albicans cells were sedimented for 10 minutes at 6'000 rpm and washed twice with 1 ml Tris-HCl (0.1 M, pH 9). Fungal cell pellets were then resuspended in 20 μl PBS and images with a Leica SP8 inverse confocal microscope.

Quantification of tongue neutrophils and inflammatory monocytes by flow cytometry
Mice were anesthetized with a sublethal dose of Ketamine (100 mg/kg, Streuli), Xylazin (20 mg/kg, Streuli), and Acepromazin (2.9 mg/kg, Fratro S.p.A.) and perfused by injection of PBS into the right heart ventricle prior to removing the tongue. Tongues were cut into fine pieces and digested with DNase I (200 μg/ml, Roche) and Collagenase IV (4.8 mg/ml, Invitrogen) in PBS at 37˚C for 45-60 min. Single cell suspensions were passed through a 70 μm strainer and stained in ice-cold PBS supplemented with 1% FCS, 5 mM EDTA, and 0.02% NaN3 with LIVE/DEAD Fixable Near-IR Stain (Life Technologies) and fluorochrome-conjugated antibodies against mouse CD45.2 (clone 104), CD11b (clone M1/70), Ly6G (clone 1A8) and Ly6C (clone AL-21). All antibodies were from BioLegend. Stained cells were analysed on a FACS Gallios (Beckman Coulter), or a SP6800 Spectral Analyzer (Sony). The data were analysed with FlowJo software (FlowJo LLC). The gating of the flow cytometric data was performed according to the guidelines for the use of flow cytometry and cell sorting in immunological studies [72]. including pre-gating on viable and single cells for analysis. Absolute cell numbers of lymphocyte populations were calculated based on a defined number of counting beads (BD Bioscience, Calibrite Beads), which were added to the samples before flow cytometric acquisition.

Preparation of tongue epithelial sheets and isolation of RNA for RNAseq
Epithelial sheets were prepared from infected tongues as described [20]. Briefly, the tongue was cut in half to obtain the dorsal part, which was freed from muscle tissue with a scalpel and

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa floated on PBS containing 2.86 mg/ml dispase II (Roche) for 60 min with the epithelial side facing upwards to separate the epithelial tissue from the lamina propria. Epithelial sheets were incubated in RNA-later (Sigma) for 1 min immediately after isolation. Three epithelial sheets were pooled per replicate, two replicates were generated for each condition.
For 'in vitro' samples, C. albicans hyphae were induced for 9 hours, harvested using a cell scraper, washed in PBS and rinsed in RNA-later. Two replicates were generated for each condition.
For RNA isolation, epithelial sheets and hyphae pellets were grinded in liquid N 2 and processed as described before [73] combining two phase separations and DirectZol RNA Mini-Prep kit (Zymo Research). DNase treatment was performed off-column using a DNA-free Kit (Life Technologies). RNA integrity was determined using a 2100 Bioanalyzer system (Agilent Technologies). RNA was denatured at 70˚C for 2 min prior to analysis. Samples were only included if the RNA integrity value (RIN) was above 7.5 or if two clear peaks for 18S and 25S/ 28S rRNA were present and no obvious degradation was observed.

Preparation of cDNA libraries and sequencing
cDNA libraries were generated using SureSelectXT multiplexed sequencing kit with RNA target enrichment for Illumina according to the manufacturer's instructions. In brief, mRNA was purified using poly(A) beads, fragmented, and double-stranded cDNA with ligated adapters was generated. The library was amplified using primers that match the adapters. C. albicans specific sequences were enriched using a hybridization technology based on biotinylated baits that cover 6'094 C. albicans ORFs but only a negligible fraction of mouse and human cDNA sequences from Ensembl, designed to specifically capture C. albicans-derived cDNA [27]. Amplified libraries were incubated with biotinylated baits at 65˚C for 24 h, and hybridized sequences were collected using magnetic streptavidin beads. The collected double-strand cDNA was then amplified and indexed in a separate PCR. For the analysis of RNA-seq samples from in vitro induced hyphae, 1% of RNA was used to spike RNA samples from naive mice, which were then subjected to the same enrichment procedure described above. RNA quality, fragment size and cDNA concentration were determined using a fragment analyzer automated CE system (Advanced Analytical) and a Qubit fluorometer (Invitrogen).
For the control on technical issues from the use of biotinylated baits-based technology to capture C. albicans-derived cDNA [27] for both isolates SC5314 and 101 as well as the use of SC5314 genome assembly for aligning sequencing data from both isolates, two duplicated RNA-seq samples from SC5314 (963bait and 964bait) and 101 (965bait and 966bait) grown in vitro were investigated. Similar percentage of reads were mapped to both isolates with about 98.8% for SC5314 and 98.1% for 101 (S1 Table).

RNA-seq data analysis
The sequencing was performed by Lausanne Genomic Technologies Facility (LGTF). cDNA libraries were subjected to cluster generation using the Illumina TruSeq PE cluster kit v3 reagents and sequenced on the Illumina HiSeq 2500 system with TruSeq SBS kit v3 reagents. Sequencing data were processed analogous to a previous study [27]: Data were processed

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa using Illumina Pipeline software (v1.82). Purity-filtered reads were adapters-and qualitytrimmed with Cutadapt (v1.2.1) (Martin, 2011) and filtered for low complexity with Prinseq (v0.20.3) [74]. After alignment against the Candida albicans genome SC5314 version A21-s02-m09-r07 using STAR (v2.5) [75,76], htseqcount (v0.5.4p3) was used to summarize the number of read counts per gene locus [77]. Genes with counts fewer than one per million in all samples were removed from the statistical analysis. Data normalization and differential expression analysis were performed in R (v3.2.2), using Bioconductor packages. Normalization of read count data was performed with the R package edgeR using the TMM (trimmed mean of Mvalues) method [78]. Normalized read count data was transformed to log2 counts per million by voom (implemented in the R package limma) [79]. The hierarchical clustering and principal component analysis were done in R (v3.2.2).
A linear model was built on the transformed data from the following conditions: SC_in_vitro, SC_d1, 101_in_vitro, 101_d1, 101_d3, and 101_d7 (all in duplicates). Differential expression analysis was then performed on the linear model with multiple contrasts representing the differences: (i) between the 101_in_vitro and SC_in_vitro conditions, and (ii) between the 101_d1/3/7 conditions and the SC_d1 condition. P-values produced from the differential analysis were adjusted using the Benjamini & Hochberg correction [80], and adjusted p-values (FDR) < 0.05 were considered significant. Heatmaps were generated with GENE-E (Broad Institute). GO process enrichment analysis was performed with the GO term finder integrated in the Candida Genome Database. The Gene Ontology (GO) term enrichment analysis was performed using DAVID Bioinformatics Resources 6.8 (NIAID, NIH). The enriched GO terms were hierarchically clustered with the semantic similarity between GO terms based on the graph structure of GO (Wang measure) using the R package GOSemSim (v2.12.1) [81] and plotted with the R package ComplexHeatmap (v2.2.0) [82].

Isolation of fungal RNA from infected keratinocyte monolayers
Monolayers of TR146 cells in 24-well tissue culture plates prepared as described above were infected with 4 x 10 4 -10 5 yeast cells per well. In case of infection with Dox -inducible isolates, 50 μg/ml (in case of SC5314 NRG1-OE and SC5314 crz2-OE ) or 10 μg/ml Dox (in case of 101 nrg1Δ/ pTET-NRG1 , unless stated otherwise) was added. After 24 h of incubation at 37˚C and 5% CO 2 , RNA was isolated with the RNeasy Mini Kit (Quiagen). Briefly, the cells were lyzed in RLT lysis buffer and homogenized with 0.5 mm glass beads (Sigma) using a Tissue Lyzer (Qiagen) 7 times for 2 minutes at 30 Hz interrupted by 30 sec cooling periods on ice. Cell debris were removed by centrifugation. One volume of 75% ethanol was admixed, and each sample was transferred to a RNeasy spin column. The loaded columns were washed using RW1 buffer and RPE buffer. RNA was eluted in 30 μl RNAse-free water.

Isolation of host RNA from infected TR146 cells and RHE
Host cell RNA was isolated with TRIZOL according to the manufacturer's instructions.

RT-qPCR
cDNA was generated by RevertAid reverse transcriptase (Thermo Fisher) and random nonamers. RT-qPCR was performed using SYBR Green (Roche) and a QuantStudio 7 Flex (Life Technology) instrument. All RT-qPCR assays were performed in duplicates and the relative expression levels of each gene was determined by normalization to the housekeeping gene. hG6PD was used as housekeeping gene in human and EFB1 or ACT1 in C. albicans, respectively. Using EFB1 or ACT1 as a housekeeping gene provided comparable results. The primers

PLOS PATHOGENS
C. albicans commensalism in the oral mucosa used to detect human transcripts are listed in Table 3, those to detect fungal transcripts are listed in Table 4.
Supporting information S1 Fig. (related to Fig 3). Isolate 101 persists in the oral mucosa even in presence of SC5314-induced inflammation. A. The proportion of each isolate 101-mCherry and SC5314-GFP in the infection inoculum 'mix' was analysed prior to infection. B. Gating strategy for quantification of neutrophils and inflammatory monocytes in the infected tongue on day 1 post-infection. (TIF)  Fig. (related to Fig 4). Read alignments parameters in RNAseq data analysis for SC5314 and 101. A. Gene body coverage represents the distribution of reads on each gene for each sample of strain SC5314 (963bait and 964bait) and 101 (965bait and 966bait) tested. B. Count distribution represents the distribution of gene counts between the four samples. C. In F and H, each symbol represents one animal, in I each symbol is the mean±SD of 4 animals. (TIF) S6 Fig. (related to Fig 6). Reduced expression of NRG1 in isolate 101 increases its pathogenicity. A. Representative images used for measuring hyphae length shown in Fig 6C for Fig. (related to Fig 7). Suppression of NRG1 expression in isolate 101 via a TET-off strategy drives fungal pathogenicity. A. Strategy used for generating isolate 101 nrg1Δ/pTET-NRG1 using a Dox shut-off system (pTet-off). Deletion of NRG1 was performed with a PCRgenerated KO cassette. Experimental details are given in the Methods section. FLP, Flippase; RFT, flippase recognition target; NAT1 and SAT1, nourseothricin resistance genes; CatTA, C. albicans tetracycline transactivator; HygR, hygromycin resistance. B. Functionality of the Tetoff system was tested with a luciferase reporter system. Median values of duplicate measurements for each condition are shown. RLU, relative luminescence units. C. Monolayers of TR146 keratinocytes were infected with isolates 101 nrg1Δ/pTET-NRG1 in presence of 0, 1 or 10 μg/ ml Dox for 24 hours. Expression of the indicated fungal genes was assessed by RT-qPCR. Bars are the mean+SD of 3 samples per condition from a single experiment. D. Representative images used for measuring hyphae length shown in Fig 7D for isolate 101 nrg1Δ/pTET-NRG1 with or without 10 μg/ml Dox on monolayers of TR146 keratinocytes in F12 medium. (TIF) S1 Table. (related to Fig 4). Alignment metrics of two duplicated control RNA-seq samples of SC5314 and 101. (DOCX) S2 Table. (related to Fig 4). Differential gene expression in multiple contrasts between the 101_in_vitro and SC_in_vitro conditions, between SC_day1 and SC_in_vitro conditions, between the 101_d1/3/7 conditions and the 101_in_vitro condition, and between the 101_d1/ 3/7 conditions and the SC_d1 condition.
for assessing Ece1 peptides in ECE1-overexpressing mutants; Ilse Jacobsen for sharing the KOH protocol; the Laboratory Animal Service Center of University of Zürich for animal care; the Laboratory Animal Model Pathology of University for histology; the Lausanne Genomic Technologies Facility of University of Lausanne for RNA sequencing; and members of the Lei-bundGut-lab for discussions.