Systematic gene overexpression in Candida albicans identifies a regulator of early adaptation to the mammalian gut

Abstract Candida albicans is part of the human gastrointestinal (GI) microbiota. To better understand how C. albicans efficiently establishes GI colonisation, we competitively challenged growth of 572 signature‐tagged strains (~10% genome coverage), each conditionally overexpressing a single gene, in the murine gut. We identified CRZ2, a transcription factor whose overexpression and deletion respectively increased and decreased early GI colonisation. Using clues from genome‐wide expression and gene‐set enrichment analyses, we found that the optimal activity of Crz2p occurs under hypoxia at 37°C, as evidenced by both phenotypic and transcriptomic analyses following CRZ2 genetic perturbation. Consistent with early colonisation of the GI tract, we show that CRZ2 overexpression confers resistance to acidic pH and bile salts, suggesting an adaptation to the upper sections of the gut. Genome‐wide location analyses revealed that Crz2p directly modulates the expression of many mannosyltransferase‐ and cell‐wall protein‐encoding genes, suggesting a link with cell‐wall function. We show that CRZ2 overexpression alters cell‐wall phosphomannan abundance and increases sensitivity to tunicamycin, suggesting a role in protein glycosylation. Our study reflects the powerful use of gene overexpression as a complementary approach to gene deletion to identify relevant biological pathways involved in C. albicans interaction with the host environment.

The biological circuitries under the control of TFs involved in the expression of C. albicans pathogenicity traits can be inferred from a combination of genome-wide expression and location (ChIP-chip/ Seq) technologies (Chen, Pande, French, Tuch, & Noble, 2011;Perez et al., 2013;Znaidi, Nesseir, Chauvel, Rossignol, & d'Enfert, 2013;Znaidi et al., 2014). As an example, the role of the Sfu1p and Sef1p TFs in C. albicans ability to respectively act as a commensal and a systemic-infection pathogen through modulation of iron homoeostasis has been elegantly shown by combining ChIP-chip and transcriptomics experiments (Chen, Pande, et al., 2011). The same approaches also demonstrated that RTG1, RTG3, TYE7, and LYS144 mediate GI tract colonisation by controlling the expression of genes involved in the acquisition and metabolism of specific nutrients, reflecting the importance of nutrient sensing/uptake during C. albicans commensalism (Perez & Johnson, 2014). Such systems biology-driven strategies are therefore cornerstone for mapping biological networks operating during C. albicans interaction with the host.
Gene overexpression is another powerful genetic approach for the discovery of pathways and phenotypes Douglas et al., 2012;Sopko et al., 2006). It mimics gain-of-function mutations, complements loss-of-function phenotypes, and allows the function of both essential and non-essential genes to be investigated (Prelich, 2012). Because diploidy and lack of a complete sexual cycle hamper the use of classical genetics in C. albicans, gene overexpression is regarded as an attractive alternative strategy for performing functional large-scale studies in this pathogen. We have previously established C. albicans strain collections for conditional gene-overexpression using the pNIM1 (Park & Morschhauser, 2005) and pNIMX (Chauvel et al., 2012) systems that respectively enable moderate and potent tetracycline derivative-induced expression (Chauvel et al., 2012). Strains from our collections carry unique 20-bp signature tags that allow simultaneous phenotyping in mixed-population experiments (Chauvel et al., 2012). We successfully used a collection with moderate, pNIM1-driven, overexpression of 531 open-reading frames (ORFs) for the identification of genes involved in biofilm formation, one of the major pathogenicity traits of C. albicans (Cabral et al., 2014). In the current report, we established a new library carrying 572 signature-tagged strains with potent, pNIMX-driven, conditional overexpression (Chauvel et al., 2012). We used it in a mouse model of GI tract colonisation to propose a new role for the TF Crz2p: the regulation of processes controlling the ability of C. albicans to efficiently proliferate within the host.

| Identification of genes whose overexpression alters colonisation of the mammalian GI tract
We screened our collection, in vivo, for genes whose overexpression alters C. albicans ability to colonise the murine gut ( Figure S1 for a schematic representation of our strategy). Mice were given gentamycin-and streptomycin-containing drinking water, supplemented (n = 5 mice) or not (control, n = 4 mice) with 2 mg ml −1 doxycycline, and then inoculated by gavage with~5 × 10 7 cells from the 572-strain pool ( Figure S1, see Section 4). Stools were collected 10 days post-gavage. Total genomic DNA was extracted from both faeces and inoculum samples, and the relative abundance of the strains was assessed by microarrays ( Figure S1). We found one hit out of the 572-competing strains displaying increased abundance in dox-treated mice (Figure 3a, upper panel, +Dox) and unaltered abundance in dox-untreated animals ( Figure 3a, lower panel, −Dox). This hit matched CRZ2 that encodes a zinc finger TF of the Cys 2 His 2 family FIGURE 1 Construction of a new tetracycline-inducible overexpression strain collection with enhanced overexpression efficiency. (a) Schematic representation of the StuI-linearised, signature-tagged (Barcode), overexpression vector that was used to create the 572-strain collection. Expression of each of the 572 open-reading frames (ORFs; orf19. XXXX; dark grey rectangle) is under the control of the tetracycline-inducible promoter (P Tet , black box), which is activated in the presence of doxycycline (horizontal arrow). The overexpression cassette is integrated at the RPS1 locus following a StuI digestion (3′ RPS1, 5′ RPS1). The ORF is flanked by the lambda-phage attachment sequences R1 and R2 (open triangles) that allowed recombination-mediated transfer of ORFs from an entry vector collection to the destination barcoded vectors. A unique 20-bp sequence (Barcode) serves as a strain identifier and allows the relative abundance of each strain to be quantified in mixedpopulation experiments. Every strain also carries the SacII, KpnI-linearised pNIMX construct (Chauvel et al., 2012) integrated at the ADH1 locus and carrying the Candida albicans reverse tetracycline transcriptional activator (cartTA) placed under the control of the TDH3 promoter (P TDH3 ). Transformant selection markers are depicted with open arrows. (b) Pie chart showing the functional categories of the 572 ORFs included in the overexpression strain collection. The number of ORFs is indicated between parentheses that was previously shown to be required for adaptation to low pH (Kullas, Martin, & Davis, 2007). To validate our microarray data, we performed quantitative PCR (qPCR) assays and specifically quantified the relative abundance of the CRZ2 overexpression strain in pooled mouse stools from dox-treated/untreated cages (housing three mice each) as compared with its abundance in the inoculum (Figure 3b).
The CRZ2-overexpressing strain was more abundant in stools from dox-treated mice, whereas its abundance was unaltered in stools from dox-untreated animals ( Figure 3b). Unchanged strain abundance was observed for four randomly selected strains (orf19.3088, PGA37, CNB1, and IHD1, Figure 3b). Under standard laboratory growth conditions, in vitro, the abundance (Figure 2, Table S2) or growth rate ( Figure S2A) of the CRZ2-overexpressing strain was unaltered, reflecting the specificity of our in vivo assay. We have also inspected colony size/morphology (e.g., GUT phenotype) and cellular morphology following passage through the mouse. We did not detect clear differences in morphology and/or size. We have also tested whether overexpression of CRZ2 could alter hyphal growth in both rich and hyphae-inducing media and found no difference compared with the control strain.
To test whether the CRZ2 overexpression phenotype could be reproduced in a 1:1 competition assay, we gavaged dox-treated and dox-untreated mice with an inoculum containing an equal mixture of a strain co-expressing the green fluorescent protein (GFP, under the control of P TDH3 ) and CRZ2 (under the control of P TET ) versus a strain expressing the blue fluorescent protein (BFP, under the control of P TDH3 ) and carrying an empty-vector control (see Section 4). Stools were recovered 4 days post-gavage, homogenised, and plated on gentamycin-and chloramphenicol-containing YPD medium during 2 days at 30°C. The resulting C. albicans colony-forming units (CFUs) FIGURE 2 Competitive fitness profiling of Candida albicans overexpression strains under standard laboratory growth conditions. (a) The effect of gene overexpression on cell growth was tested in liquid YPD medium at 30°C under vigorous shaking (standard laboratory growth conditions, normoxia) in the absence or presence of 40 μg ml −1 doxycycline (Dox) for 18 generations. The experiment was performed using n = 3 biological replicates. Samples were subjected to genomic DNA extraction, polymerase chain reaction-amplification, indirect fluorescent dye labelling (Doxtreated sample: Cy5-labelled; untreated control: Cy3-labelled), and hybridization to a barcode microarray with both forward and reversecomplemented probes for each tag (two black circles next to each gene name/orf19 nomenclature represent forward and reverse-complemented probe sequences). Fitness scores (Z-score for each tag) are shown on the y axis. The corresponding probe number ranked using the orf19 nomenclature in ascending order is shown on the x axis. Z-score calculations were performed using ArrayPipe v2.0. Dashed lines correspond to the Z-score values +2.0 (upper line) and −2.0 (lower line). Names or orf19 nomenclature of the genes whose overexpression alters strain fitness are shown. (b) Confirmation of the microarray data by liquid growth assay (YPD, 30°C) of strains overexpressing PPH21, orf19.1792, BUD5, FHL1, orf19.2781, and orf19.996, together with a control strain carrying the empty vector (CTRL) grown 3 times independently in a 96-well plate using a TECAN Infinite M200 device (see Section 4). Doubling time in hours (average from n = 3 biological replicates and error bars denote standard deviations) is indicated on the y axis for each strain grown in the absence (grey bar, −Dox) or presence (black bar, +Dox) of 40 μg ml −1 Dox; statistical significance was assigned (P < 0.05, asterisks) by performing a two-tailed Student's t tests. (c) Phenotypic analysis of a subset of strains overexpressing the indicated genes or control strain (CTRL). Strains were microscopically examined (40× magnification) immediately after being subjected to the fitness profiling assay described in (b) +Dox, dox treatment (40 μg ml −1 ); −Dox, untreated samples were pooled, and the relative abundance of strains carrying P TET -CRZ2 (GFP) and control plasmid (BFP) was quantified with fluorescenceactivated cell sorting (Figure 3c, see Section 4). The CFUs from doxtreated mice displayed increased median competitive index (CI:~2.9) for the GFP-positive (CRZ2) strain as compared with the BFP control strain (Figure 3c), whereas those from dox-untreated animals showed a median CI close to 1.0 ( Figure 3c). The effect of deleting CRZ2 was also examined using the same strategy, except that dox treatment was omitted from the assay because we used non-conditional CRZ2 deletion mutants (see Section 4). We tested two independently generated crz2−/− strains and found that both mutants failed to maintain efficient colonisation, as judged by the significant decrease in their median CIs (Figure 3d).
We further measured the relative abundance of the CRZ2 overexpression and deletion mutants at later time points using our 1:1 competition assay (Days 10 and 14). Surprisingly, the effect of CRZ2 overexpression and deletion on GI colonisation was not sustained over time ( Figure S2B,C), which may reflect an adaptive process confined to the upper sections of the digestive tract (see below). Yet, increased abundance of the CRZ2-overexpressing strain was still detected in our 1:571 competitive screen on Day 10 post-gavage ( Figure 3b). To explain this discrepancy, we simulated competitive growth of a strain with a fitness of either 2.2 or 2.1 relative to a strain with a fitness of 2.0 using two different ratios: 1:1 and 1:571 (Table S5). We calculated the corresponding CIs for cultures with a 1:1 ratio versus those with a 1:571 ratio (Table S5). Consistent with our observation, we found that CIs evolve differently according to the competition ratio and time, inuring to the benefit of the 1:571 comparison (see Graph in Table S5).

| Crz2p activates the hypoxic transcriptional program and modulates the expression of cell-surface genes
Crz2p was shown to regulate the expression of a subset of Zap1p targets involved in biofilm formation using nanoString transcript profiling of 293 genes (Finkel et al., 2012). To comprehensively define the regulatory network of Crz2p and better understand how CRZ2 contributes to colonisation of the GI tract, we performed whole-genome transcript profiling of the strain overexpressing CRZ2. The P TET -CRZ2 strain was grown in YPD medium at 30°C, in the absence or presence of dox, for 2 and 4 hr to investigate early (2 hr, those that could reflect direct targets) versus later (4 hr, including direct and indirect targets) transcriptional programming (see Section 4). At time point 2 hr, 110 and 66 genes were respectively upregulated (fold-change ≥1.5, P < 0.05) and downregulated (fold-change ≤−1.5, P < 0.05) in response to CRZ2 induction ( Figure 4a, Table S6). At time point 4 hr, 220 and 205 genes were respectively upregulated and downregulated using the same criteria ( Figure 4a, Table S6). Among the P TET -CRZ2 upregulated genes, we found a high proportion of those encoding (or predicted to encode) cell-surface proteins (e.g., PGA6, ECM331, PLB1, PLB4.5, KRE1, and SCW11), mannosyltransferases (e.g., MNN1, MNN24, MNN22, and BMT5), proteins involved in methionine/cysteine metabolism (e.g., MET15, MET3, MET10, and ECM17), and small molecule/amino-acid transporters (e.g., orf19.4690, SSU1, FRP3, FIGURE 3 CRZ2 contributes to Candida albicans fitness during gastrointestinal colonisation. (a) Genomic DNA was extracted from faeces of BALB/c female mice at Day 10 post-gavage with an inoculum of 572 P TET -inducible-overexpression strains. Barcodes were polymerase chain reaction-amplified from both faecal (Cy5-labelled) and inoculum (Cy3-labelled) genomic DNA and hybridized to barcode arrays ( Figure S1). Data are presented as described in legend to Figure 2a. The CRZ2 tag displays increased abundance (positive Zscore) in dox-treated mice (+Dox, n = 5 mice, P = 0.01) but not in doxuntreated animals (−Dox, Control, n = 4 mice). (b) Pooled faecal samples, at Day 10 post-gavage, from two independent cages housing three dox-treated (+Dox, black bars) or three dox-untreated (−Dox, grey bars) mice were subjected to genomic DNA extraction and up to seven quantitative polymerase chain reaction assays to determine the average fold-change abundance (y axis, error bars denote standard deviations) of strain P TET -CRZ2 (CRZ2) as compared with its abundance in the inoculum. Strains orf19.3088, PGA37, CNB1, and IHD1 were used as negative controls, and ACT1 served as a normalisation control. One representative result out of two is shown. Statistical significance was assessed by a two-tailed Student's t test (asterisk, P < 0.05). (c and d) Median competitive indexes (CIs, y axis) at Day 4 post-gavage of strains P TET -CRZ2 (green fluorescent protein [GFP]-labelled) versus control (blue fluorescent protein [BFP]-labelled) in dox-treated (+Dox, n = 6) and dox-untreated (−Dox, n = 12) mice (c) and two independent crz2Δ/crz2Δ mutants (crz2Δ/Δ #1 and #2, GFP-labelled) versus control (CTRL, BFP-labelled) or the parental wild-type strain (WT, GFP-labelled) versus control (CTRL, BFP-labelled) (d) were determined by flow cytometry analyses (see Section 4). Statistical significance was assessed using a non-parametric Mann-Whitney test (two-tailed; FRP6, and OPT1; Figure 4a, Table S6). On the other hand, many cell wall/surface genes (e.g., IFF11, RHD3, ALS4, CSP2, CSH1, ALS2, and PGA10) and genes involved in signalling (ASR2 and SRR1) and carbohydrate metabolism (PCK1, DLD2, and ARA1) were downregulated ( Figure 4a, Table S6). As a control, dox treatment alone does not significantly alter gene expression ( Figure S3). Only 13 and 9 genes were upregulated and downregulated with maximum fold-change values of 2.0 and −2.5, respectively (Table S7), including 10 genes displaying increased fold-change due to a skew in the distribution of high-intensity signal ( Figure S3B and Table S7).
To get a global view of the metabolic processes and pathways that were significantly transcriptionally modulated by CRZ2 FIGURE 4 Transcriptomic analysis of CRZ2 overexpression strains. (a) Heat maps of the 40 highest transcriptionally modulated genes (absolute n-fold changes are shown) in P TET -CRZ2 transcript profiling data at time points 2 and 4 hr post-induction with 40 μg ml −1 doxycycline (combination of three biological replicates in each condition). The most upregulated (descending signal intensity, sorted by average expression between 2 and 4 hr, left panel) or downregulated (ascending signal intensity, sorted by average expression between 2 and 4 hr, right panel) genes in dox-treated versus untreated cells are indicated with their corresponding name or orf19 nomenclature on the right side of each panel. Heat maps were constructed using Genesis version 1.7.6 (Sturn, Quackenbush, & Trajanoski, 2002). (b and c) Gene-set enrichment analyses maps at time points 2 hr (b) and 4 hr (c) post-induction of CRZ2 gene expression. Functional enrichment among the upregulated and downregulated genes is depicted with red and blue spheres, respectively. Sphere size is proportional to the number of genes. Network motif thickness is proportional to the extent of overlap in the list of genes between each node (blue/red spheres). Selected functional categories are indicated with dashed circles. Highlighted subnetworks are zoomed in (open large boxes in [b]). (d) Venn diagram of the overlap (45 genes) between genes that are induced by P TET -driven overexpression of CRZ2 (269 genes) and those that are transcriptionally upregulated in mouse gastrointestinal (GI) tract (408 genes) as identified by Rosenbach et al. (2010). Statistical significance (P = 1.1 × 10 −9 ) was assessed using a hypergeometric test. TF: transcription factor induction, we performed gene-set enrichment analyses (GSEA) and used the Cytoscape GSEA plugin (Sellam et al., 2014)  Rosenbach, Dignard, Pierce, Whiteway, and Kumamoto (2010) analysed the genome-wide transcriptional changes of C. albicans during GI tract colonisation as compared with laboratory growth conditions. They identified a set of 408 genes specifically upregulated in the mouse gut. We compared the list of genes that are significantly upregulated by P TET -CRZ2 to those identified in Rosenbach et al.
Taken together, our results indicate that CRZ2 overexpression turns off aerobic respiration and activates the hypoxic transcriptional signatures, while inducing the expression of cell wall genes together with genes involved in GI colonisation and interaction with the host.

| Crz2p binds to the promoter of mannosyltransferase-and cell wall protein-encoding genes
To determine if the CRZ2 transcriptionally modulated genes were directly regulated by Crz2p, we performed chromatin immunoprecipitation coupled with hybridization to whole-genome tiling microarrays in strains expressing a tandem affinity purification (TAP)-tagged versus a wild-type version of CRZ2, both placed under the control of P TET and grown for 4 hr in the presence of dox (see Section 4). Using the CisGenome peak finding algorithm (Ji et al., 2008), we identified 331 Crz2p-binding peaks (Tables S8-S10).
Among these peaks, 194 were located in promoter regions that were clearly associated with unique targets, 113 were located in promoters shared by two ORFs in opposite orientations, and the remaining 24 were found within ORF regions (Table S10). In many occurrences, more than one peak was found in the promoter of a given gene ( Figure

| The Crz2p regulatory network
We combined our genome-wide expression and location data to map the direct regulatory interactions occurring between Crz2p and its target genes ( Figure 6). This enabled the identification of additional bona fide Crz2p binding peaks matching 79 targets that were not detected by the peak-finding algorithm, and these were added to the list of Crz2p targets (Table S9, Figure 6). We found 102 targets that were both bound and transcriptionally induced by Crz2p ( Figure 6, red box) versus 63 genes that were both bound and transcriptionally repressed ( Figure 6, green box), indicating that Crz2p acts as both activator and repressor of gene expression.
A high proportion of the Crz2p directly induced genes encodes FIGURE 5 Genome-wide location of transcription factor Crz2p. (a) Crz2p occupancies along 25-kb intervals of selected locations from the Candida albicans genome (Assembly 20, the corresponding chromosome numbers are indicated at the left of each panel). Plotted are the relative signal intensities of the 60-bp probes covering the whole C. albicans genome following enrichment of the tandem affinity purification (TAP)-tagged Crz2p-coimmunoprecipitated DNA relative to DNA from a mock immunoprecipitation (i.e., in an untagged-strain background). Data from one ChIP-chip experiment out of two are shown. Some binding-enrichment signals extend beyond the maximum graduations (−5.0 and +5.0 fold-enrichment). The orientation of each open-reading frames is depicted by the arrowed black rectangle. Binding maps were generated using the Integrated Genomics Viewer genome browser (Thorvaldsdottir et al., 2012). (b) Quantification of DNA enrichment following immunoprecipitation of TAP-tagged Crz2p at the promoters of CRZ2, MNN1, ECM331, MNN22, RHD1, and RHD3 by quantitative polymerase chain reaction assays in strains C251, C252 (tagged), C255 and C257 (control strains; untagged). Bars represent absolute relative enrichment values (n-fold, y axis) of TAP-Crz2p coimmunoprecipitated DNA as compared with DNA from mock immunoprecipitation. Error bars denote standard deviations from the mean (using data from three quantitative polymerase chain reaction assays with two biological replicates in each assay, assumed as n = 6). All enrichment values were statistically significant using a two-tailed Student's t test that compares binding enrichment values at the indicated loci to those at the ACT1 locus (negative control; * P < 0.05; ** P < 0.01). (c) Motif logos of conserved sequences in Crz2p-enriched DNA fragments. DNA sequences encompassing ±250 bp around peak summits in Crz2p binding data were used as input for motif discovery by Regulatory Sequence Analysis

| CRZ2 mutants exhibit altered respiration
Our GSEA revealed that CRZ2 overexpression turns on and off the hypoxic and cellular respiration transcriptional signatures, respectively (Figure 4b). We challenged growth of the CRZ2 overexpression strain with Antimycin A, as C. albicans strains with reduced respiratory activity should show reduced sensitivity to this respiratory chain complex III inhibitor (Desai, van Wijlick, Kurtz, Juchimiuk, & Ernst, 2015; Figure 7a). Strikingly, a P TDH3 -CRZ2 strain overexpressing CRZ2 under the control of the constitutive P TDH3 promoter (Table S11)   The Crz2p regulatory network. Venn diagrams of the overlap between genes that are transcriptionally modulated by P TET -CRZ2 at time points 2 and 4 hr (gene expression fold-change ≥1.5; P < 0.05) and bound by Crz2p. Numbers in Venn diagrams indicate the number of genes, and those between parentheses indicate the total number of upregulated (light red circles), downregulated (light green circles), and bound (light blue circle) genes. Circled numbers indicate the number of genes that are both bound and transcriptionally modulated by Crz2p. The name of these genes (or their orf19 nomenclature) and the functional categories to which they belong are shown in the linked red (103 bound and upregulated genes) and green (63 bound and downregulated genes) boxes FIGURE 7 CRZ2 mutants exhibit altered respiration. (a) Antimycin A susceptibility of CRZ2-deficient (crz2Δ/crz2Δ) and CRZ2-overexpressing (CIp10-P TDH3 -CRZ2 and CIp10-P TET -CRZ2) strains was tested by spot assay on SD plates supplemented with 20 μg ml −1 antimycin A or with 20 μg ml −1 antimycin A + 3 μg ml −1 anhydrotetracycline (aTc). The CIp10-carrying (CIp10 and CIp10-P TET -GTW) and CRZ2/CRZ2 strains were respectively used as a control. Plates were incubated at 30°C for 2 days. (b) Tetrazolium salt-(2,3,5-triphenyltetrazolium chloride) containing overlay was poured on patches of the same strains to reconfirm respiration deficiency (final 0.05% triphenyltetrazolium chloride) A panel, compare CIp10-P TET -GTW to CIp10-P TET -CRZ2). In contrast, loss of CRZ2 resulted in increased sensitivity to Antimycin A (Figure 7 a, compare CRZ2/CRZ2 and crz2Δ/crz2Δ). We further confirmed the respiration deficiency phenotype of the CRZ2-overexpressing strains using the triphenyltetrazolium chloride (TTC) overlay assay (Figure 7 b), whereby reduction of TTC by the electron transport chain leads to the formation of a red product that accumulates within cells (Rich, Mischis, Purton, & Wiskich, 2001). As expected, the P TDH3 -CRZ2 strain lacked the characteristic red/pink colour indicative of efficient respiratory activity in both aTc-free and aTc-supplemented media (Figure 7b, compare CIp10 to CIp10-P TDH3 -CRZ2), whereas the P TET -CRZ2 strain displays respiratory deficiency only in the presence of aTc (Figure 7b, compare CIp10-P TET -GTW to CIp10-P TET -CRZ2). Thus, CRZ2 expression appears to control respiratory activity in C. albicans.

| CRZ2 deletion alters the expression of Crz2p targets in response to hypoxia at 37°C
Based on the altered respiration of the CRZ2 overexpression and deletion strains and the impact of CRZ2 overexpression on the induction of the hypoxic program, we hypothesised that CRZ2 could contribute to regulating adaptation to hypoxia when C. albicans is exposed to the GI tract environment. We analysed the transcriptome of a wild-type strain relative to that of a crz2Δ/crz2Δ mutant, both grown under normoxia 30°C or under hypoxia 37°C (Figure 8a, Table S6, see Section 4, 37°C being combined with hypoxia to mimic the GI tract environment). We found that the expression levels of CRZ2 itself werẽ 10-fold induced (Figure 8a, Table S6), correlating with those reached using P TET -driven overexpression (approximately eightfold induction, Figure 4a and Table S6). Importantly, the expression of many Crz2p bound targets was altered in the crz2Δ/crz2Δ mutant as compared with the wild-type strain (Figure 8a, asterisks). The crz2Δ/crz2Δ mutant failed to fully activate or maintain the expression of a subset of P TET -CRZ2-upregulated genes, such as MNN1, RME1, PLB4.5, UTR2, PGA6, orf19.6350, and orf19.3988 (Figure 8a and Table S6). Similarly, a subset of the P TET -CRZ2 downregulated genes displayed altered expression in the crz2Δ/crz2Δ mutant, including WH11, IFE2, ALD6, GAL10, PHO84, RHD1, orf19.3721, orf19.2959.1, and orf19.3722 ( Figure 8a and Table S6). We performed a qRT-PCR assay to confirm our observations using primers that specifically amplify reverse-transcribed RNA from MNN1, UTR2, PGA6 (upregulated in P TET -CRZ2), PHO84, RHD1 (downregulated in P TET -CRZ2), and ACT1 (control, Figure 8b). Taken together, our data indicate that an oxygen-scarce environment at 37°C induces transcriptional regulation by Crz2p.
FIGURE 8 CRZ2 deletion alters the expression of Crz2p targets in response to hypoxia at 37°C. (a) Heat maps of the 30 most upregulated (descending signal intensity, left panel) or downregulated (ascending signal intensity, right panel) genes in a CRZ2/CRZ2 wild-type strain (Wild-type) following a shift from normoxia at 30°C to hypoxia at 37°C and their corresponding fold-change intensities in the CRZ2 homozygous mutant derivative (crz2Δ/crz2Δ) subjected to the same environmental perturbation (normoxia 30°C to hypoxia 37°C). All displayed genes show significant differential expression in the wild-type strain as compared with the crz2Δ/crz2Δ mutant by analysis of variance analysis (P < 0.05). Genes are indicated with their corresponding name (or orf19 nomenclature) on the right side of each panel. Asterisks (*) indicate genes whose promoter was bound by Crz2p in ChIP-on-chip assay (Table S9). Heat maps were constructed using Genesis version 1.7.6 (Sturn et al., 2002). (b) The expression of MNN1, UTR2, PGA6, PHO84, RHD1, and ACT1 (control) during growth under hypoxia at 37°C relative to their expression under normoxia at 30°C in the wild-type strain (Wild-type, light grey bars) versus the crz2Δ/crz2Δ mutant (crz2Δ/crz2Δ, black bars) was quantified by quantitative reverse transcription polymerase chain reaction. Bars represent the average relative change in RNA abundance of the indicated genes, and error bars denote standard deviations (n = 3 independently grown strains in each condition). Asterisks indicate significantly different gene expression levels as compared with the corresponding ACT1 control using a standard Student's t test ( * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001) 2.9 | Genetic perturbation of CRZ2 under hypoxia 37°C alters Candida albicans sensitivity to highly acidic pH and bile salts Candida albicans adapts to different niches of the GI tract by optimising its growth and metabolism according to various parameters, including oxygen availability, pH variation (acidic in upper GI tract), bile-salt, and (micro-) nutrient availability (Noble, 2013;Perez & Johnson, 2014;Prieto, Correia, Pla, & Roman, 2016). Our observation that CRZ2 is required for early colonisation of the GI tract could reflect an adaptive response of C. albicans to the upper section of the GI tract, including the stomach and proximal intestine, where pH is highly acidic and bile salts are abundant, respectively (Begley, Gahan, & Hill, 2005). We tested whether genetic perturbation of CRZ2 alters C. albicans susceptibility to highly acidic pH (pH 3, Figure 9). We combined different temperature and oxygen-availability parameters to further examine the specificity of the CRZ2 phenotypes to hypoxia 37°C. As previously observed by Kullas et al. (2007), two independent crz2Δ/crz2Δ mutants already displayed sensitivity to acidic pH (Figure 9, upper panel). This phenotype was independent of growth temperature and oxygen availability (Figure 9, upper panel).
Strikingly, strains overexpressing CRZ2 showed marked resistance to acidic pH only under hypoxia 37°C (Figure 9, upper right panel, hypoxia 37°C), reinforcing the notion that CRZ2 exerts its optimal activity under this condition.
We similarly tested the susceptibility of CRZ2 overexpression and deletion strains on bile-salt containing medium (Figure 9, lower panel).
Growth of all tested strains was unaltered under normoxia 30°C, normoxia 37°C, and hypoxia 30°C (Figure 9, lower left and middle panels). Importantly, under hypoxia 37°C, the CRZ2 deletion strains displayed increased sensitivity to bile salts, whereas the CRZ2 overexpression strains were markedly resistant to the same compound ( Figure 9, lower right panel, hypoxia 37°C). Both pH 3 and bile-salt phenotypes were unlikely to reflect a more general CRZ2-dependent stress sensitivity occurring under hypoxia 37°C, as growth of the corresponding strains on media containing cadmium (ER stress inducer) or copper (oxidative stress inducer) was not altered ( Figure S5).
Taken together, our results further reinforce the requirement of hypoxia 37°C for Crz2p to optimally exert its function and suggest an important role of CRZ2 in C. albicans adaptation to stresses encountered in the upper sections of the GI tract.
2.10 | CRZ2 overexpression confers sensitivity to the N-glycosylation inhibitor tunicamycin and alters phosphomannan abundance Our finding that Crz2p directly regulates the expression of both mannosyltransferase-and cell wall-encoding genes led us to hypothesise that Crz2p could exert its protective role against stresses encountered in the upper GI tract by interfering with pathways affecting protein glycosylation. We tested the susceptibility of the CRZ2 overexpression strain to the N-glycosylation inhibitor tunicamycin, which blocks N-glycosidic protein-carbohydrate linkages, using a microtiter plate assay (Figures 10a, see Section 4). Under normoxia 30°C, two independent P TDH3 -CRZ2 strains were more susceptible to increasing concentrations of tunicamycin (Figure 10a, upper panel) than was the control or the crz2Δ/crz2Δ mutant strains. As this phenotype could also be a consequence of tunicamycin-induced ER stress (e.g., unfolded protein response), we also examined the susceptibility of the P TDH3 -CRZ2 strain to another potent inducer of ER stress, FIGURE 9 CRZ2 is required for adaptation to acidic pH and bile salts under hypoxia at 37°C. pH 3-(upper panels) and bile salt-(lower panels) susceptibility phenotypes of two independent crz2Δ/crz2Δ (#1 and #2) and the corresponding wild-type control (CRZ2/CRZ2) strains were analysed together with two independent P TDH3 -driven CRZ2 overexpressers and the matched control strain (P TDH3 -GTW) by spot assay on SD plates supplemented with 150 mM HEPES at pH 3 and 0.1% bile salts, respectively. Plates were incubated under normoxia at 30°C or at 37°C and during hypoxia at 30°C or at 37°C for 2 and 7 days, respectively dithiothreitol (Zhang, Heitman, & Chen, 2012) and found no difference in growth rate inhibition (Figure 10b, upper panel). Under hypoxia 37°C, overexpression and deletion of CRZ2 respectively increased and decreased tunicamycin sensitivity (Figure 10a, lower panel). The phenotypes were further validated by spot assay on solid medium containing tunicamycin ( Figure S5). No major growth rate difference was observed when strains were grown under the same conditions in the presence of increasing concentrations of the ER stress-inducing agent dithiothreitol (Figure 10b, lower panel), reinforcing again the specificity of the CRZ2 phenotype to tunicamycin treatment.
To investigate whether the tunicamycin phenotype correlates with altered glycosylation, we tested the effect of CRZ2 overexpression on N-glycan outer chain elaboration, by quantifying phosphomannan abundance using cell affinity to Alcian Blue dye (Hobson et al., 2004). We found that CRZ2 overexpression significantly decreased phosphomannan abundance under both normoxia 30°C and hypoxia 37°C (Figure 10c), unlike the CRZ2-deficient and wild-type strains. Taken together, our data indicate that CRZ2 overexpression interferes with pathways involved in protein glycosylation.

| DISCUSSION
Systematic gene overexpression is a powerful approach for linking genotypes to phenotypes and associating genes to biological pathways Sopko et al., 2006). Gene overexpression can particularly stimulate a specific activity and mimic gain-of-function mutations (Prelich, 2012). Our collection includes 572 strains (near 10% genome-coverage), each overexpressing one single ORF, using the potent pNIMX tetracycline-inducible expression system (Chauvel et al., 2012). Our previous study identified only five genes whose overexpression altered C. albicans fitness out of 531 competitively grown signature-tagged strains, in vitro, using the moderate pNIM1 overexpression system (Cabral et al., 2014). Here, in addition to the previously identified genes (Cabral et al., 2014), we found 21 overexpressers with altered fitness, indicating that increasing overexpression levels correlates with increased sensitivity of our assay.
These data also indicate that at least~10% of the C. albicans genome is still relatively robust to genetic perturbation by gene overexpression, even if we witnessed an increase in the number of hits (5 out of 531 vs. 25 out of 572; Cabral et al., 2014). Consistent with our previous findings, we did not detect genes whose overexpression increased fitness, probably because cells were grown under optimal in vitro growth conditions (YPD, 30°C, normoxia). Our assay also revealed that overexpression of a significant number of genes alters cellular morphology and this appears to translate into a fitness cost (Figure 2a,c). Many of the genes whose overexpression altered cell morphology were also found in our previous screen that analysed both fitness and morphogenesis of individually grown strains, including RAD53, FKH2, BEM1, and YCK2 (Chauvel et al., 2012). This further validates our competitive assay and our readout approach.
Although 572 overexpression strains were competitively screened in vivo for altered GI tract colonisation, only one hit was found to affect this process: CRZ2. This finding was unexpected, knowing that our collection includes a subset of some additional regulators FIGURE 10 CRZ2 overexpression interferes with pathways involved in protein glycosylation. (a and b) Tunicamycin (a) and dithiothreitol (b) susceptibilities of two independent parental (CRZ2 +/+), CRZ2 deletion (crz2 −/−), and CRZ2 overexpression (P TDH3 -CRZ2) strains under normoxia at 30°C and hypoxia at 37°C were determined by microtiter plate assay. The data are presented as the relative growth of the cells in tunicamycin-or dithiothreitol-containing medium (drug concentration range is shown on top of each panel) as compared with growth of the same strain in drug-free medium. The growth rate values (%) are illustrated using heat maps generated with Genesis version 1.7.6 (Sturn et al., 2002). (c) Alcian Blue binding assay. The reference strain BWP17 (wild-type control, CRZ2/CRZ2), two independent crz2 deletion strains (crz2Δ/crz2Δ #1 and crz2Δ/crz2Δ #2) together with strains overexpressing CRZ2 from the TDH3 promoter (CIp10-P TDH3 -CRZ2 #1 and CIp10-P TDH3 -CRZ2 #2), and the corresponding parental strain harbouring the empty vector control (CIp10) were grown under normoxia 30°C (open bars) or under hypoxia 37°C (filled bars) and then subjected to Alcian Blue dye binding assay as described in Section 4. Strains are indicated on the x axis, and the corresponding Alcian Blue binding values (μg/OD 600 ) are indicated on the y axis. The assay was performed 3 times independently with averaged values ± standard deviations involved in GI tract colonisation, such as RTG1, CPH2, LYS144, and EFH1 (Table S1). It is possible that the inherent noise generated through animal experimentation prevented the clear detection of these genes in our screen. Alternatively, longer colonisation time (>10 days) might have been required to resolve their overexpression phenotype, and this was shown in a previous loss-of-function screen, where the identified hits altered GI colonisation at different time points of stool collection (Perez et al., 2013). For instance, RTG1 deletion phenotype showed up on Day 9 following strain-pool inoculation, whereas those of LYS144 and EFH1 required extended time (Day 21 and upwards, Perez et al., 2013). As most of the known regulators of GI tract colonisation were identified through loss-of-function screens, another explanation is that gene overexpression does not necessarily lead to inversion of the gene-deletion phenotype (Hoon et al., 2008).
We could also have selected for strains with disadvantage in colonisation, including perhaps the fitness defect genes that were already detected in vitro (Figure 2). We believe that, in vitro, cells are under optimal growth in rich medium (with preferential carbon source, complete nutrient broth), which allows the detection of subtle phenotypes.
In the gut, however, the different stresses encountered during colonisation and adaptation to those stresses could have masked the effect of the fitness-defect genes that were seen in vitro.
CRZ2 encodes a zinc finger TF of the C 2 H 2 family specifically present among species of the CTG clade that also include commensal species residing in the gut of insects, such as Pichia stipitis, Candida tenuis, and Spathaspora passalidarum (Wapinski, Pfeffer, Friedman, & Regev, 2007). Although Crz2p is homologous to the calcineurin target Crz1p, it is not involved in calcium signalling (Karababa et al., 2006;Kullas et al., 2007), and its function has likely diverged from Crz1p through duplication of a common ancestor (Wapinski et al., 2007).
Here, we propose that Crz2p contributes to C. albicans adaptation during early days of GI tract colonisation, because the 1:1 competitive fitness advantage and defect of the CRZ2 overexpression and deletion strains, respectively, was not maintained during prolonged colonisation (i.e., >4-10 days, Figures 3c,d and S2B,C). A similar observation was reported for Wor1p, the regulator of the GUT phenotype (Pande, Chen, & Noble, 2013), where WOR1 overexpression increased competitive GI colonisation only within 14-21 days post-inoculation (Prieto, Roman, Alonso-Monge, & Pla, 2017). Failure of WOR1 overexpressers to maintain normal levels of GI colonisation during early days was explained by their increased sensitivity to bile salts (Prieto et al., 2017), suggesting that they are more fit for lower sections of the gut rather than upper sections where bile salts are excreted. Consistent with the hypothesis of preferential colonisation of the upper digestive tract, we show here that CRZ2 deletion and overexpression respectively increases and decreases C. albicans sensitivity to bile salts and acidic pH (Figure 9), which probably explains the early competitive growth advantage and defect of CRZ2 overexpression and deletion strains, respectively (Figure 3c,d). This reinforces the notion that different stages of fungal colonisation occur in the mammalian gut environment, as previously described (Prieto & Pla, 2015;Prieto et al., 2017 White et al., 2007), and Crz2p suggests that C. albicans has evolved an extended array of regulators probably acting as relays needed for efficient and sustained growth during early and later processes of C. albicans outgrowth in the GI tract.
We show here that CRZ2 is transcriptionally induced by hypoxia at 37°C (Figure 8a) and that forced overexpression of CRZ2 allows to confer resistance to the respiratory chain inhibitor Antimycin A ( Figure 7a) in addition to altering reduction of TTC ( Figure 7b); reinforcing the notion that the optimal activity of Crz2p occurs under conditions where oxygen is limited. Following a shift to hypoxia at 37°C, the expression of CRZ2 reached levels (~10-fold, Figure 8a) similar to those observed upon P TET -driven overexpression (approximately eightfold at 4 hr post-induction, Figure 4a). This indicates that our overexpression system successfully recapitulated the levels attained via physiological activation of Crz2p by hypoxia 37°C, reinforcing the biological relevance of our experimental strategy. How exactly Crz2p adapts to hypoxic conditions awaits further investigations.
One of the direct Crz2p targets that may contribute to adaptation to hypoxia and modulation of carbohydrate metabolism is TYE7 (Figure 6), encoding a basic helix-loop-helix TF involved in transcriptional regulation of glycolytic genes (Askew et al., 2009;Bonhomme et al., 2011). Indeed, Tye7p function was shown to be essential when the respiratory pathway is disrupted, such as in the presence of respiratory chain inhibitors or during growth in a hypoxic environment (Askew et al., 2009). Importantly, Tye7p plays a role in adherence to host cells and favours commensal colonisation (Bohm et al., 2017;Finkel & Mitchell, 2011;Pierce et al., 2013). It is tempting to hypothesise that TYE7 could be required, at least in part, for Crz2pmediated GI tract colonisation.
CRZ2 is part of a set of adherence regulators required for adhesion of C. albicans to abiotic substrates, some of them modulating the expression of CSTAR (cell-surface targets of adherence regulators) and HYVIR (hyphal growth or virulence) genes, but CRZ2 could not be linked to a specific target subgroup (Finkel et al., 2012). One could speculate that CRZ2 contributes to early gut colonisation by transiently increasing C. albicans adherence to components of the GI environment. Many putative mannoprotein-encoding genes are direct targets of Crz2p, including PGA6, PGA54, PGA38, IHD1, PHR2, PLB1, PLB4.5, ECM331, SAP9, UTR2, SUN41, and INT1 ( Figure 6) and carry the N-glycosylation signature Asn-X-Ser/Thr in their predicted protein sequence (predicted using NetNGlyc version 1.0). Some of them encode putative adhesins such as PGA6, PGA38, and FGR41 or were shown to affect C. albicans adhesion to host cells (SAP9, UTR2, and SUN41; Alberti-Segui et al., 2004;Albrecht et al., 2006;Hiller, Heine, Brunner, & Rupp, 2007). INT1 encodes an integrin-like surface protein required for efficient caecal colonisation in mice (Bendel et al., 2000;Kinneberg et al., 1999); however, the pleiotropic phenotypes of INT1 make it an unlikely contributor to CRZ2-specific function. At least three Crz2p direct targets (Sun41p, Tos1p, and Scw11p), functioning as secreted β-1,3-glucan-modifying enzymes, were shown to be highly abundant in culture supernatants of C. albicans grown under a variety of growth/stress conditions (Heilmann et al., 2013), suggesting that they rather participate in more general cell-wall remodelling processes.
However, Crz2p direct targets Utr2p, Plb4.5p, and Pir1p become much more abundant upon lowering the environmental pH (Klis & Brul, 2015). Indeed, we show here that CRZ2 confers survival to highly acidic and oxygen-scarce environments (Figure 9) similar to those encountered in the stomach. Although the mean intestinal pH in rodents is lower than that in man, the mouse stomach is particularly highly acidic (pH 3 to 4; McConnell, Basit, & Murdan, 2008). Crz2p directly controls the expression of PHR2 (Figure 6), encoding a cellwall β-glycosidase that is expressed at acidic pH (Muhlschlegel & Fonzi, 1997) and was shown to be required for virulence in a rat model of vaginal infection (De Bernardis, Muhlschlegel, Cassone, & Fonzi, 1998). Many Crz2p direct targets were also expressed at the C. albicans cell wall under hypoxia at 37°C in a vagina-simulative medium (Sosinska et al., 2008) as well as in an in vitro system that mimics colonisation of mucosal surfaces at acidic pH, including PHR2, ECM331, PIR1, UTR2, and TOS1 (Sosinska et al., 2011), further reinforcing the notion that Crz2p is an important contributor to C. albicans ability to grow in low-pH environments.
Candida albicans adaptation to the mammalian gut is tightly linked to nutritional requirements, such as the need for preventing iron toxicity (Chen, Blyth, Sorrell, & Slavin, 2011), the use of amino acids/allantoate and carbohydrates as a source of energy (Perez et al., 2013), and the catabolism of fatty acids/N-acetylglucosamine or phosphate uptake (Pande et al., 2013;Pierce et al., 2013). It is also linked to the ability of preventing bile-salt toxicity, particularly in the upper section of the gut. CRZ2 confers increased resistance to bile salts (Figure 9), which is considered a major selective pressure that shapes the structure of the microbial communities residing in the gut (Ridlon, Harris, Bhowmik, Kang, & Hylemon, 2016). Many bacterial species that colonise the GI tract play a role in bile salt metabolism and are able to make bile salts available as substrates for further modifications by the intestinal microbiota, mainly through the activity of bile salt hydrolases (Urdaneta & Casadesus, 2017). Our inspection of the fungal genomes suggests that the genome of only one Candida species, Candida kefyr, encodes a putative bile salt hydrolase (GenBank accession # AFC60678.1), whose BLAST analysis against the C. albicans proteome did not return any significant hit.
Bile-salt resistance is also conferred by the expression of multidrug efflux pumps and alteration of membrane lipid/protein composition (Ridlon et al., 2016). At least 15 Crz2p direct targets encode membrane transporters (Figure 6), two of which are members of the major facilitator superfamily: QDR3 and orf19.7554. Qdr3p is similar to Escherichia coli MdtM, a major facilitator superfamily drug/H + antiporter of the DHA1 subfamily that confers bile-salt resistance through catalysis of electrogenic bile salt/H + antiport (Paul et al., 2014). Whether QDR3 confers Crz2p-mediated resistance to bile salts awaits further investigation. As CRZ2 genetic perturbation could affect cell-wall integrity, one could speculate that it may also alter the composition of membrane lipids and proteins with the consequence of decreasing C. albicans susceptibility to bile salts.
We propose that the combined action of the hypoxic environment within the GI tract and the required adaptation to low pH and bilesalt toxicity in the upper digestive tract are crucial for CRZ2's role in C. albicans ability to thrive within the gut. Our findings and those already reported by others suggest that C. albicans adaptation to the gut is a complex process requiring tight and combined control of sensing abilities, extended morphological and metabolic activities, and specific adaptive responses.

| Candida albicans strains
The C. albicans signature-tagged overexpression collection was constructed as described previously (Cabral et al., 2014;Chauvel et al., 2012; see Table S1 for a list of all ORFs included and corresponding barcode sequences and Table S11 for additional strains used in this study). Briefly, the respective ORFs were PCR amplified using previously designed chimeric primers (Cabral et al., 2014) followed by recombination-mediated transfer into the Gateway donor vector pDONR207 (Invitrogen). The set of pDONR207 derivatives was fully sequenced to ascertain that no unintended mutations were introduced during PCR amplification. The pDONR207-ORF plasmids were then used in a Gateway LR reaction together with barcoded derivatives of the CIp10-P TET -GTW vector (Cabral et al., 2014;Chauvel et al., 2012). All barcoded overexpression vectors were linearised with StuI and used to transform strain CEC2908, a derivative of the BWP17AH strain that carries the pNIMX plasmid (isogenic to CEC2907 described in Chauvel et al., 2012, Table S1). Transformants were selected and checked by PCR for correct integration yielding 572 overexpression strains (Table S1). Seven strains carried altered barcode sequences and were therefore omitted from the microarray analyses (Table S1, strains marked in red). The barcoded CIp10-P TET -CRZ2 or CIp10-P TET -GTW control plasmids were used to transform strains CEC3783 or CEC3781 (Cabral et al., 2014) carrying the pNIMX plasmid for doxycycline-regulated expression from the P TET promoter and either a P TDH3 -BFP or P TDH3 -GFP gene fusion for constitutive expression of BFP or GFP, respectively, generating strains CEC4442 or CEC4439, respectively (Table S11).
The C. albicans crz2Δ/crz2Δ mutants were generated in the BWP17 (ura3, arg4, and his1 auxotroph) background by successive replacement of the complete ORF from both alleles using PCRgenerated disruption cassettes flanked by 100 bp of target homology region as previously described (Gola, Martin, Walther, Dunkler, & Wendland, 2003). The disruption cassettes were amplified using oligonucleotides CRZ2_F_KO and CRZ2_R_KO (see Table S12 for primers used in this study) and ARG4-or HIS1-bearing plasmids. The resulting transformants were verified by PCR, and two independent clones were selected for subsequent transformation with plasmid CIp10 or plasmid CIp10-P TDH3 -GFP yielding the prototrophic crz2Δ/crz2Δ mutants C89 and C90 (two independent CIp10 plasmid integrants) or CEC4263 and CEC4265 (CIp10-P TDH3 -GFP integrants that constitutively express GFP), respectively (Table S11). Strain CEC155 (Table S11; Firon et al., 2007) was first transformed with a PCRamplified DNA fragment carrying both P TDH3 -BFP fusion and HIS1 marker such that the whole cassette integrates between the PGA62 and PGA59 loci, as previously described (Cabral et al., 2014). The resulting strain was transformed with plasmid CIp10 to generate the prototrophic strain CEC4425 (Table S11) that we used as a BFPlabelled control for our 1:1 competitive assays (Figure 3c,d). Strains overexpressing CRZ2 from the constitutive TDH3 promoter were generated by transferring the CRZ2 ORF from the respective pDONR207 derivative to the Gateway-compatible CIp10-P TDH3 -GTW plasmid (Legrand et al. 2018) followed by transformation of strain BWP17AH (CEC161; Chauvel et al., 2012) to generate three independent P TDH3driven CRZ2-overexpression clones (C26, C27, and C28; Table S11).
The CEC161 strain was also transformed with CIp10-P TDH3 -GFP, generating strain CEC4267 (Table S11), to serve as a control for phenotypic assays.
To create a TAP epitope-tagged version of CRZ2, the CRZ2 ORF was transferred from the corresponding pDONR207 to the Gateway-compatible CIp10-P TET -TAP-GTW that allows P TET -driven expression of N-terminally-tagged ORFs (Legrand et al., 2018, in press). The ura3 auxotrophic derivative of the crz2Δ/crz2Δ mutants was first transformed with plasmid pNIMX then with CIp10-P TET -TAP-CRZ2 or CIp10-P TET -CRZ2 (untagged control) to generate strains C251, C252 (two independent TAP-tagged clones) and C255, C257, C258 (three independent untagged control clones), respectively (Table S11). Expression of the TAP-Crz2p fusion was confirmed by Western blotting as described in Znaidi et al. (2013).

| Preparation of strain pools
The 572 signature-tagged overexpression strains were thawed on Nunc omnitray plates (Thermo Scientific) containing YPD (1% Yeast Extract, 2% Bacto-Peptone, 2% D-glucose)-agar supplemented with 50 μg ml −1 gentamycin using a 96 pin replicator and allowed to grow for 4 days at 30°C. No major colony size alterations were noticed. Five millilitres of YPD were added to each plate, and colonies were scraped off using a cell spreader. Strains were pooled in~100 ml YPD/15% glycerol at a concentration of~50 or~132 OD 600 (optical density at 600 nm) units/ml, aliquoted in 1.5-ml tubes and frozen at −80°C.

| In vitro competitive fitness assay
An aliquot (1.6 μl) from the frozen 50 OD 600 -unit strain-pool was used to inoculate 2 ml of YPD medium (starting OD 600 = 0.0625) and grown at 30°C with agitation (200 rpm) for 20 generations, in the absence or presence of 40 μg ml −1 doxycycline. Genomic DNA was extracted using the MasterPure Yeast DNA Purification Kit (Epicentre) and quantified using a NanoVue Plus device (GE Healthcare Life Sciences).
Barcodes were PCR-amplified using primers CipSAC2-UP-2 and CipSAC2-DWN-2 (3 min at 94°C; followed by 35 cycles of 30 s at 94°C, 30 s at 50°C, and 30 s at 72°C; and a final step of 7 min at 72°C; Table S12). The PCR products were then purified and subjected to indirect differential fluorescent dye labelling (Cy5 for Dox-treated, Cy3 for untreated pools). Labelled DNA was resuspended in 50 μl DigEasy Hyb solution (Roche), incubated at 95°C for 5 min, snapcooled on ice, and directly deposited on a barcode microarray that we previously described (Agilent Technologies, GEO platform # GPL17420; Cabral et al., 2014) containing (a)~12 on-chip replicates of both sense and antisense DNA sequences complementary to 657 tags (representing 572 strain tags +78 unused tags) and (b) different negative control spots (Agilent reference). Hybridization was performed overnight at 25°C, followed by washing and scanning of the arrays using GenePix 4200 AL scanner (Molecular Devices). This experiment was performed 3 times independently. Microarray data were analysed using ArrayPipe v2.0 (Hokamp et al., 2004). Z-score (i.e., number of standard deviations from the population mean) calculations were performed using ArrayPipe v2.0, and thresholds for considering significant deviation from the population were set at absolute Z-score values ≥2 and P values <0.05. Only strains with tags that met thresholds for both sense-and antisense-barcode fluorescence signals were considered as altered in cell fitness. Microarray data have been deposited at GEO under accession number GSE67215, and Z-scores and fold-change data are available in Table S2.

| Confirmation of the in vitro fitness-profiling data by liquid growth assay or microscopic examination
Strains were individually grown 3 times independently in 96-well plates at a starting optical density (OD 600 ) of 0.1 in 100 μl of YPD supplemented or not with 40 μg ml −1 doxycycline. The OD 600 was measured every 5 min using a TECAN Infinite M200 reader. The temperature was set at 30°C. TECAN OD 600 readings were converted into "flask OD 600 " reading using the following formula: OD Flask = OD Tecan × 12.2716 -1.0543 (Ericson, Hoon, St Onge, Giaever, & Nislow, 2010), and doubling times were calculated within the exponential growth interval as previously described (St Onge et al., 2007). Strains displaying altered morphology or cell-cell aggregation phenotypes were microscopically examined with a Leica DM RXA microscope (Leica Microsystems) at a 40× magnification.

| In vivo competitive GI tract colonisation screen
Microarray data have been deposited at GEO under accession number GSE67215, and Z-scores and fold-change data are available in Tables S3 and S4. An overview of the design of the competitive fitness profiling experiment in mice is shown in Figure S1.

| Validation of the GI tract colonisation screen
Q-PCR assays were performed to quantify the relative abundance of strains carrying P TET -CRZ2 (test), P TET -PGA37, P TET -orf19.3088, P TET -CNB1, and P TET -IHD1 (controls) using genomic DNA extracted from pooled stools at Day 10 post-gavage from two independent doxtreated and dox-untreated cages (each housing three mice) relative to the inoculum. Forward primers CRZ2-BC, PGA37s-BC, orf19.3088-BC, CNB1-BC, and IHD1-BC (Table S12) complementary to the 20-bp barcode sequence of the corresponding strains were used in combination with the reverse primer CIPSAC2-DWN3 (Table S12) (White et al., 2007). Competitive assays using P TET were performed up to 4 times independently totalling n = 5 to 12 dox-untreated mice and n = 5 to six dox-treated animals (Figures 3c and S2B, six doxtreated mice showed signs of illness and were removed from the analysis). Two reproducible competition assay experiments were performed for the crz2Δ/crz2Δ mutants #1 and #2 versus wild-type strain (n = 5 mice in each condition, one representative experiment is shown in Figures 3d and S2C). A two-tailed non-parametric Mann-Whitney test was used for assigning statistical significance (P < 0.05).

| Whole-genome transcript profiling experiments
For the P TET -CRZ2 microarray experiment (Figure 4), total RNA was extracted from three independently generated crz2Δ/crz2Δ strains carrying the P TET -CRZ2 fusion (C255, C257, and C258; Table S11), pre-grown overnight in 10 ml YPD at 30°C and then diluted in fresh YPD medium supplemented or not with 40 μg ml −1 doxycycline to an OD 600 of 0.3 and regrown for 2 and 4 hr. For the normoxia 30°C-to-hypoxia 37°C shift microarray experiments (Figure 8), strains CEC369 (WT) and C90 (crz2Δ/Δ; Table S11) were grown overnight in 10 ml YPD at 30°C and then diluted to an OD 600 of 0.16 in 250-ml flasks containing 50 ml of YPD medium. The flasks were incubated in a BBL GasPak anaerobic jar at 37°C for 24 hr without shaking (hypoxia 37°C) or in a 30°C incubator for 24 hr under vigorous shaking (normoxia 30°C). Total RNA was extracted from 50 OD units using the hot phenol method as described previously (Znaidi et al., 2013), followed by first-strand cDNA synthesis and Cy5 (dox-treated cDNA samples for the P TET -CRZ2 experiment and hypoxia 37°C-treated cDNA samples for the normoxia 30°C-to-hypoxia 37°C shift experiment)/Cy3 (untreated cDNA samples for the P TET -CRZ2 experiment and normoxia 30°C-treated cDNA samples for the normoxia 30°Cto-hypoxia 37°C shift experiment) labelling from 20 μg total RNA, using the SuperScript III indirect cDNA labelling system (Invitrogen).
Purified labelled samples were mixed and hybridized to a C. albicans expression array (Agilent Technologies) designed such that two nonoverlapping probe sets target each of 6,105 C. albicans ORFs from Assembly 19 for a total of 15,744 probes, thereby allowing two independent measurements of the mRNA level for a given gene (Znaidi et al., 2013). As an additional control experiment, we also compared the transcriptome of strain C90 (Cy5-labelled) to that in crz2Δ/Δ (CEC369, Cy3-labelled) under normoxia 30°C and under hypoxia 37°C (Table S6). Hybridization was performed as described elsewhere (Znaidi et al., 2013). Images of the Cy5 and Cy3 fluorescence were generated by scanning the expression arrays using an Axon Autoloader 4200AL scanner (Molecular Devices, Downingtown, PA, USA).
Images were analysed with the GenePix Pro 6.1.0.2 software (Molecular Devices). GenePix Results files were imported into the ArrayPipe 2.0 software for spot filtering, background subtraction (limma normexp BG correction), and Lowess global normalisation of signal intensities (Hokamp et al., 2004). Replicate arrays (n = 3) were combined, and fold-change and P values (standard Student's t test within group) were calculated. The complete expression profiling datasets are available in Table S6. Expression microarray data have been deposited at GEO under accession number GSE67226.

| Chromatin immunoprecipitation and ChIP-chip
Strains C251, C252 (TAP epitope-tagged), and C255 and C257 (untagged control; Table S11) were grown overnight in 2 ml YPD at 30°C, diluted to an OD 600 of 0.3 in 50 ml YPD medium supplemented with 40 μg ml −1 doxycycline and grown during 4 hr at 30°C. The subsequent steps of DNA cross-linking, DNA shearing, and chromatin immunoprecipitation (ChIP) were conducted as described in Liu et al. (2007), with some modifications. Briefly, cultures were treated with 1% formaldehyde (cross-linking) and snap-frozen in liquid nitrogen.
ChIP-chip raw data were deposited at GEO under accession number GSE67233.
Assays were performed (a) at least 2 times using three biological replicates for the qRT-PCR experiments or (b) 3 times independently with two biological replicates each time. A two-tailed Student's t test was applied by comparing, for a given gene/ChIP target, the n-fold relative gene-expression values to the corresponding n-fold values of the ACT1 control ( Figures S4 and 8b) or, for ChIP-qPCR, the n-fold enrichment values of the selected target gene (CRZ2, MNN1, ECM331, MNN22, RHD1, or RHD3) to those of the corresponding ACT1 control (Figure 5b). Statistical significance threshold was P < 0.05.

| Alcian Blue binding of cell wall mannans
Alcian Blue binding assay was performed as described in Odani, Shimma, Tanaka, and Jigami (1996). Briefly, cells were grown in YPD medium at 30°C for 48 hr or at 37°C for 48 hr under hypoxic conditions and then harvested by centrifugation, washed with 0.02 M HCl, and resuspended in 1 ml of 100 μg ml −1 Alcian Blue HCl solution (Merck Millipore, Germany). After incubation at room temperature for 10 min, cells were pelleted, and absorbance (OD 600 ) of the supernatant was determined in a microplate reader. Alcian Blue binding was calculated following the formula: (μg/OD600) = 61.3 × (OD 600 original solution − OD 600 supernatant)/OD 600 cell density.
Growth was measured spectrophotometrically at OD 600 after 24 hr of incubation at 30°C in a humid chamber or after 4 days under hypoxia at 37°C.
Briefly, the GSEA PreRanked tool (http://www.broadinstitute.org/ gsea/) was used (default parameters) to determine if our lists of ranked genes (P TET -CRZ2 data at 2 and 4 hr sorted from the highest upregulated genes to the highest downregulated ones) display a significant bias with any gene sets from a compendium of published microarray data, GO term categories/pathways, and data derived from S. cerevisiae resources (Sellam et al., 2014). The Cytoscape EnrichmentMap plugin was used to further visualise the GSEA network using the default parameters (Figure 4b,c). For motif discovery analyses, peak summit location files generated by CisGenome (Tables S8 and S9; Ji et al., 2008) were imported into the Galaxy NGS analysis pipeline (https:// main.g2.bx.psu.edu/), and DNA sequences encompassing ±250 bp around peak summits in Crz2p datasets were extracted using the Extract Genomic DNA tool version 2.2.2. The resulting sequences together with an equivalent set of randomly chosen sequences from the C. albicans genome (used as a negative control) were used as input for motif discovery with the Regulatory Sequence Analysis Tools were the recipients of post-doctoral fellowships from the European