Transcription Profiles Associated with Inducible Adhesion in Candida parapsilosis

ABSTRACT Candida parapsilosis has emerged as a frequent cause of invasive candidiasis with increasing evidence of unique biological features relative to C. albicans. As it adapts to conditions within a mammalian host, rapid changes in gene expression are necessary to facilitate colonization and persistence in this environment. Adhesion of the organism to biological surfaces is a key first step in this process and is the focus of this study. Building on previous observations showing the importance of a member of the ALS gene family in C. parapsilosis adhesion, three clinical isolates were cultured under two conditions that mimic the mammalian host and promote adhesion, incubation at 37°C in tissue culture medium 199 or in human plasma. Transcriptional profiles using RNA-seq were obtained in these adhesion-inducing conditions and compared to profiles following growth in yeast media that suppress adhesion to identify gene expression profiles associated with adhesion. Overall gene expression profiles among the three strains were similar in both adhesion-inducing conditions and distinct from adhesion-suppressing conditions. Pairwise analysis among the three growth conditions identified 133 genes that were differentially expressed at a cutoff of ±4-fold, with the most upregulated genes significantly enriched in iron acquisition and transmembrane transport, while the most downregulated genes were enriched in oxidation-reduction processes. Gene family enrichment analysis identified gene families with diverse functions that may have an important role in this important step for colonization and disease. IMPORTANCE Invasive Candida infections are frequent complications of the immunocompromised and are associated with substantive morbidity and mortality. Although C. albicans is the best-studied species, emerging infections by non-albicans Candida species have led to increased efforts to understand aspects of their pathogenesis that are unique from C. albicans. C. parapsilosis is a frequent cause of invasive infections, particularly among premature infants. Recent efforts have identified important virulence mechanisms that have features distinct from C. albicans. C. parapsilosis can exist outside a host environment and therefore requires rapid modifications when it encounters a mammalian host to prevent its clearance. An important first step in the process is adhesion to host surfaces. This work takes a global, nonbiased approach to investigate broad changes in gene expression that accompany efficient adhesion. As such, biological pathways and individual protein targets are identified that may be amenable to manipulation to reduce colonization and disease from this organism.

in the United States (1). Infections caused by C. albicans continue to dominate in many settings, but non-albicans Candida species such as C. parapsilosis, C. glabrata, C. krusei, C. tropicalis, and C. auris are emerging as significant causes of systemic infection, especially in the setting of immune compromise (2). Resistance against first-line and second-line antifungals is increasing in C. parapsilosis and C. glabrata, and multidrug resistance has been reported for C. auris (3,4). Non-albicans Candida species exhibit unique biology and distinct mechanisms of pathogenesis but are poorly understood.
Among the non-albicans Candida species, C. parapsilosis has generated interest based on unique aspects of its biology and emergence in many parts of the world as an invasive pathogen nearing or exceeding the incidence of C. albicans (5). It is also a frequent cause of disease in neonatal intensive care units (6). Among its clinically relevant attributes are the ability to form biofilms on central venous catheters and other medical devices and its capacity for horizontal transmission from environmental sources, including soil, water, and the hands of health care workers (7)(8)(9). Because C. parapsilosis maintains a reservoir in the environment, rapid modifications in gene expression are likely to be necessary to adapt to niches within a host. Adhesion is a necessary first step in successful colonization and potential invasive disease in a susceptible host, leading our group and others to investigate the mechanisms involved in adhesion of the organism to mucosal and biotic surfaces both in vitro and in vivo (10)(11)(12)(13).
We recently described an adhesion assay to measure binding of C. parapsilosis to extracellular matrix proteins such as albumin and fibronectin under fluid shear (13). We observed that certain clinical isolates of C. parapsilosis exhibited enhanced adhesion when grown in mammalian tissue culture medium 199 (M199) or human serum. The ALS gene family is well known to encode important adhesins in C. albicans (14), and recent work has defined the unique features of the ALS family in C. parapsilosis (15). We found that CpALS7 (CPAR2_404800 or CpALS4800 [15]) was significantly upregulated when a clinical C. parapsilosis isolate was grown under conditions that led to adhesion (13). Because these conditions model the host environment to which yeast must adapt to successfully cause systemic disease, we hypothesized that clinical isolates of C. parapsilosis differentially regulate many genes that are important for virulence in the presence of these host stimuli. In the current study, we used RNA-seq to compare the transcriptional profiles of three adhesive clinical isolates of C. parapsilosis under adhesion-promoting or adhesion-suppressing growth conditions to identify gene expression patterns associated with these phenotypes.

RESULTS
Adhesion of C. parapsilosis isolates. To confirm that adhesion was induced in the same samples that were to undergo transcriptional profiling, adhesion of all 3 clinical isolates of C. parapsilosis to immobilized bovine serum albumin (BSA) was measured under physiologic fluid shear as previously reported (13). All three isolates showed relatively weak adhesion when grown in yeast extract-peptone-dextrose (YPD). After 3 h growth in M199, all isolates showed significantly increased adhesion compared to YPD (Fig. 1). JMB81 showed a moderate increase in adhesion in M199, but JMB77 and Ro75 showed much greater increases, with 80 to 90% of the entire channel surface area carpeted with yeast.
Transcriptional profiling of C. parapsilosis by RNA-seq. Total RNA was prepared from all three isolates grown in YPD, M199, or pooled human plasma in three independent biological replicates. Next-generation sequencing using the Illumina platform was completed as described in Materials and Methods. Summary characteristics of the RNA-seq data are provided in Table 1.
The gene expression profiles of three adhesive C. parapsilosis isolates were compared in adhesion-inducing or adhesion-noninducing growth conditions. Pairwise differential gene expression analyses of isolates grown in each of the three growth conditions showed that there were 133 differentially expressed genes (log 2 -fold change 62, adjusted P , 0.001) among the three conditions (Table S1). The most upregulated genes in adhesion-inducing conditions were significantly enriched (corrected P # 0.05) in iron acquisition and transmembrane transport, while the most downregulated genes were enriched in oxidation-reduction processes. Profiles of differentially expressed genes from isolates grown in adhesion-inducing growth media, M199 and plasma, were similar to each other ( Fig. 2A). Their profiles were significantly different from those grown in YPD as illustrated by a correlation plot of the normalized expression counts for the 133 differentially expressed genes (Fig. 2B).
Volcano and MA plots (Fig. 2C) of the expression data show that the majority of differentially expressed (DE) genes in our comparisons were those differentially expressed between nonadhesive and adhesive growth conditions, YPD versus M199 (78 DE genes) and YPD versus plasma (96 DE genes), while relatively few genes were differentially  Transcription in C. parapsilosis Adhesion expressed between the two adhesion-inducing growth media, M199 and plasma (28 DE genes). The MA plots show that many of the genes that were differentially expressed in both of the adhesive versus nonadhesive condition comparisons were among the most highly expressed genes in the genome. The 20 genes that demonstrated the highest upregulation based on fold increase in expression in adhesive conditions (M199, plasma) compared to nonadhesive conditions (YPD) are listed in Table 2 along with their respective normalized count values from the differential expression analysis R package, DESeq2. They were ranked by a ratio of mean fold change between adhesive and nonadhesive conditions in descending order. Homologues of each gene in C. albicans, and Saccharomyces cerevisiae are indicated where known. The genes that showed the highest upregulation in adhesive conditions included homologues of INO1, PHO89, PHR1, and CDG1. Each showed a 50-fold or greater increase in mean normalized count values. Complete tables of differential expression results for each of the three pairwise comparisons are included in the supplemental material (Data Set S1).
Genes that have been previously suggested to have a role in adhesion of Candida species were identified using Gene Ontology annotations from the Candida Genome Database. The expression of the homologous genes in C. parapsilosis was compared in YPD, M199, and plasma in this subset of genes (Fig. 3). As expected, a number of these genes showed a greater than 2-fold change in expression in M199 or plasma compared to YPD. Very few genes reached this threshold when expression in M199 and plasma was compared.
A comparative analysis of the results of each of the three differential expression comparisons using a three-set Venn diagram (Fig. 4) showed that, of the 105 genes which were differentially expressed in isolates grown in adhesive growth conditions compared to those grown in nonadhesive growth conditions, 50 genes were differentially expressed in both YPD versus M199 and in YPD versus plasma. There were 35 unique genes which were differentially expressed in YPD versus plasma and 20 unique genes which were differentially expressed in YPD versus M199.
The most enriched GO terms among each of the differential expression comparisons for each set of differentially expressed genes are included in the supplemental material (Table S2).
Gene family enrichment analysis. Since C. albicans orthologue annotations are typically used to identify gene family members, genes that are unannotated in C. albicans or unique to C. parapsilosis may be overlooked which exist in biologically significant gene families. Gene families of differentially expressed genes in C. parapsilosis for each of the three condition comparisons were identified using BLASTp similarity searches and the Markov clustering algorithm (16-18) as described in Materials and Methods. For each of the three comparisons, gene families of differentially expressed genes were arranged from highest to lowest change in expression (Fig. 5). The FRE, CFEM, PHR, siderophore transporters, and MNN4 families were among the most highly upregulated gene families in adhesion-inducing conditions. Detailed gene family clustering results are included in the supplemental material (Table S3). Validation by qPCR. To provide independent validation of the RNA-seq data, the expression of 26 genes that showed various degrees of changes in expression was measured by qPCR specific for each gene. Figure 6 shows correlation plots comparing qPCR (DC t relative to actin) to RNA-seq data (fragments per kilobase per million [FPKM] values for each gene minus actin). In each isolate, the correlation between qPCR and RNA-seq was high, with R 2 values greater than 0.74 for all isolates. qPCR of select differentially expressed genes in a nonadhesive strain. Genes that are differentially expressed in adhesion-inducing conditions are strong candidates to have a role in adhesion. Identifying genes that are strongly upregulated in adhesive strains but to a lesser extent in strains that do not have an inducible adhesive phenotype is one method to further narrow the list of candidate genes. To explore the potential utility of this approach, qPCR was performed to measure the relative expression of a subset of the strongly upregulated genes identified in the adhesive strains compared to a strain that demonstrates only weak inducible adhesion, CLIB214 (Fig. 7). Several genes (Cpar2_400510, Cpar2_701230, Cpar2_108340) showed considerably higher upregulation in all three adhesive strains than in CLIB214. Analyses such as these will be helpful in prioritizing additional mechanistic studies.
Comparison of transcriptional responses of C. parapsilosis and C. albicans. The C. albicans transcriptome was evaluated previously in response to a variety of conditions known to trigger filamentation compared to yeast growth in YPD alone at 30°C Blue circles represent genes that were differentially expressed by at least 64-fold. The most upregulated and downregulated families are listed on each side of the plots. The sequential order on the x axis, in which the gene families appear from left to right, is listed in parentheses next to each gene family label.
Transcription in C. parapsilosis Adhesion for 3 h (19). While the growth conditions used in that experiment were not identical to those in our study, their growth conditions were the most similar to ours among the data sets which were made available in public repositories. To allow a first approximation of similarities and differences in the response to these stimuli between species, the differentially expressed genes for C. albicans grown in liquid RPMI or YPD containing 10% fetal bovine serum at 37°C were compared to the differentially expressed genes identified in the current study for C. parapsilosis grown in M199 or human plasma. Genes that were coordinately up-or downregulated in both species as well as genes that were divergently regulated are shown in Table 3.

DISCUSSION
Previous studies have examined the C. parapsilosis transcriptome. These reports have defined responses to different growth media, temperatures, hypoxia (20,21), quorum sensing molecules (22,23), and changes associated with acquired resistance to azole drugs (24). Indeed, these studies have contributed substantially to detailed annotation of the C. parapsilosis genome, and the effects of mutations in key elements of virulence pathways have been globally defined. Several studies have focused on   Gene names are provided where available; otherwise, the Cpar2_ numeric designation is listed (45).
Transcription in C. parapsilosis Adhesion specific transcription factors and their targets (21,(25)(26)(27)(28)(29). An additional study characterized the response to a monocyte cell line in vitro to identify potential virulence mechanisms (30). While these reports have provided important insights into the gene regulation and pathobiology of C. parapsilosis, the majority have focused on one or two strains (CLIB214 or CDC317). However, a study comparing genomic sequence data from 3 additional isolates that varied in source and geographic region demonstrated substantial genomic variability, including variations in copy number, rearrangements, and single nucleotide polymorphisms (31). The phenotypic variability in adhesion observed in the current report in clinical isolates is striking and may reflect genomic or gene expression variability that has not been previously detected. Additionally, in many studies of the C. parapsilosis transcriptome mentioned above, yeasts were grown in standard laboratory medium and at 30°C. We have focused our studies on conditions that are likely to more closely mimic a host environment, including growth in mammalian tissue culture medium or human plasma and at physiologic temperatures of 37°C. These environmental cues are likely to reveal additional gene expression changes that are relevant to virulence.
In the current study, to model the early transcriptional response within the host, we used a strategy of examining adhesive clinical isolates, maintained at low passage number and grown under conditions modeling the host environment. We previously reported that growth in M199 or fetal bovine serum (FBS) at 37°C resulted in enhanced adhesion to immobilized substrates under fluid shear compared to growth in YPD (13). In the current study, human plasma was substituted for FBS to more closely mimic human conditions and include any potential influence of plasma components that would be lost during serum preparation (32). Although an effect similar to that of FBS is likely, we were unable to experimentally confirm the adhesion phenotype in vitro because the microfluidics adhesion assay is sensitive to shear forces that were significantly altered due to the increased viscosity of plasma. Comparison of the transcriptome under these conditions allowed identification of genes that were differentially expressed during the switch to an adhesive phenotype. In each of the growth conditions tested (YPD, M199, or plasma), common sets of genes were differentially expressed in all three strains, although some variations were also observed among them. Interestingly, for all three strains, transcriptional profiles in M199 and plasma resembled one another more closely than those in YPD. In combination with our earlier observation of increased adhesion (13), this result indicates that some of the differentially expressed genes in M199 and plasma are likely to represent adhesins or adhesion-related factors, although many others appear to encode transporters and genes involved with nutrient acquisition.
The ALS gene family is well known for its role in adhesion in C. albicans (33). Recent work using a combination of approaches to avoid pitfalls associated with accurate assembly of genome sequence data for these genes having multiple similar loci and tandem copies of repeated sequences has defined 5 ALS genes in several C. parapsilosis strains (15). Because these genes were substantively different from the C. albicans ALS genes, a naming scheme was proposed to capture their genomic arrangement (CpALS4770, CpALS4780, CpALS4790, CpALS4800, and CpALS660) (15). Careful assessments of C. parapsilosis ALS gene expression in this study demonstrated considerable variation among strains, which is much in line with our findings. Consistent with our previous report (13), CpALS4800 (CPAR2_404800 or CpALS7) was upregulated in JMB81 in M199 and plasma. However, in JMB77 and Ro75, CpALS4800 expression was consistently high under all conditions despite phenotypic differences in adhesion. This result suggests either that CpALS4800 does not contribute to adhesion under the conditions of our assay in these strains or that its adhesive function is being modulated by another factor. The other ALS family members did not show differential expression in any isolate and were generally expressed at low levels. CpALS4790 is notable in that it was the most abundant constitutively expressed ALS gene in JMB81 (13) and in the strains studied by Oh et al. (15), leading the authors to suggest an abundance of CpAls4790 on the cell surface. Further, a CpALS4790 null mutant showed a loss of adhesion to epithelial cells in vitro and a reduction in fungal burden in a mouse model of vaginal candidiasis (10). However, no CpALS4790 expression was detected by RNA-seq from JMB77 or Ro75, although adhesion is robust in these clinical isolates, and they were capable of colonization and/or invasive disease in the human infants from whom they were obtained. This finding is also consistent with our previous study, in which genomic sequences for CpALS4790 were not detected in JMB77 and Ro75 (13). Taken together, these findings lend further support to the heterogeneity of the ALS family among isolates of C. parapsilosis and represent additional evidence to support the noted genomic variability among strains of this species, including the CpALS genes (31). The observations also underscore the complexity of the adhesive phenotype and the difficulty in attributing it to a single gene product or family.
Several members of the CFEM (common in fungal extracellular membranes) gene family were strongly upregulated in adhesive conditions-CFEM2, CFEM3, and CFEM6 (CPAR2_402910, CPAR2_402900, and CPAR2_300120, respectively). These genes are syntenic with the C. albicans genes RBT5, PGA10, and CSA1 and are involved in heme binding and iron acquisition, a role that is conserved between the species. These genes have been shown to be upregulated in response to BCR1, a transcription factor that regulates biofilm formation (25). Iron acquisition from the host is increasingly being recognized for its importance in virulence among pathogenic fungi and is integrated with alterations in the cell surface related to morphogenesis and adhesion (34). Further, it has been shown that iron modulates chromatin activation of known adhesins and putative adhesin-like cell surface proteins in C. albicans (35). The finding that genes involved in iron acquisition were so strongly upregulated in our experimental conditions supports the notion that these conditions mimic the host environment and are useful for identification of virulence pathways.
The PHR family of transglycosylases is important in the virulence of C. albicans and is necessary for processing of b-1,3-glucans in the yeast cell wall (36). In the current study, CpPHR1 (CPAR2_302140) and CpPHR2 (CPAR2_109660) showed reciprocal transcriptional regulation. The C. albicans PHR genes exhibit inverse pH-dependent expression (36), and our results suggest that they behave similarly in C. parapsilosis. A PHR1 null mutant in C. parapsilosis forms biofilms with reduced numbers of cells, implying a role in this function. (37) Another glucan-processing enzyme that was upregulated under adhesive conditions was CpSKN1 (CPAR2_807200). SKN1 serves as a b-1,6-glucan synthase in C. albicans and functions in concert with the constitutive b-1,6-glucan synthase, KRE6. (38) Consistent with our findings, expression of SKN1 in C. albicans is induced upon induction of hyphal growth. The current study supports the possibility that polysaccharide remodeling of the cell wall, which is necessary for hyphal growth in C. albicans, may be important in adhesion of C. parapsilosis.
Other groups have examined the C. albicans transcriptional response, under similar but not identical conditions. Azadmanesh et al. compared the profile of SC5314 grown in YPD at 30°C for 3 h to growth in RPMI or in YPD containing 10% FBS at 37°C for 3 h (19). Despite these differences, our data show similar responses between C. albicans and C. parapsilosis for many of the top differentially expressed genes, including PHO89, CDG1, and PHR1 (Table 3). These results raise the possibility of a generic Candida response to host stimuli. However, several genes show a disparate response between the two species. CUP9 encodes a transcription factor that was upregulated in C. parapsilosis under adhesive conditions but is downregulated in C. albicans, where it represses filamentation and expression of hypha-specific genes such as HWP1 and ECE1. (39) Genes encoding two cell wall proteins, HWP1 and RBT5, are both strongly upregulated in C. albicans but show little change in C. parapsilosis, perhaps an effect of the upregulation of CpCUP9. Thus, these results support both general Candida spp. responses and C. parapsilosis-specific responses.
In summary, we describe a new approach to identifying C. parapsilosis adhesins, utilizing clinical isolates that show strong adhesion in vitro when exposed to host factors.
Using this approach, several potential adhesins and transcription factors were identified that may play a role in adhesion. It is likely that additional adhesion pathways or host stimulus responses may dominate in different isolates. It will also be useful to compare nonadhesive isolates to identify genes that are not relevant to adhesion.

MATERIALS AND METHODS
Strains and growth conditions. The C. parapsilosis clinical isolates used in the current study (JMB81, JMB77, and Ro75) have been previously described, and CLIB214 is a type strain (13). Isolates were confirmed as C. parapsilosis sensu stricto using PCR probes specific to the ITS1 region (40). Yeasts were grown overnight in YPD medium (1% yeast extract, 2% peptone, 2% dextrose) with vigorous agitation at 37°C. Stationary-phase cultures were washed with sterile water and resuspended in fresh YPD medium, tissue culture medium M199 (BioWhittaker/Lonza, no. BW12-117Q), or pooled human blood plasma (sodium heparin anticoagulated) at 3 Â 10 6 /ml and further incubated at 37°C for 3 h without shaking. Under these growth conditions, filaments (hyphae or pseudohyphae) were not observed. Parallel samples were tested in adhesion assays (Fig. 1) to confirm the adhesive phenotype, or yeast were centrifuged and pellets were flash frozen in liquid N 2 and stored at 280°C for later RNA extraction for RNA-seq or quantitative PCR (qPCR) validation.
Adhesion assays. A microfluidics multichannel adhesion assay was used to measure adhesion of yeast under fluid shear, using the Bioflux 200 platform (Fluxion Biosciences) as previously described (13). Briefly, channels were coated with BSA for 48 h and then washed with Hanks' balanced salt solution with calcium and magnesium (HBSS1) at 2 dynes/cm 2 to remove unbound protein. Yeasts grown in YPD or M199 were loaded into the channels and passed through at 5 dynes/cm 2 for 30 min. Excess yeasts were then removed from inlets and outlets, and Dulbecco's phosphate-buffered saline with calcium and magnesium (DPBS1) containing 5 mM calcofluor white was added to the channels. The flow was resumed for another 10 min at 5 dynes/cm 2 to remove unbound yeast. The channels were then imaged using a Nikon Ti-E inverted fluorescence microscope with a motorized stage to automate sequential imaging of the channel area. Yeast binding was measured in the DAPI (49,6-diamidino-2-phenylindole) (calcofluor) fluorescence channel (395 nM excitation, 440/40 emission). An intensity threshold was used to determine the percentage of the channel surface area that was covered with yeast (expressed as adhesion index in Fig. 1). Data from 4 independent experiments with duplicate channels for each condition were pooled. Statistical analysis was conducted using SigmaPlot v13.0.
RNA extraction, library preparation, and sequencing. The three strains used in this study were cultured in YPD, M199, or plasma as described above in three biological replicates. Total RNA was extracted from each and purified using a RiboPure yeast extraction kit (Ambion/Life Technologies). RNA integrity was assessed using agarose gels and an Agilent bioanalyzer. RNA-seq library preparation was conducted using the NEBNext Ultra RNA library prep kit for Illumina (New England Biolabs, Ipswich, MA, USA) according to the manufacturer's instructions, and sequencing was performed following rRNA depletion using paired-end reads (Illumina HiSeq2000), 2 Â 150 bp, from all samples.
RNA-seq pipeline. Samples were checked for quality control using FastQC v0.11.5 (http://www .bioinformatics.babraham.ac.uk/projects/fastqc/). Samples that contained remaining adaptor sequence were trimmed with Skewer v0.2.1. (41) Reads were then aligned to the genome (20) and quantified using Salmon v1.0.0 (42) in mapping-based mode. Salmon transcript quantification was performed by applying the -gc-bias and -seq-bias flags for correcting fragment-level GC content bias and sequence-specific random hexamer priming biases, respectively. The tximport v1.6.0 R package (43) was used to import transcript quantification files and bias correction matrices into the differential expression analysis workflow. Differential expression analysis using the DESeq2 v1.18.1 R package (44) was performed to identify differentially expressed genes in a series of pairwise comparisons between each of the sets of replicates of three samples in each of the three growth conditions. Default parameters for DESeq2 were used for differential expression analysis.
Data analysis and annotation. Gene ontology enrichment was performed using the Candida Genome Database's GO Term Finder tool. (45) Genes which were significantly upregulated or downregulated under each condition comparison were used as a query set for GO term enrichment. GO terms with a Bonferroni corrected P # 0.05 were considered to be significantly enriched. C. albicans and S. cerevisiae gene names which are included in the results as annotations were obtained from the Candida Gene Order Browser (46,47).
The gplots v3.0.1 and DESeq2 v1.18.1 R packages were used to create heat maps of the expression data. All R analyses were run using R v3.4.3. To create the heat maps of top differentially expressed genes, variance stabilizing transformation (VST) was first applied to counts of differentially expressed genes which were normalized by DESeq. Then, z scores per expression value were calculated using row means across all samples for each gene and z scores were plotted using the heatmap.2 function contained in the gplots package. The heat map data were clustered using default heatmap.2 settings, which use Euclidean measures to obtain a distance matrix and the complete agglomeration method for clustering, before scaling the data. Finally, the data dendrograms were reordered based on row and column means (https://stat.ethz.ch/R-manual/R-patched/library/stats/html/reorder.dendrogram.html).
Data for the Venn diagram (Fig. 4) were generated through a series of comparisons of the numbers of differentially expressed genes in each of the three DESeq2 results sets. Thresholds of gene expression fold change $ 64 and P # 0.001 were used to select genes to include in this comparison. The distribution of shared and unique genes among DESeq2 results sets was displayed as a Venn diagram using the VennDiagram v1.6.20 R package. Additionally, the correlation coefficients of all VST transformed normalized gene counts for each sample were plotted as a clustered heat map using the cor() function of the base R stats library to generate a correlation matrix model and the pheatmap v1.0.12 library to plot a clustered heat map of the correlation coefficient between all samples.
Gene family enrichment analysis of differentially expressed genes was performed by including genes with an expression fold change of $64 and P # 0.001 from each of our three condition comparisons in our gene family enrichment analysis workflow. First, same-species homologous protein sequences were determined using BLASTp for the three sets of differentially expressed genes from each DESeq2 comparison (18). Next, significantly enriched gene groups were identified using the Markov clustering algorithm with an inflation factor of 2.1 (16,17). Finally, the newly identified gene clusters were annotated with known gene family labels using the UniProt database of proteins (48). In some cases, gene family labels did not exist for C. parapsilosis genes, so gene family annotations for homologues in C. albicans or S. cerevisiae were used to identify our gene groups instead. The log 2 -fold change was plotted for genes within significantly enriched gene families using the ggplot2 v3.0.1 R package, and the gene families were ordered from highest to lowest mean log 2 -fold change for each gene group. The gene groups were spaced evenly apart by an arbitrary increment on the x axis for enhanced visibility.
qPCR. Quantitative PCR was performed as previously described (13). Primer sequences are listed in supplemental material (Table S4).
Data availability. The raw gene expression count files generated from our RNA-seq data are freely available and hosted in the Gene Expression Omnibus (GEO) RNA-seq data repository (accession number GSE159274).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. DATA SET S1, XLSX file, 9.8 MB.