Correlation between Biofilm Formation and the Hypoxic Response in Candida parapsilosis

Abstract

The ability of Candida parapsilosis to form biofilms on indwelling medical devices is correlated with virulence. To identify genes that are important for biofilm formation, we used arrays representing approximately 4,000 open reading frames (ORFs) to compare the transcriptional profile of biofilm cells growing in a microfermentor under continuous flow conditions with that of cells in planktonic culture. The expression of genes involved in fatty acid and ergosterol metabolism and in glycolysis, is upregulated in biofilms. The transcriptional profile of C. parapsilosis biofilm cells resembles that of Candida albicans cells grown under hypoxic conditions. We therefore subsequently used whole-genome arrays (representing 5,900 ORFs) to determine the hypoxic response of C. parapsilosis and showed that the levels of expression of genes involved in the ergosterol and glycolytic pathways, together with several cell wall genes, are increased. Our results indicate that there is substantial overlap between the hypoxic responses of C. parapsilosis and C. albicans and that this may be important for biofilm development. Knocking out an ortholog of the cell wall gene RBT1, whose expression is induced both in biofilms and under conditions of hypoxia in C. parapsilosis, reduces biofilm development.

Candida parapsilosis is closely related to the major human fungal pathogen Candida albicans: both are members of the Saccharomycotina CTG clade and translate CTG as serine (16). Whereas C. albicans is the most common cause of candidiasis, other species are isolated with increasing frequency. C. parapsilosis is now responsible for 8 to 10% of Candida bloodstream infections (45). The incidence of C. parapsilosis infection has been associated with the increased use of caspofungin (17), most likely because C. parapsilosis and closely related species are less susceptible to treatment with echinocandins due to a polymorphism in the target protein Fks1 (β-1,3-glucan synthase) (19).

Unlike other Candida species, C. parapsilosis is often isolated from the hospital environment and, in particular, from the hands of health care workers (12, 30, 34, 60). There have been several reports of infection outbreaks, from hospitals in the southern United States (12, 48) to neonatal intensive care units in Finland (51) and Taiwan (60). Children are particularly susceptible, and C. parapsilosis is responsible for approximately 20% of Candida infections in infants less than 1 year old (3, 13).

Infection with C. parapsilosis is strongly correlated with its ability to grow as biofilms on indwelling medical devices (29). Although there have been relatively few studies with C. parapsilosis, biofilm development by C. albicans has been well characterized (8). Cells first adhere to a surface (such as a catheter) and then build up a layer through cell-to-cell contact. At some point following adherence, the cells undergo a switch from yeast to hyphal growth, which is strongly correlated with the development of a structured biofilm (5, 47). The mature biofilm is surrounded with an extracellular matrix and is much less susceptible to antifungal drugs. Several groups have determined the transcriptional changes that occur during biofilm growth of C. albicans (20, 38, 64). In general, there is an increase in levels of expression of genes involved in glycolysis and amino acid and lipid metabolism.

There are considerable differences between the biofilms generated by C. parapsilosis and those generated by C. albicans. Most reports indicate that C. parapsilosis biofilms are smaller and less complex (25, 29), although this depends on the concentration of glucose available (53) and the colony morphology (32). C. parapsilosis does not form hyphae, and its biofilms are composed of yeast and pseudohyphal cells only (15, 18, 29). Biofilm production is inhibited in the presence of exogenous farnesol (32, 50), as in C. albicans (46), but unlike C. albicans, C. parapsilosis does not secrete farnesol (62). However, there are also many similarities between biofilms in the two species, in particular the fact that the transcription factor Bcr1 is a major regulator of biofilm formation in both of them (15, 40, 41). In C. albicans, Bcr1 regulates the expression of several cell wall and adhesion genes, and at least some targets are conserved in C. parapsilosis (15).

We describe here the use of partial genomic microarrays to determine the transcriptional changes that occur during biofilm development in C. parapsilosis. We observed an upregulation in the expression of genes involved in glycolysis, fatty acid metabolism, and ergosterol synthesis. Some of these changes are similar to those observed in C. albicans cells grown under low-oxygen conditions (52). We therefore also profiled C. parapsilosis cells growing in a hypoxic environment, and we characterized the effect of deleting RBT1, a gene induced both in biofilms and under conditions of hypoxia.

MATERIALS AND METHODS

Strains and media.

All experiments were carried out using C. parapsilosis CLIB214 and derivatives (Table 1). For the biofilm microarray studies, cells were grown 37°C in SD medium (0.67% yeast nitrogen base) with 2% dextrose supplemented with methionine (20 mg/liter), arginine (50 mg/liter), and histidine (20 mg/liter) where required. For pH experiments, SD medium was buffered with 50 mM succinate, and the pH was adjusted to 5.4 with HCl. For hypoxic profiling, cells were grown in SD medium at 37°C. To enable comparisons with C. albicans, the hypoxic profile of C. parapsilosis cells was also determined after growth in YPD medium (1% yeast extract, 2% peptone, and 2% dextrose) at 30°C. Two percent agar was used for gene disruption selection. Transformants were selected as described previously (15) except that YPD medium was used to recover cells after electroporation. Hypoxic experiments were carried out in a dedicated chamber (In Vivo₂ 400 workstation) with 1% O₂.

TABLE 1.

Yeast strains used in this study

Strain	Genotype	Description, source, or reference
C. parapsilosis
CLIB214	Wild type	Type strain
CDR1	RBT1/rbt1Δ::SAT1-FLP	This study, derived from CLIB214
CDR14	RBT1/rbt1Δ::FRT	This study, derived from CDR1
CDR210	rbt1Δ::SAT1-FLP/rbt1Δ:: FRT	This study, derived from CDR14
CDR212	rbt1Δ::FRT/rbt1Δ::FRT	This study, derived from CDR210
CDR311	rbt1Δ/rbt1Δ::SAT1-FLP- RBT1	This study, derived from CDR212
CDRbt8	rbt1Δ/rbt1Δ::RBT1	This study, derived from CDR311
74/046	Wild type	56
81/041	Wild type	56
C. albicans SC5314	Wild type	22

Biofilms on 96-well plates and silicone squares.

Ninety-six-well polystyrene microtiter plates (catalog number 167008 from Nunc or from Techno Plastic Products AG) were inoculated with 100 μl per well of a C. parapsilosis culture grown overnight diluted to an A₆₀₀ of 0.05 in either SD medium supplemented with 50 mM dextrose or RPMI 1640 with Glutamax medium (Invitrogen) buffered with 50 mM HEPES and incubated at 37°C for 1 h to allow adherence. C. albicans cells were grown in SC medium supplemented with 2% dextrose. The supernatant (planktonic and nonadherent cells) was then removed, and adherent cells were washed twice with 200 μl of 1× phosphate-buffered saline (PBS). The wells were filled with 200 μl of fresh medium and incubated at 37°C. After 24 h, the plate was washed with 1× PBS using a Hydroflex plate washer (Tecan). Biofilm quantification was performed using the fluorescein diacetate (FDA) method (26) and was verified with a crystal violet assay as described previously by Laffey and Butler (32). For the first method, 100 μl of FDA (40 μg liter⁻¹ in PBS) was added per well, and the plate was incubated for 1 h at 37°C. Fluorescence was measured with an Infinite M200 plate reader (Tecan) using an excitation filter of 486 nm and an emission filter of 535 nm.

Transcriptional profiling.

The partial arrays used to determine the transcription profile of biofilm cells contain one probe from each of 3,849 open reading frames (ORFs), spotted in duplicate (50). For the hypoxia experiments, arrays representing 5,834 ORFs were designed and manufactured using tools available from Agilent eArray. Each gene was represented by two probes, both spotted in duplicate. Four copies of each array were printed on a single slide and hybridized individually. Where possible, the assigned gene names (CPAG [C. parapsilosis gene] designations) are based on an annotation of the C. parapsilosis genome by the Broad Institute (http://www.broad.mit.edu/annotation/genome/candida_group/Info.html).

Biofilm growth in continuous flow was carried out in a microfermentor system as described previously by Garcia-Sanchez et al. (20). Briefly, this consists of a glass vessel with a 40-ml incubation chamber with tubes allowing entry and evacuation and continuous medium flow of 0.6 ml/min (SD medium supplemented with methionine and arginine). Air under pressure was supplied to allow medium to flow through the chamber. The continuous flow allows cells to grow as a biofilm on a slide positioned in the chamber and minimizes planktonic growth. Thermanox slides (Nunc) were first immersed into an inoculum of the appropriate culture at an A₆₀₀ of 1 for 1 h at 37°C to allow adherence and were then transferred into the microfermentor system. Experiments were performed at 37°C.

For planktonic growth, flasks containing 50 ml of the same medium used for the biofilm were inoculated at an A₆₀₀ of 0.2 from a culture grown overnight. Cells were grown at 37°C with orbital shaking to an A₆₀₀ of 1. For quantitative reverse transcription (RT)-PCR (qRT-PCR), additional sampling was performed after 24 h, when cells had reached almost maximum absorbance and were therefore judged to be in stationary phase.

For the hypoxic study, cultures of C. parapsilosis CLIB214 cells grown overnight were diluted to an A₆₀₀ of 0.2 in 300 ml of SD medium supplemented with 2% dextrose at 37°C under normoxic (atmospheric-oxygen) conditions. Two 100-ml aliquots were removed after 3 h, and the cells were collected by centrifugation and resuspended in preconditioned medium (incubated under hypoxic conditions for 12 h before use). One flask was incubated under normoxic conditions (21% O₂) in a standard orbital shaker, and the other was incubated under hypoxic conditions (1% O₂ and 99% N₂) in a dedicated chamber (In Vivo₂ 400 workstation). Both cultures were incubated at 37°C and at 200 rpm for 2 h.

RNA was isolated using an RNeasy kit (Qiagen) for the biofilms and a Ribopure kit (Ambion) for the hypoxia cultures according to the manufacturers' instructions. The quality and concentration of the isolated RNA were analyzed using an Agilent 2100 Bioanalyzer. For the biofilm experiments, RT, cDNA labeling, and probe purification were carried out using an Atlas Powerscript fluorescent labeling kit (Takara) according to the manufacturer's instructions, starting with 5 μg of total RNA. Partial genomic microarrays representing 3,849 putative ORFs described previously by Rossignol et al. (50) were used. Hybridization, washing, and scanning were carried out as described previously by Rossignol et al. (50). Four independent biological replicates were compared; in two replicates, the biofilm sample was labeled with Cy5, and in the other two replicates, the biofilm sample was labeled with Cy3.

For the hypoxia experiments, 24 μg of total RNA was labeled as described previously (50). The experimental and control samples were mixed and applied to genomic microarrays representing 5,834 ORFs (manufactured by Agilent Technologies, design 015742). The hybridization, wash, and scanning protocols were the same as those used for the biofilm arrays. Six independent biological replicates were compared; in four replicates, the hypoxic samples were labeled with Cy5, and in two replicates, the samples were labeled with Cy3.

Data analysis.

The data from both the biofilms and the hypoxic conditions were statistically analyzed using the LIMMA package from the Bioconductor project (54). The data sets were preprocessed using Loess normalization and no background correction (as suggested in reference 65). Probes with a change lower than 1.5-fold were discarded. For the biofilm arrays, probes with adjusted P values of greater than 0.05 were discarded, resulting in 185 genes that are differentially expressed. In the hypoxia experiments, probes with adjusted P values higher than 0.005 were excluded. For this data set, only genes with three or four significant probes were included. The final list contained 341 differentially expressed genes.

GO analysis.

The Bioconductor package topGo (2) was used to identify enrichment of Gene Ontology (GO) terms in both data sets. C. parapsilosis orthologs in C. albicans were first identified using gene family and reciprocal BLAST analysis (16). For the biofilm experiment, 3,573 (94.3%) of the spotted probes and 185 (94.6%) of the differentially expressed genes had identifiable orthologs in C. albicans. For the hypoxia experiment, 5,091 (87.3%) orthologs of the entire C. parapsilosis gene set and 300 (88%) orthologs of the differentially expressed genes were identified. The most recent GO annotation (version 1.493, 13 May 2008) for C. albicans was downloaded from the Candida Genome Database (http://www.candidagenome.org/). GO terms associated with processes were assigned to 2,272 genes on the biofilm arrays and 3,017 genes on the hypoxia arrays. For the differentially expressed genes, GO terms were assigned to 121 genes in the biofilm experiment and 174 genes in the hypoxia experiment. Enrichment of categories was determined using two statistical approaches, classic (Fisher's exact test compares the number of observed occurrences with the number of expected occurrences) and elim (uses the GO tree structure to identify significant terms) (2). GO terms with the 50 lowest adjusted P values for the classic approach are shown in Tables S4 and S5 in the supplemental material, and categories with adjusted P values lower than 0.05 for both classic and elim are shown in Tables 2 and 3. Approximately 70% of the genes are shared between the two array platforms. We used statistical analysis to show that none of the significantly changing categories are overrepresented on the smaller (biofilm) arrays.

TABLE 2.

Biological processes enriched in biofilms^a

GO term	Significance value
Alcohol metabolic process	6.60E−10
Monocarboxylic acid metabolic process	4.20E−08
Protein folding	6.30E−07
Monosaccharide catabolic process	2.40E−06
Fatty acid metabolic process	7.50E−06
Acetyl coenzyme A biosynthetic process	5.70E−04
Ergosterol metabolic process	8.70E−04
Glycolysis	1.25E−03

^aRNA was extracted from C. parapsilosis biofilms after 50 h, and the transcriptional profile was compared to that of planktonic culture for independent replicates. GO enrichment analysis was carried out using topGO (2) to determine GO classes that are overrepresented in the differentially expressed data. The GO terms listed are sorted by adjusted P values using the classic (Fisher's exact test) approach and are also significantly different using the adjusted P values from the elim (GO term hierarchy) approach. The full gene list is available in Table S2 in the supplemental material, and the GO enrichment analysis is available in Table S4 in the supplemental material.

TABLE 3.

Differentially expressed genes identified by comparing the levels of gene expression of cultures of six independent replicates of C. parapsilosis CLIB214 cells exposed to normoxic (21% O₂) or hypoxic (1% O₂) environments for 2 h^a

GO term	Significance value
Alcohol metabolic process	6.60E−23
Ergosterol biosynthetic process	6.40E−13
Glycolysis	4.80E−07
Galactose metabolic process	8.60E−07
Fatty acid metabolic process	1.30E−04
Heme biosynthetic process	3.40E−04
Pyruvate metabolic process	8.90E−04
O-linked mannosylation	3.31E−03

^aGO enrichment analysis was carried out as described in the legend of Fig. 1 and Table 1. The full gene list is available in Table S3 in the supplemental material, and the GO enrichment analysis is available in Table S5 in the supplemental material.

qRT-PCR.

For the biofilm experiments, a supplementary DNase treatment (Turbo kit; Ambion) was carried out with purified RNA to ensure the absence of DNA, which was verified by RT-PCR. cDNAs were synthesized using 2 μg of RNA mixed with 0.5 μg of oligo(dT) (Invitrogen) by incubation at 85°C for 5 min, 65°C for 20 min, and 42°C for 10 min. The final mix was added to a volume of 20 μl containing RT buffer and 150 U Superscript II (Invitrogen), 15 U of RNAseout (Invitrogen), and 3 mM of deoxynucleoside triphosphate. RT was performed at 42°C for 60 min followed by 10 min at 70°C. Quantitative PCR were performed with the Power SYBR green master mix kit (Applied Biosystem) on an Abi Prism 7000 apparatus (Applied Biosystem) in a 20-μl reaction mixture containing 5 μl of cDNA. The results shown are an average of five or six replicates from two independent biological samples. For the hypoxic experiments, real-time PCR was carried out as described previously (37). For all qRT-PCR experiments, primers were designed using Primer 3.0 Plus (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi), and TUB4 was used as an endogenous control. For the ALS genes, each primer was tested for specificity by comparison to the C. parapsilosis genome. Each primer pair was also used to amplify C. parapsilosis genomic DNA, and the PCR products were sequenced. This confirmed that the primer combinations are specific for the individual family members. The oligonucleotide primers used are shown in Table S1 in the supplemental material. ΔC_T values were calculated by subtracting the threshold cycle (C_T) value for TUB4 from the C_T value of the gene of interest. Relative expression was calculated from the ratio of the 2^ΔCT value for the test condition relative to the 2^ΔCT value for the control condition. Standard deviations were calculated from the replicates (a minimum of three) and divided by the value for 2^ΔCT for the control condition.

Generation of gene knockouts.

Both alleles of RBT1 in C. parapsilosis were deleted using the SAT1 flipper cassette as previously described (15). Oligonucleotides Rbt1Kpn and Rbt1Apa, which introduce the restriction sites KpnI and ApaI, respectively, were used to amplify a 550-bp fragment upstream from RBT1 (including part of the ORF), and oligonucleotides Rbt1SII and Rbt1SI, which introduce the restriction sites SacII and SacI, were used to amplify a 587-bp fragment downstream from RBT1 (including part of the ORF). The two PCR products were cloned into the relevant sites of pCD8 (15) to generate pCD40. A purified KpnI/SacI fragment from pCD40 was used to transform C. parapsilosis CLIB214 by electroporation, and nourseothricin-resistant colonies were selected. The SAT1 flipper cassette was recycled from the first disrupted allele by inducing expression of the recombinase, and the process was repeated to disrupt the second allele. The deletion construct was confirmed by PCR and Southern blotting. PCR amplification was performed using oligonucleotide Rbtup, which binds upstream of the flanking region of RBT1, and But237, which binds within the SAT1 flipper cassette. Correct integration at RBT1 yields a 1,536-bp PCR product (CDR1, CDR210, and CDR311), which is absent from the wild-type strain (CLIB214) and recycled strains (CDR14, CDR212, and CDRbt8). The presence of at least one intact RBT1 gene was tested using oligonucleotides Rbtup and Rbtmid, which generate a 1.0-kb product from the wild-type allele only (CLIB214, CDR1, CDR14, CDR311, and CDRbt8). Oligonucleotides RbtKpn and RbtSI were used to confirm recycling of the SAT1 flipper cassette from both alleles. The wild-type allele generates a 3.0-kb PCR product (CLIB214 and CDR1), and recycled strains generate a 1.0-kb PCR product (CDR14, CDR210, and CDR212).

An intact RBT1 allele was reconstituted by amplifying the entire ORF (including 190 bp of promoter sequence and 40 bp of termination sequence) using oligonucleotides Rbt1Apa5 and Rbt1int_Apa, which both introduce ApaI restriction sites. The PCR fragment was cloned into plasmid pCD39, which contains the SAT1 cassette and a 587-bp fragment from the region downstream of RBT1. An 8-kb PvuI/SacI fragment encompassing RBT1 and the SAT1 nourseothricin marker was introduced into C. parapsilosis CDR212. Nourseothricin-resistant colonies were selected, and integration at the rbt1 deletion was confirmed using oligonucleotides Rbt1up/Rbtmid and Rbt1SII/But237, which generate fragments of 1.0 kb and 850 bp, respectively. The SAT1 marker was then removed by inducing recombination.

The constructs were also confirmed by Southern hybridization using a probe amplified from the cassette using oligonucleotides Rbt1sII and Rbt1sI. Labeling and hybridization were carried out using DIG-High Prime DNA Labeling and Detection Starter Kit II (Roche) according to the manufacturer's instructions.

Genomic DNA was isolated from a 5-ml culture grown overnight and was digested using NcoI and XmaI overnight at 37°C. A 4.8-kb NcoI fragment (representing the wild-type RBT1 allele) was detected in CLIB214, CDR1, and CDR14. A 3.8-kb NcoI/XmaI fragment was detected after integration at one RBT1 allele in strains CDR1 and CDR210. Recycling of the integration cassette produces a 2.7-kb NcoI fragment, which is present in strains CDR14, CDR210, and CDR212. Reintroducing the RBT1 allele (CDRbt8) generates a 5.0-kb fragment (4.8 kb from the wild-type allele plus a 200-bp duplication generated during construction of the vector). We also observed a nonspecific fragment around 3.0 kb, which is present in all strains.

Fluorescence microscopy.

Fluorescence microscopy was performed using biofilms grown for 50 h in the same medium and conditions as those used for the microarray experiments. The biofilms were stained in a 5-ml solution of 50 μg ml⁻¹ Alexa Fluor 594 conjugate of concanavalin A (Invitrogen) for 1 h in the dark and observed without washing. Images were obtained with an Upright Wide Field Microscope Axioplan (Zeiss) and Axiovision 4.5 (Zeiss) software, with a 40× immersion objective. An HBO mercury short arc lamp was used with a XF 43 filter (excitation wavelength of 563 to 587 nm and emission wavelength of 615 to 645 nm; Omega Optical). z-stack images were acquired at 1-μm step intervals. Image deconvolution was performed using Huygens software (SVI), and three-dimensional reconstruction was done with Imaris software (Bitplane).

Microarray data accession numbers.

The transcriptional data and a description of the arrays used here have been deposited in the Gene Expression Omnibus database under accession numbers GPL7693, GSE13717, GSE13722, and GSE13832.

RESULTS

Gene expression in biofilms.

There have been numerous studies of biofilms formed by C. albicans on various surfaces including plastic and glass (reviewed in references 8 and 39). For this study, we used the microfermentor model described previously by Garcia-Sanchez et al. (20). This allows the development of large quantities of biofilm on plastic Thermanox slides under continuous flow with unlimited nutrients and under aerobic conditions. To facilitate comparisons with C. albicans biofilms, which were developed over 48 h (20), RNA was extracted from C. parapsilosis biofilm cells after 50 h of growth, and gene expression was compared to that of planktonic cultures in exponential-phase growth.

Transcriptional profiling was carried out using microarrays representing 3,849 ORFs from C. parapsilosis, which were based on a genome sequence survey (35). One hundred eighty-five genes showed reproducible changes in gene expression: 122 genes with increased levels of expression in biofilms and 63 genes with reduced levels of expression (see Table S2 in the supplemental material). Where possible, the C. albicans orthologs were identified using reciprocal BLAST analysis, and these were assigned to GO processes. Enrichment of specific GO categories in the differentially expressed data was determined using topGo within the Bioconductor package (21). Two measurements of statistical reproducibility were used, the classic Fisher's exact test, which compares the number of observed incidences with the number expected, and elim, which compensates for overlapping categories by removing genes mapped to significant GO terms from higher-level GO terms (2). GO categories that are significantly enriched include metabolism of ergosterol, fatty acids, and glucose (Table 2). The levels of expression of ergosterol pathway and glycolytic genes in particular are increased, whereas levels of expression of the gluconeogenic enzymes FBP1 and PCK1 are decreased (see Table S2 in the supplemental material).

Changes in levels of expression of several genes were confirmed by quantitative RT-PCR (Fig. 1A and B). Expression in biofilms was compared to that in planktonic cells in exponential and in stationary growth phases to ensure that the observed differences are not an artifact of the culture conditions of the control. The levels of expression of the ergosterol pathway genes ERG1 and ERG11 and the glycolytic genes PGK1 and PFK2 are increased in biofilms compared to those in planktonic cells in both exponential and stationary phases (Fig. 1B). Similarly, we also confirmed that the level of expression of the cell wall gene RBT1 is increased in biofilms.

FIG. 1.

Transcriptional profile of C. parapsilosis biofilm cells. (A) The expression of several genes was confirmed by qRT-PCR by comparing expression levels in planktonic cells in exponential-phase growth (Exp) and in stationary-phase growth (Stat) with 50-h biofilm samples. At least three biological replicates were used. (B) The effect of pH was determined by comparing the level of gene expression in unbuffered planktonic cultures (pH 3.8) with that in buffered cultures (pH 5.4), which more closely resembles the pH of biofilm cells. The scale is broken to allow the inclusion of the high level of expression of PHR1.

In addition, changes in levels of expression of a number of genes associated with pH regulation (PUT1, PUT2, PHR1, CCP1, and RIM101) led us to investigate the role of pH in the planktonic cultures. In unbuffered medium, the pH of the planktonic culture drops to 3.8, whereas it remains stable at pH 5.4 in the continuous-flow fermentor system. We therefore determined if the difference in pHs under planktonic and biofilm conditions could have contributed to the variation of levels of expression of some of our target genes. Figure 1B shows that buffering of the medium has no effect on the expression of the ergosterol pathway genes or on the expression of PGK1. The level of expression of PFK2, however, is increased at pH 5.4 relative to that at pH 3.8. The level of expression of PHR1 is also strongly increased at pH 5.4, whereas the level of expression of PHR2 is unchanged, as expected. The expression of RBT1 is known to be regulated by pH, at least in C. albicans (6), but levels of expression in C. parapsilosis are essentially the same at pH 5.4 and pH 3.8 under the conditions used (P value of 0.067 by t test).

Gene expression in hypoxia.

The increase in levels of expression of genes involved in ergosterol metabolism and in glycolysis is reminiscent of gene expression changes reported previously for C. albicans cultures during growth under conditions of low oxygen (52). We therefore determined the gene expression changes that occur when C. parapsilosis is grown under hypoxic (low-oxygen) conditions. For these experiments, we utilized the emerging whole-genome sequence of C. parapsilosis (http://www.sanger.ac.uk/sequencing/Candida/parapsilosis/). Microarrays were manufactured by Agilent, representing 5,834 genes identified in an in-house annotation of the genome sequence.

The response of C. parapsilosis to hypoxic conditions was first investigated by comparing the growth on YPD medium in 21% oxygen with that on YPD medium in 1% oxygen (Fig. 2A). Low oxygen has a dramatic effect on growth and colony size in three different C. parapsilosis isolates, whereas the growth of C. albicans under the same conditions is only slightly reduced. To measure changes in gene expression that are related to hypoxia rather than reduced growth, we restricted the incubation time of C. parapsilosis cultures under low-oxygen conditions. Cells were grown to exponential phase in atmospheric oxygen in SD medium at 37°C. The cultures were then split and incubated for 2 h in either atmospheric oxygen or a hypoxia chamber at 1% oxygen, and RNA was isolated. Analysis of enrichment of specific GO categories was done as described above for the biofilm experiments. We once again observed an overrepresentation of genes involved in ergosterol metabolism and glycolysis (Table 3; see Table S3 in the supplemental material). In addition, the expression levels of genes required for heme synthesis are increased. Changes in levels of expression of selected genes were confirmed using qRT-PCR (Fig. 2B). The level of expression of ergosterol pathway genes (ERG1 and ERG11) was increased four- to fivefold. The levels of expression of glycolytic genes (PFK2 and PGK1) were similarly affected. The level of expression of the cell wall gene RBT1 was also increased, by approximately 30-fold. One of the genes with the highest induction in expression in the array experiments (CPAG_05061) has no ortholog in any other Candida species and, indeed, no obvious similarity to any sequenced gene. qRT-PCR confirmed that expression is induced (by approximately 40-fold) under hypoxic conditions, but we do not know the function of this gene.

FIG. 2.

Transcriptional response of C. parapsilosis to hypoxic conditions. (A) C. parapsilosis (CLIB214, 74/046, and 81/041) and C. albicans (SC5314) isolates were grown overnight at 30°C in liquid YPD medium. Approximately 50 cells were plated onto YPD agar plates and incubated for 5 days under either normoxic (21% O₂ in a standard laboratory incubator) (top) or hypoxic (1% O in a hypoxia chamber) (bottom) conditions at 30°C. All plates were photographed on the same day and at the same magnification. The same effect was observed during growth at 37°C. (B) The expression levels of several potential genes under hypoxic conditions were confirmed using qRT-PCR. Ergosterol genes (ERG1 and ERG11) and glycolytic genes (PFK2 and PGK1) were induced four- to sixfold and are shown in one panel. The data for expression of a cell wall gene (RBT1) and a gene unique to C. parapsilosis (CPAG_05061), which are induced 30- to 45-fold, are shown separately. Three biological replicates were used, and each sample was analyzed in duplicate. (C) Distribution of genes with differential expression in biofilms and under hypoxic conditions. Sixty genes are differentially regulated under both conditions; the full list is available in Table S6 in the supplemental material.

In total, 60 genes with differential expression were shared between the biofilm and hypoxic experiments, 57 with the same patterns (i.e., expression levels increased or decreased both in biofilms and under conditions of hypoxia) (Fig. 2C; see Table S6 in the supplemental material). The vast majority of upregulated genes in both experiments are involved in fatty acid or ergosterol synthesis or in glycolysis (Fig. 3; see Tables S2 and S3 in the supplemental material).

FIG. 3.

Gene expression changes in biofilms and hypoxia. C. albicans orthologs of C. parapsilosis genes with altered expression in biofilms or under hypoxic conditions were identified as described in Materials and Methods. GO terms with significant enrichment were determined using GeneSet enrichment analysis. The figure shows selected GO processes and the associated genes that are enriched in the biofilm and hypoxia arrays. C. albicans gene names are used, and genes highlighted in gray are common to both experiments. Several genes are associated with more than one process. CoA, coenzyme A. (The structure of the figure is based on a similar diagram in reference 1.)

RBT1 is required for biofilm development.

Setiadi et al. (52) previously observed that hypoxia induces the expression of several hypha-specific genes in C. albicans, including HWP1. C. parapsilosis does not generate true hyphae, but the genome does include several members of the Hwp1 family. The level of expression of one of these members, an ortholog of C. albicans RBT1, is increased both in biofilms (Fig. 1) and under conditions of hypoxia (Fig. 2B). We therefore tested the effect of knocking out this gene on biofilm development. Both RBT1 alleles were deleted using an SAT1 flipper cassette (15), and in one construct, the RBT1 gene was reintroduced at the same location (Fig. 4). The wild-type and rbt1 knockout isolates were incubated on Thermanox slides in the fermentor system, and the structure of the resulting biofilms was determined using fluorescent imaging (Fig. 4C). Whereas the wild type generates biofilms of approximately a 300-μm depth, this is reduced to less than 40 μm in the rbt1 knockout. Biofilm mass (measured using an FDA assay) was also reduced when cells are grown in 96-well plates (Fig. 4D). However, there was no obvious difference in biofilms generated on silicone squares (not shown). The heterozygous strains CDR14 and CDRbt8 that have only one allele of RBT1 have slightly reduced biofilms (Fig. 4D), although the growth rate is the same for all the strains, including the wild type and the double-knockout mutant (not shown). Two RBT1 alleles are therefore required for full biofilm development, at least on some surfaces. Deleting rbt1 has no observed effect on the hypoxic growth of C. parapsilosis (not shown).

FIG. 4.

Construction of an rbt1 deletion of C. parapsilosis. (A) The RBT1 gene was disrupted as described previously (15). Briefly, upstream and downstream sequences from the RBT1 gene were amplified using primer pairs Rbt1kpn/Rbt1apa and Rbt1sII/Rbt1sI and inserted at either side of the SAT1 flipper cassette in plasmid pCD8, which contains a nourseothricin resistance gene and an FLP (flippase recombination enzyme), surrounded by two FRT (flippase recognition target) sites. The entire fragment was used to replace one RBT1 allele in C. parapsilosis CLIB214 by homologous recombination. The SAT1 cassette was removed by inducing the expression of the recombinase, used again to delete the second RBT1 allele, and the cassette was finally recycled. A reconstituted strain was generated by cloning a wild-type copy of RBT1 upstream from the SAT1 flipper cassette. The entire fragment was used to replace one rbt1 allele by homologous recombination, and the cassette was recycled by inducing recombination between the FRT sites. (B) All steps in the construction were confirmed using PCR (top and middle) and by Southern hybridization (bottom). The order of the strains for each reaction shown at the top is as follows: lane 1, CLIB214; lane 2, CDR1 (integration at the first allele); lane 3, CDR14 (first recycle); lane 4, CDR210 (integration at the second allele); lane five, CDR212 (second recycle). In the middle, lanes 1 and 2 are CDR311 (reconstituted strain), and lanes 3 and 4 are CDRbt8 (reconstituted strain following recycling of the cassette). M, marker. (C) Fluorescence microscopy of biofilms on Thermanox slides. Biofilms were grown for 50 h and stained with a fluorescent conjugate of concanavalin A. z-stack images were acquired at intervals of 1 μm, and a three-dimensional reconstruction was generated using Imais software. (CLIB214, wild type; CDR212, rbt1Δ/rbt1Δ.) (D) Biofilm quantification in 96-well plates. Cells were first inoculated at and A₆₀₀ of 0.05 for 1 h, wells were washed two times with PBS, and fresh RPMI 1640 with Glutamax medium (Invitrogen) buffered with 50 mM HEPES was added. The plates were then incubated for 24 h at 37°C. Cells were washed with PBS, and biofilms were quantified using the FDA assay. (CLIB214, wild type; CDR14, RBT1/rbt1Δ; CDR212, rbt1Δ/rbt1Δ; CDRbt8, RBT1/rbt1Δ::RBT1.)

ALS genes.

Increased levels of expression of the ALS adhesin family have been associated with biofilm formation in C. albicans (11, 20, 23, 44, 64). We identified little change in levels of expression of the family in C. parapsilosis biofilms in the array experiments. However, not all family members were represented on the arrays, and there is likely to be some cross-hybridization due to sequence similarities. We identified five members of the ALS family in C. parapsilosis, and we measured levels of expression in biofilms at 24 h and 50 h using qRT-PCR (Fig. 5). Levels of expression of CPAG_05314, CPAG_00368, and CPAG_00369 were essentially unchanged. The level of expression of CPAG_05054 was induced approximately threefold at 24 h but not at 50 h, and the level of expression of CPAG_05056 was slightly increased in 50-h biofilms. The changes in levels of expression are much lower than those observed for RBT1 (which is not a member of this family) in 50-h biofilms. The expression of the ALS family is also not greatly affected by hypoxia, the level of expression of CPAG_05056 is reduced, and that of CPAG_05314 is slightly increased (Fig. 5; see Table S3 in the supplemental material). It is, however, possible that levels of expression of gene families (such as the ALS genes) may vary in different isolates of C. parapsilosis, which we have not yet tested.

FIG. 5.

Expression of the ALS gene family in C. parapsilosis. (A) RNA samples were extracted from planktonic cultures and from 24-h and 50-h biofilm samples. The levels of expression of five ALS family members and RBT1 were measured using qRT-PCR. At least three biological replicates were used. (B) RNA samples grown under normoxic conditions (21% O₂) and under hypoxic conditions (1% O₂) were extracted. The expression levels of five ALS family members were measured using qRT-PCR. At least three biological replicates were used.

DISCUSSION

Transcriptional profiling has made major contributions to our understanding of biofilm formation, particularly by bacteria (4). In C. albicans, array analysis led to the identification of the role of the transcription factor Gcn4 and the role of sulfur metabolism in biofilm development (20, 38, 64). Our analyses of gene expression changes reveal substantial overlaps during biofilm growth in C. parapsilosis and that in C. albicans. At least four glycolytic genes have increased levels of expression in biofilm versus that in planktonic cells in C. parapsilosis. This was confirmed for PFK2 and PGK1 using qRT-PCR, where we showed that the level of expression is higher in biofilms than in planktonic cells in exponential- or in stationary-phase growth. A large increase in levels of expression of glycolytic genes in C. albicans, particularly during early biofilm development, was also reported (64). Levels of expression of glycolytic genes also increase during planktonic growth, but the difference is greater in biofilms (64). In contrast, levels of expression of glycolytic and ergosterol metabolism genes are decreased in stationary-phase cells in C. albicans (59), suggesting that the changes observed in biofilm cells do not result from reaching stationary-phase growth. Changes in carbohydrate metabolism are therefore important for biofilm development in both species.

Surprisingly, we did not observe any change in sulfur amino acid metabolism or, indeed, in general amino acid metabolism in C. parapsilosis biofilms. This is a feature of most of the C. albicans-profiling experiments reported to date (20, 38, 64). Our experimental conditions are generally comparable to those reported previously by Garcia-Sanchez et al. (20), who identified a pivotal role for the amino acid regulator Gcn4 in C. albicans biofilms. Our biofilm experiments measured levels of expression of only a portion (3,789 ORFs) of the C. parapsilosis gene repertoire, but this does include most of the amino acid biosynthetic genes. In addition, we did not observe major changes in genes required for protein synthesis reported in C. albicans (20, 64). This result suggests that the growth stage (or the protein synthesis needs) under the two conditions tested here (planktonic exponential phase and biofilm) are similar, in contrast to that observed for C. albicans. The difference between the expression profiles in the two species may be linked to the growth rate; we observed that C. parapsilosis grows approximately twofold slower than does C. albicans. The biofilms generated are therefore unlikely to be mature, even after 50 h, and the biomass increases with longer incubation times (not shown). It is therefore likely that there are some significant differences between the metabolic profiles of C. parapsilosis and C. albicans biofilms.

We also observed an increase in levels of expression in ergosterol genes in the C. parapsilosis biofilms. Altered expression of the ergosterol pathway in C. albicans biofilms has been associated with increased antifungal resistance (36), and the level of expression of ERG10 is increased in the early stages (64). However, we observed changes in levels of expression of several genes, including ERG1, ERG11, ERG25, and ERG5, that act in the oxygen-dependent postsqualene part of the ergosterol biosynthesis pathway (61). When we compared the upregulated genes to data from several C. albicans experiments (using List-to-List at http://candida.bri.nrc.ca/l2l/), the closest match was to the data set with increased levels of expression under conditions of hypoxia (from reference 52). This prompted us to compare the transcriptional responses of C. parapsilosis biofilms to gene expression changes that occur under low-oxygen conditions.

C. parapsilosis appears to be more susceptible to low-oxygen conditions than C. albicans, as colonies grown on YPD plates in 1% oxygen are much smaller than C. albicans colonies grown under the same conditions (Fig. 2A). The transcriptional response of C. parapsilosis following short-term exposure to hypoxia is, however, very similar to that observed for C. albicans. We determined the hypoxic profile of C. parapsilosis cells grown in SD medium at 37°C to mimic the conditions used for biofilm development, and we also analyzed the response of cells growing in YPD medium at 30°C to facilitate a direct comparison with previously published results for C. albicans (52). In both species, irrespective of the media and growth temperature used, low oxygen induces the expression of fatty acid and ergosterol metabolism, glycolysis and fermentation, heme biosynthesis and iron metabolism, and cell wall genes (see Table S3 in the supplemental material) (52). Similar pathways respond to anaerobiosis in Saccharomyces cerevisiae (31). We compared the genes that are upregulated under conditions of hypoxia in C. parapsilosis to gene lists generated from several profiling experiments with C. albicans (http://candida.bri.nrc.ca/l2l/). The most similar profiles (found in C. parapsilosis cultures grown in both SD and YPD media under conditions of hypoxia) are C. albicans genes downregulated in deletions of efg1 (24), ace2 (37), and pmt6 (10) and genes upregulated in hypoxia (from reference 52). Efg1, Ace2, and Pmt6 are all required for the expression of glycolytic genes during normoxia, although Efg1 at least is not required for hypoxic induction (10, 37, 52).

In both C. albicans and S. cerevisiae, exposure to low-oxygen in rich medium (YPD medium at 30°C) results in decreased levels of expression of genes in the tricarboxylic acid cycle and in the electron transport chain (31, 52). We did not observe similar changes in expression in C. parapsilosis cultures grown in SD medium at 37°C in 1% oxygen, and in fact, levels of expression of several cytochrome c oxidase (COX) genes were increased (see Table S3 in the supplemental material). However, when C. parapsilosis cultures were grown in YPD medium at 30°C, there was a decrease in levels of expression in most of the tricarboxylic acid enzymes (LSC1, LSC2, CIT1, IDH1, IDH2, IDP2, FUM12, and MDH1) and in components of the F₁-ATP synthase (ATP7 and ATP14) (data not shown). The profile of downregulated genes in C. parapsilosis cells grown in YPD medium under conditions of hypoxia most closely resembles the profile of C. albicans genes downregulated under conditions of hypoxia, whereas there is little obvious similarity between the profile of downregulated genes in C. parapsilosis cultures grown in SD medium and any other C. albicans profile. A reduction in respiration in low oxygen during growth on rich medium is therefore a conserved response across several fungal species, but the medium used (and perhaps the growth temperature) can have a major effect.

One important feature of the hypoxic response in S. cerevisiae is the induction of the seripaurin family of mannoproteins (14). There is no expansion of this family in Candida species, but hypoxia does induce the expression of other cell wall genes in C. albicans, including RBT5 and other members of the CFEM family and HWP1, a member of the RBT1 family (52, 55). Surprisingly, the expression of the CFEM family is not induced by hypoxia in C. parapsilosis either in the array experiments (see Table S3 in the supplemental material) or when measured by RT-PCR (not shown).

One of the closest relatives of HWP1 in the C. parapsilosis genome is CPAG_00831, an ortholog of C. albicans RBT1, which is induced both under conditions of hypoxia and in biofilms. The expression in biofilms is unlikely to be attributable to the difference in pH between the planktonic and biofilm cultures (Fig. 1). Knockout analysis confirms that RBT1 is important for biofilm development (Fig. 4). RBT1 is induced during filamentation in C. albicans (9). An rbt1 mutant in C. albicans is defective in virulence in a rabbit cornea model (9) and is partially attenuated in a mouse cornea model (27). However, there is no effect on hyphal formation in C. albicans, suggesting that the reduction in the virulence of rbt1 is not associated with a defect in hyphal formation. RBT1 has not been associated with biofilm growth in C. albicans, but HWP1 is required (42). The Hwp1/Rbt1 family is therefore implicated in biofilm development, and possibly in virulence, in both species.

Several genes that are important for biofilm formation in C. albicans have been identified. Ace2, which regulates the expression of cell wall genes, is important for adherence (28). Mutations in some genes (such as SUV3, NUP85, MDS3, and KEM1) cause defects in hyphal development that may be important for biofilm formation (49). Efg1 and Tec1, major regulators of hyphal growth, are also important for biofilm development (41, 47). Tec1 regulates the expression of BCR1, which is required for biofilm development in both C. albicans and C. parapsilosis (15, 40, 41). The level of expression of CPH2 (which regulates the expression of TEC1 in C. albicans) is increased both in biofilms and under conditions of hypoxia in C. parapsilosis (see Tables S2 and S3 in the supplemental material). The expression of TEC1 is also induced under conditions of hypoxia (see Table S3 in the supplemental material). It is therefore likely that the Cph2/Tec1/Bcr1 pathway plays a conserved role in biofilm development in Candida species independent of the yeast/hyphal transition. However, there are also distinct differences. The Bcr1-dependent regulation of ALS genes is important for biofilm formation in C. albicans (40), but Bcr1 in C. parapsilosis plays no obvious role in regulating the expression of ALS genes (15). We identified five members of the ALS family in C. parapsilosis (compared to seven members in C. albicans), but we observed only minor changes in the levels of expression of two of these members in C. parapsilosis biofilms (Fig. 5). Nobile et al. (43) recently demonstrated that Als1, Als3, and Hwp1 act as redundant adhesins in biofilm formation in C. albicans. It is possible that Rbt1 in C. parapsilosis plays a role similar to that of Hwp1, but determining if the ALS family has any function will require more investigation.

There is not complete overlap between the gene sets induced in biofilms and those induced under hypoxic conditions (Fig. 3). Levels of expression of heme biosynthesis enzymes are increased under conditions of hypoxia but are not obvious in biofilms. The expression of genes requiring molecular oxygen (HEM13 and HEM14) and also the expression of earlier steps in the pathway, including HEM1 (5-aminolevulinate synthase) and HEM4 (uroporphyrinogen III synthase), are induced (see Table S3 in the supplemental material). The expression of heme biosynthesis is also induced under conditions of hypoxia in C. albicans and S. cerevisiae (31, 52, 57). In S. cerevisiae, decreasing of levels of heme biosynthesis is at least one method used to sense lowering oxygen concentrations.

Oxygen availability is important for biofilm development by bacterial species. In biofilms formed by Pseudomonas aeruginosa, low-oxygen conditions result in decreased protein synthesis, and under aerobic conditions, the concentration of available oxygen decreases across the biofilm (63). Growth in low oxygen inhibits adhesion and biofilm formation by Escherichia coli (33). Several bacterial species respond to lower oxygen levels in biofilms by altering the expression of the respiratory pathway (reviewed in reference 4). C. albicans regularly forms biofilms in low-oxygen environments, such as on dentures, but biofilm development is generally reduced compared to that in aerobic environments (7, 58). Some Candida species appear to form more biofilms under anaerobic conditions (58). Ours is the first study to our knowledge that suggests that the hypoxic environment of biofilms results in an altered transcriptional response in Candida species, at least for C. parapsilosis.