Candida albicans OPI1 Regulates Filamentous Growth and Virulence in Vaginal Infections, but Not Inositol Biosynthesis

ScOpi1p is a well-characterized transcriptional repressor and master regulator of inositol and phospholipid biosynthetic genes in the baker’s yeast Saccharomyces cerevisiae. An ortholog has been shown to perform a similar function in the pathogenic fungus Candida glabrata, but with the distinction that CgOpi1p is essential for growth in this organism. However, in the more distantly related yeast Yarrowia lipolytica, the OPI1 homolog was not found to regulate inositol biosynthesis, but alkane oxidation. In Candida albicans, the most common cause of human candidiasis, its Opi1p homolog, CaOpi1p, has been shown to complement a S. cerevisiae opi1∆ mutant for inositol biosynthesis regulation when heterologously expressed, suggesting it might serve a similar role in this pathogen. This was tested in the pathogen directly in this report by disrupting the OPI1 homolog and examining its phenotypes. It was discovered that the OPI1 homolog does not regulate INO1 expression in C. albicans, but it does control SAP2 expression in response to bovine serum albumin containing media. Meanwhile, we found that CaOpi1 represses filamentous growth at lower temperatures (30°C) on agar, but not in liquid media. Although, the mutant does not affect virulence in a mouse model of systemic infection, it does affect virulence in a rat model of vaginitis. This may be because Opi1p regulates expression of the SAP2 protease, which is required for rat vaginal infections.


Introduction
Candida albicans is a commensal organism that lives as a benign resident of the microflora of the human oral, gastrointestinal, and vaginal tracts as well as the skin. It can shift from a commensal to a pathogenic state in response to environmental stimuli that trigger developmental programs that induce the expression of virulence factors. Virulence factors exhibited by C. CPH1. Therefore, we wished to investigate if C. albicans OPI1 has a similar role in inositol regulation to ScOPI1 and CgOPI1, or if it has possibly been transcriptional rewired.
In this communication, we report that C. albicans OPI1 does not regulate the inositol biosynthetic gene INO1, but affects the SAP2 expression and virulence of C. albicans in a rat vaginitis model. In addition, OPI1 affects morphogenesis at 30°C. These results illustrate that the regulation of inositol biosynthesis in C. albicans and S. cerevisiae is different. From now on, in this paper, all genes from C. albicans will be referred to by their simple names such as OPI1 or INO1, whereas genes from other organisms such as S. cerevisiae will be referred to as ScOPI1 or ScINO1. Strains and growth media C. albicans strains used in this study are shown in Table 1. Media used in this study include YPD (yeast extract-peptone-dextrose: 1% yeast extract, 2% peptone, 2% glucose), defined medium 199 (M199, Invitrogen, pH7.0 adjusted by 150mM HEPES), Spider (1% nutrient broth, 1% mannitol, 0.2% dipotassium phosphate, 1.35% agar), YPD containing 10% fetal bovine serum, YCB-BSA (1.17% yeast carbon base-Difco, 0.2% bovine serum albumin-Sigma),YCB-BSA-YE (1.17% yeast carbon base, 0.2% bovine serum albumin, 0.1% yeast extract, pH 5.0) [28,29]. Unless otherwise stated, agar plates were solidified with 2% agar (granulated, Fisher). Strain construction

Ethics Statement
The C. albicans OPI1 gene (OPI1) was disrupted by using the CaNAT1-FLP cassette [30] ( Table 2). For the OPI1 disruption construct, the 379 base pair (bp) 5' non-coding region (NCR) of OPI1 was amplified with primers TRO522 and TRO526 (Table 3), and cloned as a KpnI-ApaI digested 228bp fragment into pJK863 5' of the CaNAT1-FLP cassette (Fig. 1A). The 448 bp 3' NCR of OPI1 was amplified with primers TRO524 and TRO525 which introduced SacII and SacI sites, and was cloned into pJK863 3' of the CaNAT1-FLPcassette (Fig. 1A). This created the OPI1 knock out construct plasmid pYLC36 ( Table 2, Fig. 1A), which was cut with KpnI and SacI to release the disruption construct (5' NCR of OPI1-CaNAT1-FLP-3' NCR of OPI1) which was transformed into the wild type SC5314 strain by electroporation [31]. The disruption construct was used to sequentially disrupt both alleles of OPI1. The OPI1 reconstitution construct was made by amplifying a 1.7 Kb fragment containing the OPI1 ORF and 5' NCR from SC5314 genomic DNA using primers (JCO12 and JCO14) that introduced KpnI and SalI sites. This fragment was ligated into the pRS316 vector along with another 1.7 Kb fragment containing the NAT1-3'NCR of OPI1 amplified from plasmid pYLC36 using primers JCO50 and TRO42 which introduced SalI and SacI sites. This resulted in the OPI1 reconstitution plasmid pYLC37 (  [32] and resulted in pYLC219. The SAP2 ORF was then cloned to XmaI-digested pYLC219 with primers JCO131 and JCO132, and resulted in pYLC221, which can constitutively express SAP2 under the control of the ACT1 promoter. To transform this constitutive construct into wild type and opi1Δ/Δ strains, a PpuMI-digested linear plasmid pYLC221 was integrated at the URA3 site of Candida genome, and resulted in OPI1/OPI1 URA3::P ACT1 -SAP2 (YLC223) and opi1Δ/Δ URA3::P ACT1 -SAP2 (YLC226).

Northern blot analysis
Northern blotting for SAP2 and INO1 expression was performed as described [33,34] with the following exceptions. Strains grown in YCB-BSA or YCB-BSA-YE medium at 37°C for 12 hrs (for SAP2) and in liquid medium 199 (pH 7.0) at 37°C for 2 hrs (for INO1) were collected for total RNA extraction by the hot phenol method. The PCR product containing bps 17-571 of the SAP2 ORF (primers JCO35 and JCO36) and bps 76-581 of the INO1 ORF (primers TRO562 and TRO563) were used as probes. Expression was normalized against C. albicans ACT1 gene expression probed on the same membrane. The ACT1 probe was generated with the primers JCO48 and JCO49.

Southern blot analysis
Hybridization conditions for the Southern blot analysis were similar to those for Northern blot analysis, except that the Techne Hybrigene oven was set to 60°C for the incubation step, and 42°C and 60°C for washing steps. The cells were grown in liquid YPD at 30°C overnight. The Table 3. Primers used in this study. genomic DNA was extracted using the Winston-Hoffman method [35] and 20μg of genomic DNA were subjected to Southern blotting. The genomic DNA of the wild type and opi1Δ/ Δmutants was cut by KpnI and SphI restriction enzymes. PCR products containing the~500bp 3' NCR of OPI1(primers TRO524 and TRO525) were used as probes for Southern blot confirmation (Fig. 1C).

Mouse bloodstream infection studies
Five-to six-week-old male CD1 mice (18 to 20 g) from Charles River Laboratories were used in this study. Mice were housed at five per cage. For infection, colonies from each C. albicans strain were inoculated into 20 ml of YPD. Cultures were grown overnight at 30°C with shaking in YPD, washed twice with 25 ml of sterile water, counted by hemocytometer, and resuspended at 10 7 cells per ml in sterile water. Mice were injected via the tail vein with 0.1 ml of the cell suspension (10 6 cells), and the course of infection was monitored for up to 14 days. The survival of mice was monitored twice daily, and moribund mice (body weight reduced by 30%, unable to eat/drink, or severely hunched) were euthanized with CO 2 . Cells were also plated on YPD to determine the viability. At least two independent infections were performed for each strain. The statistical analysis was done using Prism 5.03 software (GraphPad Software). For the mouse model of systemic infection, Kaplan-Meier survival curves were compared for significance using the Mantel-Haenszel log rank test. Statistical significance was set at P< 0.05.

Rat vaginitis studies
The protocol of estrogen-dependent rat vaginal infection model adapted from De Bernardis et al. [8] was used throughout this study. Briefly, oophorectomized female Wistar rats (80-100 g; Charles River, Calco, Italy) were injected subcutaneously with 0.5 mg of estradiol benzoate (Benzatrone; Samil, Rome). Six days after the first estradiol treatment, the animals were inoculated intravaginally with 10 7 yeast cells of each C. albicans strain in 0.1 mL. The inoculum was dispensed into the vaginal cavity through a syringe equipped with a multipurpose calibrated tip (Combitip; PBI, Milan, Italy). The yeast cells had been previously grown in YPD broth at 28°C on a gyratory shaker (200 rpm), harvested by centrifugation (1500 g), washed, counted in a hemocytometer, and suspended to the required number in saline solution. The results of two independent experiments are each represented separately. A third experiment involving all of the strains is not shown, but gave similar trends. In each experiment, each Candida strain was inoculated into 5 rats. Kinetics of C. albicans growth in, and clearance from, the vaginal cavity was measured by colony forming unit (CFU) enumeration after culturing 100 μl of vaginal samples, taken by washing the vaginal cavity by gentle aspiration of 100 μl of sterile saline solution, repeated four times, at 1:10 serial dilutions on Sabouraud agar containing chloramphenicol (50 μg/ml). CFUs were enumerated after incubation at 28°C for 48 h.

C. albicans OPI1 does not regulate INO1 expression
When heterologously expressed in an S. cerevisiae Scopi1Δ mutant, the C. albicans OPI1 gene has been demonstrated to repress expression of a reporter gene that contains the inositol/ choline responsive element (ICRE) found in ScINO1 and other ScOpi1p-ScIno2p-ScIno4p target genes [24]. This data suggested that C. albicans Opi1p may regulate the cognate C. albicans INO1 gene, as its homolog does in S. cerevisiae. In order to test this both copies of the C. albicans OPI1 gene were disrupted in C. albicans using the CaNAT1-FLP cassette [30] as described in Fig. 1.
The wild type and opi1Δ/Δ strains were then compared to see if the opi1Δ/Δ mutant would fail to repress INO1, as expected, if it acts like the homologous S. cerevisiae mutant, Scopi1Δ [22]. First, the strains were grown in Medium 199, pH 7.0, which contains low levels of inositol (~10 μM), which should result in high expression of INO1, and it was found that both upregulated INO1 to similar levels (Fig. 2). Then, they were grown in the same medium supplemented with 75 μM inositol, which should repress INO1 expression in wild-type, but not in the opi1Δ/Δ mutant, if it cannot repress the gene. However, in both strains, INO1 was similarly repressed, suggesting that inositol biosynthesis is regulated by different transcription factors in C. albicans.

The opi1Δ/Δ mutant exhibits hyperfilamentous growth in filamentinducing media at 30°C
It has been shown that ScOPI1 is necessary to activate invasive growth and ScFLO11 expression in S. cerevisiae [34]. It was therefore hypothesized that OPI1 would affect filamentous growth in C. albicans. Three filament-inducing media were used to test this hypothesis. In contrast to the situation with the Scopi1Δ mutant in S. cerevisiae, it was found that the opi1Δ/Δ mutant exhibited hyperfilamentous growth rather than hypofilamentous growth, but only at 30°C on solid filament-inducing agar plates (Fig. 3). This effect was not observed at 37°C on similar  media. These phenotypes were also not seen in liquid forms of the same filament-inducing media at either 30°C or 37°C. In order to control for a possible effect from some other unlinked mutation, a copy of the C. albicans OPI1 gene was reintegrated into the opi1Δ/Δ mutant (Fig. 1B), and it was found that the phenotype was restored when the wild-type copy of OPI1 was present (Fig. 3), indicating that the hyperfilamentous growth at 30°C is linked to the loss of OPI1 gene.

OPI1 does not affect virulence in a mouse model of systemic infection
The opi1Δ/Δ mutant appears to affect the ability of the fungus to repress filamentation at lower temperatures. Some hyperfilamentous mutants such as nrg1Δ/Δ and tup1Δ/Δ have been found to be attenuated in virulence in mouse models of infection [36]. Therefore, a mouse model of systemic infection was used to test the role of OPI1 in virulence. However, the OPI1 gene does not contribute to the virulence in this model since the opi1Δ/Δ mutant exhibits a similar phenotype to wild-type on the survival curves of mice (Fig. 4).

OPI1 is involved in establishing infection in the rat vaginitis model
In addition to infections of the bloodstream, C. albicans can also cause infections of mucosal surfaces including the vaginal tract [8]. A rat vaginitis model was used to determine if the opi1Δ/Δ mutation would play a role in the establishment of infection in this host niche. It was demonstrated that OPI1 was involved in establishing rat vaginitis. In this model C. albicans cells are injected into the rat vaginal tract, and then over time the level of colonization is measured based on the recovery of colony counts. It was discovered that the opi1Δ/Δ mutant is quickly cleared by the host compared to the wild type (Fig. 5). The opi1Δ/Δ::OPI1 reintegrant strain and opi1Δ/OPI1 heterozygous mutant had an intermediate phenotype between the opi1Δ/Δ mutant and wild type (Fig. 5).
It has been shown that deletion of the C. albicans SAP2 proteasegene [8] causes a similar clearance to the opi1Δ/Δ mutant, and a sap2Δ/Δ mutant (SAP2MS4A) [37] was included in this experiment as a control. Our results confirmed that an independently constructed sap2Δ/Δ mutant (gift from Joachim Morschhäuser), behaved like a previously constructed sap2Δ/Δ mutant, and exhibits reduced colonization in the rat vaginal tract (Fig. 5), suggesting the importance of SAP2 in the rat vaginitis model. Our results also indicate that OPI1 plays a critical role in establishing infection in the rat vaginal tract (Fig. 5).

OPI1 affects rat vaginal establishment through regulating SAP2
The similarity of the phenotypes of the opi1Δ/Δ mutant with the sap2Δ/Δ mutant suggested that OPI1 might act through SAP2. In wild-type cells, SAP2 is upregulated in bovine serum albumin (BSA) media. We performed reverse transcriptase (RT) real-time PCR to detect if OPI1 controls SAP2 expression in YCB-BSA medium. The opi1Δ/Δ mutant showed 5.5 fold reduced SAP2 expression compared to wild type (Fig. 6), indicating that OPI1 controls SAP2 expression. The opi1Δ/Δ::OPI1 reintegrant strain can restore the SAP2 expression and actually shows~3 fold higher expression of SAP2 than the wild type.
In order to test if OPI1 affects colonization of the rat vaginal tract through SAP2, an epistasis experiment was performed in which the SAP2 gene was overexpressed in the opi1Δ/Δ mutant via the ACT1 promoter (P ACT1 -SAP2). This overexpression was confirmed by Northern blotting (S1 Fig.). If opi1Δ/Δ blocked rat colonization by compromising SAP2 expression, then overexpression of SAP2 from an independent promoter should suppress the phenotype. In contrast to the opi1Δ/Δ mutant, the opi1Δ/ΔURA3::P ACT1 -SAP2 mutant was suppressed for its defect in rat vaginal infection, and behaved similarly to the wild type (Fig. 7). This implicates the OPI1 gene as a regulator of SAP2 in the vaginal tract of the rat.

Discussion
Our results show that the OPI1 gene of C. albicans, unlike its homologs in S. cerevisiae and C. glabrata [22,25], does not affect INO1 expression, but does repress filamentous growth at low temperature (Fig. 3) and regulates virulence in the rat vaginitis model (Figs. 5 and 7). The latter phenotype appears to be mediated by changes in SAP2 expression. The opi1Δ/Δ mutant exhibits reduced SAP2 expression compared with the wild type in liquid YCB-BSA medium (Fig. 6), and in vivo SAP2 overexpression can restore the opi1Δ/Δ mutant's vaginal colonization defect, when under the control of the constitutive ACT1 promoter. Epistasis experiments are inherently challenging to interpret, as overexpression of a target gene such as SAP2 could lead to enhanced colonization by a mechanism that bypasses the actual defect caused by the opi1Δ/ Δ mutation. Based on our data, this possibility cannot be completely ruled out. It is also possible that OPI1 controls colonization in the rat vaginal tract by regulating one of the other SAPs (e.g. SAP1, SAP3-10). However, as the differential expression of nine other SAPs in the opi1Δ/ Δ mutant compared to the wild type using the same condition (i.e. YCB-BSA liquid medium and RT real time PCR) was not detected, we do not know if these others are affected in vivo, and this remains to be examined (unpublished data). Meanwhile, further studies will be needed to test if SAP4, SAP5 and SAP6 genes are regulated by OPI1 under hypha-inducing conditions since SAP4-6 are hypha-specific genes [38,39].
In S. cerevisiae, ScOpi1p is the master regulator of ScINO1 and other phospholipid genes [15][16][17]20]. ScOpi1p controls expression in response to cellular inositol levels by binding to Ino2p in the Ino2p-Ino4p heterodimer and repressing its activation of ScINO1, among other targets. When inositol is plentiful, PI is efficiently synthesized from CDP-DAG and inositol by the ScPis1p enzyme [40,41]. In this circumstance, the endoplasmic reticulum (ER) localized pool of phosphatidic acid (PA), which is the precursor for CDP-DAG, is consumed, and ScOpi1p is translocated to the nucleus. There, it binds ScIno2p and represses ScINO1 with help from the global repressor Sin3p via a direct interaction involving the N-terminal Sin3p binding domain of ScOpi1p [16,24]. When inositol is not plentiful in the environment, cellular stores drop and PI synthesis slows causing a build-up of precursors including PA. ScOpi1p binds to PA in the ER via its basic domain, and the ER membrane protein Scs2p via its FFAT domain [20,42]. This sequesters ScOpi1p to the ER, and then Ino2p-Ino4p activate transcription of ScINO1 so inositol can be synthesized for PI production.
A previous report demonstrated that OPI1 from C. albicans could complement an Scopi1Δ mutant in S. cerevisiae, and it could repress the ICRE promoter element found on ScINO1 and other phospholipid biosynthetic genes when expressed heterologously in S. cerevisiae [24]. However, we found that in C. albicans, OPI1 does not regulate INO1 expression. This overlap in function of CaOPI1 when expressed heterologously in S. cerevisiae, but not endogenously in C. albicans, may be due to the conservation of some key domains required for ScOpi1p function, but not the conservation of other domains (S2 Fig.). In particular, the C. albicans Opi1p has very little conservation with ScOpi1p in the large N-terminal ScSin3p binding domain [16]. However, CaOpi1p does have some conserved sequences with the C-terminal ScIno2p interaction domain of ScOpi1p, including two out of three residues (ScOpi1p aas 358-360) that were shown to be crucial for ScIno2p-ScOpi1p interactions in S. cerevisiae [16]. In contrast, CaOpi1p shares very few residues in common with ScOpi1p in the PA-binding basic domain, and no residues of the FFAT domain that binds to ScScs2p [20,42]. It does, however, carry a leucine zipper motif with some isoleucine substitutions that has been shown to be crucial for ScIno2p-ScOpi1p interactions [15]. Thus, this conservation of some domains, but not others may help explain why CaOpi1p can complement a Scopi1Δ mutant for ScINO1 repression [24], but not act the same within C. albicans itself. Further support for our findings comes from the observation that CaINO2 and CaINO4 do not appear to regulate CaINO1 either, but may actually regulate ribosomal genes [26]. This is in marked contrast to CgOPI1 from C. glabrata, which does regulate CgINO1 with help from CgINO2 and CgINO4 [25]. Consistently, CgOpi1p has close conservation of all of the important regulatory domains of ScOpi1p (S2 Fig.). Interestingly, one other ScOpi1p homolog has been characterized, and this is Yas3p from Yarrowia lipolytica. Yas3p also does not have a number of domains conserved with ScOpi1p, and like CaOpi1p does not regulate YlINO1, but does, along with Ino2p and Ino4p homologs Yas1p and Yas2p, respectively, regulate hexane metabolism genes [43]. The C. albicans regulators of CaINO1 are currently unknown, and this will be interesting to elucidate, as expression of CaINO1 is regulated by extracellular inositol levels, but not by CaOpi1p and apparently not by CaIno2p or CaIno4p either.
Finally, the role Opi1p in repressing filamentation at 30°C on solid media remains elusive (Fig. 3). The opi1Δ/Δ mutant exhibits hyperfilamentous growth in filament-inducing agar plates including medium 199, spider, and 10% serum at 30°C, but not 37°C. These results indicate that OPI1 might be a low temperature repressor of filamentous growth. It has been demonstrated that C. albicans CPP1, a tyrosine phosphatase, is required tor repress the yeast to hyphal transition at 23°C in contact with solid surfaces [44,45]. The cpp1Δ/Δ mutant exhibited hyperfilamentous growth on spider and a wide variety of rich and defined solid media including Lee's medium, YPD, YPM, and 10% serum at 23°C, but not at 37°C. The germ tube formation defect of the cpp1Δ/Δ mutant was not observed at liquid culture at 37°C, an effect similar to opi1Δ/Δ mutant. In contrast to opi1Δ/Δ, the cpp1Δ/Δ mutant exhibited reduced virulence in mouse systemic infection and mouse mastitis models [44][45][46]. The relationship between Opi1 and Cpp1 is unknown and needs further studies in C. albicans. Taken together, our data suggest that, when compared to its homolog in S. cerevisiae, C. albicans has a transcriptionally rewired regulator, OPI1, which does not regulate INO1 expression but affects morphogenesis, SAP2 expression and virulence in a rat vaginitis model. It also makes it clear that identification of ScOpi1p homologs in other fungi does not clearly implicate them for roles in regulating inositol biosynthesis in these microbes. Rather, the Opi1p family members, which are conserved in a wide variety of fungi appear to have a diversity of functions.
Supporting Information S1 Fig. SAP2 is overexpressed in the P ACT1 -SAP2 construct. Expression was tested by Northern blotting in YPD media, and it was confirmed that the P ACT1 -SAP2 construct overexpressed SAP2.