The fungal clinical and laboratory strains used in this study are listed in the Tables S1, S2, respectively. C. albicans and other yeast strains were routinely maintained at 30°C on YPD (1% yeast extract, 2% peptone, 2% dextrose, with 50 mg/ml uridine) or synthetic complete (SC; 0.67% yeast nitrogen base with ammonium sulfate, 2.0% glucose, and 0.079% complete supplement mixture) or RPMI (RPMI-1640 with 0.3 g/L-glutamine) media. Acidic pHs used for VPA susceptibility were obtained using hydrochloric acid.

Valproic acid (VPA; Sigma-P4543) was dissolved in sterile water (50 mg/ml). Standard antifungals used for VPA-synergy assessment are: Fluconazole (FCZ; Sigma-F8929), Caspofungin (CSP; Sigma-SML0425), Voriconazole (VCZ; Sigma-PZ0005), Amphothericin B (AMB; Sigma-A488), Itraconazole (ITZ; Sigma-I6657), Clotrimazole (CTZ; Sigma-C6019), Teroconazole (TCZ; Toronto Research Chemicals-T110600), Miconazole (MCZ; Sigma-PHR1618), Terbinafine (TRB; Sigma-T8826), Nystatin (NST; Sigma-N4014), and Micafungin (MCF; McKesson Canada-205666). Antifungals were prepared using DMSO for Amphothericin B (30 mg/ml), Fluconazole (300 mg/ml), Terbinafine (10 mg/ml), Clotrimazole (9 mg/ml), Nystatine (5 mg/ml), Miconazole (30 mg/ml), Terconazole (1 mg/ml); water for caspofungin (10 mg/ml), Voriconazole (10 mg/ml), Micafungin (10 mg/ml), and chloroform for Itraconazole (50 mg/ml).

The pH-dependant effect of VPA on C. albicans was evaluated as follows: The reference clinical strain SC5314 was grown overnight in YPD medium at 30°C in a shaking incubator. Cells were then resuspended in fresh SC at an optical density at 595 nm (OD_595nm) of 0.05. The pHs of SC media were adjusted using sodium hydroxide or hydrochloric acid for alkaline and acidic pHs, respectively. A total volume of 99 μl C. albicans cells was added to each well of a flat-bottom 96-well plate in addition to 1 μl of the corresponding stock solution of VPA. Plates were incubated in a Sunrise-Tecan plate reader at 30°C with agitation and OD_595nm readings were taken every 10 min over 24 h. Experiments were performed in triplicate, and average values were used for analysis. VPA effect on other fungal species at acidic pH was performed in a similar fashion.

The Minimal Inhibitor Concentration (MIC) was determined following the CLSI recommendations (CLSI, 2008). Briefly, 50 μl of VPA or standard antifungals at two-fold the final concentration prepared in RPMI was serially diluted in flat-bottom 96-well plates (Costar-Corning) and combined with 50 μl of an-overnight culture of C. albicans and other yeasts at 10⁴ cell/ml. Plates were incubated at 30°C with shaking and OD_595nm readings were taken after 24 h using the Sunrise-Tecan plate reader. The MIC was determined as the first well with growth reduction of >10% based on OD_595nm values in the presence of VPA or conventional antifungals as compared to untreated control cells. Time-kill was performed as described by Sanglard et al. (2003). Briefly, C. albicans SC5314 strain cultures were grown in RPMI pH 4.5 at 30°C under shaking in the presence of different concentration of VPA for defined time periods (6, 24, and 48 h). Fractions of cultures were removed at each exposition time and the colony forming units (CFU) counts were ensured by serial dilution in YPD-agar.

Evaluation of synergistic interactions between VPA and standard antifungals was performed using RPMI-1640 medium buffered at pH 4.5. Synergism was assessed by calculating the fractional inhibitory concentration (FIC) index as described by Epp et al. (2010). The FIC index was calculated as follows: (MIC of VPA in combination/MIC of VPA alone) plus (MIC of a standard antifungal in combination/MIC of a standard antifungal alone).

Biofilm formation and XTT (2,3-bis(2-methoxy-4-nitro-5-sulfo- phenyl)-2H-tetrazolium-5-carboxanilide) assays were carried out as previously described by Askew et al. (2011). Overnight YPD cultures were washed three times with PBS and resuspended in fresh RPMI supplemented with L-glutamine (0.3 g/l) to an OD_595nm of 1. C. albicans yeast cells were allowed to adhere to the surface of 96-well polystyrene plate for 3 h at 37°C in a rocking incubator. Non-attached cells were washed from each well three times with PBS and fresh RPMI supplemented with VPA was added for 24 h at 37°C for biofilm formation. The plates were then washed and fresh RPMI supplemented with 100 μl of XTT-menadione (0.5 mg/ml XTT in PBS and 1 mM menadione in acetone) was added. After 3 h incubation on the dark at 37°C, 80 μl of the resulting colored supernatants were used for colorimetric reading (OD_490nm) to assess metabolic activity of biofilms. A minimum of four replicates were at least performed.

Damage of vaginal epithelial cells was assessed using the lactate dehydrogenase (LDH) cytoxicity detection kit (Sigma) based on the release of LDH in the surrounding medium following the manufacturer's protocol. VK2/E6E7 (ATCC- CRL-2616) vaginal epithelial cell line was grown on a keratinocyte-serum free medium (supplemented with 0.1 ng/ml recombinant epidermal growth factor and 50 μg/ml bovine pituitary extract) as a monolayer to 95% confluency on a 96-well culture plate and incubated at 37°C with 5% CO_2. VK2/E6E7 cells were infected with 2 × 10⁴ of C. albicans SC5314 blastospores for 24 h. A total of 100 μl supernatant was removed from each experiment and LDH activity in this supernatant was determined by measuring the absorbance at 490 nm (OD_490nm). LDH activity were calculated as the mean of, at least, three independent biological replicates.

A total of 2371 mutants from the transcription factors (Homann et al., 2009) (365 strains), transcriptional regulators (Vandeputte et al., 2012) (509 strains), kinases (Blankenship et al., 2010) (165 strains), and generalist collections (Noble et al., 2010) (1,332 strains) were screened for VPA-sensitivity. These mutant libraries were obtained from the Fungal Genetics Stock Center (FGSC). With the exception of the kinase collection where genes were disrupted by transposon insertions, mutants of the other collections were created through gene deletion of the complete ORF. In most cases and for each gene, at least two independent transformants were screened. Mutant strains were grown overnight in SC with pH 4.5 on flat-bottom 96-well and were plated on SC-agar pH 4.5 medium with or without VPA (50 μg/ml) using a 96-well blot-replicator. Mutants exhibiting more than fold-fold growth reduction based on colony diameter were compiled together in a 96-well plate and their sensitivity were confirmed to different concentration of VPA (10, 50, and 100 μg/ml) following the same procedure. Mutant strain with established VPA sensitivity were individually reconfirmed by serial dilution spot assay. A complete listing of VPA-sensitive mutants is shown in Table S3. The overrepresentation of specific GO terms associated with the function of gene required for VPA tolerance was determined with GO Term Finder using a hypergeometric distribution with multiple hypothesis correction (http://www.candidagenome.org/cgi-bin/GO/goTermFinder) (Inglis et al., 2012). Descriptions related to gene function in Table S3 were extracted from CGD (Candida Genome Database) database (Inglis et al., 2012).

C. albicans vacuole integrity was assessed using the lipophilic vacuole membrane dye MDY-64 (Molecular probes, Fisher Scientific) following the manufacturer's recommended procedure. Briefly, cells were grown overnight on RPMI liquid medium with pH 4.5 at 30°C. Cells were pelleted and washed twice with fresh RPMI pH 4.5 and resuspended in the same medium at an OD₅₉₅ of 0.1. VPA was added at different concentrations (10, 50, and 100 μg/ml). Cells were incubated for 2 h at 30°C under agitation. Aliquots were taken from VPA-treated and non-treated cultures and the MDY-64 was added at a final concentration of 10 μM. Cells were incubated at room temperature for 3 min prior to confocal microscopy visualization. Images were acquired with a 1.3-numerical-aperture (NA) 63x objective on a Leica DMI6000B inverted microscope connected to a Hamamatsu C9100-13 camera.

Pan1-green fluorescent protein (GFP), End3-GFP and LIFEACT-GFP (Epp et al., 2013) were visualized using confocal microscopy as follow: an overnight culture was diluted in SC supplemented with 10 or 50 μg/ml VPA to an OD_595nm of 0.05 and grown for four generations at 30°C under agitation. Cells were imaged as described for the vacuole staining experiments.

Antifungal activity of VPA on C. albicans was evaluated by monitoring OD_595nm of cultures exposed for 24 h to increased concentration of VPA in SC media at different pHs. VPA exerted an inhibitory effect that was exaggerated in acidic pH (Figure 1A). Antifungal activity of VPA was also assessed in other clinically relevant Candida species including C. glabrata, C. tropicalis, C. parapsilosis, and C. krusei in addition to the yeast S. cerevisiae. The obtained data demonstrates that VPA inhibited the growth of all tested fungal species, with C. albicans exhibiting higher sensitivity at elevated VPA concentrations (>216 μg/ml) (Figure 1B). Thus, VPA is a potent antifungal compound against C. albicans at acidic pH.

To test whether VPA had fungistatic or fungicidal activity on C. albicans at acidic pH, time-kill curve assays were performed. Two high concentrations of VPA which correspond to 125x (1,000 μg/ml) and 375x (3,000 μg/ml) of the MIC for the C. albicans reference strain SC5314 (Table 1) were tested. VPA exhibited a concentration-independent fungistatic activity (Figure 1C). Lower VPA concentrations ranging from 7.8 (MIC for the SC5314 strain) and 500 μg/ml were also tested and the obtained results demonstrates a similar fungistatic activity (result not shown).

Since VPA was highly potent against C. albicans, we wanted to test whether its antifungal activity can be expanded to other clinically sensitive and resistant strains of this yeast. Several azole-resistant strains with different resistant mechanisms, were selected (Table S1) in addition to echinocandin-resistant isolates. A total of four sensitive and 11 resistant strains (six azole- and five echinocandin-resistant strains) were examined using broth microdilution assay as specified by CLSI at both neutral or acidic pHs. The sensitivity of C. albicans isolates to VPA was pH-dependant and MICs ranged from 3.5 to 15.6 μg/mlfor both resistant and susceptible strains (Table 1). The range of MICs was also similar when comparing azole-resistant and echinocandin-resistant clinical strains separately (3.5–15.6 μg/ml). Overall, these results demonstrate that VPA may be of use to tackle therapeutic limitations related to acquired clinical resistance of C. albicans. Furthermore, comparable VPA-sensitivity in susceptible and resistant strains indicates that the mechanisms that confer resistance to azoles and echinocandins are distinct from those that may cause VPA resistance.

To verify whether VPA exerts protective antifungal activity during host cell invasion, interaction of C. albicans with the human epithelial vaginal cell line VK2/E6E7 were performed as described in the method section. C. albicans-mediated damage of VK2/E6E7 cells were quantified based on the LDH release. Two different concentrations of VPA (7.8 and 78 μg/ml) corresponding to the MIC and 10x MIC for C. albicans SC5314 strain were used. In accordance with our in vitro data, the VPA had no significant protective effect at pH 7 (Figure 2). At pH 5, 7.8, and 78 μg/ml of VPA prevented 55 and 100% of VK2/E6E7 damage, respectively, as compared to the control. Intermediate protective activity was perceived at pH 6 where 28 and 52% damage reduction was obtained with 7.8 and 78 μg/ml of VPA, respectively. In support of in vitro data, these results demonstrate that VPA confers a protective antifungal activity during the invasion of vaginal epithelial cells.

Different standard antifungals used against C. albicans and other human fungal pathogens were screened to identify drugs that could potentiate the anti-Candida activity of VPA. Interactions of VPA with other 11 antifungal agents including azoles (Fluconazole, Voriconazole, Itraconazole, clotrimazole, Terconazole, and Miconazole), polyenes (Amphothericin B and Nystatin), echinocandins (Caspofungin and Micafungin), and the allylamine, Terbinafine were tested. Based on the appreciation of the FIC index in the clinical strain SC5314, VPA was found to exhibit an apparent synergistic interaction with terbinafine (Table 2; FIC index < 0.5). VPA-terbinafine combinations were also synergistic in azole and echinocandin resistant clinical strains (Table 2).

The effect of VPA on biofilm formation was evaluated using the metabolic colorimetric assay based on the reduction of XTT at acidic and neutral pHs. At neutral pH, no VPA anti-biofilm activity was noticed of all tested concentrations for the C. albicans SC5314 reference strain (not shown). In contrast, at pH 4.5, biofilm inhibition was apparent at 1.44 μg/ml of VPA with ~5% of inhibition as compared to the control (Figure 3). The MIC of VPA on the SC5314 strain was evaluated at 7.2 μg/ml. The effect of VPA on biofilm formation was also tested in two azole-resistant strains (S2 and F5) with different resistance mechanisms in addition to two echinocandin-resistant isolates (DPL-1008 and DPL-1010). As for the SC5314 sensitive strain, the four resistant strains exhibited a clear reduction in metabolic activity at 1.44 μg/ml of VPA (Figure 3). The MIC values for the azole-resistant strains were similar (2.88 μg/ml VPA) and slightly decreased as compared to the SC5314 susceptible strain. The two echinocandin-resistant strains DPL-1008 and DPL-1010 were highly sensitive to VPA as compared to other strains and their MIC was noticed at 1.44 μg/ml of VPA. These results demonstrate that, in addition to its antifungal activity on planktonic cells, VPA is also active on sessile forms of C. albicans at acidic pH.

To gain insight into the mechanism of action of VPA associated with its antifungal property, a comprehensive regulatory and generalist mutant collections of C. albicans were screened for their sensitivity to VPA. Among the 947 unique mutants that were screened, 55 were confirmed to be hypersensitive to VPA (Table S3). To identify the functional categories that are associated with mutations affecting VPA susceptibility, we performed gene ontology (GO) enrichment analysis. Our data demonstrated that VPA sensitive mutants are defective in genes related primarily to vacuole transport (p = 1.72e-08) and organization (p = 8.86e-09) (Table 3, Table S4). This include mutants of vacuolar protein sorting (vps15, vps34, vps64, and ypt72), proteins associated with the retromer complex (pep7 and pep8), and proteins required for vacuole inheritance, and organization (cla4, pep12, vam6, vps41, and pep12). Requirement of vacuolar functions for VPA tolerance was also reported previously in S. pombe (Zhang et al., 2013) and S. cerevisiae (Deranieh et al., 2015) where genome-wide screens demonstrated that retromer complex and vacuolar ATPases, respectively, were associated with VPA sensitivity. Taken together, our chemogenetic screen provides a rational for mechanistic investigation into the effect of VPA on fungal vacuole.

Our chemogenetic screen demonstrated clearly that C. albicans sensitivity to VPA were exaggerated in mutant of vacuolar transport, organization and inheritance. The requirement of intact vacuolar pathways for VPA tolerance suggests that VPA might alters the function and/or the integrity of the vacuole. To verify this hypothesis, the integrity of C. albicans vacuoles were assessed using the vacuole membrane marker, MDY-64, in cells treated or not with different concentrations of VPA at pH 4.5. A dominant fraction of non-treated cells internalized the MDY-64 dye and exhibited well-structured vacuoles with two to four compartments comprising discernable lumens (Figure 4A). However, cells treated with either 10 or 50 μg/ml of VPA displayed an altered vacuole structure with a foamy fluorescence pattern and indistinguishable lumens (Figure 4B). These findings suggest that VPA affect the morphology and the integrity of vacuoles in C. albicans.

Perturbation of vacuoles by VPA might be either a direct consequence where VPA act in situ on vacuole or by impacting indirectly other process required for proper vacuole biogenesis and organization. Indeed, vacuolar integrity and homeostasis depends on the proper functioning of different cellular processes including actin filaments organization (Eitzen et al., 2002), endocytosis (Michaillat and Mayer, 2013), and the phosphatidylinositol phosphate signaling (Michell et al., 2006). Recent work by Deranieh et al. (2015) indicated that VPA led to cellular depletion of inositol, which disrupts the vacuolar phosphoinositide, PI(3,5)P2 (Phosphatidylinositol 3,5-bisphosphate) homeostasis and consequently compromise vacuole morphology. In this regard, we checked whether VPA-mediated vacuole alteration can be bypassed by inositol supplementation or deprivation. Our data demonstrate that the altered vacuole phenotype was not influenced by lack or excess of inositol (data not shown), suggesting that the probable VPA-induced inositol depletion in C. albicans is unlikely to account for the toxicity of this compound under our conditions.

To check whether the vacuole alteration caused by VPA is related to a defect in endocytosis or actin filament organization, we have used a Pan1- and End3-GFP fusions as clathrin-coated vesicles markers, and LIFEACT-GFP (Epp et al., 2013) to monitor actin patches and cables. VPA treatment did not cause any apparent alteration of the organization of actin filaments or endocytic vesicles (Figure S1). Taken together, these results support that the direct alteration of vacuole is the cellular mechanism underlying the antifungal activity of VPA.

A provisional patent application for an antifungal vaginal gel formulation related to this work (Patent number US 62/532,603) has been filled by Université Laval. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.