Fungal strains used in this study are presented in Table 1. Yeast peptone dextrose agar (YPD) was purchased from BD Biosciences (San Jose, CA). Auranofin (Enzo Life Sciences, Farmingdale, NY), fluconazole (Acros Organics, New Jersey), and flucytosine (TCI chemicals, Tokyo, Japan) were purchased from commercial vendors. XTT-sodium salt, menadione, RPMI powder, and MOPS were purchased from Sigma-Aldrich (St. Louis, MO). Concanavalin A–conjugated with FITC 488 dye was acquired from Thermo Fisher Scientific Inc. (Waltham, MA).

Antifungal susceptibility testing was carried out as per the National Committee for Clinical Laboratory Standards M-27A3 (NCCLS) guidelines (da Silva et al., 2016). Briefly, the inocula were prepared from 24 h old cultures of Candida spp. or 48 h old cultures of Cryptococcus spp. in YPD plates. Five colonies were then transferred to 5 mL of sterile 0.9% saline (PBS). The suspensions were adjusted to McFarland standard 0.5 and then diluted 1:2000 in RPMI 1640 buffered to pH 7.0 with 0.165 M MOPS (RPMI-MOPS) to yield an inoculum of 5.0 × 10², −2.5 × 10³ CFU/mL. An aliquot (100 μL) of the resulting suspension was incubated with serially diluted fluconazole, flucytosine, and auranofin for 24 h for Candida spp and 72 h for Cryptococcus spp. The minimum inhibitory concentration (MIC) of fluconazole and flucytosine were determined as the prominent decrease (~50%) in visible growth compared to untreated controls, as per NCCLS guidelines. Similarly the MIC of auranofin was determined as the lowest concentration resulting in 50% reduction in visible growth. All experiments were carried out in triplicate wells.

Fungal cultures of Candida albicans and Cryptococcus neoformans were diluted approximately to 5 × 10⁵ CFU/mL and treated with 5 × and 10 × MICs of auranofin and fluconazole (in triplicate) in RPMI-MOPS, at 35°C. Samples were collected at indicated time points and serially diluted in PBS and plated onto YPD plates. Plates were incubated at 35°C for 24–48 h prior to counting fungal colony forming units (CFU), as described elsewhere (Cantón et al., 2004).

C. albicans ATCC 10231 was grown in YPD broth at 35°C for 24 h. Cells were washed with PBS and resuspended in RPMI-MOPS at 10⁶ cells/mL (Pierce et al., 2008; Rane et al., 2012). An aliquot (100 μL) of cell suspension was transferred to wells in a 96-well tissue culture plate. After 48 h incubation (at 37°C), wells were washed with PBS and drugs (auranofin, fluconazole, and flucytosine) were added at indicated concentrations. After 24 h of incubation, the supernatant was removed and 100 μL of XTT/menadione solution was added to each well. The plates were covered with aluminum foil and incubated at 37°C for 1 h. Aliquots (75 μL) were taken from each well and the absorbance (OD₄₉₅) was measured using a spectrophotometer. The antifungal activity of each drug was expressed as a percentage of metabolic activity of treatment groups relative to the DMSO-treated control groups. The experiment was performed using triplicate samples for each treatment regimen.

C. albicans ATCC 10231 was seeded on FBS-coated glass cover slips in 6-well tissue-culture plates and grown in RPMI-MOPS medium with 0.2% glucose at 37°C (Dongari-Bagtzoglou et al., 2009). After 48 h, wells were washed with PBS and drugs (auranofin, fluconazole, and flucytosine) were added at indicated concentrations. After 24 h of treatment, wells were washed with PBS and stained with concanavalin A–conjugated with FITC 488 dye (25 μg/mL in PBS) for 45 min at 37°C. After incubation, the coverslips were washed three times with PBS and mounted on glass slides. Stained biofilms were observed using Leica confocal laser scanning microscopy. Images were reconstructed using IMARIS software.

Initial testing of Saccharomyces cerevisiae sensitivity to auranofin was performed with the wild-type BY4743 diploid strain, the isogenic parent to the heterozygous diploid deletion collection. BY4743 was grown in YPD in 96-well plates with 1% DMSO or auranofin in concentrations ranging from 10 to 200 μM to determine a suitable level of growth inhibition. Auranofin (75 μM) was used for haploinsufficiency profiling because it delayed growth by 30% compared to the no drug control half-maximal optical density (OD). All experiments were performed at 30°C and cultures were shaken at 300 rpm. The heterozygous deletion set was purchased in a pooled format (Thermo Fisher Scientific, Waltham, MA). A frozen aliquot (200 μL) was thawed and used to inoculate 2 mL of YPD and grown for 9 h to reach an OD₆₀₀ of 4.0. The culture was diluted to an OD₆₀₀ of 0.13 and either 1% DMSO or 75 μM auranofin was added (three replicates each, 1 mL) and grown for 7 h. The cultures were grown again by diluting to an OD of 0.13 in 1 mL YPD with DMSO or 60 μM auranofin and grown for 8 h. Cultures were harvested and genomic DNA extracted using the YeaStar Genomic DNA kit (Zymo Research, Irvine, CA). The UPTAGs were amplified by PCR with Phusion Hot Start II High-Fidelity DNA polymerase at 0.02 U/μL (Thermo Fisher Scientific, Waltham, MA) using 0.5 ng/μL genomic DNA. Primers are listed (Table S1). The PCR reactions were electrophoresed on an agarose gel and the 267 bp product extracted using a QIAquick Gel Extraction Kit (Qiagen, Valencia, CA). Purified DNA was measured using a Qubit instrument and samples were normalized and mixed together to a final concentration of 10 nM. Strains were grown and maintained on media according to standard practices (Amberg et al., 2005).

The pooled PCR products were sequenced using standard Illumina sequencing in a HiSeq 2500 instrument. The reads were separated based on a 5 base multiplex tag unique for each experiment and an average of 5 million reads per replicate was obtained. The UPTAG barcodes in each experimental sample were separated based on a reference database of recharacterized barcode sequences (Smith et al., 2009).

The resulting strain counts were imported into R and analyzed with edgeR (Robinson et al., 2010). Sequencing library sizes were normalized using the default parameters. Only strains with one or more counts in three or more samples were analyzed further. Differential representation of strains was determined using the quantile-adjusted conditional maximum likelihood (qCML) method. False discovery rates were determined to control for multiple testing.

Overnight grown yeast cells were diluted (OD₆₀₀ ~ 0.03) and grown in the presence and absence of auranofin, at indicated concentrations. Growth was monitored using a spectrophotometer (OD₆₀₀) at indicated time points and the results were expressed as percent growth rate for each strain compared to the untreated control group. To assess growth on solid medium, 5 μL of ten-fold diluted yeast cells were spotted onto YPD agar containing DMSO or auranofin (6.25 μg/mL). Growth of yeast strains was monitored after incubating the plates for 48 h, as described elsewhere (Gamberi et al., 2015).

Mitochondria were purified from yeast cells grown on YPEG as described previously (Hasson et al., 2010). Oxygen consumption measurements with isolated mitochondria were performed using an oxygen electrode (Hansatec) as described previously (Dabir et al., 2013). Membrane potential measurements of purified mitochondria were performed with fluorescent 3, 3′-dipropylthiadicarbocyanine iodide dye [DiSC3(5)]. 1% DMSO, carbonyl cyanide m-chlorophenyl hydrazone (CCCP), MB-6, or MB-7 was added to mitochondria in import buffer (0.6 M sorbitol, 2 mM KH₂PO₄, 60 mM KCl, 50 mM HEPES-KOH, 5 mM MgCl₂, 2.5 mM EDTA, 5 mM L-methionine, pH 7.1) for 10 min. Subsequently 0.2 μM DiSC3(5) in import buffer was added, incubated for 5 min, and fluorescence was measured at excitation and emission length of 620 and 670 nm, respectively.

Mitochondria were purified from wild-type yeast or yeast overexpressing Erv1 with a hexahistidine tag ([a2up] Erv1) grown in YPEG as described previously (Glick and Pon, 1995; Dabir et al., 2007). Yeast cultures were kept at 25°C with vigorous shaking during growth. Mitochondria concentration was measured by BCA assay and stored at 25 mg/mL at −80°C. Mitochondria with increased levels of Erv1 were purified from a strain in which Erv1 was overexpressed from a 2-micron plasmid (Dabir et al., 2007).

Prior to import into purified mitochondria, [³⁵S]-methionine and cysteine labeled proteins were generated with TNT Quick Coupled Transcription/Translation kits (Promega) and plasmids carrying the genes of interest. Transcription of genes was driven by either a T7 or SP6 promoter. Import reactions were conducted as previously described (Hasson et al., 2010; Dabir et al., 2013). After frozen mitochondria aliquots were thawed and added to the import buffer at a final concentration of 100 μg/mL, 1% DMSO or the small molecule was added as indicated. A final concentration of 1% DMSO was used in all experiments. Following incubation at 25°C for 15 min, import reactions were initiated by the addition of 5–10 μl of translation mix. Aliquots were removed at intervals during the reaction time course and import was terminated with addition either of cold buffer or 25 μg/mL trypsin, or the combination. If trypsin was added to digest non-imported precursor protein, soybean trypsin inhibitor was subsequently added in excess after 15 min incubation on ice. After a final recovery of by centrifugation (12,000 × g, 6 min), mitochondria were disrupted in Laemmli sample buffer. Samples from import reaction time points were resolved by SDS-PAGE and visualized by autoradiography. For experiments to investigate the Cmc1-Mia40 intermediate, non-reducing conditions were used. The import reactions were stopped in the presence of 20 mM iodoacetamide and mitochondria disrupted in Laemmli sample buffer lacking β-mercaptoethanol. The imported products were separated by non-reducing SDS-PAGE.

L4-stage worms of C. elegans AU37 (sek-1; glp-4) strain (glp-4(bn2) were used to examine the antifungal efficacy of auranofin as described elsewhere (Mylonakis et al., 2002; Thangamani et al., 2015d). Briefly, L4-stage worms were infected with C. neoformans NR-41292 for 2 h at room temperature. After infection, worms were washed with M9 buffer and treated either with DMSO or drugs (auranofin, fluconazole, and flucytosine), at a concentration of 8 μg/mL. After 24 h, worms were washed with PBS and disrupted using silicon carbide particles (Thangamani et al., 2015d). The final suspensions were plated onto YPD agar plates containing ampicillin (100 μg/mL), streptomycin (100 μg/mL), and kanamycin (45 μg/mL) to determine the colony forming unit (CFU) per worm (Li et al., 2013).

Statistical analyses were done using GraphPad Prism 6.0 (GraphPad Software, La Jolla, CA). P values were calculated via the Student t-test and P-values of ≤0.05 were deemed significant.

The antifungal activity of auranofin was tested against various clinical isolates of Candida and Cryptococcus. Auranofin was very active in inhibiting the growth of all strains of C. albicans, C. glabrata, C. tropicalis, and C. parapsilosis with the MIC ranging from 1 to 16 μg/ml (Table 1). Auranofin also displayed potent activity against both C. neoformans and Cryptococcus gattii inhibiting growth of these fungal species at a concentration ranging from 0.5 to 8 μg/ml (Table 1).

A time-kill assay was employed to investigate the killing kinetics of auranofin against both C. albicans and C. neoformans. Similar to fluconazole, auranofin (at the 5 × MIC) exhibited fungistatic activity against C. albicans ATCC 10231 and C. neoformans NR-41296 (Figure 1). However, at the 10 × MIC, auranofin and fluconazole exhibited fungistatic activity against C. albicans ATCC 10231, whereas unlike fluconazole, auranofin (at the 10 × MIC) completely kills C. neoformans NR-41296 after 48 h of incubation (Figure 1).

The antibiofilm activity of auranofin, against C. albicans, was evaluated using the XTT reduction assay in order to measure the metabolic activity of fungal cells post-treatment. The metabolic activity of C. albicans was reduced by more than 70% with the treatment of auranofin at 8 × MIC (Figure 2A). However, the control antifungals fluconazole and flucytosine were ineffective (less than 10% reduction observed) at reducing metabolic activity of C. albicans biofilm, even at a concertation equivalent to 32 × MIC when compared to the DMSO-treated control groups (Figure 2A).

The effect of auranofin on reducing fungal biofilm density was further evaluated using confocal microscopy. Fungal cells stained with ConA conjugated with FITC revealed that auranofin (8 × MIC) eradicates a considerable portion of Candida cells in comparison to the control group (Figure 2B). However, treatment with fluconazole and flucytosine, even at 32 × MIC, appear similar to control group (Figure 2B). These results correlate with the results of the XTT reduction assay.

To investigate the MOA, we subjected auranofin to chemogenomic profiling using the S. cerevisiae heterozygous deletion collection. Haploinsufficiency profiling (HIP) allows for the simultaneous assessment of the sensitivity of the pooled genome-wide set of heterozygous deletion strains because each strain is uniquely identified with a synthetic DNA barcode. The method is an unbiased approach to survey the genome-wide strain set in order to identify the strains with the most sensitivity to auranofin. We first identified the concentration that reduced wild-type growth by 30% and used 75 μM to profile the pooled heterozygous strains in biological samples. PCR was used to amplify the unique UPTAG DNA barcodes located at the gene deletion site and we tracked the barcode abundance with Illumina sequencing. The resulting counts were normalized and visualized using EdgeR (Figure 3A). We identified 85 heterozygous deletion strains that were under-represented based on an FDR ≤ 0.1 when comparing auranofin treatment to DMSO. These 85 strains were analyzed to identify associated gene ontology cellular component annotations and found to be enriched in several categories including the mitochondrial intermembrane space and chromatin components (Supplementary data excel file). Five heterozygous deletion strains within these enriched categories (mia40Δ, acn9Δ, coa4Δ, rad18Δ, and nsi1Δ) were selected to validate sensitivity to auranofin using a variety of growth assays (Figure 3A). These strains were randomly selected on the basis that they represented genes in the highest significance gene ontology categories including “regulation of translational initiation” (p-value = 0.00033) and “mitochondrial intermembrane space” (p-value = 0.0072) (Supplementary data excel file). In addition, as outlined in the following section, the mia40Δ strain was tested because it was implicated as sensitive to auranofin in a previous study (Lee et al., 2014).

Growth of these five heterozygous deletion strains and the wild-type (BY4743) strain were monitored in the presence of different concentrations of auranofin (6.25, 12.5, and 25 μg/mL) in a liquid growth assay. The result indicated that only three heterozygous deletion strains (mia40Δ, acn9Δ, and coa4Δ) exhibited drug-induced haploinsufficiency under these conditions. The growth of these deletion strains was suppressed, even in the presence of low concentrations of auranofin (6.25 μg/mL) (Figure 3B). However, auranofin does not induce haploinsufficiency in the other two deletion strains (rad18Δ and nsi1Δ) as growth of these strains, in the presence of auranofin, mimics the pattern observed with the wild-type strain (Figure 3B). These two deletion strains were not affected possibly because of the concentration used in our validation studies or because they may be false positives. For each strain, the growth of cells (OD₆₀₀ after 24 h) incubated with auranofin (6.25 μg/mL) was determined in relation to DMSO treatment. The growth of three heterozygous deletion strains (mia40Δ, acn9Δ, and coa4Δ) was drastically suppressed by more than 50% in the presence of auranofin (6.25 μg/mL). However, the remaining two deletion strains (rad18Δ and nsi1Δ) had a modest reduction in growth of~10% compared to the wild-type strain (Figure 3C).

The growth of these five deletion strains was further confirmed by spotting serial dilutions of cultures on solid agar. As shown in Figure 3D, growth of the wild-type and five heterozygous deletion strains was normal in agar containing DMSO. However, the heterozygous deletion strain, mia40Δ, exhibited a nearly two-fold reduction in colony forming units when spotted onto YPD agar containing auranofin (6.25 μg/mL).

A study conducted by Lee et al. (2014) previously analyzed a heterozygous deletion pool, representing essential genes, using haploinsufficiency profiling with hundreds of compounds including auranofin. Examination of theauranofin-generated data set shows that they identified 17 strains as possibly sensitive. Comparison of Lee et al.'s results with our 85 strains showed that two strains, rho1Δ and mia40Δ, overlapped between the data sets (Figure 3E). An additional study by Gamberi et al. (2015) specifically assessed sensitivity and resistance of haploid deletion strains involved in mitochondrial function and found them to be differentially effects to auranofin compared to the haploid wild type parental strain. Based on studies by Gamberi et al. (2015) and Lee et al. (2014), we next moved to examine sensitivity of the corresponding heterozygous deletion strains involved in mitochondrial function and redox homeostasis that are possibly sensitive to auranofin.

Heterozygous deletion strains with genes deleted in mitochondrial function and redox homeostasis experienced a significant growth reduction when treated with auranofin (6.25 μg/mL) relative to DMSO-treated cells (Figure 4A–green and gray bars), including ndilΔ, atp2Δ, citlΔ, sdh4Δ, gsh1Δ, gsh2Δ, prx1Δ, trr1Δ, erv1Δ, toa2Δ, arp7Δ, ydl63wΔ, and yjl086cΔ. These results are in agreement with Gamberi's et al. (2015) results in that mitochondrial function appears to be a target of auranofin. It should be noted that Gamberi et al. used haploid deletion strains sensitivity, which generally does not inform on the direct target of a compound as opposed to the heterozygous deletion strains used in our study. In addition, Gamberi et al. examined resistance of haploid deletion strains and varied media conditions, resulting in their conclusion that Pos5 was the target—however, several other haploid deletions strains also demonstrated slight resistance to auranofin and heterozygous strain sensitivity was not examined. Our results demonstrate that the heterozygous pos5Δ strain is not sensitive to auranofin in liquid or agar conditions suggesting Pos5 is unlikely to be the direct target of auranofin (Figures 4A,B). The heterozygous strains that were identified by Lee et al. (2014) were generally not as sensitive in liquid growth compared to mitochondrial or redox-related strains (Figure 4A—brown bars). Because this set of strains contained deletions of genes with a wide variety of functions it is likely that they not specific hits from the screen and shows the importance of testing individual strains to confirm sensitivity to auranofin. These results were confirmed using the YPD agar-spotting assay in that many strains were not sensitive. Interestingly, heterozygous deletion strains involved in ROS response and redox homeostasis (sdh4Δ, gsh1Δ, gsh2Δ, and prx1Δ) which had significant growth reduction in liquid medium did not demonstrate considerable reduction in growth when spotted onto YPD agar containing auranofin (6.25 μg/mL) (Figure 4B).

As noted earlier, heterozygous deletion strains that encode genes required for mitochondrial function (including ndilΔ, atp2Δ, citlΔ, and erv1Δ), showed a considerable decrease in colony count (almost one-fold log reduction) when spotted onto YPD agar containing auranofin (6.25 μg/mL) (Figure 4B). Interestingly, a deletion strain (erv1Δ), which forms a complex with Mia40 (Rissler et al., 2005), showed considerable sensitivity to auranofin, which coincides with Lee et al.'s findings (Figure 4B). Taken altogether, our results as well as the overlap with the previous haploinsufficiency profiling (Lee et al., 2014), supports the notion that Mia40/Erv1 is the probable antifungal target of auranofin.

To further confirm the specific inhibition of the Mia40-Erv1 pathway by auranofin we employed several biochemical experiments using purified yeast mitochondria similar to a previous study that investigated the effect of several small molecule inhibitors of redox-regulated protein import into mitochondria (Dabir et al., 2013). A possible indirect mechanism of inhibition of mitochondrial function and the Mia40-Erv1 pathway is by the disruption of membrane potential or diminished oxidative phosphorylation. Maintenance of membrane potential was determined by mitochondrial uptake of DiSC3 (5) dye and subsequent quenching in the presence of membrane potential. Auranofin had no effect on the membrane potential compared to DMSO whereas the uncoupling agent, CCCP, caused a 4-fold increase in fluorescence indicating uncoupled mitochondria (Figure 5A). The effect on mitochondrial respiration was determined by measuring dissolved oxygen in a chamber with purified mitochondria and respiration was initiated with NADH resulting in an oxygen consumption rate (−0.45 O₂ nmol/s) consistent with well-coupled mitochondria. The addition of DMSO did not increase respiration rate and auranofin at a concentration of 34 μg/mL only slightly increased the respiration rate (−0.64 O₂ nmol/s) (Figure 5B and Table S2). As a control, the addition of CCCP resulted in a severe increase in consumption rate (−1.15 O₂ nmol/s) suggestive of uncoupled mitochondria (Figure 5B).

To confirm that auranofin targets the Erv1/Mia40 pathway we measured the effect of compound on the import of mitochondrial protein substrates compared to control compounds previously identified as Erv1 inhibitors (Dabir et al., 2013). Radiolabeled precursor proteins were incubated with mitochondria in the presence of small molecules or DMSO and the reaction was terminated with protease and subsequently analyzed by gel electrophoresis. Protein substrates from different import pathways were assessed including the Tim23 substrate, Su9-DHFR, and the Mia40 substrate, Cmc1. Auranofin at a lower concentration of 6.8 μg/mL inhibits import of Su-DHFR to a 60% level and Cmc1 to a 25% level compared to untreated samples (Figures 5C,D). These results indicate the preferential activity of auranofin toward inhibiting Cmc1 import compared to Su9-DHFR, which is expected because Cmc1 is directly imported by Mia40/Erv1. Strikingly, auranofin exhibits more potent activity than control compound, MB-7 with a drastic difference in import efficiency observed between the compounds at 10 μM (6.8 μg/mL for auranofin and 8.5 μg/mL for MB-7; Figure 5D). Although auranofin does inhibit Su9-DHFR import at high concentrations, these results demonstrate the compound has specificity toward the Mia40 pathway and increased potency compared to previously identified inhibitors from a large-scale chemical library screen (Dabir et al., 2013). It is not surprising that the import of Su9-DHFR is mildly inhibited because mitochondrial import pathways are interconnected.

Mia40 has previously been demonstrated to form an intermediate with Cmc1 as part of the import process (Bourens et al., 2012; Neal et al., 2015). The effect of compounds on the formation of a disulfide intermediate between Mia40 and Cmc1 was monitored in organello. Auranofin inhibits radiolabeled Cmc1 from interacting with Mia40 in a similar dose dependent manner to MB-7 (Figure 5E). The addition of another control, MB-6, causes the accumulation of the intermediate. In sum, auranofin inhibits the heterodimer formation of the Mia40-Cmc1 intermediate and is a potent inhibitor of the Mia40 import pathway.

To further validate the Mia40 pathway as a target of auranofin, import of Cmc1 was performed with mitochondria from WT and Erv1 overexpressing yeast. Erv1 overexpression is expected to maintain the Mia40 pool in an oxidized state, which is required for the interaction with substrate proteins (Mesecke et al., 2005; Dabir et al., 2007) and hence should be more resistant to auranofin inhibition. As predicted, the Erv1 overexpressing mitochondria were resistant to auranofin (3.4 μg/mL) treatment as evidenced by the increased level of Cmc1 (60%) import compared to WT (30%) mitochondria providing further confirmation of Mia40 as a target (Figure 5F).

To investigate if the in vitro antifungal activity of auranofin translates in vivo, the antifungal efficacy of auranofin was examined in a C. neoformans-infected C. elegans animal model. As shown in Figure 6, treatment of infected C. elegans with fluconazole, flucytosine, and auranofin, at 8 μg/mL, produced a significant reduction (P ≤ 0.01) in mean fungal load when compared to the untreated control groups. Strikingly, C. elegans treated with auranofin (8 μg/mL) generated the largest reduction in C. neoformans CFU (0.87 ± 0.03 log₁₀), followed by fluconazole (8 μg/mL) (0.82 ± 0.03 log₁₀) and flucytosine (8 μg/mL) (0.58 ± 0.11 log₁₀) (Figure 6).

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.