All N. crassa strains used in this study are listed in Table 1. N. crassa single-gene deletion mutants were purchased from Fungal Genetic Stock Center (FGSC, University of Kansas Medical Center, Lawrence, KS, USA). Double mutants Δp23Δsti1 and Δp23Δaha1 were generated by crossing FGSC#01792 (Δp23) and FGSC#00714 (Δsti1), and FGSC#01792 (Δp23) and FGSC#04087 (Δaha1), respectively, with the method described before (Zhang et al., 2012). Vogel's minimum medium, supplemented with 2% (w/v) sucrose for slants or 2% (w/v) glucose for plates, was used for culturing N. crassa strains. The slants were incubated at 28°C in the dark for 2 days and then in light for 5 days for conidiation. Antifungal compounds were added when needed. Fusarium verticillioides wild type M-3125 (mating type: MAT-1) and knockout mutants were cultured on potato dextrose agar at 28°C in the dark.

Ketoconazole (KTC), fluconazole (FLU), itraconazole (ITC), amphotericin B (AMB), caspofungin (CSP), and geldanamycin (GA) were dissolved in DMSO. The final DMSO concentrations in media were below 0.25% (v/v; Sun et al., 2013). Menadione and H₂O₂ were diluted in distilled water. The agents were aseptically added to media before plates were made. The final concentrations of KTC, FLU, ITC, AMB, CSP, GA, menadione, and H₂O₂ in the media were 2, 25, 10, 0.05, 1, 1, 2 mg/L, and 2 mM, respectively. Two microliters of conidial suspension (2 × 10⁶ spores/ml) were inoculated on plates (Φ 90 mm, 12 mL medium each) with or without drugs and the plates were incubated at 28°C in the dark. Heat sensitivity tests were carried out at 28 and 42°C, respectively. Each test was duplicated and was repeated at least three times.

The MICs of KTC and FLU for each strain were determined in 96-well microtiter plates according to previous methods (Zhang et al., 2012). Briefly, 100 μl of 2 × KTC or FLU solution and 100 μl of conidial suspension media were added to each well. The final KTC concentrations were 0, 0.25, 0.5, 1, 2, 3, 4, 5, and 6 μg/ml; the final FLU concentrations were 0, 5, 10, 20, 30, 40, 50, and 60 μg/ml. The final conidial concentration was approximately 1 × 10⁵ conidia/ml. The plates were incubated at 28°C for 24 h. The minimum inhibitory concentration (MIC) values were determined as the lowest KTC concentration that inhibited growth observed by naked eye and dissecting microscope.

All primers used in this study are listed in Table S1. To create a complementary p23 plasmid, the full-length p23 coding sequence (947 bp) with 1733 bp upstream and 935 bp downstream was amplified using the primers Nc-p23-com-1(ClaI) and Nc-p23-com-2(EcoRV), yielding a 3734 bp complementary fragment. The PCR product was inserted into the plasmid pCB1532 (Sweigard et al., 1997), which harbors a sulfonylurea resistant allele of Magnaporthe grisea ILV1 as a selective marker, at ClaI and EcoRV sites to create pCB1532-p23. For sti1, a 4174 bp fragment, consisting of 1567 bp upstream, 1868 bp coding sequence, and 723 bp downstream of sti1, was created by PCR using the primers Nc-sti1-com-1(XbaI) and Nc-sti1-com-2(BamHI). The fragment was then inserted into pCB1532 at BamHI and XbaI sites to create pCB1532-sti1. Similarly, a 3707 bp PCR fragment, including 1771 bp upstream, 1189 bp coding sequence, and 747 bp downstream of aha1, was amplified with the primers Nc-aha1-com-1 and Nc-aha1-com-2, and then ligated with linearized pCB1532, by EcoRV digestion, to generate pCB1532-aha1. The constructs pCB1532-p23, pCB1532-sti1 and pCB1532-aha1 were transformed into Δp23, Δsti1, and Δaha1 mutants, respectively, using the method described previously (Sun et al., 2012) Transformants were screened on Vogel's medium with 20 μg/ml chlorimuron ethyl and verified by PCR.

The culture conditions for RNA extraction were the same as previously described by Sun et al. (2012). Briefly, wild type and mutant conidia were separately inoculated into 20 mL liquid medium in a plate (Φ 90 mm) and incubated for 24 h at 28°C in the dark to form mycelia mats on the surface of the liquid medium. The mycelia mats were then cut into small pieces (Φ 10 mm), which were then torn into fragments (< 5 mm²) and transferred to 100 mL liquid media in 150 mL Erlenmeyer flasks. Cultures were incubated at 28°C with shaking at 180 rpm for 12 h. KTC was then added into the medium at a final concentration of 2.5 mg/L, as required. After 24 h incubation, mycelia were harvested and total RNA was extracted and reverse transcribed using FastQuant RT Kit (with gDNase) (Tiangen Biotech Co. Ltd., Beijing, China). The synthesized cDNA (2 μg in 20 μl) was diluted to 200 μl, and quantitative PCR was carried out on a CFX-96 Multicolor Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) with SYBR-Green detection (SYBR Master Mix, TOYOBA Biotechnology Co., Ltd., Osaka, Japan); 0.8 μl each forward and reverse primers, 6 μl diluted cDNA, and 10 μl SYBR master mix were mixed in a 20 μl reaction, obtaining a 3 ng/μl final cDNA concentration. Each cDNA sample was analyzed in duplicate and repeated at least three times, and the average threshold cycle was calculated. Relative expression levels were calculated using the 2^−ΔΔCt method (Livak and Schmittgen, 2001). The results were normalized to the level of β-tubulin.

Sterol extraction and liquid chromatography–mass spectrometry (LC-MS) analysis were performed as described by Sun et al. (2013). Briefly, 0.1 g dried mycelia powder with 25 μg fluconazole was extracted in 1.7 ml chloroform for 12 h with 10 min ultrasonic treatments before and after extraction. The extracts were dried and dissolved in 300 μl methanol under ultrasonication for 30 min. The extracts were subjected to LC-MS analysis as previously described (Sun et al., 2013) after filtered with 0.22 μm Millipore filters. The derived sterols were identified with reference molecular weight and fragmentation spectra for known standards. The amounts of sterols were normalized by relative peak area with fluconazole as reference.

The homologs of Hsp90 co-chaperones were identified in the National Center for Biotechnology Information (NCBI) by protein-protein BLAST (http://www.ncbi.nlm.nih.gov/BLAST) from the following species: N. crassa (taxid: 5141), F. oxysporum (taxid: 5507), A. fumigatus (taxid: 746128), C. albicans (taxid: 5476), S. cerevisiae (taxid: 4932), C. neoformans (taxid: 5207), and Homo sapiens (taxid: 9606). To reveal the sequence conservation and phylogenic relationship of homologs of Hsp90 and its three co-chaperones in fungi and human, the peptide sequences were aligned with DNAman software, and the results of multiple alignment were shown by phylogenetic trees generated with the Neighbor-Joining method (bootstrap = 1000) under support of MEGA 7.0 (Kumar et al., 2016).

To knockout the sti1 and p23 homologs (sti1: FVEG_00423, the E-value, query coverage and identity with N. crassa sti1 are 8e⁻⁷², 62 and 72%, respectively; p23: FVEG_11505, the E-value, query coverage and identity with N. crassa p23 are 0, 100 and 76%, respectively) in F. verticillioides, the upstream and downstream flanking sequences were amplified and ligated to the hygromycin phosphotransferase gene by fusion PCR. Then, the deletion cassettes were transformed into F. verticillioides wild type M-3125, producing deletion mutants. Fungal transformation followed the protocol described by Sun et al. (2013). Transformants were checked using the primers Fv-sti1-check-1/2 and Fv-p23-check-1/2, respectively.

To investigate the roles of the Hsp90 co-chaperones in azole-induced stress responses, we tested the susceptibility of 24 single-gene deletion mutants or heterokaryons to 2 mg/L ketoconazole (KTC; Figure 1, Table 1). These genes were presumed to be involved in the Hsp90 cycle and calcineurin pathway. Without KTC, growth rates of all tested mutants were similar to that of wild type. On plates with 2 mg/L KTC, three mutants displayed much severer growth inhibition than the rest of mutants and wild type. The corresponding genes of these three mutants encode Hsp90 co-chaperones P23, Aha1, and Sti1, respectively (Figures 1, 2A, Table 2). Heterokaryons deficient for hsp90 (the gene name is hsp80), or cna1 (calcineurin catalytic subunit) or cnb1 (calcineurin regulatory subunit), showed increased susceptibility to KTC. Deletion of hsf1, the gene coding a heat shock transcriptional factor, also increased susceptibility to KTC (Figure 1, Table 2). However, the mutants lacking either PPIases and heterokaryons deficient for Hsp70 showed wild-type KTC susceptibility (Figure 1, Table 1).

Based on the above results, Hsp90 co-chaperones P23, Aha1, and Sti1 were chosen for further study. Complementation of Δsti1, Δaha1, and Δp23 recovered wild-type susceptibility to KTC (Figure 2B). The susceptibilities of mutants (Δp23, Δaha1, and Δsti1) to fluconazole and itraconazole were also tested. The results obtained from fluconazole treatment corresponded to the results in KTC. On plates with 25 mg/L fluconazole, the growth of wild type was slower than that on the plates without drug; the growth of Δp23 was completed arrested; Δaha1 and Δsti1 formed only tiny colonies after 50 h incubation (Figure 2A). On plates with 10 mg/L itraconazole, the growth rates of these mutants and wild type were not dramatically different (Figure 2A).

To further understand the relationship among these three co-chaperones under azole stress, double mutants Δp23Δsti1 and Δp23Δaha1 were generated. Interestingly, these double mutants displayed greater growth inhibition than the respective single gene deletion mutants on plates with itraconazole. On the plates with 10 mg/L itraconazole, the growth of Δp23Δsti1 was completely arrested and Δp23Δaha1 could only form tiny colonies, which were significantly smaller than those of Δp23 or Δaha1 (Figure 2A, Table 2). Since the susceptibility difference to KTC and fluconazole between each of double mutants and its respective single gene deletion mutants was difficult to show based on colony growth, MIC analysis was further conducted. Results showed that the double mutants Δp23Δsti1 and Δp23Δaha1 had MIC values to fluconazole significantly lower than their respective single-mutation lines (Table 3). For KTC, MIC values of these double mutants and their respective single gene mutant lines were not significantly different (Table 3). Since the complete suppression of single gene mutants' growth requires very low concentration of KTC (0.7–1.7 μg/ml), accurate measurement of differences in KTC susceptibility between each double mutant and the single gene mutant lines is difficult. Nevertheless, based on the results in itraconazole and fluconazole, it could be concluded that each of these Hsp90 co-chaperones has its independent role in the basal azole resistance.

To further reveal the relationship between Hsp90 and its co-chaperones, the sensitivity of N. crassa strains to the Hsp90 inhibitor geldanamycin (GA) was tested. As shown in Figure 3, 2 mg/L GA affected N. crassa growth, but its growth effects in Δp23 and Δaha1 were similar to that of wild type. For Δsti1, although its growth was only partially repressed, its growth was significantly slower than wild type. Thus, the simultaneous disruption of Hsp90 and its co-chaperones did not dramatically affect the hyphal growth on plates without antifungal azoles. Addition of 1 mg/L KTC slightly affected the growth of wild type but completely suppressed the growth of Δp23. Although growing slower than wild type, Δaha1 and Δsti1 could form colonies on plates with 1 mg/L KTC after 42 h. On plates with both 2 mg/L GA and 1 mg/L KTC, the growth of all mutants was arrested. None of them formed visible colony even after 120 h incubation. Wild type could form colonies, but its growth was significantly slower than that on plates with 2 mg/L GA or 1 mg/L KTC alone. The above results suggest that Hsp90 and its co-chaperones P23, Aha1, and Sti1 play more important roles under azole stress than under normal conditions.

In addition to azoles, polyenes, and echinocandins are also widely used for treating fungal infections. To test whether Sti1, Aha1, and P23 are also required for the basal resistance to polyenes and echinocandins, susceptibilities of their mutants to amphotericin B and caspofungin were tested. As shown in Figure 2A, compared with wild type, neither single gene deletion mutants, including Δsti1, Δaha1, and Δp23, nor double mutants, including Δp23Δsti1 and Δp23Δaha1, displayed increased susceptibility to amphotericin B or caspofungin. These results indicate that Sti1, Aha1, and P23 are not generally required for the basal resistance to a wide range of antifungal drugs.

Hsp90 prevents ROS generation and stimulates anti-oxidative defenses under thermal stress conditions in the fungal species Paracoccidioides brasiliensis (Matos et al., 2013). We tested the roles of Hsp90 and its related genes under heat and oxidative stress. As shown in Figure 1, the growth rates of Δhsp80 and all tested mutants related to Hsp90 grew as normal as wild type at 28°C. At 42°C, Δp23, Δaha1, and Δsti1, grew significantly slower than wild type (Figure 2A, Table 2). At 42°C, the growth of wild type was inhibited by 44.5% relative to its growth at 28°C, while the growth of Δsti1, Δaha1, and Δp23 was inhibited by 75, 55.5, and 70.9%, respectively, compared to their growth at 28°C. These results indicate that these Hsp90 co-chaperones also participate in stress responses induced by heat.

H₂O₂ and menadione were used to test the roles of Hsp90 co-chaperones under oxidative stress. On plates with 2.5 mM H₂O₂, the growth of wild type was inhibited by 51.3%, relative to its growth on control plates (Table 2), while the growth of Δhsp80 and Δcnb1 were completely arrested (Table 2), which further demonstrated that Hsp90/calcineurin system had critical function under oxidative stress.In contrast, Δsti1 and Δp23 were more resistant than wild type. On plates with 2.5 mM H₂O₂, the growth of Δsti1 and Δp23 were inhibited by only 26.3 and 28.7%, respectively, which were significantly lower than that of wild type (by 51.3%; Table 2). Δaha1 displayed wild-type sensitivity to H₂O_2. On plates with 2 mg/L menadione, the growth of Δp23, Δaha1, and Δsti1 were comparable to wild type. Thus, these Hsp90 co-chaperones do not positively contribute to the basal resistance to oxidative stress as they do to azole stress. In contrast, Sti1 and P23 negatively regulate H₂O₂ resistance.

Hsp90 governs the stress response network in fungi, and its co-chaperones are essential for the Hsp90 cycle (Walton-Diaz et al., 2013). To clarify how the co-chaperones affect the Hsp90 cycle at transcriptional level in N. crassa, we analyzed the expression of NCU04142 (hsp80, the Hsp90 coding gene), NCU00714 (sti1), NCU04087 (aha1), NCU01792 (p23), and NCU08512 (hsf1) using quantitative RT-PCR. In wild type N. crassa, KTC treatment did not affect the transcription of these genes (Data not shown), which is consistent with previous RNA-seq data (Sun et al., 2014). However, hsp80 mRNA levels were more than 2-fold higher (2.7- and 2.3-fold, respectively) in both non-KTC and KTC treatments in Δsti1 than those of wild type; deletion of p23 also caused elevated transcription of hsp80 in non-KTC treatment (Figure 4). The increase of hsp80 expression may be a compensatory effect of the loss of the co-chaperones, due to the Hsp90 auto-regulatory loop control (Leach et al., 2012).

Azoles target ERG11, directly blocking 14α-demethylation of lanosterol, resulting in sterol biosynthesis through the eburicol bypass and an accumulation of the toxic intermediate 14α-methyl-3,6-diol, which can damage membrane integrity (Kelly et al., 1995). The efflux pump CDR4 has been shown to be the major contributor to azole resistance among four Pdr5p-like ABC transporters in N. crassa (Zhang et al., 2012). We analyzed the expression of the key genes in ergosterol synthesis, including NCU02624 (erg11) and NCU03006 (erg6), as well as the efflux pump gene NCU05591 (cdr4), in their mutants and wild type. Quantitative RT-PCR revealed that in wild type, erg11, erg6, and cdr4 were up-regulated by 6.0-, 7.6-, and 9.0-fold after 24 h of KTC (2.5 mg/L) treatment, respectively. These results are consistent with previous RNA-seq data (Sun et al., 2013). In contrast, the level of erg11 in Δsti1, Δaha1, and Δp23 mutants was increased only 1.5-, 4.5-, and 1.7-fold compared with non-KTC treated wild type, respectively, and the levels of erg6 in Δsti1 and Δp23 mutants were increased only 2.7- and 1.6-fold compared with non-KTC treated wild type, respectively. In Δaha1, erg6 levels were reduced by 0.33-fold without KTC treatment and increased 4.4-fold with KTC treatment compared with non-KTC treated wild type. The mRNA levels of cdr4 in three mutant strains were similar to that of wild type in the medium without KTC. After KTC treatment, cdr4 transcripts in mutants and wild type were elevated to a similar level (Figure 3). These data suggest that Sti1, Aha1, and P23 are required for the adaptive responses by genes involved in ergosterol synthesis under azole stress.

Previous studies have shown that the mechanism of action of antifungal azoles includes decreasing ergosterol levels and/or the accumulation of 14α-methyl-3,6-diol (Kelly et al., 1995; Sun et al., 2014). Therefore, we tested the sterol composition in the mutants and wild type after 24 h culture with or without KTC (2.5 mg/L) using LC-MS with fluconazole as a reference to normalize the total amount of sterols. In wild type, ergosterol was the primary sterol (2.9, relative amount) and small amount of eburicol (0.2) was detected in untreated samples; when KTC was added, ergosterol levels were slightly reduced (2.0) but eburicol significantly increased (1.6) and 14α-methyl-3,6-diol was detected (0.4). The untreated mutants accumulated ergosterol as the primary sterol, similar to wild type. Following KTC treatment, ergosterol was reduced in Δp23, Δaha1, and Δsti1 (2.6/3.5, 2.9/4.0, and 2.8/4.1, respectively, KTC treatment/untreated); ergosterol was reduced more in the double mutants Δp23Δsti1 and Δp23Δaha1 (2.3/4.1 and 2.6/4.4, respectively). Meanwhile, eburicol and 14α-methyl-3,6-diol were sharply increased in these mutants following KTC treatment: Δp23: 2.3/1.2; Δaha1: 1.4/0.6; Δsti1: 2.0/0.9; Δp23Δsti1: 3.0/1.4; Δp23Δaha1: 2.5/0.9 (eburicol/14α-methyl-3,6-diol; Figure 5). Thus, the excessive accumulation of toxic 14α-methyl-3,6-diol might be an important cause to the reduced azole resistance in these mutants.

Multiple sequence alignment revealed that the homologs of Hsp90 from seven fungal species, including N. crassa, S. cerevisiae, C. albicans, F. verticillioides, F. oxysporum, A. fumigatus, and C. neoformans, and H. sapiens, are highly conserved in amino acid sequences (Figure S1). However, the phylogenetic analysis shows that Hsp90s from fungi are distantly related to Hsp90s from H. sapiens: fungal Hsp90s were clustered to one big clade while human Hsp90s were clustered to another big clade (Figure S1). The similar results were obtained when homologs of Sti1, Aha1, and P23 were analyzed (Figures S2–S4).

To test the functional conservation of Sti1 and P23 among different fungi under azole stress, the gene deletion mutants for the sti1 (FVEG_00423) homolog and the p23 (FVEG_11505) homolog in plant pathogen F. verticillioides, were respectively created. As shown in Figure 6, on the medium without KTC, the growth rates of sti1 and p23 deletion mutants were similar to that of wild type. When grown on the medium with 2 mg/L KTC, both sti1 and p23 deletion mutants displayed increased sensitivity compared with wild type, indicating that Sti1 and P23 homologs also contributes to the basal azole resistance in F. verticillioides and their roles in azole adaptation should be conserved among fungi.

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.