Jerveratrum-Type Steroidal Alkaloids Inhibit β-1,6-Glucan Biosynthesis in Fungal Cell Walls

ABSTRACT The limited number of available effective agents necessitates the development of new antifungals. We report that jervine, a jerveratrum-type steroidal alkaloid isolated from Veratrum californicum, has antifungal activity. Phenotypic comparisons of cell wall mutants, K1 killer toxin susceptibility testing, and quantification of cell wall components revealed that β-1,6-glucan biosynthesis was significantly inhibited by jervine. Temperature-sensitive mutants defective in essential genes involved in β-1,6-glucan biosynthesis, including BIG1, KEG1, KRE5, KRE9, and ROT1, were hypersensitive to jervine. In contrast, point mutations in KRE6 or its paralog SKN1 produced jervine resistance, suggesting that jervine targets Kre6 and Skn1. Jervine exhibited broad-spectrum antifungal activity and was effective against human-pathogenic fungi, including Candida parapsilosis and Candida krusei. It was also effective against phytopathogenic fungi, including Botrytis cinerea and Puccinia recondita. Jervine exerted a synergistic effect with fluconazole. Therefore, jervine, a jerveratrum-type steroidal alkaloid used in pharmaceutical products, represents a new class of antifungals active against mycoses and plant-pathogenic fungi. IMPORTANCE Non-Candida albicans Candida species (NCAC) are on the rise as a cause of mycosis. Many antifungal drugs are less effective against NCAC, limiting the available therapeutic agents. Here, we report that jervine, a jerveratrum-type steroidal alkaloid, is effective against NCAC and phytopathogenic fungi. Jervine acts on Kre6 and Skn1, which are involved in β-1,6-glucan biosynthesis. The skeleton of jerveratrum-type steroidal alkaloids has been well studied, and more recently, their anticancer properties have been investigated. Therefore, jerveratrum-type alkaloids could potentially be applied as treatments for fungal infections and cancer.

and echinocandins. Plant diseases are caused by diverse phytopathogenic fungi and affect a wide range of crops, such as wheat, rice, pepper, rapeseed, potatoes, soybeans, and fruits (3,4), reducing crop yield and quality and resulting in enormous economic losses (3,5). The emergence of resistant strains is a major problem in both mycoses and plant diseases (6)(7)(8)(9). Therefore, fungal infections pose a serious threat to public health, and new and effective antifungal drugs are needed.
The budding yeast Saccharomyces cerevisiae has cell wall components similar to those of human-pathogenic fungi such as Candida and Aspergillus and constitutes a powerful model system for developing new antifungal agents. We previously used chemical reagents and S. cerevisiae to discover a novel antifungal agent derived from plant lignocellulose, named poacic acid (19). We profiled 13,524 compounds using genomic techniques and predicted that some compounds target cell wall biosynthesis and assembly. Further analysis of the morphology and wall components of chemicaltreated cells revealed that eight compounds affected the cell wall constituents (20). Four of these compounds (NP157, NP293, NP329, and NP413) are pseudojervines with a skeleton of jerveratrum-type steroidal alkaloids. Jervine was isolated from Veratrum californicum in 1943 (21), and it has been studied as an anticancer agent together with another jerveratrum-type steroidal alkaloid, cyclopamine (22)(23)(24)(25). The name "cyclopamine" is derived from "cyclops," and this alkaloid is a cause of the developmental defect cyclopia (26). Jervine affects yeast cell wall biosynthesis (20); however, it is not yet clear how this bioactive compound functions at the molecular level.
An analysis of cell wall components revealed that jervine inhibits the biosynthesis of b-1,6-glucan. Drug susceptibility tests using mutant strains affecting b-1,6-glucan biosynthesis showed that jervine acts on both Kre6 and Skn1. Also, we found that jervine was effective against human and phytopathogenic fungi. Thus, jerveratrum-type steroidal alkaloids represent a new class of antifungals.

RESULTS
Effect of jervine on the yeast cell wall. We predicted by chemical-genomic analysis that the intracellular targets of pseudojervines are involved in cell wall construction (20). Further phenotypic analysis indicated that jervine (Fig. 1A) and pseudojervines induce cell wall phenotypes (20). Therefore, we compared the chemical-genetic profiles of jervine and pseudojervines with those of other cell wall agents using a functional pool of 310 haploid gene deletion mutants. There was a significant correlation between the chemicalgenetic profile of jervine and those of the pseudojervines NP329 and NP293 (Pearson's correlation coefficient R . 0.8, P , 0.05 after Bonferroni correction, noncorrelation test) and between those of jervine and D75 (Pearson's correlation coefficient R = 0.93, P , 0.05 after Bonferroni correction, noncorrelation test) ( Fig. 1B and C). The correlations of jervine with micafungin, calcofluor white, and tunicamycin (TN) were not significant ( Fig. 1B and C). Thus, the chemical-genetic profile of jervine is most similar to the chemical-genetic profile of D75, a specific b-1,6-glucan biosynthesis inhibitor (18).
Next, we compared the morphological phenotypes of yeast cells treated with the cell wall agents after quantifying cell, actin, and nuclear morphology using CalMorph (27). Jervine had effects on yeast cell morphology similar to those of D75 (Pearson's rank correlation R = 0.54, P , 0.01 after Bonferroni correction, noncorrelation test) but not to those of EB, TN, or nikkomycin Z (NZ) ( Fig. 2A and B). Treatment with jervine or D75 resulted in slightly smaller cells with a wider neck ( Fig. 2A). Cells treated with jervine and EB showed different b-1,3-glucan staining patterns (Fig. 2C). EB-treated cells had weak b-1,3-glucan staining in buds due to decreased b-1,3-glucan biosynthesis. In contrast, jervine-treated cells had strong b-1,3-glucan staining in buds (P , 0.01 after Bonferroni correction, t test) (Fig. 2C), probably due to a compensatory mechanism. Like jervine, cyclopamine and D75 also showed strong b-1,3-glucan staining in buds, whereas TN and NZ had little effect on b-1,3-glucan staining (Fig. S1A in the supplemental material). Therefore, jervine has effects on yeast cells similar to the effects of D75 but different from those of other cell wall-targeting agents, such as EB, TN, and NZ. Chemical-genetic profiles of jervine, D75-4590, NP329, NP293, calcofluor white, and tunicamycin. Chemical-genetic screening was conducted through a nonessential deletion collection of S. cerevisiae mutants. Barcode sequencing data were quantified using BEAN-counter software. Hierarchical clustering was done on both compounds and genes and visualized using Java TreeView with the contrast level set to 5. Blue, negative chemical-genetic interactions, indicating hypersensitivity of the mutants to a compound. (C) Pearson correlation matrix showing the similarities between chemicalgenetic profiles. Pearson correlation coefficients (PCC) were calculated for the chemical-genetic profiles based only on negative chemical-genetic interactions. The PCC between two compounds is depicted by a square at the intersection of the compounds. Values closer to 1 are yellow, while those closer to 21 are blue. The chemical-genetic profiles of jervine, D75-4590, NP329, and NP293 are very similar (PCC . 0.8).
K1 killer toxin has a two-step mechanism of action. In step 1, K1 killer toxin binds to b-1,6-glucan in the cell wall. In step 2, the toxin disrupts membrane integrity, which leads to cell death (28,29). Yeast cells having walls with a high b-1,6-glucan content exhibit a large zone of growth inhibition by K1 killer toxin, whereas mutants that are defective in terms of b-1,6-glucan biosynthesis are often resistant to toxin-mediated cell death (30). The kre6D strain (with markedly reduced b-1,6-glucan content) had no visible growth inhibition zone (P , 0.01 after Bonferroni correction, t test) (Fig. 4A). The growth inhibition zone was significantly decreased on plates containing 5 and 10 mg/ml of jervine (P , 0.05 after Bonferroni correction, t test) (Fig. 4A). We also analyzed the incorporation of 14 C-labeled glucose into b-1,6-glucan, b-1,3-glucan, and chitin fractions after extraction by mild alkaline lysis and Zymolyase treatment (18). Jervine markedly reduced the radioactivity in only the b-1,6-glucan fraction (Fig. 4B). Wild-type cells were cultured with 1% DMSO, 10 mg/ml jervine, or 4 mg/ ml EB for 2 h and stained with aniline blue. The arrowhead (or arrow) indicates increased (or decreased) b-1,3-glucan. More than 150 budded cells were observed and quantified according to three phenotypes: b-1,3-glucan intensity is identical in bud and mother cells (normal), the intensity of bud cells is higher than that of mother cells (bud), and the intensity of mother cells is higher than that of bud cells (mother). **, P , 0.01 after Bonferroni correction, t test.
The reduction in radioactive label incorporation was dose dependent-a significant reduction was observed at 10 mg/ml (P , 0.05 after Bonferroni correction, t test) (Fig. 4B). These results suggest that jervine specifically inhibits b-1,6-glucan biosynthesis in yeast.
Effects on yeast strains defective in b-1,6-glucan biosynthesis. We next examined the effect of jervine on yeast mutant strains defective in b-1,6-glucan biosynthesis. Figure 5A summarizes the genes involved in b-1,6-glucan biosynthesis. Kre6p and Skn1p are putative membrane-associated subunits of related, partially redundant b-1,6-glucan synthases (31). Except for KRE6 and SKN1, the genes involved in b-1,6glucan biosynthesis are essential, and therefore, we used conditionally lethal, temperature-sensitive (TS) mutants in this analysis. All of the TS mutants examined showed cell wall phenotypes similar to those of kre6D and jervine-treated cells; the population of cells with accumulated b-1,3-glucan at the buds increased in all TS mutants incubated at the restrictive temperature, with significant increases in the big1-5001, keg1-1, kre5-ts2, and rot1-5001 strains (P , 0.01 after Bonferroni correction, t test) (Fig. 5B). These TS mutants were then tested for jervine susceptibility. Compared with the control wildtype strain (half-maximal inhibitory concentration [IC 50 ] = 9.7 mg/ml), TS mutant strains defective in b-1,6-glucan biosynthesis exhibited jervine-hypersensitive phenotypes at 25°C. The IC 50 s were 0.1 to 1.0 mg/ml, significantly lower than that of the wild-type strain (likelihood ratio test, P , 0.05 after Bonferroni correction) ( Fig. 5C and Table 1).
Effects on KRE6(F552I) and SKN1(F604I) mutants. Because a point mutation in KRE6 that changed phenylalanine to isoleucine at position 552 [KRE6(F552I)] induced resistance to D75, KRE6 was formerly considered the target gene of D75 (18). We report Aniline blue staining of the wild type (WT), b-1,6-glucan synthase mutant (kre6D), b-1,3-glucan synthase mutant (fks1-1154), chitin synthase mutant (chs3D), and mannosyltransferase mutant (mnn9D). The yeast strains were cultured to log phase in YPD medium at 25°C and stained with aniline blue. Because the fks1-1154 strain is a temperature-sensitive mutant (TS mutant), it was cultured overnight at 25°C and cultured for 4 h at restricted temperature (37°C). Arrowhead indicates accumulation of b-1,3-glucan in the bud. (B) Cells with increased b-1,3-glucan in the bud. More than 150 budded cells were observed; the proportions of cells exhibiting b-1,3-glucan accumulation in the bud are shown. Error bars indicate standard deviations. Significant differences from the wild-type strain are indicated with asterisks (**, P , 0.01 after Bonferroni correction, t test). (C) b-1,3-Glucan contents of the cell wall. More than 150 cells were observed, and b-1,3-glucan contents were quantified using ImageJ (n = 3; a.u., arbitrary units). Error bars indicate standard deviations. Significant differences from the wild-type strain are indicated (*, P , 0.05, and **, P , 0.01, after Bonferroni correction, t test).
here that the mechanism of action of jervine is similar to that of D75. However, the kre6D strain showed jervine sensitivity, excluding the possibility that Kre6 is the only target of jervine. KRE6 has a homolog, SKN1 (67% amino acid sequence homology), and the kre6D skn1D double deletion mutation is lethal (31). Therefore, to determine whether both Kre6 and Skn1 are targets of jervine, a possible resistance mutation, Phe604Ile, of SKN1 (corresponding to Phe552Ile of KRE6) was examined.
Antifungal spectrum. The antifungal spectrum of jervine was investigated using human-pathogenic fungi and phytopathogenic fungi. Candida species-including Candida albicans, Candida glabrata, Candida tropicalis, Candida parapsilosis, and Candida krusei-are the major causative fungi of invasive human mycoses (2). We found that jervine was highly effective against C. parapsilosis and C. krusei (Table 3). Specifically, jervine was more effective than fluconazole (FLC) against C. krusei and more effective than EB against C. parapsilosis (Table 3). Therefore, jervine can be used as an alternative when FLC or EB is not sufficiently effective. The germination inhibitory effect of jervine on the phytopathogenic fungi Botrytis cinerea, Puccinia recondita, and Pyricularia oryzae was next investigated. Jervine at 50 mg/ml inhibited the germination of B. cinerea significantly, by 80% (P , 0.05 after Bonferroni correction, t test) (Fig. 7A). Jervine inhibited P. recondita germination by 60% (P , 0.01 after Bonferroni correction, t test) (Fig. 7B). However, the phytopathogenic fungus P. oryzae, which lacks b-1,6-glucan, was resistant to jervine (Fig. 7C). A similar tendency was observed for D75 ( Fig. 7A to C). These results unveiled the antifungal spectrum of jervine against human-pathogenic fungi and phytopathogenic fungi.
Combination therapy. There are only four types of antifungal agents in clinical use. If a single agent is not effective, combination therapy is attempted (33). The combined use of amphotericin B and flucytosine is effective in cryptococcosis (34). Here, we examined the efficacy of combinations of jervine with EB or FLC in S. cerevisiae. Jervine with EB ( Fig. 8A) was effective; more importantly, jervine with FLC showed synergistic effects on the growth rate (fractional inhibitory concentration [FIC] index , 0.5) (Fig. 8B). These results suggest that jervine is effective in combination with EB and FLC. Structural moieties essential for activity of jerveratrum-type steroidal alkaloids. We aimed to identify the steroidal alkaloid residues important for antifungal activity. The IC 50 s of jervine and cyclopamine against S. cerevisiae were similar (Table 4), indicating that the ketone group of jervine is not important for its antifungal activity. The IC 50 of cyclopamine-N-Boc (di-tert-butyl dicarbonate) was ;10-fold that of cyclopamine (P , 0.01 after Bonferroni correction, likelihood ratio test) ( Table 4). This indicated that the amine in the piperidine skeleton is important for antifungal activity, likely forming hydrogen bonds with target molecules. When NP157, NP293, NP329, and NP413 were compared with jervine, there was little difference (Table 4). This may constitute useful information for studies on structure-activity relationships aiming to improve water solubility (35). Cyclopamine-O-Ac (acetyl), in which the same modification site was acetylated in cyclopamine, had a slightly but significantly lower IC 50 (P , 0.01 after Bonferroni correction, likelihood ratio test) ( Table 4), demonstrating that the secondary alcohol in the A ring has some effect on antifungal activity. Therefore, modification of this secondary alcohol via ester and carbamate bonds may increase activity. The basic structure common to these jerveratrum-type steroid alkaloids is the 6-6-5-6 ABCD ring system and its spiro-connected aza-bicyclo[4.3.0]nonane ring. Because jervine has an a,b-unsaturated ketone, it may have a more planar CD ring than cyclopamine. In addition, ring expansion modifications of the basic skeleton by construction of 6-6-5-7 and 7-6-5-7 ABCD ring systems enhance cyclopamine activity (36)(37)(38)(39). The structure-activity relationships of jerveratrum-type steroid alkaloids (Fig. 8C) will contribute to the design of molecules that are more selective for pathogenic fungi.

DISCUSSION
The fungal cell wall is a target for antifungals because it has components absent in humans and plants (12,13). We presented evidence that jerveratrum-type alkaloids are antifungal agents that inhibit fungal cell wall biosynthesis via a mechanism different from those of EB and NZ and affect b-1,6-glucan biosynthesis. Jervine acts on Kre6 and Skn1, both involved in b-1,6-glucan biosynthesis. A combination drug test for human mycosis suggested the antifungal potential of jervine. The skeleton of jerveratrum-type steroidal alkaloids has been studied for over 100 years (40), and more recently, its anticancer potential has been investigated (22)(23)(24)(25). Therefore, jerveratrum-type alkaloids have potential as antifungals, linking cancer and antifungal treatment.
Mechanism of b-1,6-glucan biosynthesis. Although many factors involved in b-1,6-glucan biosynthesis have been identified, its biosynthesis remains unclear. One reason is delayed development of biochemical technology. For example, aniline blue stains b-1,3-glucan, but there was no stain specific for b-1,6-glucan and its intermediate products. From this perspective, the recent development of b-1,6-glucan-specific probes is promising (32). Kinetic studies of b-1,3-glucan biosynthesis have been conducted by tracing the formation of glucose chains (43), but the biosynthesis of b-1,6-glucan chains is unclear. EB likely binds directly to the catalytic subunit Fks1 of b-1,3-glucan synthase (44,45), but the catalytic subunit of b-1,6-glucan synthase has not been identified. Because jervine acted on Kre6 and Skn1, we plan to study the involvement of Kre6 and Skn1 in b-1,6-glucan biosynthesis.
Comparison of jervine and D75. Although jervine and D75 (18) and its derivatives (D11-2040 and D21-6076) (46,47) have different chemical structures, their antifungal activities have the same mechanism. This implies that b-1,6-glucan biosynthesis inhibitors have at least two types of maternal skeleton (jervine, steroid skeleton; D75, heterocyclic skeleton). The development of new antifungals that target the cell wall is desired. As a result, b-1,3-glucan biosynthesis inhibitors have been developed, but only one type of echinocandin is in use (10). Because there is only one echinocandinbased maternal skeleton (cyclic peptide) and various echinocandin-based derivatives were developed by comprehensive chemical synthesis targeting side chains (10), the FIG 8 Checkerboard assays of jervine and the structure required for jervine activity. The wild-type strain was incubated in YPD in the presence of 0 to 64 mg/ml EB and 0 to 64 mg/ml jervine (A) or 0 to 256 mg/ml FLC and 0 to 64 mg/ml jervine (B) at 25°C for 18 h. The degree of proliferation was quantitated using the OD 600 ; that of the control (2% DMSO) was set as 1 (n = 3). Yellow regions represent higher cell densities. The FIC index represents the effect of the combined use of the two compounds and is expressed as the mean value from three biological replicates. (C) Structural moieties essential for the activity of jerveratrum-type steroidal alkaloids. echinocandin-based derivatives may not be effective against fungi that have acquired resistance to echinocandin itself. In contrast, b-1,6-glucan biosynthesis inhibitors have at least two types of chemical structure. Drug discovery research on D75 has been conducted (46,47), and jervine, with its significantly different maternal skeleton, will likely be effective against D75-resistant fungi. Simultaneous development of jervine and D75 may lead to the discovery of additional antifungal agents.
Jervine is effective against NCAC. Non-Candida albicans Candida species (NCAC) are on the rise as a cause of mycosis (48)(49)(50). Many antifungal drugs are generally considered to be less effective against NCAC (51-53), so the available therapeutic agents are limited (52). NCAC include C. parapsilosis and C. krusei, the growth of which jervine inhibited in vitro at low concentrations. Candida parapsilosis has marked biofilm-forming ability (2) and, as a result, is less susceptible to echinocandin-based derivatives (54), implying that our findings are of importance. D75 is not effective against C. parapsilosis (18). Therefore, jervine is the only alternative drug for C. parapsilosis mycosis when FLC is not sufficiently effective. It should be noted that jervine is more selective toward fungal than human cells; jervine concentrations higher than 160 mg/ml are required for the inhibition of nontumor epithelial cells (25). Jervine is not recommended for use in humans due to the teratogenic potential of jerveratrum-type steroidal alkaloids (55). However, jervine could be explored as a scaffold for the development of novel antifungal agents in the future.
Conclusion. In this study, we demonstrated the antifungal activity of jerveratrumtype steroidal alkaloids, such as jervine and cyclopamine, for the first time. These compounds inhibited the growth of the human-pathogenic fungi C. parapsilosis and C. krusei, as well as the phytopathogenic fungi B. cinerea and P. recondita. Jervine did not impact the growth of P. oryzae, which lacks b-1,6-glucan. We found that jervine significantly inhibits b-1,6-glucan biosynthesis in S. cerevisiae. Jervine acts on Kre6 and Skn1, which are involved in b-1,6-glucan biosynthesis. Furthermore, jervine acts synergistically with fluconazole. These findings support the future development of jerveratrumtype steroidal alkaloids as antifungal agents.

MATERIALS AND METHODS
Strains. An S. cerevisiae his3D(MATa his3::kanMX4 leu2 met15 ura3) strain, a derivative of BY4741 harboring a kanMX4 cassette at the his3 locus, was used as the wild-type strain unless otherwise indicated. Various MATa haploid gene deletion strains (with deletions of his3D, kre6D, chs3D, mnn9D, and skn1D) were obtained from the European Saccharomyces cerevisiae Archive for Functional Analysis (Euroscarf). The Candida strains (C. albicans ATCC 24433, C. krusei ATCC 6258, C. tropicalis ATCC 750, and C. parapsilosis ATCC 22019) were obtained from National BioResource Project Pathogenic eukaryotic microorganisms. The other yeast strains used are listed in Table 5.
Morphological analysis. Logarithmic-phase wild-type cells were fixed and stained with fluorescein isothiocyanate-concanavalin A (FITC-ConA) for mannoprotein, rhodamine-phalloidin for actin, and DAPI (4,6-diamidino-2-phenylindole) for nuclear DNA. Images were acquired at room temperature using a fluorescence microscope (Axioplan 2; Carl Zeiss AG, Oberkochen, Germany). A cooled charge-coupled device camera (CoolSNAP HQ; Roper Scientific Photometrics, Tucson, AZ, USA) was used for image capture. Yeast cell image analysis was performed using CalMorph software (version 1.2) as described previously (27). CalMorph automatically characterizes each yeast cell by calculating 501 morphologic parameters based on data from more than 200 cells. Morphological data for 4,718 nonessential-gene deletion mutants, the wild type (126 replicates), and cells treated with EB, TN, and NZ were obtained previously (27,56,57). Data were analyzed using R (http://www.r-project.org/). High-content image profiling, including the Pearson product-moment correlation analysis, was described previously (57).
Chemical-genomics analysis. Chemical-genomics profiling using a pool of 310 deletion mutant strains was performed as described previously (20). Briefly, pooled cultures were treated with the indicated concentrations of jervine (25, 12.5, and 6.25 mg/ml) and D75 (12.5, 6.25, and 3.13 mM) and grown for 48 h at 30°C. Purification of genomic DNA from harvested cells, PCR amplification of barcodes using multiplex primers, and gel purification of barcodes were carried out as described previously (20). Barcodes were sequenced on an Illumina MiSeq using MiSeq reagent kit version 3 (150 cycles; Illumina, Inc., San Diego, CA, USA). The barcode counts detected for each deletion mutant were quantified using BEAN-counter software to generate fitness-based chemical-genetic interaction scores (58). To compare multiple chemical-genomic profiles, we employed chemical-genomic profiles for jervine (12.5 mg/ml) and D75 (6.25 mM) as representatives (Table S1). The chemical-genomics profiles for the pseudojervinerelated compounds (NP157, NP293, NP329, and NP413) and control compounds (TN, calcofluor white, and micafungin) were reported previously (20).
b-1,3-Glucan staining. b-1,3-Glucan staining was performed as described previously (19,59). Yeast cells were cultured overnight at 25°C in YPD medium to ;1 Â 10 7 cells/ml. The cells were washed in phosphate-buffered saline (PBS) and added to 5 mg/ml aniline blue. The signal intensity of aniline blue was quantified using ImageJ software version 1.49v. We binarized the image manually by the default method (Image ! Adjust ! Threshold color). Furthermore, a 1-pixel hole was filled (Process ! Binary ! Close) and a small area considered noise was deleted (Process ! Binary ! Open). The particle areas of cell were added to ROI Manager (Analyze ! Analyze particles), and the mean fluorescence intensity of particle areas was measured using ROI Manager (Measure).
Uptake of [ 14 C]glucose into the cell wall. The wild-type strain was cultured overnight at 25°C in YPD medium to ;1 Â 10 7 cells/ml and adjusted to 1 Â 10 7 cells/ml in low-glucose YPD containing jervine (0, 2.5, 5.0, and 10.0 mg/ml) and 0.624 mCi [ 14 C]glucose (ARC0122; American Radiolabeled Chemicals, Maryland Heights, Mo, USA). Cells were cultured at 25°C for 2 h, and b-1,6-glucan, b-1,3-glucan, and chitin fractions were prepared using a slightly modified protocol of Kitamura et al. (18). Five hundred microliters of 10% trichloroacetic acid (TCA) was added, and the culture was incubated on ice for 10 min. After centrifugation at 15,000 Â g for 1 min, samples were washed twice with distilled water (DW). The pellet was suspended in 500 ml of 1 N NaOH and incubated at 75°C for 1 h. The mixture was centrifuged at 15,000 Â g for 3 min, and the supernatant was discarded. After washing the cells with Tris buffer (10 mM Tris-HCl, pH 7.5), pellets were added to 100 ml of Zymolyase buffer (5 mg/ml Zymolyase 100 T, 10 mM Tris-HCl) and incubated at 37°C for 20 h. The mixture was centrifuged at 15,000 Â g for 15 min, and the supernatant was loaded on a separation filter (UFC 501096; Merck Millipore, Burlington, MA, USA). Tris buffer was added to the pellet, and the pellet was centrifuged at 5,000 Â g for 15 min. The supernatant was placed on a b-1,6-glucan separation filter. The pellet was suspended in 100 ml of Tris buffer to prepare a chitin fraction. The separation filter was centrifuged at 14,000 Â g for 30 min, and the filtration fraction was the b-1,3-glucan fraction; the concentrated fraction was the b-1,6-glucan fraction. Five milliliters of scintillation cocktail (6013329; PerkinElmer, Waltham, MA, USA) and three fractions were added to a plastic vial and assayed using a scintillation counter (LSC-6100; Hitachi Aloka, Tokyo, Japan) for 3 min.
K1 killer toxin susceptibility test. The test strains (his3D and kre6D mutants) were cultured in YPD medium for 10 h at 30°C and diluted with YPD to an optical density at 600 nm (OD 600 ) of 0.01 (4 Â 10 5 cells/ml). Cells (620 ml) were put on a low-pH YPD plate (pH 4.5) and dried. The K1 killer toxin-producing strain (S. cerevisiae NCYC 232) was cultured overnight at 20°C in YPD to an OD 600 of 1.4, and cells (5 ml) were spotted on a plate. After incubation for 2 days at 25°C, the growth inhibition circle was quantified using ImageJ software version 1.49v. We binarized the image manually by the default method (Image ! Adjust ! Threshold color). A 1-pixel hole was filled (Process ! Binary ! Close), and a small area considered noise was deleted (Process ! Binary ! Open). The straight distance (from the end of the colony of NCYC 232 to the test strain) of the growth inhibition circles was manually drawn by 'Straight.' The straight distance was measured at four places per sample using 'Measure' in ROI Manager. A distance of 0 pixels was assumed to be 84.7 mm because 1 pixel = 84.7 mm.
Quantification of b-1,6-glucan. Wild-type and mutant S. cerevisiae strains were grown in YPD at 25°C with shaking at 200 rpm to 1 Â 10 7 cells/ml. The samples were centrifuged at 15,000 Â g for 3 min, and the supernatant was discarded. The pellet was washed, suspended in PBS to adjust it to 1 Â 10 6 cells/ml, and then autoclaved for 20 min. After centrifugation at 15,000 Â g for 1 min, the pellet was further extracted. The b-1,6-glucan was extracted from the pellet using a slightly modified version of the protocol of Kitamura et al. (18). First, 500 ml of 10% TCA was added to the culture, which was incubated on ice for 10 min. After centrifugation at 15,000 Â g for 3 min, the samples were washed twice with DW. The pellet was suspended in 500 ml of 1 N NaOH and incubated at 75°C for 1 h. The solution was mixed with 500 ml of 1 M HCl and Tris buffer (10 mM Tris-HCl, pH 7). After centrifugation at 15,000 Â g for 1 min, the supernatant was stored on ice. The total amounts of b-1,6-glucan were measured according to the method of Yamanaka et al. (32). Briefly, a 96-well white plate was coated with Neg1-E321Q-His (2 mg/ml), followed by overnight incubation at 4°C. The plate was washed with PBS containing 0.05% Tween 20 (PBST) and incubated for 1 h with PBST plus 1% bovine serum albumin (BPBST). After washing, the diluted specimen and standard b-1,6-glucan (pustulan; InvivoGen, San Diego, CA, USA) were added to the plate and incubated for 1 h at room temperature. Biotinized Neg1-E321Q-His (2 mg/ml) in PBS containing BPBST was added to the washed plate, which was incubated for 1 h. The plate was then washed and treated with streptavidin-horseradish peroxidase (HRP) (R&D Systems, MN, USA) in BPBST for 20 min. After removing the unbound enzyme, the peroxidase substrate (SuperSignal ELISA Femto substrate; Thermo Fisher Scientific, Waltham, MA, USA) was added, and luminescence signals were measured using a microplate reader (GloMax; Promega, Madison, WI, USA).
Antifungal susceptibility test using S. cerevisiae mutants. Saccharomyces cerevisiae wild-type and mutant cells were grown in YPD at 25°C or 30°C with shaking at 200 rpm overnight to logarithmic phase (1 Â 10 7 to 5 Â 10 7 cells/ml). Overnight cultures were diluted with YPD, inoculated into YPD containing 3% DMSO (with and without drugs) to 1 Â 10 5 to 5 Â 10 5 cells/ml, and incubated at 30°C in a static incubator. Jervine and D75 concentrations ranged from 0 to 128 mg/ml and 0 to 256 mg/ml, respectively. After 18 h of incubation in 96-well flat-bottom microtiter plates (Corning, Corning, NY, USA), the cell suspension was stirred with a Titramax 1000 rotator (Heidolph, Schwalbach, Germany). The OD 600 was measured using a SpectraMax plus 384 plate reader (Molecular Devices, San Jose, CA, USA). The IC 50 was estimated using the 4-parameter logistic equation in the drc package in R (60). To test whether two pairs of IC 50 s were statistically different, dose-response curves were generated according to Markov chain Monte Carlo methods with the rstan package (https://mc-stan.org/users/interfaces/rstan) and the 4-parameter log-logistic equation was reparametrized using the drc package in R (60). Next, we employed a likelihood ratio test between the full model and null model. The full model and null model assumed differences among all conditions in all four parameters of the log-logistic equation and differences except for the IC 50 between the conditions, respectively. The P value was calculated from the chi-squared distribution after Bonferroni correction.
Antifungal susceptibility testing in Candida species. The susceptibility test for each fungal strain was measured by the Clinical and Laboratory Standards Institute microdilution method (CLSI M60 [61]). Saccharomyces cerevisiae wild-type and Candida (C. albicans, C. tropicalis, C. krusei, and C. parapsilosis) cells were grown on Sabouraud glucose agar at 30°C or 35°C for 24 to 48 h. Cells were washed with saline and diluted in RPMI 1640 to 2.5 Â 10 3 cells/ml. Diluted cells and 1% DMSO (with/without drugs) were added to 96-well round-bottom microplates and incubated at 25°C or 30°C in a static incubator. Jervine concentrations ranged from 0 to 64 mg/ml. Saccharomyces cerevisiae cells were grown at 30°C because they did not grow at 35°C. After 24 h of incubation, cell growth was assessed visually as follows: MIC 90 , optically clear; MIC 50 , prominent decrease in turbidity.
Growth inhibition assay of phytopathogenic fungi. Botrytis cinerea spores were harvested on potato dextrose agar and filtered through cloth. The spore concentration was determined, and the suspension diluted to 2 Â 10 4 spores/ml in DW. Puccinia recondita spores were harvested on wheat and stirred with 2,500-fold-diluted Tween 20. The spore concentration was determined, and the suspension was diluted to 6 Â 10 4 spores/ml. Pyricularia oryzae spores were harvested on oatmeal agar. The spore concentration was determined, and the suspension diluted to 2 Â 10 4 spores/ml in DW. Fifty microliters of diluted spores with 1% DMSO (with/without drugs) was added to 96-well flat-bottom microplates and incubated at 25°C for 48 h, and spore germination was evaluated under an optical microscope.
Checkerboard assay. Synergy was tested by the checkerboard method, a two-dimensional array of serial concentrations of test compounds, which is frequently used to assess combinations of antifungal agents in vitro (62). The tested dilutions were selected based on the MICs of each substance. Each fungal strain was exposed to various concentrations of jervine (0 to 64 mg/ml) in combination with fluconazole (0 to 256 mg/ml) or echinocandin B (0 to 64 mg/ml). The checkerboard test was used to calculate the FIC index (62)  General synthesis procedures. All reagents were of the highest commercial grade and applied directly unless otherwise stated. All reactions were performed under a nitrogen or argon atmosphere unless otherwise stated. Tetrahydrofuran (THF), toluene, hexane, dichloromethane, and ethyl acetate were purchased from Kishida Chemical (Osaka, Japan). Column chromatography was performed with silica gel (Wakogel 60N; Fujifilm Wako Pure Chemicals, Osaka, Japan). Analytical thin-layer chromatography (TLC) was performed with glass TLC plates (silica gel 70 F 254 Plate-Wako; Fujifilm Wako Pure Chemicals, Osaka, Japan). The plates were visualized with UV light and phosphomolybdic acid and subsequent heating. Nuclear magnetic resonance (NMR) spectra were recorded using the ECX-400 instrument (JEOL, Tokyo, Japan). Chemical shift values are reported in ppm (d ) downfield from tetramethylsilane (0 ppm) with reference to an internal residual solvent [ 1 H NMR, CHCl 3 (7.26)]. The coupling constants (J) are reported in Hertz (Hz). The following abbreviations are used to designate the multiplicities: s, singlet; d, doublet; t, triplet; m, multiplet; and br, broad or combination peaks.
Synthesis of cyclopamine-N-Boc. Di-tert-butyl dicarbonate (Boc; 5.3 mg, 0.0243 mmol) was added to a solution of cyclopamine (10 mg, 0.0243 mmol) in THF/H 2 O/toluene (1:1:2, 10 ml) at room temperature and stirred at 90°C overnight. The reaction mixture was cooled to room temperature and concentrated under reduced pressure in vacuo to remove all solvents. This crude compound was purified by silica-gel column chromatography (2:1 ratio of hexane and ethyl acetate). The fractions containing the desired compounds were combined and concentrated under reduced pressure in vacuo to yield cyclopamine-N-Boc as a colorless solid (12.7 mg, ,100%); R f = 0.36 (hexane:ethyl acetate =2:1); 1  Synthesis of cyclopamine-O-Ac. Acetic anhydride (0.15 ml) and diisopropylethylamine (0.23 ml) were added to a mixture of cyclopamine-N-Boc (5.6 mg [0.0109 mmol]) and dichloromethane (1 ml). The reaction mixture was stirred at 50°C for 5 h and cooled to room temperature; saturated aqueous NH 4 Cl (1 ml) was then added. The mixture was washed twice with ethyl acetate (5 ml), dried over sodium sulfate, filtered, and concentrated under reduced pressure. The crude compound was purified by silica gel column chromatography (4:1 ratio of hexane and ethyl acetate). The fractions containing the desired compounds were combined and concentrated under reduced pressure in vacuo, to yield cyclopamine-N-Boc-O-Ac as a colorless solid: Trifluoroacetic acid (0.03 ml) was added dropwise to a mixture of cyclopamine-N-Boc-O-Ac (3.1 mg [5.6 mmol]) and dichloromethane (1 ml) at 0°C. The reaction mixture was stirred for 10 min and concentrated under reduced pressure to remove trifluoroacetic acid. The crude compound was purified by silica-gel column chromatography (ratio of dichloromethane/methanol of 30:1 to 10:1). Fractions containing the desired compounds were combined and concentrated under reduced pressure in vacuo, to yield cyclopamine-O-Ac as a colorless solid: (0.87 mg, 34% yield); R f = 0.50 (dichloromethane/methanol = 5:1); 1

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.4 MB. SUPPLEMENTAL FILE 2, XLSX file, 0.01 MB.