Mutations in hmg1, Challenging the Paradigm of Clinical Triazole Resistance in Aspergillus fumigatus.

Aspergillus fumigatus is the predominant pathogen of invasive aspergillosis, a disease state credited with over 200,000 life-threatening infections each year. The triazole class of antifungals are clinically essential to the treatment of invasive aspergillosis, both as frontline and as salvage therapy. Unfortunately, resistance to the triazoles among A. fumigatus isolates is now increasingly reported worldwide, and a large proportion of this resistance remains unexplained. In this work, we characterize the contributions of previously identified mechanisms of triazole resistance, including mutations in the sterol-demethylase-encoding gene cyp51A, overexpression of sterol-demethylase genes, and overexpression of the efflux pump-encoding gene abcC, among a large collection of highly triazole-resistant clinical A. fumigatus isolates. Upon revealing that these mechanisms alone cannot substantiate the majority of triazole resistance exhibited by this collection, we subsequently describe the identification and characterization of a novel genetic determinant of triazole resistance. Mutations in the 3-hydroxy-3-methyl-glutaryl-coenzyme A (HMG-CoA) reductase-encoding gene, hmg1, were identified in a majority of triazole-resistant clinical isolates in our collection. Introduction of three different hmg1 mutations, predicted to encode residue alterations in the conserved sterol sensing domain of Hmg1, resulted in significantly increased resistance to the triazole class of agents. Additionally, correction of a hmg1 mutation in a pan-triazole-resistant clinical isolate of A. fumigatus with a novel Cas9-ribonucleoprotein-mediated system was shown to restore clinical susceptibility to all triazole agents. Mutations in hmg1 were also shown to lead to the accumulation of ergosterol precursors, such as eburicol, by sterol profiling, while not altering the expression of sterol-demethylase genes.IMPORTANCE Aspergillus fumigatus is the predominant pathogen of invasive aspergillosis, a disease state credited with over 200,000 life-threatening infections annually. The triazole class of antifungals are clinically essential to the treatment of invasive aspergillosis. Unfortunately, resistance to the triazoles among A. fumigatus isolates is now increasingly reported worldwide. In this work, we challenge the current paradigm of clinical triazole resistance in A. fumigatus, by first demonstrating that previously characterized mechanisms of resistance have nominal impact on triazole susceptibility and subsequently identifying a novel mechanism of resistance with a profound impact on clinical triazole susceptibility. We demonstrate that mutations in the HMG-CoA reductase gene, hmg1, are common among resistant clinical isolates and that hmg1 mutations confer resistance to all clinically available triazole antifungals.

INTRODUCTION

Aspergillus fumigatus is the predominant pathogen isolated from patients with invasive aspergillosis, the most common human invasive mold infection. Invasive aspergillosis is responsible for more than 200,000 life-threatening infections each year and afflicts up to 10% of patients with acute leukemia as well as patients receiving stem cell or solid organ transplantation (1–3). Unfortunately, the mortality rates associated with invasive aspergillosis remain unacceptably high, often exceeding 50%, even when appropriately diagnosed and treated (3–6).

Contributing to the poor outcomes associated with the treatment of invasive aspergillosis is the relative paucity of clinically available antifungal agents with demonstrated clinical efficacy. Of the four major classes of antifungals, only the triazoles and amphotericin B are recommended as monotherapy for the treatment of aspergillosis, with the latter relegated to second-line or salvage treatment due to significant toxicities and the lack of oral formulations (7). The triazole class of antifungals therefore has proven essential to the treatment of invasive aspergillosis, comprising both frontline treatment options such as voriconazole and isavuconazole and salvage options such as posaconazole and itraconazole (7). These agents are generally accepted to exert fungicidal activity against Aspergillus through the competitive inhibition of sterol-demethylase, encoded by the genes cyp51A (Afu4g06890) and cyp51B (Afu7g03740) in A. fumigatus, which results in abrogation of the synthesis of ergosterol, the major fungal membrane sterol (8–11). However, the treatment of invasive aspergillosis has recently been further complicated by the global emergence of triazole-resistant disease. Over the past decade, triazole resistance among clinical isolates of A. fumigatus has been increasingly reported on six continents, with resistance rates exceeding 20% now being reported in some studies (12–15).

Considerable research has led to the identification of three distinct molecular pathways to clinical triazole resistance in A. fumigatus. These three paths largely mirror the established mechanisms of resistance previously identified in other pathogenic fungi such as Candida albicans: (i) mutations in the sterol-demethylase gene cyp51A, (ii) overexpression of the sterol-demethylase gene cyp51A, and (iii) overexpression of the efflux pump-encoding gene abcC (16–25). However, it remains unknown to what extent these mechanisms explain the degree of triazole resistance observed among clinical isolates of A. fumigatus, and resistant clinical isolates exhibiting none of these three mechanisms have been repeatedly identified (22). Thus, the molecular mechanisms underpinning much of the clinical resistance to this essential class of anti-Aspergillus agents remain unexplained, limiting the discovery of therapeutic strategies to overcome triazole-resistant aspergillosis.

In this work, we perform a comprehensive characterization of the direct contributions of previously identified mechanisms of triazole resistance in a large collection of triazole-resistant clinical isolates of A. fumigatus. We correct cyp51A mutations in 10 different triazole-resistant clinical isolates using a novel Cas9-mediated transformation system to delineate the direct impact these mutations have on clinical triazole susceptibility, and we assess the potential impact of overexpression of cyp51A, cyp51B, and abcC on triazole susceptibility in a collection of 21 triazole-resistant clinical isolates. We subsequently demonstrate that these mechanisms alone fail to explain the resistance observed in this collection. We further describe here, for the first time, a novel mechanism of clinical triazole resistance which was found to be present in a majority of triazole-resistant clinical isolates in our collection. Mutations in the 3-hydroxy-3-methyl-glutaryl-coenzyme A (HMG-CoA) reductase-encoding gene, hmg1 (Afu2g03700), were found to result in dramatically increased resistance to all clinically available triazole agents. Additionally, the restoration of hmg1 to the wild-type sequence in a clinical A. fumigatus isolate exhibiting high-level pan-triazole resistance was found to restore clinical susceptibility to the triazole class. We demonstrate that mutations in hmg1 result in sustained or even increased ergosterol production as well as accumulation of multiple ergosterol precursors, including eburicol, the substrate of the sterol-demethylase enzymes, which the triazole antifungals competitively inhibit.

RESULTS

Mutations in sterol-demethylase alone poorly explain clinical triazole resistance.

Mutations in the sterol-demethylase gene cyp51A are the most frequently identified mechanism of triazole resistance in A. fumigatus (6, 15, 26). We sought to delineate the proportion of triazole resistance observed in individual clinical isolates explained by this mechanism. To accomplish this, a collection of 26 clinical isolates of A. fumigatus originating from the United States was obtained from the Fungus Testing Laboratory at the University of Texas Health Science Center at San Antonio, including 21 multitriazole-resistant isolates with previously reported clinical origins and characterized cyp51A genotypes and 5 triazole-susceptible control isolates (15). High-quality genomic DNA was extracted from each of the 26 isolates, whole-genome sequencing was performed targeting a read depth of approximately 100 reads per base aligned to the Af293 A. fumigatus reference genome assembly, and a total of 48,552 SNPs and insertions or deletions (indels) unique to triazole-resistant isolates were identified within 8,669 genes with annotations available. Of the 21 triazole-resistant isolates in this collection, 16 were identified as possessing peptide-sequence-altering cyp51A mutations unique to resistant isolates. Conversely, no unique cyp51B mutations were identified. We selected 10 of these 16 isolates with cyp51A mutations, which also exhibited resistance to all clinically available triazole agents, for correction of the cyp51A sequence to that of the wild-type consensus sequence. These isolates included examples of previously characterized cyp51A mutations (TR34/L98H, TR46/Y121F/T289A, and G448S), mutations at residues where alternative mutations have previously been characterized (G138C and F219S), and mutations never before characterized to our knowledge (M263I and I367F) (Table 1) (17, 24, 25, 27, 28).

TABLE 1

Triazole MICs of clinical isolates and derivative cyp51AWT strains

Parent clinical isolate orderivative cyp51AWT strain	cyp51A genotype	MIC (mg/liter) of triazole:
		Voriconazole	Itraconazole	Posaconazole
DI15-117	F219S	2	≥32	2
DI15-117 cyp51AWT	WT	4	≥32	4
DI15-100	M263I	8	≥32	1
DI15-100 cyp51AWT	WT	8	≥32	0.5
DI15-116	TR34/L98H	8	≥32	1
DI15-116 cyp51AWT	WT	1	0.25	0.125
DI15-102	TR34/L98H	8	≥32	2
DI15-102 cyp51AWT	WT	4	≥32	0.5
DI15-108	G138C	4	≥32	1
DI15-108 cyp51AWT	WT	4	≥32	0.125
DI15-95	M263I	16	≥32	2
DI15-95 cyp51AWT	WT	16	≥32	1
DI15-98	I367F	16	16	2
DI15-98 cyp51AWT	WT	16	≥32	1
DI15-96	TR46/Y121F/T289A	≥32	4	1
DI15-96 cyp51AWT	WT	≥32	0.5	0.25
DI15-106	TR46/Y121F/T289A	≥32	4	1
DI15-106 cyp51AWT	WT	8	2	0.5
DI15-120	G448S	≥32	≥32	1
DI15-120 cyp51AWT	WT	≥32	≥32	1

MICs shown in bold are greater than epidemiologic cutoff values as published by the Clinical Laboratory and Standards Institute (1 mg/liter for voriconazole and itraconazole; 0.25 mg/liter for posaconazole). WT, cyp51A sequence matches the Af293 wild-type consensus sequence.

In order to test the direct contribution of these mutations to clinical triazole resistance in these isolates, the wild-type cyp51A promoter region and open reading frame from the triazole-susceptible reference strain Af293 were first cloned upstream of the hygromycin resistance cassette in plasmid pCR-HygR (29). Fifteen hundred bases of the cyp51A terminator was then cloned downstream of the hygromycin resistance cassette, yielding the plasmid pCyp51A-HygR. Then, the cyp51AWT transformation repair template was created by PCR using primers which amplified the wild-type cyp51A allele (including 500 bases of promoter sequence), the hygromycin resistance cassette, and 500 bases of the cyp51A terminator. Importantly, this repair template extends approximately 200 bases upstream of the promoter region where characterized regulatory elements and tandem repeat sequences have been identified (21). Protoplasts of each of the 10 triazole-resistant clinical isolates, as well as the well-characterized triazole-susceptible laboratory strain akuBKU80 (CEA17 ΔakuBKU80), which has a wild-type cyp51A allele, were then transformed with Cas9-ribonucleoprotein (RNP) complexes, consisting of Cas9 protein and CRISPR RNA (crRNA) and trans-activating CRISPR RNA (trRNA) duplexes targeting sequences immediately upstream and downstream of the cyp51A locus. The resulting transformants were selected on Aspergillus minimal medium containing hygromycin as previously described (29, 30). Integration of the repair template at the native cyp51A locus was confirmed by PCR, and acquisition of the cyp51AWT sequence was verified by Sanger sequencing.

Clinical susceptibility testing was performed for each of the 10 triazole-resistant clinical isolates and the corresponding cyp51AWT derivative strains for voriconazole, itraconazole, and posaconazole according to the Clinical and Laboratory Standards Institute M38-A2 standards (Table 1) (31). All of the resistant clinical isolates exhibited triazole MICs consistent with those previously reported for these isolates under clinical susceptibility testing conditions, and all MICs for each agent tested were above epidemiologic cutoff values previously reported by CLSI (1 mg/liter for voriconazole and itraconazole and 0.25 mg/liter for posaconazole) (15, 32). Among the cyp51AWT strains, only one strain (DI15-116 cyp51AWT) was observed to have full restoration of wild-type triazole susceptibility, and two strains (DI15-96 cyp51AWT and DI15-108 cyp51AWT) were observed to have restoration of wild-type susceptibility to at least one triazole. Additionally, two strains (DI15-102 cyp51AWT and DI15-106 cyp51AWT) were observed to have at least one triazole MIC decrease by more than 1 dilution. Five of the 10 cyp51AWT derivative strains exhibited no change in susceptibility to any triazole agent tested. Importantly, replacement of the native wild-type cyp51A allele with the cyp51AWT cassette in the akuBKU80 background had no effect on triazole MIC. Taken together, these data demonstrate that while mutations in cyp51A are common among triazole-resistant clinical isolates of A. fumigatus, and a number of these mutations have been demonstrated to increase triazole MICs when introduced into a triazole-susceptible background, mutations in cyp51A alone poorly explain the high level of triazole resistance observed in this collection of clinical isolates.

Analysis of the expression of the sterol-demethylase genes cyp51A and cyp51B among triazole-resistant clinical isolates.

Triazole-resistant clinical isolates of A. fumigatus exhibiting overexpression of cyp51A and cyp51B, relative to susceptible comparator isolates, have previously been reported (22, 33, 34). Additionally, both the formation of tandem repeats (TR34, TR46, and TR53) in the promoter region of cyp51A and a mutation (encoding the amino acid substitution P88L) in hapE, a CCAAT-binding complex protein-encoding gene which participates in the negative regulation cyp51A expression, have been shown to increase both cyp51A expression and triazole MICs (20, 21, 24, 25). Following the observation that mutations in cyp51A alone do not explain the degree of triazole resistance observed in this collection of clinical isolates, we sought to evaluate the potential contributions of sterol-demethylase gene overexpression to clinical triazole resistance. To accomplish this, each of the 26 clinical isolates in our collection was grown overnight at 37°C in Aspergillus minimal medium in biological triplicate, and RNA was extracted from mature hyphae following liquid nitrogen crush as previously described (35). Reverse transcription-quantitative-PCR (RT-qPCR) was then utilized to quantify the relative expression of both cyp51A and cyp51B in each of the 21 triazole-resistant clinical isolates compared to the expression level observed among the 5 triazole-susceptible control isolates.

The expression of cyp51A and cyp51B in all clinical isolates was compared to the triazole-susceptible isolate DI16-6 (voriconazole MIC, 0.5 mg/liter). Among triazole-susceptible control isolates, the expression of cyp51A ranged from 0.5- to 1.6-fold that of isolate DI16-6 (Fig. 1a), with isolates DI16-5, DI16-7, and DI16-9 exhibiting statistically significant lower expression of cyp51A. The expression of cyp51A among triazole-resistant clinical isolates ranged from 0.7- to 3.5-fold that of the susceptible comparator isolate. Twelve of the 21 triazole-resistant isolates were observed to have cyp51A expression that exceeded the highest expression level observed among triazole-susceptible isolates. Of these 12 isolates, 8 were found to have cyp51A expression levels statistically greater than the susceptible comparator isolate. This included 2 isolates (DI15-102 and DI15-116) which possess the TR34/L98H cyp51A mutation, 2 isolates (DI15-96 and DI15-106) which possess the TR46/Y121F/T289A cyp51A mutation, and one isolate (DI15-105) which possesses the previously characterized P88L-encoding mutation in hapE. However, no previously characterized mutations known to contribute to cyp51A overexpression were identified in the remaining 3 isolates. The expression of cyp51B among the triazole-susceptible control isolates ranged from 0.7- to 1.3-fold that of isolate DI16-6 (Fig. 1b), with isolate DI16-8 exhibing statistically greater expression of cyp51B (1.3-fold). The expression of cyp51B among triazole-resistant clinical isolates ranged from 0.7- to 2.1-fold that of the susceptible comparator isolate, with 4 isolates (DI15-95, DI15-97, DI15-102, and DI15-108) exhibiting expression of cyp51B exceeding that of the highest expression level observed among triazole-susceptible isolates, 3 of which (DI15-95, DI15-102, and DI15-108) were statistically significant.

FIG 1

Relative expression of cyp51A and cyp51B among triazole-resistant clinical isolates. (a) Expression of cyp51A in clinical isolates relative to triazole-susceptible isolate DI16-6; identified mutations known to contribute to cyp51A overexpression are shown on the corresponding bar. (b) Expression of cyp51B in clinical isolates relative to triazole-susceptible isolate DI16-6; triazole-susceptible isolates are shown as open bars, and triazole-resistant isolates are shown as filled bars. Triazole-resistant isolates shown in order of escalating voriconazole MICs from left to right. Horizontal dotted lines denote highest and lowest expression levels observed among the 5 triazole-susceptible isolates. Comparisons with the susceptible control isolate DI16-6 with statistically significant (P < 0.05) results are noted with an asterisk.

Constitutive overexpression of cyp51B decreases triazole susceptibility.

Overexpression of cyp51A has previously been shown to contribute to triazole resistance, but it remains unknown if overexpression of the paralogous gene, cyp51B, has a similar effect on triazole susceptibility. We sought to employ a novel Cas9-RNP mediated promoter replacement system to delineate the direct impact of constitutive overexpression of cyp51B on triazole MIC. To accomplish this, the previously characterized, strong, constitutive promoter from the heat shock protein-encoding gene, hspA (Afu1g07440), was cloned upstream of the hygromycin resistance cassette in the plasmid pCR-HygR (36). A transformation repair template for the cyp51B promoter replacement was then constructed by amplifying the hspA promoter and hygromycin resistance cassette from the resulting plasmid (pJMR2) with primers that introduced approximately 50 bases of homology for the 3′ end of the native promoter region and the 5′ end of the cyp51B open reading frame. Protoplasts of the well-characterized triazole-susceptible laboratory strain akuBKU80 were then transformed with Cas9-ribonucleoprotein (RNP) complexes, consisting of Cas9 protein and crRNA-trRNA duplexes targeting sequences immediately upstream of the cyp51B start codon, yielding the strain akuBKU80_P (Fig. 2) (37). As a comparator, the same methodology was employed to create the constitutive cyp51A-overexpressing strain akuBKU80_P.

FIG 2

Schematic of Cas9-RNP-mediated promoter replacement system. Constitutive overexpression repair templates were amplified from pJMR2 using primers incorporating microhomology targeting genes of interest (GOI) and then mixed with Cas9-RNP targeting immediately upstream of the start codon of the GOI, and protoplasts were created from the strain akuBKU80.

The relative expression of both cyp51A and cyp51B in the strains akuBKU80_P and akuBKU80_P was then compared to the parental akuBKU80 strain both under normal growth conditions and following voriconazole treatment. Conidia from each strain were grown in Aspergillus minimal medium at 37°C until germination and then were transferred to fresh Aspergillus minimal medium containing either no voriconazole or 0.125 mg/liter of voriconazole (half of the MIC of the parental akuBKU80 strain) for an additional 6 h at 37°C. RNA was extracted from each sample as previously described, and RT-qPCR was performed. The expression of cyp51B in akuBKU80 was 2-fold higher after treatment with voriconazole than under untreated conditions (Fig. 3a). Expression of cyp51B in the akuBKU80_P strain was 35.9-fold higher under untreated conditions and 36.9-fold higher following voriconazole exposure than the expression measured in the parental akuBKU80 under untreated conditions. This increased expression was statistically greater than in the parental strain under both treated and untreated conditions. The expression of cyp51B in the akuBKU80_P strain approximated that of akuBKU80 under both untreated and voriconazole-treated conditions. The expression of cyp51A in akuBKU80 was 4.1-fold higher following treatment with voriconazole than under untreated conditions. Expression of cyp51A in the akuBKU80_P strain was 14.1-fold higher under untreated conditions, and 18.5-fold higher following voriconazole exposure, than the expression measured in the parental akuBKU80 under untreated conditions. This increased expression was statistically greater than in the parental strain under both treated and untreated conditions. The expression of cyp51A in the akuBKU80_P strain was similar to that of akuBKU80 under both untreated and voriconazole-treated conditions.

FIG 3

Relative expression of cyp51A and cyp51B and voriconazole MICs in promoter replacement strains akuBKU80_P and akuBKU80_P. (a) The expression of cyp51A (left) and cyp51B (right) in akuBKU80, akuBKU80_P, and akuBKU80_P was determined by RT-qPCR following growth in Aspergillus minimal medium at 37°C both with voriconazole (0.125 mg/liter) treatment for 6 h and without voriconazole treatment. The expression level for each sample is shown relative to that of the respective gene in akuBKU80 without voriconazole treatment. (b) Voriconazole MICs for akuBKU80, akuBKU80_P, and akuBKU80_P. Comparisons with statistically significant (P < 0.05) results are noted with an asterisk, while those that are not are noted with “ns.”

Upon confirmation of the constitutive overexpression of cyp51A and cyp51B in the akuBKU80_P and akuBKU80_P strains, respectively, voriconazole MICs were determined in triplicate according to CLSI M38-A2 standards (31). The parental akuBKU80 strain exhibited a voriconazole MIC of 0.25 mg/liter. The akuBKU80_P strain was observed to have a 2-fold increase in voriconazole MIC (0.5 mg/liter), which is similar to the decrease in susceptibility previously reported when expression of this gene is increased as a result of either the formation of tandem repeats in the cyp51A promoter or a mutation in hapE (Fig. 3b) (20, 21, 24, 25). Interestingly, the voriconazole MIC of the akuBKU80_P strain was 1 mg/liter, 4-fold higher than the parental isolate. Taken together, these results demonstrate that overexpression of either sterol-demethylase gene can decrease triazole susceptibility. However, even when the expression of either of these genes increased by as much as 36.9-fold, the MIC did not exceed the epidemiological cutoff value for voriconazole reported by CLSI (1 mg/liter) (32).

Analysis of the expression of the efflux pump-encoding gene abcC among triazole-resistant clinical isolates.

In addition to sterol-demethylase-mediated mechanisms of triazole resistance, increased constitutive expression of the ATP-binding cassette efflux pump-encoding gene, abcC (Afu1g14330; also known as abcB, cdr1B, and abcG1), has been associated with triazole resistance in A. fumigatus (22, 23, 38). Furthermore, deletion of abcC in both triazole-susceptible laboratory strains of A. fumigatus and a triazole-resistant clinical isolate has been shown to increase triazole susceptibility (22, 23, 38). As cyp51A- and cyp51B-mediated mechanisms alone do not adequately explain the high level of triazole resistance in this collection of clinical isolates, we sought to characterize the expression of abcC and its potential impact on triazole susceptibility. To accomplish this, each of the 26 clinical isolates in our collection was grown overnight at 37°C in Aspergillus minimal medium in biological triplicate, and RNA was extracted as previously described. RT-qPCR was then utilized to quantify the relative expression of abcC in each of the 21 triazole-resistant clinical isolates compared to the expression level observed among the 5 triazole-susceptible control isolates.

The expression of abcC in all clinical isolates was compared to the triazole-susceptible isolate DI16-6. Among triazole-susceptible control isolates, the expression of abcC ranged from 0.7- to 2-fold that of isolate DI16-6 (Fig. 4a), with both isolates DI16-5 and DI16-8 exhibiting statistically greater abcC expression (1.3- and 2.0-fold, respectively) than DI16-6. The expression of abcC among triazole-resistant clinical isolates ranged from 1.2- to 33.2-fold that of the susceptible comparator isolate. Nineteen of the 21 triazole-resistant isolates were observed to have abcC expression that was statistically greater than the expression observed in the susceptible control isolate DI16-6, 14 of which exhibited abcC expression which also exceeded the highest expression level observed among triazole-susceptible isolates (2-fold). While the majority of these 14 isolates exhibited modestly elevated abcC expression (2.4- to 4.1-fold that of the comparator susceptible isolate), isolates DI15-110, DI15-120, and DI15-106 exhibited markedly higher levels of abcC expression (6.5-, 10.4-, and 33.2-fold, respectively). Notably, the pan-triazole-resistant isolate DI15-110, which exhibited the highest level of abcC expression, possesses no mutations in cyp51A and did not exhibit increased constitutive expression of either sterol-demethylase gene.

FIG 4

Relative expression of abcC in triazole-resistant clinical isolates and the constitutive overexpression strain akuBKU80_P. Triazole-susceptible isolates are shown as open bars; triazole-resistant isolates are shown as filled bars. Triazole-resistant isolates shown in order of escalating voriconazole MICs from left to right. Horizontal dotted lines denote highest and lowest expression levels observed among the 5 triazole-susceptible isolates. (a) Expression of abcC relative to triazole-susceptible isolate DI16-6. (b) The expression of abcC in strains akuBKU80 and akuBKU80_P was determined by RT-qPCR following growth in Aspergillus minimal medium at 37°C both with voriconazole (0.125 mg/liter) treatment for 6 h and without voriconazole treatment. The expression level for each sample is shown relative to that of the respective gene in akuBKU80 without voriconazole treatment. Comparisons with statistically significant (P < 0.05) results are noted with an asterisk, while those that are not are noted with “ns.”

As previous studies associating increased triazole resistance with abcC have entirely relied upon deletion of the abcC gene among either clinical or laboratory isolates of A. fumigatus, we next sought to utilize our Cas9-RNP-mediated promoter replacement system to delineate the direct impact of constitutive overexpression of abcC on triazole MIC. To accomplish this, protoplasts of the triazole-susceptible laboratory strain akuBKU80 were transformed with Cas9-ribonucleoprotein (RNP) complexes, consisting of Cas9 protein and crRNA-trRNA duplexes targeting sequences immediately upstream of the abcC start codon, as well as the P transformation repair template, which contains both the hspA promoter and hygromycin resistance cassette flanked by 50 bases of microhomology targeting the abcC locus, yielding the strain akuBKU80_P Subsequently, the relative expression of abcC in both akuBKU80_P and the parental strain akuBKU80 was assessed by RT-qPCR following growth in Aspergillus minimal medium at 37°C both with voriconazole treatment (0.125 mg/liter) for 6 h and without.

In the parental akuBKU80 strain, abcC expression was significantly (9.7-fold) higher following voriconazole treatment than was observed without voriconazole treatment (Fig. 4b). By comparison, the akuBKU80_P strain exhibited 7.8-fold- and 10.4-fold-higher abcC expression, under untreated and voriconazole-treated conditions, respectively, than the expression of abcC measured in the parental akuBKU80 under untreated conditions. abcC expression was not found to be significantly different between akuBKU80_P and the parental strain when both had been treated with voriconazole for 6 h. Surprisingly, even with an approximately 10-fold increase in the constitutive expression of abcC, the akuBKU80_P strain did not demonstrate any change in voriconazole susceptibility relative to the parental akuBKU80 strain (MIC, 0.25 mg/liter).

Mutations in hmg1 are common among triazole-resistant clinical isolates of A. fumigatus.

Following the finding that the previously characterized mechanisms of triazole resistance did not explain the high level of triazole resistance observed in this collection of clinical A. fumigatus isolates, we sought to interrogate the previously generated whole-genome sequencing data to identify novel mechanisms of resistance. As the triazoles exert antifungal activity through the inhibition of biosynthesis of the predominant fungal membrane sterol ergosterol, initial whole-genome sequencing analysis targeted genes encoding proteins known to participate in ergosterol biosynthesis. In addition to the previously identified cyp51A mutations, mutations unique to triazole-resistant isolates were also observed in erg3B (Afu2g00320), erg3C (Afu8g01070), erg4B (Afu1g07140), erg5 (Afu1g03950), and erg6 (Afu4g03630). However, mutations occurred in these genes among only a small number of the clinical isolates (see Table S1 in the supplemental material). Intriguingly, 11 of the 21 triazole-resistant clinical isolates (52%) were found to possess mutations in the hmg1 gene resulting in peptide sequence changes not observed among the triazole-susceptible control isolates (Fig. 5a). The A. fumigatus hmg1 gene encodes a HMG-CoA reductase enzyme previously reported to be essential. HMG-CoA reductase catalyzes the first committed step in ergosterol biosynthesis and has previously been shown in other eukaryotic organisms to participate in the regulation of sterol biosynthesis through direct interactions between sterols and a conserved sterol-sensing domain (39–42). Of particular interest, eight of the 11 isolates with hmg1 mutations were found to have mutations affecting residues predicted to reside within the conserved sterol-sensing domain of Hmg1 (Fig. 5b). We additionally accessed publicly available whole-genome sequencing data for 11 triazole-resistant clinical isolates (https://www.ebi.ac.uk/arrayexpress/experiments/E-SYBR-1/samples/) and found 5 to have mutations in hmg1, 2 of which alter residues in the predicted sterol-sensing domain (P309L and I412T). As mutations in hmg1 were observed in a large proportion of clinical triazole-resistant isolates, and hmg1 mutations have previously been observed among laboratory isolates of A. fumigatus grown in the presence of triazole antifungals and the closely related agricultural sterol-demethylase inhibitors, the observed mutations in hmg1 were selected for further interrogation (43, 44).

FIG 5

Mutations in hmg1 observed in this collection of triazole-resistant clinical A. fumigatus isolates. (a) hmg1 mutations observed in each clinical isolate relative to the Af293 reference genome, with mutations affecting residues predicted to reside within the conserved sterol-sensing domain shown in bold. (b) Schematic depicting the predicted transmembrane domains of A. fumigatus Hmg1. Residues in light blue comprise the predicted sterol-sensing domain, residues in dark blue represent those affected by hmg1 mutations observed in this collection of clinical isolates, and residues shown in green represent those affected by hmg1 mutations observed in this collection of clinical isolates and further characterized in this work.

Mutations in other genes involved in ergosterol biosynthesis which are unique to triazole-resistant isolates. Download Table S1, DOCX file, 0.1 MB.

Mutations in hmg1 confer resistance to the triazole class of antifungals.

To test the hypothesis that mutations in hmg1 directly contribute to clinical triazole resistance in A. fumigatus, three mutations encoding amino acid substitutions or in-frame codon deletions in different transmembrane regions of the predicted hmg1 sterol-sensing domain (F262del, S305P, and I412S) were selected for characterization (Fig. 5b). Each of these mutations and a wild-type control allele were directly introduced to the native hmg1 locus of the triazole-susceptible laboratory strain akuBKU80 using a novel Cas9-RNP editing technique incorporating a split hygromycin B resistance marker (Fig. S1). Briefly, hmg1 alleles including the open reading frame and approximately 500 downstream bases were amplified by PCR from DI15-98 (S305P), DI15-100 (I412S), DI15-105 (F262del), and akuBKU80 (wild-type control) using a 3′ primer which introduced the terminal 80 bases with homology to the 3′ end of the hphR hygromycin B resistance gene open reading frame. A partial hygromycin B resistance cassette, including the gdpA promoter and a truncated hphR gene lacking the terminal 40 bases, was then amplified by PCR from the pUCGH plasmid using primers that introduced approximately 70 bases of homology with the downstream region of hmg1 (45). Protoplasts of akuBKU80 were then transformed with RNP complexes, consisting of Cas9 protein and crRNA-trRNA duplexes targeting sequences immediately upstream and approximately 500 bases downstream of the open reading frame of hmg1, and 2 μg of each portion of the split marker repair template (a single hmg1 allele and the truncated hygromycin B resistance cassette), to produce hmg1 mutant strains akuBKU80 hmg1F262del, akuBKU80 hmg1S305P, akuBKU80 hmg1I412S, and akuBKU80 hmg1WT.

Schematic of Cas9-ribonucleoprotein (RNP) editing technique incorporating a split hygromycin B resistance marker for creation of hmg1 mutant strains. Download FIG S1, PDF file, 0.1 MB.

Antifungal susceptibility testing was subsequently performed for akuBKU80 and each of the derivative strains for voriconazole, isavuconazole, itraconazole, posaconazole, and amphotericin B according to the Clinical and Laboratory Standards Institute M38-A2 standards (Fig. 6a) (31). The parental akuBKU80 strain was susceptible to all agents with MICs of 0.25, 0.5, 0.125, 0.06, and 0.5 mg/liter for voriconazole, isavuconazole, itraconazole, posaconazole, and amphotericin B, respectively. The akuBKU80 hmg1WT control strain exhibited MICs exactly matching those of the parental strain. Conversely, all strains harboring a clinically derived mutation in hmg1 were found to exhibit a 4- to 8-fold increase in MICs for all tested triazoles, exceeding the CLSI reported epidemiologic cutoff values for at least one agent in each mutant strain. Notably, amphotericin B MICs were observed to decrease by 2-fold in both the akuBKU80 hmg1S305P and akuBKU80 hmg1F262del strains (0.5 versus 0.25 mg/liter) but not in akuBKU80 hmg1I412S.

FIG 6

Impact of hmg1 mutations on antifungal susceptibility. (a) Triazole MICs for akuBKU80 and derivative mutant hmg1 strains. (b) Triazole MICs for DI15-105 and derivative hmg1WT strain.

In an effort to delineate the extent of hmg1 mutation-mediated triazole resistance within a clinical isolate from our collection, the mutant hmg1 allele in the pan-triazole-resistant clinical isolate DI15-105 was replaced with the wild-type hmg1 allele from akuBKU80 using the same Cas9-RNP-mediated gene editing described above. The resultant DI15-105 hmg1WT strain exhibited complete restoration of triazole susceptibility with an 8-fold or greater decrease in MIC for each triazole agent (Fig. 6b). Intriguingly, the amphotericin B MIC was observed to increase by 2-fold in DI15-105 hmg1WT relative to DI15-105 (0.125 versus 0.25 mg/liter).

Mutations in hmg1 lead to accumulation of ergosterol precursors.

As the sterol-sensing domain of HMG-CoA reductase has previously been shown to interact with ergosterol precursors and participate in the negative regulation of HMG-CoA reductase activity in both human cells and the model fission yeast Schizosaccharomyces pombe, we hypothesized that the mutations identified in the sterol-sensing domain of A. fumigatus hmg1 may be leading to dysregulation of the ergosterol biosynthesis pathway resulting in the observed triazole resistance (Fig. 7a) (39–42). To test this hypothesis, comprehensive sterol profiling including assessment of the relative distribution of cell sterols and the measurement of total ergosterol content per dry weight was performed on akuBKU80, akuBKU80 hmg1F262del, akuBKU80 hmg1S305P, akuBKU80 hmg1I412S, and akuBKU80 hmg1WT.

FIG 7

Relative distribution of sterols and total ergosterol content of hmg1 mutant strains. (a) A. fumigatus ergosterol biosynthetic pathway. (b) Relative fold change of cellular sterols lanosterol, eburicol, 4,4-dimethylergosta-8,24(28)-dienol, fecosterol, episterol, and ergosterol in hmg1 mutant strains compared to the parental akuBKU80 strain. (c) Total cellular ergosterol per dry weight for hmg1 mutant strains and the parental akuBKU80 strain. Comparisons with the susceptible parental isolate akuBKU80 with statistically significant (P < 0.05) results are noted with an asterisk.

To assess the relative distribution of sterols among A. fumigatus strains harboring hmg1 mutations, freshly harvested conidial stocks of each strain were grown in six biological replicates in RPMI medium supplemented with 0.2% glucose and buffered with MOPS (pH 7.0) for 24 h before cells were flash frozen with liquid nitrogen, dry weights were obtained, and nonsaponifiable lipids were extracted, derivatized, and analyzed by gas chromatography-mass spectrometry as described previously (46, 47).

In agreement with the hypothesis that mutations in hmg1 precipitate dysregulation of the ergosterol biosynthetic pathway, the distribution of sterols was dramatically altered in all hmg1 mutant strains relative to akuBKU80 (Fig. 7b). Most notably, a significant decrease in the relative proportion of ergosterol, along with corresponding increases in the proportions of several ergosterol precursors downstream of Hmg1, including eburicol, 4,4-dimethylergosta-8,24(28)-dienol, and fecosterol, were observed in all mutant hmg1 strains. Additionally, the relative proportion of episterol was observed to be significantly elevated in akuBKU80 hmg1F262del and akuBKU80 hmg1S305P. Importantly, no significant difference in the relative distribution of sterols was observed between akuBKU80 and akuBKU80 hmg1WT (data not shown).

While the distribution of cellular sterols among A. fumigatus strains harboring hmg1 mutations demonstrated a significant decrease in the relative proportion of cellular ergosterol, total ergosterol content per dry weight was not lower among any of the mutant hmg1 strains (Fig. 7c). In fact, relative to the parental akuBKU80, akuBKU80 hmg1F262del was observed to exhibit a significant increase in total ergosterol (4.56 versus 7.06 μg ergosterol/mg [dry weight]; P < 0.01). Taken together, these findings demonstrate hmg1 mutations precipitate a relative increase in ergosterol precursors downstream of Hmg1, while maintaining or in some cases even increasing cellular ergosterol, leading to significantly altered cellular sterol profiles among mutant hmg1 strains.

Mutations in hmg1 do not lead to increased expression of sterol-demethylase genes.

As our sterol profiling studies revealed significant accumulations of ergosterol precursors both upstream and downstream of sterol-demethylase, and we had previously observed that increased expression of either sterol-demethylase gene contributes to triazole resistance, we sought to investigate if the triazole resistance associated with mutations in hmg1 was at least in part mediated by altered expression of sterol-demethylase genes. To accomplish this, conidial suspensions of akuBKU80, akuBKU80 hmg1F262del, akuBKU80 hmg1S305P, akuBKU80 hmg1I412S, and akuBKU80 hmg1WT were grown in biological triplicates under the same conditions used for sterol profile analysis, and RNA was extracted from liquid nitrogen-pulverized samples, as previously described (35). RT-qPCR was then performed to assess expression of the sterol-demethylase genes cyp51A and cyp51B among mutant hmg1 strains relative to the parental akuBKU80. No significant differences in expression of either sterol-demethylase gene were observed for any of the mutant hmg1 strains or the akuBKU80 hmg1WT control strain (Fig. 8). This finding demonstrated that while mutant hmg1 strains are observed to have relatively greater proportions of ergosterol precursors, expression levels of both cyp51A and cyp51B remain unchanged. Thus, resistance in these strains is not due to increased expression of these genes.

FIG 8

Expression of sterol-demethylase genes is not elevated among mutant hmg1 strains. Fold expression of cyp51A and cyp51B in mutant hmg1 strains relative to the parental akuBKU80 strain as determined by reverse transcription-quantitative PCR.

DISCUSSION

Resistance to the triazole class of antifungals represents a critical threat to the current treatment of invasive aspergillosis, a devastating disease state primarily afflicting the expanding populations of immunocompromised patients worldwide. While considerable research has been performed to interrogate the mechanisms by which pathogenic fungi such as Aspergillus become resistant to triazole antifungals, the overwhelming majority of resistance has been ascribed to three canonical paths that were originally described among pathogenic fungi decades ago: mutations in sterol-demethylase genes, overexpression of sterol-demethylase, or overexpression of efflux pump-encoding genes. However, these mechanisms alone or even in combination poorly explain a large proportion of the triazole resistance observed among clinical A. fumigatus isolates today.

In this work, we performed a comprehensive characterization of the direct contributions of previously identified mechanisms of triazole resistance in a collection of triazole-resistant clinical isolates of A. fumigatus. We observed that correction of cyp51A mutations in 10 pan-triazole-resistant clinical isolates restored wild-type susceptibility, to even a single triazole agent, in only 3 isolates. We then demonstrated that while constitutive overexpression of cyp51A, and to a lesser extent cyp51B, was common among isolates in this collection, increasing expression of either sterol-demethylase gene by as much as 18- to 36-fold in the triazole-susceptible akuBKU80 background had a limited effect on triazole MICs. Analogously, we then observed that while the expression of the ABC-type efflux pump-encoding gene abcC was higher among 14 of the triazole-resistant isolates in this collection, constitutive overexpression of this gene in the akuBKU80 background did not have any effect on voriconazole MIC. Upon revealing that the previously known mechanisms of clinical triazole resistance explain only a small fraction of the resistance observed in this collection, we then utilized whole-genome sequencing to identify a previously uncharacterized mechanism of resistance present among the majority of the isolates.

Here, we describe for the first time a novel and noncanonical mechanism of triazole antifungal resistance. Identified through whole-genome sequencing, mutations in the HMG-CoA reductase-encoding gene, hmg1, were found among the majority of triazole-resistant A. fumigatus clinical isolates in our collection. Of the observed mutations, most occurred in the conserved sterol-sensing domain of hmg1, which has been previously demonstrated to participate in the regulation of sterol biosynthesis in other eukaryotic organisms, including S. pombe. When three of these clinically derived hmg1 mutations were independently introduced into a well-characterized triazole-susceptible strain of A. fumigatus using an optimized Cas9-mediated system to replace the native hmg1 locus, resistance to all clinically available anti-Aspergillus triazole antifungals greatly increased. To confirm that mutations in hmg1 are a novel mechanism of clinical triazole resistance, we employed the same system to correct a hmg1 mutation identified in a pan-triazole-resistant clinical isolate of A. fumigatus and observed restoration of susceptibility to all triazoles.

As the sterol-sensing domain of HMG-CoA reductase has been shown to be essential to the regulation of sterol biosynthesis, we hypothesized that amino acid substitutions in this domain of Hmg1 may lead to impaired negative regulation of Hmg1 activity. In an effort to test this hypothesis, we performed comprehensive sterol profiling on the mutant hmg1 strains. This analysis identified significant accumulations of ergosterol precursors, including eburicol, 4,4-dimethylergosta-8,24(28)-dienol, fecosterol, and episterol, while total cellular ergosterol content was seen to be sustained or, in the case of the hmg1F262del mutation, significantly increased. However, this increase in ergosterol precursors did not result in a subsequent increase in sterol-demethylase expression levels in any of the mutant hmg1 strains. Thus, hmg1 mutation-mediated triazole resistance is not the result of increased sterol-demethylase gene expression.

Taken together, the results of this study identify mutations in the A. fumigatus HMG-CoA reductase gene, hmg1, as a novel and noncanonical genetic determinant of clinical triazole resistance, present among a large proportion of resistant clinical isolates. While the exact mechanism by which mutations in the sterol-sensing domain of hmg1 impart clinical triazole resistance remains unknown at this time, it is tempting to speculate that the negative regulation of Hmg1 activity, which has been shown to be dependent on the sterol-sensing domain in other eukaryotic organisms, may be altered by residue substitutions in the sterol-sensing domain of A. fumigatus Hmg1.

MATERIALS AND METHODS

Isolates, media, and growth conditions used in this study.

Twenty-one previously characterized multitriazole-resistant clinical isolates of A. fumigatus, and five triazole-susceptible control clinical isolates of A. fumigatus, originating from the United States were obtained from the Fungus Testing Laboratory at the University of Texas in San Antonio. The laboratory strains akuBKU80 and Af293 were obtained from the Fungal Genetics Stock Center. All strains and clinical isolates used in this study were maintained on glucose minimal medium (GMM) agar at 37°C. Transformants were selected for using sorbitol (1.2 M)-supplemented GMM agar (SMM) containing 150 mg/liter of hygromycin. All conidia were harvested in sterile water from 3-day-old growth plates, and conidia were enumerated visually using a hemocytometer.

Whole-genome sequencing.

For each isolate, genomic DNA was extracted using the Qiagen DNeasy Plant minikit as previously described (29). DNA concentrations were quantified using both the Qubit fluorometer and the NanoDrop spectrophotometer using the manufacturers’ protocols. Whole-genome libraries were prepared and sequenced at the University of Tennessee Health Science Center Molecular Resource Center. Bioinformatics services were provided by code4DNA. Sequence reads for each sample were aligned using the STAR sequence alignment tool (v2.5.0b) to the A. fumigatus reference sequence (version A_fumigatus_Af293_version_s03-m05-r05) which was downloaded from http://aspgd.org/. Duplicate alignments were marked using Picard (v1.119), and variants were called using Freebayes (v1.1.0) using the haploid population-based model. The Freebayes output vcf file was split into individual sample vcf files using bcftools (v1.3), and variants for each sample were annotated with snpEff (v4.2) using the A. fumigatus reference sequence annotations. Visualization of Hmg1 peptide sequence was performed using Protter (49).

Construction of promoter replacement repair templates and Cas9-RNP for cyp51A, cyp51B, and abcC.

The plasmid pJMR2 was constructed by cloning the proximal 1 kb of the promoter sequence from the A. fumigatus heat shock protein-encoding gene, hspA, into the plasmid pCR-HygR (29). Then promoter replacement transformation repair templates for cyp51A, cyp51B, and abcC were created by PCR using primers (Table S2) which amplified both the hygromycin resistance cassette and the hspA promoter from the plasmid pJMR2, while also introducing approximately 40 bases of homology targeting sequences immediately upstream and downstream of the start codon for the respective gene of interest. PCR products were subsequently purified using the Gene Clean II kit (MP Biomedicals). Cas9-RNP complexes targeting sequences immediately upstream of the open reading frame of each gene of interest were assembled as previously described (29).

Oligonucleotides used in this study. Download Table S2, DOCX file, 0.02 MB.

Assessment of cyp51A, cyp51B, and abcC expression by RT-qPCR.

For assessment of expression in clinical isolates, conidia from each isolate were allowed to germinate overnight in Aspergillus minimal medium incubated at 37°C on an orbital shaker at 250 rpm. For assessment of expression in constitutive overexpression strains constructed in the akuBKU80 background, conidia from each isolate were allowed to germinate overnight in Aspergillus minimal medium incubated at 37°C on an orbital shaker at 250 rpm and then transferred to fresh Aspergillus minimal medium either with voriconazole at 0.125 mg/liter (treated) or without voriconazole (untreated) for an additional 6 h at 37°C on an orbital shaker at 250 rpm. Then RNA was extracted from mature hyphae following liquid nitrogen crush as previously described (35). The RevertAid RT kit (Thermo Scientific) was utilized to synthesize cDNA. PCR master mix and SYBR were utilized to amplify A. fumigatus cyp51A, cyp51B, and abcC from cDNA by PCR per the manufacturer’s instructions. Table S2 lists the gene-specific primers used for PCR. Conditions used for PCR were as follows: AmpliTaq Gold activation at 95°C for 10 min, 40 cycles of denaturation at 95°C for 15 s, and annealing/extension at 60°C. The dissociation curve and threshold cycle (C) was determined using the CFX96 real-time PCR system (Bio-Rad). Changes in gene expression among isolates were then calculated using the 2−ΔΔ method. All experiments were performed in triplicate from biological triplicates. As previously described, ΔC values were used to calculate the standard error (35). Statistical analysis was performed using unpaired, two-tailed, Student’s t test in Prism 8 for Mac OS by GraphPad Software Inc. with significance set at 0.05 and degrees of freedom equal to 10.

Construction of allele replacement repair templates and Cas9-RNP for the cyp51A locus.

The plasmid pCyp51A-HygR was created by cloning the proximal 1.5 kb of promoter sequence and open reading frame of the wild-type cyp51A allele from the genomic reference strain Af293, upstream of the hygromycin resistance cassette in the plasmid pCR-HygR, and then cloning the proximal 1.5 kb of the wild-type cyp51A terminator downstream of the hygromycin resistance cassette (29). Then, the cyp51AWT transformation repair template was created by PCR using primers which amplified the wild-type cyp51A allele (including 500 bases of promoter sequence), the hygromycin resistance cassette, and 500 bases of the cyp51A terminator. Cas9-RNP complexes targeting sequences immediately upstream and approximately 500 bases downstream of the open reading frame of cyp51A were assembled as previously described (29). Target-specific guide sequences are listed in Table S2.

Construction of allele replacement repair templates and Cas9-RNP for the hmg1 locus.

Two-component repair templates consisting of a split hygromycin resistance marker (hphR) and hmg1 alleles of interest were prepared by PCR. Briefly, hmg1 alleles of interest including the open reading frame and approximately 500 downstream bases were amplified by PCR from DI15-98 (S305P), DI15-100 (I412S), DI15-105 (F262del), and akuBKU80 (wild-type control) using a 3′ primer which introduced the terminal 80 bases with homology to the 3′ end of the hphR hygromycin B resistance gene open reading frame. A partial hygromycin B resistance cassette, including the gdpA promoter and a truncated hphR gene lacking the terminal 40 bases, was then amplified by PCR from the pUCGH plasmid using primers that introduced approximately 70 bases of homology with the downstream region of hmg1. Primers used are listed in Table S2. PCR products were subsequently purified using the Gene Clean II kit (MP Biomedicals). Cas9-RNP complexes targeting sequences immediately upstream and approximately 500 bases downstream of the open reading frame of hmg1 were assembled as previously described (29). Target-specific guide sequences are listed in Table S2.

Aspergillus protoplast transformations.

Transformation of A. fumigatus protoplasts was performed as previously described with minor modifications (29). Approximately 2 μg of each portion of applicable transformation repair templates was then mixed with 200 μl of protoplasts, 26.5 μl of Cas9-RNP complexes, and 25 μl of polyethylene glycol (PEG)-CaCl2 buffer (60% [wt/vol] PEG 3350, 50 mM CaCl2·H2O, 450 mM Tris-HCl, pH 7.5) as previously described. Following an incubation on ice for approximately 1 h, 1.25 ml PEG-CaCl2 was added and the mixture was incubated for an additional 20 min at room temperature. The mixture was then diluted to a total volume of 3 ml with STC buffer (1.2 M sorbitol, 7.55 mM CaCl2·H2O, 10 mM Tris-HCl, pH 7.5) and plated on SMM agar plates. Plates were then incubated overnight prior to being overlaid with SMM top agar (GMM supplemented with 1.2 M sorbitol and 0.7% [wt/vol] agar) supplemented with hygromycin (final concentration of 150 g/ml), and the plates were incubated at 37°C for 3 days. Transformants were isolated to separate plates containing selective agar, genomic DNA was extracted, and successful CRISPR-Cas9 editing was confirmed by PCR screens using gene-specific primers and Sanger sequencing.

Clinical antifungal susceptibility testing.

Susceptibilities for amphotericin B, voriconazole, isavuconazole, itraconazole, and posaconazole were determined for all isolates in accordance with CLSI M38-A2 methodology utilizing broth microdilution in RPMI (31). Each antifungal was obtained from the appropriate manufacturer. All agents were suspended in dimethyl sulfoxide (DMSO) for preparation of stock solutions.

Comprehensive sterol profiling.

Fresh conidial suspensions of each strain to be studied were prepared in saline with Tween 80 from Aspergillus minimal medium agar plates. Conidia were grown in 25 ml RPMI supplemented with 0.2% glucose buffered to pH 7.0 with MOPS for 24 h in an orbital shaker at 180 rpm and 35°C. Cells were then flash-frozen using liquid nitrogen, dry weights were obtained, and alcoholic KOH was utilized to extract nonsaponifiable lipids. A vacuum centrifuge (Heto) was then used to dry samples, prior to derivatization by the addition of 200 μl of anhydrous pyridine (Sigma), 100 μl of N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA)–10% trimethylsilyl (TMS) (Sigma), and 2 h of heating at 80°C. Gas chromatography-mass spectrometry (GC-MS) (Thermo 1300 GC coupled to a Thermo ISQ mass spectrometer; Thermo Scientific) was then used to analyze and identify TMS-derivatized sterols. Known standards were referenced for fragmentation spectra and retention times. Sterol profiles for each isolate were then created using Xcalibur software (Thermo Scientific) to analyze GC-MS data (48). Statistical analysis of both sterol profiles and total ergosterol per dry weight was performed using GraphPad Prism 7. In all cases, 6 independent biological replicates were measured and included in analysis, and two-tailed unpaired t tests were performed in Prism 8 for Mac OS by GraphPad Software Inc. with significance set at 0.05 and degrees of freedom equal to 10.

Data availability.

Whole-genome sequencing data files for the 26 Aspergillus fumigatus isolates have been deposited in NCBI SRA under the accession number PRJNA491253.