Multiresistance to Nonazole Fungicides in Aspergillus fumigatus TR₃₄/L98H Azole-Resistant Isolates

ABSTRACT

Drug resistance is a worldwide problem affecting all pathogens. The human fungal pathogen Aspergillus fumigatus coexists in the environment with other fungi targeted by crop protection compounds, being unintentionally exposed to the selective pressure of multiple antifungal classes and leading to the selection of resistant strains. A. fumigatus azole-resistant isolates are emerging in both clinical and environmental settings. Since their approval, azole drugs have dominated clinical treatment for aspergillosis infections and the agriculture fungicide market. However, other antifungal classes are used for crop protection, including benzimidazoles (methyl benzimidazole carbamates [MBCs]), strobilurins (quinolone oxidation inhibitors [QoIs]), and succinate dehydrogenase inhibitors (SDHIs). Mutations responsible for resistance to these fungicides have been widely researched in plant pathogens, but resistance has not been explored in A. fumigatus. In this work, the genetic basis underlying resistance to MBCs, QoIs, and SDHIs was studied in azole-susceptible and -resistant A. fumigatus strains. E198A/Q and F200Y mutations in β-tubulin conferred resistance to MBCs, G143A and F129L substitutions in cytochrome b conferred resistance to QoIs, and H270R/Y mutations in SdhB conferred resistance to SDHIs. Characterization of susceptibility to azoles showed a correlation between strains resistant to these fungicides and the ones with tandem-repeat (TR)-based azole resistance mechanisms. Whole-genome sequencing analysis showed a genetic relationship among fungicide multiresistant strains, which grouped into subclusters that included only strains carrying the TR-based azole resistance mechanisms, indicating a common ancestor/evolution pattern and confirming the environmental origin of this type of azole-resistant A. fumigatus.

INTRODUCTION

Aspergillus fumigatus is an opportunistic human fungal pathogen, with a worldwide distribution, that can affect immunocompromised individuals, causing a broad range of clinical manifestations encompassed under the name of aspergillosis (1, 2). The most serious clinical manifestation of aspergillosis, invasive pulmonary aspergillosis (IPA), represents a major cause of morbidity and mortality, largely due to the difficulty of diagnosis and late initiation of antifungal therapy (3). First-line treatment and prophylaxis for aspergillosis rely on the employment of an antifungal class called azoles (4) that target the enzyme 14-α sterol demethylase (Cyp51) involved in the ergosterol biosynthesis (5).

Drugs targeting the 14-α sterol demethylase enzyme, also called demethylation inhibitors (DMIs), are used not only in the clinical setting but also in the environment, preventing crop damage by plant pathogens (6). A. fumigatus azole resistance is acquired through selective pressure that can develop in two different scenarios: a clinical route generated during long periods of exposure to azole treatments (7) or an environmental origin due to the extended use of DMI drugs in agriculture (8). In both scenarios, strains with different resistance mechanisms and different azole susceptibility profiles can be selected, although in both cases, the acquired resistance to azole antifungals is based on mutations in the cyp51A gene (5). Currently, the most frequent azole resistance mechanisms are based on tandem-repeat (TR) insertions in the promoter with mutations in the coding sequence of the cyp51A gene (TR₃₄/L98H and, less frequently, TR₄₆/Y121F/T289A and TR₅₃) (5). Strains harboring these resistance mechanisms can infect azole-naive immunocompromised patients, seriously compromising their treatment options (9, 10). Although the number of cases of azole-resistant A. fumigatus recovered from clinical samples is still limited, azole resistance mechanisms continue to spread worldwide, threatening the effectiveness of this important antifungal class in aspergillosis treatment (11).

A. fumigatus can grow in a wide variety of ecological niches and plays a role in recycling organic matter from the soil (1). In the environment, A. fumigatus is exposed to strong antifungal selective pressure from not only DMI fungicides but also other classes of fungicides used for crop protection. Different classes of antifungals are currently commercialized, with each of them targeting different mold species and with diverse mechanisms of action (12). DMIs, targeting the fungal enzyme Cyp51, have dominated the agricultural market since their approval in the 1970s (13), but other drugs such as methyl benzimidazole carbamates (MBCs), quinolone oxidation inhibitors (QoIs), and succinate dehydrogenase inhibitors (SDHIs) are widely employed as well (14).

The continuous exposure of fungal species, including A. fumigatus, to chemical compounds in the field favors the emergence of strains resistant to different classes of fungicides. DMIs, MBCs, QoIs, and SDHIs are all site-specific inhibitors, and resistance to them arises due to point mutations of the target site, which can happen only if the activity of the protein is not significantly affected and the fitness of the strain is not compromised (14–16). Reports of resistance to DMIs in A. fumigatus associated with their use in agriculture are being acknowledged (6, 8, 13); however, only one study has recently reported this phenomenon regarding other classes of antifungal compounds in environmental strains of A. fumigatus (17).

In 1966, the SDHI fungicide class was added to the market (12, 18, 19). This fungicide class acts by blocking the Krebs cycle by binding to the ubiquinone-binding site of mitochondrial complex II, disrupting the electron transport chain (12, 20). Complex II is a heterotetramer, and point mutations in the sdhB subunit have been associated with the development of resistance to SDHIs in several fungal pathogens (18–20). MBCs, introduced to the market in the 1970s, are heterocyclic compounds that target the protein encoded by the β-tubulin gene (benA) (21–23). Point mutations associated with resistance to MBCs in the coding sequence of benA have been acknowledged in more than 100 fungal species (24–26); thus, MBCs are currently considered fungicides with a high risk of developing resistance (http://www.frac.info/home). QoIs, or strobilurins, launched in the market in 1996, target the cytochrome bc₁ enzyme complex (complex III) in the mitochondrial electron transport chain (15, 27). Resistance mechanisms against QoIs have been described in other fungal species as point mutations in the mitochondrial cytochrome b gene (cytB) (28, 29).

To date, the effects on resistance development produced by these antifungal drugs have been studied in several plant pathogens, but to our knowledge, they have only recently started to be studied in the human fungal pathogen A. fumigatus (17). We hypothesized whether A. fumigatus DMI resistance is related to resistance to other antifungals frequently used for crop protection. In this work, the susceptibility of a large collection of A. fumigatus clinical strains, including DMI-susceptible and -resistant strains, was tested against compounds of three antifungal classes, MBCs, SDHIs, and QoIs. Molecular characterization was performed by sequencing the three antifungal targets benA, cytB, and sdhB. In addition, we use whole-genome sequencing (WGS) analysis to compare the genetic relationships of multifungicide-resistant A. fumigatus strains from different geographic origins. A phylogenetic tree showing the relationship among A. fumigatus strains resistant to the four fungicide classes was constructed.

RESULTS

Antifungal susceptibility to MBCs targeting the benA gene and sequence analysis of the A. fumigatus benA gene.

A. fumigatus susceptibility to benomyl (BNY) and carbendazim (CBZ) was tested using 60 strains (Table 1). The MIC ranges to consider BNY and CBZ susceptible were established based on the MIC values obtained with A. fumigatus wild-type (WT) reference strains used in the mycology laboratory. Using these ranges, two-thirds of the A. fumigatus strains included in this study could be considered BNY and CBZ susceptible. However, 23 strains had MIC values of >32 mg/liter for both BNY and CBZ and therefore were considered methyl benzimidazole carbamate (MBC) resistant (Table 1).

TABLE 1.

Amino acid substitutions in Cyp51A, β-tubulin, cytochrome b, and SdhB and MIC or MEC ranges of agricultural antifungal drugs in a collection of 60 A. fumigatus strains, including azole-resistant and azole-susceptible isolates^a

Strain category (no. of strains tested) and no. of strains	Amino acid substitution(s) in:				MIC/MEC range of agricultural antifungal drug (mg/liter)
	Cyp51A	β-Tub	CytB	SDHB	MBC		QoI		SDHI
					BNY	CBZ	AZB	PYB	BCL	FLP
Azole-susceptible strains with no mutations (14)
14	WT	WT	WT	WT	2	0.25–1	0.25–2	0.125–1	0.25–2	1–8
Azole-resistant strains with no mutations (21)
4	R-non-Cyp51A	WT	WT	WT	2	0.25	0.25–0.5	0.125–1	0.5–2	0.5–2
9	Point mutationsb	WT	WT	WT	2–4	0.25–1	0.125–2	<0.064–2	0.5–4	1–8
8	TR34/L98H	WT	WT	WT	2	0.25	0.125–1	0.25–0.5	0.5–2	1–4
Azole-resistant strains with mutations in β-tub (12)
4	TR34/L98H	E198A	WT	WT	>32	>32	1–2	0.5–4	0.25–1	2
2	TR34/L98H	E198Q	WT	WT	>32	>32	0.125–2	<0.064–0.5	0.125–2	0.25–8
6	TR34/L98H	F200Y	WT	WT	>32	>32	0.5–2	0.5–1	0.25–1	0.25–4
Azole-resistant strains with mutations in CytB (2)
2	TR34/L98H	WT	G143A	WT	2	0.25–1	>32	>32	1	1–2
Azole-resistant strains with mutations in β-tub and CytB (4)
1	TR34/L98H	F200Y	F129L	WT	>32	>32	>32	2	1	2
2	TR34/L98H	F200Y	G143A	WT	>32	>32	>32	>32	0.25–0.5	0.5–2
1	TR34/L98H	E198A	G143A	WT	>32	>32	>32	>32	0.5	2
Azole-resistant strains with mutations in β-tub and SDHB (4)
1	TR34/L98H	F200Y	WT	R51G	>32	>32	0.5–4	0.5–1	0.5–2	1–2
3	TR34/L98H	F200Y	WT	H270R	>32	>32	0.5–1	0.25–0.5	>32	1–8
Azole-resistant strains with mutations in β-tub, CytB, and SDHB (3)
1	TR46/F121Y/T289A	F200Y	G143A	H270R	>32	>32	>32	>32	>32	1–2
2	TR46/F121Y/T289A	F200Y	G143A	H270Y	>32	>32	>32	>32	>32	0.25–8

^aβ-tub, β-tubulin; CytB, cytochrome b; SDHB, succinate dehydrogenase; MBC, benzimidazoles; QoI, strobilurin; R-non-Cyp51A, azole resistant strains without Cyp51A mutations; SDHB, succinate dehydrogenase B; SDHI, succinate dehydrogenase inhibitor; BNY, benomyl; CBZ, carbendazim; AZB, azoxystrobin; PYB, pyraclostrobin; BCL, boscalid; FLP, fluopyram.

^bPoint mutations include isolates with mutations G54, M220, and G448.

PCR amplification and sequencing of the benA gene from 138 A. fumigatus strains showed two different amino acid substitutions: the substitution of a glutamic acid (E) for an alanine (A) or a glutamine (Q) at position 198 of the amino acid sequence and the change of a phenylalanine (F) for a tyrosine (Y) at position 200 of the protein. The E198A/Q substitution was harbored by 4 and 2 strains, respectively, and the F200Y mutation was present in 16 isolates. Strains harboring these substitutions showed resistance to MBCs by antifungal susceptibility testing (AFST) (Table 1).

Antifungal susceptibility to QoIs and sequence analysis of the A. fumigatus cytB gene.

The results of AFST for the quinolone oxidation inhibitors (QoIs) azoxystrobin (AZB) and pyraclostrobin (PYB) are shown in Table 1. MIC ranges for AZB and PYB susceptibility were established based on the MICs obtained for the A. fumigatus wild-type reference strains. Based on these ranges, we can consider most of our strains susceptible to AZB, except for nine strains that showed MICs of over 32 mg/liter and therefore were considered AZB resistant. All AZB-resistant strains but one were also PYB resistant (Table 1).

PCR amplification and sequencing of the cytB gene in the A. fumigatus collection of 138 strains revealed several polymorphisms responsible for synonymous mutations. The I119V polymorphism, in which an isoleucine (I) is substituted for valine (V) at position 119 of the protein, was present in QoI-susceptible and -resistant strains, so its implication in resistance to QoIs was discarded. Two polymorphisms were found in the strains that showed resistance to both QoIs: the change of a glycine (G) for an alanine (A) at position 143 (G143A) present in eight strains and the substitution of a phenylalanine (F) for a leucine (L) at position 129 (F129L) in one strain. This strain harboring the F129L substitution was resistant to AZB but not to PYB.

Antifungal susceptibility to SDHIs and sequence analysis of the A. fumigatus sdhB gene.

AFST was performed against the drugs boscalid (BCL) and fluopyram (FLP). MIC ranges for susceptibility testing were compared to the MIC values obtained from the A. fumigatus reference strains to consider susceptibility. Based on these criteria, six strains were resistant to succinate dehydrogenase inhibitors (SDHIs), with MICs of over 32 mg/liter (Table 1).

PCR amplification and sequencing of the sdhB gene in 138 A. fumigatus strains revealed three amino acid modifications. The R51G substitution was present in only one strain and was demonstrated not to have any impact on susceptibility to BCL or FLP. An amino acid substitution at position 270 of histidine (H) for an arginine (R) or a tyrosine (Y) (H270R/Y) was present in four and two strains, respectively, which seems to be associated with resistance only to BCL but not to FLP.

In addition, all A. fumigatus strains with mutations of benA, cytB, or sdhB had mutations of Cyp51A (TR₃₄/L98H or TR₄₆/F121Y/T289A) and were also multiresistant to clinical azole drugs as well as to environmental DMIs (8).

Polymorphisms of benA, cytB, and sdhB genes in A. fumigatus strains analyzed by whole-genome sequencing.

We extended the search for mutations in the benA, cytB, and sdhB genes to a larger collection of 205 A. fumigatus genomes, including 163 whole-genome sequences that had been previously sequenced in our laboratory or downloaded from public databases (30) and 42 strains that were received later (Table 2).

TABLE 2.

Analysis of the benA, cytB, and sdhB polymorphisms and amino acid substitutions found in our set of 205 A. fumigatus azole-susceptible and azole-resistant strains^a

Gene	Nucleotide position (cDNA)	Codonb	Amino acid change	No. of Azl-S strains	No. of Azl-R strains	% of strains with mutation
benA	A593C	gAg/gCg	E198A	0	5	2.5
	G592C	Gag/Cag	E198Q	0	3	1.5
	T599A	tTc/tAc	F200Y	1	29	14
cytB	A355G	Ata/Gta	I119V	18	57	36.6
	C387A	ttC/ttA	F129L	0	1	0.5
	G428C	gGt/gCt	G143A	0	8	3.9
sdhB	A151G	Agg/Ggg	R51G	0	1	0.5
	C808T	Cac/Tac	H270Y	0	3	1.5
	A809G	cAc/cGc	H270R	0	4	2

^aOnly polymorphisms involving nonsynonymous mutations are shown. Azl, azole; S, susceptible; R, resistant.

^bUppercase letters in the codon column indicate the base change in the codon.

Genome analysis revealed that benA is a highly conserved gene, and the same two nonsynonymous mutations previously detected were found (E198A/Q and F200Y). In total, the E198A substitution was present in 5 strains (2.5%), E198Q was found in 3 strains (1.5%), and the F200Y mutation was present in 29 strains (14%) (Table 2). Both amino acid substitutions were found in azole-resistant A. fumigatus strains harboring the resistance mechanism TR₃₄/L98H or TR₄₆/Y121F/T289A, except for one Japanese azole-susceptible strain that harbored an F200Y mutation in benA.

Three nonsynonymous mutations were found in the cytB gene. The I119V mutation found in 36.6% of the strains was present in azole-susceptible as well as azole-resistant strains. The F129L mutation was present in only one azole-resistant strain (0.5%), and the G143A mutation was identified in eight azole-resistant strains (3.9%) (Table 2). Both substitutions were always found in azole-resistant A. fumigatus strains harboring the resistance mechanism TR₃₄/L98H or TR₄₆/Y121F/T289A.

The sdhB gene presented a lower frequency (4%) of nonsynonymous mutations. Of the total genomes included in this analysis, the R51G mutation was found in only one azole-resistant strain (0.5%), the H270Y substitution was detected in three strains (1.5%), and the H270R substitution was detected in four strains (2%) (Table 2). All the amino acid substitutions were found only in azole-resistant A. fumigatus strains harboring the resistance mechanism TR₃₄/L98H or TR₄₆/Y121F/T289A.

Phylogenic tree representation of a collection of 163 A. fumigatus genomes.

The collection of 163 A. fumigatus strains included in the whole-genome sequencing analysis (30) were clustered according to their genetic proximity. Figure 1 represents the relatedness among strains, indicating their susceptibility to azoles and the existence of mutations in the cyp51A (separating single point mutations from tandem-repeat insertions), benA, cytB, and sdhB genes (see also Table S2 in the supplemental material).

FIG 1.

Phylogenic tree representation of a whole-genome alignment of a collection of 163 A. fumigatus genomes clustered according to their genetic proximity. SP, Spain; PT, Portugal; CN, Canada; UK, United Kingdom; IT, Italy; JP, Japan; NT, The Netherlands; DN, Denmark; IN, India; FR, France. Azole resistance is marked in red, strains harboring azole resistance mechanisms based on tandem-repeat insertions in the promoter of the cyp51A gene are marked in blue, and azole resistance mechanisms based on point mutations in the cyp51A gene are marked in pink. Mutations in the three fungicide targets are also color-coded: green for benA, orange for cytB, and yellow for sdhB.

The A. fumigatus strains were divided into four clear clusters. Cluster I included azole-susceptible strains and azole-resistant strains with single point mutations in the coding sequence of cyp51A. Cluster II included azole-susceptible strains and azole-resistant strains with resistance mechanisms based on tandem-repeat insertions in the promoter of cyp51A as well as strains with single point mutations. However, strains harboring tandem-repeat insertions were grouped into three well-supported subclusters within cluster II (subclusters II.1, II.2, and II.3), in which only strains with TR-based Cyp51A mutations were included. A set of azole-susceptible strains with five cyp51A modifications (F46Y, M172V, N248T, D255E, and E427K) grouped into cluster III, while cluster IV included azole-susceptible strains with three cyp51A modifications (F46Y, M172V, and E427K); neither of them had any mutations in the benA, cytB, or sdhB gene.

All except two of the strains with mutations in the benA, cytB, and sdhB genes belonged to cluster II, and all but one of them also harbored a TR insertion in the promoter of cyp51A. In addition, all of the strains with mutations in the benA, cytB, and sdhB genes were grouped in subclusters II.1 and II.3 along with tandem-repeat strains. As mentioned above, only two strains with a tandem-repeat insertion in cyp51A and mutations in the genes benA, cytB, and sdhB were detected in cluster I (TR₄₆/Y121F/T289A and wild-type Cyp51A), and no other strains with mutations in these three genes were found outside subclusters II.1 and II.3, in neither azole-susceptible nor azole-resistant strains. All of the strains included in subclusters II.1 and II.3 had a substitution in benA, except one (F16216). Substitutions in cytB and sdhB were less frequent (Fig. 1).

DISCUSSION

Aspergillus fumigatus is an opportunistic human pathogen that coexists in its natural environment with plant-pathogenic molds and as a consequence is exposed to strong selective pressure from different fungicide classes. The hypothesis that the continuous exposure of A. fumigatus to DMI antifungals in the environment favors the development of strains resistant to clinical azoles is being acknowledged by several authors (5). However, only one study has recently reported resistance to other classes of fungicides in A. fumigatus from environmental origins (17). In the present study, a collection of A. fumigatus strains, including azole-susceptible and azole-resistant isolates, was tested against three fungicide classes, MBCs, QoIs, and SDHIs, including 38.4% of strains resistant to DMIs (8).

Eighteen percent of the A. fumigatus strains included in our study had mutations in the benA gene and were resistant to MBCs. The amino acid substitutions E198A, E198Q, and F200Y found in β-tubulin have been previously described in several plant-pathogenic fungi such as Venturia inaequalis, Botrytis cinerea, or Podosphaera xanthii, with very high prevalences that can reach up to 90% (15, 26, 31, 32). All of them showed higher prevalences of resistance than that observed in A. fumigatus. However, we have to consider that the A. fumigatus azole-resistant strains in this study were mainly obtained from the clinical setting. An environmental search looking specifically for A. fumigatus isolates resistant to any of the fungicide classes will probably provide higher percentages of resistance.

Resistance to QoIs was present in 4.4% of the strains harboring two different mutations in the cytB gene (G143A and F129L). The amino acid substitution G143A in the cytochrome b protein has been one of the most described mutations in plant-pathogenic isolates resistant to QoIs to date, including Mycena galopoda, Erysiphe necator, or V. inaequalis (33, 34). Frequencies of resistance to QoIs are also quite high in plant pathogens, with percentages that can go up to 80 to 90% (15). Furthermore, the F129L mutation has also been detected in resistant isolates of several fungal species, including Cercospora beticola, Pythium aphanidermatum, Pyricularia grisea, and the genus Pyrenophora (29, 34, 35). According to the results of the susceptibility testing carried out in this study, the F129L mutation is responsible for resistance to strobilurin but not pyraclostrobin, which correlates with data from previous studies (29, 34, 35).

The SdhB substitution H270R/Y was found in 3.5% of the A. fumigatus isolates. This mutation confers resistance to boscalid but not fluopyram, as previously described for other plant pathogens (15). The location of the SdhB amino acid at position 270 in A. fumigatus changes depending on the fungal species under study. In B. cinerea, the corresponding H272Y/R substitution has also been related to resistance to boscalid but not fluopyram (20). Rates of resistance of A. fumigatus to SDHIs, as with other fungicides, are considerably lower than those of other plant pathogens (15). To our knowledge, the R51G substitution has not been described in the literature previously. In fact, the susceptibility results obtained with boscalid and fluopyram do not lead us to think that the R51G substitution has any relevance in conferring resistance to SDHIs in A. fumigatus.

QoIs and SDHIs are fungicides that block the electron transport chain by targeting mitochondrial genes, which are less exposed to DNA repair mechanisms than nuclear DNA (16), favoring the rapid emergence of mechanisms of resistance against mitochondrial target drugs. However, when mitochondrial genes are implicated, the level of resistance to a drug depends not only on the nature of the amino acid substitution itself but also on the heteroplasticity of the cell, which is defined by the number of mitochondria inside the cell that harbor the mutated allele (36–38). In the case of the G143A mutation, the risk of strobilurin resistance is determined by the percentage of A143 alleles in the fungal population and not only the presence of this substitution (39).

In mammals, mitochondrial DNA does not have introns (40); however, in fungi, the existence of introns in the mitochondrial DNA depends on the species (36). In some species such as Puccinia spp., Alternaria alternata, or Saccharomyces cerevisiae, an intron in the cytB gene at a position immediately after the G143 amino acid has been described (29). As a consequence, the A143 allele is not possible in these species since its base position is located at the intron boundary, which is crucial for correct intron splicing (29). In the case of A. fumigatus, we ruled out the existence of an intron at this position (results not shown). Furthermore, we have shown that cytB with an A143 allele encodes a functional protein since resistance to azoxystrobin and pyraclostrobin was demonstrated in isolates with this mutation.

The selection of resistant isolates depends on two factors: the selection pressure of the fungicide and the fitness cost associated with the change of protein functionality. In general, mutated strains tend to have a fitness disadvantage in the absence of antifungal drugs. Previous studies about the relationship between resistance and fitness cost in fungal species are contradictory, as in some species, like Phakopsora pachyrhizi, the G143A mutation in CytB is correlated with a fitness penalty, while in others, like C. beticola or B. cinerea, it is not (20, 35, 41). Similarly, in Pyrenophora teres, the presence of the F129L substitution in CytB did not correlate with a fitness cost (34). The same has been described for mutations in the sdhB gene; depending on the sdhB amino acid substitution and the fungal species implicated, results of fitness studies show contradictory results (20). In A. fumigatus, adaptive mutations might favor the survival of the mutated strains in the environment for generations without any associated fitness cost. The rapid dispersal of azole-resistant A. fumigatus strains with the TR₃₄/L98H genotype in Asia also supports the hypothesis that in natural environments, these strains have fitness comparable to or even higher than that of wild-type strains (42). Nevertheless, this hypothesis needs further consideration and an in-depth study.

WGS analysis of A. fumigatus strains from very diverse geographic origins (Europe, Japan, Canada, and India), isolated 20 years apart, showed four differentiated clusters including azole-susceptible and azole-resistant strains (Fig. 1). Clusters I and II could be further divided into well-defined subclusters. A remarkable finding was that all genomes harboring the TR₃₄/L98H Cyp51A alleles were grouped into cluster II, and more specifically, they were included in three subclusters within cluster II (clusters II.1, II.2, and II.3) that harbored only strains with TR insertions in cyp51A. This particular clustering of TR strains was independent of their geographic origin, isolation year, or clinical or environmental origin, and it suggests that the selective phenomenon occurring among these isolates does not imply a lack of fitness, but further experiments are necessary to clarify this important subject. Our A. fumigatus WGS collection has only two genomes belonging to strains with the TR₄₆/Y121F/T289A Cyp51A azole resistance mechanism, one located in cluster I and the other located in cluster II together with TR₃₄/L98H strains. The absence of more strains with TR₄₆/Y121F/T289A makes it difficult to draw conclusions about this specific resistance mechanism, although the presence of mutations in all the benA, cytB, and sdhB genes strongly supports evolution similar to that of the TR₃₄/L98H strains. However, until we have more whole genomes of these strains sequenced, we cannot reach any further conclusions.

A remarkable finding was that, with the exception of two strains included in cluster I, strains harboring mutations in the benA, cytB, and sdhB genes, resistant to MBCs, QoIs, and SDHIs, respectively, grouped into two subclusters of cluster II (subclusters II.1 and II.3) composed of TR₃₄/L98H Cyp51A strains. Subclusters II.1 and II.3 contained 26 and 6 strains, respectively. Among these 32 azole-resistant isolates, 31 were also resistant to MBCs, 13 were resistant to QoIs, and 6 were resistant to SDHIs. Only two strains included in subcluster II.1 were resistant to all four antifungals tested, harboring mutations in all the fungicide target genes (Fig. 1). Multifungicide resistance is a well-known phenomenon in some plant pathogen species (14) (http://www.frac.info/home), with some fungal species showing resistance to up to six fungicide classes (43). The A. fumigatus isolates included in this work showed 25% resistance to more than one antifungal class, and some particular isolates even had a profile of resistance to four different fungicidal classes. This percentage is quite high considering that A. fumigatus is not the target pathogen of these fungicides, the origin of these strains was mainly clinical, and these drugs are never used in the clinical setting. A recent study performed in the United Kingdom looking for environmental hot spots of A. fumigatus azole resistance reported results for multifungicide resistance similar to those of our study. Some of the mutations found, F200Y in benA, G143A in cytB, and H270Y in sdhB, were also found in combination with TR-based Cyp51A resistance mechanisms (17). The results of this study on environmental strains and our results obtained from clinical isolates confirm the environmental origin of strains carrying this type of azole resistance mechanism.

In plant pathogens, the application of single-site-specific fungicides, including rotations and mixtures for crop protection, seems to favor the selection of multidrug-resistant isolates (44). The development of multifungicide resistance in fungi is believed to be via independent rounds of selection (45). This is supported by the accumulation of mutations in the target genes of the respective fungicides causing resistance to them (46). Although some fungi showed a predisposition to selection for resistance in isolates that were already resistant to an unrelated fungicide (44, 46), this phenomenon implies that fungi might be selecting for not only resistance but also increased genetic plasticity that enables accelerated resistance development (45). Many of these nonazole fungicides have activity against A. fumigatus, and although they are not employed to target A. fumigatus, the selection of resistant strains seems to mirror what is happening with most of the plant pathogens. The multiresistance profile of the strains within subclusters II.1 and II.3 (Fig. 1) carrying several mutations suggests a common origin based on selection due to exposures to multiple environmental fungicides. Future studies aimed at understanding correlations between mutator genotypes and multifungicide-resistant phenotypes are needed. The genetic relationship among the multiresistant strains supports the idea that these A. fumigatus strains have a common background and may have a common ancestor reflecting biological adaptations to the selective pressure of fungicide applications. Whether or not this A. fumigatus genetic background implies improved fitness for the strain needs to be studied and clarified.

In conclusion, the application of fungicides for crop protection seems to be favoring the selection of A. fumigatus multidrug-resistant isolates. Whatever the mechanism for resistance selection applied, this finding would confirm the environmental origin of the TR-based azole resistance mechanisms since the antifungals tested here (MBCs, QoIs, and SDHIs) are not used in clinical settings. The current evidence supporting environmental resistance selection is a very important finding that will help to design and change fungicide applications in the environment and medical applications in order to contain the spread of drug resistance. If improved strain fitness is associated with fungicide multiresistance, this finding could have severe consequences in the future, for both clinical and agricultural uses of azoles and other antifungals.

MATERIALS AND METHODS

A. fumigatus strain collection and DNA extraction.

A total of 100 Spanish unrelated azole-resistant A. fumigatus strains with known azole resistance mechanisms were included in this work, as were 38 azole-susceptible strains. The A. fumigatus reference strains AF293 and CBS144.89 (CEA10), the whole genomes of which have been sequenced, were used as control strains. In addition, other reference stains such as A. fumigatus ATCC 204305 (reference strain for European Committee on Antimicrobial Susceptibility Testing [EUCAST] susceptibility testing) and strains CM237 and ATCC 46645, frequently used in A. fumigatus laboratories, were also included. Apart from the 138 strains available in our laboratory, other genomes downloaded from databases were used to perform this study. See Fig. S1 in the supplemental material for an explanation of how the isolates were selected and used.

Conidia from each strain were cultured in 3 ml of broth containing 0.3% yeast extract and 1% peptone (Difco, Soria Melguizo, Madrid, Spain) with 2% glucose (Sigma-Aldrich Química, Madrid, Spain) (GYEP broth) and grown overnight at 37°C, after which mycelium mats were harvested and genomic DNA was extracted as described previously (47). All isolates were identified to the species level by PCR amplification and sequencing of the internal transcribed spacer 1 (ITS1)-5.8S-ITS2 regions and a portion of the β-tubulin gene (48).

Characterization of molecular mechanisms of azole resistance in a collection of A. fumigatus strains.

To study the mechanisms associated with azole resistance, the full coding sequence of the cyp51A gene, including its promoter sequence, was amplified using PCR conditions described previously (49). A 1-kb molecular DNA ladder (Promega, Spain) was used for all electrophoresis analyses, and both strands were sequenced with the BigDye terminator cycle sequencing kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s instructions. To exclude the possibility that any change identified in the sequences was due to PCR-induced errors, each isolate was independently analyzed twice. All cyp51A DNA sequences were edited and assembled using the Lasergene software package (DNAStar Inc., Madison, WI, USA), and the A. fumigatus reference strain CBS144.89 (NCBI accession number AAK73659.1) was used for comparison.

PCR and sequencing conditions for benA, cytB, and sdhB gene amplification.

For the determination of gene modifications associated with fungicide resistance, the full coding sequences of the benA, cytB, and sdhB genes were amplified and sequenced. PCR mixtures contained 0.5 μM each primer, 0.2 μM deoxynucleoside triphosphate (Roche, Madrid, Spain), 5 μl of 10× PCR buffer, 2 mM MgCl₂, 5.2% dimethyl sulfoxide (DMSO), 2.5 U of Taq DNA polymerase (Applied Biosystems, CA, USA), and 100 to 200 ng of DNA in a final volume of 50 μl. The samples were amplified in a GeneAmp 9700 PCR system (Applied Biosystems, CA, USA). The parameters used were 1 cycle of 5 min at 94°C and then 35 cycles of 30 s at 94°C; 45 s at 60°C for benA, 56°C for cytB, or 58°C for sdhB; and 2 min at 72°C, followed by a final cycle of 5 min at 72°C. PCR products were analyzed and sequenced as described above. Primers used to amplify and sequence the genes included in this study are listed in Table S2 in the supplemental material.

Clinical antifungal drug susceptibility testing.

Antifungal susceptibility testing (AFST) was performed according to EUCAST broth microdilution reference method 9.3.1 (50). A representation of 60 A. fumigatus isolates among the collection of 138 strains was tested, including a group of 14 wild-type strains (wild type for all genes under study, cyp51A, benA, cytB, and sdhB) and all the strains that harbored mutations in any of the genes of interest (Fig. S1). Antifungal drugs used were amphotericin B (Sigma-Aldrich Química, Madrid, Spain), itraconazole (Janssen Pharmaceutica, Madrid, Spain), voriconazole (Pfizer SA, Madrid, Spain), posaconazole (Schering-Plough Research Institute, Kenilworth, NJ), and isavuconazole (Basilea Pharmaceutica, Basel, Switzerland). The final concentrations tested ranged from 0.03 to 16 mg/liter for amphotericin B and 0.015 to 8 mg/liter for the four azoles tested. Aspergillus flavus ATCC 204304 and A. fumigatus ATCC 204305 were used as quality control strains in all tests performed. MICs were visually read after 24 and 48 h of incubation at 37°C in a humid atmosphere. MIC determinations were performed at least twice for each isolate (biological duplicates). Clinical breakpoints for interpreting AFST results established by EUCAST (51) were used for classifying the A. fumigatus strains as azole susceptible or resistant.

Environmental fungicide susceptibility testing.

AFST was also performed against six different fungicides used in agriculture according to the EUCAST methodology described above. The antifungals tested were (i) two methyl benzimidazole carbamate (MBC) compounds targeting β-tubulin, benomyl and carbendazim; (ii) two quinolone oxidation inhibitors (QoIs), azoxystrobin and pyraclostrobin, targeting cytochrome b of mitochondrial complex III; and (iii) two succinate dehydrogenase inhibitors (SDHIs), boscalid and fluopyram, targeting succinate dehydrogenase. All antifungal compounds were purchased from Sigma-Aldrich Química, Madrid, Spain.

All drugs were dissolved in DMSO and autosterilized for 30 min at room temperature, as stated in the EUCAST protocol for clinical azoles (51). The final concentrations tested ranged from 0.06 to 32 mg/liter. Visual reading at 24 h was used to determine the MICs or minimum effective concentrations (MECs). As breakpoints for interpreting AFST results for environmental fungicides have not been established yet, isolates were considered susceptible or resistant based on the MIC/MEC shown by the group of A. fumigatus WT reference strains. AFST was performed at least twice for each isolate (biological duplicates). The MIC was recorded for benomyl and carbendazim, whereas the MEC was used for the other four fungicides. The MIC is defined as the lowest concentration of drug that yields no growth, and the MEC is the lowest concentration of drug that results in macroscopic changes of filamentous growth to microcolonies or granular growth compared with growth control wells (52).

AFST had been previously performed against a set of demethylation inhibitor (DMI) drugs for the collection of 138 A. fumigatus strains (8) according to the same procedures as the ones described above for clinical and environmental fungicides.

Aspergillus fumigatus whole-genome sequence alignment.

A search for variants or mutations in the benA, cytB, and sdhB genes was performed using a collection of 163 A. fumigatus whole-genome sequences that had been previously sequenced in our laboratory using the Nextera XT library prep kit (Illumina Inc., San Diego, CA, USA) as described previously or obtained from public databases (Fig. S1) (30). All data used in this analysis included information about clinical antifungal susceptibility and azole resistance mechanisms of the strains (Table S1).

The Illumina reads were trimmed using Trimmomatic (version 0.32) (53). The sequencing adapters and sequences with low-quality scores on 30 ends (Phred score [Q], <20) were trimmed. Raw Illumina WGS reads were quality checked by performing quality control with FastQC (version 0.11.3; Babraham Institute). Data sets were analyzed against the A. fumigatus A1163 reference genome (GenBank accession number ABDB00000000.1) using WGS-Outbreaker v1.0 (Instituto de Salud Carlos III, Madrid, Spain) (https://github.com/BU-ISCIII/WGS-Outbreaker) with default parameters. The pipeline comprised all steps needed for single nucleotide variant (SNV) analysis using whole-genome sequencing data. Mapping against the genome reference was performed with bwa mem (version 0.7.12-r1039) (54), duplicated reads were removed using Picard (version 1.140) (http://broadinstitute.github.io/picard), and the bedtools coverage v2.26 program (55) was used to perform further quality controls. Here, in order to identify genetic variations among strains, SNV detection (variant calling) and SNV matrix generation were performed using GATK version 3.8.0 (56) with best-practices parameters. The ENSEMBL variant effect predictor script (version 88) was used for variant annotation.

Phylogenetic analysis and determination of modifications of genes of interest.

The final step of the WGS-Outbreaker pipeline comprised maximum likelihood tree construction using RAxML software (version 8.2.9) (57) with the General Time Reversible CAT (GTRCAT) model and 100 bootstrap replicates. The phylogenetic tree was visualized and annotation was performed using the ggtree R package (58). In order to see if the population structure could be based on particular genomic modifications, some genes that have previously been described as being important in A. fumigatus biology were analyzed in depth in each of the A. fumigatus population clusters formed from the SNV comparisons. The genes analyzed were the cyp51A gene, including its promoter (AFUB_063960); the benA gene (AFUB_010330); the cytB gene (AfuMt00001); and the sdhB gene (AFUB_057960). The modifications found in these genes were used to determine the fungicide susceptibility phenotype based on the resistance mechanisms.

Data availability.

Data from the whole-genome sequencing project have been deposited in the NCBI SRA database (project accession number SRP151231).