A Novel Combination of CYP51A Mutations Confers Pan-Azole Resistance in Aspergillus fumigatus

Strain identification, clonality, and antifungal susceptibility testing.

Based on their calmodulin (CaM) gene sequences, the three isolates recovered from one cystic fibrosis patient were identified as A. fumigatus sensu stricto. To investigate whether these strains were different, their multilocus sequence type (MLST) patterns were established (17). Strains A and B (obtained from sputum samples in October 2015 and November 2016, respectively) shared the same MLST pattern (ANX4, allele 7; BGT1, allele 1; CAT1, allele 3; LIP, allele 2; MAT1-2, allele 1; SODB, allele 2; ZRF2, allele 1), while strain C (recovered from the patient in June 2017) showed a different one (ANX4, allele 1; BGT1, allele 1; CAT1, allele 3; LIP, allele 1; MAT1-2, allele 5; SODB, allele 1; ZRF2, allele 2). These strains exhibited reduced susceptibility to triazole drugs (Table 1). Considering the published epidemiological cutoff (ECOFF) values (18, 19), all strains were considered non-wild type for itraconazole and voriconazole (>16 and MICs ranging from 2.00 to 8.00 μg/ml, respectively), while strain B was non-wild type for all the tested azole drugs (isavuconazole and posaconazole; 4.00 μg/ml) (Table 1). When voriconazole susceptibility was established using agar diffusion, strains A and C showed a 10-mm inhibition diameter, while strain B showed no inhibition at all (Table 1; Fig. 1). On the other hand, the strains showed low amphotericin B, caspofungin, and anidulafungin MIC values (Table 1).

Sequence analysis of CYP51A genes and their promoters.

The three studied strains showed two point mutations at their CYP51A open reading frames, namely, t364a and t960a. These mutations were responsible for the amino acid substitutions L98H and S297T, respectively. In addition, the three strains also contained a 34-bp tandem duplication sequence in the CYP51A promoter (between positions −288 and −322). Strain B had an extra base change at CYP51A (a194g), leading to the substitution R65K (Table 1). Clustal alignment of different Cyp51Ap/Erg11p sequences showed that the R65 residue is conserved in a great diversity of fungal species, including yeast, molds, and dimorphic ascomycetes and several basidiomycetes. On the other hand, important pathogenic fungal species, such as Candida glabrata and Coccidioides posadasii, harbor a polymorphism (arginine by glutamine) at their Cyp51Ap/Erg11p site at the equivalent position (Fig. 1).

CYP51A gene replacement.

To establish the implication of the newly observed CYP51A mutation combination on triazole MIC increases, we initially transformed by electroporation the azole-susceptible A. fumigatus LMDM-31 strain (akuB_KU80Δ) (20, 21) with four linear PCR fragments named TR34-65K-98H, TR34-98H, TR34-65K, and 65K. These fragments carry the promoter alteration plus the substitutions at the 65 and 98 residues (1.1 kb), the 34-bp promoter repetition together with the substitution at the 98 residue (1.1 kb), the promoter alteration and the substitution at position 65 (2.2 kb), and the mutation that confers the R65K substitution alone (1.6 kb), respectively.

First, transformations were performed using strain LMDM-31 as the recipient strain and itraconazole 1 μg/ml was used as the selector. Homologous recombination and integration of two of the PCR fragments (TR34-98H and TR34-65K-98H) were obtained, and the integration was confirmed phenotypically by PCR and by sequencing in several mutants (n = 5 for TR34-98H and n = 4 for TR34-65K-98H) (Fig. 2). These transformants were obtained in two separate transformation experiments performed in different weeks. On the other hand, and after eight transformation experiments, no transformants were recovered when TR34-65K and 65K PCR fragments were intended to be introduced into LMDM-31 conidia by transformation. Thus, we decided to use an azole-hypersusceptible strain (A. fumigatus LMDM-32) as the recipient strain (21, 22) and using lower itraconazole concentrations (0.12 μg/ml) as the selector. Doing so, we tried to establish if subtle azole MIC shifts were produced by this newly described mutation. Using these new conditions, transformants were obtained with the other two PCR fragments (TR34-65K and 65K).

Studied transformants were named LMDM-1427 (LMDM-31::TR34-L98H), LMDM-1428 (LMDM-31::TR34-R65K-L98H), LMDM-1599 (LMDM-32::TR34-R65K), and LMDM-1600 (LMDM-32::R65K). These transformants were subjected to antifungal susceptibility testing by microdilution and by agar diffusion. LMDM-1427 and LMDM-1428 strains showed the same triazole susceptibility patterns as the clinical strains A (and C) and B, respectively. Both transformant strains showed a 266-fold increase in itraconazole MICs compared with the parental strain (LMDM-31). On the other hand, the transformant harboring the TR34/R65K/L98H allele (LMDM-1428) had a 4-fold increase in isavuconazole, posaconazole, and voriconazole MICs compared with the already azole-resistant (itraconazole/voriconazole) LMDM-1427 mutant strain (TR34/L98H) (Table 1). The reduction in voriconazole susceptibility was also evident for LMDM-1428 when the susceptibility was evaluated by using disk diffusion. The inhibition zone was reduced from 32 mm (parental LMDM-31 strain) to 10 mm (LMDM-1427) and was inexistent for LMDM-1428 (TR34-R65K-L98H) (Fig. 3). The described mutant phenotypes exactly mimicked the phenotypes of the clinical strains. On the other hand, transformants that do not harbor the L98H substitution (LMDM-1599 and LMDM-1600) showed slight or no azole MIC increases (Table 1).

Samples, strains, and media.

Three fungal isolates recovered from different sputum samples of an Argentinean cystic fibrosis patient obtained in October 2015, November 2016, and June 2017 (identified as A, B, and C, respectively) were sent to the Centro de Micología, Instituto de Microbiología y Parasitología Médica, Universidad de Buenos Aires-CONICET. The patient received itraconazole treatment (400 mg/day) during 2015 to 2016. In 2017, antifungal treatment was switched to voriconazole (200 mg/day). Isolates were identified to section level by macroscopic and microscopic morphology and to species level by sequencing of the CaM gene (30). A. fumigatus ATCC 204305 and Aspergillus flavus ATCC 204304 were used as quality-control strains for susceptibility testing. A. fumigatus LMDM-31 (akuB_KU80Δ) (20) and A. fumigatus LMDM-32 (LMDM-31:: CYP51AΔ) were used as recipient strains in electroporation assays (21).

PCR amplification, clonality, and sequencing of the CYP51A and CaM genes.

Conidia from each strain were inoculated into GYEP broth (2% glucose, 0.3% yeast extract, and 1% peptone), and DNA extraction was performed as previously described (5). CaM gene amplification was performed using the set of primers CMD5 and CMD6 (30). The full sequences of the CYP51A genes and its 5′ untranslated region (UTR) and 3′ UTR regions were amplified (31), and sequences were analyzed using the Unipro UGENE v1.29.0 software (32). CYP51A sequences were aligned with wild-type A. fumigatus CYP51A (GenBank accession number AF338659). The three clinical isolates underwent multilocus sequence typing (MLST) using the Bain et al. MLST scheme (17) and the Aspergillus PubMLST (https://pubmlst.org/cgi-bin/mlstdbnet/mlstdbnet.pl?page=allseq&file=af_profiles.xml).

Aspergillus fumigatus transformation vectors, transformation method, and integration confirmation.

A. fumigatus LMDM-31 was transformed by electroporation with two linear DNA fragments of 1.1 kb long each, following a method published earlier (16, 21, 22). One of the fragments was named TR34-L98H and contained the cyp51A promoter sequence of CYP51A with the 34-bp tandem repeat in combination with a t364a mutation (leading to a L98H substitution). The second fragment was designated TR34-R65K-L98H and included the same promoter and open reading frame (ORF) mutations described for the fragment TR34-L98H but with a second nucleotide change (g194a) that generates the substitution R65K. Both fragments were amplified by PCR using the primer set A7-A10 and the genomic DNAs from A. fumigatus A and B strains. Primer A7 (5′-TCATATGTTGCTCAGCGG -3′) hybridizes to the AfCYP51A 5′ UTR 462 nucleotides (nt) downstream of the CYP51A start codon, and primer A10 (5′-GGACATCTCTGCGGCAAT-3′) hybridizes 604 nt upstream of the start codon. Transformants were selected on minimal medium using itraconazole (1 μg/ml).

Strain LMDM-32 was transformed with two different linear PCR fragments (2.1 kb and 1.6 kb) obtained by fusion PCR and cloned later into two pGEM-T easy vectors (Promega, Biodynamics Argentina). Vectors were named TR34-65K and 65K. The last vector (cyp51A with the a194g mutation) was generated by amplifying the DNA obtained from LMDM-31 and introducing the a194g mutation by fusion PCR. The first two PCR amplifications were done using the following primer pairs: A7 together with R65K-fusion-R (5′-CTACAATCTTGAGACTTGCCTTTTCcCTGCACGCAAAGAAGAACTTGTAGG-3′) and R65K-fusion-F (5′-CCTACAAGTTCTTCTTTGCGTGCAGgGAAAAGGCAAGTCTCAAGATTGTAG-3′) together with P450-2-R (21) (5′-CTGTCTCACTTGGATGTG-3′) (lowercase letters indicate the inclusion of the nucleotide a194g change). Later, these two fragments (682 nt and 1,449 nt, respectively) were fused by PCR using the A7 and P450-2-R primers and cloned into a pGEM-T easy vector.

The transformation vector TR34-65K was generated using DNA obtained from the clinical strain A and the vector 65K. The first DNA was used to PCR amplify a 692-nt CYP51A fragment (496 nt downstream from the start codon of the CYP51A to 220 nt upstream of the ATG, including a TR34 duplication) using the primers A7 and R65K-fusion-R. The vector 65K was used as the DNA template to amplify a second portion of the CYP51A gene (220 nt to 1,619 nt) by using the R65K-fusion-F together with P450-2-R (5′-CTGTCTCACTTGGATGTG-3′) (21). These fragments were fused using the A7 and P450-2-R primers and cloned into a pGEM-T easy vector.

An itraconazole resistance phenotype was expected after the incorporation of a functional mutant cyp51a gene. The integration of the DNA fragments was phenotypically confirmed by plating the strains in duplicate in minimal medium and minimal medium containing 1 μg/ml and 0.12 μg/ml itraconazole to select transformants with LMDM-31 and LMDM-32 as the genetic background, respectively. Genomic DNA from itraconazole-resistant transformants and parental strains was obtained. Incorporation of the cassettes in the A. fumigatus genome was screened by a gradient PCR (temperature [T] annealing from 56°C to 66°C), using primers A7 and A5R (5′-CTTCTTTGCGTGCAGAGA-3′). Primer A5R hybridizes 197 nt upstream of the start codon of AfCYP51A. The mutation R65K (g194a nucleotide change) produced a negative PCR result since primer A5R would not hybridize at the upper limit of the T annealing gradient (Ta, >65°C). On the other hand, another PCR was designed to assess the presence of a cyp51a promoter alteration using the primers A7 and A10. If the TR fragments were incorporated in the genome, a 1,100-nt PCR band would be observed. In contrast, a 1,066-nt PCR fragment would be obtained for the parental strain or for nontransformant strains. PCR amplifications were performed in an Applied Biosystems thermocycler (Tecnolab-AB, Argentina) in a 25-μl final volume by using Pegasus DNA polymerase (PBL, Argentina). The cycling program used included an initial 2-min step at 95°C, followed by 30 cycles of 30 s at 95°C, 30 s at the melting temperature of the primer set used (56°C for A7/A10 and a gradient from 56°C to 66°C for A7/A5R to assess the presence of the g194a mutation), and 90 s at 72°C. A final cycle of 10 min at 72°C was included. The CYP51A genes, including the promoter area from the transformants, were sequenced to confirm their incorporation.

Antifungal susceptibility testing.

Antifungal susceptibility testing by microdilution was performed following the CLSI M38 3rd ed. document (33). Amphotericin B, anidulafungin, caspofungin, itraconazole, isavuconazole, voriconazole, and posaconazole were evaluated. Drugs were obtained as standard powders from their manufacturers or from Merck-Sigma-Aldrich (Argentina). All samples were analyzed at least three times on different days. Susceptibility testing results were interpreted by using the epidemiological cutoff (ECOFFs) values published in the CLSI M59 2nd ed. document for amphotericin B, caspofungin, itraconazole, isavuconazole, and voriconazole (18). For posaconazole, the ECOFFs were those published by Espinel-Ingroff et al. (19). Voriconazole susceptibilities were also assessed by agar diffusion following the guidelines published in the CLSI document M51A (34).