ABSTRACT
Invasive aspergillosis (IA) is a severe condition mainly caused by Aspergillus fumigatus, although other species of the genus, such as section Nigri members, can also be involved. Voriconazole (VRC) is the recommended treatment for IA; however, the prevalence of azole-resistant Aspergillus isolates has alarmingly increased in recent years, and the underlying resistance mechanisms in non-fumigatus species remain unclear. We have determined the in vitro susceptibility of 36 strains from section Nigri to VRC, posaconazole (POS), and itraconazole (ITC), and we have explored the role of Cyp51A and Cyp51B, both targets of azoles, in azole resistance. The three drugs were highly active; POS displayed the best in vitro activity, while ITC and VRC showed MICs above the established epidemiological cutoff values in 9 and 16% of the strains, respectively. Furthermore, expression studies of cyp51A and cyp51B in control condition and after VRC exposure were performed in 14 strains with different VRC susceptibility. We found higher transcription of cyp51A, which was upregulated upon VRC exposure, but no correlation between MICs and cyp51 transcription levels was observed. In addition, cyp51A sequence analyses revealed nonsynonymous mutations present in both, wild-type and non-wild-type strains of A. niger and A. tubingensis. Nevertheless, a few mutations were exclusively present in non-wild-type A. tubingensis strains. Altogether, our results suggest that azole resistance in section Nigri is not clearly explained by Cyp51A protein alteration or by cyp51 gene upregulation, which indicates that other mechanisms might be involved.
INTRODUCTION
Aspergillus is a widely distributed genus able to cause human opportunistic infections (1), with invasive aspergillosis (IA) being the most problematic condition in terms of mortality, which ranges from 50 to 100% (2). Immunocompromised patients, such those suffering from neutropenia and cancer and patients undergoing organ or hematopoietic stem cell transplantation, are the most susceptible (3, 4). Although A. fumigatus is the most frequent causal agent of IA (5), species from other sections of the genus, such as Flavi, Terrei, and Nigri, can also be involved (6). In particular, members of the section Nigri, including A. niger and the cryptic species A. tubingensis, A. brasiliensis, A. awamori, A. japonicus, and A. carbonarius, among others (1), have been reported as the second leading cause of IA, along with sections Flavi and Terrei (6–11).
In the current guidelines for diagnosis and management of IA, voriconazole (VRC) constitutes the first-line therapy (12). However, in recent years there has been an alarming increase in Aspergillus isolates showing resistance to azoles, which is correlated with poor therapeutic outcome, entirely reducing treatment options (5, 13–15). In the particular case of Aspergillus, which generally displays elevated azole susceptibility (16, 17), cases of azole resistance have been constantly increasing since the first reports in the 1990s and have been associated with both azole-treated and azole-naive patients (13). This acquired resistance can be explained by the fungistatic effect of azoles but also by azole exposure in the clinical and environmental settings (5, 13, 18).
The azole mode of action consists of noncompetitive binding to the Cyp51 enzyme, a 14α-demethylase from the ergosterol biosynthetic pathway, which leads to ergosterol synthesis inhibition and, consequently, cell membrane disruption (19). In this sense, several azole resistance mechanisms in A. fumigatus have been postulated. The presence of amino acid substitutions in the Cyp51 target proteins and overexpression of the cyp51 genes or efflux pumps, such as the ATP binding cassette transporter Cdr1B, constitute the most important mechanisms described thus far (7, 20).
In this context, the study of cyp51 genes has emerged as a main topic in azole susceptibility profiling in fungal pathogens. Some filamentous fungi, including Aspergillus, carry multiple cyp51 genes in their genomes. This genetic redundancy is widely believed to facilitate adaptation to fungicides exposure, leading to an increase in azole resistance (21). Specifically, A. fumigatus, A. terreus, and A. niger contain two cyp51 paralogs (cyp51A and cyp51B), while A. flavus contains three cyp51 genes (cyp51A, cyp51B, and cyp51C) (7, 22). Moreover, previous studies conducted in A. fumigatus have indicated functional diversification and different contribution to azole resistance of cyp51 paralogs (22).
Since azole resistance mechanisms in non-fumigatus species, such as the members of section Nigri, remain practically unknown, characterization of cyp51 genes might lead to increase our knowledge on azole resistance emergence.
In this study, we performed gene expression studies of cyp51A and cyp51B in a set of clinical and environmental strains belonging to Aspergillus section Nigri, as well as sequence analyses of the cyp51A gene, in order to potentially correlate gene expression and/or nonsynonymous mutations with VRC resistance in members of this section.
RESULTS
Antifungal susceptibility.The MICs of itraconazole (ITC), posaconazole (POS), and VRC were determined in 36 strains belonging to section Nigri. The results are shown in Fig. 1 and also Table S1 in the supplemental material. All the assayed drugs were active against the tested strains, with the lowest modal MIC being that of POS (0.06 μg/ml), followed by ITC (1 μg/ml) and VRC (2 μg/ml). In all strains the POS MICs were below the epidemiological cutoff value (ECV) established for this drug against section Nigri, whereas the ITC and VRC MIC values were lower than the ECVs established in 9 and 16% of the strains, respectively. The ITC MIC was equal to the ECV in only one strain of A. tubingensis, while in the case of VRC, this was observed in eleven strains (species A. niger, A. tubingensis, A. neoniger, and A. acidus).
Box plot representation of Aspergillus section Nigri susceptibility pattern to the triazoles ITC (a), POS (b), and VRC (c). The box edges represent the first and third quartiles, respectively. Thick lines denote median values, while minimum and maximum values are depicted by horizontal lines.
Cross-resistance between ITC and VRC was observed in four strains (9% of the total) belonging to A. niger, A. tubingensis, and A. brasiliensis species.
In silico identification and sequence analysis of cyp51 genes in section Nigri species.We conducted an in silico analysis in the A. niger, A. tubingensis, and A. brasiliensis genomes to identify cyp51 genes and their deduced amino acid sequences. The results of BLASTP using the amino acid sequence of A. fumigatus Cyp51A (Afu4g06890) as a query revealed high homology with A. niger An11g02230, A. tubingensis Asptu1_0056695, and A. brasiliensis Aspbr1_0190913, showing a 79.7 to 80.2% identity range. Likewise, BLASTP search using A. fumigatus Cyp51B (Afu7g03740) as a query revealed high homology with A. niger An13g0009, A. tubingensis Asptu1_0051806, and A. brasiliensis Aspbr1_0137500, showing a 84.2 to 85.1% identity range.
In addition, sequence alignment of Cyp51A and Cyp51B revealed moderate pairwise similarity among these paralogous proteins in every species, displaying identity values of 62.5% for A. fumigatus, and from 63 to 63.8% for the non-fumigatus species studied. The detailed view reveals conserved residues involving the putative azole binding site among both proteins in all species.
Unlike A. fumigatus, which contains a single transmembrane domain, A. niger, A. tubingensis, and A. brasiliensis exhibit two predicted transmembrane domains in both Cyp51A and Cyp51B, comprising amino acids 5 to 24 and 37 to 59 and amino acids 20 to 42 and 54 to 73, respectively. Protein models are represented in Fig. 2. The sequence alignment, azole-binding residues, transmembrane domains, and the heme prosthetic group are represented in Fig. S1 (Cyp51A) and Fig. S2 (Cyp51B). The Cyp51A and Cyp51B sequence alignment of every species is represented in Fig. S3.
Protein topology of A. fumigatus (a) and A. niger, A. tubingensis, and A. brasiliensis (b) Cyp51A and Cyp51B associated with the endoplasmic reticulum membrane. The amino acid positions comprised by each protein structure are also indicated.
Expression analysis of cyp51A and cyp51B in azole-inducing conditions.To analyze the expression patterns of cyp51A and cyp51B in control condition and during exposure to VRC, transcription levels of both genes were determined in 14 strains of section Nigri by quantitative reverse transcription-PCR. The results show higher expression of the cyp51A gene compared to that of the cyp51B in both conditions and all strains. We observed remarkable fluctuations of cyp51A transcription among strains, ranging from 2 × 10−1 to 5 × 102 times fold with respect to the housekeeping gene (Fig. 3). Exposure to VRC resulted in statistically significant cyp51A induction (P ≤ 0.008) in six strains (43%) categorized as wild type (wt) and non-wild type (non-wt). In addition, another 43% of the tested strains showed a non-statistically significant increased transcription of this gene upon drug exposure. Exceptionally, cyp51A was similarly expressed in the control condition and under VRC exposure in the A. brasiliensis strain (Fig. 3a). Unexpectedly, the strain displaying the highest VRC MIC showed low levels of cyp51A transcription, which were significantly reduced in VRC presence. Therefore, we could not establish any pattern of expression of cyp51A along the increasing MIC scale.
Expression analysis results of cyp51A (a) and cyp51B (b) genes in control condition and after exposure to 4 μg/ml VRC for 8 h. actA was used as the housekeeping gene to normalize transcription levels. MIC values expressed in μg/ml are shown below isolates labels. Statistical significance (P ≤ 0.05) is marked by an asterisk (*).
Regarding cyp51B, expression levels ranged from 3 × 10−1 to 7 times fold respect to the housekeeping gene, and VRC exposure did not result in significant overexpression (Fig. 3b). These data suggest that cyp51B gene expression is not inducible by VRC, in contrast to the cyp51A gene.
cyp51A sequencing and amino acid substitutions identification.In order to identify nonsynonymous mutations in cyp51A that could be correlated with VRC resistance within the section Nigri, Cyp51A protein sequence was analyzed in the 14 selected strains. All A. tubingensis strains but one and all A. niger strains displayed amino acid substitutions in the Cyp51A protein compared to the reference sequences.
In the particular case of A. tubingensis, the amino acid change H467Q was exclusively detected in environmental non-wt strains in combination with substitutions K64E or V377I. In addition, other mutations were found in both wt and non-wt strains of this species, such A140V and P413S, whereas the amino acid change A321T was only found in an environmental strain that displayed elevated susceptibility to VRC (Table 1). Regarding A. niger, none of the substitutions found in the Cyp51A protein could be linked to resistant phenotypes. Q228R was a common change in all strains, which was frequently found in combination with T57A, located in the second transmembrane domain of the A. niger Cyp51A. In addition to these, an environmental wt strain showed the S346R change. Multiple substitutions were observed in the case of the clinical wt strain FMR 11894 beyond Q228R and T57A: I172V, A185S, T242M, I378V, M420L, and V422I. The A. brasiliensis non-wt strain did not display any amino acid substitution in the Cyp51A protein.
Nonsynonymous nucleotide mutations in gDNA and their corresponding amino acid substitutions in the deduced Cyp51A protein sequencea
The detected nucleotide and amino acid modifications compared to the deposited sequences in the Aspergillus Genome Database are summarized in Table 1. In addition, Cyp51A sequence alignments with domain identifications and the amino acid changes found in our strains are marked in Fig. S1.
DISCUSSION
Although resistance mechanisms in A. fumigatus have been extensively studied, there are very few studies carried out on members of Aspergillus section Nigri. Thus, we sought to test the in vitro susceptibility of clinical and environmental isolates of relevant species of Aspergillus section Nigri, as well as to characterize possible resistance mechanisms present in our set of strains.
The three azoles tested here (VRC, PSC, and ITC) showed good activity, with POS being the most active one, a finding in accordance with previous studies (23–25). The low modal POS MIC and its narrow MIC range are also indicative of its elevated in vitro activity against these fungi. Both ITC and VRC exhibited good in vitro activity, even though low susceptibility to VRC was more common than to ITC, which is also in line with previous results (26). In addition, ITC non-wt isolates were found more frequently among A. tubingensis isolates than in the other species of this section that we tested, as already described (23, 26, 27). In accordance with previous works (23, 27), cross-resistance was not common in our isolates.
Since Cyp51 proteins constitute the targets for azoles, a high level of their transcripts could confer azole resistance by maintaining the cellular ergosterol levels. In fact, it has been documented that the introduction of extra copies of the A. nidulans cyp51A gene (pdmA) in A. fumigatus caused a decrease in ITC susceptibility in the latter (28), suggesting that cyp51A overexpression might confer azole resistance. However, this mechanism does not seem to be the key factor conditioning azole resistance, since no overexpression of cyp51A was detected in a set of non-wt A. fumigatus strains in a subsequent study (29). Great variability in expression patterns was also described in studies focusing on other species of the genus, thus providing controversial results. Specifically, in A. flavus, cyp51 transcript levels were not correlated to VRC-resistant phenotypes in a study (30), whereas 80% of non-wt strains displayed cyp51A overexpression in another one (31), although in the latter work resistance was linked to multidrug efflux pump overexpression as well.
Our results suggest, for the first time in section Nigri to our knowledge, that cyp51A actively participates in the transcriptional response to azole stress in these fungi, which is clearly upregulated after exposure to VRC. Nevertheless, transcript levels of cyp51A showed no correlation to the VRC MICs in this section, meaning that azole resistance is not explained by cyp51A overexpression. This has also been discussed in a recent study in which the basal expression of cyp51A in a few species of section Nigri did not correlate with the susceptibility patterns they displayed (26).
In contrast, cyp51B showed notably lower and stable expression in all the conditions we tested, suggesting a minor role in azole response. Accordingly, cyp51A has also been described to be the key player in A. fumigatus azole response, since its disruption results in reduced resistance to azoles, while azole susceptibility patterns are not altered in cyp51B knockout mutants (22, 32). However, this perception is not shared by everyone, since it has been suggested that cyp51B also contributes to azole resistance in A. fumigatus due to its induction upon ITC exposure, although only in one strain after ITC exposure (29). This was also observed in our study since a slight but not significant upregulation of cyp51B was detected upon VRC exposure. Considering that these are isolated facts, the involvement of cyp51B in azole resistance seems negligible.
Interestingly, an opposite pattern was detected in a transcriptome analysis performed on a strain of A. fumigatus with elevated azole susceptibility in which both cyp51 genes decreased their expression after VRC exposure (33). Surprisingly, we have observed this inhibition in the strain showing the highest VRC MIC (FMR 11900), which displayed a drastic reduction of cyp51A transcription under VRC. These controversial data reinforce the hypothesis that other mechanisms and not cyp51 expression could be contributing to the appearance of resistant phenotypes in Aspergillus.
In another line, differences in the Cyp51A protein sequence and structure are also important parameters to be considered, as changes in the conformation of the protein can vary its stability and flexibility, reducing azole binding affinity (34, 35). It should be noted that the detailed study of the Cyp51 proteins revealed some structural and sequence differences among the studied species of section Nigri and A. fumigatus, such as the presence of two transmembrane domains in section Nigri in contrast to only one of A. fumigatus. Although the transmembrane domain present in A. fumigatus has been described already, there are slight differences between the previously reported amino acids that comprise the transmembrane region in this species and our predicted data (36).
In A. fumigatus, many cases of azole resistance have been linked to cyp51A single point mutations that lead to amino acid alterations in the Cyp51A protein, as well as tandem repeats in the promoter region of the cyp51A gene that enhance its expression (37–41). Interestingly, no amino acid substitutions in the Cyp51B protein have been correlated to azole resistance so far, which is also indicative of its smaller role in azole response.
However, there has been some controversy, since some of the Cyp51A mutations previously described have also been found in highly azole-susceptible A. fumigatus and other Aspergillus spp. (23, 27, 42, 43). In fact, some of these amino acid differences believed to confer azole resistance have been related to different regional origins rather than resistance events (42).
Regarding our results in the species of section Nigri, sequence analysis of Cyp51A revealed no general correlation between amino acid changes and azole resistance, since most of the mutations were present in both wt and non-wt strains. This is the case for A140V and P413S substitutions in A. tubingensis or T57A and Q228R in A. niger, which were previously reported by others as well (23, 26). Moreover, amino acid A321 in A. tubingensis Cyp51A has been described as a residue that could confer azole resistance (26). Our results though, seem to contradict this, since five of our strains with different VRC MICs presented Ala in this position and a non-wt strain presented Thr in the same position. This straightforwardly suggests that amino acid changes in this position are not related to azole resistance. In addition, we found that two non-wt strains belonging to A. tubingensis (FMR 14635) and A. brasiliensis (FMR 15386) did not present any amino acid substitution, meaning that elevated VRC MICs found for these strains could be due to other mechanisms. Nevertheless, other non-wt strains (FMR 14712 and FMR 17207) presented in an exclusive manner a few Cyp51A substitutions. To our knowledge, the mutations H467Q, V377I, or K64E found in these isolates have not been reported to date; however, it remains unclear whether they are the cause of azole resistance in these strains, and further studies are needed to clarify their exact role. In particular, the substitutions K64E, V377I, and H467Q do not correspond to any characterized azole binding site residues, and therefore azole union should not be altered. Nonetheless, it remains unknown whether these mutations have any effect on the conformation of the protein or its binding pocket, preventing optimal binding.
Although cyp51 expression levels and protein affinity to VRC may be crucial to dealing with azole toxicity inside the cell, our data show that neither cyp51A transcription nor amino acid changes are straightforwardly linked to resistant phenotypes in Aspergillus section Nigri. This contrasts with what occurs in A. fumigatus (26, 27). Together, these findings suggest that resistance is a complex phenomenon in which other molecular mechanisms are involved, especially in non-fumigatus species. In this sense, more studies with larger sets of strains are needed, and potential new resistance mechanisms should be further explored.
MATERIALS AND METHODS
Strains, media, and growth conditions.A total of 36 environmental and clinical strains belonging to Aspergillus section Nigri (A. niger, A. tubingensis, A. brasiliensis, A. awamori, A. japonicus, A. carbonarius, A. neoniger, and A. acidus), previously identified by ITS, benA, and cam gene marker sequencing, were included in the in vitro susceptibility study (see Table S1 in the supplemental material).
Fourteen of these strains, belonging to A. niger, A. tubingensis, and A. brasiliensis, were selected for further studies on the basis of their different in vitro VRC susceptibility to characterize potential resistance mechanisms (Table S1). Five of the strains were clinical and had been previously isolated from sputum or skin lesions in Texas, Brazil, and Spain. The rest had been isolated from environmental sites in different countries (Spain, Venezuela, Mexico, Vietnam, and Brazil).
Strains were stored as mineral oil cultures or lyophilized cultures and before used they were grown on potato dextrose agar (PDA) media (Conda-Pronadisa) for 3 to 5 days at 35°C twice. For conidiation, strains were grown for 3 to 5 days on PDA at 35°C, and conidia were collected by flooding the PDA culture plates with 5 ml of 0.1% Tween 20 solution and gentle agitation. For genomic DNA (gDNA) extraction, cultures were incubated in potato dextrose broth (PDB) as previously described with some modifications (44). Briefly, conidial suspensions were incubated in PDB for 24 h at 37°C and 180 rpm, and mycelia were collected by monodur filtration, washed with distilled sterile water, and stored at –80°C until use. For gene expression analyses, 1 × 107 conidia/ml were germinated in YG medium (0.5% yeast extract, 2% glucose) for 12 h at 37°C and 180 rpm. Mycelia were harvested by filtration and aseptically transferred to fresh YG medium supplemented with VRC (Pfizer, Inc.) at a final concentration of 4 μg/ml when required. Cultures were incubated at 37°C and 180 rpm for 8 h and mycelia were collected by monodur filtration, washed with distilled water, and stored at –80°C until use.
Antifungal susceptibility testing.In vitro susceptibility testing was carried out according to Clinical and Laboratory Standards Institute (CLSI) protocol M38 for broth microdilution (45). Briefly, stock solutions of ITC (Janssen Pharmaceutica), POS (Schering-Plough Research Institute), and VRC (Pfizer, Inc.) were prepared in sterile water or dimethyl sulfoxide. All drugs were diluted in RPMI 1640 medium and dispensed in 96-well microdilution trays, which were inoculated with a conidial suspension of every strain previously adjusted by using a hemocytometer count. MICs of triazoles were determined after 48 h of incubation at 35°C without light nor agitation by direct visualization with an inverted mirror. MICs corresponded to the lowest drug concentration that completely inhibited fungal growth. A. flavus strain ATCC 204304 was used as the quality control strain.
Strains were classified as wt and non-wt according to the currently established ECVs for ITC (4 μg/ml), POS (2 μg/ml), and VRC (2 μg/ml) and section Nigri, since no clinical breakpoints are available yet for Aspergillus (46). The strains displaying ITC, POS, or VRC MICs above the ECVs were categorized as non-wt, while those displaying MICs below the ECVs were categorized as wt.
DNA and RNA isolation.For DNA and RNA extraction, mycelia were ground in liquid nitrogen with mortar and pestle. Total gDNA was extracted from ground mycelia according to a previously reported protocol (47). Total RNA was extracted from ground mycelia with TRIzol reagent (Thermo Fisher) as previously described (48). RNA was purified with a NucleoSpin RNA kit (Macherey-Nagel, Germany), including on-column DNase digestion. The resulting DNA and RNA pellets were resuspended in DNase- and RNase-free purified water, and the quantity and quality were checked by running aliquots in RedSafe-stained agarose gels and spectrophotometric analysis in NanoDrop 2000 spectrophotometer (Thermo Fisher).
Gene and protein identification and analysis in section Nigri.cyp51A, cyp51B, and actin (actA) orthologs were identified in A. niger, A. tubingensis, and A. brasiliensis by BLASTP searches (Basic Local Alignment Search Tool) in the Aspergillus Genome Database, using A. fumigatus Af293 protein sequences as templates. Alignments of the deduced amino acid sequence were performed with ClustalW algorithm of MegaLign software (DNASTAR). Sequence analysis and identification of the putative azole-binding residues and heme prosthetic group in A. niger, A. tubingensis, and A. brasiliensis was performed by comparison with A. fumigatus protein structure deposited in the SWISS-MODEL database (IDs: 4uyl.2, 4uym.1, and 5frb.1) and data published by others (49–51). Prediction of transmembrane helices and topology of proteins was performed with the TMHMM v2.0 server and the InterPro platform.
Retrotranscription and quantitative real-time PCR.For gene expression studies, total RNA was reverse transcribed into first-strand complementary DNA (cDNA) using the iScript cDNA synthesis kit (Bio-Rad) and following the manufacturer’s indications. Reverse transcription-PCRs were performed in a StepOne Plus real-time PCR system (Applied Biosystems) using 7.5 μl of FastStart Universal SYBR green Master mix (Roche Diagnostics), 6.9 μl of cDNA template, and 300 nM of each gene-specific primer (Table 2) in a final reaction volume of 15 μl. All primer pairs amplified products of 162 to 207 bp. The thermal cycling conditions were as follows: 94°C (5 min); 45 cycles of 94°C (30 s), 60°C (30 s), and 72°C (30 s); and finally 80°C (20 s). A melting curve was obtained immediately after PCR completion to check amplification specificity under the following conditions: 95°C (15 s), 60°C (15 s), and 95°C (15 s). Expression levels were calculated by normalizing cyp51A and cyp51B cycle thresholds (CT) with those of the housekeeping gene actA (52) using the efficiency (E) corrected ΔCT method (53). With this method, the relative expression ratio is calculated from the quantitative real-time PCR (RT-qPCR) E value assessed for the three genes in the study (Fig. S4), which were within an acceptable E value range (54), and the CT in each case.
Primers used in this study designed with Oligo7 softwarea
cyp51A sequencing and protein analysis.Amplification of cyp51A was carried out by standard PCR using the following thermal cycling protocol: 98°C (1 min); followed by 35 cycles of 98°C (10 s), 62°C (30 s), and 72°C (1.5 min); and finally 72°C (10 min).
Amplicons were sequenced with the primers listed in Table 2 at Macrogen Europe (Macrogen, Inc., Madrid, Spain), and sequences were analyzed using SeqMan (DNASTAR) and MEGA version 7 software. Alignments of the deduced amino acid sequences were performed using the ClustalW algorithm of MegaAlign software (DNASTAR).
Statistical analysis.A Mann-Whitney test was used to compare relative gene expression values. All statistical analyses were performed with GraphPad Prism 6.0 for Windows. P values of ≤0.05 were considered statistically significant.
ACKNOWLEDGMENTS
This research received no specific grant from any funding agency in the public, commercial, or nonprofit sectors.
A.P.-C. is the recipient of a FI fellowship from Generalitat de Catalunya (Spain).
FOOTNOTES
- Received 12 March 2019.
- Returned for modification 20 April 2019.
- Accepted 2 May 2019.
- Accepted manuscript posted online 6 May 2019.
Supplemental material for this article may be found at https://doi.org/10.1128/AAC.00543-19.
- Copyright © 2019 American Society for Microbiology.