1. Introduction
Aspergillus fumigatus is a globally distributed saprophytic mold that plays a major role in recycling environmental carbon and nitrogen. Its primary ecological niche is decaying organic matter, but it is also commonly found in the air, water and soil. The fungus has an abundant asexual reproduction cycle and produces a prolific number of asexual spores, known as conidia [
1]. The conidia’s hydrophobic surface facilitates air dispersion and the spores can remain dormant and/or germinate in a wide range of environmental conditions [
2].
A. fumigatus has a ubiquitous presence in the air that can reach up to 10
9 conidia/m
3 in certain environments [
3]. Population genetic studies using simple sequence repeat (SSR) markers suggest that airborne dispersal by conidia has likely played a major role in the global population structure of
A. fumigatus [
4,
5]. However, evidence for geographic specific SSR alleles and drug resistance profiles have also been reported, including for the Hamilton, Canada population of this species [
6,
7,
8]. At present, whether the genetic uniqueness of geographic samples based on certain SSR markers reflects their whole-genome distinctiveness remains unknown. The main objectives of this study are to investigate the potential genetic uniqueness of the Hamilton population at the whole-genome sequence level and the potential relationship between genome sequence variation and susceptibility to the antifungal drug, amphotericin B (AMB).
A. fumigatus is among the most important opportunistic fungal pathogens in humans. Due to their high abundance in the air, conidia of
A. fumigatus are inhaled by humans daily and are small enough (2 to 3 μm) to reach the lung alveoli. This can lead to a spectrum of fungal infections generally termed as aspergillosis. The disease can range from simple allergic reactions to severe invasive infections [
1]. In immunocompetent individuals, inhaled conidia are cleared by the pulmonary immune system and rarely cause any harm. However, in immunocompromised individuals, incomplete killing of the fungi can lead to conidia germination followed by invasion of hyphae into tissue [
2].
A. fumigatus is considered the primary cause of invasive aspergillosis, a life-threatening mold infection with high morbidity and mortality rates in immunosuppressed patients. Depending on factors such as patient population type, site of infection and treatment regimen, mortality rates associated with invasive aspergillosis range from 60 to 90 percent [
3].
Among all antifungal agents, the triazole antifungals are usually first-choice drugs for treating aspergillosis because their use has been associated with better clinical response, lower infusion-related toxicity, lower nephrotoxicity, and increased survival [
9]. However, the emergence of multiple-triazole resistant
A. fumigatus strains throughout the world has been a growing public health concern and an increasing problem in treatment of patients in certain geographic areas. For cases of triazole-resistant
A. fumigatus strains, amphotericin B (AMB) has been recommended by experts as core therapy [
10].
AMB is a polyene antifungal agent that has been around since the 1950s for treatment against invasive fungal infections [
11]. Although AMB has been in use for almost 70 years, its mechanism of action has not been completely elucidated. Instead, multiple mechanisms of action have been suggested over the years, such as the ion-channel model, the production of oxidative stress, the surface absorption model, and the sterol sponge model. The major benefits for AMB use are that it possesses a broad spectrum of action, being effective against most human pathogenic fungi, and that mycological resistance to AMB has been very uncommon. The antifungal is also fungicidal, in contrast to most azoles that are fungistatic. With the majority of affected patients being immunocompromised, the fungicidal effect of AMB is a desirable property in treatment and would be preferable for immunosuppressed patients where killing target fungal cells in a short period of time is needed. Indeed, resistance to AMB is less common than resistance to azoles across human fungal pathogens [
12].
Recently, however, AMB resistance has been found in several
A. fumigatus populations [
13,
14,
15]. For example, the
A. fumigatus population in Hamilton, Ontario, Canada was found to have a very high rate (96%) of AMB resistance [
7]. Interestingly, also different from the majority of other geographic populations of
A. fumigatus, the rate of triazole resistance in Hamilton was very low. Indeed, none of the 196 strains from three ecological niches (farm fields, city parks, and patients) was resistant to voriconazole or itraconazole, the two main triazole drugs used to treat
A. fumigatus infections [
8]. This high rate of AMB resistance in Hamilton was surprising because AMB has rarely been used in this region. Therefore, the observed AMB resistance in this and other regions may represent geographic-specific intrinsic resistance by the local populations. In addition, as AMB is the commonly recommended last-line drug of treatment and with the rising incidence of triazole resistance, the observed AMB resistance represents a major challenge for the healthcare system. At present, the mechanism(s) for the high rate of AMB resistance in Hamilton as well as other geographic regions are unknown.
Whole-genome sequencing (WGS) is a powerful approach to develop an understanding of population structure and antifungal resistance mechanisms in human fungal pathogens [
16,
17,
18]. For example, WGS has been performed on
A. fumigatus strains to investigate mutations causing azole resistance [
19,
20,
21], however, a study focusing on AMB resistance in
A. fumigatus has yet to be conducted. Indeed, there is little information about the mechanisms of AMB resistance in most human fungal pathogens. The aim of this study was to use a WGS approach to investigate genes associated with AMB resistance in
A. fumigatus, with a focus on the Hamiltonian
A. fumigatus isolates where the highest rate of AMB resistance has been reported. In addition, publicly available genome sequences of strains from a wide range of regions were included in our comparison to determine the relationships among strains from different geographic regions and to reveal the potential of the genetic uniqueness of the Hamiltonian
A. fumigatus population at the genome level.
4. Discussion
In this study, we investigated the overall phylogenomic diversity of
A. fumigatus using whole-genome sequences of 196 strains collected from 11 different countries: Canada, India, Ireland, Japan, Netherlands, Peru, Portugal, Singapore, Spain, United Kingdom and the United States. Using 404,021 SNP sites, our phylogenetic analysis identified a group of strains that had a high level of divergence from all other
A. fumigatus strains. This clade consisted of strains from Spain (
n = 12), Canada (
n = 1), India (
n = 1) and Peru (
n = 1). Our results are consistent with what had been reported by Garcia-Rubio and colleagues [
38]. Their study examined 101
A. fumigatus genome sequences from seven countries (Canada, India, Japan, Netherlands, Portugal, Spain, and the United Kingdom) and included the same strains from Spain (
n = 12) and Canada (
n = 1). They conducted a phylogenetic analysis using all 101 genomes. The result was that these 13 isolates clustered together and were identified as belonging to the most distant cluster. The cluster was also found to have the greatest homogeneity, with differences between strains being the lowest of all other groupings [
38]. Our analyses further extended their analyses for this clade and expanded its geographic distribution into South America (Peru) and Asia (India). The significant divergence of strains in this cluster from other strains suggest that Cluster 1 strains likely represent a cryptic species within the A. fumigatus sensu stricto complex. In addition, the relatively limited sequence different among geographically distant strains in this cluster suggests recent dispersals of Cluster 1 strains among geographic regions.
Our analyses focused on the 71 strains with known AMB susceptibilities to investigate potential genomic regions associated with AMB resistance. The 71 strains were obtained from 8 countries: Canada, India, Ireland, Japan, Netherlands, Spain, United Kingdom, and the United States. Clustering analysis was conducted using 117,770 SNP sites from CDS regions, and it showed the existence of three genetic clusters, consistent with the results based on all SNPs from the 196 genomes. However, our results are slightly different from that based on 101 genomes by Garcia-Rubio and colleagues [
38]. They determined that their
A. fumigatus samples were divided into four clusters, while our analyses showed three clusters. This difference in cluster number was because our Cluster 2 was a combination of their Clusters I and III. This combination was likely due to the addition of 95 genomes representing more geographic regions which could have resulted in intermediates that linked these two clusters (Clusters I and III) together to form one larger cluster (our Cluster 2). In addition, our clustering result showed a noticeably lower number of genetic clusters from what was reported earlier based on SSR markers e.g., [
4]. For example, in the study by Ashu et al. [
4], they analyzed the population structure of 2025
A. fumigatus isolates from 13 countries (Australia, Belgium, China, Cuba, France, Germany, India, Italy, Netherlands, Norway, Spain, Switzerland and the United States) using nine SSR markers. Although their study analyzed more strains (2025) than this study (196 strains), the number of markers used in the current study (404,021 SNPs for the whole sample of 196 strains and 117,770 SNPs for the 71 strains) is much greater than the nine microsatellite loci. Nonetheless, the Ashu et al. [
4] study had samples from 13 countries (vs. 11 countries in this study) and included strains from 9 countries that we did not have in our study, which may account for some of the discrepancy.
Our study also examined non-synonymous SNPs in 22 genes of interest and the
p-values for each SNP were calculated using Fisher’s exact test. We found 12 missense mutations that were significantly associated with AMB resistance. The SNPs were located in six genes,
CatA, ERG3, Fos1, MpkB, MpkC, and
TcsB. These 22 genes were chosen based on the known mechanisms of AMB resistance in other fungi [
39]. Previous studies have found a correlation between alterations in the ergosterol synthesis pathway and AMB resistance in several fungal species. Here, we found two SNPs in
ERG3 that were significantly associated with AMB resistance.
ERG3 is involved in the ergosterol biosynthesis pathway and encodes for C-5 sterol desaturase. The role of
ERG3 in AMB resistance has also been validated in AMB resistant
Candida lusitaniae strains where
ERG3 expression was found to be reduced [
40]. Additionally, other studies have found an association between oxidative stress tolerance and AMB resistance. Therefore, we also examined genes involved in the two ROS-detoxifying systems, specifically catalases and superoxide dismutases. Moreover, in a previous study on
Candida dubliniensis and
Candida albicans, superoxide dismutase and catalase activities were significantly higher in AMB resistant strains [
41]. There are three functional catalases in
A. fumigatus:
Cat1, Cat2 and
CatA. Our results showed that two SNPs in
CatA were significantly associated with resistance. The final category we looked at were genes involved in the HOG MAPK signaling pathway, as the pathway plays a role in oxidative stress response in many fungal pathogens. This pathway has also been observed to regulate genes involved in ergosterol biosynthesis. For example, inhibition of the HOG pathway increased the expression levels of
ERG11 in
Cryptococcus neoformans [
42]. Our analyses found four SNPs in
TcsB to be associated with AMB sensitivity. SNPs associated with AMB resistance were also found in
Fos1 (
n = 1),
MpkB (
n = 2) and
MpkC (
n = 2). To our knowledge, these four genes have not been implicated in AMB resistance, and they represent promising candidates for further analyses.
To investigate AMB resistance in
A. fumigatus on a broader scale, whole-genome sequences of the 71 strains were used in GWAS. Of the top 15 SNPs significantly associated with AMB MIC, two significant SNPs in genic regions were identified: a synonymous variant in
Afu6g04940 and a missense variant in
Afu3g03350.
Afu6g04940 is annotated as the cytokinesis protein SepA/Bni1. Although not much is known about its role in
A. fumigatus, its ortholog has been studied in the closely related model species
Aspergillus nidulans. The
A. nidulans ortholog,
SepA, is involved in polarized growth, specifically apical growth and septation [
43].
SepA deletion mutants are unable to undergo septation due to their inability to assemble actin rings at division sites [
44]. The mutants still form hyphae, but they are wider and have an abnormal dichotomous branching pattern due to defects in hyphal polarity [
44]. The second discovered gene,
Afu3g03350 or
SidE, encodes a bimodular peptide synthetase [
45].
SidE was previously thought to be involved in siderophore production but is now found to be involved in fumarylalanine (FA) production [
45]. Although the biological function of fumarylalanine remains unknown, due to the structural similarity of FA to established pharmaceuticals with immunomodulatory activity, the gene may be involved in host immunosuppression and
A. fumigatus virulence [
45]. Expression of
SidE was also found to be significantly transcriptionally upregulated at increased temperatures, under both presence and absence of iron, as well as by oxidative stress in the presence of iron (in iron-replete or high-iron conditions) [
45].
At present, the origin of the high rate of AMB resistance in Hamilton, Canada and several other countries remains unknown. However, the results of our study have brought into light several possibilities. In the first, the genes identified here as related to AMB resistance, including
Afu6g04940 and
Afu3g03350, suggest the possibility that AMB resistance could be related to iron and/or temperature conditions in the environment, where mutations that occurred in response to these stresses led to increased AMB tolerance as a by-product. External cell wall stress has also been found to induce alternative cytokinesis and septation strategies in fungal species. Additionally, iron acquisition and the oxidative stress response are known to be related, with iron shown to play an essential role in the oxidative stress response [
46]. The second possibility is related to mutational rates. Several studies have suggested that environmental stress could increase the mutational rate [
47]. It is thought that the maintenance and repair pathways function at a lower capacity during stress, which in turn leads to more mutations [
47]. This would ultimately result in a higher chance for the appearance of mutations that confer resistance [
47]. The high allelic and genotypic diversity of the Hamilton strains at both the SNP and microsatellite loci are consistent with the high mutation rates in this population. Finally, the Hamilton strains were found mainly in Cluster 2 (90%), which could suggest a potential evolutionary predisposition of this groups of strains being more likely to develop AMB resistance. These three possibilities are not mutually exclusive of each other and all could contribute to the observed high prevalence of AMB resistance. Indeed, the close relationships among clinical and environmental strains at the genome level as shown in both
Figure 2 and
Figure 5 suggest that selection pressure on environmental strains of
A. fumigatus can have significant impacts on clinical practices on antifungal drug treatments.
The incidence of invasive aspergillosis has been increasing, due in part to the expanding immunocompromised population. Invasive aspergillosis is becoming an important infectious disease, with
A. fumigatus accounting for the majority of cases. Over the past decade, incidence of invasive aspergillosis in low-risk patients and immunocompetent individuals have also been increasingly reported [
48,
49,
50]. Therefore, antifungal susceptibility patterns and information on resistance mechanisms is becoming more important. Understanding the underlying mechanisms leading to the origin and persistence of AMB-resistant strains could help us developing measures to reduce their emergence and spread. Furthermore, while our current knowledge on AMB resistance in
A. fumigatus is still fragmentary, GWAS analysis in this study identified a large number of SNPs significantly associated with AMB resistance. We highlighted the significant SNPs associated with AMB susceptibility in eight genes,
Afu6g04940,
Afu3g03350, ERG3, Fos1, CatA, TcsB, MpkB, and
MpkC. The role of these eight genes as well as other genes in AMB resistance still need to be confirmed through additional experiments and functional assessments. Nonetheless, the results obtained from this study provide a foundation for further research on AMB resistance. From a clinical perspective, the results could aid in the development of diagnostic molecular markers for AMB resistance screening. This information would help predict the outcome of AMB therapy and determine whether it remains the best option as the last line drug in treatment. With prompt treatment being crucial in cases of invasive aspergillosis, the potential diagnostic markers to be developed in this study could aid in and speed up clinical decision making.