Evolving moldy murderers: Aspergillus section Fumigati as a model for studying the repeated evolution of fungal pathogenicity

Introduction Species in the genus Aspergillus are saprophytic filamentous fungi that are most commonly found in soil and litter environments of subtropical and warm temperate latitudes [1]. Inhalation of asexual spores produced by Aspergillus fumigatus and a few other species in the genus cause a group of diseases collectively referred to as aspergillosis [2]. The most severe form of aspergillosis is invasive aspergillosis, which primarily affects individuals with compromised immune systems or preexisting lung conditions [3]. Since drugs targeting invasive aspergillosis are not always effective due to our lack of understanding of how they function inside the human host [4] and the evolution of drug resistance [5, 6], infected individuals suffer high morbidity and mortality [7]. Collectively, Aspergillus fungi affect millions of patients and cause hundreds of thousands of life-threatening infections every year [8, 9]. Not all pathogenic Aspergillus species exhibit the same infection rates [10, 11]. Approximately 70% of all Aspergillus infections are caused by A. fumigatus, whereas the remaining 30% of infections stem from other species in the genus [12]. Some of these other pathogenic species–for example, Aspergillus flavus, Aspergillus niger, and Aspergillus terreus–are distantly related to A. fumigatus [12]; each of these three species belong to different Aspergillus sections (note “section” is a taxonomic rank in-between the genus and species ranks) and show extensive genomic divergence [13]. However, there are also pathogenic Aspergillus species, such as Aspergillus lentulus and Aspergillus udagawae, that belong to the same section as A. fumigatus (section Fumigati) and are much more closely related [14–16]. In contrast, the vast majority of Aspergillus species, including many close relatives of the pathogens mentioned above, are either not pathogenic or very rarely cause disease [16, 17]. Interestingly, A. fumigatus and a few other species in the genus can also cause opportunistic infections, mainly in other mammals and birds, and occasionally in other vertebrates and invertebrates [18]. Even though a great deal is known about some aspects of A. fumigatus pathogenicity [2, 19, 20], we have only recently begun to examine why pathogenicity varies so dramatically across the entire genus and the traits and genetic elements that contributed to this variation. Addressing this question requires that we consider the fact that pathogenic Aspergillus species are not


Introduction
Species in the genus Aspergillus are saprophytic filamentous fungi that are most commonly found in soil and litter environments of subtropical and warm temperate latitudes [1]. Inhalation of asexual spores produced by Aspergillus fumigatus and a few other species in the genus cause a group of diseases collectively referred to as aspergillosis [2]. The most severe form of aspergillosis is invasive aspergillosis, which primarily affects individuals with compromised immune systems or preexisting lung conditions [3]. Since drugs targeting invasive aspergillosis are not always effective due to our lack of understanding of how they function inside the human host [4] and the evolution of drug resistance [5,6], infected individuals suffer high morbidity and mortality [7]. Collectively, Aspergillus fungi affect millions of patients and cause hundreds of thousands of life-threatening infections every year [8,9].
Not all pathogenic Aspergillus species exhibit the same infection rates [10,11]. Approximately 70% of all Aspergillus infections are caused by A. fumigatus, whereas the remaining 30% of infections stem from other species in the genus [12]. Some of these other pathogenic species-for example, Aspergillus flavus, Aspergillus niger, and Aspergillus terreus-are distantly related to A. fumigatus [12]; each of these three species belong to different Aspergillus sections (note "section" is a taxonomic rank in-between the genus and species ranks) and show extensive genomic divergence [13]. However, there are also pathogenic Aspergillus species, such as Aspergillus lentulus and Aspergillus udagawae, that belong to the same section as A. fumigatus (section Fumigati) and are much more closely related [14][15][16]. In contrast, the vast majority of Aspergillus species, including many close relatives of the pathogens mentioned above, are either not pathogenic or very rarely cause disease [16,17]. Interestingly, A. fumigatus and a few other species in the genus can also cause opportunistic infections, mainly in other mammals and birds, and occasionally in other vertebrates and invertebrates [18].
Even though a great deal is known about some aspects of A. fumigatus pathogenicity [2,19,20], we have only recently begun to examine why pathogenicity varies so dramatically across the entire genus and the traits and genetic elements that contributed to this variation. Addressing this question requires that we consider the fact that pathogenic Aspergillus species are not a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 dependent on their hosts for survival and their pathogenic effects are entirely accidental or opportunistic. Thus, understanding the evolution of pathogenicity in the genus requires that we understand how variation in the traits that enable Aspergillus species to survive in their natural soil and litter environments has rendered a few of these species capable to establish infections inside human hosts.
In this pearl article, we focus on section Fumigati, a lineage of~60 species that includes A. fumigatus and its close relatives [16,21], to discuss the latest advances in our understanding of the evolution of pathogenicity in Aspergillus. More broadly, given that the ability to cause human disease has repeatedly evolved across the fungal tree of life, and that the vast majority of human fungal pathogens have non-pathogenic close relatives [22], understanding the evolution of Aspergillus pathogenicity can serve as a model for studying fungal pathogenicity in general.

Pathogenicity in Aspergillus section Fumigati fungi has evolved multiple times independently
Only a handful of species in section Fumigati are considered pathogenic [10,16,21], and the distribution of these pathogens on the section's phylogeny [16] suggests that the ability to cause human disease has evolved at least 5 times independently (Fig 1). For example, whereas A. fumigatus infects >300,000 humans per year [9], its close evolutionary relative A. fischeri, whose protein sequences exhibit, on average, 95% similarity to their A. fumigatus orthologs, has only rarely been reported to cause human disease and is not considered clinically relevant [23,24]. Interestingly, evolutionary reconstruction shows that A. fumigatus pathogenicity likely evolved after the species diverged from either A. fischeri or from its even closer nonpathogenic relative A. oerlinghausenensis, suggesting that the last common ancestor of these three species was non-pathogenic (Fig 1). The same is true for most other pathogens in the section. For example, the pathogenic A. udagawae accounts for a few thousand infections per year [25][26][27], but its close relatives (e.g., Aspergillus aureolus, Aspergillus acrensis, and Aspergillus wyomingensis) are not considered clinically relevant (Fig 1), suggesting that the pathogenicity of A. udagawae evolved independently and that the common ancestor of A. udagawae and its close relatives was non-pathogenic.

The observed spectrum of pathogenicity cannot be explained by ecology or ascertainment bias and is likely to have a genetic component
Several ecological attributes, such as the global ubiquity of their small and readily airborne asexual spores, are thought to contribute to the pathogenicity of Aspergillus molds in general [11], and to the pathogenicity of A. fumigatus in particular [19,32]. Although these ecological attributes are undoubtedly important for infecting and causing disease in a human host, the observed spectrum of pathogenicity among section Fumigati species cannot be solely explained by the known differences in species' ecologies [1,[33][34][35]. For example, previous studies have shown that A. fischeri, one of the closest non-pathogenic relatives of A. fumigatus, can be frequently isolated from a variety of locales, including soils, fruits, and hospitals [33][34][35]. Case in point, approximately 2% of the fungi isolated from the Beijing Hospital environment were A. fischeri [34], but only a handful of infections caused by this fungus have ever been reported [10,23,24].
Another ecological attribute known to be associated with fungal pathogenicity is thermal tolerance [36]. However, all species in section Fumigati that have been tested can grow, dependent on the growth medium used, at 37˚C [14]. These data suggest that pathogenicity in Aspergillus section Fumigati is not simply due to species' abilities to grow at the human body temperature. Nevertheless, species in the section do show substantial differences in how well they can grow at 37˚C [14], but these differences are likely to have a genetic basis (see below). It would be highly interesting for future studies to examine growth curves of closely related pathogenic and non-pathogenic species in specific stressful and human infection-relevant conditions (e.g., at 37˚C, with limited nutrient availability, low levels of oxygen and high levels of oxidative stress).
Another potential explanation for some of the observed differences in the spectrum of pathogenicity among species is ascertainment bias. In the context of Aspergillus pathogenicity, ascertainment bias is a term that describes systematic deviations from the true incidence of disease caused by a given species. These systematic deviations stem from the methods used to taxonomically identify (ascertain) clinical isolates and estimate how often they cause disease, i.e., from our failure to measure the true numbers of infections caused by so-called cryptic species, namely organisms that are morphologically similar to major pathogens, such as A. fumigatus, but genetically distinct from them [10]. The true burden some of these cryptic species, including species currently thought to not be clinically relevant, place on human health is unknown and may be in several cases underestimated [22]. However, numerous molecular typing studies of clinical isolates from diverse countries routinely identify the known pathogens in section Fumigati (Fig 1), but not the non-pathogens [25][26][27], indicating that the observed variation in pathogenicity is not solely an artifact of species misdiagnosis. These data suggest that the differences in pathogenicity observed across Aspergillus section Fumigati have, at least partially, a genetic basis. Support for the role of genetic differences in contributing to the observed spectrum of pathogenicity is provided by the numerous traits, and their underlying genes and pathways, that are required for pathogenicity in A. fumigatus [2,19,20,37] and have been found to exhibit substantial genetic and phenotypic diversity among section Fumigati species. These traits include thermotolerance, the ability to respond to multiple environmental stresses, including antifungal drugs, and the capacity to biosynthesize a range of structurally diverse secondary metabolites [14,30,38,39].

Two models for the evolution of pathogenicity in Aspergillus molds
One useful approach for gaining insights into the genetic foundations of the multiple, independent origins of pathogenicity in Aspergillus section Fumigati is the development of conceptual models that describe the differences that we would expect to observe in genomic comparisons involving pathogenic and non-pathogenic species. We propose two alternative, although not necessarily mutually exclusive, models, which we have named the "conserved pathogenicity" model and the "species-specific pathogenicity" model (Fig 2). The conserved pathogenicity model posits that A. fumigatus and other pathogenic species in section Fumigati share common pathogenicity traits and genetic elements (or shared differences in genetic elements) that are absent in non-pathogens (e.g., traits / elements E1 and E2 in Fig 2) or vice versa (e.g., trait / element E3 in Fig 2). In contrast, the species-specific model posits that each pathogen

Fig 1. Reconstruction of the ability to cause human disease on the Aspergillus section Fumigati phylogeny suggests that pathogenicity evolved multiple times independently in the lineage.
For the trait reconstruction inference, Biosafety Level (BSL) 2 organisms were considered pathogenic and BSL1 organisms or organisms that so far lack BSL labelling were considered non-pathogenic; these transitions to a pathogenic lifestyle (i.e., from BSL1 to BSL2) are labelled by red bars on the figure. Note that clinical isolates from humans or other mammals from a few additional species in the section have been identified [10,28]; this handful includes relatively newly described species that some authors consider to have pathogenic potential (e.g., Aspergillus novofumigatus [29]) as well as organisms thought to be on the non-pathogenic end of the spectrum (e.g., A. fischeri [30,31]). The phylogeny of the section was redrawn from Hubka et al. [16]. contains a unique suite of traits and genetic elements (or unique differences in genetic elements) that distinguish it from its non-pathogenic relatives; these traits / elements could be ones that are uniquely present in a given pathogen but absent in related pathogens and nonpathogens (e.g., traits / elements E4 and E5 in Fig 2) or vice versa (e.g., trait / element E6 in Fig 2). Note that these shared or species-specific genetic elements (or differences in genetic elements) are not limited to differences in gene content but to any type of genetic variation that alters pathogenicity trait values. These variants can range from, for example, differences in a single or in a handful of nucleotides within otherwise conserved protein-coding or non-coding (regulatory) regions to larger-scale differences concerning the presence of entire genetic pathways and networks.
Discerning which model explains the repeated evolution of pathogenicity is key for developing research strategies to understand the underlying molecular mechanisms in the genus and, more broadly, in filamentous fungi and beyond. For example, the conserved pathogenicity model would predict that pathogenicity stems from the action of conserved genetic elements, suggesting that known genetic determinants of virulence in A. fumigatus [40] would be great candidates for involvement in virulence in other pathogenic Aspergillus species. It is the adoption of the conserved pathogenicity model that underlies recent examinations of the degree to which genetic elements known to contribute to A. fumigatus pathogenicity are conserved in other species [29,30,41]. In contrast, the species-specific model would predict the opposite, namely that the genetic determinants of virulence are unique to each pathogen, suggesting that extrapolations of knowledge on virulence mechanisms from one pathogenic species to another would be futile. Genomic comparisons between the major pathogen A. fumigatus and its close non-pathogenic relative A. fischeri [30,31] as well as broad comparisons of select species across the genus [29,41] have begun to shed light on the validity of, and provide support for, both of these models.

Support for the conserved pathogenicity model
Consistent with one of the predictions of the conserved pathogenicity model, examinations of dozens of A. fumigatus genes known to be associated with virulence [40] have shown that most of these genetic determinants of virulence are highly conserved in closely related species [29,30]. For example, a recent genomic comparison of the major pathogen A. fumigatus with its close, non-pathogenic relative A. fischeri showed that 48 of 49 known genetic determinants of A. fumigatus virulence (e.g., CrzA, the C 2 H 2 -type zinc finger transcription factor involved in calcium ion homeostasis, or LaeA, a methyltransferase and master regulator of secondary metabolism) were highly conserved in A. fischeri [30]. However, these results also suggested that the differences in virulence among organisms spanning the pathogenicity spectrum may not be primarily due to differences in gene content, which virtually all genomic comparisons of fungal pathogens and non-pathogens in the genus [29][30][31]41] and beyond [42][43][44] have heavily focused on. How these conserved genetic determinants of virulence function in other pathogenic, as well as in non-pathogenic, species is an interesting future direction of inquiry.

Support for the species-specific pathogenicity model
Comparisons of gene content between closely related Aspergillus species have also identified numerous genes that appear to be species-specific. For example, a broad scale comparison of A. fumigatus strains Af293 and A1163 against A. fischeri and A. clavatus found that approximately 8.5% of genes were unique to A. fumigatus and absent from the other two species [31]. These A. fumigatus-specific genes tend to reside near the ends of chromosomes (i.e., are subtelomeric) and have functions associated with metabolism (e.g., secondary metabolism, transport, and detoxification), raising the hypothesis that some of them may aid A. fumigatus survival inside the human host [31]. For example, more than two thirds of A. fumigatus biosynthetic gene clusters are absent from the closely related non-pathogen A. fischeri [30,38]; however some of these biosynthetic gene clusters are found in other species in section Fumigati, suggesting that they were lost in A. fischeri rather than originated within A. fumigatus. Several other clusters and their secondary metabolites appear to have uniquely evolved in A. fumigatus, or their loss in species closely related to A. fumigatus was so widespread that they are now uniquely present in A. fumigatus.
One recent, striking example in support of the species-specific model was the discovery of the gene hypoxia-responsive morphology factor A or hrmA (Afu5g14900) [45]. While homologs of hmrA are present in other distantly related fungi, this subtelomeric gene is polymorphic within A. fumigatus and is absent from the genomes of all sequenced Aspergillus section Fumigati species. Investigation of hrmA function shows that it likely regulates a cluster of genes, which also appear to be absent from the genomes of other Aspergillus section Fumigati species, that collectively contribute to the generation of a morphotype that facilitates adaptation to very low oxygen conditions encountered by the fungus inside human lungs [45].

Concluding remarks
This pearl has focused on Aspergillus section Fumigati, outlining two general models for the repeated evolution of pathogenicity in the section and how they could aid in the design of experiments aimed at elucidating the underlying molecular changes responsible. The same approach could also be employed for developing strategies for the burgeoning problem of drug resistance (e.g., to what extent are mechanisms of drug resistance conserved across pathogenic species?). But the utility of studying the evolution of Aspergillus section Fumigati pathogenicity goes beyond, as Carl Sagan so aptly put it, understanding "the present". By posing questions such as "are species that are currently considered non-pathogenic but contain conserved genetic determinants of virulence likely to emerge as new pathogens?", we believe that an evolutionary approach-by identifying the presence of constellations of genes and traits associated with pathogenicity in non-pathogens-also holds promise for predicting the emergence of new pathogens in the future.