Itˈs not all about us: evolution and maintenance of Cryptococcus virulence requires selection outside the human host

Abstract

Cryptococcus is predominantly an AIDS-related pathogen that causes significant morbidity and mortality in immunocompromised patients. Research studies have historically focused on understanding how the organism causes human disease through the use of in vivo and in vitro model systems to identify virulence factors. Cryptococcus is not an obligate pathogen, however, as human–human transmission is either absent or rare. Selection in the environment must thus be invoked to shape the evolution of this taxa, and directly influences genotypic and trait diversity. Importantly, the evolution and maintenance of pathogenicity must also stem directly from environmental selection. To that end, here we examine abiotic and biotic stresses in the environment, and discuss how they could shape the factors that are commonly identified as important virulence traits. We identify a number of important unanswered questions about Cryptococcus diversity and evolution that are critical for understanding this deadly pathogen, and discuss how implementation of modern sampling and genomic tools could be utilized to answer these questions.

Introduction

Cryptococcus neoformans is the most frequent cause of meningitis among adults in Africa, causing 15–20% of all AIDS-related diseases (Park et al., 2009). Owing to the rapid emergence and extremely high mortality of cryptococcal meningitis, the academic Cryptococcus field has been designed around a common, unified goal: to improve patient survival and develop new treatments through understanding how the organism causes disease. The field contains a comparatively high number of medical professionals (alongside academic researchers) and thus access to clinical samples, patient data and clinical outcome is available. Tremendous effort has been expended to identify the fungal virulence factors that enable disease at various stages of the infection process, and to develop in vivo systems that model human disease. This intense focus on virulence factors has left unanswered many questions about how this fungal pathogen has evolved to cause disease in humans and how environmental stressors ultimately selected for traits important for pathogenesis. Here we seek to examine and define parallels between what is known about the myriad stressors a cryptococcal genotype must be able to survive in both the ecological environment and the human host.

Not all strains are created equal

Cryptococcus strains vary significantly with respect to virulence and their ability to cause human disease. Early studies based on a wide variety of methods identified considerable genotypic variation among strains [reviewed in (Hagen et al., 2015)], with early differentiation into four different serotypes [serotypes A, B, C, D (Evans, 1950; Wilson et al., 1968)] later followed by further differentiation into two species (including two varieties) based on mating compatibility (serotype A: C. neoformans var grubii; serotype D: C. neoformans var. neoformans; serotypes B and C: C. gattii). Differences in human virulence between the two species rapidly became apparent – C. neoformans was implicated in the majority of disease in immunocompromised individuals globally, while C. gattii strains were more common in apparently healthy individuals in tropical and sub-tropical regions as well as a current outbreak in North America (Kidd et al., 2004). The most recent molecular techniques have led to further reclassification, with the two C. neoformans varieties designated as separate species, renaming C. neoformans var. grubii as C. neoformans and C. neoformans var. neoformans as C. deneoformans, and the identification of five separate species of C. gattii (Hagen et al., 2015). As differences in ecological niche and thus selective pressures are likely to be significantly different between these closely related species, here we will focus solely on C. neoformans and C. deneoformans.

Cryptococcus species are heterothallic with two mating types, MATα and MATa. Interestingly, greater than 90% of strains identified to date are MATα and this extreme skew in mating type was originally thought to limit the frequency of sexual reproduction relative to asexual budding. Recent studies have identified the existence of unisexual reproduction (MATα × MATα), however, and population genetic analyses consistently reveal a low level of sexual recombination within and between the two species (Lin et al., 2005; Lin et al., 2007), suggesting at least a low level of gene flow. The relative proportion of species in clinical isolates varies worldwide, with a much higher proportion of C. deneoformans strains (and C. neoformans/C. deneoformans hybrids) isolated in Europe compared with the Americas (Desnos-Ollivier et al., 2015; Viviani et al., 2006). C. neoformans strains cause >90% of all infections and 99% of cryptococosis in AIDS patients, while C. deneoformans strains (and the hybrids) are generally less virulent (Casadevall and Perfect, 1998).

Evolution of an environmental pathogen (itˈs not all about humans)

An important characteristic that distinguishes Cryptococcus from obligate human fungal pathogens (such as Candida albicans or Pneumocystis jiroveci), is that it is a saprophytic environmental organism, i.e. it does not rely on a human or other animal host to survive (Casadevall et al., 2003). Futhermore, unlike the majority of opportunistic human fungal pathogens such as Aspergillus and the dimorphic fungi, Cryptococcus belongs to the Basidiomycetes rather than the Ascomycetes. Cryptococcus is thus evolutionarily distinct from other human pathogens and it is not clear that insights gained from these other taxa can be applied to Cryptococcus.

A complete picture of the Cryptococcus life cycle includes life within the environment as well as multiple distinct areas in the human host. Humans are first exposed to environmental Cryptococcus spores that enter the lungs through inhalation (Botts and Hull, 2010). Cryptococcus then establishes a pulmonary infection that is either cleared or remains latent in healthy individuals. In immunocompromised individuals the organism can become pathogenic and spread from the lungs through the bloodstream to the central nervous system where it can eventually cross the blood–brain barrier (BBB) to enter the brain and cause cryptococcal meningitis. These different stages can be thought of as acting as a unidirectional genotypic filter (Figure 1). There may exist isolates that thrive in the environment yet are unable to grow within the lungs. Similarly, isolates capable of causing a latent lung infection need not possess the ability to cross the BBB. Thus, if it were possible to sample the global pool of genotypes, there should be a decrease in genotypic diversity at each step along this path from environment to brain. As the different invertebrate and vertebrate model organisms capture different aspects of strain virulence [Box 1 (He et al., 2012; Lee et al., 2016)], it is not possible to gain a comprehensive view of how a particular strain will behave during a human infection using a single model system.

Figure 1.

Selection on cryptococcal strains is a unidirectional filter from the environment to the brain. As all strain genotypes need to survive in the environment, selection should be strongest at the top of the pathway and weakest at the bottom; this should be reflected in genetic diversity and strain performance

Box 1. Model systems to study cryptococcosis.

Invertebrate models

Interactions with these organisms may mimic microbial biotic interactions occurring in the environment; assays are often performed at ambient temperatures, but cannot examine processes involving adaptive immune responses.

Amoebae Thought to mimic biotic interactions that occur in the environment. Some Cryptococcus genes important for survival in macrophages are also expressed in amoebae (Derengowski Lda et al., 2013).

Wax moth larvae Assay was developed as a low-cost alternative to other invertebrate models. Galleria mellonella has been used successfully in many studies as a preliminary screen for genes involved in virulence (Mylonakis et al., 2005).

Nematodes Perhaps the simplest and most high-throughput of the invertebrate models. Cryptococcus is lethal to Caenorhabditis elegans when used as the sole nutrient source for the worm (Mylonakis et al., 2002).

Fruit flies Drosophila melanogaster can control the infection unless Cryptococcus is grown under nutrient limitation prior to assay (Apidianakis et al., 2004) similar to subclinical or dormant infections thought to exist in humans.

Vertebrate models

Owing to their multi-organ systems, the presence of an adaptive immune response, and ability to target specific aspects of the infection, vertebrate models are considered the gold standard to analyse virulence of Cryptococcus.

Zebrafish Newly developed model. This basal vertebrate has functional organ systems, an innate immune system, and is transparent during early development allowing real-time analysis of host–pathogen interactions (Bojarczuk et al., 2016; Tenor et al., 2015)

Mice Most commonly used vertebrate model owing to consistency of infections and ability to examine impact of host genetic background. Route of infection (inhalation, intravenous, intracranial) can be varied to analyse specific aspects of the infection (Casadevall and Perfect, 1998).

Rats Granuloma formation in the lungs that mimics the subclinical or dormant infection thought to exist in humans. As with human disease, immune suppression can stimulate dissemination.

Rabbits Cryptococcus can be injected intracisternally into the subarachnoid space to establish a central nervous system infection. Cerebral spinal fluid can be serially sampled allowing analysis of the infection at multiple timepoints (Perfect et al., 1980).

Selection to maintain viability for these different Cryptococcus life stages can be thought of as akin to selection at the different reproductive stages of a multicellular organism [first put forward by Williams (Williams, 1957)]. In the case of a multicellular organism, selection on a genotype is strongest between zygote and reproductive maturity, with a rapid decline shortly after maturity (intuitively, one needs to stay alive to reproduce, with much weaker selection post-reproduction). In an analogous fashion, selection on a Cryptococcus genotype should be predominantly for survival in the ecological environment. A Cryptococcus genotype can thus have extremely high fitness even if it is not able to cross the BBB and cause human mortality. This line of thinking suggests that virulence traits that enable Cryptococcus to be a successful human pathogen are likely to be pleiotropic and should also confer a selective advantage to abiotic or biotic pressures in the ecological environment. This idea has previously been termed ‘dual-use virulence’ or ‘accidental virulence’ (Casadevall and Pirofski, 2007; Casadevall et al., 2003), and is very similar to the process of exaptation, a term that describes traits which have been co-opted for usage in a way other than that they were selected for (Gould and Vrba, 1982). To comprehensively understand the virulence traits that Cryptococcus utilizes to cause human infection, there is much to be gained from understanding how selection acts in the ecological environment outside of the human.

Whither the environmental diversity?

From a historical prospective, the earliest isolates of Cryptococcus from the late 1800s reflect the full diversity of known environments, including a clinical human isolate, an environmental isolate from peach juice and an isolate from a non-human animal host – the lymph node of an ox (Casadevall and Perfect, 1998). Accordingly, C. neoformans is often referenced as ubiquitous in the environment. Although it may well be true that C. neoformans has a cosmopolitan distribution, studies that have sought to isolate it from environmental samples generally report a far greater number of negative samples than positive samples, typically no higher than 10% (Mitchell et al., 2011). An extensive review by Mitchell and colleagues (Mitchell et al., 2011) that examined the niches of C. neoformans and C. deneoformans identified avian (particularly pigeon roosting areas and droppings) and arboreal habitats (including bark, tree-base soil, tree hollows and the flowers of a multitude of tree species) in tropical and semitropical climates as the predominant, perennial environments. Yet these two species have also been isolated from a large number of additional environmental niches including a variety of solid materials (soil, sand, rock, dust, roads, tires, shoe soles) and liquid environments (fresh water, sea water, brackish water, swamp debris, mud, tree sap), from plants and vegetation, associated with animal droppings and habitats, and from the air [see references within (Mitchell et al., 2011)]. Thus, a comprehensive picture of the true geographic distribution and incidence of Cryptococcus probably suffers significantly from both publication bias and ascertainment bias: owing to the tremendous effort that is required to conduct environmental sampling, studies are biased towards sampling avian and arboreal habitats and thus projects that examine only these habitats continue to be published. Furthermore, the spectrum of sites that have failed to yield positive samples remains largely unknown owing to frequent non-reporting (Mitchell et al., 2011).

Given these sampling limitation caveats, we can examine the abiotic characteristics of the two predominant ecological niches (avian and arboreal environments) to seek clues about the nature of abiotic selection acting on environmental strains. A number of factors have been identified that correlated with the presence of Cryptococcus strains within small spatial scales. Studies of pigeon droppings have repeatedly identified a significant relationship between positive sample isolates in dry/dessicated droppings compared with moist droppings (Granados and Castaneda, 2005; Mseddi et al., 2011; Ruiz et al., 1981), and an overabundance of strains from protected areas that favour desiccation compared with unprotected areas (Montenegro and Paula, 2000). Climactic variables including moderate to high rainfall, high temperature and high humidity are also correlated with the presence of Cryptococcus isolates (Granados and Castaneda, 2005; Li et al., 2012; Mak et al., 2015), although note that, at least in the case of high humidity, this is contrary to the overabundance of Cryptococcus cells found in dessicated droppings compared with moist droppings. The ‘typical’ abiotic environment of Cryptococcus might thus directly select for the ability of C. neoformans to rapidly acclimate to different aspects of the clinical environment including variation in nutrient limitations, fluctuations in pH and hypoxia tolerance (Kronstad et al., 2012).

Following the logic in Figure 1, the diversity of environmental strains should be greater than (or equal to) the diversity of clinical strains from a given geographic region, and all clinical strains should be represented in the environmental strain set. Studies that have directly examined this question have generally failed to identify this expected relationship, often even finding the opposite pattern (Figure 2). A recent study of Botswana isolates compared 125 clinical isolates (cultured from 64 patients with HIV/AIDS and cryptococcal meningitis) and 179 environmental isolates (from 77 samples taken predominantly from trees, but also avian droppings and soil samples). The clinical isolates were significantly more diverse than the environmental isolates, and some clinical isolates were very rare or entirely absent from the environmental strain sets (Chen et al., 2015). Parallel results have also been identified from strains isolated in Cuba (Illnait-Zaragozi et al., 2010), the southern US (Litvintseva et al., 2005), Brazil (Zhu et al., 2010), Japan (Yamamoto et al., 1995) and Asia (Khayhan et al., 2013). It thus seems likely that a primary niche (or niches) remains to be discovered and that we are missing a major reservoir for environmental isolates.

Figure 2.

(A) Environmental strains compose the source of all Cryptococcal strains (as human–human transmission is not observed); all clinical strain genotypes should be present in the global set of environmental strains. (B) This expected pattern is not observed. Studies that have compared environmental and clinical strain genotypes often find clinical genotypes that are not represented in the environmental set. Thus, the true diversity of environmental strains must be larger than we are currently sampling (dashed line) and we are likely missing at least one major abiotic or biotic niche

As predicted by our model (Figure 1), not all environmental strains are capable of causing disease in humans. The degree of filtering from the global population of environmental isolates to isolates that are capable of causing disease seems highly dependent on the study. A set of 10 environmental isolates from bird excreta in Brazil found that all strains were capable of infecting murine lungs and brain, with 9/10 strains capable of killing 5/5 mice by 33 days (Pedroso et al., 2010). In contrast, only 1/11 environmental strains from North Carolina and California pigeon excreta from the USA caused disease in mice within 60 days of post-intranasal infection, compared with 7/10 clinical isolates from a diversity of patients (Litvintseva and Mitchell, 2009). Other studies have found intermediate results (da Silva, 2006; Fromtling et al., 1989; Movahed et al., 2015).

Variation in strain ability to cause disease might theoretically arise from multiple factors. From a methodological perspective, the way in which ‘disease-causing’ is assessed could greatly influence whether a strain is classified as virulent or avirulent. Based on our proposed selection regime outlined above, a strain that is capable of establishing a lung infection may not disseminate or cause death. Thus, strains may be classified differently depending on the infection model utilized to analyse virulence (Box 1; e.g. whether mice are infected intranasally vs. through the bloodstream). There could also, however, be environmentally derived variation in the strain set that is examined – it may well be that strains isolated from different niches (e.g. avian droppings vs. an arboreal habitat) or different geographic locations may be more or less prone to cause disease. Additional studies that utilize random sampling techniques and extend the diversity of sites that are sampled are required to address these questions.

Abiotic selection in the environment: backdoor to human virulence?

The overarching picture that has emerged is that, although a significant fraction of environmental strains are capable of causing disease, the underlying factors that differentiate virulent from avirulent strains remain unclear. This suggests that selection in the environment maintains or promotes these same traits that assist Cryptococcus in the colonization of different body sites, dealing with the immune system and causing human disease. For example, the ability to grow at 37 °C (a trait that is shared by all human pathogens) can be readily explained by the propensity of Cryptococcus cells to be found in high-temperature environments. However, of the estimated 1.5 million fungal microbial taxa (Hawksworth, 2006) (a number that may well be lower than the true number), a small minority cause disease in animals, and even fewer [~300 (Taylor and Gurr, 2014)] cause disease in humans. Furthermore, when one considers the myriad selection environments required for a Cryptococcus cell to traverse the human body from lungs to brain (Figure 1), it seems unlikely that the abiotic environment could select for these abilities. Instead some aspect of environmental selection may pre-dispose C. neoformans, and to a lesser extent C. deneoformans, to cause human disease.

To seek clues about the basis of human disease we need to examine virulence traits in the context of environmental selection. The two best-studied virulence traits are the ability to form a polysaccharide capsule and the production of melanin. In a clinical setting, the capsule has multiple functions including a reduction of host immune responses, regulation of phagocytosis by macrophages and providing an antioxidant defence mechanism inside macrophages [see references in (OˈMeara and Alspaugh, 2012)]. Although polysaccharide capsules are also important virulence factors in Gram-negative and Gram-positive bacterial pathogens [reviewed in (Park et al., 2015)], Tremella mesenterica, a close non-pathogenic relative of C. neoformans that grows primarily on dead tree branches, also produces a similar polysaccharide capsule (De Baets et al., 2002), as does Cryptococcus liquefaciaens, a related species commonly found in the high Arctic and hypersaline waters (Araujo, 2012). The capsule is thought to play an important role in environmental desiccation resistance (Granger et al., 1985; Park et al., 2014), which is consistent with the characterization of the ‘typical’ abiotic environment of Cryptococcus identified above. Importantly, genes that cause defective capsule production also interfere with spore formation (Kronstad et al., 2011), perhaps providing a positive-feedback mechanism to maintain capsule production independent of its role in human infection.

Melanin production by laccase functions as a defence mechanism inside the human host. Laccase enzymes have been widely identified across the phylogenetic kingdoms, are involved in reproduction (through conidiation, fruiting body production and sporulation), play a role in degradation (releasing cell nutrients for consumption), the protection of lignin-like pigments (that can influence cell wall-protective mechanisms) and act as production against oxidizing toxic compounds [reviewed in (Chen and Williamson, 2001)]. Melanin production is also widely dispersed across bacteria, animal, plant and fungal kingdoms [reviewed in (Butler and Day, 1998)], and is found in other fungal species capable of causing human disease including the dematiaceous fungi (‘darkly pigmented fungi’) [reviewed in (Wong and Revankar, 2016)]. In contrast to the majority of fungal taxa, however, C. neoformans can only make melanin in the presence of exogenous substrate (Casadevall et al., 2000). In the environment, melanin is thought to protect C. neoformans cells against environmental stresses such as UV light (Wang and Casadevall, 1994), oxygen radicals (Jacobson et al., 1994) and has been shown in vitro to influence survival under large temperature extremes (at or above 45 °C and at or below −20 °C) (Rosas and Casadevall, 1997). In other fungal taxa melanin has been shown to offer protection against enzymatic lysis in natural soils and to bind metals (preventing the entry of toxic metals and/or concentrating and bioabsorbing essential metals), and may play a role in protection against dessication [reviewed in (Butler and Day, 1998; Treseder and Lennon, 2015)]. C. neoformans in pigeon excreta is melanized (Nielsen et al., 2007; Nosanchuk et al., 1999), thus further supporting the assertion that melanization is beneficial and probably selectively advantageous to Cryptococcus strains in the environment.

There is not a straightforward link between defects in the genes that are known to affect capsule or melanin formation and clinical parameters of disease such as lung colonization, uptake by macrophages or crossing of the BBB (Griffiths et al., 2012; Liu et al., 2008; Shea et al., 2006; Tefsen et al., 2014). Although some degree of capsule and melanin seem to be required for survival in the host, there does not seem to be a qualitative relationship between nuances in trait presentation and the degree of virulence in animal models (McClelland et al., 2005; OˈMeara and Alspaugh, 2012), and recent studies examining in vitro and in vivo capsule and melanin production in the context of human disease have shown mixed results (Beale et al., 2015; Boulware et al., 2016; Sabiiti et al., 2014; Wiesner et al., 2012). Furthermore, in very few cases are environmental strains, including avirulent strains, entirely lacking the ability to form a polysaccharide capsule or produce melanin. It thus seems likely that neither of these traits are the sole basis for Cryptococcus human virulence and therefore other virulence traits are necessary.

A number of secreted extracellular enzymes are also commonly invoked as important virulence factors that play diverse and important roles throughout the course of a human infection (Almeida et al., 2015; Wozniak et al., 2015). These enzymes are not specific to pathogenic fungi, yet in a number of cases there may be evidence that they have evolved specialized function independent of their historical role in non-pathogenic taxa. The factor (or factors) that have pre-disposed C. neoformans to pathogenicity might well not be something novel to these taxa, but something that has been altered to be specialized in a particular way that aids their role as a pathogen.

Phospholipases, for example, have been identified as important virulence factors in multiple fungal pathogens, yet are also broadly found in non-pathogenic and industrial yeasts (Djordjevic, 2010). In Cryptococcus infections they are necessary for the initiation of pulmonary infection as well as the dissemination from the lung to the bloodstream (Santangelo et al., 2004). Phospholipase B homologues in Cryptococcus and Candida seem to be specialized to play a role in human infection in unique ways compared with homologues from non-pathogenic yeast. For example, C. neoformans and C. albicans Plb1 enzymes exhibit differences in their preferred substrates compared with non-pathogenic yeast with 10- to 200-fold higher specific activities for cell membranes and surfactants and are more likely to digest cell membranes and surfactants than non-pathogenic Plb1 (Djordjevic, 2010). In addition, C. neoformans Plb1 has an acidic pH optimum that may be adapted for utilization inside macrophages (Chen et al., 2000).

In addition to the phospholipases, there are many examples of other dual-purpose enzymes. Urease, which catalyses the hydrolysis of urea into carbon dioxide and ammonia, is common among many human bacterial and fungal pathogens (Cox et al., 2000; Rutherford, 2014). For Cryptococcus, urease is important for central nervous system dissemination and crossing the BBB (Olszewski et al., 2004), and promotes a non-protective T2 immune response (Osterholzer et al., 2010). In the environment, urease probably assists C. neoformans in converting urea into a usable nitrogen source in pigeon guano (Cox et al., 2000). DNAse production is found in bacterial pathogens (Sanchez and Colom, 2010), as well as non-pathogenic Cryptococcus, Rhodotorula and Tremella (Cazin et al., 1969). Interestingly, higher activity (and greater diversity) of DNAse was observed from clinical C. neoformans strains compared with environmental strains (Sanchez and Colom, 2010). Streptococcus DNAse activity functions in avoidance of neutrophil-killing, and it has been hypothesized that it may play a similar role for C. neoformans (Sanchez and Colom, 2010). Superoxide dismutases catalyse the dismutation of toxic superoxide anions into hydrogen peroxide, and thus neutralize toxic levels of reactive oxygen species. These metalloenzymes differ among species primarily in terms of gene number and metal cofactor (Frealle et al., 2005), and have been identified as virulence factors in the majority of human fungal pathogens (Cox et al., 2003). Two superoxide dismutase genes (SOD1 that utilizes Cn/Zn and SOD2 that uses Mn) have been shown in C. neoformans to be involved in oxidative stress and high-temperature growth, and to facilitate growth within macrophages (Cox et al., 2003; Frealle et al., 2005).

In addition to these molecularly defined cellular characteristics and functions, recent studies have highlighted the diversity of morphological transitions that C. neoformans undergoes in response to the host environment [reviewed in (May et al., 2016)]. Large titan cells are generated in response to the pulmonary environment (Okagaki et al., 2010) and their production causes an overall reduction in phagocytosis, the major mechanism of pathogen clearance in the lungs (Okagaki and Nielsen, 2012). Cell wall changes in titan cells also modulate the host adaptive immune response to promote dissemination and disease (Crabtree et al., 2012; Wiesner et al., 2016; Wiesner et al., 2015). Similarly, small drop cells that are also produced during infection appear to be highly adapted to growth within host phagocytes (Alanio et al., 2015). The molecular mechanisms underlying the appearance and activity of these novel cell types have yet to be explored and may involve production of compounds and enzymes described above. That these cell morphologies are uniquely produced in response to the complex host environment suggests that they may have specifically evolved in response to selective agents within a biotic host organism.

The biotic environment isnˈt sterile

Enzymatic specialization and pathogenesis-related morphological transitions raise the possibility of virulence factors that are specialized to navigate through the varied stages of clinical infection, and that biotic stressors in the environment also played a role in the evolution of these traits. The realized niche of Cryptococcus environment strains is also likely to be heavily influenced by the presence (or absence) of other microbes, and biotic selective pressures may also directly or indirectly play an important role in the maintenance or acquisition of virulence factors. Indeed, decreasing the interaction with microbial competitors was the original hypothesis to explain the overabundance of Cryptococcus cells in desiccated pigeon droppings (Ruiz et al., 1981). Now that genomic methods are available to more comprehensively sample a microbial community, it would be interesting to test whether the number of competitor species is indeed higher in moist droppings.

Early work that isolated other microbes from-pigeon droppings (including P. aeruginosa, B. subtilis and the amoeba, A. palestinenesis) demonstrated that these organisms displayed anticryptococcal activity (Ruiz et al., 1982), suggesting that these microbial predators could directly select for traits that enable Cryptococcus to survive and thrive within the human host. The use of the soil amoeba, Acanthamoeba castellanii, as a model to study cryptococcal virulence (Steenbergen et al., 2001) (Box 1) is predicated on this notion, as C. neoformans has been shown to interact with amoeba in their environment. Importantly, a number of parallels have been identified between the interactions of C. neoformans with amoeba and macrophages (Chrisman et al., 2010; Steenbergen et al., 2001). Approximately one-third of the genes that were differentially expressed by C. neoformans after 6 h of incubation with murine macrophages were similarly modulated after 6 h of incubation with A. castellannii (Derengowski Lda et al., 2013).

An important trait that is not easily explained by the identified abiotic pressures or biotic interactions is the ability of Cryptococcus to evade the human adaptive immune system (May et al., 2016). Although no vertebrate animal host has emerged as a significant (nor requisite) component of the cryptococcal life cycle, Cryptococcus has been isolated from an ever-increasing range of animals, including koala (Connolly et al., 1999), dogs and cats (Danesi et al., 2014; Malik et al., 1997; Sykes et al., 2010), ferrets (Morera et al., 2014), squirrels (Iatta et al., 2015), foxes (Staib et al., 1985) and even dolphins (Miller et al., 2002; Rotstein et al., 2010). Whether contact with these or other organisms is frequent enough to act as a biotic selective agent for the traits required for human infection remains unknown. Furthermore, for selection to be efficient, not only would contact with a host animal need to be frequent, but Cryptococcus strains would also require the potential to re-enter the environment after animal contact and infection. This is in important contrast to their interaction with humans, which is essentially a ‘dead-end’ in the life cycle.

Future Directions

Since the 1980s, Cryptococcus research has made great strides in understanding how this pathogen causes disease in humans, and the factors that are important for virulence. Yet we still know very little about the selective pressures, both within the environment and within the host, that have driven the evolution of pathogenicity. Recent studies have begun tracking genotypes in patients to link genetic differences back to disease outcome (Beale et al., 2015; Wiesner et al., 2012). Without linkage back to the environmental sources of the infection, however, these genotypic studies cannot tell us how strains evolved to cause disease. Similarly, an understanding of the pressures within the host that select for increased disease (i.e., dissemination to the brain to produce meningitis) is still lacking.

Application of modern genomic tools has the potential to address these important questions. Affordable whole genome sequencing allows better resolution in population genetic analysis of differences between environmental and clinical strains. Paired with unbiased environmental sampling to identify the major environmental reservoir for clinically relevant genotypes, and deep sequencing within environmental reservoirs to identify biotic interactions, whole genome sequencing studies have the potential to define the selective pressures that generate virulent Cryptococcus strains. Development of singe-cell genomic techniques in Cryptococcus is necessary to understand in vivo selective pressures that lead to meningitis and to determine whether strains are continually evolving within patients. These future studies will lead to a comprehensive understanding of the evolutionary and selective pressures that promote Cryptococcus disease in humans, and will inform which virulence traits are critical for disease progression.