The Insect Pathogens

Estimates of the number of arthropod species vary between 1,170,000 and 10 million, accounting for over 80% of all known living animal species. One arthropod subgroup, insects, is the most species-rich member of all ecological guilds in land and freshwater environments (1). As arthropods were emerging as the dominant animals they are today, fungi were also colonizing the land. Over the past 400 million years fungi and insects have coevolved a wide array of intimate interactions (2, 3). These interactions include mutualistic endosymbiosis (4); fungi as obligate food sources, such as those found in fungus-gardening ants (5); sexually and behaviorally transmitted parasites, such as Laboulbeniales (6); and the most common disease-causing agents of insects (7). Entomopathogenicity has evolved independently and repeatedly in all the major phyla of the Kingdom Fungi (3). The heterogeneity of entomopathogenic fungi probably derives from both they and their hosts having short generation times, i.e., rapidly driving new diversity with each generation, and from their occupation of a wide range of habitats, with near ubiquity in the soil and on plants. Interactions among fungi, hosts, and the environment are therefore diverse and dynamic, which complicates comparisons between different fungi infecting different insects since their interactions may be necessarily disparate. Historically, this quandary was dealt with by intensively studying the host pathogen interactions of a couple of experimentally tractable fungal species, and then extrapolating these results to distantly related species. Consequently, most of what we know about the biochemical and molecular basis of interactions between fungi and insects has been determined with the experimentally tractable hypocrealean ascomycete genera Metarhizium (family Clavicipitaceae) and Beauveria (family Cordycipitaceae). Metarhizium, in particular, has also emerged as an excellent model to explore a broad array of questions in ecology and evolution, host preference and host switching, and the mechanisms of speciation.

Comparative genomics offers a way forward for assessing poorly characterized species. Already, comparative genomics has facilitated the identification of fungal fitness traits and the selective forces that act upon them, improving our understanding of how entomopathogenic fungi interact with insects and the environment. In particular, sequence data have provided crucial information on the poorly understood ways that these organisms reproduce and persist in the environment, identified the genes involved in ecologically relevant traits, and illuminated the nature, timing, and architecture of the genomic changes governing the origin and processes of local adaptation. Alongside the recent availability of genomic resources, the wide array of experiments that can be performed with entomopathogenic fungi makes them ideal models for answering basic questions on the genetic and genomic processes behind adaptive phenotypes (a “Holy Grail” in biology). In addition, Meyling and Hajek have described how insects and their fungal pathogens could be used as model species for exploring metapopulation theory via experimental and predictive models: consideration of the interactions insect pathogenic fungi have with their host and the broader community enables ecological questions to be probed from a unique perspective (8).

There are a lot of detailed reviews dealing with entomopathogenic fungi and their development as microbial control agents, including those from Lacey et al. (9), Vega et al. (10), and several chapters in Lovett and St. Leger (11). Relevant chapters in Roy et al. (12) detail the crucial roles fungal entomopathogens play in natural ecosystems, and for methods and techniques used to study these pathogens consult the chapters in Stock et al. (13). Studies on the evolutionary genetics of insect immunity and on insect-pathogen coevolution are reviewed by Hu and St. Leger (14), and entomopathogen genomes are reviewed by Wang et al. (15). This review highlights recent advances in our understanding of insect pathogenic fungi deriving in large part from a plethora of genome sequences that have illuminated the nature of adaptive differences by which novel pathogens emerge and form. These studies have increased the utility of these fungi as important model and applied organisms, allowing us to focus attention on their convergent evolution, pathogenic strategies, and host specificity, which frequently involves manipulation of host behavior.

Early interest in insect pathogenic fungi grew mostly from economic concerns. For example, Agostino Bassi described Beauveria bassiana in 1835 as the cause of the devastating muscardine disease of silkworms, and it was instrumental in his development of the germ theory of disease (16). There is still considerable research interest in protecting beneficial insects, particularly bees, from highly potent fungal pathogens, such as the causative agent of chalkbrood, Ascosphaera apis (17). The research interest in insect pathogens has also been propelled by the enormous ability of insects to negatively impact human society. It is estimated that insects destroy approximately 18% of the world’s annual crop production, and vector-borne diseases kill millions every year (18). Since populations of most insects are regulated by density-dependent factors, the diseases, pathogens, and immune responses of insects have been a long-standing research interest. In 1880, the pioneer immunologist Elie Metchnikoff was among the first to propose practical methods of microbial biological control of an insect crop pest, initiating trials of the fungus Metarhizium anisopliae against grain beetles (19). Entomopathogenic fungi are particularly well suited for development as biopesticides because, unlike bacteria and viruses that have to be ingested to elicit disease, fungi are contact insecticides and typically infect insects by directly penetrating the surface of the insect (cuticle) and multiplying in the insect body cavity (hemocoel) (Fig. 1). Approximately 170 pest control products have been developed based on at least 12 species of fungal entomopathogens (20).

Killing the host slowly is adaptive for many entomopathogenic fungi because it allows them to replicate in large numbers in the living host, but this is a severe limitation in modern agriculture or when attempting to control a disease vector. Insect pathologists have been trailblazers in utilizing genetic engineering to improve the efficacy of beneficial fungi by improving their tolerance to environmental stresses and increasing their virulence (21, 22). Up until now, most “useful” genes for strain enhancement have come from pathogens themselves, from host insects, or from arthropods, such as spiders, that produce insect-specific toxins (Fig. 2A) (23–27, 155). Each of these sources provides a vast array of biologically active metabolites. In addition, a Metarhizium strain has been engineered that expresses a single-chain antibody fragment that blocks transmission of malaria parasites (27). Recombinant antibodies provide an additional vast array of potential anti-insect effectors that could target, for example, hormone receptors. The efficacy of Metarhizium strains expressing arthropod toxins or antimalarials is currently being tested in a purpose-built “Mosquito Sphere” (a compartmentalized facility with double-walled mosquito netting designed to test experimental fungi and insects in a natural ecosystem) in the malaria endemic region of Soumousso, Burkina Faso. The desire to use these new tools safely has driven studies on monitoring and mitigation methodologies for confinement of genetically engineered (GE) Metarhizium; development of bioconfinement strategies; survivability profiles and fitness of GE Metarhizium in the wild; studies of transgene stability over many generations; and the “evolvability” of transgenic strains if they escape containment (28). A review of future prospects for biocontrol highlights the application of transgenic Metarhizium to mosquitoes as providing GE biopesticides with the high profile necessary for widespread acceptance as safe and sustainable alternatives to chemicals (29).

As well as their direct benefit to agriculture and vector control as insect pathogens, Cordyceps/Ophiocordyceps spp., in particular, are medicinally valued because they are generally prolific producers of enzymes and diverse secondary metabolites with activities against insects, fungi, bacteria, viruses, and cancer cells (30–34). Enzymes from Metarhizium and Beauveria spp. are frequently exploited as industrial catalysts (35, 36), even though the responsible genes for these products are rarely identified. The study of infection and immunity in insects has achieved considerable prominence with the appreciation that their host defense mechanisms share many fundamental characteristics with the innate immune system of vertebrates (14). Studies on the highly tractable model organism Drosophila, in particular, have led to a detailed understanding of conserved innate immunity networks, such as Toll (37).

The recent discovery (38) that common entomopathogenic fungi, including B. bassiana and Metarhizium spp., have epiphytic interactions with plants has generated a plethora of promising avenues of research investigating commonalities and distinctions between these two lifestyles (28, 39, 40). The same molecular and genetic tools developed to probe and manipulate these fungi to better kill insects are quickly allowing development of fungi with altered persistence in the environment and that can function as biofertilizers (40–42).

Fungal-insect pathogens represent lifestyle adaptations that have evolved numerous times, and there are significant differences in host range and pathogenic strategies between the major groups (Fig. 3). Ascomycetes, the most speciose phylum, and Entomophthoromycota are known to infect a diverse array of insects, Basidiomycota infect mostly Hemiptera, and the basal zoosporic groups that produce motile spores that swim to reach their hosts (e.g., aquatic chytrids and blastoclads) usually infect Diptera (3). Araújo and Hughes (3) emphasize that, despite their ecological importance and potential applications, fungal-arthropod associations remain an understudied area of fungal biodiversity and likely harbor one of the largest reservoirs of undocumented taxonomic, functional, and genetic diversity within the Fungi.

The best-studied entomopathogenic fungi are the Entomophthoromycota and three ascomycete families within the order Hypocreales (Cordycipitaceae, Clavicipitaceae, and Ophiocordycipitaceae). In terms of pathogenic strategies, the ascomycetes are characteristically hemibiotrophic, switching from a parasitic, biotrophic phase in the hemocoel to a saprophytic phase colonizing the host cadaver. Conidia are produced on the cadaver, but, unlike the Entomophthoromycota, are not actively discharged. In some genera of Entomophthoromycota, if the forcibly discharged primary infective spores (ballistospores) miss their aerial targets, they have evolved the ability to form secondary sticky spores (capillispores), thereby creating “minefields” around cadavers to entrap crawling targets (3, 156).

Both Entomophthoromycota and Hypocreales produce resting structures for persistence in the absence of new hosts or under adverse environmental conditions. The Hypocreales have both sexual (teleomorph) and asexual (anamorph) forms, although most research has focused on the anamorphs, which tend to be more amenable to in vitro techniques and have greater practical applications. Species in the Entomophthoromycota are characteristically biotrophic (exhibiting no somatic growth after death) with a narrow host range and are common among foliar arthropods in temperate environments. Eilenberg and Pell (43) list a number of host-pathogen systems in which the ecology of Entomophthoromycota has been studied.

Entomopathogenicity evolved independently in the Cordycipitaceae, Clavicipitaceae, and Ophiocordycipitaceae, and these hypocrealean genera cluster among closely related phytopathogens, endophytes, and mycoparasites. These ancestral associations are consistent with repeated transitions (host switching) between plant, fungi, and insect hosts (4). Their teleomorphs are the most common fungi encountered in association with arthropods in tropical forests (Fig. 2) (43). Most appear to have a very restricted host range; e.g., the zombie-ant fungus, Ophiocordyceps unilateralis species complex (carpenter ant-specific teleomorphs), may have a level of specificity of one fungal species per insect species (Fig. 4) (44–46). In contrast, their anamorphic counterparts are often generalist pathogens with broad host ranges and the ability to pursue both pathogenic and saprophytic lifestyles (47). However, even within anamorph genera, there is considerable variation: thus, Metarhizium robertsii can infect hundreds of insect species spanning different orders, while M. album and M. acridum are specific to hemipterans (true bugs) and acridids (locusts and grasshoppers), respectively (48). In terms of the evolution of fungal host specificity, the study of Metarhizium species with different host ranges has shown a directional trajectory of speciation from being specialists to becoming generalists, and the specialists may have cryptic sexual stages (48). As discussed later, this raises the basic question of what is driving the host specificity of the sexual forms to a degree not observed in asexual forms of the same species. Other basic questions that remain unanswered include: why do teleomorphic ascomycetes not occur widely in temperate habitats, and are the teleomorphic ascomycetes utilizing the functional niches in the tropics that Entomophthoromycota occupy in temperate zones (49)?

Researchers studying the Entomophthoromycota, which are difficult to mass-produce, have focused on the ecology of these organisms and their role as causative agents of mass epizootics. Hajek and Delalibera (50) conclude that they have been used more frequently than other types of pathogens in classical biological control and provide a sustainable avenue for controlling arthropod pests, especially the increasing numbers of invasive species. As of February 2016, 10 strains of entomophthoromycotans have sequencing projects listed on the Genomes OnLine Database (GOLD): three are Entomophthoraceae, three are Basidiobolaceae, and four are Ancylistaceae. Leveraging the full power of genomics in the study of Entomophthoromycota will be shortly forthcoming, but currently the only complete genome in this collection is that for a strain of Conidiobolus coronatus (all others are draft genomes or incomplete). Studies on anamorphs within the Ascomycota, particularly the genera Beauveria and Metarhizium (family Cordycipitaceae and Clavicipitaceae, respectively), dominate the entomopathogen literature because they have been developed as inundative control agents, which can be applied en masse to control a pest population. To date, there is much more genomic information on ascomycete insect pathogens relative to the Entomophthoromycota because sequences are available from nine Metarhizium strains (48, 51–53), B. bassiana (54), Cordyceps militaris (32), Ophiocordyceps sinensis (anamorph, Hirsutella sinensis) (33), O. unilateralis (55), Tolypocladium inflatum (57), and Hirsutella thompsonii (56).

The comparative genome analysis of seven Metarhizium genomes (48) confirmed it is a monophyletic lineage that diverged from clavicipitacean plant pathogens and endophytes about 231 million years ago (MYA) and placed the hemipteran-specific M. album as basal in the Metarhizium clade with an estimated divergence time about 117 MYA (48). It has been suggested that the close physical proximity of the plant-associated ancestor of M. album to plant-sap-sucking hemipteran bugs may have facilitated this particular host switch to insects (3, 48).

Several Metarhizium spp. retain complex relationships with plants and endophytically colonize plant roots and the rhizosphere (the layer of soil influenced by root metabolism) (38, 39, 57, 58). B. bassiana also forms intimate endophytic relationships with a broad range of plant species, although it usually colonizes the aerial parts of the plant (59–62). Field trials showed that a rhizosphere competent, avirulent mutant M. robertsii strain survived better in grassland soil than an insect pathogenic mutant unable to adhere to root surfaces, providing experimental evidence for the importance of plant roots in maintaining populations of M. robertsii (28). Stress may play an important part in adaptive evolution to soil and root conditions (63). Cell wall and stress response genes evolved at an accelerated rate, increasing the fitness of field isolates, whereas virulence determinants were unaltered (28). M. robertsii trades insect-derived nitrogen (64) for plant-derived carbohydrates from its plant host (65). Moonjely et al. (39) suggest that, rather than shifting hosts, many of these fungi broadened their host range to exploit insects as a nitrogen source while maintaining mutualistic plant endosymbiosis. They hypothesize that the coupling of these dual lifestyles was the driving force behind this evolution: as conduits of insect-derived nitrogen, these fungi become an indispensable partner underground. As shown by their antagonism to plant pathogenic fungi (66), ability to survive exposure to lead and other heavy metals (67), and pathogenicity to soil amoebae (68), at least some Metarhizium isolates have additional unexpected flexibility in their trophic capabilities.

Cumulative evidence suggests that genes involved in the specificity of some entomopathogens to a narrow range of insects could control adhesion to the cuticle surface; exploitation of cuticle surface conditions (nutrients, humidity, specific recognition factors); ability to overcome structural and chemical barriers to penetration; and toxin production (48, 51, 69). Much research has been directed at identifying determinants of specificity and virulence, because a major goal in the development of fungal insect pathogens as biocontrol agents is to be able to control these parameters.

Entomopathogenic fungi provide many classic examples of pathogens manipulating the behavior of their hosts, the so-called fungal extended phenotype (132–134). In fact, the Fungal Kingdom stands alone in the range, extent, and complexity of their manipulation of arthropod behavior (3). This ability fascinates both scientists and nonscientists alike: it is fascinating because it inspires the macabre and touches on core philosophical issues, such as the existence of free will, and it is of enormous practical importance because parasites are ubiquitous and many have a predilection for the “immunologically privileged” site of the central nervous system. However, this location also provides a parasite with direct “access to the machinery” to alter host behavior, and a very large number take advantage of this. Unfortunately, the significance of such parasitic infections of the brain and “mind” is still underappreciated with limited information on their public health impact, and their overall impact in many areas of ecology, physiology, and evolution. Fungi probably represent a special case in this general field because of several peculiar features that usually include causing host death at the end of an infection cycle. At this stage, the pathogen depends on efficient transmission to maximize its fitness, so selection will favor positioning the dying host in time and space to achieve this. Numerous studies have documented that entomophthoralean fungi kill insect hosts during the late afternoon or evening (135, 136). This ensures that cadavers sporulate under the humid conditions of night to infect new hosts. Rapid transmission is required because the longer a cadaver is present in the field, the greater the chance that it will fall, be overgrown with saprophytic fungi, or be eaten by a scavenger (134). In many cases, the final interactions between a host and pathogen involve the pathogen directing the infected insect to seek an elevated position where wind currents and gravity can effectively disseminate conidia (the so-called “summit disease”). Summit disease has evolved multiple times within both the Entomophthoromycota and Ascomycota, and often involves an induced death-grip behavior involving the insect’s legs and/or jaws with wings outstretched to allow spores to be discharged from the body (132). Carpenter ants (Camponotus spp.) are probably the best known victims (132). An O. unilateralis infection transforms these ants into “zombie ants”; ultimately the fungus directs the ant to wander upward and, around noon (possibly after a solar cue), secure itself by biting down on foliage (twigs in temperate woods and leaf margins in rainforests) (Fig. 4) (136, 137). After the death of the host, the fungus sprouts a fruiting body (ascoma) from the back of its head and, after a few weeks of subsequent growth, this starts to rain sexual ascospores down on passing ants as they move on the forest floor or on branches below (132, 138). The modus operandi of Pandora formicae, an entomophthoromycotan infecting ants in the genus Formica, is very similar to the zombie-ant behavior manipulation seen in O. unilateralis: P. formicae manipulates ants to leave their nest, wander up onto vegetation, and use their mandibles to secure themselves so fungi are well placed for active discharge of conidia from the soft areas of the ant carcass (139, 140).

Very little is known about the mechanisms behind behavioral manipulation, and research has more often ruled out mechanisms than proved them. For example, adult flies killed by entomophthoromycotan fungi are often highly sexually attractive, facilitating spread of infectious propagules, but this is not due to the production of a sex pheromone (141). However, recent transcriptomic studies on ants infected with Pandora and Ophiocordyceps offer major insights into the mechanisms and extent to which fungi can manipulate their hosts (55, 142). The in insecta transcriptome of P. formicae elucidated the machinery that articulates its extensive morphological changes in the host, and revealed upregulation of catalases, which protect against host oxidative defenses, along with host-degrading subtilisin- and trypsin-like proteases (142). These proteases were similarly upregulated by O. unilateralis in the death-grip phase of infection, along with enzymes affecting oxidation-reduction processes and modulating serotonin and dopamine, thus, likely also modulating behavior (55). O. unilateralis sensu lato also secretes a plethora of putative compounds that have an effect on behavior, including, potentially, ergot, enterotoxins, alkaloids, polyketides, and nonribosomal peptides (55). Ergot famously causes serotonergic overstimulation of the central nervous system in humans and livestock (143), and polyketides, nonribosomal peptides, and alkaloids can be neuromodulatory agents (144). It has also been hypothesized that muscle atrophy seen during an O. unilateralis infection may be due in part to fungal enterotoxins (138). When incubated with the brains of two host and two nonhost ant species, O. unilateralis produced a different chemical cocktail for each ant species, in a manner that suggested it “knows” the brains of its target hosts and reacts accordingly (44). This goes some way to explaining why different Ophiocordyceps species seem to infect only certain ants.

Also of particular interest are potential commonalities for behavior manipulators across kingdoms (55). O. unilateralis upregulated the secreted enzyme protein-tyrosine phosphatase (PTP) by >110-fold in its ant hosts. In baculoviruses, PTP has been shown to cause the migration of caterpillars upward into plant foliage before death: a convergent viral approximation of the fungal summit disease (145, 146). The activity of PTP in fungal manipulations of ant behavior has not yet been experimentally confirmed.

Summit disease is not the only behavioral manipulation produced by insect pathogens. In possibly the most highly evolved interactions, as they indicate a very high level of adaptation of the pathogen to the host, the fungus keeps the host alive and controls its flight behavior so that the insect becomes an aerial vehicle for spore release. There are many examples reviewed by Roy et al. (134), but the entomophthoromycotan fungal genera Strongwellsea and Massospora are the best understood. In the case of Strongwellsea castrans, the hosts are adult flies (3). The fungus causes a large circular hole to develop on the host’s abdomen, and conidia are ejected from the flying insect through this hole. Massospora cicadina also initiates sporulation when its cicada host is still alive (Fig. 5) (147, 148). Eventually, the abdomen is entirely consumed, leaving just the head and thorax of the living insect. The cicada’s ability to fly is retained, increasing dispersion of spores in the environment, especially in the case of infected male cicadas, which attempt to attract and copulate with females and even continue to feed (3, 149).

Teleomorph hypocrealean states have much narrower host ranges than anamorphs, prompting the interesting question of how closely related fungi could differ so fundamentally in ecology depending on sexual state (49). Comparative genomics of entomopathogenic fungi has confirmed they exhibit diverse reproductive modes that often determine the rates and patterns of genome evolution and are linked as cause or effect with pathogenic strategies (15). The three modes of sexual reproduction in ascomycetous fungi, i.e., heterothallic, homothallic, and pseudohomothallic, are governed by the mating-type (MAT) locus (150, 151). The haploid genome of heterothallic species carries only one of the MAT loci, and, thus, they require a haploid partner with a compatible MAT locus to complete the sexual cycle. Homothallic species (self-fertile) have both loci in their haploid genomes, while the pseudohomothallic species are similarly self-fertile, but they contain two compatible haploid nuclei within their sexual spores (152). Most members of the three hypocrealean entomopathogen families (Clavicipitaceae, Cordycipitaceae, and Ophiocordycipitaceae) are heterothallic and are therefore potentially outcrossing fungi. Signs of sex in supposedly asexual species include footprints of Repeat Induced Point (RIP) mutations, a genome defense mechanism specific to fungi, occurring only during the sexual stages on repeated sequences. The consequences of RIP mutations are that repeated DNA segments, such as those that would result from the transposition of a retrotransposon or the duplication of a gene, are inactivated by mutations. Calculations of RIP indices indicated that RIP mutations occur in narrow-host-range species, M. album and M. acridum, but not in the broad-host-range species such as M. robertsii, which suggests retention of sexuality in specialists, although their sexual stages have not been verified (48). Intriguingly, therefore, asexuality in Metarhizium spp. is associated with broad host ranges and sexuality with narrow host ranges, consistent with the narrow host range of known teleomorph states. Unlike C. militaris, which is specific to lepidopteran pupae, the closely related B. bassiana has a wide host range. B. bassiana, like generalist Metarhizium spp., lacks the RIP mutations, consistent with the sexual cycle being rare (54). RIP is incompatible with gene duplication events, so its absence is consistent with expanded gene families and more transposable elements (TEs) in the B. bassiana and M. robertsii genomes, relative to C. militaris and M. acridum. It is possible, therefore, that losing sexuality, and therefore RIP, was a prerequisite for generalists expanding gene families.

In contrast to the other sequenced insect pathogens, which are all heterothallic, O. sinensis contains two compatible MAT loci in the genome and is sexually self-fertile, i.e., homothallic (33). It is likely that inbreeding is an adaptation by O. sinensis to its small population size resulting from a very specialized lifestyle and the extreme environmental conditions in its small geographical range. The O. sinensis genome size is approximately three times larger (∼120 Mb) than other ascomycete insect pathogens, but, whereas they contain more than 9,500 protein-coding genes, O. sinensis has only 6,972 genes (33). The RIP mechanism is dysfunctional in O. sinensis, which has probably contributed to a massive proliferation of retrotransposable elements, and thus genome size inflation, and the large number of retrotransposed and fragmented pseudogenes in the genome implicates retrotransposition in most of the gene losses in O. sinensis. It has been proposed that plant pathogenic fungal lineages with large and flexible genomes are likely to adapt faster during coevolution with hosts (153). The massive proliferation of transposable elements in the inbreeding O. sinensis, may provide a trade-off between advantages of increased genetic variation independent of sexual recombination and deletion of genes dispensable for its specialized pathogenic lifestyle. As O. sinensis has lost many genes for expanding its host range, future transitions away from its current lifestyle seem unlikely, indicating that, while retrotransposition may facilitate rapid adaptation, it may also contribute to stable host interactions.

Nearly all fungal phyla have representatives that have convergently evolved pathogenicity to insects. Identifying commonalities and differences between these fungi is providing a comprehensive picture of the mechanisms they use for pathogenicity, and how new pathogens emerge with different host ranges. Bioinformatics and sequencing technology have already contributed greatly to the study of insect pathogenic fungi, particularly the hard to culture obligates, and promise further breakthroughs in some of the more intractable aspects of this field, e.g., behavior manipulation and evolution of host specificity. Bolstered by this influx of information, entomopathogenic fungi are emerging as models for understanding how pathogens in general respond to and evolve in changing environments, initiate host invasion, colonize tissues, and counter host immune responses. These pathogens offer us a glimpse into the diverse answers fungi have to the question, how best to kill an insect?

This work was supported by the U.S. National Science Foundation Grant IOS-1257685 and by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under award number RO1 AI106998 to Raymond St. Leger.