Symbiont-Mediated Protection of Acromyrmex Leaf-Cutter Ants from the Entomopathogenic Fungus Metarhizium anisopliae.

Many fungus-growing ants engage in a defensive symbiosis with antibiotic-producing ectosymbiotic bacteria in the genus Pseudonocardia, which help protect the ants' fungal mutualist from a specialized mycoparasite, Escovopsis. Here, using germfree ant rearing and experimental pathogen infection treatments, we evaluate if Acromyrmex ants derive higher immunity to the entomopathogenic fungus Metarhizium anisopliae from their Pseudonocardia symbionts. We further examine the ecological dynamics and defensive capacities of Pseudonocardia against M. anisopliae across seven different Acromyrmex species by controlling Pseudonocardia acquisition using ant-nonnative Pseudonocardia switches, in vitro challenges, and in situ mass spectrometry imaging (MSI). We show that Pseudonocardia protects the ants against M. anisopliae across different Acromyrmex species and appears to afford higher protection than metapleural gland (MG) secretions. Although Acromyrmex echinatior ants with nonnative Pseudonocardia symbionts receive protection from M. anisopliae regardless of the strain acquired compared with Pseudonocardia-free conditions, we find significant variation in the degree of protection conferred by different Pseudonocardia strains. Additionally, when ants were reared in Pseudonocardia-free conditions, some species exhibit more susceptibility to M. anisopliae than others, indicating that some ant species depend more on defensive symbionts than others. In vitro challenge experiments indicate that Pseudonocardia reduces Metarhizium conidiospore germination area. Our chemometric analysis using matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) reveals that Pseudonocardia-carrying ants produce more chemical signals than Pseudonocardia-free treatments, indicating that Pseudonocardia produces bioactive metabolites on the Acromyrmex cuticle. Our results indicate that Pseudonocardia can serve as a dual-purpose defensive symbiont, conferring increased immunity for both the obligate fungal mutualist and the ants themselves. IMPORTANCE In some plants and animals, beneficial microbes mediate host immune response against pathogens, including by serving as defensive symbionts that produce antimicrobial compounds. Defensive symbionts are known in several insects, including some leaf-cutter ants where antifungal-producing Actinobacteria help protect the fungal mutualist of the ants from specialized mycoparasites. In many defensive symbioses, the extent and specificity of defensive benefits received by the host are poorly understood. Here, using "aposymbiotic" rearing, symbiont switching experiments, and imaging mass spectrometry, we explore the ecological and chemical dynamics of the model defensive symbiosis between Acromyrmex ants and their defensive symbiotic bacterium Pseudonocardia. We show that the defensive symbiont not only protects the fungal crop of Acromyrmex but also provides protection from fungal pathogens that infect the ant workers themselves. Furthermore, we reveal that the increased immunity to pathogen infection differs among strains of defensive symbionts and that the degree of reliance on a defensive symbiont for protection varies across congeneric ant species. Taken together, our results suggest that Acromyrmex-associated Pseudonocardia have evolved broad antimicrobial defenses that promote strong immunity to diverse fungal pathogens within the ancient fungus-growing ant-microbe symbiosis.

INTRODUCTION

Microbial symbionts provide ecological and physiological services, from supplying nutrients to protection from pathogens. Symbiont-derived protection can be crucial for the health and fitness of some hosts, particularly in insects (1–8). While the primary models of symbiont-derived protection in insects involve intracellular and gut symbiotic bacteria (9–16), less is known about insect-associated ectosymbiotic bacteria and their effects on host defense (17–20). Some insect-associated ectosymbiotic bacteria produce antimicrobial compounds that protect their insect host, its offspring, or other symbiotic partners, such as in termites, southern pine beetles, beewolves, bees, and fungus-growing ants (21–28).

Leaf-cutter ants (Acromyrmex and Atta species) are the most derived species of the attine fungus-growing ants, are dominant herbivores in the Neotropics, and are important agricultural pests (29, 30). The ants cultivate a fungal garden by supplying it with fresh plant material, and in return, the fungus Leucoagaricus gongylophorus produces hyphal swellings on which the ants feed (29, 31, 32). The fungal garden is susceptible to infection by a specialized mycoparasite in the genus Escovopsis (33–35). To help protect their gardens against pathogens, the ants utilize both physical and chemical measures, such as grooming the garden (36) and using antibiotic-secreting metapleural glands (36–40). In the genus Acromyrmex, garden-tending workers carry an antibiotic-producing ectosymbiotic Pseudonocardia bacterium (Actinobacteria) on specialized structures on their exoskeleton that help to protect the fungus garden from Escovopsis (36, 41, 42).

Pseudonocardia symbiont transmission occurs from caregiver nestmates to callow workers and occurs within a narrow window of time after worker eclosion (43, 44). This quick transmission likely helps ensure the specificity and fidelity of native symbionts and reduces the opportunity for colonization by other bacteria. As predicted by this specialized mode of transmission and narrow window of acquisition, studies report evidence for the presence of just a single strain of Pseudonocardia on individual ants and within colonies (43, 45, 46). Since the origin of the attine-Pseudonocardia symbiosis, the ants have acquired free-living Pseudonocardia strains multiple times, and over the evolutionary history of this association, switches between ant species and some Pseudonocardia lineages have occurred (22, 47).

In many species of Acromyrmex, nestmate workers harbor an abundance of Pseudonocardia. The coverage of the ectosymbiont increases exponentially from initial inoculation through 10 to 15 days posteclosion, resulting in the colonization of virtually the entire ant exoskeleton. Pseudonocardia abundance then decreases 25 days posteclosion (44). The heavy abundance of the Pseudonocardia symbiont likely facilitates the application of antimicrobial compounds for the protection of fungal gardens from Escovopsis. Additionally, given that many entomopathogens infect their host by penetrating the exoskeleton, it has been suggested that Pseudonocardia coverage may prevent spore germination of fungal pathogens on workers by acting as a physical barrier on the ant exoskeleton (33, 41). Mattoso and colleagues (82) showed, using a single colony of Acromyrmex subterraneus subterraneus, that removing Pseudonocardia from the exoskeleton of workers increases their susceptibility to the entomopathogenic fungus Metarhizium anisopliae, a common insect pathogen. Although much is known about the ability of Pseudonocardia to inhibit Escovopsis, our understanding of the protective role of Pseudonocardia against Metarhizium across Acromyrmex ants is limited.

Here, we explore the role of Pseudonocardia in the Acromyrmex leaf-cutter ants as a protective partner against the entomopathogenic fungus Metarhizium. Specifically, we evaluate whether (i) Pseudonocardia-derived protection against Metarhizium is found broadly across Acromyrmex ant species, (ii) there is a trade-off between Pseudonocardia and metapleural gland (MG) secretions on ant susceptibility to the pathogen, (iii) the acquisition of nonnative Pseudonocardia affects individual ant survival, (iv) the abundance of nonnative and native Pseudonocardia increases protection, and (v) Pseudonocardia protects the ants by the production of bioactive compounds in vitro and in situ.

RESULTS

Pseudonocardia-mediated protection from Metarhizium.

Ant survival was significantly greater for control-treated ants than for those exposed to Metarhizium (Wald = 32.45, df = 1, P < 0.0001). Overall, ants carrying their own Pseudonocardia (native) survived 2.24 times more than Pseudonocardia-free ants (Wald = 59.67, df = 1, P < 0.0001) (Fig. 1). However, Acromyrmex laticeps and Acromyrmex niger ants did not differ in their susceptibility to infection when they carried their own Pseudonocardia or under Pseudonocardia-free conditions (Fig. 1C and D). In contrast, A. echinatior and Acromyrmex octospinosus ants reared under Pseudonocardia-free conditions (Fig. 1A and B) were 2.28 times more susceptible to infection than when they carried their own Pseudonocardia. When comparing a within-group of ants carrying their own Pseudonocardia, we found that all ants carrying their own Pseudonocardia differed in survival when challenged with Metarhizium (Wald = 13.79, df = 3, P = 0.003). Acromyrmex echinatior, A. laticeps, and A. niger were 1.8, 2.3, and 2.0 times more susceptible to infections by Metarhizium than A. octospinosus (Fig. 1).

FIG 1

Pseudonocardia as a defensive symbiont in four species of Acromyrmex leaf-cutter ants. Survivorship curves of Acromyrmex workers with A. echinatior (A), A. octospinosus (B), A. laticeps (C), and A. niger (D) carrying Pseudonocardia from their own colony (native) (■ and □) or under a Pseudonocardia-free condition (● and ○), exposed to Metarhizium (solid symbols) or a control solution (open symbols) of sterile deionized water + 0.01% Tween 20. Error bars represent standard error. Letters represent significant differences from one treatment to another at a P value of <0.05 in pairwise comparisons using a Kaplan-Meier pairwise test.

Metapleural glands and Pseudonocardia effects on ant defense.

Ant survival was significantly greater for control-treated ants than that for ants exposed to Metarhizium (Wald = 95.54.71, df = 1, P < 0. 001). Interestingly, there were significant main effects within MG treatments (Wald = 9.27 df = 1, P = 0.002) and symbiotic condition (Wald = 7.37 df = 1, P = 0.007), but there were no significant differences for any interaction effect (P > 0.05). A. echinatior ants with MGs open have a significantly lower risk (58.6%) of death than ants with MGs closed (Fig. 2A). In contrast, A. echinatior ants with a Pseudonocardia-free condition were 2.25 times more likely to die than those ants carrying Pseudonocardia. Metarhizium-treated ants had a 72.5% lower risk of death when associated with Pseudonocardia (Wald = 25.41, df = 1, P < 0. 0001) and 35% lower risk when the MGs were open (Wald = 4.37, df = 1, P = 0.036) (Fig. 2A). When ants were exposed to the control solution, there were significant effects within MG treatments (Wald = 5.81, df = 1, P = 0.016), where ants with MGs open had a lower risk (72.9%) of death than those with MGs closed (Fig. 2B).

FIG 2

Metapleural glands and Pseudonocardia effects on ant defense. Metapleural glands and Pseudonocardia effects on A. echinatior workers infected with Metarhizium (A) and a control solution (B). Ants were reared under either Pseudonocardia-carrying conditions (+P) or Pseudonocardia-free conditions (−P) with either MGs sealed (−MG) or MGs open (+MG). A control solution was made of sterile deionized water + 0.01% Tween 20. Error bars represent standard error. Letters represent significant differences at a P value of <0.05 in pairwise comparisons using a Kaplan-Meier pairwise test. The inset graph shows the effects of the acrylic solution, which was used to block the MGs, on gaster-painted ants (red bar) and unpainted ants (pink bar).

Nonnative Pseudonocardia acquisition and bacterial coverage effects on ant individual susceptibility.

In the nonnative experiment, all ants exposed to the control solution survived, while worker ant mortality was significantly higher in the Metarhizium infection treatments, both for the control native pairing (Wald = 60.36, df = 1, P < 0.0001) and for A. echinatior ants carrying Pseudonocardia from different ant species (conditions reared) (Wald = 62.9, df = 6, P < 0.0001) (Fig. 3A). There was no correlation between the visible abundance of Pseudonocardia and ant survivorship (Wald = 0.02, df = 1, P = 0.885) nor between the phylogenetic clade of origin (see Fig. S1 in the supplemental material) for Pseudonocardia and ant survival (Wald = 1.77, df = 1, P = 0.83). Although pairwise comparisons of the survival distribution were variable within treatments (Fig. 3A), the hazard ratio for Metarhizium-treated ants was significantly higher for A. echinatior carrying either nonnative Pseudonocardia or carrying native (own colony) Pseudonocardia than that for Pseudonocardia-free A. echinatior ants (reference group) (see Table S1 in the supplemental material).

FIG 3

Nonnative Pseudonocardia acquisition and bacterial coverage effects on ant individual susceptibility. (A) Survivorship curves of A. echinatior workers carrying the Pseudonocardia ectosymbiont from different Acromyrmex species after being exposed to Metarhizium. Species names on the legend denote the ant host species from which each Pseudonocardia strain was derived. The A. echinatior Pseudonocardia is the native strain. A. echinatior ants raised by A. cephalotes ants are Pseudonocardia free. Letters represent significant differences from one treatment to another at a P value of <0.05 in pairwise comparisons using the Kaplan-Meier pairwise test. (B) Effects of different strains of Pseudonocardia isolated from different species of Acromyrmex ants on the conidial germination area of Metarhizium. Streptomyces coelicolor (a common soil-dwelling Actinobacteria) was used for comparative effects. (C) Micrograph illustrating the interactions between Metarhizium (left) and Pseudonocardia (right). Error bars represent standard error.

Maximum likelihood phylogenetic tree using elongation factor-Tu sequences of Pseudonocardia isolates. Pseudonocardia isolates from the experimental colonies are indicated by solid circles and other phylogenetically closely related attine-associated Pseudonocardia. The name of the Pseudonocardia isolates represents the attine species in which the bacteria was obtained, and GenBank accession number are shown in parenthesis. Clade (IV and VI) indicates the position of the Pseudonocardia isolates relative to previous studies (7, 8). The number before branch points supports bootstraps from 1,000 resampled datasets with values less than 50% not shown. Streptomyces griseus and Streptomyces sampsonii were used as outgroups. The scale bar indicates 0.02 change per nucleotide. Download FIG S1, DOCX file, 0.4 MB.

Survivorship of Acromyrmex echinatior workers carrying different Pseudonocardia ectosymbionts after being treated with Metarhizium anisopliae Table S1, DOCX file, 0.02 MB.

To compare the effects of acquiring nonnative Pseudonocardia on A. echinatior, we excluded Pseudonocardia-free ants from the Cox regression model and included only A. echinatior ants carrying their native Pseudonocardia (native) as a reference group. There were significant differences in ant survival within treatments (Wald = 21.32, df = 5, P = 0.001). The hazard ratio did not differ between A. echinatior ants carrying their own Pseudonocardia and A. echinatior ants carrying nonnative Pseudonocardia from the colonies of Acromyrmex hispidus fallax (P = 0.256), A. laticeps (P = 0.655), A. niger (P = 0.275), or A. octospinosus (P = 0.838). However, A. echinatior ants carrying Pseudonocardia from an Acromyrmex volcanus colony showed a significantly higher risk (2.4 times, P = 0.002) of death than A. echinatior ants carrying their own Pseudonocardia.

Coverage of Pseudonocardia acquired from native and nonnative colonies in Acromyrmex echinatior.

Overall, there were significant effects of the Pseudonocardia coverage within ant species (F5, 488 = 161.93, P < 0.0001), conditions reared (cross-fostered and native) (F5, 488 = 122.17, P < 0.0001), and the interaction between condition reared and ant species (F5, 448 = 25.30, P < 0.0001). A post hoc comparison showed that Pseudonocardia coverage was higher for Acromyrmex ants raised by their native colonies than that for A. echinatior ants raised by allospecific Acromyrmex ants (P < 0.0001, for all). However, Pseudonocardia coverage between cross-fostered and native ants did not differ with A. octospinosus (P = 0.55) (see Fig. S2 in the supplemental material).

Graph comparing the difference in Pseudonocardia coverage (mean ± SE) on Acromyrmex echinatior workers carrying the Pseudonocardia ectosymbiont from different leaf-cutter ant species (cross-fostered) and when these Acromyrmex species are reared by their conspecific nestmates (native colony). The bar within the dashed lines represents A. echinatior ants raised only by their own colony, and A. cephalotes, which naturally does not carry Pseudonocardia, was used as negative control of Pseudonocardia acquisition. Download FIG S2, DOCX file, 0.07 MB.

Pseudonocardia-Metarhizium in vitro challenges.

There was a significant effect of Actinobacteria strain on the reduction of Metarhizium mycelial growth area (F5, 29 = 4.31, P = 0.0046) and on the reduction of the conidium germination area (F5, 29 = 19.21, P < 0.0001). Post hoc comparisons revealed that Streptomyces coelicolor significantly reduced mycelial growth (P < 0.05) compared with the rest of the Actinobacteria treatments, except for the strain of Pseudonocardia isolated from A. niger (P = 0.107). Pseudonocardia from A. laticeps, A. octospinosus, and S. coelicolor showed a greater reduction of conidium germination area than the rest of the Actinobacteria (Pseudonocardia strains from A. echinatior, A. niger, and A. hispidus fallax; P < 0.001, for all cases) (Fig. 3B). However, S. coelicolor and the Pseudonocardia strains isolated from A. laticeps did not differ significantly (P = 0.25), nor did Pseudonocardia strains from A. niger and A. echinatior (P = 0.08). The conidium germination coverage did not show significant differences between treatments with Pseudonocardia from A. hispidus fallax and treatments with strains from both A. laticeps (P = 0.07) and A. echinatior (P = 0.82) (Fig. 3B).

MALDI imaging.

MALDI imaging detected 41,724 peaks with an average of 1,490.1 peaks per sample. A heat map revealed that the distribution of the putative metabolites associated with ants carrying Pseudonocardia with MGs open and infected with Metarhizium (Fig. 4). Partial least-squares discriminant analysis (PLS-DA) data clearly showed discrimination between groups, particularly between Pseudonocardia-carrying treatments and Pseudonocardia-free treatments (see Fig. S3 in the supplemental material).

FIG 4

In situ imaging mass spectrometry of Acromyrmex ants. Heat map shows the 50 top putative metabolites (row) of importance. Metabolites were selected based on an ANOVA test between treatments and clustered by similarities. Ants were reared under either Pseudonocardia-carrying conditions (+P) with MG opened (pink) and MG opened control (yellow) or Pseudonocardia-free conditions (−P) with MGs opened (light blue), MGs sealed (blue), beheaded (red), and MG opened control (green). Metarhizium-treated ants are denoted by the inset green square. Control denotes uninfected treatments. The color key indicates metabolite relative intensity (blue, lowest; red, highest).

PLS-DA of the profile of metabolites associated with the ants exoskeleton treated with Metarhizium (inset green square) and control treatments. Dots on the graph represent a single ant per treatment clustered by a 95% confidence interval. Ants were reared under either Pseudonocardia-carrying conditions (+P) with MG opened (pink) and MG opened control (yellow) or Pseudonocardia-free conditions (−P) with MGs opened (light blue), MGs sealed (blue), beheaded (red), and MG opened control (green). Control denotes uninfected treatments. Download FIG S3, DOCX file, 0.08 MB.

DISCUSSION

Our study explores the epidemiological and chemical dynamics involved in Pseudonocardia-derived protection against a generalist entomopathogen (Metarhizium anisopliae) in Acromyrmex leaf-cutter ants. Overall, our results indicated that the ecosymbiont Pseudonocardia helps confer increased protection to ant workers from entomopathogenic diseases. Specifically, we experimentally show a substantial reduction in Metarhizium-induced mortality in Acromyrmex workers with Pseudonocardia. Furthermore, our findings that (i) Pseudonocardia-carrying ants produced more compounds in response to Metarhizium infection than Pseudonocardia-free ants, (ii) in vitro Pseudonocardia reduced Metarhizium conidium germination area, and (iii) nonnative Pseudonocardia still conferred protection to A. echinatior ants against Metarhizium, despite reduced physical coverage of workers in some switches, support the hypothesis that Pseudonocardia-derived antifungal compounds are involved in helping protect ants from an entomopathogenic fungus. Our results reflect an apparent complex evolutionary dynamic between the specialized chronic parasite Escovopsis and a cosmopolitan insect-pathogenic fungus and a selective pressure for acquiring Pseudonocardia partners to protect their hosts from pathogen outbreaks.

It is notable that when A. echinatior and A. octospinosus ants were reared under Pseudonocardia-free conditions, they were more susceptible to Metarhizium, while A. laticeps and A. niger ants had a similar susceptibility to Metarhizium under either condition (i.e., with or without Pseudonocardia). This result indicates that an ant’s individual susceptibility to Metarhizium and the level of Pseudonocardia-derived protection can differ between Acromyrmex species. However, interpreting this variation requires an understanding of the interaction between host and symbiont genotypes and natural variation in resistance against Metarhizium (48–50). Indeed, natural variation in resistance to Metarhizium has been observed among colonies of A. echinatior (51) and other social insects, such as termites (52). Another source of natural variation against pathogens can be derived from different symbiont genotypes. For example, distinct strains of the endosymbiotic bacterium Wolbachia can provide different levels of protection against several pathogenic RNA viruses in Drosophila flies (53). Variable protection among distinct symbiont genotypes has also been reported in pea aphids by the defensive symbiont Hamiltonella defensa against parasitoid wasps.

Experimental manipulation of the presence/absence of MGs and Pseudonocardia symbiont on ants show that Metarhizium-treated A. echinatior ants had a larger reduction in pathogen-induced mortality (72.5%) when associated with Pseudonocardia than when the MGs were open (35%). Furthermore, A. echinatior ants that did not carry Pseudonocardia were at an increased risk of death regardless of whether their MGs were open or sealed. These results provide further support that Pseudonocardia helps protect workers from entomopathogens, such as Metarhizium, and that the antibiotics secreted by Pseudonocardia appear to play a larger role in attenuating diseases than MG-secreted antibiotics in some Acromyrmex ants. Ants can selectively apply MG secretions to themselves (self-grooming) and other nestmates (allogrooming) as a social prophylaxis strategy to reduce the transmission of diseases within the colony (36, 54–58). Several studies have demonstrated that MG secretions display antimicrobial activity in attine ants (55, 59–61) and other ant species (62, 63). However, it has been shown in leaf-cutter ants (i.e., Atta and Acromyrmex) that the role of MG use differs quantitatively and qualitatively in the spectrum of compounds secreted between species, colonies, and worker caste (39, 63, 64). It was argued recently that these differences may be a consequence of the trade-off between MG secretions and Pseudonocardia in the production of antimicrobial metabolites (39, 57, 60, 61). While Acromyrmex ants depend on producing specialized Pseudonocardia-derived antibiotics for suppressing the fungal garden pathogen Escovopsis, it seems that Atta ants compensate for the lack of Pseudonocardia by producing broad-spectrum antimicrobials in their MGs. According to this compensatory hypothesis and our data, it is possible that there is a high selection pressure in Acromyrmex ants to select Pseudonocardia strains capable of producing complementary compounds, in addition to those that inhibit Escovopsis, that suppress generalized pathogens in compensation for reduced protective effects of MG secretions. Further research should be undertaken to investigate the chemical and physiological relationship between the bioactive compounds derived from MG secretions and Pseudonocardia across attine ants, particularly in those that lack the association with Pseudonocardia, such as Sericomyrmex.

One interesting finding is that the antifungal protective effect of Pseudonocardia was extended to A. echinatior ants even when they acquired nonnative Pseudonocardia from geographically distant species. This finding suggests that the acquisition of nonnative Pseudonocardia may not affect individual performance (i.e., survival) of the ants and further shows evidence that horizontal transmission may be plausible as previous phylogenetic studies argue (22, 47). These findings are consistent with Armitage et al. (83) and Andersen et al. (65) who reported that sympatric Panamanian Acromyrmex ants (A. echinatior, A. octospinosus, and A. volcanus) can exchange their Pseudonocardia strains in the laboratory.

It has been reported that colonies of sympatric Panamanian Acromyrmex ants are colonized predominantly by only one of two possible Pseudonocardia lineages (clade IV and VI), indicating prevalence and stability within ant-Pseudonocardia populations (22, 45, 46, 65). However, these previous studies were focused on a small area around the Panama Canal Zone, Gamboa, Panama. In contrast, one interesting finding in our study is that A. echinatior ants, which carry Pseudonocardia strain VI, were similarly resistant to Metarhizium when they acquired nonnative Pseudonocardia strains IV and VI from both South American and Central American Acromyrmex species. It is possible that both clades are equally protective against pathogens. Consequently, a possible explanation for this persistence might be that these strains (clade IV and VI) are coadapted to Acromyrmex ants, in accordance with the data of Cafaro et al. (2011) (22), and are under high selection pressure to combat pathogens (24, 45, 46, 66). A future approach for understanding the evolutionary dynamic between ant-associated Pseudonocardia and ant-infecting pathogen diversity is the incorporation of whole-genome analysis across regional and local levels.

When A. echinatior pupae were cross-fostered with allospecific caregiver workers, the ants acquired a lower abundance of Pseudonocardia in contrast to A. echinatior ants reared by their own colony. These findings indicate a certain degree of ant-Pseudonocardia affinity, which was also found by Armitage et al. (2011) and Andersen et al. (2015). We found no correlation between the relative visible coverage of Pseudonocardia and ant mortality. These results suggest that the bacterial coverage does not explain the susceptibility to Metarhizium. However, it has been demonstrated that the presence of Escovopsis within fungus gardens can result in corresponding increases in Pseudonocardia coverage on workers (41, 61). It is difficult to interpret whether the filaments of the bacterium are physically protecting the ant body because we observed similar beneficial effects in all nonnative Pseudonocardia transmitted to A. echinatior ants regardless of the bacterial coverage. However, our in vitro and in situ results suggest that chemical protection derived from Pseudonocardia plays a significant role in protection.

The chemical protection of Pseudonocardia can differ when the bacterium is associated with the ant cuticle or when it is grown under in vitro conditions. For example, we noticed that nonnative Pseudonocardia strains from A. hispidus fallax and A. niger conferred protection to A. echinatior ants. However, these Pseudonocardia strains differed in their suppressing effect on the spore germination of Metarhizium under in vitro conditions. These differences need to be treated with caution because the cuticle of the ant represents a limited nutritional environment where cuticular crypts supply nutrients to Pseudonocardia (67–69). In contrast, a synthetic medium provides a nutritionally rich environment. How these different conditions affect the expression of antimicrobial compounds against Metarhizium is unknown, but it is more likely that the production of antibiotics occurs under stress conditions, such as nutrient depletion (70–72). Furthermore, it is likely that Pseudonocardia responds to the antimicrobial demands of its ant host (73).

Within Actinobacteria, the family Pseudonocardiaceae is recognized for producing bioactive antimicrobial compounds. Consequently, it might be expected that Pseudonocardia produces bioactive secondary metabolites to protect the ants against other pathogens besides Escovopsis. There have indeed been many studies showing that the metabolic potential of Pseudonocardia ranges from a broad spectrum to a narrow spectrum of antimicrobial activity. For example, Barke et al. (2010) isolated a nystatin-like polyene from a colony of A. octospinosus with weak activity against Candida albicans and Escovopsis (74). In contrast, Meirelles et al. (2014) showed that several Pseudonocardia strains isolated from Trachymyrmex ants may have a generalized inhibitory activity against Escovopsis (38). Other previous studies have documented similar activities (41, 42, 75–77). Taken together, these studies suggest that there is high variability in antimicrobial activity by Pseudonocardia (42).

We found a high number of putative metabolites that might have a defensive role against the insect pathogen Metarhizium. However, characterizing the identity of these metabolite signals, expressed across different treatments, is difficult given the apparent complexity in the chemical response. Nevertheless, the results of the Pseudonocardia-Metarhizium challenges and MALDI-MSI support the conclusion that the protective benefits we observed are likely conferred by chemical protection. Additionally, the PLS-DA and heat map reveal that ants associated with Pseudonocardia secrete a distinct and more complex profile of compounds than ants lacking Pseudonocardia and not challenged with Metarhizium. This result indicates that the production of different compounds is down- and upregulated during Metarhizium infections. Because Metarhizium conidial adhesion to the insect cuticle is necessary for infection (78), it is possible that the high diversity of biomolecules is acting synergistically and contingently to reduce conidial germination rate. Insect-associated microbial symbionts depend on secondary metabolite products to mediate the host-symbiont association and inhibit both host-targeting pathogens and potential symbiont competitors. Therefore, it is not surprising that Pseudonocardia responded chemically to Metarhizium. In general, it seems that attine ants prevent and ameliorate disease using a complex arsenal of bioactive compounds whose functions need to be further studied.

This research extends our knowledge of Pseudonocardia-mediated protection in Acromyrmex ants. The results of this research support the idea that Pseudonocardia promotes ant resistance to pathogens by inducing an antimicrobial coating effect on the ant exoskeleton. It would be interesting to assess the collective effect (social immunity) of Pseudonocardia on the colony level. Mature Acromyrmex ant colonies maintain thousands of workers that harbor mutualistic Pseudonocardia; therefore, further experimental investigations are needed to estimate the amount of secondary metabolites secreted in the colony. As an analogy to the health care industry in which antimicrobial agents are applied to the surface of a material to reduce the growths of microorganisms, we suggest that Pseudonocardia acts as an antimicrobial surface agent promoting colony health by providing biotherapeutic medication on the individual and colony level.

MATERIALS AND METHODS

Study species.

Seven species of leaf-cutter ants were used in our experiments. Four species that coexist sympatrically in Central America were used, as follows: Acromyrmex echinatior (AL050505-11; Panama), Acromyrmex octospinosus (ST04116-01; Panama), Acromyrmex volcanus (CR2014; Costa Rica), and Atta cephalotes (AL050513-22; Costa Rica). The following three South American species were used: Acromyrmex laticeps (UGM030330-05; Argentina), Acromyrmex hispidus fallax (UGM030327-02; Argentina), and Acromyrmex niger (CC030327-02; Argentina). The Acromyrmex species used in this study carry abundant visible Pseudonocardia on their exoskeleton. Colonies were kept in large plastic containers in the Microbial Science Building at the University of Wisconsin-Madison.

Cross-fostering technique.

We used a cross-fostering technique to manipulate Pseudonocardia acquisition in which medium-sized pupae from a leaf-cutter ant colony were removed and raised by caregiver ants from either conspecific (own colony) or allospecific (Atta or Acromyrmex) workers. Using this approach, we generated aposymbiotic (i.e., Pseudonocardia free) workers and switched ant-Pseudonocardia combinations (i.e., switch Pseudonocardia symbionts between ant hosts). Members of the genus Atta do not have Pseudonocardia on their exoskeleton, and so Atta caregiver workers do not transmit the ectosymbiont (43). These Pseudonocardia-free ants were used as controls (see below).

Individual-level susceptibility across Acromyrmex.

To evaluate whether Pseudonocardia helps protect different species of Acromyrmex ants from infection by Metarhizium, we examined the susceptibility of four Acromyrmex species under Pseudonocardia-free conditions. We generated Pseudonocardia-free ants by rearing pupae of Acromyrmex ants with workers of A. cephalotes. For the experiment, we set up two subcolonies (n = 8) per species consisting of a weigh boat placed in a petri plate with a ring of moist cotton to maintain the humidity, 3 g of A. cephalotes fungal garden, and 50 focal medium-size pupae of either A. echinatior, A. niger, A. laticeps, or A. octospinosus. One subcolony of each species was placed with either 70 workers (15 major, 25 medium, and 30 minima workers) of its own species (own colony) or A. cephalotes. We monitored each subcolony every 3 days to remove and replace dead caregiver workers and fungus garden rejected by the ants; rejected pupae were not replaced. Unless otherwise specified, we used the fungus garden from A. cephalotes in all our experiments to control for any nutritional differences among gardens that may have affected the ant performance (79). After 15 to 18 days posteclosion, all ants were each placed into an individual Petri plate with a ring of moist cotton. Then, the propleural plate of each ant was inoculated with Metarhizium. M. anisopliae was applied using 1 μl of ca. 1.00 × 107 conidia ml−1 suspension + 0.01% Tween 20 by using a micropipette. These treatment setups were repeated and inoculated with a control solution of sterile, deionized water + 0.01% Tween 20. The survival of the ants was monitored every 24 h posttreatment for 10 days.

M. anisopliae is a broad-spectrum insect pathogen, known to infect many insect species (80). In our research, we were interested in the generalized capacity of Metarhizium to infect many different arthropods. The M. anisopliae strain used in our experiments was isolated from dead bee workers of Apis mellifera and showed the ability to infect ants. For the experiments, conidia were taken from recently sporulating cultures on potato dextrose agar and suspended in a solution of sterile deionized water containing 0.01% Tween 20. The conidial concentration was quantified using a hemocytometer and diluted to a concentration of ca. 1.00 × 107 conidia ml−1.

Metapleural gland and Pseudonocardia effects on ant defense.

To test the effects of the compounds produced by the MGs and Pseudonocardia on ant susceptibility to Metarhizium, we reared A. echinatior ants with or without Pseudonocardia and MGs open or MGs sealed. To produce Pseudonocardia-free ants, we cross-fostered A. echinatior pupae with A. cephalotes workers as described above. We set up two subcolonies as described above containing 5 g of fungal garden from A. cephalotes, 60 focal pupae from A. echinatior, and 70 workers of either A. echinatior or At. cephalotes. All ants were monitored, and worker and fungus garden were replaced as mentioned above. After 15 to 18 days posteclosion, we blocked the MG of the ants reared under both the Pseudonocardia-free condition (n = 30) and Pseudonocardia-carrying condition (n = 30) by applying a harmless acrylic solution with a paintbrush (Fig. 2, inset graph). All ants (Pseudonocardia-free ants with MG open and sealed and Pseudonocardia-carrying ants with MG open and sealed) were each placed into an individual Petri plate with a ring of moist cotton. Ants were then treated with Metarhizium or control solution and monitored as described above.

Nonnative Pseudonocardia acquisition and coverage effects on ant susceptibility.

To examine whether the acquisition of nonnative Pseudonocardia affects the individual susceptibility of the ants to Metarhizium, we manipulated Pseudonocardia acquisition by rearing pupae from a colony of A. echinatior with ants from seven leaf-cutter ant species (A. echinatior, A. octospinosus, A. hispidus fallax, A. niger, A. laticeps, A. volcanus, and A. cephalotes) that significantly differ in the visible abundance of Pseudonocardia on their exoskeleton, ranging from not visible to highly abundant (22, 44, 57).

For this experiment, we set up one subcolony (n = 7) per species as described above containing 6 g of fungal garden; 50 focal medium-size pupae from A. echinatior; and 70 workers of either A. echinatior (own colony), A. octospinosus, A. hispidus fallax, A. niger, A. laticeps, A. volcanus, or A. cephalotes. All ants were monitored, placed in individual subcolonies, and treated with Metarhizium or the control solution as mentioned previously. Before infecting the ants, we scored Pseudonocardia coverage via stereomicroscope using the scale designed by Poulsen et al. (44), which scores from 0 (no visible Pseudonocardia) to 12 (ant cuticle is totally covered with Pseudonocardia).

To experimentally examine the antifungal properties of Pseudonocardia against Metarhizium, we performed pairwise bioassays in which pure cultures of Pseudonocardia were challenged with Metarhizium isolates. For each microbial bioassay, we inoculated Pseudonocardia in the center of a Petri plate (100 mm by 15 mm) containing yeast malt extract agar (YMEA) and allowed it to grow for 10 weeks. The Metarhizium isolate was then inoculated 1 cm from the Pseudonocardia area using a suspension solution of 1 μl of ca. 1.5 × 106 conidia ml−1. At 8 days postfungal inoculation, each Petri plate was photographed and the mycelial growth area and conidia germination coverage relative to the mycelial growth area were measured using the software ImageJ (http://rsbweb.nih.gov/ij/). We tested the inhibitory proprieties of Pseudonocardia using pure isolates from the experimental colonies A. echinatior, A. octospinosus, A. hispidus fallax, A. niger, and A. laticeps. We also measured the inhibitory properties of Streptomyces coelicolor, a common soil-dwelling Actinobacteria (n = 5 to 8 per pairing). As a positive control, we challenged Metarhizium against antifungal disks soaked in nystatin with a concentration of 10,000 units ml−1. Pseudonocardia was isolated from worker cuticles using a method from Cafaro and Currie (2005) (47). Despite extensive efforts, we were unable to isolate Pseudonocardia from the colony of A. volcanus using this method.

MALDI-Orbitrap imaging.

We then assessed whether Pseudonocardia produces in vivo compounds that inhibit Metarhizium. To do so, 28 medium-size pupae were collected randomly from the top of an A. echinatior fungal garden. The ants were split into two groups, as follows: A. echinatior ants (n = 19) reared under a Pseudonocardia-free condition and A. echinatior ants (n = 9) reared with their own Pseudonocardia using the cross-fostering approach in the experiment above. Fifteen days posteclosion, each ant was placed in a single Petri plate with a small piece moist cotton. To reduce other sources of secondary metabolites, we tested Pseudonocardia-free ants under the following treatments: removal of the ant’s head (n = 5), sealing of the MGs (n = 5), and keeping MGs open (n = 5). The ant’s propleural plate was then inoculated with Metarhizium using the following treatments: Pseudonocardia-free ants (beheaded, MG sealed, and MG open) and Pseudonocardia-carrying ants with MG open (n = 5). Two groups, namely, Pseudonocardia-free ants with MG open (n = 4) and Pseudonocardia-carrying ants with MG open (n = 4), were inoculated with sterile-deionized water as a control. The other ants were inoculated with a 1 μl of ca. 1.00 × 107 conidia ml−1 suspension of M. anisopliae + 0.01% Tween 20. Twenty-four hours postinfection, ants were collected and stored at −20°C. All ants were transported to the School of Pharmacy at the University of Wisconsin-Madison for mass spectrometry imaging (MSI) analysis (84).

DNA extraction, sequencing, and phylogenetic analysis.

DNA from was extracted, and a partial length sequence of the nuclear elongation factor gene (EF-Tu) was amplified using primers 52F and 920R (46). All sequences were aligned and a maximum likelihood-based phylogeny was generated. Twenty-six sequences in the GenBank database from previous studies were added to the analysis for comparison (see electronic supplemental material).

Statistical analysis.

We analyzed ant survivorship using a Cox regression model. The Cox regression model produces a survival function (i.e., hazard) that predicts the probability of death associated with a variable(s) at a specific time. The hazard ratio was estimated for the experiments using the categorical variables as follows: fungal treatment (Metarhizium versus control), symbiotic condition (Pseudonocardia-carrying ants and Pseudonocardia-free ants), condition reared (Pseudonocardia-donor ant species, namely, A. echinatior, A. octospinosus, A. volcanus, A. laticeps, A. hispidus fallax, A. niger, and A. cephalotes), Pseudonocardia clade (clade V versus clade IV), and visible abundance (i.e., coverage) of Pseudonocardia was used as a covariable. The relative risk reduction parameter was estimated to quantify the relative decrease in the risk of death (see Table S1). A full model was performed to estimate the main effects and their interaction effects. Separately, Metarhizium-treated ants were analyzed to estimate and compare the effects of resistance across the treatments. To compare within treatments, we used the Kaplan-Meier analysis. A two-way analysis of variance (ANOVA) was used to analyze the abundance of Pseudonocardia between ant species and condition reared (cross-fostered and native) and the interaction effect. For Pseudonocardia-Metarhizium challenges in vitro, a one-way ANOVA was used to analyze mycelium growth area and conidium germination area. A post hoc comparison between treatments was made using the Tukey’s honestly significant difference (HSD) test. All statistical analyses were performed in SPSS v. 22.

Mass spectrum results of the samples were statistically analyzed using MetaboAnalyst 3.0 (http://www.metaboanalyst.ca) (81). The low-intensity peaks (noise) were replaced by half of the minimum positive values of the data set, and the standard deviation option was applied to filter data as recommended by Hackstadt and Hess (85). The data were normalized using a reference control sample (control, uninfected Pseudonocardia-carrying ants) and transformed using a generalized logarithm transformation. The relative change of the masses between treatments was applied with the auto-scaling option. A heat map and principal-component analysis (PCA) were constructed to explore the dimension of the data set, followed by a partial least-squares discriminant analysis (PLS-DA) to reduce dimensionality.

Data availability.

Raw sequencing data have been deposited in the GenBank database (accession numbers OL630658 to OL630663). The data sets generated during the current study are available from the corresponding authors on request.

Contains expanded materials and methods of procedures, including DNA extraction, phylogenetic analysis methods, and MALDI-Orbitrap imaging. In addition, it contains expanded references for methods. Download Text S1, DOC file, 0.05 MB.