CAMERON R. CURRIE*†, JAMES A. SCOTT†, RICHARD C. SUMMERBELL‡§ & DAVID MALLOCH†
Nature 398, 701 - 704 (1999) © Macmillan Publishers Ltd.
The well-studied, ancient and highly evolved mutualism between fungus-growing ants and their fungi has become a model system in the study of symbiosis1-5. Although it is thought at present to involve only two symbionts, associated with each other in near isolation from other organisms1-5, the fungal gardens of attine ants are in fact host to a specialized and virulent parasitic fungus of the genus Escovopsis (Ascomycotina)6. Because the ants and their fungi are mutually dependent, the maintenance of stable fungal monocultures in the presence of weeds or parasites is critical to the survival of both organisms. Here we describe a new, third mutualist in this symbiosis, a filamentous bacterium (actinomycete) of the genus Streptomyces that produces antibiotics specifically targeted to suppress the growth of the specialized garden-parasite Escovopsis. This third mutualist is associated with all species of fungus-growing ants studied, is carried upon regions of the ants' cuticle that are genus specific, is transmitted vertically (from parent to offspring colonies), and has the capacity to promote the growth of the fungal mutualist, indicating that the association of Streptomyces with attine ants is both highly evolved and of ancient origin.
Because few organisms cultivate their own food, fungus-gardening by ants (Attini: Formicidae) is considered to be a major breakthrough in animal evolution7. These ants forage on a variety of substrates that they use for the cultivation of the vegetative mycelium of a fungus, their dominant food source. Fungus cultivation evolved apparently only once in the attines, over 50 million years ago, with the domestication of a fungus in the family Lepiotaceae (Agaricales: Basidomycotina)3,4,8. Other lepiotaceous lineages, and in one case a distantly related non-lepiotaceous basidiomycete, were domesticated in subsequent evolutionary history4. The success of fungal cultivation by the attine ants is illustrated by the leaf-cutting genera, Acromyrmex and Atta, which are the dominant herbivores in the neotropics9.
Certain areas of the cuticle of fungus-growing ants are coated with what appears to the naked eye to be a powdery, whitish-grey crust (Fig. 1). This has been dismissed previously as a 'waxy bloom', implying that its aetiology was cuticular exudate10. Micromorphological and biochemical examination reveals that this coating is in fact formed from masses of a filamentous bacterium (actinomycete) of the genus Streptomyces (Fig. 2a; see Methods). Actinomycetes are mostly soil-dwelling organisms of great abundance and ecological importance that produce an array of secondary metabolites, many of which have specific antibacterial or antifungal properties11,12. In fact, most antibiotics developed for human pharmaceutical use are actinomycete metabolites, many derived from the genus Streptomyces11,12. In light of the unique biochemical properties of actinomycetes as a group, we proposed that the Streptomyces associated with fungus-growing ants may have an important function in this symbiosis, that of suppressing the growth of potentially devastating pathogens.
|
Figure 1 Photograph showing the presence of the third
mutualist, Streptomyces, on the cuticle of Acromyrmex octospinosus.
|
 |
Figure 2 Scanning electron micrographs of fungus-growing ants, showing the location of Streptomyces. |
To ascertain the distribution and prevalence of this bacterium, both within and between species and genera of fungus-growing ants, we studied 22 species of attine ants representing 8 genera for the presence of the actinomycete. The bacterium was associated with all species studied, from the phylogenetically basal genera Myrmicocrypta and Apterostigma to the highly derived, leaf-cutting genera Acromyrmex and Atta. All 112 colonies from Panama sampled for the presence of the actinomycete in 1997 and 1998 showed this bacterial association. In all cases, the actinomycete was concentrated on genus-specific areas of the ant integument that appear to be modified for the maintenance and growth of the Streptomyces, conceivably to facilitate the distribution of bacterial metabolites throughout the garden (Fig. 2, Table 1). In workers and queens of Myrmicocrypta and Apterostigma, the actinomycete mutualist occurs under the forelegs (Fig. 2b); in the more phylogenetically derived genera (Table 1), the actinomycete mutualist is most prominent on the laterocervical plates of the propleura (Fig. 2c), ventral, collar-like structures immediately posterior to the mouth-parts. The consistent association of the actinomycete with phylogenetically diverse attine ants, as well as its location on the ant integument (that is, conserved within species but unique between genera), indicates that this association is highly evolved and may be of ancient origin.
We studied the presence of the actinomycete on foundress
queens (gynes) during their mating flights to determine whether, like the fungal
mutualist, this bacterium is transmitted vertically between parent and offspring colonies.
We examined 74 foundress queens and 15 males collected from throughout Gamboa, Panama,
during the mating flight of Acromyrmex octospinosus on 9 May 1997. Streptomyces
was present on the cuticle of all gynes examined, whereas it was absent from all males
collected during the same mating flight. Examination of female and male reproductive
individuals in natal gardens of Trachymyrmex cf. zeteki before mating
flights revealed the presence of the cuticular actinomycete on females only (
To determine whether the attine-associated actinomycete
produces compounds with beneficial antimicrobial properties, we performed bioassays in
which taxonomically diverse sets of fungi were challenged with attine Streptomyces
isolates. Streptomyces isolates obtained from Acromyrmex octospinosus lacked
detectable inhibitory effects on the growth of generalist saprotrophic fungi,
entomopathogenic fungi, and other fungi commonly used for antibiotic screening. However,
it showed potent inhibitory effects towards Escovopsis, a fungal genus (anamorphic
Hypocreales: Ascomycotina)13
identified recently as a specialized virulent parasite of the attine fungal gardens6. The actinomycete completely
suppressed spore germination of Escovopsis isolates in 25% of bioassays. Linear
fungal growth in the remaining challenges was inhibited by 
3.0% s.e.m.
 |
Figure 3 Bioassay challenge between Streptomyces and Escovopsis, the specialized parasite of attine fungal gardens, associated with Acromyrmex octospinosus, illustrating the substantial zone of inhibition of fungal growth. |
We tested growth-enhancing effects of the actinomycete on the
basidiomycete mutualist in broth culture bioassays, using a Streptomyces isolate
obtained from the phylogenetically basal attine genus Apterostigma. We observed
significant increases in basidiomycete vegetative biomass in the presence of the
actinomycete culture filtrate (vegetative biomass in presence of actinomycete culture
filtrate averaged 7.6 mg2.4 mg
Several lines of evidence indicate that this newly discovered bacterial symbiont of the attine ants is a third mutualist in an ancient symbiosis among the ants, the domesticated fungi, the parasitic fungi, and the antibiotic-producing bacteria. First, this actinomycete is associated with all attine species and all colonies studied, representing the generic diversity of the Attini. Second, the actinomycete is transmitted vertically from parent to daughter nest, as is the fungal mutualist, and third, it promotes the growth of the fungal mutualist in vitro. Fourth, and most important, this Streptomyces produces highly potent antibiotics that selectively inhibit the growth of the garden parasite Escovopsis. Although many antibiotics developed for medicinal use have fairly broad-spectrum effects, it is likely that most fungicidal secondary metabolites produced by microbes evolved toward specific targets, such as competitors and pathogens17,18. As the production of secondary metabolites is energetically costly and requires complex, genetically based biosynthetic pathways, its evolution and subsequent maintenance is presumed to impart a substantial selective advantage to the microbe17. Thus the production of antibiotics by the attine-associated Streptomyces that specifically target the growth of Escovopsis is compelling evidence that the bacterium is a highly evolved mutualist. The ants use this bacterium, because of its production of antibiotics, to treat microbial infection of their garden, and, in exchange, the ants disperse the actinomycete and appear to provide some form of nourishment for its growth.
Although the ant-fungus mutualism is often regarded as one of the most fascinating examples of a highly evolved symbiosis, it is now clear that its complexity has been greatly underestimated. The attine symbiosis appears to be a co-evolutionary 'arms race' between the garden parasite, Escovopsis, on the one hand, and the tripartite association amongst the actinomycete, the ant hosts, and the fungal mutualist on the other. The evolution of a mutualistic association between the attine ants and an actinomycete that suppresses parasites is perhaps not surprising. The benefits of such a symbiosis are illustrated by the paramount part played by therapeutic antibiotics in human biomedical history. Our results indicate that microbes and their metabolites may be key components in the regulation of other symbiotic associations between higher organisms, and thus a more detailed analysis of their functions promises to illuminate the general dynamics of symbiosis. Study of the presumably highly evolved chemical interactions between symbionts may provide valuable theoretical and practical insights regarding the identification, production and application of antibiotics19-21.
Methods
Identification of bacterium. To identify the bacterium, we used micromorphological
parameters as well as accepted biochemical criteria. We analysed the cell-wall fatty-acid
methyl esters by gas-liquid chromatography22. The absence of tuberculosteric acid and related
10-methyl esters excluded the genera Actinomadura and Nocardiopsis, which
are morphologically similar to Streptomyces.
Attines. The presence of actinomycete was determined in fungus-growing ant species from the canal region of Panama (112 colonies, 17 species) and the Napo province of Ecuador (25 colonies, 5 species). Attine genera studied included a representative sampling of both 'lower' (phylogenetically basal) and 'higher' (phylogenetically derived) attines, namely Myrmicocrypta (two species), Apterostigma (four species), Mycocepurus (two species), Cyphomyrmex (two species), Sericomyrmex (one species), Trachymyrex (five species), Acromyrmex (three species) and Atta (three species).
Antibiotic-bioassay challenges. Bioassays were done in Petri dishes using an actinomycete isolate obtained from a phylogenetically derived attine species, Acromyrmex octospinosus. Saprotrophic fungi isolated from attine gardens, including taxa closely related to Escovopsis, were tested, as were generalist entomopathogens. We also assayed a diverse, representative set of fungi used for general anti-fungal screening. Finally, we challenged representative strains of the specialist garden parasite Escovopsis. Specifically, two strains of Metarhizium anisopliae and one strain of Beauveria bassiana were challenged with actionomycete, as were the following common microfungi: Absidia sp., Ascobolus crenulatus, Aspergillus fumigatus, Coprinus patouillardii, Cryptococcus albidus, Drechslera biseptata, Exophiala spinifera, Fusarium oxysporum, Mucor pyriformis, Penicillium sp., Pythium aphanidermatum, Schizophyllum commune, Sordaria fimicola and Trichoderma sp.
Each Streptomyces-fungal challenge was replicated
three times and done on Czapek yeast autolysate agar. The actinomycete was inoculated on
Petri dishes and grown to a diameter of 
1.5 cm; fungal strains were then point-inoculated near the
edge of the culture. Challenges were monitored every two days and growth inhibition of
tested fungi was scored as a reduction of growth rate as compared with growth of fungal
cultures in the absence of the Streptomyces, or as complete suppression of growth.
We assayed possible antibiotic production specific to the specialized parasite Escovopsis
in the same way that we assayed antibiotic production specific to other potential
contaminants, except that each challenge to Escovopsis was replicated five times.
Four strains of Escovopsis isolated from the gardens of different Acromyrmex
octospinosus colonies in Panama in 1997 were tested against Streptomyces. We
also studied the production of antibiotics specific towards Escovopsis in other
attine species, including Cyphomyrmex longiscapus, Atta colombica and Atta
cephalotes. The presence of a zone of inhibition in bioassays indicates first, the
production of diffusible metabolites by the actinomycete, and second, the susceptibility
of the test fungus to these compounds. As inhibition is dose dependent, the detection of
partial inhibition implies the existence of a dose that could impart complete inhibition.
Growth-promotion bioassays. Broth cultures of the attine fungus isolated from an Apterostigma colony were grown with extracts from the Streptomyces isolated from this species. Actinomycete extracts were obtained by growing Streptomyces in Czapek yeast autolysate broth for 2 weeks and then passing the broth through a low protein-binding, sterilizing filter unit (Millipore, Millex) to remove bacterial biomass. We replicated each bioassay five times and used 50 ml Czapek yeast autolysate broth per bioassay.
Received 20 November 1998;
accepted 18 February
1999.
References
Acknowledgements. This work was supported by
Smithsonian and NSERC predoctoral awards (to C.R.C.) and an NSERC grant (to D.M.). C.R.C.
thanks the Smithsonian Tropical Research Institute and ANAM of the Republic of Panama for
assisting with the research and granting collecting permits, and U. G. Mueller for
guidance, support and encouragement. We thank I. Ahmad, N. Alasti-Faridani, G. de Alba,
S. Barrett, E. Bermingham, A. Caballero, J. Ceballo, S. Dalla Rosa, L. Ketch, M.
Leone, G. Maggiori, S. Rand and M. Witkowska for logistical support; C. Ziegler for
the photograph in
Correspondence and requests for materials should be addressed to C.R.C. (e-mail: currie@botany.utoronto.ca).