Adaptation via Symbiosis: Recent Spread of a Drosophila Defensive Symbiont

See allHide authors and affiliations

Science  09 Jul 2010:
Vol. 329, Issue 5988, pp. 212-215
DOI: 10.1126/science.1188235

Offsetting the Cost of Parasitism

Fruit flies, like most animals, are vulnerable to infection by a range of organisms, which, in co-infections, can interact with sometimes surprising effects. Jaenike et al. (p. 212) discovered that a species of Spiroplasma bacterium that is sometimes found in flies, and that is transmitted from mother to offspring, protects its host from the effects of a nematode worm parasite, Howardula aoronymphium. The worm sterilizes the female flies and shortens their lives, but when flies were experimentally infected with Spiroplasma, their fertility was rescued. Similarly, in wild populations of fruit flies infected with worms, those also infected with Spiroplasma had more eggs in their ovaries. The bacterium inhibits the growth of the adult female worms, but such is the advantage of this bacterial infection in offsetting the burden of nematodes on reproductive fitness, Spiroplasma appears to be spreading rapidly through populations of fruit flies in North America.


Recent studies have shown that some plants and animals harbor microbial symbionts that protect them against natural enemies. Here we demonstrate that a maternally transmitted bacterium, Spiroplasma, protects Drosophila neotestacea against the sterilizing effects of a parasitic nematode, both in the laboratory and the field. This nematode parasitizes D. neotestacea at high frequencies in natural populations, and, until recently, almost all infections resulted in complete sterility. Several lines of evidence suggest that Spiroplasma is spreading in North American populations of D. neotestacea and that a major adaptive change to a symbiont-based mode of defense is under way. These findings demonstrate the profound and potentially rapid effects of defensive symbionts, which are increasingly recognized as major players in the ecology of species interactions.

The ancient origin (13) yet ongoing rapid evolution (46) of genes involved in defense against pathogens and parasites indicate that infective agents have been and continue to be major selective factors for virtually all organisms. In addition to the arsenal of nuclear genes encoding diverse and sophisticated mechanisms of defense, some organisms carry symbiotic microbes that provide defense against natural enemies (7, 8). In insects, for example, maternally transmitted symbionts have recently been shown to provide protection against parasitoid wasps, fungal pathogens, and RNA viruses (913).

Nematodes are probably the most abundant, diverse, and destructive macroparasites of plants and animals (1417). Nematodes commonly attack Drosophila (18), and at least 10 mushroom-feeding species of Drosophila are parasitized by the nematode Howardula aoronymphium (Allantonematidae, Tylenchida) (18). Mated female Howardula infect Drosophila larvae, persist to the adult stage of flies, and release offspring that are passed from the fly via the gut and ovipositor into mushrooms, where the nematodes mate to renew the cycle (fig. S1). In the eastern United States, the most commonly infected Drosophila species is D. neotestacea, with a mean prevalence of parasitism of 0.23 around Rochester, New York (fig. S2) (19). Infections are severe, generally rendering females completely sterile, as well as reducing adult survival and male mating success (18). Given the high frequency and virulence of infection, there must be strong selective pressure on D. neotestacea to evolve defenses against these nematode parasites.

D. neotestacea is also infected with two maternally transmitted bacterial endosymbionts, Spiroplasma and Wolbachia, neither of which acts as a reproductive parasite in this species (20). To test whether the endosymbionts confer defense against nematode parasites, we exposed replicate iso-female lines of D. neotestacea to parasitism by H. aoronymphium in the laboratory, using lines co-infected with Spiroplasma and Wolbachia (SW), infected with Spiroplasma only (S), infected with Wolbachia only (W), or uninfected (U) (21). The fertility of nematode-parasitized females was greater if they were infected with Spiroplasma, but not Wolbachia, indicating that such flies have greater tolerance of parasitism [as defined in (22)] (Fig. 1A; nematode parasitism × Spiroplasma infection: F1,723 = 239, P < 0.0001; parasitism × Wolbachia interaction: F1,723 = 0.04, P = 0.84). We failed to detect any bacterial symbionts other than Spiroplasma in the S lines of D. neotestacea by cloning and sequencing 16S ribosomal DNA, suggesting that Spiroplasma alone is responsible for the fertility rescue.

Fig. 1

(A) Fertility (measured as number of eggs at stage 10B or later per ovary) in laboratory-reared females of D. neotestacea as a function of Howardula aoronymphium parasitism, Spiroplasma infection, and Wolbachia infection. Hatched bars, nematode-parasitized flies; unhatched bars, unparasitized flies. Spiroplasma infection increases the fertility of females parasitized with H. aoronymphium (compare U to S and W to SW). In contrast, Wolbachia has no such effect (compare U to W and S to SW). (B and C) Fertility of 2008 field-collected D. neotestacea females parasitized (B) or not parasitized (C) by Howardula. Gray, infected with Spiroplasma; white, uninfected. The fertility of unparasitized flies was independent of Spiroplasma infection, whereas for parasitized flies, those carrying Spiroplasma were more fertile than uninfected flies. (D and E) Fertility of 1989 field-collected D. neotestacea females parasitized (D) or not parasitized (E) by Howardula. Mean number of eggs per ovary = 0.5 ± 0.2 and 14.5 ± 0.7 for parasitized and unparasitized flies, respectively. These flies were not scored for Spiroplasma infection, but note that the fertility of Howardula-parasitized flies in 1989 was similar to that of parasitized flies in 2008 that were not infected with Spiroplasma, suggesting a low level of Spiroplasma infection in 1989. (F) Nematode motherworms are significantly smaller in flies that harbor Spiroplasma. Motherworm size within one-week-old Spiroplasma-infected (gray) and uninfected (white) individuals of D. neotestacea experimentally parasitized with Howardula. N, number of flies (B to E) or motherworms (F).

To test whether Spiroplasma is associated with tolerance of nematode parasitism in the wild, we dissected D. neotestacea collected from natural populations and scored them for Howardula parasitism and the number of mature eggs per ovary. We then used polymerase chain reaction (PCR) to screen these flies for infection with Spiroplasma and Wolbachia. Female fertility was significantly affected by a nematode parasitism × Spiroplasma interaction (F1,211 = 4.45, P = 0.036; Fig. 1, B and C). Among nematode-parasitized flies, females infected with Spiroplasma had >10× greater fertility than those not infected with this symbiont (means = 11.13 ± 0.89 and 0.95 ± 0.35 eggs per ovary, respectively; F1,129 = 19.34, P < 0.0001). Among unparasitized flies, females harboring Spiroplasma carried a mean of 17.2 ± 0.8 eggs per ovary, and those without Spiroplasma carried 17.2 ± 1.5 (F1,87 = 0, P = 0.99), indicating that Spiroplasma has little effect on the fertility of unparasitized flies. All main effects and interactions involving Wolbachia were nonsignificant (P > 0.5), indicating that it does not play a role in defense against nematode parasitism.

To explore how Spiroplasma confers tolerance of nematode parasitism, we measured the sizes of motherworms (inseminated adult female worms within flies) in experimentally parasitized one-week-old D. neotestacea females. Using antibiotics, we selectively cured a doubly infected SW line of D. neotestacea from wild populations of either Wolbachia only or both Spiroplasma and Wolbachia; therefore, experimental flies had similar nuclear genetic backgrounds. At the motherworm stage, size is a good indicator of a nematode’s potential reproductive output and impact on the host, as a Howardula motherworm is largely a sack of embryos and developing juveniles (23). As measured by surface area, motherworms were only half as large in flies infected with Spiroplasma as in those that were uninfected (means = 0.44 ± 0.04 mm2 and 0.80 ± 0.04 mm2, respectively; F1,58 = 45.2, P < 0.0001), indicating that Spiroplasma adversely affects motherworm growth and reproduction (Fig. 1F).

Thus, in both the wild and the laboratory, Spiroplasma is associated with tolerance of nematode parasites that would otherwise cause sterility in D. neotestacea females, and it appears to do so by impairing, through an unknown mechanism, the growth of Howardula within parasitized flies. To our knowledge, this is the first example of Spiroplasma acting as a mutualist. Spiroplasma are among the most widespread bacterial associates of arthropods (24), including Drosophila (25, 26), but the role of Spiroplasma strains that experience solely vertical transmission has remained largely elusive. Although some Spiroplasma are reproductive parasites (27), mutualistic benefits may be responsible for the persistence of Spiroplasma in many host species. It is interesting to note that Spiroplasma occurs not only within cells of its invertebrate hosts but also within the hemocoel (28), where most parasitic nematodes, such as Howardula, reside (29).

Four lines of evidence independently suggest that the Spiroplasma infection is dynamic and spreading within natural populations of D. neotestacea. First, we PCR-screened museum specimens of D. neotestacea collected in the eastern United States in the early 1980s for Spiroplasma, Wolbachia, and, as a control for DNA quality, Drosophila cytochrome c oxidase subunit 1 (COI). Of the 20 flies, 18 (90%) were PCR-positive for Wolbachia, similar to current levels of Wolbachia infection in this species (20). In contrast, Spiroplasma was not detected, suggesting a prevalence in the 1980s in the range of 0 to 0.14 (the 95% confidence interval around 0 out of 20 Spiroplasma-infected flies). This is well below the current infection prevalence in eastern North America, which ranges from 0.5 to 0.8 at sites from Maine to Minnesota (20).

Second, almost all nematode-parasitized females of D. neotestacea collected in New York in the 1980s were sterile (18, 30). The fertility distribution of nematode-parasitized flies collected in 1989 (Fig. 1D) was similar to that of nematode-parasitized flies that were uninfected with Spiroplasma in 2008 (Fig. 1B). A small fraction of parasitized flies in the 1980s carried 10 or more eggs, suggesting they may have been infected with Spiroplasma. Thus, in populations of D. neotestacea in Rochester, Spiroplasma appears to have increased from a low frequency in the 1980s to ~0.8 in less than 20 years (fig. S3 and table S6) (20).

Third, there is a continent-wide cline in the prevalence of Spiroplasma infection in D. neotestacea (Fig. 2A). In contrast, there is much less geographic variation in the infection prevalence of Wolbachia, suggesting that it is close to equilibrium across North America (Fig. 2B). D. neotestacea is parasitized by Howardula throughout its range, from Maine to British Columbia. Howardula infection frequencies in coastal British Columbia, where Spiroplasma is absent, were 0.21 in 2008 (n = 296 flies surveyed for parasitism) and 0.25 in 2009 (n = 132), similar to the long-term 0.23 prevalence of parasitism near Rochester, New York (19). Thus, nematode parasites probably impose selection in favor of Spiroplasma infection across the range of D. neotestacea.

Fig. 2

Geographic variation in D. neotestacea mtDNA haplotype and symbiont status. (A) Prevalence of Spiroplasma infection, showing a significant decline in the prevalence of infection from east to west across North America (among-sites rank correlation between infection prevalence and longitude ρ = 0.89, P < 0.0001). (B) Prevalence of Wolbachia infection in D. neotestacea, showing a lack of longitudinal variation (among-sites rank correlation between infection prevalence and longitude ρ = 0.41, P = 0.13). Spiroplasma and Wolbachia infection prevalence data for sites east of the Rocky Mountains (except Montana) are from (20). (C) Molecular phylogeny of mtDNA haplotypes in D. neotestacea. Geographical distribution of the haplotypes, with sites arrayed from west to east along the top [Victoria, British Columbia (BCVi); Vancouver, British Columbia (BCVa); MacKenzie Bridge, Oregon (OR); Jasper, Alberta (ABJa); Columbia Falls/Big Sky, Montana (MT); Edmonton, Alberta (ABEd); Winston Churchill Provincial Park, Alberta (ABWc); Minot, North Dakota (ND); The Pas, Manitoba (MB); Bemidji, Minnesota (MN); Munising, Michigan (MI); Samuel de Champlain Province Park, Ontario (ON); Rochester, New York (NY); and Chebeague Island, Maine (ME)]. The Spiroplasma infection prevalence for each site and haplotype is indicated by the proportion of black shading in the pie diagrams. Samples sizes are indicated by the numbers within each pie diagram. The clinal variation in Spiroplasma prevalence, the association of Spiroplasma infection with “eastern” mtDNA haplotypes, and the clines in Spiroplasma prevalence within haplotypes all suggest that Spiroplasma is spreading from east to west across North America. The lack of longitudinal variation in Wolbachia is consistent with a long-term infection and is close to equilibrium prevalence everywhere.

Finally, the Spiroplasma infection status of flies carrying different mitochondrial haplotypes reveals a Spiroplasma infection not yet at species-level equilibrium. We previously found a perfect association between Spiroplasma haplotype and mitochondrial DNA (mtDNA) haplotype within Rochester, New York, populations of D. neotestacea, indicating that horizontal transmission of Spiroplasma is rare or nonexistent (20). Consequently, mtDNA variation can be used to infer the history of Spiroplasma infection in D. neotestacea. At equilibrium between natural selection favoring Spiroplasma infection and imperfect maternal transmission resulting in loss of Spiroplasma, the prevalence of Spiroplasma infection should be similar among flies with different mtDNA haplotypes; flies carrying all major mtDNA haplotypes should be infected, and all individuals, whether infected or not, should be descended from infected females (31, 32). This is not the case for D. neotestacea, as Spiroplasma is common in flies carrying certain mtDNA haplotypes (notably 1, 5, and 8) but absent from all flies carrying “western” haplotypes (e.g., haplotype 16; Fig. 2C). We identified 16 individuals, collected in Oregon and British Columbia, carrying the most common mitochondrial haplotypes in eastern North America, and none were infected with Spiroplasma, indicating that the absence of Spiroplasma in the west is not due to the absence of mitochondrial clades that elsewhere harbor Spiroplasma. Mean within-population mitochondrial diversity is greater in populations where Spiroplasma is absent (Embedded Image = 0.0056 ± 0.0007) than where it is present (Embedded Image = 0.0025 ± 0.0005; F1,12 = 11.07, P = 0.006), consistent with theoretical expectations that Spiroplasma has not been present in western populations of D. neotestacea in the recent evolutionary past (33). Taken together, these four patterns suggest that Spiroplasma has recently increased in frequency in the eastern populations of D. neotestacea and may now be spreading from east to west across North America.

The equilibrium infection prevalence of maternally transmitted endosymbiont isEmbedded Image, where β is the fidelity of maternal transmission of the symbiont, and s is the selective advantage of infected over uninfected cytoplasmic lineages (31). Using wild-caught females, we have estimated that β = 0.97 (20) and s = 0.17. The estimate of s is based on the fertility of wild females as a function of Howardula parasitism and Spiroplasma infection, weighted by the probability of nematode parasitism (21). The expected equilibrium prevalence, Embedded Image, is similar to that observed in populations in eastern North America, suggesting that Spiroplasma prevalence is at or approaching an equilibrium based largely on a balance between imperfect maternal transmission and a selective advantage due to tolerance of nematode parasitism. Our estimates of β and s are also consistent with a hypothesized increase in Spiroplasma infection around Rochester from ~10% in the 1980s to ~80% today (fig. S4).

Does the apparent recent increase of Spiroplasma result from recent colonization of D. neotestacea by Spiroplasma, a recent favorable Spiroplasma mutation conferring tolerance to an existing parasite challenge, or the imposition of a new selective pressure? The occurrence of Spiroplasma in flies carrying three different mtDNA haplotypes (Fig. 2C) suggests that the colonization of D. neotestacea by Spiroplasma was not a recent event. We previously found a perfect match between two slightly different Spiroplasma variants and two closely related mtDNA haplotypes, indicating that sufficient time has elapsed since the original infection for mutations in both Spiroplasma and mtDNA to have accumulated in the infected cytoplasmic lineages (20). Thus, Spiroplasma was probably present within D. neotestacea long before its recent increase. We can also rule out a recent favorable mutation, as both of these Spiroplasma variants were associated with tolerance to nematode parasitism. Among nematode-parasitized flies collected in 2008 and for which spoT was sequenced, the mean egg numbers for flies carrying the two variants were 13.5 ± 1.0 and 17.2 ± 2.2, both of which were much greater than the 0.95 ± 0.35 eggs in parasitized flies that did not carry Spiroplasma (F1,90 = 41.9 and F1,29 = 83.2, respectively; both P values < 0.0001). Finally, we previously hypothesized that H. aoronymphium had recently colonized North America (34), based on our finding of no DNA sequence variation (mtDNA COI) among North American samples of H. aoronymphium, as well as sequence identity between North American and European samples of this species (35). Thus, the apparently rapid spread of Spiroplasma is most likely due to recently imposed selection on D. neotestacea to evolve tolerance of these sterilizing parasites. The presumed beneficial function of Spiroplasma in D. neotestacea before the arrival of H. aoronymphium is unknown.

Our results show that Spiroplasma rescues D. neotestacea females from the sterilizing effects of nematode parasitism and that this endosymbiont appears to have recently increased to high frequency in eastern North America and is now spreading from east to west across the continent. Thus, D. neotestacea is undergoing a major change to a symbiont-based mode of defense against nematode parasites. This is the first report of natural symbiont-mediated defense against nematodes, the most widespread macroparasites of plants and animals. From an applied perspective, these findings suggest novel measures for nematode control (36); for instance, river blindness and lymphatic filariasis are caused by nematodes that are transmitted by various species of flies (37). If Spiroplasma impaired the development of filarial nematodes within their insect vectors, this could reduce nematode transmission and, thus, incidence of disease in human populations. With respect to natural communities, this study demonstrates the profound and potentially rapid effects of defensive symbionts, which are increasingly recognized as major players in the ecology of species interactions (713).

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S4

Tables S1 to S6


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. This work was supported by NSF grants DEB-0542094 and DEB-0918872 to J.J. and a Natural Sciences and Engineering Research Council of Canada Discovery Grant to S.J.P. S.J.P. is a Scholar in the Canadian Institute for Advanced Research. We thank C. Cornish, P. Gibas, and K. Dyer for collecting some of the North American D. neotestacea used in this study; A. Marshall and L. Harris for technical assistance; N. Polet for the cloning work; D. Grimaldi for providing museum specimens of D. neotestacea; and J. Coyne, R. Minckley, K. Oliver, and D. Presgraves for comments on the manuscript. Sequences have been deposited in GenBank under accession numbers GU552299 to GU552304, HM126644 to HM126667, and HM133591 to HM133593.
View Abstract

Stay Connected to Science

Navigate This Article