The Hologenomic Basis of Speciation: Gut Bacteria Cause Hybrid Lethality in the Genus Nasonia

See allHide authors and affiliations

Science  09 Aug 2013:
Vol. 341, Issue 6146, pp. 667-669
DOI: 10.1126/science.1240659

Microbes → Host Speciation?

No living organism is an individual—an individual's microbiota can outnumber the host's somatic cells. Working in parasitoid wasps, Brucker and Bordenstein (p. 667, published online 18 July) now suggest that the gut microbiota can play a crucial role in speciation and hybrid lethality. In a clade of parasitoid wasps, interspecies hybrids survived when reared on antibiotic-treated sterile food (thus eliminating gut microbiota), but experienced high mortality when reared on conventional diet or host material.


Although the gut microbiome influences numerous aspects of organismal fitness, its role in animal evolution and the origin of new species is largely unknown. Here we present evidence that beneficial bacterial communities in the guts of closely related species of the genus Nasonia form species-specific phylosymbiotic assemblages that cause lethality in interspecific hybrids. Bacterial constituents and abundance are irregular in hybrids relative to parental controls, and antibiotic curing of the gut bacteria significantly rescues hybrid survival. Moreover, feeding bacteria to germ-free hybrids reinstates lethality and recapitulates the expression of innate immune genes observed in conventionally reared hybrids. We conclude that in this animal complex, the gut microbiome and host genome represent a coadapted “hologenome” that breaks down during hybridization, promoting hybrid lethality and assisting speciation.

Although the gut microbiome influences numerous fitness traits in animals, little attention has been given to how the microbiome is structured between closely related species and how the microbiome contributes to the origin of new species. By incorporating the microbiome into the biological species concept (1) and the Bateson-Dobzhansky-Muller model of hybrid incompatibilities (2), theoretical evidence specifies that negative epistasis between host genes and the host microbiome can accelerate the evolution of hybrid lethality and sterility (3).

We recently established that when environmental factors such as diet are controlled for, species of the Nasonia wasp complex harbor phylosymbiotic gut microbiotas—a term introduced here to denote microbial community relationships that recapitulate the phylogeny of their host (4). Similar to phylogenomics, phylosymbiosis asserts that the relationships of microbiomes across host species maintain an ancestral signal of the host's evolution. The phylosymbiotic signal could be a consequence of host immune genes that rapidly evolve in a continual arms race with components of the microbiome.

We hypothesize that the phylosymbiotic gut microbiome within species breaks down in hybrids via epistatic interactions between the microbiome and nuclear and/or cytoplasmic genomes, leading to hybrid lethality. Using the parasitoid wasp genus Nasonia, we set out to experimentally determine the influence of the microbiome on hybrid lethality. The Nasonia genus consists of several species of haplodiploid parasitoid wasps that are readily hybridized. Nasonia vitripennis diverged approximately 1 million years ago from the ancestor of Nasonia giraulti and Nasonia longicornis, which themselves diverged less than 400,000 years ago (5). In the laboratory, all three species are reared on the fly host Sarcophaga bullata under identical conditions. Nasonia offspring are oviposited by the mother inside the puparium of the fly, where the eggs develop to adulthood before emerging from the fly host in about 14 days. In the absence of Wolbachia infections, reciprocal crosses between species produce fertile, diploid, F1 hybrid females (6). However, hybrid lethality is observed in F2 male offspring, because they are haploid recombinants of their grandparents. Interspecies crosses of N. vitripennis and N. giraulti or N. longicornis result in F2 hybrid males exhibiting up to ~90% lethality during larval development, whereas N. giraulti and N. longicornis hybrids only exhibit ~8% hybrid lethality (610).

In this study, we scored the average number of hybrid and nonhybrid eggs, larvae, pupae, and adults produced by Nasonia females (n = 48 females, each parasitizing one S. bullata host per developmental period) and determined that 78% of hybrid lethality occurs between the first- and the fourth-instar (L4) larval stages (Fig. 1A). There was a slight asymmetry in the number of surviving N. vitripennis/N. giraulti and N. giraulti/N. vitripennis hybrids (the F2 hybrid genotype denotes grandfather or grandmother), but the difference was not statistically significant (P = 0.36, Mann-Whitney U test). In contrast, hybrids of the younger sister species N. giraulti and N. longicornis exhibited little to no F2 hybrid lethality, as previously reported (6) (fig. S1A). The hybrid lethality between N. vitripennis and N. giraulti is often diagnosed by larval melanization—a prominent immune response to pathogens in arthropods (Fig. 1B). We postulated that because parental Nasonia species assemble phylosymbiotic gut microbiomes (4) (Fig. 2), the melanization and lethality in larval hybrids result in part from altered gut microbiomes.

Fig. 1 The symbiotic and genetic basis of hybrid lethality.

(A) Average number of F2 males (±SEM) within the S. bullata host during development of the egg, L1 larvae, L4 larvae, yellow red-eye pupae, and eclosed adults conventionally reared on S. bullata hosts. The F2 hybrid genotype is indicated as paternal/maternal where v = N. vitripennis and g = N. giraulti. Mann-Whitney U test, **P < 0.001. n = 48 replicates per developmental stage. (B) (Top) N. vitripennis L3 larva that is healthy and alive; (bottom) hybrid v/g L3 larva that is melanized and dead. (C) Percent of survival (±SEM) from egg to pupae of conventionally reared (red), germ-free (blue), and inoculated (purple; germ-free individuals inoculated with Providencia and Proteus bacteria in the NRM) N. vitripennis (v) and N. giraulti (g) parental species and hybrids. Mann-Whitney U test, **P < 0.001, F1,46 = 23.863 and 12.962, ***P < 0.001, an average of triplicate experiments with n = 48 hosts for egg and pupae counts per conventional cross, and n = 24 wells containing 12 to 42 larvae per germ-free and inoculated cross. (D) Average hybrid v/g gene expression relative to conventionally reared hybrids for the total genome, OXPHOS genes, and immunity genes (t test, **P < 0.001). RPKM denotes reads per kilobase per million mapped reads. Conventional n = 14, germ-free n = 20, and inoculated n = 20 individuals were sequenced and averaged.

Fig. 2 Phylosymbiosis and speciation in Nasonia.

(A) Simplified phylogeny of Nasonia (bold lines) and S. bullata flies. N. vitripennis (v), N. giraulti (g), and N. longicornis (l). (B) A weighted, to abundance of each OTU, UniFrac cluster analysis depicting the microbial relationships of the three Nasonia species (bold lines) and two hybrids for the L2 larval microbiota, as well as the unparasitized Sarcophaga host pupa. The tip of each branch is a pie chart depicting the abundance of each genus of bacteria within the host insects. Genera of the top 10 most abundant bacterial genera are listed in the key. (C) An unweighted UniFrac cluster analysis depicting the microbial relationships; pie charts at the tip of each branch represent the genera level, bacteria diversity sampled.

First, we tested the hypothesis that bacterial community differences occur between larval hybrids and nonhybrids during the L2 larval stage, just before the point of F2 hybrid male lethality. We focused on the N. vitripennis/N. giraulti hybrid microbiota, because this genotype elicits elevated lethality as compared to the reciprocal cross; however, this asymmetry was insignificant in our experiments. As hypothesized, the microbiota of N. vitripennis/N. giraulti hybrids was unlike that of either parental species in both bacterial abundance (Fig. 2B) and diversity (Fig. 2C), whereas the negative control N. longicornis/N. giraulti hybrids that survived (6) (fig. S1A) had a parental-like microbiota. Both hybridizations had a substantial number of novel and rare operational taxonomic units (OTUs, P = 0.06, chi square test with Tates correction, fig. S3). The single major difference in the N. vitripennis/N. giraulti hybrid microbiota was a shift in the dominant bacterial OTU from Providencia sp. IICDBZ10 in pure species controls (81 and 96% of the reads in N. vitripennis and N. giraulti, respectively) to Proteus mirabilis strain SNBS in the N. vitripennis/N. giraulti hybrid (86% of reads). These two species are natural residents of the Nasonia parental species. P. mirabilis SNBS is the dominant species in N. longicornis, and Providencia sp. IICDBZ10 is the dominant species in N. vitripennis and N. giraulti.

Second, to test whether the N. vitripennis/N. giraulti F2 lethality in hybrids is conditional on the microbiome, we reared conventional (on a normal S. bullata host), germ-free (without bacteria), and bacteria-inoculated (without bacteria first and subsequently inoculated with specific bacteria) hybrids and nonhybrids. For germ-free rearing of Nasonia, we used Nasonia rearing medium (NRM), a liquid medium that we recently developed for culturing Nasonia without its S. bullata fly host (11). If hybrid lethality in larvae is intrinsically based on negative epistasis between host incompatibility genes, then the null hypothesis is that N. vitripennis/N. giraulti and N. giraulti/N. vitripennis hybrid lethality occurs regardless of germ-free or conventional rearing conditions. However, if the lethality is conditional on the microbiome, then we expect that germ-free rearing of hybrids will rescue hybrid lethality.

A comparison of results indicates a near-complete rescue of hybrid lethality in germ-free hybrids relative to pure species controls under germ-free conditions (Fig. 1C, F1,46 = 1.207, P = 0.277 for N. vitripennis to N. giraulti/N. vitripennis, and F1,46 = 0.824, P = 0.369 for N. giraulti to N. vitripennis/N. giraulti). Therefore, hybrids that would typically show severe lethality under conventional rearing conditions exhibit a striking increase in survival. In a subsequent experiment, when F1 female hybrids oviposited their F2 male hybrid offspring into germ-free (GF) S. bullata fly hosts, there was also a marked increase in survival relative to hybrids that were reared on conventional (Cv) S. bullata fly hosts (fig. S1B). In contrast, in N. longicornis/N. giraulti and N. giraulti/N. longicornis hybrids, survival values of germ-free hybrids and nonhybrid controls were expectedly high and similar to each other (fig. S1A, Mann-Whitney U test, P > 0.5).

If the NRM generally increased survival, it should also affect nonhybrid survival as it would that of hybrids. However, average survival rates of controls decreased insignificantly on NRM. Further, germ-free hybrids that were reared on NRM inoculated with bacterial strains once again yielded higher lethality as compared to parental controls (Fig. 1C). The common Nasonia bacteria Providencia rettgeri strain IITRP2 and Proteus mirabilis strain SNBS were isolated from parental Nasonia and added to germ-free NRM. Upon consuming a 1:1 inoculum of these bacteria, germ-free hybrids exhibited severe lethality in comparison to nonhybrids (Fig. 1C, F1,46 = 23.863, P < 0.001) and at levels similar to those of conventionally reared hybrids (Fig. 1C, Mann-Whitney U test, P > 0.5). Furthermore, mono-inoculants of antibiotic-resistant (AR) P. rettgeri and Enterococcus faecalis strain XJALT-127-2YG1 isolated from Nasonia, as well as green fluorescent protein (GFP)–expressing E. coli, also recapitulated significant hybrid lethality (fig. S1B, F1,46 = 36.372, P < 0.001, fig. S2A, F1,46 = 7.281, P < 0.01; see the supplementary text).

The requirement of gut bacteria for interspecific hybrid lethality in Nasonia is remarkable given that several studies previously mapped quantitative trait loci with hybrid lethality to the wasp chromosomes and mitochondria (5, 7, 8, 10). Indeed, marker transmission ratio distortion (MTRD) analyses in surviving Nasonia F2 hybrids demonstrate an allelic bias at several loci in the genome toward one parental genome, indicating that an incompatible allele from the other parental species contributes to hybrid lethality (5, 911). Based on our observation that germ-free hybrid Nasonia exhibit a significant increase in survival, we predicted that MTRD would revert to near-Mendelian inheritance ratios in the germ-free hybrids. We selected four markers—three within genomic regions associated with hybrid lethality and one control locus that is expected to have 50:50 inheritance ratios—and genotyped N. vitripennis/N. giraulti and N. giraulti/N. vitripennis hybrid Nasonia L4 larvae. Larvae reared conventionally yielded the expected distorted frequencies for each of the MTRD markers (5), whereas germ-free hybrids exhibited typical Mendelian inheritance at all markers except for MM5.03 (table S1). A large bias against N. vitripennis alleles remained (a frequency of 0.08), although the frequency was significantly higher from the MTRD under conventional rearing (N. vitripennis frequency of 0.20, Z test of proportions, P < 0.001).

One explanation for a microbial basis for hybrid lethality is that negative epistasis (mismatched gene-gene interactions) occurs between chromosomal genes and the microbiome. These interactions can accelerate the number of potential hybrid incompatibilities under the Bateson-Dobzhansky-Muller model of genetic incompatibilities (3). To better understand the mechanisms behind host-microbe interactions that underscore hybrid lethality, we compared the transcriptomes of germ-free L2 hybrid larvae with those of conventionally reared and bacteria-inoculated L2 hybrid larvae (just before lethality). The genome-wide expression patterns and the oxidative phosphorylation (OXPHOS) family of genes, a family thought to be causative in hybrid lethality, were similar across all three rearing conditions (Fig. 1D and supplementary text). However, the 489 innate immune genes in Nasonia that we previously annotated (12) yielded, on average, a significant decrease in transcript levels in germ-free individuals relative to conventional or inoculated hybrids (Fig. 1D, t test, P < 0.001). Specifically, 39.7% of the immune genes were underexpressed by twofold or greater in germ-free hybrids relative to conventional and inoculated hybrids, and 4.9% were overexpressed (fig. S4 and table S2). Conventionally reared and inoculated hybrids, in turn, have similar immune gene expression (Fig. 1D, t test, P = 0.104). However, it is important to note that immune genes may be only one of several possible functional categories that break down between the host and microbiome during hybrid lethality.

Causes for the postzygotic hybrid lethality in Nasonia have traditionally been attributed to host cytonuclear interactions and host gene-by-environment interactions (710). However, this study shows that severe hybrid lethality in larvae can also be due to gene-microbe interactions with beneficial members of the phylosymbiotic gut microbiome. In this light, the phylosymbiotic microbiome can be understood as an addition to the coadapted genomes of a host organism rather than an arbitrary amalgam. Linking the microbiome and host genome underscores the hologenome as a unit of evolution and blurs the lines between what biologists typically demarcate as the environment (13) and the genotype of a species. Based on the mounting evidence for speciation by symbiosis (3), it is becoming clearer that a unified theory of evolution that considers the nuclear genome, cytoplasmic organelles, and microbiome as interacting components in the origin of new species is an emerging frontier for biology.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S4

Tables S1 to S6

References (1426)

References and Notes

  1. Acknowledgments: Sequences are available at the Dryad Digital Repository, doi:10.5061/dryad.3c190. We thank S. Bordenstein and A. Williams for technical assistance during the development of the in vitro cultivation method; L. Funkhouser, D. Sutherland, and C. Wogsland for their assistance in genotyping Nasonia hybrids; R. Pauly for bioinformatic assistance; and B. Jovanovic, L. Funkhouser, K. Jernigan, and N. Renner for providing feedback on an earlier version of the manuscript. We apologize in advance to colleagues whose papers we could not cite due to space restrictions. This research was made possible by NSF award DEB 1046149 to S.R.B.
View Abstract

Navigate This Article