Hybrid Incompatibility and Speciation

Hybrids between closely related species are often inviable or, if they live, they are sterile. This hybrid inviability and sterility, collectively known as hybrid incompatibility, can reduce the exchange of genetic variants between species. Thus, hybrid incompatibility can be important in the process of speciation by acting as a reproductive isolating barrier (Coyne & Orr, 2004). Hybrid incompatibility, as a phenotype, is unusual, as it serves no apparent function and yet is so common. Naturalists and philosophers since Aristotle have been puzzled by its existence. Charles Darwin argued that hybrid incompatibility does not evolve because of direct natural selection for hybrid incompatibility, but is instead a by-product of other evolutionary divergence between the species (Darwin, 1859). Contemporary studies have confirmed Darwin's insight.

What is the genetic basis of hybrid incompatibility? Changes in ploidy number certainly are an important factor in incompatibilities seen in plant hybrids, though less so for animal hybrids (Coyne & Orr, 2004). Chromosomal rearrangements also sometimes contribute to hybrid sterility and inviability in both animals and plants. Increasing evidence, however, points to much hybrid incompatibility being the result of alleles at different genetic loci that do not function well together (Johnson, 2000). In recent years, evolutionary geneticists have mapped many hybrid sterility and hybrid inviability factors to increasingly smaller regions of chromosomes. In fact, a handful of genes that contribute to hybrid incompatibility (mainly in the genus Drosophila) have been characterized and sequenced (Wu & Ting, 2004; Oritz-Barrientos et al., 2007). The nature of these hybrid incompatibility genes has begun to shed light on the evolutionary processes that have led to hybrid sterility and inviability.

Two geneticists, Russian-born American evolutionary geneticist Theodosius Dobzhansky and American Nobel Laureate Hermann Joseph Muller, dominated the early period of genetic studies of hybrid incompatibility. During the 1930s and early 1940s, these scientists used genetic techniques (such as backcrosses), as well as cytogenetic and embryological techniques, to provide proof of the principle that hybrid incompatibility, like other traits, could be subject to genetic analysis. Dobzhansky and Muller worked with species of Drosophila: Muller with the genetic model species D. melanogaster and D. simulans, and Dobzhansky with the North American species D. pseudoobscura and its relatives.

Dobzhansky (1937) and Muller (1942) independently formulated a model of how hybrid incompatibility could evolve. They both realized that hybrid inviability or sterility was unlikely to arise from a single change at one locus. Their reasoning was based on evolutionary thinking. It is conceivable that hybrid inviability could be due to hybrids being heterozygous for different alleles at a single locus (underdominance). In other words, individuals that are AA or aa would be viable, but individuals that are Aa would be inviable. The question, then, is how hybrid incompatibility via underdominance could ever evolve. If two populations that share a common ancestor are both homozygous for different alleles at the same locus (AA in one population, and aa in the other), an allelic substitution (either A for a or a for A) must have taken place in one of the populations. In this process, the population that changed would have had to go through an intermediate stage in which it had a high proportion of heterozygotes. These heterozygotes, by definition of the model, have low fitness. Thus, such changes could not take place by natural selection. It is possible that a population could change from being predominately AA to predominately aa (or vice versa) via genetic drift, but this would require one or more of the populations persisting at a very low population size.

Figure 1: Dobzhansky-Muller model of hybrid incompatibility.

How does genic incompatibility between species evolve without simultaneously causing defects in pure species? A popular explanation is the Dobzhansky-Muller (DM) model of hybrid incompatibility. In the ancestral population, the genotype is AA BB. When the population is split into two, A evolves into a in one population and B evolves into b in the other. a and b are mutually incompatible. As the a–b interaction is not present in the pure species, the evolution of incompatibility is possible. Detailed genetic analysis on hybrid male sterility, however, has shown that hybrid incompatibility often involves conspecific genic interactions as well. The DM model is far too simple in that respect. On the far right (hybrid), the divergence process is indicated by the black double-headed arrows and the incompatibility is indicated by the green double-headed arrow. The relationship between the black and green arrows should be the essence of the evolution of reproductive isolation. A deficiency of the DM model is that it does not consider the divergence process that is indicated by the black arrows but focuses instead on the incompatibility, which is a byproduct of that divergence. See PDF for original figure and associated references.

© 2004 Nature Publishing Group Wu, C. & Ting, C. Genes and speciation. Nature Reviews Genetics 5, 118 (2004). All rights reserved.

Figure Detail

The solution formulated by Dobzhansky and Muller was that hybrid incompatibility was the result of more than one change and involved multiple genetic loci (Figure 1). In the simplest case, two genetic changes are required. For instance, consider two loci (A and B). Assume that any combination of an a allele at locus A with a b allele at locus B causes hybrid inviability due to interactions between these two alleles. Crosses between a population with AAbb individuals and a population with aaBB individuals would yield AaBb offspring, which would be inviable due to the combined presence of a and b alleles. In contrast to the case of the one-locus underdominance previously discussed, the AAbb population and the aaBB population could have evolved from a common ancestral population without the generation of any unfit individuals as they evolved. Suppose the ancestral population was AABB, and it was split in two. One population could have evolved from AABB to AAbb (population 1) and the other could have evolved from AABB to aaBB (population 2). Unfit genotypes (like AaBb) would not have been required in the evolution of either population 1 or population 2. Other possible evolutionary paths also exist. For instance, AAbb could have been the ancestral population, and one population evolved stepwise from AAbb to AABB to aaBB, while the other population retained the ancestral genotype. The key point is that these so-called Dobzhansky-Muller incompatibility models assume that hybrid incompatibility arises from the interactions of more than one genetic change.

One example of Dobzhansky-Muller incompatibility occurs in hybrids between closely related species of fish. Many, but not all, individuals of the platyfish Xiphophorus maculatus have spots on their dorsal fins, while a related species (the swordtail, X. helleri) lacks spots (Coyne & Orr, 2004). In some backcross hybrids of these species, the spots are enlarged and develop into malignant tumors, which reduce the organisms' lifespans. What is the genetic basis of this hybrid incompatibility? Those platyfish with spots have an X-linked gene that produces spots, and all platyfish have an autosomal repressor that checks the expression of the spot-producing gene. In contrast, the swordtail lacks both the spot-producing gene and the repressor. In backcrosses, some of the hybrids receive the spot-producing gene, but not the repressor. These individuals are the ones that develop malignant tumors, because the expression of the spot-producing gene is not properly regulated. This spot-producing gene is a duplicated copy of the Xiphophorus melanoma receptor kinase (Xmrk). Xmrk-1 is present in all individuals of both fish species and doesn't lead to the formation of spots, while Xmrk-2 is the spot-producing gene. The repressor to Xmrk-2 has yet to be identified.

After the early 1940s, studies of the genetics of hybrid incompatibility waned and did not pick up again until the early 1980s, when Jerry Coyne reenergized the field. Coyne and his then-graduate student H. Allen Orr expanded the number of species pairs of Drosophila in which genetic studies of hybrid incompatibility had been examined. Coyne and his colleagues focused on three species of Drosophila that are closely related to D. melanogaster: D. simulans, D. mauritiana, and D. sechellia. Like D. melanogaster, D. simulans is found in most places around the world. The other two species, however, are restricted to small islands near Madagascar (the Seychelles Islands for D. sechellia, and Mauritius for D. mauritiana). Crosses between these species lead to F1 hybrid sterile males, but fertile F1 females. The fertility of the F1 females permits easy backcrossing. Successive backcrossing enabled Coyne to map several genetic factors contributing to hybrid sterility to small regions of the chromosome, and one of these factors to within a centimorgan (1 map unit) (Coyne & Charlesworth, 1986).

During the early 1990s, Chung-I Wu and his colleagues refined Coyne's studies in the same fly species trio by using what is known as the introgression approach. This technique enables mapping of the genetic factors contributing to hybrid sterility (or any hybrid trait) to progressively smaller regions of the chromosome, and eventually down to individual genes. The first step in the introgression approach is creation of a series of introgression lines; males of these introgression lines have a small part of the chromosome from one species (e.g., D. mauritiana) in the genetic background of another species (e.g., D. simulans) and are more or less genetically homogeneous. Creation of introgression lines involves a series of backcrosses guided by visible and molecular markers; sometimes, chromosomal inversions and other genetic tools are used to limit recombination. The fertility or lack thereof of among males is then assessed. Because males from a given introgression line are relatively genetically homogeneous, quantitative measures can also be made (i.e., one can assess the degree of the fertility or any other trait of the introgression males). The last step is to determine the nature of the introgression—that is, what specific part of the genome comes from which species. This is achieved through the use of molecular markers such as restriction fragment length polymorphisms (RFLPs) or microsatellites (Figure 2; Wu & Ting, 2004).

Figure 2: Positional cloning of hybrid incompatibility genes.

a) Recombination mapping: the coloured horizontal bars indicate introgressed chromosomes. The vertical orange bar marks the position of the gene that is responsible for hybrid sterility (for example, the Drosophila gene OdsH). Different sterile and fertile hybrids are typed for markers on the introgressed chromosomes that are known to be polymorphic between the two species (M1-M8). The pattern of marker distribution among the different hybrids allows the location of the hybrid sterility gene to be mapped at a resolution that is dependent on the concentration of informative markers in the region. b) Deficiency mapping: the phenotypes that are associated with various overlapping chromosome deficiencies (indicated by the gaps in the bars) show the position (marked with the vertical orange bar) of the hybrid incompatibility gene (for example, the Drosophila gene Nup96). Viable and inviable hybrids with different chromosome deletions are typed for markers that are known to be polymorphic between the two species (M1-M8). The pattern of marker distribution among the different hybrids allows the location of the hybrid inviability gene to be mapped at a resolution that is dependent on the concentration of informative markers in the region.

© 2004 Nature Publishing Group Wu, C. & Ting, C. Genes and speciation. Nature Reviews Genetics 5, 117 (2004). All rights reserved.

Figure Detail

The introgression approach enables a fine-scale mapping of genetic factors that contribute to hybrid incompatibility. This is a first step toward identifying the genes involved. Fine-scale mapping also provides information about the genetic architecture (the number of genes involved, their location, their dominance effects, etc.). From the introgression studies, we know that a large number of genes can contribute to hybrid incompatibility, even between closely related species. An estimated 100 genes contribute to the sterility of male hybrids between D. mauritiana and D. simulans, species that are separated by about half a million years (Wu et al., 1996; Johnson & Kliman, 2002). More genes contribute to hybrid male sterility than to hybrid female sterility or to hybrid inviability of either sex. This is not because of mutational differences between sterility and inviability; indeed, mutations affecting viability are much more common than those affecting sterility (Wu et al., 1996). Rather, genes that contribute to hybrid incompatibility can be found on all chromosomes but are found at higher densities on the X chromosome (Masly & Presgraves, 2007).

The Dobzhansky-Muller model predicts that hybrid incompatibility will be due to interactions between at least two genes (one from each species), but both Dobzhansky and Muller recognized that these interactions could be more complex. Hybrid sterility (or inviability) could require interactions of three or even more genes. Wu and his colleagues found that such complex interactions are not only possible, but are common in hybrid sterility between D. mauritiana and D. simulans (Cabot et al., 1994; Palopoli & Wu, 1994; Johnson, 2000).

A handful of genes that contribute to hybrid incompatibility in Drosophila have been characterized and sequenced. The first of these, named Odysseus (Ods), was sequenced by Chau-Ti Ting in Chung-I Wu's lab. When introgressed into D. simulans, Ods from D. mauritiana contributes to sterility of hybrid males. The reciprocal introgression, Ods from D. simulans into D. mauritiana, does not lead to hybrid male sterility. The protein encoded by Ods has a homeodomain, a 60-amino acid motif that binds to DNA and is found in many important developmental genes. This structure strongly suggests that Ods plays a role in regulating the expression of other genes, though the identity of those genes is not known (Ting et al., 1998).

In subsequent years, other genes that contribute to hybrid sterility and inviability between species of Drosophila have been characterized and sequenced. The data obtained from sequencing these incompatibility genes in multiple individuals in both species enable biologists to make inferences regarding the evolutionary forces operating on these genes. Most hybrid incompatibility genes, including Ods, show sequence patterns indicating that positive natural selection has been acting on these genes repeatedly during the divergence of the species (Wu & Ting, 2004; Coyne & Orr, 2004). The reason for the selection is not yet known.

Based on the Dobzhansky-Muller model, hybrid incompatibility should arise from the interactions of at least two genes. In nearly all cases, we do not yet know the identities of both interacting Dobzhansky-Muller genes. In some cases, researchers have mapped the interacting partners of those hybrid incompatibility genes that have been identified to small regions of the chromosome. In 2006, Cornell University researchers led by Daniel Barbash reported the first case of identification of a pair of interacting genes that result in incompatibility (Brideau et al., 2006). The Hybrid male rescue (Hmr) gene has a sequence that strongly suggests that it is a transcription factor, a gene whose protein affects the expression of other genes. The Lethal hybrid rescue (Lhr) gene produces a protein that is associated with heterochromatin, a region of the chromosome that is made up of repetitive DNA. Inviability results from the combination of the D. melanogaster allele of Hmr and the D. simulans allele of Lhr. Like Ods, the sequence data from Lhr and Hmr also suggest that positive natural selection has driven the divergence between species of these genes (Brideau et al., 2006). Identification of more such Dobzhansky-Muller pairs of genes in hybrid incompatibility is underway.

Recently, evolutionary geneticists have examined hybrid incompatibility as a consequence of divergence of regulatory genetic networks. Several of the hybrid incompatibility genes thus far characterized (like Ods and Hmr) have regulatory functions. Given that genes interact in regulatory genetic networks, these would seem natural places for Dobzhansky-Muller incompatibilities to arise. Theoretical models also suggest hybrid incompatibility may arise in geographically isolated populations when a trait determined by a regulatory genetic pathway is subjected to directional natural selection (selection to change trait value) (Johnson & Porter, 2000). Microarray and other techniques used to characterize gene expression have uncovered several genes that are misregulated (either overexpressed or, more frequently, underexpressed) in hybrids between species (Oritz-Barrientos et al., 2007). Such studies may be useful in uncovering interacting genes involved in hybrid incompatibility.

To date, most fine-scale genetic studies of hybrid incompatibility have been in Drosophila, and in particular, those species most closely related to D. melanogaster. In other groups of organisms, either the absence of genetic tools or the lack of closely related species has limited genetic studies. This is changing, especially with the advent of genomic tools. For instance, monkeyflowers in the genus Mimulus are rapidly becoming a model system to investigate the genetics of hybrid incompatibility in plants (Sweigart et al., 2006).