Human Inversion Under Selection
Chromosomal inversions have been studied for nearly a century by geneticists and cytologists. Early researchers could visualize inversions in Drosophila via polytene chromosomes, and Sturtevant (1926) realized that inversions suppress recombination in laboratory experiments. This helped Dobzhansky and others develop the theory of coadaptation to explain how inversions can maintain favorable combinations of alleles in cis (i.e., on the same chromosome). There has also been extensive work done on inversions and sex ratio biases in Drosophila -- from both a molecular and evolutionary perspective. Recent work on speciation has shed light on the role inversions can play in patterns of nucleotide diversity between diverging taxa as well as actively encouraging speciation by suppressing recombination in heterokaryotypic individuals.
I do not want to give the impression that polymorphic genome rearrangements are a quirk relegated to Drosophila and research on these chromosomal rearrangements does not have implications outside of Dipterans. It just so happens that a majority of the work was performed on these convenient laboratory creatures, but research in marsupials, mice, and now human (see here and here for reviews) suggests that chromosomal rearrangements segregate in natural populations of mammals as well. The question to ask is "What role do these inversions and Robertsonian fusions play in evolution?" Recent work on chromosomal inversions and human and chimp speciation has been inconclusive (see here and here for critiques and here for a rebuttal); in my opinion, the main limitation is that the model assumes sympatry or parapatry, and we have no conclusive evidence that humans and chimps speciated with gene flow.
New findings by Stefansson et al show that natural selection may act on human inversion polymorphisms and shape their frequencies in natural populations. They discovered this inversion polymorphism by accident when they were examining a gene associated with Parkinson disease. By comparing the sequence from multiple individuals in the region surrounding the gene, they were able to determine that one of the haplotypes was inverted relative to the other. (There is also a duplication associated with the inversion.)
Inversion polymorphisms are expected to maintain different alleles in different arrangements because the inversions suppress recombination. This can be seen in the phylogenies created using sequence from within the inverted region.
H1 is the wild-type arrangement, and H2 is the inverted arrangement. As you can see, in both phylogenies, the H2 allele is found outside of the H1 clade. This suggests that the two groups (H1 and H2) diverged a long time ago and that the inversion is fairly ancient. In fact, the authors date the inversion event to 3 million years ago -- half as long ago as the divergence between humans and chimps.
The authors also show that individuals carrying the inversion have more children than wild-type parents. They say that this may be due to increased recombination in mothers who carry the inversion. This may be confusing, since I told you that inversions suppress recombination. Inversions only suppress recombination within the inversion and in the area surrounding the inversion. In this study, the researchers show that females who carry the inversion have an elevated recombination rate in the rest of their genome (the inverted region is fairly small so it probably plays a minor role in suppressing recombination).
If this inversion increases fitness (by increasing fecundity), shouldn't it be found throughout the world? This is especially true if it is as old as the authors claim it is. The authors found that it is common in Europe, but rare in the rest of the world:
The black wedges on the pie charts indicate the approximate frequency of the inversion in each of the populations sampled. One possible explanation is that the inversion is under balancing selection (natural selection favoring heterozygotes), which leads to a stable polymorphism. Another explanation is that the inversion only confers a fitness benefit in the European population, but it is neutral or deleterious in the rest of the world.
If this allele is beneficial (as the authors claim) and has been under directional selection due to the fitness effect it confers on its carriers, the estimates of divergence time may be flawed. The authors assumed a neutral process when calculating divergence times. Directional selection leads to an elevated rate of evolution, and, in turn, an overestimate of divergence times. Given the data the authors present, it seems highly unlikely that this inversion really is 3 million years old. It is more probable that natural selection has acted on this recent inversion event causing it to increase in frequency in European populations. This would explain the low frequency of the allele in the rest of the world and the high frequency within Europe as well as the ancient divergence time estimate (an overestimate).
The error in the paper is minute, and this finding is still quite important. It could very well be possible that this human inversion maintains alleles that interact favorably, much like what has been Drosophila inversions. It is difficult to detect polymorphic inversions in mammals because they do not produce polytene chromosomes. More inversions may be found through rigorous examination of particular loci. Through detailed comparisons of sequence data from different inversion types, we may be able to locate signatures of selection at the sequence level (decreased polymorphism, excess non-synonymous substitutions, elevated linkage disequilibrium) and test whether Dobzhansky's theory of coadaptation applies to taxa outside of Drosophila, such as mammals.