High prevalence of Prdm9-independent recombination hotspots in placental mammals

In many mammals, recombination events are concentrated into hotspots directed by a sequence specific DNA-binding protein named Prdm9. This protein facilitates chromosome pairing and its inactivation has been shown to induce fertility losses in mice and rats. Intriguingly, Prdm9 has been lost several times in vertebrates, and notably among mammals, it has been pseudogenized in the ancestor of canids (dogs, wolves foxes). When this gene is inactive, either naturally in dogs, or through knock-out experiments in mice, recombination hotspots still exist, but they tend to occur in promoter-like features such as CpG islands. It has thus been proposed that one role of Prdm9 could be to direct recombination away from those Prdm9-independent hotspots. However, the ability of Prdm9 to direct recombination hotspots has been assessed only in a handful of species, and a clear picture of how much recombination occurs outside of Prdm9-directed hotspots in mammals is still lacking. In this study, we derived an estimator of past recombination activity based on signatures of GC-biased gene conversion in substitution patterns. We applied it to quantify recombination activity in Prdm9-independent hotspots in 52 species of boreoeutherian mammals. We observed a wide range of recombination rate at these loci: several species (such as mice, humans, some felids or cetaceans) show a deficit of recombination, while a majority of mammals display a clear peak of recombination. Our results demonstrate that Prdm9-directed and Prdm9-independent hotspots can co-exist in mammals, and that their co-existence seem to be the rule rather than an exception.