Revisiting recent challenges to the ancient fish-specific genome duplication hypothesis

The discovery of seven unlinked Hox gene clusters in zebrafish (Danio rerio) and two HoxA gene clusters in the pufferfish (Takifugu rubripes) led to the hypothesis that a genome duplication event occurred early during the evolution of ray-finned fishes (Actinopterygii) [1]. Shortly thereafter, Wittbrodt and co-workers [2] uncovered many additional examples of genes that appeared to have been duplicated in fish. The fish-specific genome duplication hypothesis was further supported by the discovery of synteny within the zebrafish genome (i.e. the observation that many duplicated zebrafish genes map to the same pairs of chromosomes) [3,4] and by the discovery of seven unlinked Hox gene clusters in the Japanese medaka (Oryzias latipes) [5]. However, in several recent papers Robinson-Rechavi et al. [6–8] have argued that an ancestral whole-genome duplication event might not be responsible for the abundance of duplicated fish genes. These authors counted orthologous genes in fish and mouse and, where extra genes were found in fish, compared the number of gene duplications occurring in a single fish lineage to the number of gene duplications shared by more than one lineage [6,7]. They found that most mouse genes surveyed occurred only once in fish. Duplicated fish genes were detected, but most were the products of lineage-specific duplication events and not an ancient duplication event. Here we discuss three major problems with the approach used by Robinson-Rechavi et al. First, to test the ancient fish-specific genome duplication event, only genes that were available in the Hovergen database [9] from mouse and at least three major fish lineages (i.e. orders) were employed. Few genes met this three-lineage criterion: 33 in one study [6] and a mostly overlapping set of 37 genes in another [7]. The authors argue that their analyses of this short list of genes provide very little evidence for the ancient fish-specific genome duplication hypothesis: in these phylogenetic analyses only 7 gene families out of 37 (19%) follow a pattern consistent with an ancestral whole genome duplication origin. On the other hand, for 11 (30%), all detected duplications happened after the divergence of fish lineages. Finally for 19 gene families (51%) no duplication was observed among fish [7]. However, the failure to find many examples of gene duplicates shared by at least two lineages (i.e. 'lineage-shared' gene duplicates) in a very limited dataset should not be considered evidence against an ancient genome duplication event. As Robinson-Rechavi et al. [6] point out, …