Zebrafish hox Clusters and Vertebrate Genome Evolution

See allHide authors and affiliations

Science  27 Nov 1998:
Vol. 282, Issue 5394, pp. 1711-1714
DOI: 10.1126/science.282.5394.1711


HOX genes specify cell fate in the anterior-posterior axis of animal embryos. Invertebrate chordates have one HOXcluster, but mammals have four, suggesting that cluster duplication facilitated the evolution of vertebrate body plans. This report shows that zebrafish have seven hox clusters. Phylogenetic analysis and genetic mapping suggest a chromosome doubling event, probably by whole genome duplication, after the divergence of ray-finned and lobe-finned fishes but before the teleost radiation. Thus, teleosts, the most species-rich group of vertebrates, appear to have more copies of these developmental regulatory genes than do mammals, despite less complexity in the anterior-posterior axis.

HOX cluster genes encode DNA binding proteins that specify fate along the anterior-posterior axis of bilaterian animals (1). Remarkably, the order of HOX genes along the chromosome reflects the order they act along the body (2). Invertebrate chordates have one HOX cluster and little axial diversity, but tetrapods have four clusters and substantial axial complexity (3). Tetrapod clusters arose by duplications of an ancestral cluster containing 13 genes (4). Although it is widely assumed that vertebrates have four HOXclusters, initial studies of teleost fish, the most diverse group of vertebrates, revealed unexpected HOX genes (5–8). To understand this problem, we isolatedhox clusters from the zebrafish Danio rerio.

To complement previous surveys of zebrafish hox gene fragments (7, 8), we identified genomic DNAs in P1 artificial chromosomes (PACs), using degenerate primers to amplify homeoboxes (9). We then identified overlapping PACs in chromosome walks, inventoried their hox gene content using redundant primers, sequenced gene coding regions, and analyzed gene phylogenies (10). These experiments identified sevenhox clusters containing 40 of the 41 previously identified zebrafish hox genes, seven new hox genes, onehox pseudogene, and evx1 (Fig. 1). Although we tried to find all genes in each cluster, it is possible that additional genes or pseudogenes exist that do not amplify with our primers.

Figure 1

Organization of vertebrateHOX clusters. Each horizontal thick line represents a cluster, designated by species abbreviation followed by cluster name. Species designations are as follows: zebrafish (Dre), black squares;Fugu (Fru), gray squares; mouse (Mmu), pale gray squares. Parts (A) to (D) display HOX clusters from different species. Clusters are organized from the 5′ end (paralogy group 13) to the 3′ end (paralogy group 1), with theeven-skipped homologs of the evx family at the 5′ end of the clusters. Clones from the PAC library (19) are shown above each zebrafish cluster. The known content of each PAC is represented by black or gray squares (genes) and open squares (pseudogenes). The orphan hox genes hoxx4,hoxx9, and hoxy6 (7, 8) are synonymous with hoxa4a, hoxa9a, and hoxc6b. Chromosome walks show that genes formerly thought to represent thehoxa cluster (7) are split into thehoxab and hoxbb clusters.

Phylogenetic analysis of sequence data (11) assigned zebrafish genes to one of 13 paralogy groups. Groups 4 and 9 appear in each mammalian cluster and in four zebrafish clusters, so we joined the nucleotide sequences of these groups, removed nonalignable sequence, and constructed a phylogenetic tree. The results showed (Fig. 2A) that each of these four clusters is orthologous to one of the four mammalian clusters. Hence, the duplication events that produced the four mammalian clusters occurred before the divergence of ray-finned and lobe-finned lineages about 420 million years ago (12).

Figure 2

Phylogenetic analysis. (A) The tree constructed by joining homeodomain sequences of group 4 and 9 genes shows that zebrafish (Dre) has orthologs of each human (Hsa) and mouse (Mmu) HOX cluster. (B) The group 6 tree shows that zebrafish has two copies of mammalian HOXB andHOXC clusters. Furthermore, Fugu (Fru)Hoxc6 is closely related to just one of the zebrafish genes, suggesting that duplication occurred before the divergence ofFugu and zebrafish lineages. This tree is rooted on the lamprey (Petromyzon marinus, Pma)hox-6w sequence. (Xla, Xenopus laevis;Nvi, Notophthalmus viridescens.) (C) The group 5 tree confirms cluster orthologies and duplications. (D) The tree constructed by joining homeodomains of groups 9, 11, and 13 shows that zebrafish has two orthologs of the mammalian HOXAcluster, and the Fugu Hoxd cluster branches withHOXA clusters of other vertebrates. Numbers at nodes indicate bootstrap values for 1000 runs.

Further analysis revealed the origin of the other three zebrafishhox clusters. The group 6 tree showed that zebrafish has two orthologs of mammalian HOXB6, calledhoxb6a and hoxb6b (Fig. 2B). The group 5 nucleotide tree confirmed duplicate hoxb clusters (Fig. 2C). Likewise, zebrafish has two orthologs of mammalian HOXC6, called hoxc6a and hoxc6b (Fig. 2B). To investigate HOXA clusters, we joined and aligned the homeodomains of groups 9, 11, and 13, which allows comparison with the pufferfish Fugu (for which only the amino acid sequence of the homeobox is available). This tree (Fig. 2D) shows that zebrafish has two clusters orthologous to the mammalian HOXA cluster. These data suggest that all hox clusters duplicated in the lineage that led to zebrafish after it diverged from the lineage that led to tetrapods, with subsequent loss of one hoxd cluster. The divergent Fugu Hoxd cluster (5) branches with high bootstrap value (965) with the HOXAclusters of other vertebrates (Fig. 2D). We conclude thatFugu has two orthologs of the tetrapod HOXAcluster and no described Hoxd cluster.

Comparative analysis of cluster content illuminates the history of HOX cluster duplication. The (AB)(CD) model (13) suggests two sequential duplications, giving a proto-AB cluster and a proto-CD cluster after the first event. The alternative (D(A(BC))) model (14) suggests three duplications, the first producing the D and proto-ABC clusters, the second giving the A and proto-BC clusters, and the third providing the Band C clusters. Cladistic analysis of cluster content favors the (AB)(CD) model (Fig. 3). For example, loss of group 12 is a shared derived characteristic of teleost and tetrapod HOXA and HOXB clusters, and loss of groups 2 and 7 unites HOXC and HOXD clusters. This model minimizes the number of convergent gene losses and is also independently supported by sequence analysis (Fig. 2C).

Figure 3

A cladistic model for the evolution of vertebrate HOX clusters, using gene presence as character states. Assuming that gene loss is more frequent than gene gain, the ancestral state (A) had 13 HOX genes plusEVX. Duplication, probably in an agnathan fish (B), gave a proto-(AB) cluster lacking group 12 and a proto-(CD) cluster lacking groups 2 and 7. A second duplication, perhaps in an ancient gnathostome fish (C), was followed by losses of group 8 from the HOXA cluster, group 11 from the HOXB cluster, EVX from theHOXC cluster, and groups 5 and 6 from the HOXDcluster; subsequently, the tetrapod lineages lost HOXC1,HOXC3, and an EVX gene from the HOXBcluster (D). Finally, an apparent duplication event produced eight clusters in a ray-finned fish (E), followed by further shared and unique losses in zebrafish (F) andFugu (G) lineages.

Superimposed on shared gene loss is lineage-specific loss. For example, fish have lost genes present in mammals (hoxa6,hoxa7, hoxd1, and hoxd8). Reciprocally, mammals have lost paralogs present in teleosts (hoxb10a and eve1). We conclude that the degeneration of HOX clusters continued in both lineages after the divergence of ray-finned fish and the lobe-finned ancestors of tetrapods. Furthermore, hox cluster degeneration may be ongoing, at least in fish, becausehoxc1a and hoxc3a are active in zebrafish but their orthologs are pseudogenes in Fugu(5) and are absent from mammals; likewise,hoxa10a is a pseudogene in zebrafish but has normal structure in Fugu and mouse.

When did the latest HOX cluster duplication occur in the zebrafish lineage? The pattern of shared gene loss suggests that the last common ancestor of zebrafish and Fugualready had duplicated HOX clusters (Fig. 3E). Gene phylogenies support this conclusion, because Fugu Hoxc-6, the only informative full-length sequence available (5), is more closely related to zebrafishhoxc6a than it is to hoxc6b (Fig. 2B). In addition, the presence of two HOXA clusters inFugu, one related to the zebrafish hoxaa and the other to the hoxab cluster (Fig. 1), supports a shared duplication. The presence in killifish (7) of five group 9 and four group 1 genes as in zebrafish, rather than four group 9 and three group 1 genes as in mammals, is consistent with the hypothesis that the killifish lineage also experienced an “extra” duplication event. This suggests that a fish-specificHOX cluster duplication occurred before the divergence ofFugu and zebrafish lineages more than 150 million years ago (15), but after the divergence of ray-finned and lobe-finned lineages. Goldfish, salmonids, and some other teleosts have experienced additional, more recent polyploidization events (16). Genomic analysis of basally branching ray-finned fish, such as sturgeons, Amia, orPolypterus, is necessary to clarify the timing of theHOX duplication event.

To determine whether “extra” fish hox clusters result from tandem duplication or chromosome duplication in fish, or cluster loss in tetrapods, we mapped zebrafish hox clusters; cloned, sequenced, and mapped four new genes whose orthologs are syntenic with HOX clusters in mammals (dhh,evx1, eng1b, and gli); and mapped four previously unmapped zebrafish genes [dlx5,dlx6, dlx8, and pl10a; see (11)] whose orthologs are linked to HOX clusters in mammals. These experiments showed that zebrafish has two copies of each HOX chromosome segment in mammals (Fig. 4). For example, the human and mouseHOXB chromosomes have six and four genes, respectively, whose apparent orthologs map on one of the two zebrafish chromosomes containing hoxba or hoxbb (Fig. 4). Each of these two chromosomes also has one copy of other duplicate genes, includingdlx7/dlx8, rara2a/rara2b, andhbae4/hbae1 (11, 17). We conclude that zebrafish has two copies of this mammalian chromosome segment. Because similar results were obtained for the other clusters (Fig. 4), we infer that hox cluster duplication in ray-finned fish occurred by whole chromosome duplication. Although we found a singlehoxd cluster in zebrafish, mapping experiments identified the predicted duplicate chromosome segments (Fig. 4), suggesting secondary loss of one hoxd duplicate.

Figure 4

HOX cluster duplication involved large chromosome segments. The diagram shows syntenic relationships amongHOX containing chromosomes of human (Hsa), mouse (Mmu), and zebrafish linkage groups (LG). Vertical gray lines indicate a group of genes on the same chromosome (syntenic loci), with order ignored to facilitate the comparison of orthologs and paralogs. Horizontal gray lines connect presumed orthologs within chromosome groups as well as paralogs between chromosome groups.

These results suggest two rounds of HOX chromosome duplication (probably whole genome duplication) before the divergence of ray-finned and lobe-finned fishes, and one more in ray-finned fish before the teleost radiation. Because gene duplicates often have a subset of the functions of the ancestral gene (18), mutations in duplicate genes may reveal essential functions that otherwise might remain hidden. For example, if a gene is essential for distinct early and late functions, a lethal mutation knocking out the early function might obscure the late function in a mutant mammal, but both functions would be evident if the two functions assort to different zebrafish gene duplicates. The conclusion that the genetic complexity of hox clusters in teleost fish has exceeded that of mammals for more than 100 million years calls into question the concept of a tight linkage of HOX cluster number and morphological complexity along the body axis. However, because teleosts are the most species-rich group of vertebrates and exhibit tremendous morphological diversity, it is tempting to speculate that the duplication event detected here may have provided gene copies that helped spur the teleost radiation.

  • * The experiments reported here were conducted in the laboratories of these authors.


View Abstract

Stay Connected to Science

Navigate This Article