The hagfish genome and the evolution of vertebrates

Abstract

As the only surviving lineages of jawless fishes, hagfishes and lampreys provide a critical window into early vertebrate evolution. Here, we investigate the complex history, timing, and functional role of genome-wide duplications in vertebrates in the light of a chromosome-scale genome of the brown hagfish Eptatretus atami. Using robust chromosome-scale (paralogon-based) phylogenetic methods, we confirm the monophyly of cyclostomes, document an auto-tetraploidization (1R_V) that predated the origin of crown group vertebrates ~517 Mya, and establish the timing of subsequent independent duplications in the gnathostome and cyclostome lineages. Some 1R_V gene duplications can be linked to key vertebrate innovations, suggesting that this early genomewide event contributed to the emergence of pan-vertebrate features such as neural crest. The hagfish karyotype is derived by numerous fusions relative to the ancestral cyclostome arrangement preserved by lampreys. These genomic changes were accompanied by the loss of genes essential for organ systems (eyes, osteoclast) that are absent in hagfish, accounting in part for the simplification of the hagfish body plan; other gene family expansions account for hagfishes' capacity to produce slime. Finally, we characterise programmed DNA elimination in somatic cells of hagfish, identifying protein-coding and repetitive elements that are deleted during development. As in lampreys, the elimination of these genes provides a mechanism for resolving genetic conflict between soma and germline by repressing germline/pluripotency functions. Reconstruction of the early genomic history of vertebrates provides a framework for further exploration of vertebrate novelties.

Figure 1.. Phylogenetic position of hagfish and multigene support for the cyclostome hypothesis.

a, Picture of a brown hagfish, Eptatretus atami (photo credit K. Kubokawa). b, Distinct hypothesis regarding early vertebrate evolution. c, Phylogenetic reconstruction of deuterostome relationships using 176 genes and 61,939 positions selected as least saturated using a site-heterogeneous model (CAT+GTR). topology is robust to composition heterogeneity and similar to what was obtained for all 1,467 genes using site-homogeneous models (Figure S3).

Figure 2.. Genomic and syntenic architecture in cyclostomes and vertebrates.

a, Distribution of genes derived from the 18 ancestral chordate linkage groups (CLGs) in the chromosomes of hagfish, lamprey and gar chromosomes. Bins contain 20 genes and only CLGs from chromosomes showing reciprocal enrichment are plotted (fisher’s test). b, Syntenic relationship between lamprey and hagfish chromosomes showing that hagfish chromosomes are typically fusions of multiple lamprey chromosomes. Each line corresponds to a single-copy orthologue labelled by CLGs. CLG colours are the same as in a.

Figure 3.. History of genome duplications in vertebrates.

a, Alternative scenarios for whole genome duplications during early vertebrate evolution tested using WHALE with the scenario and duplication nodes receiving statistical support highlighted (see Figure S4). b, Evolutionary history of vertebrate Hox gene clusters determined by paralogon phylogeny (see also panel d). c, Timetree of deuterostomes combining results from paralogon-based molecular dating analyses. Distribution of divergence times for speciation (grey) and duplication (coloured as in a) nodes inferred for the 17 CLGs is shown at each node (Table S6). Each node is labelled with the median divergence time across CLGs and individual dates for distinct CLGs are plotted as dots with as an overlay the distribution of inferred divergence time for all CLGs. d, Example of topology and divergence times obtained using segments derived from the Hox bearing CLG B. Species and datasets used are listed in Table S5. Vertebrate species: X. tropicalis, western clawed frog; G. gallus, chicken; L. oculeatus, spotted gar; C. punctatus, brownbanded bamboo shark. Outgroups include two amphioxus (B. floridae, B. lanceolatum), hemichordates (S. kowalevskii and P. flava), and one echinoderm (S. purpuratus).

Figure 4.. Functional impact of vertebrate WGD and gene loss in vertebrates.

a, Key neural crest-related genes classified according to their paralogy status in regarding to the 1R_V. b, Enrichment of functional annotation terms (Gene Ontology) in sets of genes showing a specific pattern of retention after vertebrate WGDs. c, Subfunctionalization of paralogous genes evaluated by counting expression pattern gain-and-loss. d, Gene family loss in deuterostomes highlighting the severe loss in the hagfish lineage. Species abbreviation is described in Table S5. e, Functional enrichment (GO) in lamprey for genes lost in the hagfish lineages highlighting a simplification of visual and hormonal systems. f, Structure and gene expression for the two clusters of alpha-keratin genes expressed in the slime gland and the skin with RNA-seq signal in blue (Figure S8).

Figure 5.. Germline-specific/enriched sequences and genes in hagfish.

a, Comparison between sequence depth of germline tissue (testes) vs. somatic tissue (blood) identifies a large number of genomic intervals with evidence for strong enrichment in germline. b, Genes encoded within germline-specific regions are enriched for several ontologies related to regulation of cell cycle and cell motility (Panther Biological Processes: most specific subclass shown). c, Degree of germline enrichment and estimated span of all predicted repetitive elements, focusing on elements with a cumulative span of <4Mb (per family member). Previously identified elements (Kojima et al., 2010; Kubota et al., 1993) are highlighted by coloured circles and newly identified high-copy elements are highlighted by coloured diamonds. Additional higher copy repeats are visible in Figure S9. d, Estimated cumulative span of the eight most highly abundant repeats shown as percentage of the genome covered. Colouring scheme is the same as in panels a and b. e, FISH hybridization of high copy germline-specific repeats to a testes metaphase plate showing their spatial clustering (blue counterstaining is NucBlue: Hoechst 33342; individual pairs of probes are shown in Figure S11).