Vertebrates have greatly elaborated the basic chordate body plan and evolved highly distinctive genomes that have been sculpted by two whole-genome duplications. Here we sequence the genome of the Mediterranean amphioxus ( Branchiostoma lanceolatum ) and characterize DNA methylation, chromatin accessibility, histone modifications and transcriptomes across multiple developmental stages and adult tissues to investigate the evolution of the regulation of the chordate genome. Comparisons with vertebrates identify an intermediate stage in the evolution of differentially methylated enhancers, and a high conservation of gene expression and its cis -regulatory logic between amphioxus and vertebrates that occurs maximally at an earlier mid-embryonic phylotypic period. We analyse regulatory evolution after whole-genome duplications, and find that—in vertebrates—over 80% of broadly expressed gene families with multiple paralogues derived from whole-genome duplications have members that restricted their ancestral expression, and underwent specialization rather than subfunctionalization. Counter-intuitively, paralogues that restricted their expression increased the complexity of their regulatory landscapes. These data pave the way for a better understanding of the regulatory principles that underlie key vertebrate innovations.

All vertebrate brains develop following a common Bauplan defined by anteroposterior (AP) and dorsoventral (DV) subdivisions, characterized by largely conserved differential expression of gene markers. However, it is still unclear how this Bauplan originated during evolution. We studied the relative expression of 48 genes with key roles in vertebrate neural patterning in a representative amphioxus embryonic stage. Unlike nonchordates, amphioxus develops its central nervous system (CNS) from a neural plate that is homologous to that of vertebrates, allowing direct topological comparisons. The resulting genoarchitectonic model revealed that the amphioxus incipient neural tube is unexpectedly complex, consisting of several AP and DV molecular partitions. Strikingly, comparison with vertebrates indicates that the vertebrate thalamus, pretectum, and midbrain domains jointly correspond to a single amphioxus region, which we termed Di-Mesencephalic primordium (DiMes). This suggests that these domains have a common developmental and evolutionary origin, as supported by functional experiments manipulating secondary organizers in zebrafish and mice.

The vertebrate circulatory system is the most complex vascular system among those of metazoans, with key innovations including a multi-chambered heart and highly specialized blood cells. Invertebrate vessels, on the other hand, consist of hemal spaces between the basal laminae of epithelia. How the evolutionary transition from an invertebrate-type system to the complex vertebrate one occurred is, however, poorly understood. We investigate here the development of the cardiovascular system of the cephalochordate amphioxus Branchiostoma lanceolatum in order to gain insight into the origin of the vertebrate cardiovascular system. The cardiac markers Hand, Csx (Nkx2-5) and Tbx4/5 reveal a broad cardiac-like domain in amphioxus; such a decentralized organization during development parallels that seen in the adult anatomy. Our data therefore support the hypothesis that amphioxus never possessed a proper heart, even transiently during development. We also define a putative hematopoietic domain, supported by the expression of the hematopoietic markers Scl and Pdvegfr. We show that this area is closed to the dorsal aorta anlages, partially linked to excretory tissues, and that its development is regulated by retinoic acid, thus recalling the aorta-gonads-mesonephros (AGM) area of vertebrates. This region probably produces Pdvegfr+ hemal cells, with an important role in amphioxus vessel formation, since treatments with an inhibitor of PDGFR/VEGFR lead to a decrease of Laminin in the basal laminae of developing vessels. Our results point to a chordate origin of hematopoiesis in an AGM-like area from where hemal Pdvegfr+ cells are produced. These Pdvegfr+ cells probably resemble the ancestral chordate blood cells from which the vertebrate endothelium later originated.

Epithelial-mesenchymal interactions are crucial for the development of numerous animal structures. Thus, unraveling how molecular tools are recruited in different lineages to control interplays between these tissues is key to understanding morphogenetic evolution. Here, we study Esrp genes, which regulate extensive splicing programs and are essential for mammalian organogenesis. We find that Esrp homologs have been independently recruited for the development of multiple structures across deuterostomes. Although Esrp is involved in a wide variety of ontogenetic processes, our results suggest ancient roles in non-neural ectoderm and regulating specific mesenchymal-to-epithelial transitions in deuterostome ancestors. However, consistent with the extensive rewiring of Esrp-dependent splicing programs between phyla, most developmental defects observed in vertebrate mutants are related to other types of morphogenetic processes. This is likely connected to the origin of an event in Fgfr, which was recruited as an Esrp target in stem chordates and subsequently co-opted into the development of many novel traits in vertebrates.

Homologous long non-coding RNAs (lncRNAs) are elusive to identify by sequence similarity due to their fast-evolutionary rate. Here we develop LincOFinder, a pipeline that finds conserved intergenic lncRNAs (lincRNAs) between distant related species by means of microsynteny analyses. Using this tool, we have identified 16 bona fide homologous lincRNAs between the amphioxus and human genomes. We characterized and compared in amphioxus and Xenopus the expression domain of one of them, Hotairm1, located in the anterior part of the Hox cluster. In addition, we analyzed the function of this lincRNA in Xenopus, showing that its disruption produces a severe headless phenotype, most probably by interfering with the regulation of the Hox cluster. Our results strongly suggest that this lincRNA has probably been regulating the Hox cluster since the early origin of chordates. Our work pioneers the use of syntenic searches to identify non-coding genes over long evolutionary distances and helps to further understand lncRNA evolution.