We are interested in genetic mechanisms involved in the origins and evolution of chordates [1]. Chordates comprise three taxa, cephalochordates, urochordates or tunicates, and vertebrates [2]. These are thought to have originated from a common ancestor of the deuterostomes, together with ambulacrarians, a clade containing echinoderms and hemichordates. The organ that best characterizes chordates is the notochord, an organ that supports the beating of the muscular tail of fish-like larvae or adults [1, 3, 4]. The T-box family transcription-factor gene, Brachyury, plays an essential role in notochord formation [5]. In ascidians (urochordates), for example, Brachyury is expressed exclusively in primordial embryonic notochord cells [6, 7]. Loss of Brachyury function results in the failure of notochord formation, while its ectopic expression induces endoderm cells to become notochord cells [8, 9].

Interestingly, Brachyury is not specific to chordates, but is present in most metazoans, including non-chordate deuterostomes [5, 10]. Beside paralogs derived from ancient whole genome duplication (referred to as “ohnologs”), Brachyury is usually present as a single copy, with some exceptions (Hydra [11], calcisponge [12], and cephalochordates (see below)), and is expressed around the blastopore during gastrulation. A recent study by our group using coral embryos demonstrated that the evolutionarily conserved function of Brachyury is associated with formation of blastopore-derived organs, such as the pharynx of coral embryos [13]. In non-chordate deuterostomes, ambulacrarians, Brachyury is expressed in the archenteron invagination region in early gastrulae and in the stomodeum invagination region in later stage embryos [14, 15]. Although an overexpression experiment with sea urchin Brachyury suggested its role in gastrulation [16], the developmental role of ambulacrarian Brachyury remains to be determined.

We proposed an evolutionary scenario for Brachyury emphasizing its primary and secondary expression domains and functions [5]. Namely, ambulacrarians require Brachyury in its primary domain of expression and function associated with the blastopore (PEF), while chordates employ the gene, not only for PEF, but also for the secondary domain of expression and function associated with the notochord (SEF). In urochordates, the PEF was likely lost due to the precocious mode of embryogenesis [1, 5]. For the past decade, evolutionary developmental biologists have been asking how the chordate ancestor acquired the Brachyury SEF. Answering this question is critical to our understanding of genetic and molecular mechanisms involved in the origins of chordates. Genomes of the cephalochordates, Branchiostoma floridae and B. belcheri, each contain a set of duplicated Brachyury, Amphi-Bra1 and Amphi-Bra2 [17, 18]. These duplicated genes have both PEF and SEF, but the expression domains of each have not been determined [17, 19]. In contrast, urochordates have only a single copy [20]. Most vertebrates also have a single copy of Brachyury/T, although they have variable numbers of T ohnologs, which arose from the two rounds of genome-wide gene duplication (2R-GWGD) that occurred in this lineage [21, 22]. In mice, only a T gene exhibits PEF and SEF during early embryogenesis [22].

The occurrence of two domains of Brachyury expression and function may be explained by one of two alternative evolutionary scenarios (Fig. 1). In scenario 1, Brachyury was present as a single copy in deuterostome ancestors, while it became duplicated in an ancestor of chordates, such that the original gene retained PEF and its newly formed counterpart obtained SEF (Fig. 1a). The cephalochordate lineage retained this arrangement, whereas the urochordate and vertebrate lineages lost one of duplicates. In scenario 2 (Fig. 1b), chordates retained a single copy of Brachyury as did the non-chordate invertebrates. Duplication occurred in the lineage leading to cephalochordates, but not in the lineage leading to olfactores (urochordates + vertebrates). Evolution of vertebrate T-box family genes seems to have been complicated by the 2R-GWGD, and vertebrate Brachyury/T needs to be examined more carefully in the future (Yasuoka et al., in preparation).

In 2015, the genomes of two hemichordate species, an indirectly developing acorn worm, Ptychodera lava, and a directly developing acorn worm, Saccoglossus kowalevskii, were decoded [23]. Genomes of animals belonging to each of the four other phyla were already decoded prior to 2014; echinoderm sea urchin (Strongylocentrotus purpuratus; [24]) and starfish (Acanthaster planci [25]), cephalochordate amphioxus (Branchiostoma floridae [26] and Branchiostoma belcheri [27]), urochordates (e.g., Ciona intestinalis [20] and Oikopleura dioica [28]) and vertebrates (e.g., Homo sapiens [29]). We obtained genomic information for all five phyla of deuterostomes, enabling us to examine which of the two scenarios outlined above better explains the genetic and molecular evolution of Brachyury in relation to chordate origins.

Gene trees were estimated with an analytical pipeline implementing BLAST search and the maximum likelihood method (modified from Inoue et al. [30]).

We examined Brachyury of representative protostomes and deuterostomes, the genomes of which have been decoded (Table 1, with Brachyury gene/protein ID and information from the genomes). Phylogenetic relationships of deuterostome species are based on previous studies (e.g., [23]).

Copy number of Brachyury in deuterostomes and the chordate ancestor

Molecular cloning of mouse Brachyury was followed by identifying its orthologs in various metazoans (reviewed by [5, 40]). In addition, cloning of other T-box-containing transcription factor genes, including T-brain, Tbx1, Tbox2, and Tbx6, shows that they form a family called the T-box family (e.g., [21]). Although Brachyury has been identified and characterized in each bilaterian with a decoded genome, we examined Brachyury copy numbers in 18 species, including six protostomes and 12 deuterostomes (Table 1). To evaluate the two scenarios (Fig. 1), we did not include ohnologs in vertebrates as the copy number of Brachyury because that duplication is not associated with the acquisition of SEF.

To this end, we first carefully identified all T-box containing genes from the decoded genomes. In order to identify a candidate query sequence from non-vertebrate chordate lineages, using the human Brachyury amino acid sequence as a query, phylogenetic relationships were roughly estimated to delineate a clade comprising the Brachyury subfamily. Using queries of human and cephalochordate Brachyury amino acid sequences (Additional file 1), we next searched for the Brachyury gene in selected bilaterian genomes by estimating phylogenetic relationships (Additional file 2). Although no ortholog was found in the present genome assembly of the nematode, Caenorhabditis elegans, a single copy of Brachyury was found in each of the other bilaterians in both resultant trees based on nucleotide and amino acid datasets. An exception was that lancelets (cephalochordates) possessed two copies in their genome, as shown previously [17, 18].

In order to estimate the copy number of Brachyury in the chordate ancestor, we reconstructed phylogenetic relationships of Brachyury using only Brachyury gene sequences (Additional file 3) selected from the estimated tree of T-box containing genes (Additional file 2). Estimated gene trees based on the comparison of nuclear (Fig. 2) and amino acid (Additional file 4) sequences produced the same deuterostome relationships, except for the positions of the urochordate, Botryllus schlosseri and a clade consisting of the remaining urochordates. For subsequent discussion, we used the tree obtained from DNA analysis because of the longer sequences. The Brachyury tree (Fig. 2) differed from the species tree (e.g., [23]) in that, in the Brachyury tree, urochordate genes formed a sister clade with all remaining deuterostome genes. This may be the result of a faster evolutionary rate of urochordate proteins than in other deuterostome taxa. Our results (Fig. 2) indicate that, with bootstrap support of 97%, Amphi-Bra1 and Amphi-Bra2 of the two species of cephalochordates forms a monophyletic group, consistent with our previous study [19].

Given that cephalochordates are the only deuterostome taxon with the duplicated state of Brachyury, and that the two copies form a monophyletic clade (Fig. 2), it is highly likely that the duplication of Brachyury was specific to the cephalochordate lineage, and did not occur in the chordate ancestor. Namely Brachyury was present in a single copy in the chordate ancestor. Therefore, our results support the second of the two scenarios described above (Fig. 1b). In vertebrates (chicken and human, here), Brachyury/T forms a clade while Tbx19 forms another clade (Fig. 2). This suggests that 2R-GWGD resulted in divergence of the ancestral Brachyury into a clade including Brachyury or Tbx19.

Genomic organization of Brachyury in deuterostomes

Next, to gain a better understanding of the evolutionary changes in the genetic and genomic organization of Brachyury in relation to chordate evolution, we examined the synteny of genes in Brachyury-containing genomic regions, especially in deuterostome taxa. Although global synteny analyses have shown comparable and conserved synteny between cephalochordate and vertebrate genomes [26] and between hemichordate and cephalochordate genomes [23], no detailed analyses were carried out on genomic regions that contain Brachyury.

In the genomes of two cephalochordate species, B. floridae and B. belcheri, the two Brachyury were tandemly aligned (Fig. 3). In both genomes, although a neighboring NOTUM-like gene is present syntenically, synteny of neighboring genes is limited to the four genes. This suggests that the two copies of Brachyury arose from a tandem duplication, not from an ancient segmental duplication.

Beside the trait of tandem duplication of cephalochordate Brachyury, our synteny analysis characterized conserved synteny around Brachyury of major deuterostome lineages (Fig. 3). A global view of synteny indicates that it is comparable among ambulacrarian species, because genes with comparable sequences were found in echinoderms and an acorn worm. On the contrary, no conserved synteny was found among four species of tunicates. Although the synteny analysis among vertebrate genomes was complicated by 2R-GWGD, it is evident that there are no scaffolds with duplicated Brachyury/T or its ohnolog, Tbx19, in chicken or human genomes. Detailed synteny of the Brachyury-containing genomic region is discussed in Additional file 5.

Brachyury is thought to be the most ancient T-box family member (e.g., [21, 22]). It has been speculated that during their histories of over 500 million years, each of the five deuterostome taxa altered the genetic and genomic constitution of Brachyury from its original forms, rending analysis of the present question of Brachyury evolution more difficult than that of more recently evolved genes. Nevertheless, the tandem duplicated state of Brachyury is found only in cephalochordates. Recently, another example of tandem duplication of chordate Brachyury has been reported in an amphibian, Xenopus tropicalis [41]. It thus appears likely that cases of Brachyury duplication [1] are exceptional. As previously described, Brachyury has two domains of expression and function. The primary domain of expression is in the blastopore during gastrulation (PEF), which is shared by all metazoans [5, 13]. The secondary domain is associated with the notochord (SEF), and is specific to chordates. It is tempting to speculate that one is the original Brachyury gene with its original function and the other is a new copy with a secondary function, and that the duplication occurred very early in chordate evolution. However, as shown in the present study, Brachyury was present as a single copy in the common ancestor of chordates and a single copy of Brachyury acquired the secondary function, SEF, in the chordate ancestor.

The presence of a single copy of Brachyury in the chordate ancestor indicates that the gene duplication was not associated with the acquisition of SEF leading to the development of the notochord. Considering that copy numbers of Brachyury vary between cephalochordates and other deuterostome lineages, the question of regulatory mechanisms still remains. Intensive analyses have been carried out to elucidate 5′ upstream sequences or modules that regulate PEF and SEF, respectively [1]. Therefore, we have to keep the two cases, namely a single copy with two domains and two copies with different domains, in mind for future studies of Brachyury in relation to regulatory mechanisms.