Ancestral mesodermal reorganization and evolution of the vertebrate head

How the highly complex vertebrate head—composed of brain, head muscles, and skull—evolved from non-vertebrate ancestors is a fundamental question in current evolutionary and developmental biology [1–3]. Recent comparative studies of the cephalochordate amphioxus and vertebrates suggest that a region homologous to the vertebrate fore/mid/hind brain is also present in the rostral part of the central nervous system (CNS) of amphioxus [4–6]. Amphioxus is a basal chordate that has somites extending to the rostral end of the body, and is considered the best proxy for understanding the origin of the vertebrate body plan [7].

The homology between amphioxus and the vertebrate CNS indicates that the unsegmented vertebrate head mesoderm evolved directly from the rostral somites of amphioxus [8]. Expression of en in the rostral somites of amphioxus (Branchiostoma floridae) and en2 in the ventral part of the mandibular head mesoderm of shark (Scyliorhinus torazame) embryos supports this hypothesis [9, 10]. However, Bfpax3/7, a homologue of pax3 that serves as a somite marker in vertebrates, is expressed in the rostral somites, suggesting that the vertebrate head mesoderm did not evolve by simple modification of rostral somites of an amphioxus-like ancestor, but rather by fundamental reorganization that occurred in the dorsal mesoderm [2, 10].

During embryogenesis, the vertebrate head mesoderm derived from the rostral part of the dorsal mesoderm is regionalized along the A/P axis by a gradient of Wnt/?-catenin signalling [11, 12]. In the regionalization of the dorsal mesoderm, downstream regional marker genes of Wnt/?-catenin signalling are expressed in the progenitor domains; gsc is expressed in the head mesoderm, whereas bra is expressed in the presumptive notochord during the late-gastrula stage [13, 14]. Additionally, in the trunk mesoderm, delta expression is detected in the presumptive somite region [15, 16]. Previous functional studies have shown that overexpression of Xenopus laevis dkk1 (negative regulator of Wnt/?-catenin signalling) expands the gsc expression domain posteriorly in Xenopus embryos, whereas bra expression is activated by Wnt/?-catenin signalling [11, 17, 18]. Additionally, delta has an essential role in somitogenesis, and is under the control of Wnt/?-catenin signalling [19].

In amphioxus, gsc and bra are co-expressed in the presumptive notochordal region at the gastrula stage [20, 21]. The presumptive somite marker delta is expressed in the first and second somites in the late-gastrula stage [22]. Loss of gsc expression in the notochord and gain of gsc expression in the head mesoderm of vertebrates compared with amphioxus indicates that A/P re-arrangement of mesodermal gene expression occurred in the lineage of vertebrates. Excessive Wnt/?-catenin signalling in amphioxus embryos induced by inhibition of GSK-3?/? does not affect the expression of regional marker genes of the dorsal mesoderm, such as bra and fgf8/17/18, during the gastrula stage [23]. This suggests that, unlike in vertebrates, Wnt/?-catenin signalling does not play a role in dorsal mesoderm regionalization in amphioxus. If vertebrate embryos did evolve a rearrangement of gene expression in the dorsal mesoderm to generate the head mesodermal region, what was the key developmental event in this process? We consider that rearrangement of gene expression in the vertebrate dorsal mesoderm from an ancestral chordate evolved through a novel mesodermal cell movement present in vertebrates.

Amphioxus gastrulation occurs through simple invagination, with little mesodermal involution of the outer layer [24], whereas in vertebrates, an overt involution takes place, as observed in Xenopus and lamprey [25–27]. Thus, uniquely in amphioxus and distinct from the case in vertebrates, there is nearly no change in the relative positions of the ectoderm and mesoderm. However, it remains unclear how mesodermal involution affects A/P patterning of the dorsal mesoderm and how this change has led to two distinct types of rostral mesoderm in amphioxus and vertebrates.

To explore the molecular background of vertebrate head mesoderm evolution, we first investigated the developmental stage at which the overall molecular topography of the dorsal mesoderm becomes distinctly different between amphioxus and vertebrates. We also examined whether mesodermal involution is important for dorsal mesoderm regionalization in vertebrate embryos, as a possible developmental factor that gave rise to vertebrate head mesoderm. Finally, we examined whether the genetic program for mesodermal involution is present in amphioxus. We hypothesize that vertebrate head mesoderm evolved from an amphioxus-like ancestral mesoderm through anteroposterior reorganization of the genetic developmental architecture.