Aside from the rostralmost part of the craniofacial region and the central nervous system, the vertebrate embryonic body consists of two major, conspicuously different domains: the pharyngeal arches and the trunk. The trunk is characterized by the presence of segmented paraxial mesodermal blocks, or somites, as well as the lateral plate–derived coelomic cavity. The outer wall of the cavity, which is called the lateral body wall, is derived from the somatopleure. The lateral body wall is occupied by hypaxial muscles that are primarily derived from somites. However, some somitic muscles are located neither in the paraxial domain nor in the lateral body wall; limb and hypobranchial muscles are in this category. Being derived from occipital or rostral myotomes, the hypobranchial muscle precursors migrate along the interface between the pharynx and body cavity (pericardium), together with the hypoglossal nerve anlage, to reach the oral floor in jawed vertebrates.
This embryonic pattern has been described for various vertebrate species [38-51]. In the lamprey, similar muscle precursors arise from rostral somites, once migrate caudally and ventrally along the caudal end of the pharynx in the rostralmost part of the body wall, to grow rostrally to reach the pharyngeal wall [25,28]. Thus, the myotomal muscle precursors nor the hypoglossal axons (somatic elements) do not typically enter into the pharyngeal arches [17,52]. The pharynx and trunk therefore not only stand out conspicuously in their morphological features, but also represent distinct developmental modules, with distinct developmental environments that favor specific sets of morphological elements. This is not unique; after all, the anatomical modules known as “visceral” or “somatic” have their embryonic backgrounds, as first recognized by van Wijhe [53].
The above distinction is clearly represented by the morphological patterns of a subset of cranial nerves distributed in the pharyngeal arches. These nerves, also known as “branchiomeric nerves” (cranial nerves V, VII, IX, and X) are primarily associated with pharyngeal arches, whereas spinal nerves belong to the trunk, exhibiting a metameric pattern in alignment with that of somites. The morphological pattern of the branchiomeric nerves is primarily characterized by their lateral position, determined by the position of the epibranchial placodes as well as the dorsolateral migratory pathway of the cephalic crest cells that prefigure the proximal nerve roots (reviewed by [17]). The dorsal root ganglion of the spinal nerve, on the other hand, is patterned more medially, medial to the dermomyotome. At the head-trunk interface, the mediolateral relationship between the vagus and hypoglossal nerves is reversed and the hypoglossal nerve comes towards the surface, whereas the pathway of the vagus switches from lateral to medial, growing caudally within the medial body wall along the esophagus [17,49]. This anatomical relationship is recapitulated in the lamprey, implying that this pattern is very ancestral, possibly dating back at least to the latest common ancestor of cyclostomes and gnathostomes [17].
In the hagfish, however, the above-mentioned anatomical pattern is greatly modified, i.e., the postotic pharynx is translocated extremely caudally, leaving a coelomless axis in front. In terms of the absence of the coelom, as well as the presence of a large number of suprapharyngeal myotomes, this elongated part of the hagfish resembles the “neck” of amniotes, probably as a homoplasy. The ventral portion of this “hagfish neck” is occupied by a cyclostome-specific structure, the lingual apparatus. This “cyclostome tongue” represents another homoplasy; unlike the somite-derived tongue in gnathostomes, it is a highly specialized organ derived from the mandibular arch [11,24,54]. The vagus nerve is extremely extended anteroposteriorly, together with the glossopharyngeal nerve at the level of pharyngeal arches 3 and 4, situated medial to the spinal nerve as well as trunk muscles. The results of the present study indicate that the initial topographical relationships among the pharynx, coeloms and cranial nerves (except the hypoglossal) are perfectly matched between the hagfish and the lamprey and gnathostomes.
Evolutionarily, the hagfish-specific peculiarity can be explained most parsimoniously as secondarily introduced changes unique in the lineage of hagfishes: since the morphological pattern of the lamprey hypobranchial/neck region resembles that of the gnathostomes, the evolutionary polarity suggests the apomorphic nature of the hagfish condition. Therefore, the ORM in the hagfish most likely represent secondary fusion or assimilation of hypobranchial muscles and abaxial muscles in the trunk, and not a primitive state before the separation of these two groups of muscles. This assumption simultaneously suggests that both the occipitospinal nerves and hypobranchial muscles had already been acquired before the split between cyclostomes and gnathostomes, more than 500 million years ago (reviewed by [55]). It should nonetheless be noted that the ancestral vertebrates may have possessed the hagfish ORM–like ventral muscles in the trunk that would not have been differentiated into hypobranchial and abaxial muscles. As mentioned by Nishi [23], the hypobranchial muscle (=rectus cervicus) and rectus abdominis were thought to represent serial homologues. Developmentally as well, these muscle primordia resemble each other, especially in terms of local mesenchyme-dependent patterning [35,36], even if there exists a conspicuous difference in the source of connective tissues [56,57].
The vertebrate hypobranchial muscles are conspicuous primarily in extant jawed vertebrates, and are generally regarded as a highly specialized category of trunk skeletal muscle. Anatomically, as seen in the tongue and infrahyoid muscle complex in mammals, the hypobranchial muscles do not reside in the lateral body wall, but are situated directly outside of the visceral structures and oral cavity. Developmentally, they arise from several rostral somites, including those often called the occipital somites, and the myoblasts migrate for a long distance along the posterior edge of the pharynx and root of the pericardium, to arrive at the oropharyngeal floor. Although this pathway is recognized in the embryonic context as the rostralmost part of the lateral body wall, this environment contains cephalic crest–derived ectomesenchyme, which will later contribute to the formation of the connective tissue of the hypobranchial muscles [17,50,56-59].
Differentiation and patterning of these muscles are highly dependent on Pax3 expression, and hypobranchial myoblasts and other long-distance–migrating myoblasts of somitic muscles, including limb muscles, express Lbx1-homologues (marker of migrating muscle precursor cells; reviewed by [60-67]. Not much is known about the developmental regulation of hypobranchial myoblasts, but hepatocyte growth factor (HGF) is distributed in the embryonic environment corresponding to the above summarized pathway, and c-Met, the gene encoding the receptor for HGF, is expressed in the myoblasts, implying that HGF signaling may be involved in the pathway regulation [61]. Due to postembryonic changes, especially the retraction of the coelom and neck formation that proceed from anterior to posterior, mature hypobranchial muscle and hypoglossal nerve in late embryos of jawed vertebrates are no longer found in the body wall.
In the lamprey, although a typical hypobranchial muscle does not appear, its possible precursor, or homologue, has been identified and named the “hypobranchial muscle” for its position ventral to the gill pores. Similarly, the nerves that innervate the hypobranchial musculature have been termed the “occipitospinal nerves” or “hypoglossal nerve” [25,28,68-70]. This homology has long been assumed by comparative embryologists. Unlike in jawed vertebrates, the hypobranchial muscle of the lamprey is segmented along the anteroposterior axis, with each segment not directly aligned with a dorsal myotome, but rather spanning two successive branchial arch skeletons. Thus, this segmental configuration, unique to the lamprey, does not reflect its innate developmental pattern, but is very likely a derived feature, adapted for pure mechanical function.
Developmentally, the lamprey hypobranchial muscle appears to be derived from rostral myotomes, except, possibly, for the first two or three segments, which differentiate into supraoptic and infraoptic myotomes, which are other cyclostome-specific muscles (as for its potential homology with the cucullaris muscle, see [71]; unpublished data by [72]). The hypobranchial muscle in the lamprey develops rather late in embryogenesis, possibly as the direct elongation of the ventral edge of myotomes expressing LjPax3/7 [73]. Although this anlage is a compact mass of cells, and does not appear to be composed of actively migrating mesenchymal myoblasts, its overall morphology is very reminiscent of the hypoglossal chord in other vertebrate embryos, even if it does not pass ventral to the pharynx and lateral to the pericardium. In lamprey and shark embryos, the hypobranchial muscle anlage is thought to grow as a direct extension of myotomes, rather than from migrating myoblasts [25,62,67,74], and the hagfish hypobranchial muscle seems to fall into this same category. Late expressions of LjMyHC2, LjLbxA, and LjMRF-A have also been detected in the hypobranchial muscle anlage in the lamprey [66,75]. Although most of these gene expression patterns are shared by the ventralmost part of the myotome in the lamprey, suggesting the possibility that ventral trunk muscle and hypobranchial muscles in the lamprey share common properties, there is also a distinction between these muscle anlagen both in terms of the morphology and growth rates [67,75]. In the hagfish, as far as we observed (mainly at histological levels), there was no clear distinction or difference found between the anterior and posterior parts of ORM anlagen.
Anatomically, however, there may at least be a clear distinction between the anterior and posterior part of the oblique muscles. Namely, the pars decussata only arises at the level of the pharynx and anterior (Figure 1C), and does not appear caudal to the pharynx. Thus, morphologically speaking, the hagfish oblique muscle is more ventrally extending in the rostral part, a configuration which is very reminiscent of the developing lamprey larvae [67] (Figure 7). It appears now very likely that the rostral part of the obliquus muscle with pars decussata on the contralateral side homologizes with the hypobranchial muscle in the lamprey (Figure 7, right). The anteroposterior distinction of the rectus muscle remains enigmatic. Only its association with the lingual apparatus is suggestive of its hypobranchial muscle–like nature. Further molecular and cellular level analyses would be necessary to identify the hypobranchial muscle homologues not only in the hagfish but also in the lamprey, where distribution of the cephalic neural crest–derived ectomesenchyme has not been fully understood.
As noted above, the rostral part of the ORM in the hagfish occupies a position equivalent to that of typical hypobranchial muscles in other vertebrates, but is not innervated by the typical occipitospinal nerve whose axons are found along the circumpharyngeal space (along the postpharyngeal arc). Instead, these muscles are innervated by the segmental spinal nerves that grow vertically from the rostral spinal cord. Strangely, this trajectory falls within the domain of third and fourth pharyngeal arches; these nerves violate the rule of mutual exclusion normally observed between the pharyngeal arches and trunk. It is true that for these nerves to reach the hypobranchial muscle, rather than growing caudally first for a long distance all the way to the posterior end of the branchial apparatus, to circumvent this apparatus by making an arch that grows ventrally, and then turns to take a rostralward pathway to come back to the muscle, taking a short cut would be much easier. There are several possible hypotheses to explain this exceptional morphology. First, there could still be unknown mesenchymal rearrangement that obliterates the typical head-trunk interface as seen in other vertebrates. Second, there is an unusual delay in the timetable for the development of the hypobranchial/hypoglossal system, and the difference in embryonic environments that establishes the head-trunk interface has long been deactivated by the time this organ is being patterned. Third, muscle precursor and nerve axons for the hypobranchial/hypoglossal system have acquired properties for patterning in the pharyngeal arch–like environment, which is unique to the hagfish.
Due to the limited amount of embryonic material, the lack of information about cyclostome myogenesis, and especially the inaccessibility to experimental embryology, we cannot easily evaluate the likelihood of the above hypotheses. It would, however, be worth comparing the anatomy and embryology of the amniote “neck”, which shows some similarity to the situation we have observed in the hagfish. In both mammals and birds, anteroposteriorly extended coelom-less domains are found between the skull and the shoulder girdle. These domains are characterized by the distribution of “neck muscles”, consisting of cucullaris and hypobranchial muscles, and by the distribution of the cephalic crest–derived ectomesenchyme, which provides connective tissues for the neck muscles [57].
Although the acquisition of the neck is not entirely comparable between birds and mammals (e.g., the entire set of postotic aortic arches as well as inferior ganglia, together with the parathyroid and thymus, shift caudally to the cardiac level in birds, whereas the dorsal part of the pharynx including the inferior ganglia remains close to the skull in mammals), some notable similarities exist, as seen, for example, in the superficial layer of the neck that is predominantly formed of second arch-derived cutaneous muscles (platysma muscle in mammals; m. constrictor superficialis coli, innervated by n. VII in avians). Thus, in amniotes, the hyoid arch becomes caudally expanded in the late pharyngular stage as a “collar”, to form the surface of the neck. By that time, the pharyngeal pores are mostly diminished, except for the ectodermal cervical sinus that continues production of nodose ganglionic neurons for the vagus nerve, while at the same time, pharyngeal arches 3 and posterior are being covered laterally by the second arch. The ventral surface of the neck, however, is not innervated by sensory fibers of the facial nerve, but in both mammals and birds, the cutaneous fibers distributed in that area originate from cervical spinal nerves (by way of ansa cervicalis in mammals). In sauropsids in particular, the cutaneous sensory fibers appear as segmentally arranged nerve nets, reflecting the presence of more clearly segmented cervical dermatomes in these animals when compared with those of mammals. This latter pattern of cervical nerve distribution is highly reminiscent of the spinal nerves’ innervation of the hypobranchial muscle in the hagfish (Figure 3H). These sensory branches penetrate the hyoid arch muscle, showing no sign of visceral/somatic distinction in their axonal morphology: they do not circumvent the hydoid arch–derived cutaneous muscles.
Similarly, the amniote hypoglossal nerves and muscles migrate in the lateral body wall (pericardium) only at their earliest phase of development; they are located close to the skull, far rostral to the caudal end of the neck, in the late embryonic to adult states. Thus, these nerves no longer indicate the caudal limit of the cephalic crest cells or the domain of head-like properties.