In all jaw-bearing vertebrates, three-dimensional mobility relies on segregated, separately innervated epaxial and hypaxial skeletal muscles. In amniotes, these muscles form from the morphologically continuous dermomyotome and myotome, whose epaxial-hypaxial subdivision and hence the formation of distinct epaxial-hypaxial muscles is not understood. Here we show that En1 expression labels a central subdomain of the avian dermomyotome, medially abutting the expression domain of the lead-lateral or hypaxial marker Sim1. En1 expression is maintained when cells from the En1-positive dermomyotome enter the myotome and dermatome, thereby superimposing the En1-Sim1 expression boundary onto the developing musculature and dermis. En1 cells originate from the dorsomedial edge of the somite. Their development is under positive control by notochord and floor plate (Shh), dorsal neural tube (Wnt1) and surface ectoderm (Wnt1-like signalling activity) but negatively regulated by the lateral plate mesoderm (BMP4). This dependence on epaxial signals and suppression by hypaxial signals places En1 into the epaxial somitic programme. Consequently, the En1-Sim1 expression boundary marks the epaxial-hypaxial dermomyotomal or myotomal boundary. In cell aggregation assays, En1- and Sim1-expressing cells sort out, suggesting that the En1-Sim1 expression boundary may represent a true compartment boundary, foreshadowing the epaxial-hypaxial segregation of muscle.

It is generally held that vertebrate muscle precursors depend totally on environmental cues for their development. We show that instead, somites are predisposed toward a particular myogenic program. This predisposition depends on the somite's axial identity: when flank somites are transformed into limb-level somites, either by shifting somitic boundaries with FGF8 or by overexpressing posterior Hox genes, they readily activate the program typical for migratory limb muscle precursors. The intrinsic control over myogenic programs can only be overridden by FGF4 signals provided by the apical ectodermal ridge of a developing limb.

Recent knockout experiments in the mouse generated amazing craniofacial skeletal muscle phenotypes. Yet none of the genes could be placed into a molecular network, because the programme to control the development of muscles in the head is not known. Here we show that antagonistic signals from the neural tube and the branchial arches specify extraocular versus branchiomeric muscles. Moreover, we identified Fgf8 as the branchial arch derived signal. However, this molecule has an additional function in supporting the proliferative state of myoblasts, suppressing their differentiation, while a further branchial arch derived signal, namely Bmp7, is an overall negative regulator of head myogenesis.