Expression of AmphiHox-1 and AmphiPax-1 in amphioxus embryos treated with retinoic acid: insights into evolution and patterning of the chordate nerve cord and pharynx

Abstract

Excess all-trans retinoic acid (RA) causes severe craniofacial malformations in vertebrate embryos: pharyngeal arches are fused or absent, and a rostrad expansion of Hoxb-1 expression in the hindbrain shows that anterior rhombomeres are homeotically respecified to a more posterior identity. As a corollary, neural crest migration into the pharyngeal arches is abnormal. We administered excess RA to developing amphioxus, the closest invertebrate relative of the vertebrates and thus a key organism for understanding evolution of the vertebrate body plan. In normal amphioxus, the nerve cord has only a slight anterior swelling, the cerebral vesicle, and apparently lacks migratory neural crest. Nevertheless, excess RA similarly affects amphioxus and vertebrates. The expression domain of AmphiHox-1 (homologous to mouse Hoxb-1) in the amphioxus nerve cord is also extended anteriorly. For both the amphioxus and mouse genes, excess RA causes either (1) continuous expression throughout the preotic hindbrain (mouse) and from the level of somite 7 to the anterior end of the nerve cord (amphioxus) or (2) discontinuous expression with a gap in rhombomere 3 (mouse) and a gap at the posterior end of the cerebral vesicle (amphioxus). A comparison of these expression patterns suggests that amphioxus has a homolog of the vertebrate hindbrain, both preotic and postotic. Although RA alters the expression of AmphiHox-1 expression in the amphioxus nerve cord, it does not alter the expression of AmphiHox-1 in presomitic mesoderm or of alkali myosin light chain (AmphiMlc-alk) in somites, and the axial musculature and notochord develop normally. The most striking morphogenetic effect of RA on amphioxus larvae is the failure of mouth and gill slits to form. In vertebrates effects of excess RA on pharyngeal development have been attributed solely to the abnormal migratory patterns of Hox-expressing cranial neural crest cells. This cannot be true for amphioxus because of the lack of migratory neural crest. Furthermore, expression of Hox genes in pharyngeal tissues of amphioxus has not yet been detected. However, the absence of gill slits in RA-treated amphioxus embryos correlates with an RA-induced failure of AmphiPax-1 to become down-regulated in regions of pharyngeal endoderm that would normally fuse with the overlying ectoderm. In vertebrates, RA might similarly act via Pax-1/9, also expressed in pharyngeal endoderm, to impair pharyngeal patterning.