Oxygen-sensing cells release neurotransmitters, including serotonin, in response to hypoxia in the blood or surrounding air/water. This stimulates the glossopharyngeal and/or vagal nerves, triggering increased ventilation via the respiratory reflex. In the adult, they are located in the carotid body (glomus cells) and lung epithelia (pulmonary neuroendocrine cells) of amniotes, and in the epithelia of the gills and orobranchial cavity (‘neuroepithelial cells’) of anamniotes. Despite their physiological importance, little is known about the molecular mechanisms of their development, while the evolutionary relationships between the various oxygen-sensing cell types are unknown.

The chromaffin cells of the mammalian adrenal medulla are hypoxia-sensitive transiently during neonatal life. Both carotid body glomus cells and adrenal chromaffin cells arise from the neural crest and require the transcription factors Phox2b and Ascl1 for their development. Given these similarities, I aimed to test the hypothesis that the same molecular mechanisms underlie their development. Expression analysis of 13 sympathoadrenal pathway genes throughout chicken carotid body development revealed striking similarities with adrenal chromaffin cell development. Analysis of mouse mutants showed that the transcription factors Hand2, Sox4 and Sox11 are required for carotid body development. In addition, loss of the receptor tyrosine kinase Ret or the transcription factor AP-2β, which significantly affects sympathetic ganglion but not adrenal chromaffin cell development, has no effect on the carotid body. Adrenal chromaffin cells differentiate from neurons that migrate into the adrenal gland from ‘primary’ sympathetic ganglia at the dorsal aorta. Carotid body glomus cells were previously proposed to arise from neuronal “émigrés” from neighbouring ganglia: the superior cervical ganglion in mammals and the nodose ganglion in the chick. However, nodose neurons are considered to be nodose placode-derived. Using electroporation and grafting in the chick, I confirmed that the nodose placode does not contribute to the carotid body, identified a small population of autonomic neural crest-derived neurons in the nodose ganglion, and confirmed the existence of “bridges” of neurons between the nodose ganglion and the carotid body. My data suggest that, like adrenal chromaffin cells, carotid body glomus cells differentiate from autonomic neural crest-derived neurons in nearby ganglia, which migrate into the carotid body primordium and down-regulate neuronal markers.

The proposed evolutionary relationship between the carotid body glomus cells and the serotonin-positive neuroepithelial cells of anamniote gills has never been tested. Using vital dye labelling, neural fold grafts, genetic lineage-tracing in zebrafish and analysis of zebrafish mutants lacking all neural crest cells, I found that serotonin-positive cells in the gills and orobranchial epithelia of lamprey (jawless fish), zebrafish (ray-finned bony fish) and frog (anamniote tetrapod) are not neural crest-derived, and hence are not homologous to carotid body glomus cells. Genetic lineage-tracing in mouse and neural fold grafts in chick also confirmed that serotonin-positive neuroendocrine cells in the lung are not neural crest- derived, hence must have an endodermal origin (since the lungs are out-pocketings of the gut). My results suggest that the neuroepithelial cells of anamniotes are not related to carotid body glomus cells, but rather are homologous to the oxygen-sensing cells of the lung. Consistent with this hypothesis, I found that many genes expressed during carotid body development are not expressed by the epithelia of either chick lungs or lamprey gills.

Taken together, my data suggest that as air-breathing evolved, gut endoderm- derived cells that originally responded to hypoxia in water were maintained in the lungs to monitor oxygen levels in air, while a population of neural crest-derived chromaffin cells near the pharyngeal arch arteries was recruited to monitor oxygen levels in blood.