Termination of Somitogenesis
In contrast to some annelids such as Platynereis dumerilii, which continue to add new posterior segments for the duration of their life (Fischer and Dorresteijn, 2004), all vertebrates eventually stop forming segments during embryogenesis. How this process comes to end is not well understood. Several mechanisms of body axis elongation and segment addition termination have been proposed, although a consensus has not been reached between different vertebrate species. A simple model is that the rate NMPs exit into the spinal cord and mesoderm exceeds the rate at which they divide, depleting the NMPs until they eventually run out (Kimelman, 2016). In the zebrafish, where tailbud NMPs rarely divide, this model seems plausible, although experiments in which additional NMPs are added to embryos to determine the effect on segment number and body length has not been performed.
In amniotes, where NMPs exhibit stem cell properties and divide more frequently than in zebrafish, other models have been proposed. During mouse development, the establishment of anterior-posterior identity along the axis is tied to dynamic changes in NMP gene expression as the axis elongates (Wymeersch et al., 2019). This is opposed to the neighboring progenitor cells of the notochord, which retain a consistent gene expression program throughout axial extension (Wymeersch et al., 2019). As NMPs mature during development, they progress through a complement of Hox transcription factor expression. Hox genes are well known for imparting axial identity to the body plan of animals, with anteriorly expressed Hox genes inducing anterior fates, and posterior Hox genes inducing posterior fates (Gaunt, 2018). This suggests that axial identity of the embryo is established, at least in part, by the transcriptionally dynamic NMP population that turns on Hox genes specific for a region of the axis and then contributes descendent cells to that region (Wymeersch et al., 2019). In this way, a clock mechanism within the NMPs controlling Hox gene expression may dictate the end of axial extension when the posterior terminal Hox genes are expressed. Indeed, posterior Hox gene expression in the chick and mouse is hypothesized to induce termination of axis extension, in part due to the posterior Hox gene-mediated repression of canonical Wnt and FGF signaling, which is required for continuous axis extension (Denans et al., 2015; Olivera-Martinez et al., 2012; Young et al., 2009).
The reduction in Wnt signaling as more posterior Hox genes are expressed causes a shortening of the presomitic mesoderm territory, which reduces the distance between the most recently formed somite and the NMPs (Denans et al., 2015). The signaling molecule retinoic acid is normally produced by the most recently formed somites, and is sufficient to inhibit the maintenance of NMPs and induce neural differentiation within them (Dobbs-McAuliffe et al., 2004; Martin and Kimelman, 2010; Niederreither et al.. 1997; Olivera-Martinez et al., 2012). In the chick, as the source of RA gets closer to the NMPs and RA signaling becomes active in the tail- bud, Brachyury expression in the NMPs is repressed, leading to neural induction and the depletion of the NMPs (Olivera-Martinez et al., 2012). The retinoic acid signaling in the tailbud represses FGF signaling, which in turn causes a loss of Brachyury expression. This model does not apply to all vertebrates, however. In mouse and zebrafish, for example, loss of function of the major retinoic acid-producing enzyme aldhla2 does not affect the proper termination of axis elongation (Begemann et al., 2001; Cunningham et al., 2011).