During normal vertebrate development, the ventricular zone (VZ) of the spinal cord reduces in size during the late prenatal period. In this process, termed dorsal obliteration, dorsal VZ progenitors move
away from the VZ. Remaining ventral progenitors give rise to the ependymal zone of the adult central
canal. At present, little is understood of the mechanisms that drive dorsal obliteration. Here I analyse
dorsal obliteration in the prenatal mouse. I show that it occurs over a tight time-window, between E14-
E17. Over this period, the dorsal midline is characterised by dorsal Nestin+ radial glial (dNRG) cells that
display strong apical/endfeet expression of the apical polarity proteins CRB2 and ZO-1. Concomitantly,
adjacent sub-dorsal VZ cells downregulate CRB2, ZO-1 and aPKC, and their nuclei become located away
from the lumen. Ectopic transplantation of dNRG into the lumen of embryonic chick neural tube causes
a disruption of progenitor patterning and the appearance of ectopic progenitor cells outside the neural
tube, suggesting that dNRG cells secrete a diffusible factor that can disrupt neuroepithelial integrity.
Previous bioinformatics studies had suggested the existence of an isoform of CRB2 that lacks the
transmembrane and intercellular domain, and so may function as a secreted variant. Using a nested PCR
approach, I find evidence that a isoform of CRB2 that lacks the transmembrane domain is specifically
expressed in dNRG cells. Further, a protein encoded by this splice variant (secreted CRB2) can be
secreted, and, when presented ectopically to chick neural tube, can disrupt neuroepithelial integrity. I
therefore propose a model for dorsal obliteration in which secreted CRB2 can outcompete or interfere
with a normal homodimerisation of transmembrane CRB2, leading to the destabilisation of the CRB2
complex and apical polarity, and the consequent delamination of subdorsal VZ cells. As a first step in
testing this model, I develop an in vivo mouse slice culture assay in which to monitor cell behaviour
during dorsal obliteration in real time. These studies provide evidence that the cell bodies of sub-dorsal
cells immediately adjacent to elongating roof plate cells/dorsal Radial Glia migrate dorsally, using either
the dorsal Radial Glia as a scaffold, or, potentially, re-orienting their own processes along the scaffold,
and migrating along these. Finally, sub-dorsal cells appear to detach from the VZ in a process that may
resemble apical abscission. Together, these studies suggest that a disruption of apical polarity proteins,
driven through dorsal midline-secreted CRB2, is instrumental in dorsal obliteration. I discuss this model
within the context of embryonic neurulation and discuss the implications for potential of the adult
spinal cord.
