Origin and plasticity of the subdivisions of the inferior olivary complex.

The precerebellar nuclei (PCN) originate from the rhombic lip, a germinal neuroepithelium adjacent to the roof plate of the fourth ventricle. We first report here that, in chicken, the Brn3a-expressing postmitotic medullary cells that produce the inferior olive (ION, the source of cerebellar climbing fibres) originate from a dorso-ventral domain roughly coinciding with the hindbrain vestibular column. Whereas Foxd3 expression labels the whole mature ION but is only detected in a subpopulation of ION neuroblasts initiating their migration, we report that Brn3a allows the visualization of the whole population of ION neurons from the very beginning of their migration. We show that Brn3a-positive neurons migrate tangentially ventralwards through a characteristic dorso-ventral double submarginal stream. Cath1 expressing progenitors lying just dorsal to the ION origin correlated dorso-ventral topography with the prospective cochlear column (caudal to it) and generate precerebellar nuclei emitting mossy-fiber cerebellar afferents. We used the chick-quail chimaera technique with homotopic grafts at HH10 to determine the precise fate map of ION precursors across the caudal cryptorhombomeric subdivisions of the medullary hindbrain (r8-r11). We demonstrate that each crypto-rhombomere contributes to two lamellae of the ION, while each ION sub-nucleus originates from at least two contiguous crypto-rhombomeres. We then questioned how rhombomere identity is related to the plasticity of cell type specification in the dorsal hindbrain. The potential plasticity of ectopically HH10 grafted ION progenitors to change their original fate in alternative rostrocaudal environments was examined. Heterotopic grafts from the presumptive ION territory to the pontine region (r4-r5) caused a change of fate, since the migrated derivatives adopted a pontine phenotype. The reverse experiment caused pontine progenitors to produce derivatives appropriately integrated into the ION complex. Grafts of ION progenitor domains to myelomeres (my) 2-3 also showed complete fate regulation, reproducing spinal cord-like structures, whereas the reverse experiment revealed the inability of my2-3 to generate ION cell types. This was not the case with more caudal, relatively less specified myelomeres (my5-6). Interestingly, when heterotopically grafted cells are integrated dorsally, they do not change their phenotype. Our results support the hypothesis that positional information present in the hindbrain and spinal cord at early neural tube stages controls the specific fates of ventrally migrating PCN precursors.