Partial reconstruction of the adult Lurcher cerebellar circuitry by neural grafting.

Solid cerebellar grafts, taken from normal mouse embryos (gestational day 12-14), were transplanted into the cerebellum of adult Lurcher mice. The degree of Purkinje cell replacement was analysed one to three months after transplantation by means of immunocytochemistry (antibodies against calbindin, cGMP-dependent protein kinase and neurofilament proteins) and electron microscopy. Grafted Purkinje cells succeed in moving out of the graft and migrate into the host cerebellar cortex. They are present next to the graft in the granule cell and molecular layers, and far from the graft remnant, only in the molecular layer, indicating that, although both layers subserve Purkinje cell migration, the molecular layer is the ultimate target. In the host molecular layer, axons of transplanted Purkinje cells form thick bundles running in the frontal plane over long distances. Most of them terminate in the upper granule cell layer by enlarged bulbs resembling collapsed growth cones. Axons reaching their normal targets (the neurons of the deep cerebellar nuclei) are observed only in cases where the granule cell layer is disrupted and/or grafted Purkinje cells remain in the white matter. The projection is massive only from grafts lying in the close vicinity of the target neurons. Electron-microscopic analysis of grafted Purkinje cells populating the host cerebellar cortex reveals that their synaptic investment is abnormal. In the molecular layer, where the normal inputs are reduced, the compartmentation in proximal and distal dendritic segments is severely affected, climbing fibre synapses only form on a minority of grafted cells and "pinceau" formations are absent. In the granule cell layer, the synaptic investment is similar to that of Purkinje cells in agranular cerebellum, and even heterelogous synapses with mossy fibres have been observed. These results, compared to those previously obtained with grafting experiments in Purkinje cell degeneration mutant mouse, allow us to conclude that: (i) the Purkinje cell-deficient molecular layer of the host, despite its severe atrophy and reactive gliosis, still exerts a positive neurotropism specific for grafted Purkinje cells; (ii) the unlesioned host granule cell layer underlying the molecular layer containing grafted Purkinje cells, even if almost depleted of granule cells, remains an obstacle for the re-establishment of a corticonuclear projection; and (iii) the degree of synaptic integration of grafted Purkinje cells is directly related to the nearby presence of available host axon terminals. Hence, owing to the atrophy of the Lurcher cerebellum, the postgrafting restoration of the cerebellar cortical circuit is much less complete in this mutant.