Thalamic collaterals of corticostriatal axons: their termination field and synaptic targets in cats.

Branched cortical projections to the thalamus and striatum were investigated in cats by injecting the retrograde-anterograde tracer biotinylated-dextran amine (BDA) into the caudate nucleus. These injections gave rise to plexuses of labeled fibers and varicosities in widespread thalamic territories. For instance, the lateroposterior nucleus and pulvinar (LP-PUL) mostly contained thick axons that contributed clusters of large-sized varicosities, each forming multiple asymmetric synapses, usually with vesicle-filled dendrites. In contrast, the intralaminar nuclei mostly contained thin axonal segments that emitted small en passant varicosities that formed single asymmetric synapses with spines. Because the caudate nucleus does not project to the thalamus, this labeling had to arise from a neuronal population with branching axons to both structures. Previous findings pointed to three possible sources: brainstem monoaminergic cells, intralaminar thalamic neurons, and corticostriatal cells. The first candidate could be ruled out because monoaminergic neurons contribute small-sized terminals that usually lack membrane specializations. The second possibility was discarded because retrograde tracer injections into the LP-PUL did not give rise to retrograde labeling in the intralaminar nuclear complex but to massive retrograde labeling in deep layers of cortical areas 5 and 7. Therefore, we concluded that the thalamic anterograde labeling originated from corticostriatal neurons, with axons branching to the thalamus. In keeping with this conclusion, Phaseolus vulgaris-leucoagglutinin (PHA-L) injections into cortical areas 5-7 labeled a group of thick corticothalamic fibers that ended in clusters of large boutons in the LP-PUL. These PHA-L-positive terminals were indistinguishable from those labeled after injections of BDA into the caudate nucleus, but they were easy to distinguish from the typical corticothalamic fibers. These findings indicate that the cerebral cortex could coordinate the activity of the striatum and the thalamus via a rich axonal network that collateralizes to both structures. The extent and synaptic organization of this branched projection impose a revision of the traditional scheme of thalamic connectivity.