Characteristics of vasculature and neurovascular relations in intraventricular anterior hypothalamic transplants

Fetal hypothalami obtained from normal Long-Evans rats were transplanted to the lateral, third or fourth ventricle of adult male Brattleboro rats, homozygous for diabetes insipidus. The density of the capillary plexuses within the grafts did not vary as a function of their intraventricular location; all transplants exhibited a capillary density equivalent to that of the in situ hypothalamus. Intravascular injections of HRP resulted in retrograde neuronal labeling only in grafts that were attached to circumventricular organs of the host brain, especially the median eminence. Typically, HRP-associated label was confined to vascular and perivascular elements and not diffusely distributed within the graft parenchyma, indicating that capillaries within the transplants developed barrier properties similar to those of the native hypothalamus. Neural integration of the transplant with the recipient brain was limited; at most points of apposition the neuropil of graft and host were separated by an intervening ependymal layer or glia limitans. All surviving grafts contained neurophysin and/or vasopressin-immunoreactive (VP-ir) neurons. Three anatomically distinct populations of VP-ir neurons were identified. Magnocellular VP-ir neurons were identified in less than half of the grafts, and when present they were few in number distinct, but sparse vasopressinergic innervation of median eminence capillaries was observed in all cases where grafts containing magnocellular neurons were apposed to this structure. Most grafts contained numerous, parvicellular VP-ir neurons arranged in aggregations which resembled the suprachiasmatic nucleus (SCN). SCN-like cell groups projected to neural targets within the graft and the host brain, but they did not project onto blood vessels. A second, distinct class of parvicellular VP-ir neuron also was identified in a majority of transplants. In contrast to SCN-like ('type' I) cells, these 'type II' parvicellular neurons were somewhat larger and found in less discretely organized groups, and they projected to vascular targets; usually locally elaborated capillary plexuses intrinsic to the transplants. In the present study, there was no amelioration of the symptoms of diabetes insipidus in the host animals despite the presence of numerous VP-ir neurons in virtually all grafts. This was probably related to the limited survival of magnocellular VP-ir neurons, which appear to be the principal source of vasopressinergic projections from the graft to fenestrated capillary plexuses of the host brain.