Cell junctions in the ventricular zone (germinal matrix) of the embryonic and foetal sheep brain were examined with thin-section and freeze-fracture electron microscopy. Neuroependymal cells in the early ventricular zone (days 19-40 of embryonic development, gestation period 147 days) exhibit a novel arrangement of cell junctions that connect adjacent neuroependymal cells at their lateral cell membranes next to the ventricular system. Small but typical gap junctions were also identified from the earliest stages examined. In serial thin sections and using a goniometer with a tilting device, the cell contacts showed a tight junction-like appearance of close and continuous fusion between neighbouring cell membranes. However, they were not arranged in a belt-like fashion close to the ventricular surface, but spiralled from the ventricular pole of the cells along the lateral cell membrane towards the deeper parts of the ventricular zone. Their freeze fracture appearance was different from that of single-stranded tight junctions in that the dimensions of their ridges and grooves were generally greater and the E-face grooves contained many particles. The junctions were especially prominent where more than two cells made contact. At mid-gestation they were less prominent than earlier and at 125 days gestation the neuroependymal layer was replaced by a mature-looking normal ependymal layer in which individual ependymal cells were connected by zonulae adherentes and large gap junctions; orthogonal arrays were also prominent. The close contact between gap junctions and single-stranded junctions found early in gestation suggests that there may be some developmental relation between these two membrane specializations. The transient single-strand junctions presumably form the morphological basis for a recently described CSF-brain barrier in the early foetal sheep brain. They may also have some mechanical function in anchoring neighbouring cells together in the region of the developing brain where cells are continuously dividing and migrating.

The nature of the barriers that keep proteins out of the developing brain has been studied in tissues obtained from fetal sheep in experiments conducted under controlled physiological conditions. In anaesthetised pregnant ewes, 60 day gestation fetuses (term is 150 days) were exposed to human albumin injected intravenously for periods up to 6 h. The immunocytochemical distribution of exogenous human albumin was compared with that of endogenous sheep albumin at both the light and electron-microscopical level. Immunogold labelling of ultracryosections suggests that a tubulocisternal endoplasmic reticulum system in immature choroid-plexus epithelial cells is the route by which albumin crosses from blood to cerebrospinal fluid (CSF) in the developing brain. The integrity of the blood-brain barrier, the blood-cerebrospinal fluid barrier and the cerebrospinal fluid-brain barrier to protein, was confirmed. In addition, at the outer surface of the developing brain there also appears to be a restriction on the passage of albumin from CSF into the brain. These observations support earlier proposals that the immature brain develops within an internal environment from which proteins in plasma and CSF are largely excluded.

Blood-brain, blood-CSF and ventricular CSF-brain barriers to protein are present very early in brain development. In order to determine whether the outer pial surface of the brain also restricts free penetration of macromolecules, the dorso-lateral part of the sensorimotor cortex from rats at embryonic day 12 (E12), 14, 16, and 18, the day of birth (P0), and adult rat, was studied by electron microscopical techniques. Potassium ferrocyanide, Ruthenium Red and immunogold labelling of endogenous albumin were used to investigate junctional structures and the sites of restriction to albumin diffusion. At E12, large fenestrated sinusoids were present in the pia-arachnoid and the brain surface was formed by an incomplete layer of neuroepithelial and presumptive radial glial end feet, but capillaries in the pia-arachnoid showed no fenestrations at E14 or later. From E14, we observed the progressive appearance of distinct junctional structures between the glial end feet which, to our knowledge, have not been described before. Analysis of albumin distribution from E16 to P0 suggests that the junctions may contribute to restriction of diffusion between the subarachnoid space and the brain extracellular fluid. The restriction to the penetration of protein at both the pial and the ependymal surfaces may ensure the isolation of the neural environment during a critical phase in development of the nervous system. The changes in the structure of the junctions between E12 and P0 suggests a transitional series of embryonic junctional types, which eventually give way to the mature junctions of the adult. Parallels between the embryonic glial junctions and junctions described in adult invertebrate brain, suggest some interesting parallels in junctional development in phylogeny and ontogeny.

The population of microglial cells in the subependymal layer of the subcommissural organ is sparse in normal adult rats. The number of microglial cells was substantially increased in this area following intraventricular injection of the serotonin neurotoxin 5,6-dihydroxytryptamine (5,6-DHT). In sections of plastic embedded material, 1 micron thick, the majority of phagocytic cells scattered in the subependymal layer had an appearance similar to that described in classical studies of microglial cells. At the electron microscopic level microglial cells exhibited the characteristic elongate nucleus with peripheral chromatin condensation. The perikaryon was scanty, containing strands of rough endoplasmic reticulum. The abundant organelles in the processes included Golgi complexes, mitochondria, rough and smooth endoplasmic reticulum as well as dense and multivesicular bodies. In addition, the processes contained phagocytosed axon terminals originating from the dense serotoninergic input to the subcommissural organ, which had degenerated on accumulating the serotonin neurotoxin. A fraction of the phagocytosed material was contained in subependymal subcommissural organ cells, astrocytes and oligodendrocytes. At the light microscopic level the phagocytosed terminals were visualized histochemically with Schmorl's reaction, which resulted in Prussian Blue precipitates. This allowed screening of microglial cells in complete series of sections through the well-defined subependymal layer of the subcommissural organ.

The CNS of the newly born opossum removed in its entirety survives and maintains its electrical excitability in suitable culture media for up to ten days at 25 degrees C. The structure of the developing neonatal spinal cord has been studied in the intact animal and in the cultured CNS. The differentiation and survival of individual cells and subcellular structures were followed at the light and electron microscopic level. The expression of cell markers in neuronal and glial cells was studied immunocytochemically using commercially available antibodies. Both mono- and polyclonal antibodies raised against antigens from several other species cross-reacted with Monodelphis antigens. The spinal cord of preparations removed from three-day-old-animals showed many neuron specific enolase-positive large neurons in the ventral horn as well as vimentin- and glial fibrillary acidic protein-positive radial glial cells and numerous small diameter unmyelinated axons, abundant dendrites and synaptic structures. From post natal day 5 to post natal day 8 continued differentiation of neurons and differentiation of radial glial cells into astrocytes were apparent. Radial glial fibres and astrocytes reacted positively to antibodies against glial fibrillary acidic protein. Myelin had not appeared at 8 days. A comparison of material obtained from postnatal day 3-postnatal day 4 preparations fixed immediately after dissection and from postnatal day 3-postnatal day 4 preparations fixed after 5 days in culture showed growth with continued mitotic activity of the neuroepithelial cells and further neuronal and glial maturation in the spinal cord especially in the more rostral end. In successful experiments in vitro, the preservation of individual cells, organelles, membranes and synapses was similar in the freshly dissected and cultured preparations apart from a distinct loss of the youngest and some of the oldest neurons in the spinal cord. Also the main fibre tracts (dorsal, lateral and ventromedial funiculus) survived. Virtually all preparations that had not been damaged or injured showed these results. Possible reasons for the death or survival of individual neuronal or glial cell populations in these preparations are discussed.