Mammalian choroid plexuses develop at four sites in the roof of the neural tube shortly after its closure, in the order IVth, lateral, and IIIrd ventricles. Bone morphogenetic proteins and tropomyosin are involved in early specification of these sites and in early plexus growth. Four stages of lateral ventricular plexus development have been defined, based on human and sheep fetuses; these depend mainly on the appearance of epithelial cells and presence or absence of glycogen. Other plexuses and other species are probably similar, although marsupials may lack glycogen. Choroid plexuses form one of the blood-brain barrier interfaces that control the brain's internal environment. The mechanisms involved combine a structural diffusion restraint (tight junctions between the plexus epithelial cells) and specific exchange mechanisms. In this review, it is argued that barrier mechanisms in the developing brain are different in important respects from those in the adult brain, but these differences do not necessarily reflect immaturity of the system. Absence of a barrier mechanism or presence of one not found in the adult may be a specialisation that is appropriate for that stage of brain development. Emphasis is placed on determining which mechanisms are present in the immature brain and relating them to brain development. One mechanism unique to the developing brain transfers specific proteins from blood to cerebrospinal fluid (CSF), via tubulocisternal endoplasmic reticulum in plexus epithelial cells. This results in a high concentration of proteins in early CSF. These proteins do not penetrate into brain extracellular space because of "strap" junctions between adjacent neuroependymal cells, which disappear later in development, when the protein concentration in CSF is much lower. Functions of the proteins in early CSF are discussed in terms of generation of a "colloid" osmotic pressure that expands the ventricular system as the brain grows; the proteins may also act as specific carriers and growth factors in their own right. The pathway for low molecular weight compounds, which is much more permeable in the developing choroid plexuses, appears also to be a transcellular one, rather than paracellular via tight junctions. There is thus good evidence to support a novel view of the state of development and functional significance of barrier mechanisms in the immature brain. It grows in an environment that is different from that of the rest of the fetus/neonate and that is also different in some respects from that of the adult. But these differences reflect developmental specialisation rather than immaturity.

The nervous and the immune systems share several molecules that control their development and function. We studied the temporal and spatial distribution of the immunoreactivity of two acute-phase cytokines, TNF-α and IL-1β, in the developing sheep neocortex and compared it with the well-described distribution of fetuin, a fetal glycoprotein also known to modulate the production of cytokines by lipopolysaccharide (LPS)-stimulated monocytes and macrophages. TNF-α was present first at embryonic day 30 (E30) (term is 150 days in sheep) as a faint band of immunoreactivity between the ventricular zone and the primordial plexiform layer (preplate). IL-1β was detected at the first appearance of the cortical plate (E35-E40). Both cytokines were present on both sides of the cortical plate, which contained fetuin-positive cells, but was free from cytokine staining. By E60, TNF-α immunoreactivity was less prominent than that of IL-1β and was confined to the marginal zone and outer developing white matter; IL-1β was present in the marginal zone and in two bands of immunoreactive cells, one at the border of the cortical plate/developing layer VI (cells of neuronal morphology) and the other at the border of layer V and the developing white matter (identified as microglia). By E80, TNF-α staining had disappeared and IL-1β-immunopositive microglia were no longer detectable. By E100-E140 only a few immunoreactive cells were identified in layers V-VI; these did not co-localize with fetuin-positive cells. The differences in distribution between fetuin and the two cytokines suggest that the opsonizing role of fetuin, proposed for monocyte production of cytokines, is probably not present in the developing brain. However, early in neocortical development TNF-α and IL-1β were present in the subplate zone at a time of intense synaptogenesis.