At the beginning of the 7th week after fertilization the diencephalon is defined by the dorsal, medial and ventral thalamic sulci at the inner surface of the diencephalic neural canal which separate the four nuclear swellings of the epithalamus, dorsal thalamus, ventral thalamus and hypothalamus. The location of these sulci and developing nuclear masses can be fitted to the principal segmental organization which is found in all vertebrates. In the human brain this basic organization pattern is preserved and visualized by the demonstration of the CD15 carbohydrate antigen (1). Beginning at 10 weeks a subset of CD15+ radial glia demarcates the segments of the diencephalon and at 14 weeks an extended system is established. This is most elaborated in the ventral thalamus where registry is provided between the radial glia cells at the ventricular surface and the stem of the choroid plexus along the pial surface. The topography favours the conclusion that CD15 is involved in mechanisms which provide shape to the CNS during the period of the formation of the telencephalic hemispheres. The further development of the thalamus can therefore be analysed with respect to this principally simple topography.

Differentiation in the dorsal thalamus starts at 9 weeks, at the beginning of fetal life and can be monitored by several antibodies, e.g. parvalbumin and calretinin. In comparison with the differentiation of the ventral thalamus the differentiation of the dorsal thalamus begins late and progresses slowly until about 13 weeks when the proliferating zone is almost exhausted. Differentiation into various subdivisions follows a specific time sequence for each nucleus which is mirrored by the expression of the CD15. Good differentiation allows the definition of the differentiation zones of the various subnuclei and of the borders between them. The appearance of CD15 material between the ventricular zone and the developing nuclei is striking; this is interpreted as a sign for the migratory pathway of the respective neurons and provides evidence that the nuclei are generated by cells from circumscribed ventricular areas. CD15 also outlined several regions transiently; e.g. the perireticular nucleus and the submedial nucleus. The existence of transiently revealed nuclear structures supports the idea that the human thalamus traverses a developmental period, where the nuclear organization of the thalamus shows close similarities with that of the mature rodent. The topography of various nuclei is detailed by immunohistochemistry for other antigens, e.g. calretinin for the intralaminar, limitans, AV and mediodorsal nuclei; calbindin for the VA and VL nuclei; SMI-32 for the VPL and VA.

The synopsis of delineations of chemoarchitectonically specified areas by the various antibodies revealed extraordinary precise regional and temporal changes during development. Their registration not only enhances our understanding of the parcellation of the thalamus but reveals also insight into developmental mechanisms responsible for the complex adult organization. The close correspondence between the thalamus of the human and of rodents, which became evident in this study, opens new perspectives for achieving a unifying concept of mammalian thalamic organization.