Segmentation is a recurring theme in developmental biology, and most of us are familiar with repetitive structures such as the somites and the hindbrain rhombomeres. In 1993, Puelles and Rubenstein proposed a segmental, or neuromeric, model for the forebrain, based largely on gene expression patterns. But how do the properties of these putative neuromeres compare with those of the rhombomeres? Larsen et al. have addressed this question by looking at the developing chick diencephalon.

The diencephalon consists of three subdivisions — the ventral thalamus, the dorsal thalamus and the synencephalon. The dorsal and ventral thalami arise from the parencephalon, which becomes bisected by a prominent boundary, the zona limitans intrathalamica (zli). The subdivisions can be distinguished on the basis of neuronal distribution and axonal projection patterns, as revealed by Nissl staining. However, as Larsen et al. show, the diencephalon is not overtly segmented. Whereas the rhombomeres appear as a series of bulges (gyri) separated by ridges (sulci), only the synencephalon is delineated by sulci in the diencephalon. Each rhombomere has a distinct gene expression profile, and the diencephalic subdivisions are similar in this respect. For example, expression of the homeobox gene Prox demarcates the synencephalon, whereas expression of Gbx2 and Dlx2 is confined to the dorsal and ventral thalamus respectively. Conversely, as shown recently by the same group, the zli is defined by the absence of lunatic fringe (L-fng) expression.

The rhombomeres show alternating adhesive properties, with the result that cells do not intermingle between adjacent segments. Lineage restriction between rhombomeres is consolidated by the formation of specialized boundary cells that express chondroitin sulphate proteoglycan (CSPG), tenascin and vimentin. Larsen et al. tested whether the diencephalic boundaries also inhibit cell mixing and/or express boundary cell markers. They showed that only the cells bordering the zli and the midbrain–synencephalic junction express boundary markers and present barriers to cell mixing. The boundary between the synencephalon and the parencephalon transiently expresses boundary markers, but the cells from these two territories are able to mix freely.

Larsen et al. have shown that the subdivisions of the diencephalon do not fulfil all the criteria that would define them as true segments. So, despite efforts to ascribe neuromeric properties to other regions of the developing nervous system, it seems that the segmental pattern of the hindbrain remains the exception rather than the rule.