The Subplate, A Transient Neocortical Structure: Its Role in the Development of Connections between Thalamus and Cortex

The functioning of the mammalian brain depends upon the precision and accuracy of its neural connections, and nowhere is this requirement more evident than in the neocortex of the cerebral hemispheres. The neocortex is a structure that is divided both radially, from the pial surface to the white matter into six cell layers, and tangentially into more than 40 different cytoarchitectural areas (Brodmann 1909). For instance, within the cerebral hemispheres, sets of tangential axonal connections link neurons within a given cortical layer to each other and also link neurons of different cortical areas; sets of radial connections link neurons of different layers together. In addition, the major input to the neocortex arises from neurons in the thalamus, which in tum receive a reciprocal set of connections from the cortex. These connections are highly restricted: In the radial domain, thalamic axons make their major projection to the neurons of cortical layer 4, and the neurons of cortical layer 6 project back to the thalamus. Connections are also restricted tangentially, in that neurons located in specific subdivisions of the thalamus send their axons to specific cortical areas. For instance, neurons in the lateral geniculate nucleus (LGN) of the thalamus connect with primary visual cortex, whereas those situated in the ventrobasal complex connect with somatosensory cortex. There are also local patterns of connections within a given cortical area, for example, the ocular dominance columns in primary visual cortex of higher mammals, or the barrels in rodent somatosensory cortex (Woolsey & van der Loos 1970). The ocular dominance columns are based on the fact that the inputs of LGN axons representing the two eyes are segregated from each other in layer 4 and their terminal arbors are clustered together in patches (LeVay et al 1980). 
\n
\nA primary question is how these sets of connections form during development. The purpose of this review is to consider this question as it pertains specifically to the formation of connections between thalamus and cortex [for a more general review of the formation of connectivity, see Goodman & Shatz (1993)]. Several major steps are involved in this developmental process. First, the constituent neurons of the thalamus and cortex must be generated. Next, axons must grow along the appropriate pathways and select the appropriate targets. In the visual system, this means that LGN axons must grow up through the internal capsule, bypass many other inappropriate cortical areas, and then select visual cortex. Finally, the axons must enter the cortical plate, recognize and terminate within layer 4, and segregate to form ocular dominance columns. Thus, in addition to the general problems of pathfinding and target selection faced by all developing neurons, thalamic neurons are faced with a series of tangential and radial decisions as they form the final pattern of connections within neocortex: they must choose the correct cortical area and the correct layer, and must restrict the extent of their terminal arbors. In addition, similar problems must be solved by the neurons of cortical layer 6 as they grow towards and invade their thalamic targets. 
\n
\nA growing body of evidence suggests that the formation of connections between thalamus and cortex requires the presence of a specific and transient cell type, subplate neurons. These neurons are present early in development, but by adulthood the majority have disappeared. Here we consider their life history and review the evidence for their role in the patterning of connections.