Development of cortical circuits: Lessons from ocular dominance columns

Abstract

Neuronal development is divided into a sequence of events that leads from the initial specification of neuronal cell fate to the emergence of adult circuits. The initial organization of neural circuits relies on molecular cues that guide axons to generally appropriate regions, but the final specification of patterned connections is widely held to depend on patterns of neuronal activity generated by circuits that are intrinsic to the developing brain, or by early experience. In the central nervous system, much of this sculpting of neuronal connections is thought to occur during 'critical periods', when circuits are particularly susceptible to external sensory inputs. Despite the powerful appeal of this model, and the experimental support that accumulated over several decades, recent findings indicate that some of the assumptions underlying the conventional formulation might need to be revised. Hubel and Wiesel described ocular dominance columns in the early 1960s, noting that in the cat primary visual cortex, cells with similar eye preference were grouped together into columns, and eye dominance shifted periodically across the cortex. They distinguished between the innate mechanisms guiding the initial formation of cortical functional architecture, and the experience-dependent, competition-based mechanisms responsible for later modification during the critical period. A substantial alteration in this formulation occurred when transneuronal transport techniques made it possible to directly visualize ocular dominance columns. It was suggested that the precise organization of columns in layer 4 was not innately specified, but was gradually moulded by correlation-based synaptic competition. Early work in macaque monkeys indicated that thalamocortical afferents are arranged into functional columns by the time of birth. Retinal waves — spontaneously generated, correlated patterns of activity that course across the neonatal retina — could provide the patterns of activity necessary to segregate thalamic afferents in the cortex. However, if both eyes are removed before the layers in the lateral geniculate nucleus (LGN) have segregated, columns of layer-specific LGN afferents still form in the cortex, perhaps indicating a role for correlated activity in the LGN. Stryker and Harris used tetrodotoxin to block retinal activity in cats from postnatal day 14 (P14) to P45. In treated animals, there was no evidence of segregated columns at P45, indicating that blocking retinal activity prevented the normal activity-driven competition that should have resulted in segregation. However, it is now clear that geniculocortical afferents are already segregated by P14, so the activity blockade probably desegregated columns that were already present. A model emerges in which columns form well before the critical period and with limited production of exuberant projections. During this initial stage, ocular dominance columns do not seem to respond to changes in activity as predicted by simple Hebbian rules. The main role of visual experience during the critical period might be to reinforce and augment an already appropriately situated set of basic connections, rather than to instruct their de novo formation. To unravel how, or whether, activity cues and molecular patterning information interact to drive column formation will require a leap of faith that such patterning information actually exists. Some 40 years after Hubel and Wiesel suggested innate mechanisms for the development of cortical functional architecture, an intriguing system of specification remains to be fully elucidated.