All mammalian behavior is generated and regulated by the nervous system. In humans, the neocortex is the structure within the nervous system that is responsible for the complex integration of information, the ability to utilize language, decision-making, motivation and other high-level emotive-cognitive processes. The complexity of the neocortex emerges during development through arealization, when specific sensory and motor functional units, or areas, are formed and connected to one another and to sub-cortical nuclei through a vast and complex network of intra- and extra-neocortical connections.
Research on the developmental mechanisms that drive arealization has been influenced by two alternative hypotheses. Rakic  famously detailed his protomap hypothesis, suggesting that the fates of different neocortical regions were pre-specified in early development, by yet-to-be characterized molecules within the proliferative zone. The alternate model, coined the protocortex hypothesis, emphasized the role of neural activity, via neocortically extrinsic thalamic sensory input, in determining neocortical areal fate . In the past 20 years, a consensus has formed in the field of neocortical developmental biology that both activity-independent cortically intrinsic mechanisms, such as gene expression, and activity-dependent mechanisms that involve input from the sensory organs via the dorsal thalamus interact to form the cortical map. The exact nature of the interaction, however, is not known.
Despite this consensus, most studies focus on one side of the argument or the other. For example, the notion that the developing neocortex is patterned early in development, regardless of driven sensory input, with differential expression of genes during arealization is highly supported [1, 3–28]. Specifically, the areal patterning period (APP), or the time in which the major structural features of the developing sensory and motor areas are established, has recently been described and defined as from embryonic day (E) 16.5 to the third postnatal day (P) in mice , prior to eye opening and active whisking. Additionally, the absence of thalamocortical afferents (TCAs) during the prenatal portion of the APP (E16.5 to birth), such as in Gbx2 or Mash1 mutant mice, leaves neocortical gene expression patterns unperturbed, downplaying the role of activity in neocortical patterning [4, 5, 29]. However, Krubitzer and colleagues demonstrated the impact of the removal of a sensory receptor surface on arealization in an elegant series of very early postnatal enucleation experiments in Monodelphis domestica. In these studies, clear expansions of auditory and somatosensory cortical areas into visual cortical regions were described in the adult [30–32]. Moreover, the relative use of different sensory modalities has been correlated with relative areal size , indicating that increased activity can also alter the cortical map.
Although changes to the molecular properties of the cortex (via knockout or electroporation studies) and changes in activity from thalamic sensory inputs, including spontaneous lateral geniculate nucleus (LGN) activity [34–36], can independently affect the cortical map either during the APP or during the critical period, it is important to better understand how one factor may impact on the other. Molecular cortical gene expression and spontaneous activity occur together very early in development, during embryogenesis. In the visual system, for example, spontaneous activity from the retina is transmitted to the developing cortex via the thalamus as early as E16 in mice and continues into the postnatal stage [37, 38]. Additionally, studies of gene expression patterns suggest that the collaboration between intrinsic and extrinsic factors is extensive; we have previously shown that transcripts for several genes are present throughout embryogenesis and the first three weeks of life [28, 39]. It is likely that each may be required in an ongoing fashion to complete successive steps of arealization, including those that are influenced by incoming activity from sensory receptor surfaces. Studies of chicken ovalbumin upstream promoter transcription factor 1 (COUP-TF1) have demonstrated multiple functions in the development of the somatosensory cortex [40–42]; however, these reports did not address the possible influence of afferent sensory input.
The visual cortex has been used extensively as a model to study aspects of both the protocortex and protomap hypotheses, as changes in visual function can easily be tracked through alterations at multiple levels of the nervous system. Prior to eye opening, during preliminary developmental stages, neocortical gene expression is thought to drive initial arealization and targeting of TCAs and ipsilateral intraneocortical connections (INCs) [6, 28, 39, 43, 44], with eye opening and subsequent sensory experience guiding detailed features of visual cortex development . The organization and connections of neurons within the visual cortex of visually impaired animals display sparse abnormal subcortical inputs from the posterior nuclei and the ventral lateral, ventral posterior and anterior thalamic nuclei . Animals lacking visual input display abnormal dendritic spine density as well as altered gene expression and regulation [47, 48]. Further evidence demonstrates that bilateral enucleation during early development results in reduced overall brain size, a reduction in the size the of the primary visual cortex yet an increase in the size of the somatosensory cortex, suggesting that the relative sensory activity level determines major features of cortical organization [32, 49]. Although bilateral enucleation as an experimental manipulation can affect multiple systems, it has been used extensively as a way to remove retinal input to the developing brain [31, 32, 48, 50].
In the present study we investigated the impact of very early bilateral enucleation on the cortical expression of several genes that have been implicated in arealization or topographic patterning, either through mutation or correlative studies: Cadherin 8 (Cad8) previously shown to delineate distinct neural pathways and cortical areas ; COUP-TF1, which is required for proper regionalization and corticospinal motor neuron differentiation [42, 43]; ephrin A5, strongly expressed in the putative somatosensory cortex and implicated in topographic patterning [28, 39, 52, 53]; inhibitor of DNA binding 2 (Id2), a well-established gene to study positional identity ; LIM homeobox protein 2 (Lhx2), required for regional specification ; and retinoic acid receptor related orphan receptor beta (RORß, also known as RZRß), a highly specific marker for the primary sensory areas . We found that, with short-term survival after enucleation from P0 to P10, the dorsal LGN (dLGN), which normally receives direct retinal input, was reduced in size with altered gene expression in the nucleus. Anatomical 'errors' in the postnatal development of INCs were present at the medial somatosensory-visual (S-V) area boundary in the neocortices of enucleated mice. These aberrant INCs stemming from dye placements within the somatosensory cortex co-registered with a positional shift in ephrin A5 expression. Our results demonstrate a link between altered sensory input via bilateral enucleation and aberrant gene expression and anatomical development in the neocortex and suggest a possible role of spontaneous retinal activity in the formation of the neocortical S-V area boundary. In this report, we discuss the potential roles that gene expression and sensory activity may play in critical period plasticity as it relates to areal boundaries and physiology.