Subplate neurons: A missing link among neurotrophins, activity, and ocular dominance plasticity?

The subplate is a transient structure comprised of a subset of the earliest neurons produced in the cerebral cortex (1). Although it has now been almost 30 years since the subplate was first described (2), a definitive function for the subplate remains unproven. In general, the subplate is believed to be important for the formation of connections between thalamus and cortex. Subplate neurons have been hypothesized to pioneer both the feedforward thalamocortical and the feedback corticothalamic pathways, guide selection of the correct cortical area by thalamic axons, participate in specification of cortical areas, and provide a transient target for thalamic axons until their target neurons in the cortical plate are born (3). In addition to these early developmental functions, the subplate may influence later events in the development of functional organization within the cortical plate. Specifically, subplate neurons may play a role in the formation of ocular dominance (OD) columns in primary visual cortex (4, 5). Despite the importance of this function, the cellular mechanisms that mediate subplate influences on thalamocortical connectivity remain a mystery. The paper by Lein et al. in a recent issue of PNAS (6) uncovers a molecular link between subplate neurons and cortical activity that helps to explain the role of the subplate in formation of OD columns.

Subplate neurons are not only the first neurons generated in the cortex, they are also the earliest to mature, differentiate, and make functional connections. Subplate neurons are the first cortical neurons to receive synaptic inputs from thalamic axons (refs. 4, 7, and 8; Fig. 1). During fetal development, thalamic axons invade the subplate and wait there for a significant period of time, ranging from 2 weeks in the cat to a month in the primate, before invading the cortical plate and forming synaptic connections with layer 4 neurons (9–12). The purpose of this waiting period, what controls its duration, and the nature of the physiological properties of synapses within the subplate during this period remain unknown. In addition to receiving thalamic input, subplate neurons project into the cortical plate and reciprocally innervate the thalamus (refs. 13–19; Fig. 1). Because of this anatomy, many of the proposed functions for the subplate focus on the possibility that it forms a crucial synaptic link between thalamic axons and their final target in layer 4.

Figure 1.

Subplate neurons may influence OD column formation by forming a crucial synaptic link between thalamic axons and their final target in layer 4. By embryonic day 50 (E50) in the cat, LGN axons are present in the subplate and form synapses there (4, 7, 8, 16). Subplate neurons also project into layer 4 of the cortical plate and back to the LGN (7, 14, 15, 19). During the first postnatal week in the cat, axons from the LGN begin to leave the subplate and make synaptic connections in cortical layer 4 (4). A single geniculate axon can arborize in both subplate and layer 4 (19). Subplate neurons continue to project into layer 4 and back to the LGN (13–19), and layer 4 neurons send a transient axonal projection back into the subplate at this time (23). By postnatal week 10 in the cat, most subplate neurons have died and the geniculate axons have segregated into ocular dominance columns (3, 21).

After leaving the subplate, thalamic axons from the lateral geniculate nucleus (LGN) that encode information from the two eyes invade the cortical plate, initially overlapping in layer 4 of primary visual cortex. Over time, these axons undergo a process of remodeling and selective growth leading to their segregation into eye-specific patches called OD columns (20–22). Studies from numerous laboratories have demonstrated that this process is activity dependent and involves competition between thalamic axons from each eye for synaptic territory in layer 4 (22). However, the precise cellular and molecular mechanisms underlying this competition also remain unknown.

An intriguing hypothesis, and one that often is overlooked, is that subplate neurons play a critical role in the formation of OD columns. Subplate neurons are clearly present in the right place at the right time to influence both thalamic axons and their target layer 4 neurons (ref. 3; Fig. 1). A subset of subplate neurons elaborate extensive axon collaterals in layer 4 at the same time that LGN axons form their initial connections in layer 4 (7, 16, 19) and some layer 4 neurons send a transient collateral that branches within subplate at this time (23). The most compelling evidence that the subplate is involved in OD column formation comes from a series of studies from Shatz and colleagues (4, 5) who directly tested this hypothesis by ablating the subplate with kainic acid. Local injections of kainic acid into the subplate selectively kills subplate neurons, while leaving most cortical neurons and subcortical structures largely unharmed when performed during a specific early postnatal time window (refs. 4 and 5; see supplementary web material referenced in ref. 6). OD column formation was prevented when subplate neurons were ablated at a time after LGN axons had left the subplate and formed connections in layer 4, but before they had segregated into OD columns (4, 5).

Although these experiments indicate that subplate neurons are essential for OD column formation, the underlying cellular and molecular mechanisms have until recently been unclear. The results presented by Lein et al. (6) are significant because they demonstrate a molecular link between subplate neurons and cortical activity that may explain the effects of subplate removal on OD column formation. The authors show that deletion of subplate neurons leads to an increase in the expression of a growth factor, the neurotrophin brain-derived neurotrophic factor (BDNF), in the precise region of visual cortex in which OD columns fail to form (6). Because neurotrophins are known to alter both synaptic activity and OD column formation (24, 25), they are likely to be a key molecular signal through which subplate neurons influence thalamocortical connectivity.

Recently, much attention has been focused on neurotrophins as potential molecular signals that mediate synaptic strengthening and directed sprouting of axons during activity-dependent development (25). The neurotrophins are a small gene family made up of at least four structurally and functionally related proteins: nerve growth factor, BDNF, neurotrophin-3 (NT-3), and neurotrophin-4 (NT-4). BDNF, in particular, is an especially attractive candidate for mediating the effects of neural activity in the developing visual cortex because it is expressed in the right place and at the right time to influence OD column formation, because its expression is regulated by neuronal activity, and because it can directly alter both synaptic activity and neuronal morphology (25).

The current model for the function of neurotrophins in OD column formation is that, during the critical period when OD columns form, LGN afferents compete for a factor (BDNF) that is secreted in limited amounts and that is necessary for the maintenance or strengthening of synapses (25–27). Although the evidence is mainly circumstantial, there are a number of observations supporting the hypothesis that neurotrophins are the object of this competition. First, either adding or removing neurotrophin from visual cortex prevents OD column formation. Local cortical infusion of excess BDNF or NT-4 (28) or of TrkB-IgG fusion proteins, which remove endogenous BDNF and NT-4, prevents segregation (29). Second, neurotrophins prevent the effects of monocular deprivation on OD column formation. When animals are deprived of vision in one eye during a critical period, cortical neurons become responsive mainly to the nondeprived eye (30). In general, infusion of nerve growth factor, BDNF, or NT-4 prevents this shift (31). All of these results are consistent with the idea that too much or too little neurotrophin attenuates competition in layer 4 and thereby alters OD plasticity.

The premise for the experiments in the paper by Lein et al. (6) was that the subplate might mediate OD column formation by altering BDNF levels in layer 4 of visual cortex. Because subplate neurons provide excitatory input to layer 4 cortical neurons (7, 16, 19), ablating the subplate should alter levels of neural activity, which, in turn, will modify BDNF expression and thereby influence OD column formation (28, 29). Indeed, their primary finding is that ablation of subplate neurons causes a delayed, but enduring, increase in the expression of BDNF in all layers of the cortex (6). Conversely, NT-3 mRNA is dramatically decreased (6), an intriguing result in light of the documented decrease in NT-3 expression induced by seizures (25) and the opposing effects of BDNF and NT-3 on cortical neuron differentiation (32). The area of cortex in which BDNF expression was increased corresponded precisely with the area of disrupted OD columns, suggesting that the increased BDNF levels caused by subplate neuron ablation may directly prevent LGN axon segregation (6).

In addition to providing a molecular explanation for the effects of subplate ablation on OD column formation, the results described by Lein et al. (6) may provide important insights into the nature of the connections from subplate to cortex. Subplate neurons are believed to provide excitatory input to layer 4 (7, 16, 19) and often are assumed to form synapses on pyramidal neurons. If this is true, subplate ablation should decrease excitatory inputs into layer 4, which, in turn, should decrease BDNF expression. Yet, the results presented by Lein et al. clearly show a strong increase in BDNF mRNA (6). This discrepancy is likely a result of the limited physiological characterization of subplate connections into layer 4. It is possible that subplate neurons mainly innervate inhibitory neurons in layer 4; if so, subplate ablation would cause a decrease in cortical inhibition leading to overall increased excitation, ultimately resulting in an increase in BDNF, like that shown by Lein et al. (6). Such an increase in cortical excitation and BDNF should cause a compensatory increase in inhibition to maintain a balance of activity in cortex (33). Consistent with this idea, Lein et al. (6) also show a strong increase in immunoreactivity for a marker of inhibitory neurons (GAD-67) in response to subplate ablation.

Although the present results strongly imply a role for subplate neurons in OD column formation, the nature of this role will be difficult to determine based solely on ablation studies. Lein et al. (6) carefully document the selectivity of the effects of kainic acid on subplate neurons, but they cannot discount the possibility that their results reflect cellular interactions that do not normally occur during OD column formation. Despite the inherent limitations of ablation studies, though, it is important to remember that this approach in general has provided critical indications that a particular neuronal population is involved in a specific biological phenomenon. In fact, the experiments of Lein et al. do indicate that the subplate influences development of circuits within the cortical plate at some level. The next challenge will be to elucidate the cellular and molecular mechanisms underlying this effect. The results in the paper by Lein et al. are a first step in this direction.

How might the subplate normally influence connectivity within the cortical plate, and specifically within layer 4? It is difficult to discuss relevant models for subplate function when so little is known about circuitry within the subplate, the anatomical and physiological properties of subplate inputs to cortical neurons, and the reciprocal connections of cortical neurons back to the subplate. The anatomy and physiology of these circuits should provide strong clues as to the role of the subplate in influencing functional development within the cortical plate. Once that information is available, experiments based on more subtle manipulations of activity in the subplate will enable a better understanding of the role of the subplate in influencing intracortical connectivity.

The idea that the subplate is more than just a temporary place for thalamic axons to wait passively before entering cortex—that it may in fact play an active role in development of the functional organization within the cortical plate—is a potentially significant conceptual advance in our understanding of cortical development. Subplate projections into layer 4 are in a perfect position to modify and guide the connectivity of thalamic axons in layer 4. Yet, current models that explain OD column formation tend to ignore any role for the subplate. If the anatomy and physiology of this circuit ultimately support a strong subplate influence on thalamocortical connections, our current ideas for how OD columns form will have to be significantly modified to include an important role for the subplate. Future studies focusing on the cellular and molecular mechanisms of subplate function, like the study by Lein et al. (6), finally should provide us with a definitive function for this elusive transient cortical layer.