Gap junctions are a ubiquitous feature of the mammalian brain, and while researchers have been gaining an appreciation for their physiological complexity over the last several decades, many of their functional roles within identified neural circuits remain elusive. All five primary neuronal cell types in the retina (photoreceptors, horizontal, bipolar, amacrine and ganglion cells) make gap junction-mediated synaptic connections, providing an approachable model system for exploring the roles that gap junctions play in neural signal processing.
Already, some of these electrical connections, like the coupling of AII amacrine cells to ON cone bipolars, have been placed within known functional circuits. Along with the understanding that electrical synapses are vital in the retinal processing of a visual scene has come the realization that gap junctional coupling in the retina is plastic. A number of studies have shown that coupling strength between retinal neurons changes with mean illumination. Previous work on electrical synapses in photoreceptors, horizontal cells, amacrine cells (ACs) and ON cone bipolar cells, has pointed to a decrease in coupling with light exposure and suggests that this mechanism underlies a functional change in the processing of signals and noise. The results from some of these experimental and computational studies indicate that in dark conditions, when signals are sparse, enhanced coupling reduces noise while increasing correlations and in light-adapted conditions decreased coupling improves the independence of signalling in parallel pathways. Molecularly, dopamine has been identified as a key regulator of gap junction coupling in the retina. It has been demonstrated that in rabbit, retinal dopamine levels increase with ambient light (Mills et al. 2007) and that dopamine modulates electrical coupling between retinal neurons via activation of D1-like and D2/4-like receptors. Recent molecular and pharmacological studies indicate that the activation of these two metabotropic receptor types leads to opposing actions on common elements in the biochemical signalling cascades that ultimately lead to changes in gap junction permeability.
In a recent article in The Journal of Physiology, Hu et al. (2010) explored both function and mechanism in their study of the modulation of gap junctions in retinal ganglion cells (GCs). They targeted ON α- and OFF α-GCs in rabbit and mouse and measured connectivity in both dark- and light-adapted conditions using standard tracer-coupling techniques. In the dark-adapted retina, Hu et al. found that OFF, but not ON α-GCs, were tracer coupled to neighbouring OFF α-GCs and a set of four distinct AC subtypes. In contrast to the decreased coupling observed in other retinal cell types, OFF α-GCs in retinas that had been exposed to mesopic or photopic light levels showed a large increase in the number of coupled cells. Importantly, the coupling radius expanded, but only the same types of coupled cells were observed, and the ON α GCs remained uncoupled in the light. Extracellular recordings from pairs of GCs showed spike time correlations that were consistent with their tracer coupling results.
In an effort to assess the impact of this apparent change in coupling strength, the authors measured receptive field widths of OFF α-GCs in dark- and light-adapted conditions, but they observed no significant difference. They also performed experiments aimed at the molecular mechanism underlying the observed coupling effects using agonists and antagonists of D1- and D2-type dopamine receptors. They found that D2 antagonists dramatically increased coupling in the dark, but surprisingly, in light-adapted conditions when coupling was relatively high, D2 agonists did not reverse the coupling effect. Instead, a D1 antagonist reduced the extent of ganglion cell coupling to other ganglion cells and amacrine cells. From this they inferred a model in which D1-like and D2/4-like receptors compete for control of the common biochemical signalling components that regulate connexin phosphorylation and gap junction permeability.
The results presented by Hu et al. (2010) are intriguing, but several additional experiments could help provide a more complete functional picture of the role of dynamic gap junctional coupling. Most critically, the strength and impact of electrical coupling between neurons remains unclear. Although neurobiotin tracer coupling supports the presence of gap junctions between neurons it does not reveal their functional ability to pass electrical signals. Acknowledging this fact, the authors relied on the strength and shape of the spike cross-correlogram to assess the functionality of gap junction coupling. However, many factors may control the strength and shape of the cross-correlogram, thus limiting interpretation. For example, non-linearities in the circuitry and changes in output statistics, such as increases in firing rates, have been shown to increase measured correlations even with no increase in common input or electrical coupling (de la Rocha et al. 2007). Additionally, common input as well as gap junctions can contribute to spike correlations (DeVries, 1999; Trong & Rieke, 2008). Although the strong split peak in the example spike cross-correlograms is indicative of a gap junction-mediated contribution, it does not rule out an equally strong contribution from common input. In contrast to the current study, a previous paper reported that common input underlies strong correlations between ON α-GCs in rabbit (DeVries, 1999). Thus, properties other than electrical coupling may underlie the changes observed in the spike cross-correlogram. In future studies, whole cell patch recordings in pairs of OFF α-GCs could be used to measure directly the extent and filtering characteristics of electrical coupling between cells and shared synaptic input. Such measurements will lead to a more direct assessment of contributions from common input and/or electrical coupling to the measured spike correlations.
Future study is also needed to improve our understanding of how coupling impacts the spatial receptive field of the ganglion cells. The authors use a rectangular slit of light to probe the extent of the Gaussian receptive field profile and observe no significant difference between the light- and dark-adapted conditions. This result may seem curious because one might expect that the strong increase in correlated firing observed between neighbouring cells would necessarily cause an increase in the size of the receptive field profile (i.e. if spikes in a distant ganglion cell induce spikes in the cell you are observing, the receptive field should be extended towards this neighbouring cell). Consistent with this intuition, a previous study in these cells noted a significant contribution of correlated activity to the size of the receptive field profile (DeVries, 1999). However the results attained by Hu et al. could be explained if gap junctions alone are not sufficient to induce spikes in a neighbouring cell and if the correlated spiking relied on the presence of other synaptic conditions conducive to synchrony. Thus, the impact of electrical coupling on receptive field profiles may depend on the stimulus used to probe the receptive field.
The results concerning the molecular mechanisms of gap junction plasticity also raise a number of important questions that need to be resolved in order to provide clear support for any particular model. In contrast to a previous study (Mills et al. 2007), Hu et al. concluded that net electrical coupling is enhanced under mesopic and photopic conditions but were ambivalent as to which specific gap junctions are increased, either GC–GC, GC–AC or AC–AC. This particular confusion is in part due to the lack of sufficiently detailed information regarding the surface expression of D1- and D2/4-type receptors on any or all of the coupled neurons and the types of connexins involved. To account for their results, Hu et al. proposed a competition model between the activation of D1- and D2-type receptors on individual cells. This proposed molecular arrangement would produce the observed changes in coupling by exploiting differences in the sensitivity and desensitization/inactivation of the dopamine receptor subtypes. However, it is also possible that the dopamine receptors are not regulated by global dopamine signals but instead D1- and D2/4-type receptors are differentially modulated by local changes in dopamine release. Experiments on dissociated cells in combination with molecular assays and pharmacological manipulations could further elucidate the mechanisms by which dopamine regulates electrical connectivity. Using these tools, researchers could more accurately assess receptor expression and more precisely control extracellular dopamine concentrations.
Critique aside, the work of Hu et al. provides a set of intriguing observations and raises important questions about the functional role of these electrical synapses. The authors speculated that such synchrony may contribute to increasing the efficiency of information transmission down the limited bandwidth of the optic fibre during the daytime. In order to accurately assess the validity of this hypothesis, future work will need to show that more information about a set of stimuli can be gleaned from taking into account the synchronous activity of electrically coupled populations of GCs. On the other hand, gap junction regulation may not primarily function to increase GC synchrony but instead may act to strengthen electrical synapses between ACs and GCs. Understanding the functional roles gap junctions play in neural coding throughout the nervous system will not be easy. However, retinal studies like this one provide tantalizing clues by characterizing the cell types and stimulus conditions under which gap junctions are utilized and modulated. Thus, in the future, the retina may provide a framework for understanding dynamic electrical networks in other areas of the nervous system.