Dying to Entrain: Regulating ipRGC Spacing

Abstract

In a recent issue of Neuron,Chen et al. (2013) show that apoptosis is required to ensure the even distribution of a class of retinal ganglion cells (ipRGCs), which sense luminance both intrinsically and through input from rods and cones. Disrupting apoptosis impairs photoentrainment mediated by rods/cones, but not that mediated by ipRGC-expressed melanopsin.

The retina is a complex neural network of more than 60 cell types. Neurons of each type are usually regularly spaced across the plane of the retina. The cell bodies of different neuronal types occupy distinct and well-defined retinal layers and synapse within two distinct plexiform layers (Figure 1) (Masland, 2012). The even distribution of each retinal cell type is independent of other cell types, even when they are synaptic partners. The dendrites of each cell type overlap or tile to achieve an even mosaic that ensures efficient coverage of the retinal surface. This represents one of the fundamental principles of functional organization in the retina: regular arrays of different neurons are responsible for processing and conveying distinct light information, such as intensity, contrast, color, and motion, to the brain.

Figure 1. Disrupted ipRGC Patterning Affects Circadian Photoentrainment.

(A) ipRGCs respond for photoentrainment at low light intensity to signals coming from rods and cones. Melanopsin-mediated intrinsic response contributes at high light intensities. In melanopsin mutants, ipRGCs only depend on the rod/cone pathway.

(B) In the wild-type, even spacing of ipRGCs ensures a uniform coverage across the visual field.

(C) Bax;melanopsin mutants establish more connections with inhibitory amacrine cells.

(D) In ipRGC conditional Bax;melanopsin mutants, not all ipRGCs receive bipolar input.

Like most neuronal populations, more than half of retinal ganglion cells (RGCs) undergo programmed cell death. While apoptosis contributes to the even spatial distribution of RGCs, the functional implications of disrupting this even spacing are not known (Raven et al., 2003). In a recent issue of Neuron, Chen et al. (2013) investigated how apoptosis affects the spatial distribution of a particular class of RGCs, the intrinsically photosensitive retinal ganglion cells (ipRGCs), and how this affects retinal circuitry, light detection, and circadian photoentrainment of locomotor activity.

In the mouse retina, rods and cones transform light energy into an electrical signal and convey this information to bipolar and horizontal cells at the level of the outer plexiform layer (Figure 1). In the inner plexiform layer, bipolar and amacrine cells, which integrate and modulate the output of bipolar cells, synapse with the RGCs, the only retinal cells projecting to the brain (Figure 1). ipRGCs constitute a small percentage of RGCs (∼2%); they express melanopsin, making them capable of autonomously depolarizing to light, even in the absence of rods and cones (for review, see Schmidt et al., 2011). Normally, ipRGCs integrate information derived from their own intrinsic melanopsin-based photosensitivity with the input from rods and cones. This represents the primary conduit of irradiance information to non-image-forming centers in the brain, such as the olivary pretactal nucleus (OPN), responsible for the pupillary light reflex, and the suprachiasmatic nucleus of the hypothalamus (SCN), the master circadian pacemaker (Berson et al., 2002; Hattar et al., 2002).

Chen et al. (2013) analyzed mouse mutants, in which RGCs do not undergo apoptosis, for the proapoptotic factor Bax and asked how this affects the mosaic distribution of ipRGCs. The authors first show that ipRGCs, which are initially randomly distributed at early developmental stages and normally progress to an evenly spaced distribution in wild-type mice, become clustered in Bax mutant mice. They suggest that apoptosis induced by homotypic proximity regulates ipRGC spacing. But what are the functional consequences of this disruption of ipRGC spatial distribution?

Photoentrainment of circadian rhythms requires the detection of overall changes in ambient light. Either extrinsic rod-cone signaling or melanopsin-mediated intrinsic photoreception is sufficient for ipRGCs to regulate photoentrainment (Hattar et al., 2003). The rod-cone pathway controls pupil constriction and photoentrainment at low light intensities while melanopsin contributes at high light intensities. Because ipRGCs are directly involved in irradiance detection, it has been assumed that an even distribution of their dendritic arbors is necessary to maximize photon capture to elicit photoentrainment, a so-called “photoreceptive net” (Provencio et al., 2002). However, by analyzing the photoentrainment behavior and electro-physiological response of ipRGCs in Bax/melanopsin double knockout mice, Chen et al. (2013) show, surprisingly, that disruption of ipRGC spacing impairs only rod-cone mediated photoentrainment and not entrainment mediated by the intrinsic melanopsin response of ipRGCs.

Do these morphological changes lead to physiological deficits in rod/cone input to ipRGCs? The authors show that the electrophysiological responses from the rods/cones to ipRGCs are significantly reduced, and this is consistent with the inability of Bax mutants to entrain to light in the absence of melanopsin. The authors propose that the reduced electro-physiological response is due to an increased inhibitory input from dopaminergic amacrine cells that are also GABAergic. This could counteract the rod/cone input on ipRGCs. However, the layering of the retina appears normal in conditional mutant mice lacking Bax only in ipRGCs, while these mice still exhibit a lack of entrainment by rods and cones. This suggests that ectopic synapses of ipRGCs with amacrine cells are not responsible for the observed entrainment deficits. In these mice, there is no increase in the number of ipRGCs but there are still deficits in cell spacing. Therefore, an alternative explanation might be that the disruption of the uniformity of retinal coverage by ipRGCs is responsible for the photoentrainment defects observed in Bax;melanopsin mutant mice. Because the number of individual synapses from ON bipolar cells to ipRGCs appears not to be affected, the disrupted spacing might leave regions deprived of ipRGC processes while clustered ipRGCs might not receive enough inputs from a limited number of bipolar cells, an arrangement that would lead to smaller responses to light (Figure 1). However, because Bax might affect other retinal cell types (White et al., 1998), it will be important to determine whether bipolar cells are normal.

Bax has also been shown to have a non-apoptotic function in synaptic transmission. It remains possible that some of the deficits observed in Bax mutants are due to Bax-dependent changes in synaptic activity (Jiao and Li, 2011). Future studies aimed at better understanding how Bax regulates apoptosis and the refinement of synaptic connections during retina development are needed to confirm this possibility.

In conclusion, this work reveals the fundamental role of apoptosis in the organization of ipRGCs. It further shows that although mosaic distribution of ipRGCs is dispensable for their ability to respond to light intrinsically, it is essential to establish the proper connections to bipolar and/or amacrine cells and thereby engage normal rod/cone-dependent circadian photoentrainment.