RCS rats [98]. Messenger RNA for metabotropic glutamate receptors (mGlur6), which are likely expressed by ON bipolar cells, increase in the INL of pink-eyed RCS rats at the longest (P60 and P120) periods examined [1], suggesting an upregulation of these signal-inverting receptors. Physiological studies of RGCs in vitro have found many changes between p20 and p100, although there are conflicting results. Extracellular recordings in whole mount RCS rat retinas [86] demonstrated an increase in spontaneous activity up to P100, which coupled with decreased responses to light, resulted in significantly lower signal-to-noise levels. These investigators noted a predominance of cells with “OFF” responses by P47, and a decrease in receptive field size by P36. Intracellular recordings of RGCs in dystrophic RCS rat retinal slices, however, demonstrated a decrease in the number of cells with sustained responses. Action potentials could not be evoked in 62% of RGCs from 9 to 12-week old animals [11].

Recent studies on functional glutamate receptors in retinal degeneration [61] based on cellular uptake of organic cations (AGB) found significant and differential changes in retinal cell glutamate responses. In two models of retinal degeneration due to loss of photoreceptors, rodless/coneless mice (rd/rd cl) and rhodopsin knock-in mutation model mice (brboG), a severe loss of glutamate (kainate) sensitivity of bipolar cells was found in the late, stage 3 of degeneration and remodeling. Glutamate still activates amacrine and ganglion cells, although reduced in this late stage of degeneration. However, in small islands where apparently non-functional cones survive, bipolar cells exhibit ionotropic glutamate responses. A high number of bipolar cells activated by kainate suggest that rod bipolar cells begin to express iGluR, in comparison to normally expressing mGluR [61]. In addition, AGB uptake suggests that some amacrine and ganglion cells exhibit increased activity. In a single retinal sample from the posterior pole of a human male RP patient with 90–100% rod PR loss and remodeled cone PRs, all inner retinal cell types exhibited a robust glutamatergic response [61].

Overall, evidence from numerous studies indicates that despite decreased number and possibly excitability, surviving RGCs in retinas undergoing photoreceptor degeneration can transmit information to the brain. Glutamate receptor changes may reduce the efficacy of exogenous glutamate application, but this needs to be examined experimentally at specific points during the degenerative process.

9.4  Rationale for a Neurotransmitter-Based Retinal Prosthesis

Transmitter application may be a more effective and naturalistic means of conveying visual information to the brain, than other methods such as electrical stimulation. The effect of stimulation will depend on the location of application. In addition all types of retinal prosthesis and vision restoration strategies must be designed to stimulate the retina according to the type and stage of retinal degeneration.

Subretinal application of neurotransmitters, such as glutamate, in the normal retina would inhibit ON and activate OFF ganglion cells, but these physiological

effects would be superimposed upon the effects of continuous glutamate release from photoreceptors. In eyes with retinal degeneration, however, exogenous glutamate application could replace endogenous glutamate release lost due to PR cell death. Pulsatile glutamate application would activate OFF ganglion cells and inhibit ON cells. Following glutamate application, the disinhibition of ON bipolar cells, may elicit a rebound response. Therefore, differential stimulation of OFF and ON pathways could be achieved, but the signals would be reversed – i.e. OFF cells respond first during stimulation, and ON-cells respond at the offset. Continuous release of glutamate with reductions to mimic light responses could mimic normal photoreceptor releases. However, this would likely produce too high of a glutamate load on cellular systems which clear glutamate from the extracellular space, such as excitatory amino acid transmitter pumps (EAATs). The distance from the subretinal surface and OPL in normal retina is over 100 mm, whereas after PR loss this distance can be less than 50 mm.

Neurotransmitter application at the epiretinal surface can stimulate the retina by activating receptors in the ganglion cell layer (40–60 mm from surface) and in the IPL (60–75 mm from the surface). Epiretinal glutamate or acetylcholine application could directly activate RGCs through receptors on their somata, which are within 50 mm of the surface. Glutamate could also stimulate receptors in synapses within the IPL, including RGC dendritic fields, bipolar-ON cell synapses in IPL b, bipolarOFF cell in IPL a, rod bipolar-amacrine in IPL, and amacrine cells. Epiretinal application of GABA or glycine could be used to inhibit RGCs and amacrine cells. This may be useful if for example RGCs become highly active in degenerated retinas, as those observed in rd1 mice [97]. Inhibition of amacrine cells could result in disinhibition of other cells, including RGCs in the surround area, due to decreased inhibitory transmitter release. Our preliminary results indicate that ganglion cells exhibit robust excitatory responses to exogenously applied glutamate in 180 day RCS and s334ter line 4 rats. We also have observed that spontaneous firing rates of RGCs in these animals range from absent to high in degenerating retinas.

9.4.1  Limitations of Electrical Stimulation

Prostheses based on electrical stimulation of the retina have been under development over the past two decades. Testing in acute humans studies have had limited success in providing useful vision. Chronic human experiments have been limited to low-resolution devices, since large electrodes are required to handle the high currents required to stimulate degenerated retinal tissues. Small-diameter electrodes, required for a high-resolution prosthesis, are prone to failure due to high charge-densities that erode metals and stimulation voltages that often exceed those required to dissociate water. These facts make small-diameter electrodes more capable of inducing retinal tissue damage from free radicals that are toxic to the lipid membranes of neurons and glia. Further limiting the efficacy of current stimulation methods is the fact that electricity cannot selectively stimulate specific types