Optimal electrical activation of retinal ganglion cells

Retinal prostheses are emerging as a viable therapy option for those blinded by degenerative eye conditions that destroy the photoreceptors of the retina but spare the retinal ganglion cells (RGCs). My research sought to address the issue of how a retinal prosthesis might best activate these cells by way of electrical stimulation. Whole-cell patch clamp recordings were made in explanted retinal wholemount preparations from normally-sighted rats. Stimulating electrodes were fabricated from nitrogen-doped ultra-nanocrystalline diamond (N-UNCD) and placed on the epiretinal surface, adjacent to the cell soma. Electrical stimuli were delivered against a distant monopolar return electrode. Using rectangular, biphasic constant current waveforms as employed by modern retinal prostheses, I examined which waveform parameters had the greatest effect on RGC activation thresholds. In a second set of experiments intracellular current injection was employed to assess the effectiveness of sinusoidal current waveforms in selectively activating different RGC subsets. These recordings were also used to validate a biophysical model of RGC activation. Where possible, recorded cells were identified and classified based on 3D confocal reconstruction of their morphology. Electrodes fabricated from N-UNCD were able to electrically activate RGCs while remaining well within the electrochemical limits of the material. They were found to exhibit high electrochemical stability and were resistant to morphological and electrochemical changes over one week of continuous pulsing at charge injection limits. Retinal ganglion cells invariably favoured cathodic-first biphasic current pulses of short first-phase duration, with a small interphase interval. The majority of cells (63\%) were most sensitive to a highly asymmetric waveform: a short-cathodic phase followed by a longer duration, lower amplitude anodic phase. Using the optimal interphase interval led to median charge savings of 14\% compared to the charge required in the absence of any inter-phase interval. Optimising phase duration resulted in median charge savings of 22\%. All RGCs became desensitised to repetitive electrical stimulation. The efficacy of a given stimulus delivered repeatedly decreased after the first stimulus, stabilising at a lower efficacy by the thirtieth pulse. This asymptotic efficacy decreased with increasing stimulus frequency. Cells with smaller somas and dendritic fields were better able to sustain repetitive activation at high frequency. Intracellular sinusoidal stimulation was used to demonstrate that certain RGC subsets, defined on the basis of morphological type, stratification, and size, were more responsive to high frequency stimulation. Simulated RGC responses were validated by experimental data, which confirmed that ON cell responses were heavily suppressed by stimulus frequencies of 20 Hz and higher. OFF cells, on the other hand, were able to sustain repetitive activation over all tested frequencies. Additional simulations suggest this difference may be plausibly attributed to the presence of low-voltage-activated calcium channels in OFF but not ON RGCs. The results of my work demonstrate that (a) N-UNCD is a suitable material for retinal prosthesis applications; (b) a careful choice of electrical waveform parameters can significantly improve prosthesis efficacy; and (c) it may be possible to bias neural activation for certain RGC populations by varying the frequency of stimulation