Abstract
Purpose: :
This work is related to the efforts of the Boston Retinal Implant Project to develop a sub-retinal prosthesis to restore vision to the blind. Although, subretinal stimulation is thought to activate the retinal network, the full range of responses that can be elicited via indirect activation is not well understood. For example, the number of action potentials, their arrangement into ‘bursts’ and/or the time course over which these occur have not been described. Further, it has not yet been established whether certain parameters of stimulation are more effective than others. Here we systematically compared responses to a wide range of stimulation parameters to reveal those that mostly strongly activate the network response.
Methods: :
Cell-attached patch clamping was used to record spikes from rabbit retinal ganglion cells in the isolated rabbit retina. Electrical stimulation was delivered subretinally via a 400 μm diameter disc electrode. Stimulus waveforms were anodic-first biphasic current pulses delivered at a frequency of 0.625 Hz with an interval of 600 ms between the anodic and cathodic phases. We recorded the response of RGCs (n = 17) to phase durations of 0.1, 0.3, 1.0 and 3.0 ms with amplitude levels ranging from 0 to 100 μA. In addition, phase durations of 10 ms (0 - 30 μA) and 50 ms (0 - 20 μA) were also tested.
Results: :
For a given pulse amplitude, the strongest responses (most spikes) were elicited by 3 ms phase durations. Responses were weaker for phase durations > 3 ms even when the total charge delivered was substantially (i.e. 3x) larger. For a given charge level, responses generally increased as the phase duration decreased. Responses consisted of only a few spikes for the shortest phase durations (0.1 & 0.3 ms) regardless of stimulus amplitude. Elicited spikes were generally clustered into bursts although the number of spikes within a burst was highly variable and typically varied systematically with changes in amplitude. The latency of a given burst also varied with amplitude
Conclusions: :
Our results indicate that burst firing in RGCs is generated over a wide range of both pulse amplitudes and durations. The fact that 3 ms pulses generate the largest number of spikes in vitro raises the possibility that they may also be more effective clinically. The constant charge comparison also suggests that at a given charge density higher charge injection rates are more advantageous for activation of the retinal network.
Keywords: electrophysiology: non-clinical • ganglion cells • retinal connections, networks, circuitry