| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Departments of Vision Science and 2Bioengineering and Molecular and 3Cell Biology, University of California, Berkeley, California
Submitted 11 August 2005; accepted in final form 12 October 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
1-ms duration) elicited activity in neurons distal to the ganglion cells; this resulted in ganglion cell spiking that could last as long as 100 ms. However, short pulses, <0.15 ms, elicited only a single spike within 0.7 ms of the leading edge of the pulse. Trains of these short pulses elicited one spike per pulse at frequencies 
250 Hz. Patterns of short electrical pulses (derived from normal light elicited spike patterns) were delivered to ganglion cells and generated spike patterns that replicated the normal light patterns. Finally, we found that one spike per pulse was elicited over almost a 2.5:1 range of stimulus amplitudes. Thus a common stimulus amplitude could accommodate a 2.5:1 range of activation thresholds, e.g., caused by differences arising from cell biophysical properties or from variations in electrode-to-cell distance arising when a multielectrode array is placed on the retina. This stimulus paradigm can generate the temporal resolution required for a prosthetic device. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Curcio et al. 1996; Friedman et al. 2004). Currently there is no cure for either disease. Loss of visual function arises from a degeneration of the outer retina, primarily the photoreceptors, but these diseases do not target inner retinal neurons; large populations of inner retinal neurons remain morphologically intact, even in patients blind for many years (Humayun et al. 1999; Santos et al. 1997; Stone et al. 1992; but also see also Jones and Marc 2005; Marc and Jones 2003; Marc et al. 2003). During clinical trials in patients blind from retinal degenerative diseases, electrical stimulation from electrodes placed close to the ganglion cell layer elicited light percepts (Humayun et al. 1994, 1996; Rizzo et al. 2003a,b). These findings suggest that ganglion cells (the output cells of the retina) not only remain morphologically intact, but remain functionally viable as well.
Several research groups are developing retinal prosthetic devices with the hope of providing meaningful visual information to blind patients (Chow et al. 2004; Humayun et al. 2003; Rizzo et al. 2003a,b; Zrenner et al. 1997). To be effective, these prosthetic devices should generate patterns of activity that resemble patterns normally evoked by light. This means that the device should be capable of replicating the temporal properties of light-elicited spike trains from normal retina, while spatially confining the response to a focal region—ideally a single ganglion cell. In addition, there are 
12–15 different types of ganglion cells (Rockhill et al. 2002; Roska and Werblin 2001); populations of each type generate unique spiking patterns that are carried to distinct downstream visual sites (Roska and Werblin 2001) (both cortical and noncortical areas). This means that the prosthetic device must also be capable of targeting the appropriate spiking pattern to the correct ganglion cell subtype.
There are several challenges associated with replicating the temporal properties of normal light-elicited spike trains. For example, maximum spike rates in normal retina can exceed 250 Hz (O'Brien et al. 2002), but not all cell types respond with an equal number and/or pattern of spikes. Therefore the prosthetic device must be capable of generating a wide range of spike frequencies and patterns. Even within a single cell type, there are considerable variations in the light-elicited spike trains, such as changes in spike rates used to code information about stimulus contrast (Smirnakis et al. 1997). In addition, several studies now indicate that spike timing between multiple retinal neurons is often very precise (Mastronarde 1989; Meister et al. 1995), suggesting that prosthetic devices may need to generate spikes with high temporal precision.
Electrical stimulation of the retina activates multiple classes of retinal neurons complicating the generation of precise temporal patterns of spiking. Jensen et al. (2005a) reported that the spiking response to electrical stimulation consists of short- and long-latency components and that the latency of the late component ranged from 8 to 60 ms. They further showed that the long component arose from activation of bipolar cells, excitatory presynaptic neurons in the inner nuclear layer. In response to light, bipolar cells deliver excitatory input to both ganglion cells and amacrine cells (see Wassle 2004 for a review). Activated amacrine cells deliver inhibitory input in a feed-forward manner to ganglion cells and through feedback to bipolar cells. The precise interplay between these excitatory and inhibitory signals shapes the retinal response. Because electrical stimulation targets bipolar cells, it is possible that amacrine cells are targeted as well. Activation of amacrine cells is likely to lead to long-lasting inhibitory signals feeding both forward and backward and therefore may lead to many undesirable effects, such as increased threshold levels.
Here, we developed a method of electrical stimulation that generates precise temporal patterns of spiking activity in retinal ganglion cells. We show that the early phase response occurs within 0.7 ms of the onset of the cathodic pulse. In response to pulses of electrical stimulation, we measured excitatory and inhibitory currents in ganglion cells, indicating that both bipolar and amacrine cells were activated. Reducing the duration of the stimulus pulse reduced the excitatory and inhibitory currents, and we found that the late phase spiking response was eliminated when using very short pulses. Short pulses therefore elicit one spike per pulse with a predictable spike latency of 0.7 ms. One spike per pulse was elicited at pulse frequencies 250 Hz—comparable with the highest normal ganglion cell spike frequencies. To confirm the validity of these findings, we measured light elicited spike trains and replicated individual trains using preprogrammed arrays of short pulses.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
The care and use of animals followed all federal and institutional guidelines, and all protocols were approved by the Animal Care and Use Committee, UC Berkeley. New Zealand white rabbits, 2.5 kg, were anesthetized with injections of xylazine/ketamine and subsequently killed with an intracardial injection of pentobarbital sodium. Immediately after death, the eyes were removed. All subsequent procedures were performed under dim red illumination. The front of the eye was removed, the vitreous was eliminated, and the eye cup was dissected so that the visual streak and regions ventral were left intact—all other areas were discarded. Three rectangular pieces, each 5 x 7 mm, were extracted and stored in oxygenated Ames medium before use. Storage times ranged from 15 min to 6 h. Just before use, the retina was extracted from the eye cup and mounted, photoreceptor side down, to a 10-mm square piece of Millipore paper that was mounted with silicone grease to the recording chamber (1.0 ml volume). The Millipore paper had a 4-mm square hole in its center that allowed light from below to be projected on to the photoreceptors. Retinas were superfused continuously at 7–10 ml/min with Ames medium (Sigma; pH 7.4, 36°C), equilibrated with 95% O2-5% CO2. Kanamycin sulfate antimicrobial was added to the Ames medium.
Electrophysiology
Patch pipettes were used to make small holes in the inner limiting membrane, and ganglion cells were targeted under visual control. Light responses were used to assign the targeted cell to a known ganglion cell type. Spiking was recorded with a cell-attached patch electrode (5–6 M
), filled with superfusate. Excitatory and inhibitory input currents were measured with whole cell patch-clamp electrodes (6–7 M) that were filled with (in mM) 112.5 CsMeSO4, 1 Mg SO4, 7.8 x 10–3 CaCl2, 0.5 BAPTA, 10 HEPES, 4 ATP-Na2, 0.5 GTP-Na3, 5 lidocaine N-ethyl bromide (QX314-Br), and 7.5 neurobiotin chloride (pH 7.2). Alexa 488 (Molecular Probes) was added to the recording pipettes so that we could visualize the dendritic morphology of recorded cells after the electrical recordings were completed. We measured input currents as described (Fried et al. 2002, 2005) by voltage clamping ganglion cells at the reversal potential for chloride channels (–60 mV) or ligand-gated nonspecific cation channels (0 mV) to isolate excitatory or inhibitory synaptic currents, respectively.
Pharmacological agents were applied locally by microinjection through a glass pipette using a World Precision Instruments PV-830 Pneumatic PicoPump. This method generated quick wash-in and -out and yielded results that were consistent with bath-applied methods. Spiking was blocked with 1 mM TTX. The cocktail of synaptic blockers consisted of 5 mM curare, 1 mM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), and 10 mM DL-2-amino-7-phosphono-heptanoic acid (AP-7).
Light stimulus and data acquisition
The stimulus presentation and data acquisition software was written by E. Eizenman and G. Spor. Light stimuli were projected on to the retina from below through an LCD panel (CRL OPTO) and focused on to the photoreceptor outer segments. Retinas were light adapted before the start of electrophysiological recordings. Light stimuli consisted of stationary flashed squares (100, 200, 300, 500, or 1,000 µm), 1-s duration, and centered at the soma. Contrast levels were set to 200% unless specified. Data were analyzed in Matlab (MathWorks).
Electrical stimulation
Electrical stimulation consisted of charge balanced biphasic pulses: cathodic and anodic pulses were typically equal and opposite square wave pulses (MultiChannel Systems hardware and software). Cathodic pulses were delivered first, and intervals between phases were large enough so that the neural response to the cathodic pulse was completed before the onset of the anodic phase (typically 5 ms; range, 2–100 ms). In some of the higher stimulation frequency experiments, the duration of the anodic pulse was increased 2–10 times. The amplitude was adjusted to keep the total charge constant. Electrical pulses were delivered using Platinum-Iridium electrodes (MicroProbes, impedances 10–100 k). Stimulation was applied epi-retinally; the stimulating electrodes were placed (under visual control) 15 µm from the somata of targeted ganglion cells.
Charge density calculations
Electrodes (10 k) were conical, with a length of 125 µm and a base radius of 15 µm. The corresponding geometric surface area was 5,890 µm2. The actual surface area was higher because of the rough nature of the platinum surface. We estimated a 50% increase in surface area leading to an effective surface area of 8,836 µm2. With this electrode, typical early phase spike thresholds were 137 µA with a 0.06-ms pulse width, resulting in a calculated charge density of 0.093 mC/cm2. The threshold levels for late phase spike activation represents a maximum stimulation level and were 322 pA, resulting in charge densities of 0.219 mC/cm2. These calculated levels of charge density assume uniformity across the surface of the electrode (but also see McIntyre and Grill 2001).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
DISTINGUISHING NEURAL ACTIVITY FROM STIMULUS ARTIFACT. We measured responses to light and electrical stimulation in 65 ganglion cells from 30 different retinas using a combination of cell-attached and whole cell patch-clamp recordings (see METHODS). The initial response to pulses of electrical current consisted of large transient currents (Fig. 1A, horizontal arrows) that were temporally correlated with the onset and offset of both phases of the stimulus pulse (cathodic and anodic). The second portion of the response consisted of a series of biphasic waveforms (asterisks). These waveforms had similar magnitude and kinetics to light-elicited action potentials (Fig. 1B, inset). TTX, a blocker of neuronal spiking, eliminated the biphasic waveforms (n = 5/5), confirming that they were conventional voltage-gated Na+ spikes (Fig. 1B).
|
, although our threshold values and latency measurements differ (see DISCUSSION). The onset of the TTX-extracted pulse-elicited spike closely followed the onset of the cathodic pulse (mean = 580 µs, range = 400–680 µs, n = 5). Without TTX, it was difficult to precisely determine the onset of the elicited spike, but comparison of control records in 5 TTX and 15 non-TTX experiments were similar, providing further support that the elicited spike closely and consistently follows the stimulus pulse onset.
ALL LATE PHASE SPIKES WERE BLOCKED BY SYNAPTIC BLOCKERS.
To determine whether pulse-elicited spiking arises from direct activation of ganglion cells or whether it arises as a result of activation of presynaptic excitatory neurons, we measured responses in the presence of a pharmacological cocktail (CNQX, AP-7, and curare) that blocked all excitatory synaptic input to ganglion cells (Fig. 2). All late-phase spiking was eliminated in the presence of this cocktail (compare Fig. 2, A and B; n = 3/3), suggesting that excitatory synaptic input underlies the late-phase response (Jensen et al. 2005). In contrast, the early-phase spike was not affected by the cocktail, suggesting that this spike is not synaptically driven. Taken together, these results suggest that the spiking response to electrical pulses arises from two different mechanisms: 1) direct activation of the ganglion cell generates a single spike immediately after the onset of the cathodic pulse (early phase) and 2) activation of presynaptic excitatory neurons (likely bipolar cells) generates increased release of excitatory neurotransmitter on to ganglion cells, causing depolarization and spiking (late phase).
|
STIMULUS PULSES ELICIT EXCITATORY AND INHIBITORY CURRENTS IN GANGLION CELLS. To better understand the activation of presynaptic cells, we directly measured the input currents to ganglion cells using whole cell patch-clamp recordings. The isolated excitatory input current to ganglion cells provides a measure of the activation level of presynaptic excitatory neurons (most likely bipolar cells); increased release from these cells appears as an inward current in ganglion cells. Similarly, the isolated inhibitory current provides a measure of the activation level of presynaptic inhibitory neurons (most likely amacrine cells); increased release from amacrine cells appears as an outward current in ganglion cells. We used the magnitude and duration of the input currents to provide a more precise means of determining the activation level of presynaptic neurons in response to different stimuli configurations.
We measured robust excitatory input currents in response to electrical stimulus pulses (Fig. 3A). By convention, increases in excitatory currents are shown as downward deflections and downward peaks are depicted by vertical arrows. The presence of excitatory input currents indicates that bipolar cells were activated by electrical stimulation (n = 4/4). Stronger stimulation levels generally elicited more excitatory activity (n = 4/4), suggesting that bipolar activation increases with increasing pulse duration.
|
STIMULUS PULSES ELICIT REVERBERATING ACTIVITY AT THE INNER PLEXIFORM LAYER We examined the temporal correlation between spiking and the peak levels of excitatory and inhibitory input. Pulse-elicited spikes occurred during the onset (leading downward slope) of excitatory input currents (Fig. 3B); a similar temporal relationship exists between spiking and excitatory currents in response to light. The inhibitory current peaks were out of phase with the excitatory peaks so that inhibition was maximal in-between and after the late-phase spikes. In general, the peaks of excitation and inhibition alternated, suggesting that electrical stimulation elicits a reverberation of the underlying circuitry. For larger amplitude stimuli, we observed additional alternating peaks of excitation and inhibition lasting as long as 80 ms, although we did not quantify these responses or study the underlying mechanism (n = 3).
Amacrine cells activation disrupts the response to subsequent pulses
In healthy retina, amacrine cells deliver both feedforward inhibition to ganglion cells and feedback inhibition to bipolar cells. This raises the question of whether electrically stimulated amacrine cells will inhibit either ganglion or bipolar cells or both. Inhibitory currents can last for 100 ms (Fig. 3A), so the spiking response to pulses spaced at <100-ms intervals might be affected.
Long-lasting inhibition reduces excitation.
To evaluate the effect of prolonged inhibition, we measured the excitatory input to ganglion cells as stimulation frequency increased (Fig. 4). Because the inhibitory input to an individual stimulus pulse persisted for 100 ms, we reasoned that inhibition would not affect responses for low stimulation frequencies (<10 Hz), but would affect responses at higher stimulation frequencies. We measured excitatory currents in response to stimulation pulses delivered at 1 Hz (Fig. 4, top; n = 4) and found them to be consistent from trial to trial. When the stimulation frequency was raised to 2 Hz, however, the amplitude of the excitatory input current was reduced immediately: the magnitude of excitation was largest for the first pulse and reduced consistently for the second and subsequent pulses (Fig. 4, middle). This suggests that an inhibitory signal was acting at the release sites (bipolar cell terminals) to reduce excitation. The reduction was largest at the second pulse; excitatory currents increased over the next few pulses but did not return to original levels. This affect is most evident in the 2-Hz traces (Fig. 4, middle). At higher stimulation frequencies, the amplitude of the excitatory input current continued to decrease (Fig. 4, bottom) and by 10 Hz, the excitatory input current was no longer detectable (data not shown). The reduction of excitatory input with increasing stimulation frequency was summarized by averaging the elicited excitatory input per pulse at each stimulation frequency (Fig. 4B). These results suggest that the release of excitatory transmitter from bipolar cells is suppressed by long duration stimulus pulses and that the suppression persists for 500 ms. It is likely that this suppression is mediated by activation of amacrine cells that supply inhibition to the release sites on the bipolar cell terminals, although we did not perform experiments to support this.
|
Is it possible to independently activate either the early- or late-phase spiking responses? Our results thus far suggest that the spiking response to electrical stimulation is the result of a direct spike in ganglion cells (early phase) as well as a complex interaction between presynaptic neurons that results in a variable amount of spiking (late phase). We observed that the excitatory input elicited by electrical stimulation was a function of pulse duration: longer duration pulses generated larger excitatory inputs (Fig. 5A; n = 2/2). This suggested that decreasing the pulse duration might reduce or eliminate excitatory input to ganglion cells and therefore the late phase spikes. Short-duration pulses were not likely to eliminate the direct spike because the onset of the spike closely followed the leading edge of the cathodic pulse (Fig. 1B, inset) suggesting that the leading edge of the pulse was generating the early-phase spike.
|
ONE SPIKE PER PULSE IS CONSISTENT AT HIGHER STIMULATION FREQUENCIES.
The ability to generate a single spike from a single electrical pulse is important and suggests that we can use short pulses to generate specific predetermined patterns of spiking. Ganglion cells generate light elicited spiking at rates 250 Hz; thus we evaluated whether short pulses could generate one pulse per spike at higher stimulation frequencies (Fig. 6, A and B). High-frequency trains of short-duration stimulus pulses (Fig. 6A, top) elicited one spike (asterisks) per pulse. We extracted the time interval from 1 ms before to 2 ms after each cathodic pulse for all 250 pulses: the individual traces are overlaid in Fig. 6B. The individual responses were extremely consistent and comparison with the response in TTX (dotted trace) indicates that each pulse elicits a single spike. We observed similar responses in 12 cells at all frequencies tested (250 Hz).
|
Replicating light-elicited spike patterns. We used programmed sequences of short pulses to replicate light-elicited spike patterns (Fig. 7). We measured the spiking response to the flash of a small square of light (Fig. 7, A and B) and calculated the latency of each spike. This was used to program a sequence of short-duration pulses (Fig. 7B, top) such that the latency of individual cathodic pulses matched the latency of individual spikes; charge-balancing anodic pulses followed each cathodic pulse. The programmed sequence of pulses was delivered to the ganglion cell, and the net result was a pattern of elicited spikes whose temporal pattern precisely matched the pattern of light-elicited spikes (Fig. 7C). Jitter between individual light- and pulse-elicited spikes was <0.5 ms. Changes in the light stimulus elicited different spike patterns that were replicated with different programmed arrays of short pulses (data not shown).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
We developed a method to replicate light-elicited spiking patterns in individual ganglion cells using electrical stimulation. One short-duration cathodic pulse (0.1 ms) generated one spike, time-locked to within 0.7 ms of the pulse onset (Fig. 5C). One spike per pulse was elicited at stimulation frequencies 250 Hz (Fig. 6, A and B), the upper limit of light-elicited spiking responses. This suggests that precise temporal patterns of spiking, including those patterns elicited by light, can be generated using programmed arrays of short pulses (Fig. 7). Matching the normal light spike patterns provides a method to send a more natural (physiological) signal to the brain that may lead to the generation of more meaningful percepts.
There will be variability in the distance between individual electrodes and targeted ganglion cells when an epi-retinal array of stimulating electrodes is placed on the retinal surface. The variations arise because ganglion cell somata are situated at different depths within the ganglion cell layer and also because the stimulating array may not lie perfectly flat on the retinal surface (Humayun et al. 2003). The threshold for activation increases with increasing distance between the stimulating electrode and targeted ganglion cell (BeMent and Ranck 1969), suggesting that stimulation methods will need to be effective over a wide range of amplitude levels. The large window of effective stimulation for short pulses (2.5:1) suggests that variations in distance can be accommodated by this method (Fig. 6C).
LATENCY OF THE EARLY-PHASE RESPONSE IS0.7 ms.
We found the latency between the onset of the stimulus pulse and the onset of the elicited spike to be <0.7 ms (Fig. 5C). This value differs from a recent study (Jensen et al. 2005) that reported latencies of 3–5 ms. The most significant difference between the studies is the position of the recording electrode. We placed patch-clamp electrodes directly onto the surface of the ganglion cell soma (cell-attached patch). This measurement provided a robust spiking signal that could be visualized within the stimulus artifact; comparison with the TTX response allowed us to isolate the spike and precisely determine its latency. Previous studies (Jensen et al. 2005a) using extracellular recordings placed the recording electrode several millimeters from the stimulating electrode. Under these conditions, the measured latency included a delay that was a result of spike propagation along the axon to the recording site. The long delay is effective for separating spikes from the stimulus artifact but misrepresents the actual timing between the stimulus pulse and the spike onset.
Long-lasting activation of amacrine cells suppresses responses to subsequent stimulation
LONG PULSES ACTIVATE NEURONS PRESYNAPTIC TO GANGLION CELLS.
Long pulses of electrical stimulation (1 ms) elicited excitatory and inhibitory currents in the ganglion cell, indicating that both bipolar and amacrine cells are activated by these pulses (Fig. 3). Bipolar cell activation is supported by previous studies (Greenberg 1998; Jensen et al. 2005), in which a portion of the pulse-elicited spiking response disappeared in the presence of excitatory synaptic blockers. We were able to determine the level of bipolar cell activation by measuring the magnitude of excitatory input currents in ganglion cells and found that shorter pulses resulted in progressively weaker bipolar cell activation (Fig. 5A). This led us to test shorter pulses as a means of silencing bipolar cell-mediated spiking.
ACTIVATION OF AMACRINE CELLS SUPPRESSES PRESYNAPTIC ACTIVITY.
In the normal retina, amacrine cells synapse onto and inhibit bipolar, ganglion, and other amacrine cells. This suggests that activated amacrine cells might not only inhibit the direct early-phase response (by raising activation thresholds), but might also reduce the amount of late phase spiking by suppressing activation of bipolar cells to subsequent pulses. The long duration of the pulse-elicited inhibitory current (Fig. 3A) suggests that the response to subsequent pulses could be inhibited for 100 ms, which would interfere with stimulation at frequencies >10 Hz. We observed a reduction of excitatory input at stimulation frequencies as low as 2 Hz (Fig. 4). Our results do not allow us to unravel the mechanism by which this occurs, but it seems likely that the reduced excitatory input to subsequent pulses arises from an increased level of amacrine cell-mediated inhibition that arrives at the bipolar cell terminal. The duration of this effect was 500 ms, even longer than the 100-ms duration of the direct inhibitory signal. The difference in duration between the two inhibitory effects (direct vs. reduction of excitatory input) suggests that more than one population of amacrine cells are activated: one population delivers direct inhibitory input to ganglion cells, and the other population inhibits the release of excitatory transmitter at the bipolar cell terminal. It is possible that additional populations of amacrine cells are also activated, although we did not encounter their effects in our measurements.
The diverse effects of activated amacrine cells suggest that long stimulus pulses may compromise our ability to generate closely spaced patterns of spiking activity in retinal ganglion cells. Whereas short pulses can reliably elicit spikes every few milliseconds, long pulses elicit prolonged inhibitory activity that can adversely affect the spiking response to subsequent pulses. On the other hand, inhibitory activity from long stimulus pulses may provide a mechanism to temporarily reduce the spiking output of the retina, e.g., if spontaneous activity is increased in degenerated retina (Stasheff 2004).
Are small diameter electrodes more suitable for retinal stimulation?
SMALL DIAMETER ELECTRODES CAN RELIABLY ELICIT SPIKING AT SAFE STIMULATION LEVELS
In this study, we used small-tipped electrodes to elicit spiking in ganglion cells—the surface area of our largest electrodes was comparable with a 40-µm-diam disk electrode, considerably smaller than currently used implanted electrodes (Humayun et al. 2003). Smaller electrodes are advantageous for stimulating the retina because they generate more focal stimulation; however, they require higher charge densities that can cause breakdown of the electrode as well as adverse tissue reactions. The recommended upper limit of charge density for platinum electrodes is 0.3–0.35 mC/cm2 (Brummer and Turner 1977). In this study, charge densities at threshold were 0.093 mC/cm2. Most of our experiments were performed at 150% of threshold, corresponding to a charge density of 0.139 mC/cm2. These charge density levels assume a uniform distribution across the surface of the electrode that may not hold true for small-tipped conical electrodes (McIntyre and Grill 2001). However, even if the nonuniformities in our electrodes increase the charge density slightly beyond safe limits, further testing of small electrodes is still warranted. New electrode materials under development have higher charge density limits (Cogan et al. 2005), and electrodes with symmetrical designs (circular, hemispherical) will reduce the elevated charge densities arising from irregular shapes. Reducing electrode size will allow focal activation of small groups of neurons which will lead to higher resolution patterns of prosthetic-elicited activity that are closer to light-elicited patterns.
SHORT PULSES LOWER CHARGE DENSITY LEVELS.
Typical threshold levels for 1-ms pulses were 15 µA, corresponding to a charge density of 0.17 mC/cm2, nearly twice the current density associated with short pulses. These results are consistent with recent results from Jensen et al. (2005), who reported similar percentage increases between short and long pulses. Short pulses, by reducing the total charge delivered, will allow us to use smaller electrodes and still remain within the acceptable limits of charge density.
VARIATIONS IN THRESHOLD ACROSS STUDIES.
The thresholds reported by Jensen et al. (2003) were 1 µA, lower than our average of 15 µA. It is likely that the differences arise because of differences in electrode impedance between the two studies. Although, Jensen et al. did not report impedance levels, they describe their electrode as conical with 5 µm height and 2 µm base diameter. Our smallest electrode was 15 µm height and 7 µm base diameter. It is likely that the smaller surface area of the electrode of Jensen et al. results in higher impedance levels of their electrodes. Higher impedances result in higher current densities with corresponding lower threshold levels.
WHY ARE PERCEPTUAL THRESHOLDS HIGHER THAN SPIKE THRESHOLDS?
Current levels required to generate spikes in in vitro preparations are lower than current levels for perception in implanted patients. Why is this so? It is likely that properly coordinated physiological activity of populations of ganglion cells normally generated in response to light (Roska and Werblin 2001) are not achieved during electrical stimulation; therefore, although cells are spiking, their activity does not lead to perception. In this study, we found that temporal properties of (long) pulse-elicited spike trains do not match the temporal properties of light-elicited spike trains. In human psychophysical experiments, the large diameter stimulation electrodes (250–500 µm; Humayun et al. 2003) cannot match the precise spatial patterns generated by light stimuli. A third discrepancy between light-elicited versus pulse-elicited patterns of activation arises from the presence of multiple types of retinal ganglion cells (e.g., ON and OFF). Under normal conditions, light stimuli do not simultaneously activate ON and OFF ganglion cells, whereas large diameter electrodes likely do. It is likely that the specific features (temporal, spatial, and cell-type) of the normal patterns are necessary for optimal detection of the retinal signal. Without the precise signaling patterns, it is likely that the brain's ability to detect the retinal signal is diminished. The method we have developed here matches the temporal component of these patterns but not the spatial and cell-type components. Stimulation methods that match these two other components may reduce the thresholds for percept detection down to the levels for spiking in individual ganglion cells (in vitro).
Future work
TARGETING INDIVIDUAL CELLS.
Our results indicate that light-elicited spike patterns in ganglion cells can be replicated using trains of short pulses that each elicit a single spike. Considerable challenges remain, however, along the path toward focal stimulation of local populations of individual ganglion cells. For example, methods that limit activation to small focal regions and methods that separately target individual ganglion cell types must both be developed. In addition, the threshold for activation of ganglion cells is only slightly lower than the threshold for activation of axonal fibers (Greenberg 1998; Jensen et al. 2003). Therefore methods that avoid stimulating axons must be improved as well.
TECHNICAL CHALLENGES FOR IMPLEMENTING SHORT PULSES. Short pulses have lower total charge requirements but require higher stimulation current levels and will likely require higher power supply voltage levels. The higher current levels associated with short pulses may lead to increased heating of the retina. Finally, stimulation methods that employ short pulses to elicit individual spikes may require more sophisticated signal processing regimens. The considerable benefits associated with replicating light-elicited spike patterns with short pulses will have to be evaluated in light of these additional engineering challenges.
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: F. Werblin, 145 LSA, UC Berkeley, Berkeley, CA 94720 (E-mail: werblin{at}berkeley.edu)
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Brummer SB and Turner MJ. Electrical stimulation with Pt electrodes: II-estimation of maximum surface redox (theoretical non-gassing) limits. IEEE Trans Biomed Eng 24: 440–443, 1977.[ISI][Medline]
Bunker CH, Berson EL, Bromley WC, Hayes RP, and Roderick TH. Prevalence of retinitis pigmentosa in Maine. Am J Ophthalmol 97: 357–365, 1984.[ISI][Medline]
Chow AY, Chow VY, Packo KH, Pollack JS, Peyman GA, and Schuchard R. The artificial silicon retina microchip for the treatment of vision loss from retinitis pigmentosa. Arch Ophthalmol 122: 460–469, 2004.
Cogan SF, Troyk PR, Ehrlich J, and Plante TD. In vitro comparison of the charge-injection limits of activated iridium oxide (AIROF) and platinum-iridium electrodes. IEEE Trans Biomed Eng 52: 1612–1614, 2005.[CrossRef][ISI][Medline]
Curcio CA, Medeiros NE, and Millican CL. Photoreceptor loss in age-related macular degeneration. Invest Ophthalmol Vis Sci 37: 1236–1249, 1996.
Fried SI, Munch TA, and Werblin FS. Directional selectivity is formed at multiple levels by laterally offset inhibition in the rabbit retina. Neuron 46: 117–127, 2005.[CrossRef][ISI][Medline]
Fried SI, Münch TA, and Werblin FS. Mechanisms and circuitry underlying directional selectivity in the retina. Nature 420: 411–414, 2002.[CrossRef][Medline]
Friedman DS, O'Colmain BJ, Munoz B, Tomany SC, McCarty C, de Jong PT, Nemesure B, Mitchell P, and Kempen J. Prevalence of age-related macular degeneration in the United States. Arch Ophthalmol 122: 564–572, 2004.
Greenberg R. Analysis of electrical stimulation of the vertebrate retina - work towards a retinal prosthesis. Baltimore, MD: Johns Hopkins University, 1998. (PhD thesis).
Humayun M, Propst R, de Juan E Jr, McCormick K, and Hickingbotham D. Bipolar surface electrical stimulation of the vertebrate retina. Arch Ophthalmol 112: 110–116, 1994.[Abstract]
Humayun MS, de Juan E Jr, Dagnelie G, Greenberg RJ, Propst RH, and Phillips DH. Visual perception elicited by electrical stimulation of retina in blind humans. Arch Ophthalmol 114: 40–46, 1996.[Abstract]
Humayun MS, Prince M, de Juan E Jr, Barron Y, Moskowitz M, Klock IB, and Milam AH. Morphometric analysis of the extramacular retina from postmortem eyes with retinitis pigmentosa. Invest Ophthalmol Vis Sci 40: 143–148, 1999.
Humayun MS, Weiland JD, Fujii GY, Greenberg R, Williamson R, Little J, Mech B, Cimmarusti V, Van Boemel G, Dagnelie G, and de Juan E. Visual perception in a blind subject with a chronic microelectronic retinal prosthesis. Vision Res 43: 2573–2581, 2003.[CrossRef][ISI][Medline]
Jensen RJ, Rizzo JF III Ziv OR, Grumet A, and Wyatt J. Thresholds for activation of rabbit retinal ganglion cells with an ultrafine, extracellular microelectrode. Invest Ophthalmol Vis Sci 44: 3533–3543, 2003.
Jensen RJ, Ziv OR, and Rizzo JF III. Responses of rabbit retinal ganglion cells to electrical stimulation with an epiretinal electrode. J Neural Eng 2: S16–21, 2005a.[CrossRef][Medline]
Jensen RJ, Ziv OR, and Rizzo JF III. Thresholds for activation of rabbit retinal ganglion cells with relatively large, extracellular microelectrodes. Invest Ophthalmol Vis Sci 46: 1486–1496, 2005b.
Jones BW and Marc RE. Retinal remodeling during retinal degeneration. Exp Eye Res 81: 122–137, 2005.
Marc RE and Jones BW. Retinal remodeling in inherited photoreceptor degenerations. Mol Neurobiol 28: 139–147, 2003.[CrossRef][ISI][Medline]
Marc RE, Jones BW, Watt CB, and Strettoi E. Neural remodeling in retinal degeneration. Prog Retin Eye Res 22: 607–655, 2003.[CrossRef][ISI][Medline]
Mastronarde DN. Correlated firing of retinal ganglion cells. Trends Neurosci 12: 75–80, 1989.[CrossRef][ISI][Medline]
McIntyre CC and Grill WM. Finite element analysis of the current-density and electric field generated by metal microelectrodes. Ann Biomed Eng 29: 227–235, 2001.[CrossRef][ISI][Medline]
Meister M, Lagnado L, and Baylor DA. Concerted signaling by retinal ganglion-cells. Science 270: 1207–1210, 1995.
O'Brien BJ, Isayama T, Richardson R, and Berson DM. Intrinsic physiological properties of cat retinal ganglion cells. J Physiol 538: 787–802, 2002.
Rizzo JF 3rd, Wyatt J, Loewenstein J, Kelly S, and Shire D. Methods and perceptual thresholds for short-term electrical stimulation of human retina with microelectrode arrays. Invest Ophthalmol Vis Sci 44: 5355–5361, 2003a.
Rizzo JF 3rd, Wyatt J, Loewenstein J, Kelly S, and Shire D. Perceptual efficacy of electrical stimulation of human retina with a microelectrode array during short-term surgical trials Invest Ophthalmol Vis Sci 44: 5362–5369, 2003b.
Rockhill RL, Daly FJ, MacNeil MA, Brown SP, and Masland RH. The diversity of ganglion cells in a mammalian retina. J Neurosci 22: 3831–3843, 2002.
Roska B and Werblin F. Vertical interactions across ten parallel, stacked representations in the mammalian retina. Nature 410: 583–587, 2001.[CrossRef][Medline]
Santos A, Humayun MS, de Juan E Jr, Greenburg RJ, Marsh MJ, Klock IB, and Milam AH. Preservation of the inner retina in retinitis pigmentosa. A morphometric analysis. Arch Ophthalmol 115: 511–515, 1997.[Abstract]
Smirnakis SM, Berry MJ, Warland DK, Bialek W, and Meister M. Adaptation of retinal processing to image contrast and spatial scale. Nature 386: 69–73, 1997.[CrossRef][Medline]
Stasheff SF. Increased spontaneous activity and high frequency rhythmic firing among retinal ganglion cells accompanies outer retinal degeneration in the rd1 mouse. Invest Ophthalmol Vis Sci 45: 5070, 2004.
Stone JL, Barlow WE, Humayun MS, de Juan E Jr, and Milam AH. Morphometric analysis of macular photoreceptors and ganglion cells in retinas with retinitis pigmentosa. Arch Ophthalmol 110: 1634–1639, 1992.[Abstract]
Wassle H. Parallel processing in the mammalian retina. Nat Rev Neurosci 5: 747–757, 2004.[CrossRef][ISI][Medline]
Zrenner E, Miliczek KD, Gabel VP, Graf HG, Guenther E, Haemmerle H, Hoefflinger B, Kohler K, Nisch W, Schubert M, Stett A, and Weiss S. The development of subretinal microphotodiodes for replacement of degenerated photoreceptors. Ophthalmic Res 29: 269–280, 1997.[ISI][Medline]
This article has been cited by other articles:
|
|
 |
|
  D. J. Margolis, G. Newkirk, T. Euler, and P. B. Detwiler Functional Stability of Retinal Ganglion Cells after Degeneration-Induced Changes in Synaptic Input J. Neurosci., June 18, 2008; 28(25): 6526 - 6536. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Sekirnjak, P. Hottowy, A. Sher, W. Dabrowski, A. M. Litke, and E. J. Chichilnisky High-Resolution Electrical Stimulation of Primate Retina for Epiretinal Implant Design J. Neurosci., April 23, 2008; 28(17): 4446 - 4456. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. J. DeMarco Jr, G. L. Yarbrough, C. W. Yee, G. Y. McLean, B. T. Sagdullaev, S. L. Ball, and M. A. McCall Stimulation via a Subretinally Placed Prosthetic Elicits Central Activity and Induces a Trophic Effect on Visual Responses Invest. Ophthalmol. Vis. Sci., February 1, 2007; 48(2): 916 - 926. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEED |
