Master Thesis-Comparative Analysis of Electrical Stimulation Versus Biochemical Stimulation in rd1 Degenerated Retina As Potential Candidates For Retinal Prosthesis
0 évaluation0% ont trouvé ce document utile (0 vote)
89 vues59 pages
Understanding the various aspects of retinal prosthesis..
Titre original
Master Thesis-Comparative analysis of electrical stimulation versus biochemical stimulation in rd1 degenerated retina as potential candidates for retinal prosthesis
0 évaluation0% ont trouvé ce document utile (0 vote)
89 vues59 pages
Master Thesis-Comparative Analysis of Electrical Stimulation Versus Biochemical Stimulation in rd1 Degenerated Retina As Potential Candidates For Retinal Prosthesis
Comparative analysis of electrical stimulation versus biochemical stimulation in
rd1 degenerated retina as potential candidates for retinal prosthesis
Thesis submitted in partial fulfillment of the requirements for the degree
Master of Science
Graduate School of Cellular & Molecular Neuroscience
Faculty of Science Faculty of Medicine
Eberhard-Karls-Universitt Tbingen
Presented by
Archana Jalligampala
from Cuttack, India
Tbingen, 26-05-2011
2
Thesis Advisor: PD Dr. / Prof. Dr. Elke Guenther Department of Electrophysiology, NMI, Reutlingen
Second Reader: PD Dr. Guenther Zeck Department of Microsystems and Nanotechnology, NMI, Reutlingen
I affirm that I have written the dissertation myself and have not used any sources and aids other than those indicated.
Date / Signature:
3 Acknowledgements
The journey of my master thesis has been indeed a great learning experience, challenging yet fun and would not have been possible without the following people. Firstly Prof. Dr. Elke Guenther who gave me an opportunity to work in her lab and help me unravelling mysteries of Electrophysiology and above all about retina electrophysiology. Irrespective of her busy schedules she was kind enough to dedicate some precious time to listen out my progress and failures in the project and helping me by giving her suggestions which indeed was quite a big contribution in itself. Secondly, Dr.Udo Kraushaar, more fondly as Udo, any amount of praising would just embarrass me because I can write a whole lot of pages about his contribution to my life both personally and professionally. Without his positive encouragements and the zeal to see science take its shape this thesis would never had shaped out this well. His every little word of encouragements and his trust on me to make things work made me a strong person and more determined to get this thesis into rolling. During this course not only did I benefit scientifically from him but personally believed that if one is determined to achieve something no barriers could pull you down. How to think like a scientist , what could be the slightest loop holes in an experiment was what I learnt from him and indeed would stay with me till the end of my life. Sorry for all the frustrations you had to bear with, your contribution means a lot. Sincere thanks to Dr.Guenther Zeck, who kindly agreed to be the second reader for my thesis. The lab members Katherine, Theresa, Sven, Adrian, Nadine without whose co-operation my thesis would be incomplete. I would not have been here without the support of Graduate school for Cellular and Molecular Neuroscience. Sincere heartfelt thanks to Prof.Herbert, Katja and Tina for always being there to help and taking enormous amount of efforts on my behalf. The zeal to work in neuroscience specially retina physiology , the love for it and a wish to pursue further research in this field is due to the revered faculty members like Prof Euler, 4 Dr.Timm Schubert, Dr. Francois Paquet Durand. Thank you for bestowing such a vast ocean of knowledge and making me realize my appetite for this science. My sincere thanks to all the faculty members of the graduate school who imparted me with such an immense amount of knowledge which would be with me for the rest of my life. Though professionally one could achieve heights, without personal sanity one would never be able to give ones 100%. This was possible due to few close people without whom my life is incomplete. Firstly Varsha, she made me realize how important science is to me and that one has to believe in oneself to understand science. The letter N of Neuroscience was very kindly fed to me by her efforts. She has been a big driving force for me to pursue neuroscience as my research interest. Thank you, Varsha for all the little help and being my strength through and through the entire course of Germany. Germany would not have been the same without you. It is often said an anchor in your life makes the journey worth while. This anchor enables you to be grounded amidst the tough tides of life. My anchor none other than Sambit Biswal, a friend, philosopher , in a nutshell a person who kept me strong through this entire course of time and much before. Irrespective of our long term association the level of understanding and love bestowed by you made me a true fighter and made me fight back at times when I thought I would break. Thank you so much for being there, your presence in my life makes my life worthwhile. In ones life time we encounter people who make you realize that life is just not achieving results but at times let loosing oneself. The spiritually sanity was a big role to play during this period of my life and none other than my grandfather like Dr. S.K. Pattnaik this would not have been possible. Finding positivity in the darkness of negativity was what he made me believe and Im thankful to him for everything, as, if Im here today is due to his blessings and kind advice to follow life. Last but not the least my parents and Almighty who have been my driving force since I have stepped into this world. Without them my meaning of life would have been pretty incomplete. 5 Table of Contents
1. Abstract 6
2. Introduction 7
3. Materials and Methods 18
4. Results 28
5. Figures and Tables 36
6. Discussions and Implications 50
7. References 57
6 Abstract
In the vertebrate eye the retina is responsible for light perception. It is a sophisticated image processor and is the first processing step in vision. As a model system, the retina is a highly organised structure with defined connections and circuitry. Recent attempts to restore vision in the blind, like patients suffering from Retinitis pigmentosa (RP) or age- related macular degeneration (AMD) have met with extraordinary success. These form of blindness results in substantial loss of photoreceptors leaving the inner retinal layers unaltered to quite some extent. Although physiological and morphological changes may take place in the inner layers of the retina of the affected patients, there exists opportunity for direct excitation of the residual neuron as a mean of restoring vision which forms the basis for visual prosthesis. The main goal of the present study aimed to mimic two broad candidates of retinal prostheses, one being the electrical stimulation and the other biochemical stimulation (via focal glutamate application) of acutely explanted degenerated retina of murine on micro electrode array system. This study also aimed at comparative analysis of epi-retinal stimulation versus sub-retinal stimulation on the basis of electrical thresholds, pulse paradigms and latency. The study could help in designing an efficient candidate of the above two prostheses models, which could encode a high spatio-temporal resolution, thus enabling a stable retinal prosthesis for restoring vision in blind patients.
7 Introduction The main part in our visual system is the eye. Our ability to see is the result of a process very similar to that of a camera. A camera needs lens and film to produce an image, similarly the eyes needs lens as well as film to process an image. The film of our eye is represented by the Retina. It captures the image and sends it to the brain to be developed as shown below in the figure. Retina is responsible for light perception and forms the first processing step in vision.
Fig 1: The above figure suggests a broad comparison of a camera with the vertebrate eye, suggesting as the film captures image and is developed to generate a processed image ,similarly in our eye the retina acts as a film and captures image and sends it the higher organisation for being developed.(Fig taken from Artificial eye, www.bestneo.com)
Being the most accessible part of the vertebrate CNS it is well separated from the rest of the brain and connected by well-defined axonal projections branched from the ganglion call layer. It possesses a highly ordered structure and consists of different neuronal cell types, which are organized in different layers. The retina derives its nourishment from the Retinal Pigment Epithelium (RPE) .The photoreceptors cells are localized in the outer and inner segment (OS & IS) and its nucleus forms the outer nuclear layer (ONL). The nuclei of bipolar cells, horizontal cells and amacrine cells form the inner nuclear layer (INL).Dividing these nuclear cell layers are two synaptic layers where synaptic contacts occur i.e. inner 8 plexiform layer (IPL) & outer plexiform layer (OPL). Lastly the nuclei of ganglion cells form the ganglion cell layer (GCL). The ganglion cells then branch out to give axonal projections which form the nerve fiber layer, (as depicted in the figure below) to the lateral geniculate nucleus and further to the visual cortex for processing information.
Fig 2: The above figure suggests the highly ordered structure of retina with well characterized cellular composition. The figure in the left shows the layered structure of retina and the one in the right gives an overview of the cells which forms these layers (Fig taken from wikipedia.com)
Photoreceptor cell death (rods and cones) is the major hallmark of a group of human inherited retinal degenerations commonly referred to as Retinitis Pigmentosa (RP) pertaining to rod degeneration (Paquet-Durand et al, 2009; Christian Hamel,2006) and Age-related macular degeneration (AMD) pertaining to cone degeneration. Rods and cones form the photoreceptor layer and form the photosensitive cells in the retina. Rods form the part of periphery vision and function in less intense light. The numbers of rod cells are more in comparison to cone cells and they form the basis for night vision. Cones function best in relatively bright light and are responsible for perception of light. They are well concentrated in the macula and become sparse on moving towards periphery of the retina. They are also responsible for visual acuity i.e. the area of highest resolution. RP 9 predominantly affects the photoreceptor layer in the periphery which is the rods and eventually leads to a vision called as tunnel vision which eventually leads to blindness. In AMD there is progressive loss of macula (cone) which leads to loss of central vision, precisely the visual acuity. Together; AMD & RP affect at least 30 million people in the world (Thomas M. O Hearn et al, 2006). The figure below illustrates the appearance of image perceived by RP & AMD patients. They are the most common causes of untreatable blindness in developed countries and, currently, there are no effective measures of restoring vision. (a) (b)
Fig 3: The above figure demonstrates the perception of patients suffering from retinal degeneration. (a) shows the tunnel vision as commonly seen in patients suffering from Retinitis pigmentosa due to loss in photoreceptors(rods) and (b) shows the loss of visual acuity due to degeneration of macula in Age-related macular degeneration (Fig taken from NIH National Eye Institute)
In recent times due to absence of effective therapeutic remedies for RP & AMD, researchers and scientists are motivated to develop experimental strategies to restore some degree of visual function to affected patients. Much previous literatures argued that retinal degeneration such RP affect only sensory retina. Many approaches to retinal rescue are based on this clearly incorrect assumption. At first it may appear discouraging and even insurmountable given that the field of vision rescue has been labouring for past 20 years to overcome the loss of photoreceptors alone, only recently recognizing the scope of neural remodelling in the retina. Furthermore this 10 plasticity and continued neural signalling reveals possible mechanism for exploitation, and emphasizing the need for early diagnosis and intervention to retard remodelling. The fact of remodelling influences all rescue strategies begins before the inner retinal cells die. Of the various retinal rescues retinal prosthesis implant strategies aim to replace photoreceptors with electronics or artificial synapse to drive remnant retina for transmitting visual information. As from histological sections of the diseased retina it is evident that residual inner layers of neural retina are spared for a long time, several approaches have been designed to artificially stimulate this residual retina and thereby the visual system and this gave way to a term Retinal Prosthesis. At present, two major strategies for electrical stimulation as a retinal prosthesis have been pursued. The Epiretinal approach involves a semi-conductor based device (array) in close contact with the ganglion cell layer. In such implants information must be captured by a camera system before transmitting data and energy to the implant (Artificial eye, 2008, www.bestneo.com). Second is the Subretinal approach which involves the electrical stimulation of inner retina from the sub-retinal space (between RPE and neural retina) by implantation of semiconductor based micro-photodiode (MPDA) array into this location or focal stimulation of the inner retina using tungsten electrode and assessing the read out at the microelectrode array (Artificial eye, 2008, www.bestneo.com). In the former approach the electrical charge generated by MPDA in response to light stimulus may be used to artificially alter the membrane potentials of the neurons of the residual retinal layers in a manner to produce formed images.
11
Fig 4: The above figure gives broad overview of strategies of occular implants including the difference between epiretinal and subretinal approaches in implants (Fig taken from Bionic eye)
Significant progress has been made towards the development of long- term, implantable retinal prostheses due to rapid advancement of micro-electrode fabrication methods (one major contribution in recent times, the group of Prof.Dr.Zrenner, in Tuebingen, Germany working on Retina Implant Project who could successfully implant a MPDA enabling a patient named Mika to read specific alphabets, Subretinal electronic chips allow blind patients to read letters and combine them to words,Eberhart Zrenner et al 2010). However electrical stimulation has few limitations encoding important sensory features used in normal central visual processing. Few acute and chronic human testing revealed that patients perceived white flashes of light in response to epiretinal and subretinal stimulation. It was also seen that although the temporal feature of electrical stimuli was well correlated with electro-phosphene percepts, patients demonstrated difficulty in perceiving shapes with more complex spatial patterns of electrical stimulations. It was observed in few patients with small diameter electrodes generated free radicals like hydrogen peroxide which induced retinal tissue damage by causing toxicity to the lipid membranes of neuron and glia and those with large diameter electrodes required yielded poor resolution. (Finlayson and 12 Iezzi, IOVS, July 2010). In addition to the above limitations, electrical stimulation is non- specific i.e. it stimulates both ON and OFF retinal ganglion cells, bipolar cells and very likely amacrine cells thus reducing contrast and spatial localization within the retina. (Finlayson and Iezzi, IOVS, July 2010). Many of the neural prostheses use electricity to depolarize the neural cells, however stimulation in the nervous system is primarily through chemical neurotransmission. The native use of electrical stimulation is demonstrated by the electrical gap junctions between neurons, neurotransmitters provide specificity and flexibility of stimulus parameters that gap junction typically do not provide. (Peterman et al, 2003) To achieve useful vision, a retinal prosthesis may need to mimic the biological complexity and resolution of the retina, including the ability to address individual cell types and to stimulate the retina with the biological specificity achieved by using neurotransmitters, of which glutamate acts as a primary neurotransmitter. The other common neurotransmitters in the retina are GABA, glycine, dopamine and acetylcholine. Glutamate is the excitatory neurotransmitter released by photoreceptor, bipolar, and retinal ganglion cells, may provide a more natural means of stimulating RGCs (retinal ganglion cells). However in some horizontal cells and amacrine cells when labelled with glutamate antibodies showed weak labelling. It is believed that these cells release GABA, not glutamate as their neurotransmitter, suggesting that the weak labelling of glutamate reflects the pool of metabolic glutamate used in the synthesis of GABA (Yang, 1996). Glutamate is incorporated into these cell types through a high affinity glutamate transporter located in the plasma membrane. These transporters maintain the concentration of glutamate within the synaptic cleft at low levels preventing glutamate excitotoxicity. A major share of glutamate is taken up by the glial cells of the retina, Muller cells and rapidly broken down to glutamine. Histological techniques help in identifying the potential glutamatergic neurons by labelling neurons containing glutamate and neurons that take up glutamate. To further determine if 13 these cell types actually release glutamate as their neurotransmitter, the receptors on the post synaptic side have to be examined. Once released from the presynaptic terminal, glutamate diffuses across the cleft and binds onto the receptors located on the dendrites of the post synaptic cell(s). Two major classes of glutamate receptors have been identified. First being the Ionotropic glutamate receptors, which directly gate ion channels and Metabotropic glutamate receptors, which may be coupled to an ion channel or other cellular functions via an intracellular second messenger cascade. Various retinal neuronal cell types differentially express their subtypes. These receptor types are similar in that they both bind glutamate and glutamate binding can influence the permeability of ion channels. However, there are several differences between the two classes. Glutamate binding on to an ionotropic receptor directly influences ion channel activity because the receptor and the ion channel form one complex, thus mediating fast synaptic transmission between neurons. Two Ionotropic receptor types have been identified, NMDA receptors which bind glutamate and the glutamate analogue N-Methyl-D-Aspartate(NMDA) and non-NMDA receptors, which are selectively agonized by kianate(GluR5/6), -amino-3-hydroxy-5-methyl-4- isoxazolepropionic acid, AMPA(GluR1-4), but not NMDA. The non-NMDA receptors open non-selective cation channels more permeable to sodium and potassium ions than calcium. In retina, non-NMDA receptors have been identified on horizontal cells, OFF-bipolar cells, amacrine cells and ganglion cells. NMDA receptors also open non-selective cation channels on binding with glutamate, resulting in an increased conductance, which is due to increased permeability to calcium than sodium ions. Few retinal ganglion cells and some amacrine cells express functional NMDA receptors. The functional organisation of Ionotropic glutamate receptors is well illustrated in Fig (5a & b).
14
(a) (b)
Fig 5: (a) Non-NMDA receptors are selectively agonized by kianate, AMPA and are permeable to Na+ and K+ ions than Ca2+ ions (Mayer and Westbrook, 1987). (b) NMDA receptors are structurally complex, with separate binding sites for glutamate, glycine, Mg2+ and Zn2+ and polyamines and are permeable to Ca2+ than Na+ ions (Fig adapted from webvision.edu)
Unlike ionotropic receptors, which are directly linked to an ion channel, metabotropic receptors are coupled to their associated ion channels through a second messenger pathway. Ligand binding activates a G-protien and initiates the intracellular cascade (Fig 6). These receptors are classified into three groups (I, II &III) based on structural homology, agonist selectivity and their associated second messenger cascade. Few metabotropic receptors (like APB) are selective for the cells of the ON-pathway and some of these receptors are differentially expressed throughout the retina. Unlike NMDA and non-NMDA receptors glutamate binding onto metabotropic receptors activates a G-protein which further activates phospohdiesterase, that reduce the intracellular concentration of cyclic nucleotides leading to closure of cGMP gated non-selective cation channels. This in turn gates the entry of sodium or calcium ions into the cell thereby decreasing the conductance. 15
Fig 6: Represents the metabotropic glutamate receptors (mGluR) coupled to their associated ion channels with second messenger cascade. In the above figure when glutamate binds the mGluR receptors then the G protein is activated and the intracellular cascade of proteins are triggered thereby modulating the associated ion channel to open or close. (Figure taken from webvision.edu)
These glutamate receptors have been shown to participate in normal visual activation of RGCs including kainate, AMPA, and NMDA receptor subtypes. To work towards the neurotransmitter based retinal prosthesis many research groups( Noolandi et al,2003, Peterman et al,2003,Finlayson and Iezzi,2010) are developing different strategies to establish a device which could release neurotransmitter(glutamate) focally in small volumes in acute retina whole mounts. Such strategies involved developing artificial synapse chip (Peterman et al, 2003) with biocompatible interface for localized neurotransmitter delivery on to the retina. With recent developments in microfluidic devices there has been possibility of ejecting a pattern of neurotransmitters via an inkjet print-head adapted from a desktop printer onto the retina (Noolandi et al, 2003). In recent studies different research groups aim at developing an artificial synapse that would be nano-scaled photovoltaic driven structure like Polypyrrol (Ppy) which would be able to bind and release glutamate as the neurotransmitter from a glutamate reservoir, as shown below, via its tuneable charge. This would then form the basis of artificial biochemical neural implant (ANI) which would facilitate restoration of vision. 16
Fig 7: The above figure conceptualizes the idea of an artificial biochemical neural implant which would help in restore vision in patients suffering from RP & AMD mimicking the natural way of visual processing. The figure in the left depicts the photoreceptor loss as seen in RP & AMD. The figure in the right shows the possibility of devising a glutamate reservoir which could mimic the neurotransmitter release of photoreceptors to the inner retina thereby forming a strategy to restore vision.( Fig adapted from wikipedia.org/idea conceptualized from Prof Dr.Elke Guenther poster presentation)
In the current study both electrical as well as biochemical stimulations formed suitable candidates for retinal prostheses. The activity was seen using microelectrode array (MEA) system. The MEA system has an advantage over other systems, as one could analyse the retinal activity at multiple sites and in turn could assess the entire retinal network activity. Electrical stimulation for RP retina was done via epiretinal stimulation and subretinal stimulation. The former involved stimulating the retina at a single electrode of the MEA from the ganglion cells side and simultaneously recording electrically evoked responses from other neighbouring multiple ganglion cells. The later involved stimulating the retina on the bipolar cells side via bipolar tungsten electrode and recording from the ganglion cells side placed over the MEA. In both the stimulations the read out is at the ganglion cells side in the form of action potentials. By standardizing threshold paradigms for electrical stimulation (for epi-retinal as well as sub-retinal) I could make a preliminary comparison between the different approaches of electrical stimulation in RP retina. Biochemical stimulation was Glutamate Reservoir 17 done via local, small volume sub-retinal application of glutamate using patch pipette in flattened eye cup preparations of RP retinas to investigate the feasibility of activating ganglion cells as means of neurotransmitter-based retinal prosthesis. Aspects of the prosthesis like duration of glutamate ejection, concentration, response latency were also standardized. Further, by different application of drugs like Picrotoxin, Strychnine and Cyclothiazide the response pattern of glutamate receptors could be studied. This could elucidate some preliminary approach towards an efficient neurotransmitter-based retinal prosthesis and thus could form a platform for comparing electrical stimulation with biochemical stimulation, to reason out a better retinal prosthesis of the two with respect to spatio-temporal resolution.
18 Materials and Methods Animals Animals were housed under standard white cyclic lighting, had free access to food and water. FVB strain (rd1) female mice were used (Charles River WIGA, GmbH, Germany).All procedures were performed in accordance with the local ethics committee for the use of animals in ophthalmic and visual research. All efforts were made to minimize the number of animals used and their suffering. Because most of the dynamic changes reduce comparatively around post-natal day 60(P60), from postnatal day 28(P28) onwards, hence most of the measurements were carried out from this age and after. This inference was made from multiple measurements made at different ages of the mice (P28, P40 & P60) to obtain a stable age for applying both electrical and biochemical stimulation.
In-vitro acute retina Preparation Retinae from P60 and older rd1 mice were used for recording. In brief, animals were killed by decapitation and the eyeballs were enucleated and cut open along the orra serrata. The lens and vitreous were carefully removed. The neural retina and the pigment epithelium were gently removed from the underlying tissue. The isolated retina was cut into retinal patches of approximately 3X3 mm (Stett et al, 2000) (Fig.8) or at times were whole mounted depending upon the preparation .It was noted that the entire retina preparation was done as quick as possible with less damage to preserve the retina activity. The retinal patches/ whole mounts were placed in artificial cerebrospinal fluid (ACSF) comprised of 125mM NaCl, 25mM NaHCO 3 , 3.5mM KCl, 2mM CaCl 2, , 25mM Glucose and 1mM MgCl 2 , bubbled with 95% O 2 and 5% CO 2 ; Osmolarity 340mOsm, pH of 7.3 -7.4 at 30C , a nd then mounted onto a planar MEA. At times ACSF solution was substituted with AMES medium (Sigma Aldrich, Germany), this substitution did not affect the retina vitality or the recordings. One of the retinal patches were placed ganglion cell layer down onto the 19 recording chamber the MEA with a continuous perfusion rate of 4-6ml/min. The inflow of the medium into the chamber was maintained by a valve and simultaneously the outflow was maintained by a peristaltic pump, in order to maintain the vitality of the retina. The MEA chip was mounted to a MEA stage with integrated heating (Multichannelsystems (MCS), Reutlingen, Germany) and placed onto the table of an inverted microscope (Olympus CK2, Japan). Experiments were performed at a temperature of 37C using a temperature controller, TC02 (MCS, Reutlingen, Germany). The remaining samples were stored at room temperature (RT) and bubbled in ACSF solution until further use (a) (b) (c)
Fig 8: Retina Preparation. The above figure depicts the invitro acute retina preparation and mounting over planar MEA for recording. In fig a) the eyeball was enucleated and cut open along orra serrata and the retina was carefully removed clearing vitreous humour. Fig b) shows mounting of retina over a Planar MEA. It is taken care that the mounting is done carefully without any damage to the retina or the electrodes. Fig c) shows the retina mounted over a Planar MEA seen from an inverted microscope. (Figure adapted courtesy Prof.Dr.Elke Guenther, NMI, Reutlingen)
Micro-electrode array and data recording A Planar MEA containing Titanium nitride (TiN) electrodes (circular shape, diameters: 30 m) on a glass substrate in an 8X8 square-type grid layout (MCS, Reutlingen, Germany) was used for the recording of field potential and spiking activities from RGCs (retinal ganglion cells) It was noted that the retinal patches/ whole mount covered the entire grid layout of the MEA for better recording. A single TiN MEA electrode is made of 200um
20 nanocolumnar 3-D structure that increases the overall surface area of the TiN electrode in comparison to a standard 2-D aluminum or platinum electrode with the same electrode diameter, thereby increasing the overall capacitance and reducing the noise levels, allowing a continuous and reliable electrical stimulation over a period of several weeks, (van Bergan et al, 2003) (Fig.9a, b & c). Four electrodes at the vertices were inactive and one electrode was used as internal reference (iR) electrode. The inter-electrode spacing from centre to centre were 200 m. Impedances of the electrodes were approximately 50 k at 1 kHz. The insulation is made of Silicon Nitride (Si 3 N 4 ). Each electrode has a track to guide it to the contact pad where it contacts the amplifier. In TiN MEA electrodes the track and contact pad is made of Titanium nitride .The MEA60 data acquisition system (MCS, Reutlingen, Germany) is a 60-channel amplifier having a compact design (165 x165x 19mm) and due to the surfacemounted technology of pre- and filter amplifiers the complete electronic circuit and amplifier hardware were built into a single housing to ensure optimal signal to noise ratio of the recording. An input voltage range from -2048 mV to +2048 mV; an amplifier gain of 1100 and sampling frequency of 20 kHz were used for obtaining microelectrode recordings of retinal activity from up to 59 retinal sites (Micheal Fejtl et al, Advances in network electrophysiology using multi electrode arrays, Springer publications) The retinal activity of 59 channels could be visualized over the computer screen with the McRack software. During data acquisition a high pass filtering (second order Butterworth filter, cut off frequency of 50 Hz) was set in order to differentiate actual retinal activity from baseline drifts. Raw recorded data were stored on a hard disk. As activities of more than one RGC can be recorded in one channel of the MEA a procedure to sort multi-unit activities to single unit activity was also carried out using data analysis tools. These tools could be used simultaneously and independently during the recordings.
21 (a) (b)
Fig 9: Planar MEA: Figure (a) shows a regular TiN electrode Planar MEA form Multichannel systems, Reutlingen, Germany bearing TiN electrodes in the middle and tracks guiding the electrode to their respective rectangular contact pads in. (b) shows the grid lay out of 8X8 electrodes over a planar MEA. (c) shows a single TiN MEA electrode at m and nm resolution. The nano columnar stricture of the electrode increases the overall surface of these electrodes and thereby yielding high performance of these MEAs. (Figures taken from MCS application downloads, Reutlingen, Germany)
Electrical Stimulations Epi-retinal stimulation: This type of stimulation involved stimulating from the ganglion cells side layer at a single electrode in the centre of the MEA and recording RGCs responses from the other nearby electrode/channels. Charge-balanced biphasic constant (c ) 22 voltage rectangular pulse trains were generated (anodic pulse first with no temporal separation between two phases.). 50 trains of pulse was applied with a period of 4 sec. Pulse amplitude was varied from 0.5V to 2.0V along with pulse duration which varied from 300 s-1ms in order to visualize a reliably evoked RGC response as seen from previous work (Ryu et al,2009). The stimulation pulses were generated by STG 1004 stimulus generator (MCS, Reutlingen, Germany). McStimulus software was used to feed the stimulation paradigms and thresholds to the stimulus generator (Fig.10). It was seen that while delivering voltage pulse the recording properties of the MEA electrodes were retained and to address this, the MEA 1060-BC amplifier in combination with the MEA Select software was used to make it sure that the available MEA electrodes could be used as stimulation sites (stimulation electrode) by simple mouse clicks. Moreover in order to retain the recording properties of the MEA electrodes selected for stimulation an electronic blanking circuit (BC) has been incorporated into the amplifier to generate blanking signals which could transiently decouple all the MEA electrodes from the main amplifier input stage during the time course of the stimulation. This is so done in order to prevent any stimulus artifacts. The time for generating this blanking signal depends on the stimulus strength and waveform and could be specified via Mc_Select software. These stimulation paradigms and thresholds were common for both epi-retinal stimulation and sub-retinal stimulation as adapted from literature survey (Ryu et al, 2009; Jenson and Rizzo 2007). However these stimulations were delivered to the tissue in two different ways depending on the type of stimulation. As stated above in epi-retinal stimulation a single electrode of the Planar MEA was used for stimulating the ganglion cells and the neighboring electrodes of the Planar MEA were used for recording the RGCs responses. Sub-retinal stimulation: This type of stimulation involved stimulating the retina at the bipolar cells side and recording RGCs responses using Planar MEA. Concentric bipolar tungsten electrode with exposed metal shield (Length=76mm, 0.325mm x 0.126mm x 23 <0.2mm, tip diameter=2-3m, Resistance=1M; Microprobes Life Science, MD,USA) was used for stimulation. The tungsten electrode was mounted on a micro manipulator (Leica Microsystems, Germany) which could lower the tungsten electrode tip into the MEA recording chamber. It was carefully noted that the tip of the tungsten electrode was at the surface level of the bipolar cells without any damage to retina (possibly due to penetration of the tip into the retina) as well as the MEA (possibly due to penetration of the tip into the retina and scratching the insulation and tracks of the electrodes connecting to the contact pads thereby introducing noise to the MEA channels). The stimulus generator was used to feed stimulus paradigms and thresholds to the bipolar tungsten electrodes in turn stimulating the surface of the retina. As stated above the paradigms and thresholds were at a similar range to that of epi-retinal stimulation, as taken from literatures (Jenson & Rizzo, 2008; Thomas M O Hearn et al,2006) (a) (b)
Fig 10: Above figure (a) shows a snapshot of the McStimulus software, from MCS, showing the stimulus paradigms and thresholds to be fed to the stimulus generator, providing the required stimulation of the retina (b) depicts monophonic and biphasic stimulus pulse generated by the stimulus generator. Monophasic pulse Biphasic pulse 24 Biochemical Stimulation Biochemical stimulation involved local and small volume sub-retinal application of glutamate containing solutions in the retina. For local application of glutamate glass patch pipette(Science products GmbH,Germany) with tip openings of 1-2 m was filled with solution containing 2mM glutamate.(This concentration was considered as this was the lowest concentration that was effective in eliciting RGCs response in normal retinas, Finlayson and Iezzi,2010) dissolved in Ringer /ACSF solution. The glass patch pipettes were manipulated into the tissue using micromanipulators (Leica Microsystems, Germany). Due to sharp and fragile pipette tips it was made sure that the pipettes were positioned well in the tissue subretinally without damaging the retina or scratching the electrodes. Images of the electrode positions were captured to determine the focality and spread of the drug application. It was made sure that the position of the electrode for drug applications over period of time remained consistent as that would bring about variability in the signals. A few m variations would not affect the data variably, as the resolution of MEA is not very high. All the drug applications were applied with a pressure ejection module, the two channel Toohey Spritzer pressure system IIe (Science products GmbH, Germany). The glass patch pipette was placed in the pipette holder in the micromanipulator and connected to the pressure outlet controlled by a valve, of the Toohey Spritzer. The pressure, pulse duration were digitalized during recordings. For generating puff applications as trigger pulses during data acquisition / recording as analog data, TTL positive pulse from the Toohey Spritzer was connected to one of the three analog channels of the MEA set up. The trigger to dispense/ puff drug could be controlled with a remote attached to the Toohey Spritzer or could be done manually. The drug application was done every 1min interval (this marked the number of sweeps in McRack software) for different ejection duration (starting from 10ms -100ms). However reliable RGCs response to puff application was visible for ejection 25 duration of 20ms. A slight electronic shift delay of 20-30ms from the command pulse/ trigger was observed as this was the time delay introduced by the capacitance of the pressure system ( valve outlet) . The drug application (glutamate) was done in area of the retina displaying spontaneous activity .The RGCs responses and spatial distribution/focality of the responses was visualized using the data acquisition software, McRack. The acquired data were further analyzed using data analysis tools like Igor Pro software.
Data Analysis Electrical stimulation: Epi-retinal and Sub-retinal stimulation The acquired evoked responses were analyzed using McRack software (MCS, Reutlingen, Germany) and Neuroexplorer v 4.0 by generating Per-event raster plots from averaged data of the total number of trails. For generating a peri-event raster from acquired evoked responses a Trigger detector was set in McRack software that marked the stimulus pulse. This was so done to align a common point for the stimulus pulse over the number of trials. Further Spike sorter from the McRack software was used to sort actual retinal activity from the rest of the data by the use of a threshold detection limit over the total number of trials. It was made sure that the detection limit values were set keeping in consideration the raw data as well as filtered data to prevent any biasing. The sorted data files generated were exported to the Neuorexplorer v 4.0 software for generating peri-event rasters and peri- stimulus time histograms using the option Peri-event rasters from the data analysis tools from Neuroexplorer software. It was made sure that all the scales, time bins were set properly in the analysis window to generate relevant rasters for the data. Peri-event rasters analyzing tool helps in visualizing the retinal activity before and after the stimulus thereby giving a clear representation of the data and helps in differentiating background or spontaneous activity from stimulus-correlated activity. 26 Biochemical stimulation: The data was acquired and analyzed using McRack software and Neuroexplorer v 4.0. The averaged data was represented in form of perievent rasters over the total number of trials. During data acquisition to mark the puff application of glutamate a Trigger detector was set. The threshold for triggering on biological signals was defined from Analog Raw Data streams. The trigger event was considered at time t=0 but one could set the time scale to negative values to display pre-trigger time (in this case it was set at -100ms) i.e. how long before the trigger event the data is considered and how long after the trigger event the data recording is stopped i.e. post-trigger time (in this case it was 800ms). This total time gives the window extent (in this case being 900ms) as shown below in the figure.
Fig 11: Above figure illustrates how trigger detection works. Time t=0 specifies the trigger event ,pre and post trigger time together give the window extent displaying the time the data is being recorded. (Fig from MCS, website) Further the data stream collected is spike sorted to differentiate actual retinal activity from rest of the data using Spike sorter from McRack software. It was very essential to define the spike cut out as the data acquired was in triggered mode. To detect the spike activity in the given window frame spike cut out ranges. The spike cutout ranges were defined depending upon the detection event, which is the time when the threshold (of the spikes) has been crossed (in this case pre-trigger time was set at 1ms and post trigger time was 27 set at 2ms), additionally a Dead time was set (in this case 2ms) which determines for how long after a detection event no new detection event is accepted and to save computer performance it was recommended that the dead time was at least nearing the post trigger time. The sorted data files generated were exported to the Neuorexplorer v 4.0 software for generating peri-event rasters and peri-stimulus time histograms using the option Peri-event rasters from the data analysis tools from Neuroexplorer software. It was made sure that all the scales, time bins were set properly in the analysis window to generate relevant rasters for the data. The generated per-event rasters represent the data for the time window considered for the trigger event for total number of trials. Offline data analysis: The data generated from the peri-event rasters were exported to excel files and were analyzed offline. To determine the effect of glutamate on RGCs, all counts/ spikes above Mean+2 S.D (threshold) was considered. In order to graphically represent the response latencies with respect to spatial distribution and spike activity with respect to duration of glutamate ejection Igor Pro software was used. To generate statistical significance in spike activity with respect to duration of glutamate ejection, Kruskal-Wallis test was performed using Igor Pro software. Bar graphs representing the success ratio of different modes of stimulation were generated using Matlab.
28 Results 1. Optimization of signal to noise ratio of Planar MEA The rate limiting step involved at achieving a good signal to noise ratio over a Planar MEA. It was observed that due to fragile vitreous humour and dead cells resulting from retina preparation the RGCs signals were masked thereby yielding a poor signal to noise ratio (Fig 12a). For overcoming this issue it was necessary that a tight contact was made between the electrodes and the retina to prevent any leak of current yielding a poor signal to noise ratio. Hence a platinum grid with a closely space nylon mesh was placed over the whole mounted retina to adhere tightly over the electrode. This resulted in good signal to noise ratio thereby making feasible to record RGCs activity over MEA. (Fig 12b)
2. Electrical Stimulations 2.1 Stimulus dependent activity of rd1 degenerated retina (P60) via epi-retinal and sub-retinal stimulation: It was seen that due to loss of photoreceptors in rd1 retina and decreased retinal thickness upto 50 %( Zrenner et al, 1999; Sekirnijak et al, 2009) which leads to increased hyperexcitability thereby increasing the level of spontaneous activity. Previous work suggests that ganglion cells of the degenerated retina require high voltage to reach the threshold for stimulation of the remaining neural retina (Chen et al 2006, Jensen and Rizzo 2007). In degenerated retina, for both epi-retinal and sub-retinal stimulation, the pulse amplitude (biphasic rectangular pulse, anodic followed by cathodic pulse without any interphase delay) was varied from +/- 0.25 to 1V with a fixed duration of 900s for a period of 4s. For epi-retinal stimulation, reliable ganglion cell response was found at +/- 0.75V (+/- 750mV, n=10/50) and for sub-retinal stimulation it was seen at +/- 0.5V (+/- 500mV, n=5/40). By varying the pulse duration from 300s to 1ms with fixed amplitude of +/- 0.75V for epi-retinal and +/- 0.5V for sub-retinal, reliable time duration was found at +/- 450s 29 (t=900s) for both the stimulation. (Fig 13a and 14a). The short latency spikes (around 2ms) could not be observed probably due to the contamination with stimulus artifacts,observed were long latency spikes (around 20ms-50ms) post electrical stimulation. For both epi-retinal and sub-retinal stimulation the read out was ganglion cell response, as a result of which the stimulus and stimulus correlated signals from RGCs were similar .This long latencies, was possibly due indirect activation of retinal ganglion cell network (Jensen and Rizzo, 2007) or could be due to different type of ganglion cells responsible for the different latencies (Jensen and Rizzo, 2008). Therefore it was seen that the threshold parameters (pulse amplitude and pulse duration) and pulse paradigms (adapted from literature) were comparable for both epi-retinal and sub-retinal stimulation.
2.2 Peri-event raster plot for determining stimulus correlated activity for epi-retinal and sub-retinal stimulations: Peri-event rasters represented the spontaneous activity of the rd1 retina followed by epi- retinal stimulus locked activity of the retinal ganglion cells. It was observed that in epi- retinal stimulation the activity lasted around 100ms post electrical stimulation, before it returned to its baseline (Fig 13b). Interestingly, it was observed that in sub-retinal stimulation the retinal ganglion cells followed a stimulus correlated oscillatory pattern(n=2/4), with a frequency of around 10Hz , which was prominent for few milliseconds and then subsequently fades away (as shown in fig 14b, around 350ms). To figure out the reason for such an oscillatory pattern a bath application of strychnine and picrotoxin cocktail was done during sub-retinal stimulation (figure not shown). It was observed that the oscillatory pattern was disrupted, which suggests that since sub-retinal stimulation explores the remnant retinal circuit, these oscillatory pattern could be notably due to the inhibition of lateral pathway onto the excitatory vertical pathway. In both the type of stimulations, the latencies were measured upon the spontaneous activity. The threshold for measuring 30 activity responsible due to electrical stimulation was calculated as (mean + 2xs.d) of spontaneous activity (counts/bin) before the command pulse. Such a calculation considers the marginal errors and fluctuations in the data set and additionally could give unbiased estimation of the data set. The activity that crossed the threshold post onset of command pulse was the activity responsible due to electrical stimulations. The respective latencies were determined as the time from the command pulse to the first time bin in the peri- stimulus histograms that exceeded the threshold activity. Mean latency for epi-retinal stimulation was around 24.3 +/- 0.94ms and for sub-retinal stimulation was 27 +/- 3.2ms (Latency in terms of Mean+/- SEM, Table 1).
3. Biochemical (Glutamate) stimulations The lack of well defined spatial resolution due to electrical stimulation and lack of cellular specificity electrical stimulation had few limitations in encoding important sensory features used in normal central visual processing. To circumvent these limitations a more naturalistic means of stimulation of RGCs was hypothesized which could encode a natural vision. The natural vision is encoded as neuro-transmitter signals with glutamate as the primary transmitter. Thus to determine the feasibility to develop such a hypothesis and standardize different parameters of such a neuro-transmitter based prosthesis local and small volume application of glutamate was performed in rd1 retina invitro.
3.1 Suppressive effect of local glutamate application on rd1 retinal ganglion cells: Local exogenous glutamate application at the site of spontaneous activity induced an initial suppression of firing activity followed by increased activity /excitation (n=5/10). (Fig 15). On application of 2mM glutamate (the minimum concentration of glutamate which could elucidate retinal ganglion cell response in normal retina, Finlayson & Iezzi, 2010). The peri- event rasters show an activity window of 900ms of which 100ms is prior to the application 31 of the glutamate puff (at time, t=20ms) followed by suppression of activity. The suppression was followed by an initial increase in activity which persisted over the entire time window or at times returned to baseline activity. This suppression could be possibly due to two major reasons. Firstly, glutamate application leads to rapid desensitization of AMPA receptors present on the retinal ganglion cells (Massey and Miller, 1988). Another possible reason could be the excitation of glutamate receptors on amacrine cells by glutamate applications which in turn released GABA or glycine to induce an inhibitory effect to the ganglion cell activity (Bloomfield and Dowling, 1985).
3.2 Oscillatory behaviour of RGCs activity of rd1 retina on local glutamate application( n=2/5): In few experiments it was seen that local exogenous application of glutamate not only did induce an initial suppression of retinal ganglion cell activity followed by an increase in activity but also could show oscillatory behaviour which was persistent over the entire time window (Fig 16, t=800ms). The oscillations were at a frequency of approx. 10 Hz. The oscillations could probably be due to the activation of glutamate receptors present on the amacrine cells which could in turn release GABA or glycine over a period of time causing inhibition. This inhibitory effect shapes the activity from the retinal ganglion cells thereby resulting in an oscillatory pattern. (Bloomfield and Dowling, 1985; Yang XL, 2004) .
3.3 Spatio-temporal effectiveness of local glutamate stimulation in rd1 degenerated retina: For elucidating the effectiveness of spatio-temporal resolution of local glutamate stimulation, the response latencies were measured with respect to the distance from the site of glutamate application. The latencies were measured upon the spontaneous activity. The threshold for measuring activity responsible due to glutamate stimulation was 32 calculated as (mean +/- 2xs.d) of spontaneous activity (counts/bin) before the command pulse. Such a calculation considers the marginal errors and fluctuations in the data set and additionally could give unbiased estimation of the data set. The activity that crossed the threshold post onset of command pulse was the activity responsible due to glutamate stimulation. The respective latency was determined as the time from the command pulse to the first time bin in the peri-stimulus histograms that exceeded the threshold activity. It was seen that mean latency at the site of ejection was 16+/-3 ms. The mean latency of the onset of activity at an inter electrode distance of 200m was 20+/- 4.2ms. Additionally at a diagonal inter electrode distance of 282m the mean latency was 25+/- 2.25ms. No prominent effect of glutamate was visible at >=400m distance from the site of ejection (all values Mean+/-SEM; for n=5; Table 2). There was no significant difference between the latency responses with respect to the spatial distribution from the site of glutamate ejection (Fig 17) which suggested that the spatio-temporal resolution was not well achieved which could be due to the resolution of MEA or probably that the effectiveness does not depend upon the distance from the site of ejection of glutamate rather on the cell type of retinal ganglion cells. (Finlayson et al, 2010).
3.4 RGC response dependence on glutamate ejection duration in rd1 degenerated retina: RGC responses were directly controlled by varying the duration of glutamate ejection. The number of spikes increased with increased duration of glutamate ejection. The latency of responses, as may expected, did not vary with the duration of ejection. The number of spikes evoked by glutamate application increased with the duration of ejection to a maximum which was 100ms and then decreased with further increase of duration to 200ms. A significant difference of RGC activity from 20ms to 100ms was seen. (Kruskal Wallis test, n=3, P< 0.05, Fig 18). 33
3.5 Local glutamate application on rd1 retina in presence of GABA and glycine blockers: To reason out the suppression effect post glutamate application due to inhibitory effects from amacrine cells, local exogenous application of glutamate on the retina perfused with a bath solution of strychnine (2M) and picrotoxin (50M) in ACSF was performed. It was observed that due to blockade of GABA and glycine, the overall activity of the retinal ganglion cells increased( 39 +/- 13%,mean+/-s.d,n=5/10). However it was seen that although the suppression effect, post glutamate stimulation had reduced but was not completely removed (Fig 19 a & b). This suggested that apart from inhibitory effects from the amacrine cells and horizontal cells, there exists another possibility accounting for this suppression.
3.6 Local glutamate application on rd1 retina in presence of only cyclothiazide (CTZ): The second probable reason accounting for suppression in RGCs activity was addressed by local application of glutamate+CTZ (50M) on to the retina perfused in a bath solution of cyclothiazide in ACSF was performed. It was seen that the suppression of RGCs decreased to a considerable amount with an overall increase in activity as shown in the figure (Fig 20 a & b). CTZ is said to destabilize the AMPA receptors present on the retinal ganglion cells and also inhibit some GABAA receptors thereby preventing rapid desensitization of the AMPA receptors and increasing the overall activity (5+/- 4%, mean +/- s.d, n=4/10). Therefore the results demonstrate that a substantial portion of this suppressive actions account from the desensitization of AMPA receptors.
34 3.7 Local glutamate application on rd1 retina in presence of cyclothiazide, strychnine and picrotoxin: To further confirm the suppression effect accounted due to the above two probable reasons, i.e., the inhibitory effects due to GABA and glycine and desensitization effect of AMPA receptors due to puff application of glutamate, thereby ruling out any other possibility contributing to this suppression effect; local application of glutamate+CTZ (2mM +50M) was done in presence of a bath of ACSF containing CTZ (50M), strychnine (2M) and pictrotoxin (50M). It was seen that the suppression effect due to glutamate application completely disappeared with an overall increase in activity (30+/- 10%, mean +/- s.d, (n=3/7) (Fig 21a & b). This suggested that glutamate application is likely to have both direct and indirect effect on the retinal ganglion cells.
4. Success ratio between electrical stimulation and biochemical stimulation To sum up the entire set of successful stimulus correlated responses a normalized success ratio plot was generated with respect to different modes of stimulation. It was seen that glutamate stimulation successfully yielded stimulus correlated RGCs activity for 50% (n=5/10) of the experiments. However the success ratio was low for electrical stimulation in both epi-retinal and sub-retinal stimulation with 20 %( n=10/50) and 12.5 %( n=5/40) respectively (Fig 22a). This could suggest that glutamate application, being a more naturalistic phenomenon, has a higher probability of eliciting response in comparison to electrical stimulation. Furthermore sub-retinal stimulation was not well achieved (matter of number of successful stimulations) in comparison to epi-retinal stimulation. This suggests that further standardization with regards to positioning of the tungsten electrode in the retina is required, in order to get comparable set of recordings as epi-retinal stimulation. Further more a summary graph (Fig 22b) depicting the mean latencies of electrical (epi- retinal and sub-retinal) stimulation and biochemical stimulations showing the comparative 35 temporal effectiveness in rd1 retina. The overall spatial resolution in Table 3 showed that the biochemical stimulation although did not have a well defined spatial resolution (around 282m) but had a restrictive focality in comparison to electrical stimulation, which could again suggest that the glutamate stimulation has a better spatio-temporal effectiveness in comparison to electrical stimulation. However to draw a conclusion further investigations needs to be done.
36 Figures & Tables (a) (b)
Fig 12: Above figure (a) shows a poor signal to noise ratio obtained retinal ganglion cells placed on a planar MEA (30m electrode diameter, 200m inter-electrode distance). Fig (b) shows a good signal to noise ratio obtained from the RGCs placed on the same planar MEA. The signals in (a) were poor due to the masking of signals because of fragile vitreous humour or preparation yielding dead cells or loose contact with the MEA electrodes. In (b) post optimization with the platinum grid enabled a good contact with the MEA electrodes which yielded a good signal to noise ratio.The red circles encircles retinal ganglion cell activity.
80ms 8V 80ms 8V
37
(a)
(b)
Fig 13: Above figure (a) shows high pass (50Hz) filtered spike waveform, evoked electrically from RGCs of rd1 retina via epi-retinal stimulation. The arrow in black indicates the biphasic pulse stimulus and the red circle encircles the RGCs activity post stimulus application. Fig (b) is the generated peri-event raster of the spike response in (a), the green arrow indicates the spontaneous activity prior to stimulation, the red arrow shows clear stimulus correlated activity of the RGCs post stimulus. The stimulus correlated activity lasts for over 100ms and then returns to baseline activity. The mean latency of the first RGC spike post epiretinal stimulation was 24.3+/-0.94 ms (mean+/- SEM; n=10) -0.5 0 0.5 Time (sec) 0 40 80 Time(sec) Counts/ bin 100ms Bin width:2ms 100ms 40 counts/bin 10V Spontaneous activity 38 (a)
(b)
Fig 14: Above figure (a) shows high pass(50Hz) filtered spike waveform, evoked electrically from RGCs of rd1 retina via sub-retinal stimulation. The arrow in black indicates the biphasic pulse stimulus and the red circle encircles the RGCs activity post stimulus application. Fig (b) is the generated peri-event raster of the spike response in (a), the green arrow indicates the spontaneous activity prior to stimulation; the red arrow shows clear stimulus correlated activity of the retinal ganglion cells. The rasters depicts typical oscillatory activity, of around 10Hz, which lasts for around 350ms (as above) and then returns to baseline activity. Oscillatory pattern suggests the inhibitory effects on the excitatory pathway thereby shaping the RGCs signals. The mean latency of the first RGC spike post subretinal stimulation was 27+/-3.2 ms (mean+/- SEM; n=4) -0.5 0 0.5 Time (sec) 0 20 40 Time(sec) Counts/ bin 100ms Bin width:3ms
100ms 20 counts/bin 10V Spontaneous activity 39
Summary of latencies of RGCs of rd1 retina to epi-retinal and sub-retinal stimulations
Rd1 Retina(P60)
Epi-retinal stimulation
Sub-retinal stimulation
Mean Latency (ms)
24.3+/-0.94
27+/-3.2
No. of stimulus correlated responses(n)
10
5
Total no. of responses(N)
50
40 Value of latency: mean+/-SEM Table 1: Above table illustrates the mean response latency of stimulus correlated activity for epi-retinal and sub-retinal stimulation. It depicts also the successful number of stimulus correlated activity (electrical stimulations) out of the total number of RGCs responses (spontaneous activity). Table suggests no significant difference in latency of both the stimulation suggesting that both epi-retinal and sub-retinal stimulations stimulate the same neural element.
40
Fig 15: Peri-event raster plot showing suppression of spontaneous activity of RGCs in rd1 retina to local exogenous application of glutamate. Suppression is followed by excitation and increased firing rate. The black arrow indicates the glutamate application (2mM) and the red arrow indicates the increased excitation of RGCs post suppression. The blue arrow indicates the suppression(notch) post glutamate stimulation (n=5/10).
Fig 16: Peri-event raster plot shows oscillatory pattern of activity of RGCs post application of glutamate stimulation. Focal application of glutamate (2mM) is indicated by the black arrow. The blue arrow indicates the suppression post glutamate stimulation(notch) and the red arrow indicates the increased excitation of RGCs post suppression. The brown arrow shows the oscillatory activity over a period of time. The oscillations (around 10Hz) could be a result of activation of glutamate receptors on amacrine cells which in turn release GABA or glycine inhibiting the excitatory activity of RGCs(n=2/5).
Summary of response latencies of RGCs with respect to the distance from glutamate ejection site of rd1 retina.
Distance from ejection site(m) Rd1
Retina(P60) Site of puff application, 30m, electrode diameter Inter electrode distance, 200m Diagonal inter electrode distance, 282m
>=400m
Latency (ms)
16+/-3
20+/-4.2
25+/-2.25 No prominent effect of glutamate
No. of stimulus correlated responses(n)
5
5
5
5
Total no. of responses(N)
10
10
10
10 Value of latency: mean+/-SEM Table 2: Above table illustrates the response latencies of the RGCs activity with respect to distance from the site of ejection of glutamate. There is no significant difference between the latencies with the interelectrode distance (200m) and diagonal interelectrode distance (282m) of the MEA. No prominent effect of glutamate was seen at >=400m. It depicts also the successful number of stimulus correlated activity( glutamate stimulation) out of the total number of RGCs responses (spontaneous activity)
43
Fig 17: Above graph indicates the spatio-temporal effectiveness of local glutamate stimulation. The above graph shows response latencies of RGCs in rd1 retina with respect from distance from ejection site of glutamate (2mM). The ejection site is take at 30m considering the diameter of the MEA electrode. The response latencies did not change significantly over different interelectrode distances, suggesting a week spatio-temporal resolution. Beyond 400m no prominent effect of glutamate stimulation was seen. (values indicate mean latency+/-SEM, n=5)
30 25 20 15 10 5 0 R e s p o n s e
l a t e n c y ( m s ) 500 400 300 200 100 0 Distance from ejection site(m) No prominent effect of local glutamate 44
Fig 18: Above graph shows the excitation of RGCs (in terms of normalized counts per bin). There is increase in activity with increase in glutamate (2mM) ejection duration until 100ms (to which all the values were normalized to) in rd1 retina. With further increase in ejection duration the activity decreased (at 200ms). There was a significant increase in activity from 20ms to 100ms (P<0.05, Kruskal-Wallis test, n=3).
1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1
N o r m a l i z e d
c o u n t s / b i n 300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 Glutamate ejection duration(ms) *
45
(a) (b) Fig 19: Above peri-event rasters from the same experiment (a) shows suppression(notch) of spontaneous activity of RGCs in rd1 retina to local exogenous application of glutamate. Suppression is followed by excitation and increased firing rate. Fig (b) depicts local glutamate application in a bath solution of picrotoxin (50M)and strychnine(2M)and ACSF. The raster plot shows a level of suppression after glutamate application, however there is an overall increase in activity due to blockade of GABA& glycine .The black arrow indicates the glutamate application (2mM), the blue arrow indicates the suppression and the red arrow indicates the increased excitation of RGCs post suppression. 0 0.5 1 Time (sec) 0 10 20 30 40 ch_25_unit_0 0 0.5 1 Time (sec) 0 10 20 30 40 ch_25_unit_0
1 0.5 0 Count s/bin
0
10
20
30
40 10 counts/bin
Time(sec) Ch_25_unit_0
Time(sec) 100ms Bin width:5ms
1 0.5 0 Count s/bin
0
10
20
30
40 10 counts/bin Ch_25_unit_0 8V
100ms Bin width:5ms
46 (a)
(b)
Fig 20: Above peri-event rasters from the same experiment (a) shows suppression(notch) of spontaneous activity of RGCs in rd1 retina to local exogenous stimulation to glutamate+CTZ(50M). Suppression is followed by excitation and increased firing rate. Fig (b) depicts local glutamate application in a bath solution of CTZ and ACSF. The raster plot shows a considerable amount of reduction in the suppression effect post stimulation with overall increase in activity due to CTZ due to reduced desensitization of AMPA receptors as well as blocking subtypes of GABAA receptors. The black arrow indicates the application of Glu (2mM) + CTZ(50M). The blue arrow indicates the suppression(notch) post glutamate stimulation in (a) and the red arrow indicates the increased excitation of RGCs post suppression (n=4/10).
Time(sec) Counts/ bin 0 5 10 15 0.0 100ms Bin width:5ms
5 counts/bin Time(sec) Counts/ bin 0 0.5 1
10 0 5 15 100ms Bin width:5ms 5 counts/bin 47 (a)
(b)
Fig 21: Above peri-event rasters from the same experiment (a) shows suppression of spontaneous activity of RGCs in rd1 retina to local exogenous application of glutamate+CTZ(50M). Suppression is followed by excitation and increased firing rate. Fig (b) depicts local glutamate application in a bath solution of CTZ(50M), picrotoxin(50M), strychnine(2M)and ACSF. The raster plot shows no more suppression effect post stimulation with overall increase in activity due to blockade of GABA& glycine along with reduced desensitization of AMPA receptors .The black arrow indicates the application of Glu (2mM) + CTZ(50M). The blue arrow indicates the suppression(notch) post glutamate stimulation and the red arrow indicates the increased excitation of RGCs post suppression(n=3/7).
0 0.5 1 Time (sec) 0 10 20 0 0.5 1 Time (sec) 0 10 20 100ms Bin width:5ms
10 counts/bin 1 0.5 0
0 10 20 Counts/ bin Time(sec) 1
0.5 0 Time(sec) Counts/ bin 0 20 10 100ms Bin width:5ms
10 counts/bin Ch_21_unit_0 Ch_21_unit_0 48
(a)
(b)
Fig 22: The above bar graph (a) represents the normalized success ratio of different modes of stimulation. It is evident that biochemical stimulation (BCHEM);N=5/10 holds a greater degree of successful stimulation of the degenerated retina in comparison to electrical stimulation (both epi-retinal (EPI);N=10/50 and sub-retinal (SUB) ;N=5/40. Additionally sub- retinal stimulation has a low success ratio in comparison to epi-retinal stimulation. In bar graph (b) the mean latencies from epi-retinal stimulation (24+/-0-94 ms), sub-retinal stimulation (27+/-3.2ms) and biochemical stimulation (22+/-3ms) is shown. This shows that the mean latencies from all the stimulations are significantly comparable, with no much of difference. 49
Summary of overall spatial resolution for electrical and biochemical stimulation in rd1 retina. Overall spatial resolution Rd1 Retina Electrical stimulation Biochemical stimulation
Spatial resolution achieved
Poor
Not well defined
Focality
Random
Well defined over a given distance(282m)
Possible reasons
Random spread of retinal ganglion axons
Poor MEA resolution
Table 3: Above table illustrates the comparison between electrical and biochemical stimulation based upon the overall spatial resolution achieved. Although the above experiments could not resolve well the spatial resolution in biochemical stimulation, however the focality in biochemical stimulation was well defined in comparison to electrical stimulation. The table also shows the possible reason due to which the spatial resolution was not well achieved in both the cases. Thus, suggesting possible reasons for trouble shooting and for achieving a good spatial resolution.
50
Discussions and Implications The present study examined the spiking activity of the retinal ganglion cells in degenerated rd1 retina occurring spontaneously and evoked by repetitive electrical and biochemical stimulations. The major focus intended to study the responses of the retinal ganglion cells on stimulation of bipolar cells electrically as well as biochemically. Additionally the study included an analysis of spatiotemporal effectiveness of the retinal ganglion cells accounted due to electrical stimulation and due local application of glutamate in rd1 retina. It aimed at elucidating the parameters for electrical stimulation, such as, thresholds and pulse paradigms for stimulating rd1 retina via epi-retinal and sub-retinal stimulation. It addressed at a preliminary comparison of the epi-retinal stimulation versus sub retinal stimulation, in order to determine, which of the two approaches would be suitable enough for electrical stimulation. Simultaneously, exploration of biochemical stimulation was done during the thesis. An attempt was made at devising a locally applied glutamate application onto the degenerated retina, as an initial step to determine the feasibility of a neurotransmitter- based retinal prosthesis. This gave way to compare two potent candidates of retinal prostheses and to work towards developing an efficient model for the visual neuroprosthetics that could elucidate spatio-temporal parameters to transmit maximum visual information.
Electrical stimulation: Bionic implants fall into two categories. Sub-retinal implants and epi-retinal implants. (Humayun et al, 1996; Chow and Chow et al, 1997). Subretinal implants are engineered with the goal of replacing the lost photoreceptors and stimulating the surviving retinal circuits to restore natural visual information, which would notably include information on light adaptation, object motion, contrast vision and so on. However epi-retinal implants placed on vitreal surface of retina would theoretically drive ganglion cell 51 populations, by surpassing much of the retinal circuitry. These approaches depend explicitly upon the preservation of the neural retina. In this study I could successfully optimize the parameters for visualizing retinal ganglion cell activity on planar microelectrode array (MEA) which in itself was the rate limiting step of the thesis. Secondly I could stimulate rd1 degenerated retina, via epi-retinal as well as sub-retinal stimulation. The degenerated retina responded to repetitive electrical stimulation and was evident from the retinal ganglion cells responses using multi-electrode array. As from the above results it was evident that the stimulation thresholds and pulse paradigms were comparable in both types of stimulations. However, the response pattern differed, in sub-retinal stimulation oscillatory behaviour of the ganglion cells was observed .This could be due the co- activation of the vertical as well as the lateral pathways of the retina. The vertical pathway involves signal transduction from the bipolar layer to the ganglion cell layer which is excitatory and uses glutamate as a major neurotransmitter. The lateral pathway involves inhibition from the horizontal cells and the amacrine cells. The interaction of the lateral pathway with the vertical pathway shapes the retinal ganglion cell response yielding an oscillatory pattern ( Neuenschwander et al, 2002). Yazulla et al (Yazulla et al, 1997) have demonstrated earlier a two fold increase in GABA content of rd1 retinas which could account for such oscillatory patterns. This pattern however was not visible for epi-retinal stimulation possibly as epi -retinal stimulation surpassed the inner retinal circuitry and inhibitory circuits hence yielding a sharp rise in activity post stimulation and subsequently fading away. In matter of temporal resolution, the response latencies were also comparable for epi- retinal and sub-retinal stimulation, which suggests that both sub-retinal and epi-retinal stimulation more or less activate the same neural element and utilize the same amount of neural processing for transmitting visual information (Thomas O Hearn,et al, 2006). These spikes could be classified as fast spikes or short latency spikes which lasted for around 52 2ms and long latency spikes which was around >=15ms. In the above results only long latency spikes could be discerned as the stimulus artifacts obscured the first few milliseconds of the recordings thereby masking the short latency spikes. These variations of spike latency are possibly due to different retinal ganglion cell types as well as the different stimulation parameters. The short latency spike yielded from direct stimulation of the retinal ganglion cells and the long latency spikes could be due to the stimulation of the network of ganglion cells (Jensen & Rizzo, 2007). However due to the inherent lack of cellular specificity, electrical stimulation used in most present devices is limited in spatial distribution. Additionally electrical stimulation non- specifically stimulates somata and axons of ON & OFF retinal ganglion cells, bipolar cells and very likely also the inhibitory cells like amacrine cells in nonphysiological sequences that may reduce contrast and spatial localisation within the retina (Finlayson and Iezzi, 2010). Thus electrical stimulation of the rd1 retina has few limitations in encoding important sensory features used in normal central visual processing. In this study the spatial distribution of electrical stimulation was also addressed. The stimulus correlation observed in few channels was variable in their distribution with respect to the stimulation site which suggests that the electrical stimulation could activate a network of different ganglion cells. A probable reason could be due to continuous remodelling of the retina, which on electrical stimulation spread the stimulus pulse over the entire network activating group of cells randomly which aligns to the stimulus pulse. Another reason could be random distribution of retinal ganglion cell axons in degenerated retina (Ryu et al, 2009). To the extent of spatial resolution limitation, one could hypothesize a more naturalistic means of stimulating retinal ganglion cells (RGCs) for efficient retinal prosthesis that could decipher the spatial resolution. This was addressed by focal application of glutamate as the neurotransmitter based prosthesis that would more or less mimics the natural vision.
53 Biochemical stimulation: Due to the shortcomings of the electrical stimulations in addressing the spatial resolution, biochemical implants came into play. However several groups are working towards this prosthesis (Noolandi et al, 2003) yet the problems in localized chemical delivery has been a major hurdle in most of the devices. To achieve the resolution to stimulate at a single cell level or few cells is what the prosthesis aims at and thereby generating a high spatial resolution for achieving good vision. In the thesis the major aim was to elucidate the role of focal glutamate delivery onto rd1 retinae placed on a multi-electrode array. This gave an opportunity to analyse the activity pattern over a network and also visualize the focality of the application. From the above results it was seen that local exogenous application of glutamate at the site of spontaneous activity induced an initial suppression of firing activity followed by an increased activity (which looked more or less like notch). One would anticipate that the suppression was due to two major reasons. One, as there is AMPA receptors on the retinal ganglion cells and AMPA receptors on application of glutamate desensitize rapidly (Bloomfield and Dowling et al, 1985). The other reason could be that glutamate application depolarizes the glutamate receptors present on the amacrine cells which in turn release GABA or glycine which could lead to inhibition of the excitatory pathway. For addressing these two possibilities, application of glutamate with different bath applications was done as shown above. Bath application of strychnine and picrotoxin (gycinergic and GABAergic blockers) increased the basal activity; however there still existed suppression in activity (presence of a notch) as shown. This suggested that although there was an increase in activity, inhibitory signals did not contribute much to the suppression of the firing activity post glutamate stimulation, which lead to addressing the other possibility of the suppression of activity. Bath application of cyclothiazide alone lead to considerable decrease of suppression of activity(diminishing the notch) , with an overall increase in basal activity. This was due to reduced desensitization effect of AMPA receptors suggesting that the desensitization of 54 AMPA receptors contributed substantially to the suppression in activity. Cyclothiazide is believed to destabilize the desensitized state of the AMPA receptors (Yamada and Tang, 1993) .Additionally it inhibits GABAA mediated currents, thus increasing glutamatergic response (Deng & Cheng, 2003) thereby facilitating high frequency synaptic transmission over a longer period of time. Therefore a bath application of strychnine, picrotoxin and cyclothiazide combination removed this suppression of activity thereby suggesting that glutamate application has both direct and indirect implications on retinal ganglion cells. Furthermore, interestingly, in few experiments the RGCs responded to glutamate application in an oscillatory pattern which could suggest that glutamate application could activate the glutamate receptors of the amacrine cells which on release of GABA or glycine gave a timely inhibition, although this requires further investigation to quote it conclusively. The response latency accounted for the temporal dynamics. It was further observed that the RGCs response latencies to locally applied glutamate were relatively comparable to latencies obtained via electrical stimulation, which suggested that a naturalistic phenomenon had a temporal effectiveness like that of electrical stimulation. However this needs further investigation with age groups of advanced photoreceptor degeneration to study whether the sequelae associated with progressive photoreceptor degeneration affected the sensitivity and properties like latencies of RGCs to glutamate, inorder to draw conclusive information. The spatial effectiveness was addressed with the resolution of the MEA. From the above results it was seen that the RCG response latencies did not vary significantly with the distance from the glutamate ejection site which was not as expected. This probably could be due to the poor MEA resolution or dense network of axons running across the degenerated retina or physiologically, the spatial distribution depends on the cell type rather than the RGC distance from the site. To resolve this issue a MEA with higher resolution needs to be considered. 55 The spread of inhibition due to the inhibitory effects from the amacrine cells and horizontal cells and the temporal dynamics warrants further investigation in determining an efficient spatio-temporal resolution. Furthermore this investigation could help in determining an optimal separation of ejection sites in a prototype retinal prosthetic device. The duration of ejection did not change the firing pattern of the RGCs or the latency of response, but, however increased the level of activity this increase in activity was due to increased exposure of the retinal ganglion cells to a glutamate rich environment. The time duration of ejection is indeed an important parameter to be considered as one could modulate the RGCs activity by varying the durations of the glutamate application (Finlayson and Iezzi, 2010). Apart from the above aspects the glutamate concentration plays a role in modulating the RGC activity. At higher concentration chances of excitotoxicity due to glutamate could lead to overstimulation and death of the retina. Therefore an optimal level of glutamate concentration is required to prevent toxicity. An application of , 2mM glutamate could elicit RGCs response in rd1 retina, in vitro. This concentration was adapted from different literatures one being the Finlayson et al, 2010, which suggested that a 2mM concentration was the minimum concentration of glutamate to efficiently stimulate the RGCs in normal retinas. Although, this concentration was indeed higher than the normal synaptic condition in vivo (around 0.5mM)( ( Finlayson and Iezzi, Springer, In press).However in the above experiments a high perfusion rate of 6ml/min could rapidly remove the glutamate preventing any obvious amount of toxicity. This implies that lower concentrations of locally applied glutamate could be effective in designing an effective prosthesis in vivo. However to design a precise prosthetic device requires a lot of considerations. The goal of the thesis was to obtain a birds eye view comparative analysis of the electrical stimulation versus biochemical stimulation. Apart from doing a comparative analysis the main goal intended to elucidate different aspects of each of this prosthesis in depth. However a substantial amount of experiments needs to be performed considering other 56 dimensions like optimal concentration levels, exact location of site of ejection, multiple application of glutamate with specific time interval and so on and so forth to draw a concrete conclusion. However due to time limitation, this could not be accomplished. This gave an opportunity to reason out the possibilities and shortcomings of each of these prosthesis models. It was seen that although electrical stimulation could provide a better temporal resolution, in matter of short latency spikes, but it could not address the spatial distribution which indeed is an important parameter for any retinal implant. However it was seen that the biochemical stimulation had comparable time latency as electrical stimulation. Also the spatial distribution was addressed by biochemical stimulation to quite some extent, implying that biochemical stimulation has an advantage over electrical stimulation at addressing spatial complexity which electrical stimulation, could not. Additionally one could try reasoning out that the electrical(sub-retinal) stimulation and biochemical stimulation are comparable as they use the entire machinery of the remnant retina, which suggests that proper standardization of sub-retinal parameters (like positioning of electrode) could help in achieving a spatio-temporal resolution, via sub-retinal stimulation. However, it requires a lot of investigations before conferring any conclusions. For designing an efficient prosthesis model spatio-temporal properties have to be considered and since biochemical stimulation could attempt to answer it to quite some extent, this prosthesis model looks promising for developing an efficient retinal prosthesis model for restoring natural vision. Furthermore with development of technical sophistication in times to come one could eventually reach the goal for devising a full blown neuro-transmitter based prosthesis that could mimic the natural vision and help in rescuing vision in blind patients.
57 References 1. Artificial Eye, www.bestneo.com 2. Bloomfield SA, Dowling JE, Roles of aspartate and glutamate in synaptic transmission in rabbit retina. II. Inner plexiform layer. J. Neurophysiology (1985), 53, 714-725 3. Chow AY, Pardue MT, Chow VY, Peymann GA, Liang C, Perlman JI, Peachey NS Implantation of silicon chip microphotodiode arrays into the cat subretinal space. IEEE Trans. Neural Syst. Rehabil (2001). Eng 9:8695. 4. Chris Sekirnjak, Clare Hulse, Lauren H. Jepson, Pawel Hottowy,Alexander Sher, Wladyslaw Dabrowski, A. M. Litke, and E. J. Chichilnisky, Loss of responses to visual but not electrical stimulation in ganglion cells of rats with severe photoreceptors degeneration , J Neurophysiol (2009), 102: 3260-326 5. Dean Scribner et al , Microelectronic array for stimulation of large retinal tissue areas MIT Press ,2005 6. Deng L, Chen G. Cyclothiazide potently inhibits gamma-aminobutyric acid type A receptors in addition to enhancing glutamate responses. Proceedings of the National Academy of Sciences of the United States of America (2003). 100 (22): 130259 7. E. Zrenner, A. Stett, S. Weiss, R.B. Aramant, E.Guenther, K.Kohler, K.D. Miliczek, M.J . Seiler, H. Haemmerle, Can sub retinal micro photodiodes successfully replace degenerated photoreceptors? , Vision research(1999)39, 2555-2567 8. Finlayson and Iezzi, , Glutamate stimulation of retinal ganglion cells in normal and S334ter Rat retinas: A candidate for a neurotransmitter based retinal prosthesis.(2010),51,No:7 9. Francois PaquetDurand, Stefanie M.Hauck, Theo van Veen, Marius Ueffing and Per Ekstrom,PKG activity causes photoreceptor cell death in two retinitis pigmentosa model,J. Neurochemistry (2009)108, 796-810 58 10. Joseph F. Rizzo,1,2 John Wyatt, John Loewenstein, Shawn Kelly, and Doug Shire methods and perceptual thresholds for short-term electrical Stimulation of Human Retina with Microelectrode Arrays, Investigative Ophthalmology & Visual Science, December(2003), 44, No. 12 11. Jensen RJ, Rizzo JFIII. Responses of ganglion cells to repetitive electrical stimulation of the retina. J Neural Eng ,(2007),4, S1-S6 12. Jensen RJ,Rizzo JFIII. Activation of retinal ganglion cells in wild type and rd1 mice through electrical stimulation of the retinal neural network. Vision research(2008), Vol. 48, 1562-1568 13. Humayun MS, de Juan Jr E, Dagnelie G, Greenberg RJ, Propst RH, Philips DH (1996) Visual perception elicited by electrical stimulation of retina in blind humans. Archives of Ophthalmology 114:40_46 14. Humayun MS, de Juan E Jr, Weiland JD, Dangnelie G, Katona S, Greenberg R, Suzuki S. Pattern electrical stimulation of human retina. Vision Res, (1999) 39,2569-2576 15. Markus Meister, Leon Lagnado and Dennis A. Baylor, Concerted signalling by retinal ganglion cells. Science(1995) 270,1207-1209 16. Massey SC, Miller RF. Glutamate receptors of ganglion cells in the rabbit retina: evidence for glutamate as a bipolar cell transmitter. J Physiology(1988), 405, 635-655 17. Neuenschwander,S., M.Castelo-Branco, J.Baron and W.Singer, Feed forward synchronization : propagation of temporal patterns along the retinothalamocortical pathway. Phil.Trans.R. Soc.Lond.(2002) B 257, 1869-1876 18. Noolandi J, Peterman M.C, Hule P, Lee C, Blumenkranz M S, Fishman H.A, Towards a neurotransmitter based retinal prosthesis using an inkjet print head, Biomedical Micro devices(2003), Vol 5, No:3 ;195-199 19. Peterman MC, Bloom DM, Lee C, Bent SF, Marmor MF, Blumenkranz M S, Fishman H.A; Localized neurotransmitter release for use in a prototype retinal interface. 59 20. Ryu SB, Lee JS, Ye JH, Goo YS, Kim CH, Kim KH. Analysis of neuronal activities of retinal ganglion cells of degenerated retinal evoked by electrical pulse stimulation , J Biomed Eng Res, (2009a),30,347-354 21. Sang Baek Ryu, Jang Hee Ye, Jong Seuung Lee, Yong Sook Goo, Chi Hyun Kim and Kyung Hwan Kim, Electrically evoked neural activities of rd1 mice retinal ganglion cells by repetitive pulse stimulation , Korean J Physiology Pharmacology (2009) 13, 443-448 22. Chris Sekirnjak, Clare Hulse, Lauren H. Jepson, Pawel Hottowy,Alexander Sher, Wladyslaw Dabrowski, A. M. Litke, and E. J. Chichilnisky, Loss of responses to visual but not electrical stimulation in ganglion cells of rats with severe photoreceptors degeneration , J Neurophysiol (2009), 102: 3260-326 23. Stett A, Barth W, Weiss S, Haemmerle H, Zrenner E. Electrical multisite stimulation of the isolated chicken retina Vis Res, (2000),40, 1785-1795 24. Webvision , webvision.med.utah.edu/ 25. Yazulla, S., Studholme, K. M., & Pinto, L. H. (1997). DiVerences in the retinal GABA system among control, spastic mutant and retinal degeneration mutant mice. Vision Research, 37, 34713482 26. Zrenner E Will retinal implants restore vision? Science(2002), 295,1022-1025
Books 1. Advances in network electrophysiology using Multi-electrode arrays Makoto Taketani and Micheal Baudry 2006, Springer Science. 2. The first steps in seeing by B.W Rodeick. ,Sinauer Associates. Inc