Function and Structure in Retinal Transplants

Embryonic mammalian donor retina transplanted into the subretinal space of a mature host develops into a graft with wellorganized, but atypical retinal structure. We tested the effect of this organization on rabbitto-rabbit graft functional properties, isolating the graft to avoid contamination of graft responses by host retinal activity. Transient ON or ON-OFF spike-like responses and local electroretinograms (L-ERGs) were recorded simultaneously via a single electrode on the graft surface. These response components depended on stimulus diameter, sometimes in a way indicating antagonistic center-surround receptive field organization and spatial tuning (43%). Other times, the responses were an increasing function of stimulus diameter which saturated for large spots (57%). Response amplitudes were transplantation surgery is to be done with therapeutic aims.

transplanted into the subretinal space of a mature host develops into a graft with wellorganized, but atypical retinal structure. We tested the effect of this organization on rabbitto-rabbit graft functional properties, isolating the graft to avoid contamination of graft responses by host retinal activity. Transient ON or ON-OFF spike-like responses and local electroretinograms (L-ERGs) were recorded simultaneously via a single electrode on the graft surface. These response components depended on stimulus diameter, sometimes in a way indicating antagonistic center-surround receptive field organization and spatial tuning (43%).
Other times, the responses were an increasing function of stimulus diameter which saturated for large spots (57%). Response amplitudes were a monotonically-increasing function of light intensity over the narrow range tested. The L-ERGs were reminiscent of the proximal negative response or M-wave seen in normal retinas, INTRODUCTION Retinal transplantation as a strategy for restoration or rescue of function in retinas rendered afunctional by disease or trauma is an intriguing idea. There has been rapid growth in research focused on this strategy, partially stimulated by recent transplantation efforts in brain related to neurodegenerative diseases such as Parkinson's, Huntington' s, and Alzheimer' s/1-3/. Such studies have been hindered by the fact that the specific properties of the inputs, intrinsic processing, and outputs of the systems affected, are not welldefined. This is especially significant when attempting to study specific correlated functional and structural properties of the implanted tissue. In contrast to these studies, the present study utilizes the retina, for which there is good understanding about the inputs (complex light patterns), intraretinal processing, anatomical microcircuitry, and the output (patterns of ganglion cell activity) /31,32/. Here we demonstrate for the first time, organized, complex physiological function in combination with its underlying neuronal microcircuitry in mammalian sub-retinal transplants.
The use of embryonic or fetal donor tissue permits in sitzt development, differentiation, and persistence in the host retina of most types of replacement retinal elements and the complex circuitry required to process visual stimuli /4,5/. Anatomically, these grafts synaptically integrate intrinsically and may do so with the (mature) host retina /6/. Second, because of the relative immunological privilege of the retina/CNS and the reduced immunogenicity of fetal donor tissue, rejection has not been problematic in allografts. Standard immunosuppression therapies have even permitted widely disparate xenografting like human donor to rat host/2,5,6/.
In initial attempts at retina to retina transplantation, Turner and Blair /7/ injected fragments of embryonic donor retina into a vitreal locus apposed to a retinal lesion site. The grafts, which fused with the host retina, formed layered, folded sheets with rosettes consisting of circular arrays of photoreceptors (outer segments oriented inward) and outer neural layers/8/, del Cerro et al. /9/injected suspensions of fluorescent-tracer labeled neonate retinal cells into the subretinal space of normal and light-damaged host rats. They observed possible anatomical integration of the differentiated donor cells with host retina, and ultrastructural evidence suggestive of photoreceptor to secondorder neuron synaptic contact. A later study of retinal transplantation in rodents by this group demonstrated, using psychophysical methods, possible restoration of retina-dependent function /33/. Sheets of intact photoreceptor layer were also grafted subretinally into light-damaged adult rat retinas by Silverman and Hughes/10,11/. In these experiments already differentiated photoreceptors were gratied to rd mouse and RCS rat retina mutants, and mature human photoreceptors to immunosuppressed, light-damaged adult rats. The grafts survived and showed potential for phototransduction (opsin immunostaining), development of photoreceptor synapses, and possible involvement in pupillary reflex action/12/.
Several criteria must be satisfied in order for transplanted embryonal retinal tissue to develop into a functional retina within the host. Firstly, the fundamental retinal cell types must develop and mature such that their functionally important components (e.g. photoreceptor outer segment discs, synaptic components, etc.) are present. Secondly, these cells should be organized in a fashion which allows for appropriate cellular connectivity (e.g. photoreceptor synaptic terminals located near bipolar and horizontal cell processes, and bipolar cell terminals near amacrine and ganglion cell processes). Finally, these neurons must make both functional and appropriate synaptic contacts. We have been examining these issues using specific physiological and anatomical approaches in rabbit donor to rabbit host retinal transplants.

METHODS
Retinas from Dutch-belted rabbits served as donor and host. Mature eyes were relatively large, approximating human and primate eyes in size (ca. 20 mm diam.) and, unlike rodent eyes, in their relative volumes of lens capsule and vitreous cavity, making them easily adaptable for manipulation with ophthalmic microsurgical instruments. Aggregates of embryonic day 15 or 16 (El 5 or El6) donor retinas (birthdate, E32), were transplanted to a subretinal locus in the posterior pole of adult male hosts. After development of mature grats (45 to 60 days post transplant), an isolated graft preparation was set up in vitro as a superfused, everted eyecup preparation. The resulting isolated grafts were large (>lmm) and visible under the dissecting microscope.
The transplant surgical techniques are described in detail elsewhere /13/. Briefly, fragments of embryonic donor neural retinas are inserted into a subretinal locus using a flexible Teflon micropipette. The pipette enters the intraocular space via a pars plana approach through a metal guard tube. The guard tube and extendable micropipette are incorporated into a stainless steel surgical handpiece. Donor tissue is ejected from the pipette by hydraulic pressure from a microsyringe. Micropipette manipulation and tissue ejection are viewed through a surgical contact lens.
In the isolation procedure, the host retina is carefully stripped from its underlying RPE/ choroid/sclera base, using microforceps and a microsclerotomy knife. The gra remains attached to the base, usually separating cleanly from the retracting host retina. The host optic nerve is transected where it passes through the choroid/RPE surface. This procedure permits complete detachment and removal of the host retina, and leaves a large, highly visible grat anchored to an underlying host RPE/choroid/sclera base (refer to Fig. 2B).
For electrophysiology, electrodes were epoxyinsulated tungsten with platinum blacked tips, 4-10 megohms resistance (F. Haer & Co.). Amplifier was an Axoprobe-lA with output split into low-pass (LERGs) and high-pass (spikes) frequency bands by a variable frequency, electronic filter (AP Electronics). The electrodes were positioned with a manual micromanipulator (Brinkmann) and stepped using a hydraulic (Kopf) microdrive. Light stimuli were spots formed by aperture stops in the collimated beam of an optical system with tungstenhalogen light source, neutral-density attenuation filters and narrow-band interference filters. Position was steered by orthogonal, galvanometer-driven mirrors. Extracellular spike activity was analyzed as peristimulus spike histograms time-locked to the computer controlled stimulus trigger signal. Fourier analysis of the spike histograms by the computer generated mean amplitude, phase, and variance measures up to N (specifiable) harmonics.
In the in vitro retinal preparation, the eye was enucleated under urethane general anesthesia supplemented by local lidocaine anesthesia around the anterior segment of the globe. Animal care and surgical procedure was in accordance with institutional guidelines and approved by the Sehepens Eye Research Institute's Animal Care and Use Committee and adhered to the ARVO Resolution on the Use of Animals in Research. The lens and anterior segment were excised, vitreous removed, the eyecup flattened using peripheral radial cuts and everted and positioned over a dome support in the perfusion chamber, and held in position by an O-ring clamp. Ames medium supplemented with Nu-Serum (Becton-Diekinson) was superfused over the retina at ca. 2-5 ml/min. A circulating heater (Lauda) and organ-bath components (Radnotti) were used to maintain bath temperature at 35C. pH was monitored on-line by a microsensor (Microelectrodes Inc.) and maintained at 7.4 by adjusting the O2/CO 2 ratio of the gas mixture bubbling through the Ames medium. The physiological condition of the retina was monitored using the recorded ERG responses. Host or graft retinas had stable responses under these conditions for 4 to 6 hours or longer.
For Lucifer Yellow filling, the posterior eyecups with the transplants were lightly fixed in 1% buffered formaldehyde for [30][31][32][33][34][35][36][37][38][39][40] min and washed. The transplants, adherent to the pigment epithelium and choroid, were mounted on a black Millipore filter (type AA, 0.8 micron pore size). The tissue was stained with ethidium bromide (2,7-diamino-10-ethyl-9-phenylphenanthridinium, 1 tg/ml 2.5 gM) for 2 mins and then transferred to a specially designed stage chamber in a Zeiss UEM microscope equipped with epifluoreseence illumination. The transplants were identified by their red ethidium bromide fluorescence, and cells within them impaled with a Lucifer Yellow-filled (2-4% in distilled water) microelectrode (50-100 megohms). Cells were injected with dye by iontophoretic current (several nanoamperes, pulsed negative for several minutes) while the extent of filling was monitored with the fluorescence microscope. In favorable cases, more than 100 cells could be injected in a single transplant. After the dye injections and photographs in whole mount, the tissue was fixed as described below, serially sectioned, and checked for completeness of host retina removal.
For both light and electron microscopy, retinas were fixed with 1.6% glutaraldehyde plus 1% freshly prepared formaldehyde in 0.1 M sodium pho.sphate buffer containing 0.15 mM CaCI2, pH 7.35 for cne hour at room temperature, followed by overnight at 4C. Tissue was then washed, postfixed for one hour with 2% OsO4, dehydrated through a graded series of ethanol and embedded in Epon. One micron sections were cut for light microscopy and stained with a mixture of methylene and toluidine blue. For EM, 80 nm silver sections were cut, counterstained with uranyl acetate and lead citrate, and examined with a Philips 410 EM at 8O kV.

RESULTS
The subretinal transplant technique /13/ produces grats that are well organized morphologically and synaptically, but not identical to normal retinas. There are regions containing "rosettes" of photoreceptors surrounded by appropriately layered retinal elements (Fig. 1). The goal of the research presented in this paper is to VOLUME 5, NO. 2, 1994  Focal field responses (L-ERGs) and spike-like activity were recorded simultaneously at various regions on the graft using a platinized, tungsten extracellular microelectrode (Fig. 3). Mapping the response with a light spot showed that the receptive field was restricted to the area on the grail around the electrode tip; there was no measurable response to a 250 tm light spot, focused onto regions off the graft or even on it, but distant (>250 gm) from the electrode location (Fig. 4) Ultrastructurally, photoreceptor inner and outer segments with connecting cilia and basal bodies appeared well developed /30/. Clusters of photoreceptor terminals frequently exhibited one or several synaptic ribbons with postsynaptic processes arranged in a dyadic or triadic configuration (Fig.  6). Postsynaptic elements at these photoreceptor synapses resembled those belonging to horizontal and bipolar cells. Regions corresponding to the inner plexiform layer exhibited a relatively high synaptic density, with most synaptic terminals being well filled with conventional synaptic vesicles. Occasionally, small and large dense-cored vesicles were seen indicating the presence of biogenic amines and neuropeptides, respectively. Bipolar ribbon synapses were also common, often with postsynaptic processes in a typical dyadic arrangement (Fig. 7A). Amacrine-to-amacrine (Fig.  7B) and amacrine-to-bipolar synapses were common, sometimes in both serial and reciprocal configurations. Gap junctions involving amacrine and bipolar cell processes were also readily identified.
Using intracellular dye filling with Lucifer yellow, we identified retinal neurons with cell body and dendritic morphologies resembling those of all normal retinal cell types with the possible exception of ganglion cells. Two examples of Lucifer yellowfilled, intragraft, amacrine cells are shown in Fig. 8. The dendritic morphology of the cell in Fig. 8A resembles that described for narrow-field AIIamacrine cells, which in normal rabbit retinas act as interneurons between rod bipolar cells and cone ganglion cells /21/. An example of a wide-field amacrine cell, also found in normal rabbit retinas /22/, is shown in Fig. 8B. DISCUSSION The results indicate a high level of function and organization of underlying structure within the transplants. We have shown by light and electron microscopy/25,26/ plus intracellular dye marking /27/that rabbit retinal grats develop in a manner that provides distinct retinal regions where specific retinal processing can take place. The transplants contain in abundance the specific cellular and synaptic components which are necessary for phototransduction and for the processing of visual information in a fashion similar to that carried out by normal retinas. The isolated grafts exhibit spikelike activity organized into spatial receptive fields with excitatory centers, and either inhibitory or no surrounds. They also have light-driven local ERGs reflecting amacrine cell (M-wave) and bipolar cell (b-wave) activity. Thus, rabbit retinal grafts contain the neural components and local synaptic circuitry    necessary for the electrophysiological responses seen in these grafts and may serve for more complex physiological processing of visual information.
A question remains concerning the origin of the organized spike activity within these grafts. It is unlikely that the source of the spike-like activity recorded in the isolated grafts derives from host ganglion cells. Histological examination of the isolated graft preparations indicates the absence of residual host retina, including host ganglion cells (Fig. 2B). Although it is possible that we overlooked host ganglion cells despite examining serial sections through the full extent of the isolated graft preparations, it is highly improbable that this has happened in all the isolated grafts tested (N=9) which gave similar electrophysiological responses (N=8). Also, since we are dealing with subretinal grafts, host ganglion cells would generally be separated from the graPt by the rest of the host retinal layers. Thus any remaining host ganglion cells would likely be embedded in sufficient host retina to make them readily observable upon serial sectioning.
Perhaps ganglion cells developed in the graft are the source of the recorded spike activity. Although this is a more attractive hypothesis than their origin being host ganglion cells, there are some inherent problems with this interpretation. In normal prenatal rabbit retina, ganglion cells outside the visual streak do not begin to differentiate until ca. E20-E24 (those within the streak, even later), significantly later than the age (El 5/El 6) of the donor retinas /23/. By E30 in normal retina, elongation of ganglion cell axons towards relatively proximate tectal targets has begun within the permissive environment of the optic fiber layer/28/.
Thus, although the E15/El 6 donor ganglion cell precursors are not subject to the degenerative influences of axotomization, they do face an  Although the structure of these graft retinas is atypical of the flat laminar organization in normal retinas and their photoreceptor outer segments do not adjoin host pigment epithelium, the transplants do exhibit light responsiveness and well-organized functional properties. Such functionality may be expected to persist based on the demonstrated longterm persistence of light-driven responsiveness in ectopic retinal transplants to the optic tectum in rodents/19,20/. Those ectopic grafts also develop rosettiform organization of their photoreceptors without pigment epithelial contact.
In summary, there are several points that must be made. Foremost, this paper presents studies of the graft retina that develops from the subretinal transplantation of undifferentiated embryonic retina. The objective was to demonstrate that the retinal graft which develops exhibits the intrinsic anatomy and function spatially and temporally appropriate for use in a therapeutic strategy. For this purpose it was essential that the inherent responses of any host retina be eliminated to avoid any "contamination" of the graft's pristine responses. Graft-driven, spatially and temporally appropriate host optic nerve responses are essential for any possible therapeutic usefulness of retinal cell transplants. That is the subject of another study, currently underway, requiring hosts with defined functional deficits, and techniques to distinguish graft from host cells, processes, and synapses. Results of recent anatomical and physiological experiments in other laboratories, notably those of Silverman et al. /12// and Yamamoto et al. /34/, using subretinal photoreceptor or RPE allotransplants to appropriate host rodents (e.g. RCS rats and rd/rds mice), report a degree of restoration or maintenance of synaptic integration between graft and host retina, and possible graft-driven host responses. These results lead us to expect some integration at both the outer arid inner plexiform levels in our rabbit transplant model analogous to the aforementioned OPL integration in the rodent models.
In conclusion, the research reported in this paper may be valuable in several ways. First, to determine whether a transplantation model, using whole neural retinal donors functionally coupled to their host retinas via longitudinal and lateral preganglion cell synaptic and junctional contacts, is a viable therapeutic strategy. Such a strategy would be applicable in retinas afunctional because of retinopathies in which the ganglion cells and optic nerve link between retina and brain are spared. Second, as a model for the prenatal ontogeny of correlated function and structure in the retina. And third, to examine, based on the apparent disparities between well-organized function in the retinal grafts and their relative anatomical disorder, the role of structural redundancy and necessity for highly ordered organization in the normal retina.