Development and Characterization of an Adult Retinal Explant Organotypic Tissue Culture System as an In Vitro Intraocular Stem Cell Transplantation Model

Abstract

purpose. To develop and characterize a retinal explant culture system to facilitate investigation of novel methods of improving retinal stem cell therapy.

methods. Retinas explanted from adult rats were cultured in serum-free medium (B27/N2) or medium containing normal horse serum (NHS). Tissue viability was assessed by gross morphology, propidium iodide (PI) uptake, cell survival quantification, activated caspase-3 expression, and immunohistochemistry. Müller progenitor cells (hMIO-M1), or mesenchymal stem cells (MSC) were placed on explants, to model intravitreal cell transplantation. Explants were compared with whole eyes, with or without experimental glaucoma and/or intravitreal cell transplantation.

results. Explants cultured in B27/N2 medium were viable for at least 17 days, as assessed by the aforementioned parameters. NHS medium was associated with obvious tissue degradation, greater/more diffuse PI uptake, significant cell loss over time, and temporal increase in activated caspase-3⁺ cells. Explants in B27/N2 medium strongly expressed β-III-tubulin, neurofilament, NeuN, Brn3a, Thy-1, GFAP, vimentin, nestin, and glutamine synthetase, whereas immunoreactivity was weak in NHS medium and decreased further with time. Seven and 14 days after coculture or transplantation, glial reactivity (GFAP/vimentin expression) was highly upregulated in explants and eyes, respectively. Some grafted cells migrated into the retina, but most remained outside the inner limiting membrane.

conclusions. Retinal explants prepared using the described techniques and cultured in B27/N2 medium are viable for at least 2 weeks and mimic in vivo glial reactivity to transplantation while allowing few grafted cells to integrate. This system may be a useful in vitro model for investigating methods of enhancing retinal stem cell therapy.

Transplantation of stem and progenitor cells to treat neurodegenerative diseases of the central nervous system (CNS) is currently a subject of intense scientific investigation. In the retina, the published literature suggests that cellular grafts may be beneficial for treating sight-threatening conditions including macular degeneration, retinitis pigmentosa, and glaucoma. ^{1 2 3 4} The proposed mechanisms of therapeutic efficacy include neuroprotection of surviving tissue and/or retinal regeneration through cell replacement or enhanced neurite regrowth. ^{1 2 3 4 5 6}  

Many studies have demonstrated that cell transplantation can attenuate neurodegeneration and/or improve visual function in a variety of retinal disease and injury models, ^{7 8 9} but they also have revealed the need for novel techniques to enhance the outcome of intraocular cell injections. Extensive retinal graft integration is a prerequisite for stem cell therapy and is likely to be necessary to maximize therapeutic efficacy. Yet in reported studies to date, only a very small proportion of grafted cells have been shown to migrate into the retina, even in the context of retinal injury, which enhances retinal migration of transplanted cells. ^{10 11} Another limitation of current techniques is the inability to control the differentiation of grafted cells into mature phenotypes. Some studies have shown that grafted cells can express some of the canonical protein markers of retinal ganglion cells (RGCs) and even extend neuritic processes, ¹² but complete functional differentiation remains elusive. For retinal stem cell therapy to move forward toward clinical application, researchers must develop novel ways to improve control of the behavior of grafted cells and the recipient environment. 

The development of specific techniques to enhance the efficacy of cell transplantation could be facilitated by model systems that are quick, efficient, and easily manipulated. In vivo retinal disease models often take weeks to months for significant damage to develop, and it is often difficult to control the graft environment in vivo precisely. A further consequence of the variability of in vivo models is the need for large groups of animals in studies to demonstrate significant effects. These factors make in vivo experiments conducted to improve the efficacy of stem cell therapy time consuming, expensive, and difficult to interpret. There is therefore a need for more efficient in vitro models that facilitate experiments with a high degree of throughput and control that can be used to screen compounds and methods for improving retinal stem cell transplantation therapy. Encouraging results obtained using in vitro models could then be followed up and adapted for use in in vivo animal models. Such a combined approach would have the additional advantage of meeting the goal of reducing, replacing, and refining the number of in vivo procedures required. 

We developed and characterized an organotypic retinal explant culture system using tissue from adult rats specifically designed to model and investigate intraocular stem cell transplant for adult retinal therapy. We compared the viability of retinal explants cultured in different media formulations for periods of up to 17 days by assessing gross morphology, propidium iodide (PI) uptake, cell survival, apoptosis, and protein expression patterns and comparing them to intact eyes. We then used the optimized retinal explant system to model intravitreal transplantation of bone marrow–derived mesenchymal stem cells or Müller progenitor cells in vitro and compared this to transplantation in live animals. Although many previous studies using retinal explant culture for other purposes ^{13 14 15} have used a medium containing a high concentration of serum, we found that serum-free medium supplemented with B27 and N2 is superior in maintaining the laminar microarchitecture of the adult retina, forestalling necrosis and apoptosis, and preserving physiological expression of a wide array of canonical inner retinal protein markers. Our results demonstrate good viability of explants over time under optimal conditions and strong similarities between cultured tissue and living eyes, suggesting that the current model could be a valuable resource for future investigations in retinal stem cell therapy. 

Methods

Cell Culture

Human Müller cells from the MIO-M1 immortalized cell line (hMIOs) ^{16 17} and bone-marrow–derived mesenchymal stem cells (MSCs) isolated from transgenic GFP-expressing rats were kindly provided by the laboratories of Astrid Limb (Institute of Ophthalmology, University College London) and Siddharthan Chandran (Cambridge Centre for Brain Repair, Cambridge University), respectively. The cells were cultured in medium containing DMEM (glucose content = 4.5 g/L for hMIOs or 1 g/L for MSCs), 10% fetal bovine serum (FBS), penicillin (100 U/mL), and streptomycin (100 μg/mL). 

Animals

Adult (8–12 weeks of age) male Sprague-Dawley rats were obtained from Charles River, Ltd. (Kent, UK). The animals were handled according to the regulations formulated by the Home Office of the United Kingdom and in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Twenty-four rats underwent laser induction of unilateral ocular hypertension, of which eight rats also received MSC transplantation, eight rats received hMIO transplantation, and eight rats received PBS injection into the experimental eye. All 24 rats used for in vivo experiments received triple immunosuppressive therapy dissolved in the drinking water (cyclosporine: 150 mg/L; azathioprine: 15 mg/L; and prednisolone: 3.75 mg/L), to give approximate daily drug dosages of cyclosporine (20 mg/kg), azathioprine (2 mg/kg), and prednisolone (0.5 mg/kg). Seven rats were used to obtain retinal explants for culture. 

Laser-Induced Ocular Hypertension

Ocular hypertension was induced by the method developed by Levkovitch-Verbin et al. ¹⁸ Briefly, anesthetized rats were placed in front of a standard ophthalmic slit lamp equipped with a 532-nm diode laser that was calibrated to deliver 0.7-W pulses lasting 0.6 seconds. Fifty to 60 laser pulses were directed to the trabecular meshwork 360° around the circumference of the left cornea only. The animals were treated twice with laser sessions spaced 1 week apart. An increase in intraocular pressure to at least 35 mm Hg in each animal 24 hours after both laser administrations was confirmed with a rebound tonometer (TonoLab; Tiolat Oy, Helsinki, Finland). Contralateral fellow eyes served as the untreated control. 

Cell Transplantation

MSCs or hMIOs were suspended in sterile PBS at a concentration of 50,000 cells/μL. One week after the second of two laser treatments to induce ocular hypertension, the left eye of anesthetized rats was treated with topical anesthetic (tetracaine 1%). Intravitreal injections were administered with a 34-gauge needle on a 5-μL syringe (Hamilton, Reno, NV) under direct visualization of the posterior segment by operating microscope and coverslip coupled to the cornea with carbomer gel. When the needle was visualized to be just through the retina and into the vitreous, a 2-μL intravitreal injection was delivered. Care was taken to ensure that the lens was not damaged. Whole eyes were fixed 7 or 14 days after transplantation. Cell transplantation was performed only in eyes with ocular hypertension as healthy eyes are less receptive to cell grafts. ¹⁰  

Retinal Explant Culture

Rats with healthy, untreated eyes were killed by exposure to a rising concentration of carbon dioxide followed by cervical dislocation. Immediately, both eyes were enucleated (Supplementary Fig. S1a) and immersed in ice-cold HBSS containing penicillin (100 U/mL) and streptomycin (100 μg/mL). With an operating microscope, a circumferential incision was made around the limbus (Supplementary Fig. S1b), followed by removal of the anterior segment, lens, and vitreous body (Supplementary Figs. S1c, S1d). With an extremely fine paintbrush, the retina was carefully peeled away from the retinal pigment epithelium (Supplementary Fig. S1e), and a single cut was made at the optic nerve head, thereby severing the retina from the optic nerve (Supplementary Fig. S1f). The retina was dissected radially into four equal-sized pieces (Supplementary Fig. S1g), and the explants were separately transferred onto 12-mm diameter filters (0.4 μm pore, Millipore; Millicell Inc., Cork, Ireland; Supplementary Fig. S1h) with the RGC side facing up. The filters were placed into the wells of a 24-well plate, each of which contained 300 μL of retinal explant media. Based on preliminary experiments (data not shown), two different types of culture media were selected for further analysis. Medium containing normal horse serum (NHS) was produced based on previous reports ^{19 20 21 22} and consisted of a neuronal growth medium (Neurobasal-A) supplemented with 25% heat-inactivated NHS (Invitrogen Ltd., Paisley, UK), l-Glutamine (0.5 mM), penicillin (100 U/mL), and streptomycin (100 μg/mL). A serum-free medium (B27/N2) was modified from previous studies ^{13 14 15} and contained the neuronal growth medium (Neurobasal A) supplemented with 2% B27 (Invitrogen Ltd.), 1% N2 (Invitrogen Ltd.), l-glutamine (0.8 mM), penicillin (100 U/mL), and streptomycin (100 μg/mL). Retinal explant cultures were maintained in humidified incubators at 34°C and 5% CO₂. Half of the media were changed on day ex vivo (DEV) 1 and every second day thereafter. 

For co-culture experiments, hMIOs or MSCs were washed in HBSS before being suspended at a concentration of 500 to 1000 cells/μL in B27/N2 culture medium. With a standard P-10 pipette, 2-μL droplets of the cell suspensions were transferred directly onto the RGC side of explants at DEV 1. A lower cell density was grafted onto explants than was transplanted in vivo based on previous experiments that demonstrated extensive graft proliferation on explants only when transplanted at densities greater than 1 × 10⁴ cells/explant (data not shown). Explants were fixed 7 or 14 days after adding the cells to the explants. 

Tissue Processing and Histology

Retinal explants were fixed by replacing the culture media with 4% paraformaldehyde (PFA) and adding approximately 10 μL of 4% PFA onto the surface of the explant for 24 hours at 4°C. Whole eyes were enucleated from rats after transcardial perfusion with 4% PFA and were immersion-fixed in 4% PFA for 24 hours at 4°C. The explants and whole eyes were then cryoprotected in 30% sucrose for 24 hours at 4°C before being embedded in OCT, frozen, and stored at −20°C. The tissue was cryosectioned at 14 μm and transferred directly onto microscope slides (Superfrost; Fisher Scientific, Pittsburgh, PA). The eyes were sectioned sagittally. 

For immunohistochemistry, the slides were washed in 0.1 M phosphate-buffered saline (PBS), incubated for 90 minutes at room temperature in blocking solution consisting of 5% normal goat serum (NGS) and 0.2% Triton in 0.1 M PBS, incubated overnight at 4°C in primary antibody (Table 1)diluted in blocking solution, incubated at room temperature for 3 hours in secondary antibody (Table 1)diluted in blocking solution, and counterstained with 4′,6-diamidino-2-phenyindole (DAPI, Table 1 ) before being coverslipped with immunofluorescence mounting medium. Slides were thoroughly washed in PBS three times between each incubation step. The tissue was imaged with a standard epifluorescence microscope (Leica FW4000; Leica, Wetzlar, Germany) or a confocal microscope (TCS-SPE; Leica). 

Data Analysis

To determine the viability of cultured retinal tissue over time, wholemount micrographs were obtained for each of eight explants (four for both culture media groups) with an operating microscope at DEV 0, 1, 3, 7, 10, 14, and 17, and assessed for gross morphologic changes. The area of viable tissue was quantified in each micrograph by an investigator masked to the identity of the file using ImageJ software (developed by Wayne Rasband, National Institutes of Healthy, Bethesda, MD; http:// rsb.info.nih.gov/ij/). Viable tissue area for each explant at DEV 0 was set as 100% and the percentage remaining was calculated for each explant over time. Viability of explant tissue was confirmed by adding 1 μg/mL PI to the media for 1 hour before visualization under fluorescence microscopy. 

For each of the remaining experiments, four explants for each media type at each time point (DEV 3, 7, 10, 14, and 17) were used. First, explants stained only with DAPI were analyzed with UV epifluorescence. An investigator masked to the identity of each explant measured the length of the RGC layer within several micrographs for each explant and also counted the number of DAPI-positive nuclei in the RGC layer. The average number of nuclei per micrometer was calculated in each explant. The same investigator also counted the thickness in cells) of the inner (INL) and outer nuclear layers (ONL) under direct microscopic visualization. Five inner and outer nuclear layer cell thickness counts, each from a separate section, were obtained for each explant and averaged. 

Sections were stained for activated caspase-3, as previously described. Under direct 200× magnification with epifluorescence, a masked investigator quantified immunoreactive cells within each of the three layers per field. Five fields per explant, each from a separate section, were assessed and averaged. 

For each of the above parameters, changes over time within groups as well as differences between groups at each time point were compared using ANOVA with post-hoc Bonferroni corrections for multiple comparisons. All data are expressed as the mean ± SE. 

Results

Explant Morphology and Viability

There was no difference between the appearances of retinal explants in the two culture media groups immediately afterexplantation (Fig. 1a , Supplementary Fig. S2a). Explants in both types of media exhibited a slight increase in opacity and change in color between DEV 0 and 1 and continuing until DEV3, after which the general appearance of the tissue did not change (Supplementary Fig. S2a). Explants cultured in NHS medium displayed significant (ANOVA, P < 0.001) tissue degeneration around the entire periphery that began at DEV 10 and progressed through DEV 17, at which point only a small island of tissue, representing 26% ± 4% of the total original explant area, remained in the center of the explant (Figs. 1a 1b , Supplementary Fig. S2a). Explants cultured in serum-free medium demonstrated negligible tissue loss over the 17-day period. 

To further characterize the viability of the central tissue, retinal explants were incubated with PI (1 μg/mL) for 1 hour before visualization under epifluorescence. Significant PI uptake was confined to the far periphery of the explants. Central tissue excluded the dye at early time points, suggesting that the majority of the explanted tissue was viable at least until DEV 7 for both types of media (Fig. 1c , Supplementary Fig. S2c–e). Explants cultured in B27/N2 media continued to demonstrate PI uptake in the periphery, but only began to display focal areas of PI uptake in the central tissue beginning at DEV 14 (Supplementary Fig. S2f). By contrast, explants in NHS medium showed a more diffuse area of PI uptake that was evident as early as DEV 10 (Supplementary Fig. S2g). 

Preservation of Retinal Architecture and Cell Survival

To assess the preservation of the retinal microarchitecture over time, we stained explant sections with the nuclear marker DAPI. Overall, the laminar structure of the retina was well-preserved in the retinal explants cultured in B27/N2 medium but was gradually lost in NHS medium (Fig. 2a) . DAPI staining also facilitated the counting of nuclei within each retinal layer. Healthy eyes fixed whole immediately after enucleation served as the control and demonstrated a linear RGC density of 114 ± 4 nuclei/mm, INL thickness of 5.3 ± 0.3 cells, and ONL thickness of 13.2 ± 0.8 cells. By DEV 3, neither media group demonstrated a significant decrease in any of these parameters compared with the healthy eyes (Figs. 2b 2c 2d) . Over time, however, explants cultured in NHS medium showed significant loss of cells in each of the three nuclear layers. The effect was most dramatic in the nuclei in the RGC layer, which were significantly reduced by DEV 7 in NHS compared with B27/N2 (P < 0.05) and were almost completely lost by DEV 14 (Fig. 2b) . Nuclear counts for explants in B27/N2 media were relatively steady in comparison, with no significant nuclear loss over time when compared to DEV 3. These differences between media groups were statistically significant (Fig. 2b 2c 2d) . 

The level of apoptosis within retinal explant cells was detected by immunolabeling the sections for activated caspase-3, a key enzyme at the junction of the intrinsic and extrinsic apoptosis pathways and active in cells destined for apoptotic death (Supplementary Fig. S3). In healthy eyes, 1.6 ± 0.2 cells/mm in the RGC layer were caspase-3 positive, whereas eyes that were assessed 2 weeks after optic nerve crush showed 5.5 ± 0.2 cells/mm. The level of staining in the RGC layer of explants was initially similar between media groups, higher than in healthy eyes and lower than in eyes that underwent optic nerve crush (Fig. 3a) . Over time, caspase staining in the RGC layer of serum-free explants decreased, probably as a function of cell loss as the percentage of caspase-positive cells was relatively constant (Fig. 3b) . The number of caspase-3 positive cells in NHS explants was elevated compared with the B27/N2 explants and did not change over time (Fig. 3a) , resulting in a gradually increasing percentage of cells in the RGC layer undergoing apoptosis (Fig. 3b) . Similar differences between media groups were found for caspase staining of nuclei within the INL and ONL (Figs. 3c 3d) . 

Expression of Neuronal Markers

As intravitreal transplantation of cells would presumably be most applicable to diseases of the inner retina, we sought to look more closely at the protein expression patterns of inner retinal neurons in explants over time (Fig. 4) . Antibodies for β-III-tubulin (Figs. 4a 4b 4c 4d)and Thy-1.1 (Figs. 4i 4j 4k 4l)bound to RGC cell bodies and neurites while neurofilament-160 kDa (Figs. 4e 4f 4g 4h)was more specific for RGC axons. Immunoreactivity for each of these markers demonstrated the presence of RGC axon bundles in the retinal nerve fiber layer, though expression level of these markers appeared to be lower in explants than in whole eyes. Axon bundles were diminished in size but still present in explants at DEV 7 in both types of media and, to a lesser extent, in B27/N2 media at DEV 14. NHS medium was associated with the virtual loss of the inner retina by DEV 14 (Fig. 2a)and the absence of immunoreactivity for any of these markers (not shown). Brn-3a (Figs. 4q 4r 4s 4t)is a transcription factor specific to RGCs and the NeuN antibody (Figs. 4m 4n 4o 4p)has been shown to bind to RGC nuclei and sometimes amacrine nuclei. Control eyes demonstrated strong immunoreactivity for these markers localized to the RGC layer. Explants in both types of media showed downregulated but well-defined expression at DEV 7 with similar localization as in whole eyes; expression was maintained through DEV 14 only in B27/N2 medium. Finally, immunoreactivity for calbindin (Figs. 4u 4v 4w 4x)demonstrated the presence of amacrine and horizontal cells within control eyes. Staining patterns were similar for retinal explants, with displaced amacrine cells sometimes being observed in the RGC layer. Immunoreactivity for most markers tested, as well as the integrity of the nuclear layers in NHS explants at DEV 7, more closely resembled explants in B27/N2 at DEV 14 rather than at a similar time point (Fig. 4) . Retinal explants in B27/N2 at DEV7 strongly resembled healthy control eyes, with some reduction in immunoreactivity for various inner retinal antigens. 

Expression of Glial Markers

A potentially important barrier to transplanted cell migration into the neural retina from the vitreous is the inner limiting membrane (ILM), composed of Müller glia end feet. As such, we wanted to determine how this structure is maintained in explants over time (Fig. 5) . As protein expression patterns change within Müller glia and astrocytes in response to injury, we compared explants to both healthy control eyes and eyes with laser-induced ocular hypertension. Vimentin was expressed in control eyes mainly at the ILM with thin processes extending as far as the inner plexiform layer, with a similar pattern found in B27/N2 explants through DEV 7 (Figs. 5a 5b 5c 5d) . The marker appeared to be upregulated in glaucomatous eyes and by DEV 14, a modest upregulation of vimentin within the outer retina of explants also was evident. GFAP expression was localized to the ILM in control eyes and B27/N2 explants at all time points, although immunoreactivity was upregulated in Müller cell bodies of glaucomatous eyes (Figs. 5e 5f 5g 5h) . Glutamine synthetase was observed in the inner retina of whole eyes, with or without glaucoma, and in B27/N2 explants; the marker did not change in intensity with length of culture but ocular hypertension (OHT) was associated with an increase in reactivity in the outer retina (Figs. 5i 5j 5k 5l) . The expression of nestin was nearly undetectable in control eyes, but was moderately upregulated in glaucoma specifically near the ILM, and was highly upregulated in B27/N2 explants at both time points (Figs. 5m 5n 5o 5p) . As before, the expression patterns in NHS explants at DEV 7 were very similar to those of B27/N2 explants at DEV 14, whereas expression in NHS explants at DEV 14 was delocalized and diminished (data not shown). These results demonstrate that the retinal explants exhibited a modest level of reactive gliosis in response to explantation and culture, but the specific response was different from that seen with OHT in vivo. OHT was associated with an increase in GFAP and glutamine synthase reactivity, whereas the predominant change in glial expression in explants was a strong upregulation of nestin. 

Cellular Transplantation and Co-culture

Having validated the retinal explant culture system in B27/N2 medium as an accurate representation of the retinal environment, we used our system as a model for intravitreal stem cell transplantation by suspending hMIOs or MSCs in 2 μL medium and co-culturing them on the RGC surface of retinal explants for 7 or 14 days. We compared this to transplantation of the same cell types into eyes of living animals with laser-induced ocular hypertension. The most striking effect of transplanting or co-culturing cells that we observed was a significant reactive gliosis that occurred both in vivo when compared with an injection of PBS alone and in our model system in vitro when compared with maintaining retinal explants in the absence of co-culture grafts. This effect was similar in both MSC and hMIO transplantation and co-cultures. Transplantation and co-culture of cells was associated with a strong upregulation of GFAP (Figs. 6a 6b 6c 6d)and, more modestly, vimentin (Figs. 6e 6f 6g 6h)throughout the tissue and even in areas of the retina far from the injection site where no grafted cells were observed (Figs. 6a 6b 6c 6d 6e 6f 6g 6h) , although glial reactivity also was strong near the grafted cells (Figs. 6i 6j 6k 6l 6m 6n 6o 6p 6q 6r) . 

The behavior of grafted cells was similar in vivo and in vitro. MSCs formed either multilayered sheets (Fig. 6j)or small spheres (Fig. 6k)on the surface of the retinal explants by DEV 7. By DEV 14, proliferation of the cells had resulted in large boluses residing on the retinal surface, with minimal cellular infiltration into the explant tissue (Fig. 6l) . We also observed the formation of multilayered cell sheets lying on the surface of the retina in vivo (Fig. 6i) , but no spheres were evident. Both in vivo and in vitro, more than 99% of the cells remained outside the ILM, excluded from integrating into the retinal tissue. MSCs did not express any neural markers, indicating that no neural differentiation had taken place. 

Unlike MSCs, hMIOs were most often found as single cells on the surface of the retinal explants (Figs. 6m 6n 6o 6r) . hMIOs often extended very long processes which were usually oriented parallel to the retina, and not directed into the tissue (Figs. 6m 6n 6o) . In some cases, hMIOs were seen to have migrated through the ILM and associate with the RGC layer after only 7 days. (Figs. 6p 6q) . More rarely, hMIOs migrated into the deeper nuclear layers, and some of these cells even extended processes oriented toward the plexiform layers of the retina (Figs. 6p 6q) , although it should be noted that this phenomenon occurred more readily in areas where at least mild disruption to the retinal architecture was also observed. A subset of the transplanted hMIOs expressed vimentin, GFAP, and/or β-III-tubulin (Figs 6m 6n 6o 6p 6q 6r) . The behavior of hMIOs that were co-cultured with retinal explants for 7 (Figs. 6m 6n 6o 6p 6q)and 14 (Fig. 6r)days did not appear to be different. Many of these observations were identical with those seen after transplantation of hMIOs into living eyes with OHT, demonstrating important correlations between transplantation in vivo and in our in vitro system. 

Discussion

Retinal explant culture has an extensive history in neuroscience. ^{23 24} Investigators have used the technique to study retinal development, ^{13 23 25 26} CNS regeneration, ^{15 27 28 29} cell death and neuroprotection, ^{14 22 30} electrophysiological recording, ³¹ and genetic modification. ^{20 32} Only recently, have investigators begun to use the system as a model for cell transplantation. ^{33 34 35 36 37 38 39} Most such studies have involved tissue from embryonic or neonatal animals or from young animals with inherited retinal degeneration, all of which demonstrate a relatively high level of receptiveness to cell transplantation compared with adult tissue. ^{7 40} Despite some interesting and impressive results in adult retinal explants, ^{39 41} the viability and behavior of retinal tissue culture systems from adult rodents has not been rigorously evaluated. Protocols differ widely between groups, most notably with regard to the age of the animals from which the tissue is extracted, the composition of culture medium, the length of culture, and the laminar orientation of the tissue with regard to the substrate. Without demonstrating that the integrity of the retina and the ability to withstand infiltration from foreign cells remain intact in culture, experiments involving retinal stem cell therapy using cultured adult explants should be interpreted with caution. 

We have developed and characterized a technique to culture retinal explants that demonstrates a high degree of tissue viability for at least 17 days ex vivo. We found that the serum-free medium preparation was superior to the NHS-containing medium in maintaining the viability of retinal explants with regard to every parameter that we evaluated. This finding is important, as serum-free culture protocols are generally preferred for investigations involving stem cell therapy, because their ingredients are clearly defined and thus any components that may affect grafted cell behavior can be manipulated. Given that the procedure for explanting retinal tissue involves a complete axotomy of all RGCs, we were interested in observing the presence of axon bundles within the nerve fiber layer of explants cultured in B27/N2 media as late as DEV 14. 

The behavior of glial cells within the explants served as a functional demonstration of the viability of the culture, and it was reassuring that the cultured retina mimicked in vivo retinal responses in several respects. Retinal injury is known to trigger a modest glial reactivity, as evidenced by upregulation of GFAP and nestin. Although GFAP levels remained low in untreated explants, nestin was dramatically upregulated, indicating that, as expected, some level of stress had been registered by the tissue. Even more important, the glial reactivity in response to cell transplantation was similarly robust in vivo and in vitro. Reactive gliosis has been shown to inhibit survival, migration, neurite extension, and synapse formation by cells transplanted into the CNS. ⁴² Indeed, GFAP/vimentin double-knockout mice exhibit reduced glial reactivity to CNS injury and transplantation and are much more receptive to retinal stem cell transplantation compared with wild-type control animals. ⁴³ Because the very act of transplanting cells into the CNS appears to reinforce a significant hindrance to the success of the graft, glial reactivity represents an important obstacle to overcome. Retinal explants with preserved glial reactivity to cell transplantation may be useful in the exploration of strategies to improve graft integration. 

In the current model, cells co-cultured on top of the explant RGC layer were in contact with tissue only on one side, as opposed to cells that could be placed between the filter and the explant, resembling a subretinal graft. Co-culturing MSCs and hMIOs with retinal explants demonstrated that most of the cells remain outside the retinal nerve fiber layer. Thus, like intact eyes of living animals, retinal explants are able to withstand uncontrolled infiltration of foreign cells. However, we were able to identify a small number of cases in which co-cultured cells were found to have migrated into the neural retina. We did not quantify the exact proportion of cells that had migrated into the retina, because this phenomenon did not take place in every retinal explant assessed, was rare when it did occur, and was found almost exclusively in instances of mild to moderate disruption to the retinal structure. Further investigation with this model may demonstrate enhanced and more consistent retinal integration by other cell types or under various experimental conditions. 

The in vitro model of organotypic retinal explant culture appears to be an advantageous system in which to screen the level of integration and/or differentiation of transplanted cells. The model may be particularly relevant to glaucoma, as the process of explantation requires complete axotomy of all RGCs, therefore producing severe axonal damage. Eight retinal explants can be created from the two eyes of a single animal, thereby reducing the number of animals needed for each experiment. Further, the entire procedure of extracting and explanting eight retinal explants requires, at most, 1 hour. A minimal effort of exchanging half of the medium for fresh medium every second day is necessary to maintain the cultures. In this system, conditions can be controlled much more precisely than in in vivo glaucoma models. In addition, explant cultures are devoid of immune activity, and so the aggressive graft rejection often observed in vivo is irrelevant in vitro. Although positive results from in vitro experiments must be followed up in live animals, our retinal explant culture model may be useful for screening a large number of compounds for beneficial effects on cellular migration and differentiation, assaying the optimum dosages to achieve the desired effect. This approach could also facilitate reduction and refinement of the number of in vivo procedures required. 

 Supported by a Gates-Cambridge Scholarship (TVJ), a Fight for Sight Summer Student’s Fellowship (TVJ), and a National Institutes of Health Graduate Partnerships Program Studentship (TVJ); a GlaxoSmithKline Clinician-Scientist Award (KRM); a Glaucoma Research Foundation Grant (KRM); and the Richard Norden Glaucoma Research Fund (KRM).

 Submitted for publication December 14, 2007; revised February 15 and March 6, 2008; accepted June 2, 2008.

 Disclosure: T.V. Johnson, None; K.R. Martin, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Details of Antibodies and Fluorescent Markers Used for Immunohistochemistry

Primary Antibodies
Specificity	Cells Labeled	Source (Isotype)	Clone	Company	Working Concentration
Active caspase-3	Apoptotic cells	Rabbit	Polyclonal	Chemicon Inc., Temecula, CA	2.0 μg/mL
β-III-Tubulin	RGCs	Mouse IgG1	5G8	Promega, Inc., Madison, WI	0.5 μg/mL
Neurofilament (NF)-160kD	RGC axons	Mouse IgG1	NN 18	Chemicon Inc.	5.0 μg/mL
Thy-1.1	RGCs	Mouse IgG1	OX 7	Chemicon Inc.	5.0 μg/mL
Neuronal nuclei (NeuN)	RGCs, amacrine cells	Mouse IgG1	NeuN	Chemicon Inc.	5.0 μg/mL
Brn-3a	RGCs	Rabbit	Polyclonal	Chemicon Inc.	5.0 μg/mL
Calbindin	Amacrine, horizontal cells	Rabbit IgG	Polyclonal	Merck Chem Ltd., Nottingham, UK	Initial conc. N/A Dilution 1:5000
Vimentin	Astrocytes, Müller glia	Chicken IgY	Polyclonal	Chemicon Inc.	0.5 μg/mL
Glial fibrillary acid protein (GFAP)	Astrocytes, Müller glia	Rabbit	Polyclonal	Dako UK Ltd., Ely, UK	Initial conc. N/A
										Dilution 1:500
Glutamine synthetase	Astrocytes, Müller glia	Rabbit IgG	Polyclonal	Sigma-Aldrich Inc., St. Louis, MO	0.27 μg/mL
Nestin	Astrocytes, Müller glia	Mouse IgG1	Rat-401	Chemicon Inc.	2.75 μg/mL

Secondary Antibodies and Other Labels
Specificity	Conjugate	Source	Company	Working Concentration
Mouse IgG (H+L)	AlexaFluor 488	Goat	Invitrogen-Molecular Probes Inc., Eugene, OR	2.0 μg/mL
Mouse IgG (H+L)	AlexaFluor 568	Goat	Invitrogen-Molecular Probes Inc.	2.0 μg/mL
Rabbit IgG (H+L)	AlexaFluor 488	Goat	Invitrogen-Molecular Probes Inc.	2.0 μg/mL
Rabbit IgG (H+L)	AlexaFluor 568	Goat	Invitrogen-Molecular Probes Inc.	2.0 μg/mL
Chicken IgY	AlexaFluor 488	Goat	Invitrogen-Molecular Probes Inc.	2.0 μg/mL
PI			BD Biosciences Inc., Oxford, UK	1.0 μg/mL
DAPI			Invitrogen-Molecular Probes Inc.	0.5 μg/mL

Figure 1.

 

Gross morphology and tissue viability of cultured retinal explants. Retinal explants cultured in B27/N2 or NHS medium were maintained for up to 17 DEV. Wholemount photographs (a) and PI uptake (c) demonstrated that tissue degeneration occurred at the periphery of the explants (↑), whereas the center (#) was spared. Longitudinal quantification of healthy tissue area in explants in B27/N2 medium demonstrated negligible tissue loss through DEV 17, whereas tissue in NHS medium progressively degenerated (b, *P < 0.001 compared with DEV 0 by ANOVA with post hoc Bonferroni comparisons). Scale bar: (a) 6 mm; (c) 400 μm.

Figure 2.

 

Laminar microarchitecture and cell survival in retinal explants. Retinal explants cultured in B27/N2 medium or in NHS medium were maintained for up to 17 DEV, whereas healthy control eyes (in vivo) were used for comparison. Sectioned explants were stained with DAPI to visualize nuclei. The 3-layered retinal architecture was present at early time points in both media and maintained in B27/N2, whereas it was progressively lost in NHS (a). B27/N2 explants demonstrated good survival of cells in the ganglion cell layer (GCL, b), inner nuclear layer (INL, c), and outer nuclear layer (ONL, d), whereas cells were progressively lost from each of these structures in NHS. *P < 0.05 comparing NHS to B27/N2 at the indicated time points by ANOVA with post hoc Bonferroni comparisons. Scale bar, 100 μm.

Figure 3.

 

Apoptosis in retinal explants. Retinal explants cultured in medium containing B27 and N2 supplements (B27/N2) or NHS were maintained for up to 17 DEV, whereas healthy control eyes or eyes that underwent optic nerve crush 2 weeks earlier were used for comparison. Sectioned tissue underwent immunohistochemistry for activated caspase-3 and was counterstained with DAPI to visualize the nuclei. Caspase-3⁺ nuclei were quantified in the retinal GCL (a, b), INL (c), and ONL (d) and quantified per millimeter measured parallel to the retina or as a percentage of cells in the layer. *P < 0.05 comparing NHS with B27/N2 at the indicated time points by ANOVA with post hoc Bonferroni comparisons.

Figure 4.

 

Immunohistochemistry for inner retinal neuronal markers. Retinal explants cultured in media containing B27 and N2 supplements (B27/N2) or NHS were maintained for up to 17 DEV, and healthy control eyes were used for comparison. Sectioned tissue underwent immunohistochemistry for the indicated markers (green or red) and was counterstained with DAPI (blue). (a–d) β-III-Tubulin; (e–h) NF-160; (i–l) Thy 1.1; (m-p) NeuN; (q–t) Brn-3a; calbindin (u–x). Scale bar, 100 μm.

Figure 5.

 

Immunohistochemistry for glial markers. Retinal explants cultured in media containing B27 and N2 supplements (B27/N2) were maintained for up to 17 DEV while healthy control eyes or eyes that had undergone OHT induction by two laser treatments at 1 and 2 weeks earlier were used for comparison. Sectioned tissue underwent immunohistochemistry for the indicated markers (green or red) and was counterstained with DAPI (blue). (a–d) Vimentin; (e–h) GFAP; (i–l) glutamine synthetase; (m–p) nestin. Scale bar, 100 μm.

Figure 6.

 

Transplantation/co-culture of stem/progenitor cells. Retinal explants cultured in B27/N2 medium or whole eyes of living rats with laser-induced OHT received transplants of either GFP-labeled MSCs or GFP-labeled hMIOs and were fixed 7 or 14 days later. GFAP (a–d, red) and vimentin (e–h, green) immunoreactivities were increased by cell transplantation, even in areas where no grafted cells could be seen. Although the pictured sections (b, d, f, h) involved MSC transplantation/co-culture, the effect was similar for hMIO transplantation/co-culture. MSCs (i–l) remained on the surface of the retina in vivo and in vitro without migrating into the retinal tissue; cell proliferation appeared to occur over time on explants. hMIOs (m–r) often extended processes, expressed a variety of neural proteins, and sometimes migrated into the neural tissue. (q) Higher magnification of (p); arrow indicates a neuritic process oriented toward the inner plexiform layer. Scale bar: (a) 100 μm. All images except (l) and (q) are at the same magnification; (l) 50 μm; (q) 100 μm. B3T, β-III-tubulin.

Supplementary Materials

Supplementary Figure S1 - (PDF) 

Supplementary Figure S2 - (PDF) 

Supplementary Figure S3 - (PDF)