OUP user menu

Characterization and Survival of Long-Term Implants of Human Retinal Pigment Epithelial Cells Attached to Gelatin Microcarriers in a Model of Parkinson Disease

Joseph Flores MSc, Ivan L. Cepeda MSc, Michael L. Cornfeldt PhD, John R. O'Kusky PhD, Doris J. Doudet PhD
DOI: http://dx.doi.org/10.1097/nen.0b013e318093e53a 585-596 First published online: 1 July 2007


Previous studies have demonstrated that the intrastriatal implantation of human retinal pigment epithelial cells attached to gelatin microcarriers (hRPE-GM) ameliorates behavioral deficits in animal models of Parkinson disease. However, there are only sparse data on cell survival in the host. In this study, we characterized a variety of retinal pigment epithelial (RPE)-specific markers in vitro and used these markers to investigate the long-term survival of hRPE-GM implants. Sprague-Dawley rats (n = 22) were unilaterally lesioned with 6-hydroxydopamine (6-OHDA) and implanted with hRPE-GM without immunosuppression. Rats were euthanized at 48 hours, 7 days, 4 weeks, and 5 months postimplant and immunohistochemically processed using the following antibodies: 1) human-specific nuclear mitotic apparatus protein (NuMA-Ab2), 2) epithelial-specific extracellular matrix metalloproteinase inducer (EMMPRIN), 3) RPE cell-specific RPE65, and the inflammation markers 4) glial fibrillary acidic protein and 5) ED1 (rat CD68). Our analysis revealed NuMA-, EMMPRIN-, and RPE65-immunoreactive cells at different times postimplant. The morphologic features of hRPE cell implants (at 48 hours and 5 months) were confirmed by electron microscopy. Furthermore, despite evidence of chronic inflammation at the later time point, there is an appreciable number of surviving hRPE cells. This study suggests that hRPE-GM implants can survive in the absence of immunosuppression and can be potentially used as an alternative for treating Parkinson disease.

Key Words
  • Immunohistochemistry
  • Parkinson disease
  • Retinal pigment epithelial cells
  • RPE65
  • Transplantation


Parkinson disease (PD) is a progressive, debilitating neurologic disorder that affects millions of people worldwide. The progressive loss of striatal dopamine (DA) is attributed to the degeneration of DA neurons in the substantia nigra pars compacta (SNc). As the loss of DA progresses, patients with PD clinically exhibit characteristic symptoms including tremor, muscular rigidity, bradykinesia, hypokinesia, postural instability, and gait disturbances. Current pharmacologic therapy for PD focuses primarily on reversing striatal DA deficiencies. Levodopa (L-dopa) in combination with dopa decarboxylase inhibitors and/or DA agonists is the current treatment of choice for early-stage PD. Unfortunately, long-term L-dopa therapy leads to the development of motor complications in late-stage PD and within a few years, 50% to 80% of patients develop motor fluctuations, "wearing off," and dyskinesia. The development of motor complications is believed to be due to the intermittent stimulation of DA receptors; indeed, the maintenance of stable L-dopa levels with intravenous infusion or slow-release preparations reduces the emergence of motor complications (1). This fact suggests that a more continuous striatal L-dopa or DA delivery may have a therapeutic advantage in managing symptoms of PD.

Cell-based transplantation therapies such as human fetal cell, embryonic stem cell, or other DA-producing cell grafts have been or are being studied as alternative treatment options for PD (2-4). Fetal mesencephalic tissue transplantation, the "gold-standard" for neural transplantation in PD, has been performed in many patients over the past several years (5). This treatment can be effective in improving PD motor symptoms and restoring L-dopa availability in the striatum. However, the limited availability of fetal tissue, significant ethical considerations, and only modest improvement seen in 2 recent placebo-controlled trials (6, 7) has impeded the availability of this therapy.

To overcome some of these issues, embryonic stem cells have been recently proposed as an alternative DA source for restorative treatment of PD (8, 9). Embryonic stem cells can be maintained in culture in an undifferentiated state and have the potential to differentiate into DA neurons for striatal transplantation. A recent study has shown that embryonic stem cells, when transplanted into the striatum of parkinsonian rats, are capable of differentiating into DA neurons (10). However, embryonic stem cell therapy is still early in development, and many issues regarding long-term survival, degree of reinnervation, and potential consequences of its unregulated growth still need to be addressed.

Human retinal pigment epithelial (hRPE) cells are currently being investigated as a potential alternative source for cell therapy. hRPE cells are retinal "support" cells, forming a monolayer between the choroid capillary bed and the photoreceptor layer (11). hRPE cells are pigmented epithelial cells that produce L-dopa as an intermediate step during the synthesis of their characteristic eumelanin pigment through the activity of a tyrosine hydroxylase (TH) analog, tyrosinase. Interestingly, although L-dopa is reoxidized quickly in adult functioning, embryonic hRPE cells contain more stable concentrations of L-dopa (12).

hRPE cells can be expanded in standard cell culture conditions, allowing for extensive testing for potential contaminants. Being anchorage-dependent, they undergo apoptosis in the absence of a support matrix. However, passive attachment to biocompatible gelatin microcarriers (GMs) improves survival and viability in vitro and in vivo (13). In addition, the ability to expand and maintain hRPE cells in culture for extended periods and to perform GM attachments in vitro allows for the development and optimization of specific methods for hRPE cell identification.

hRPE-implanted parkinsonian patients continue to show behavioral improvement 4 years postimplant (14). Early studies in hemiparkinsonian nonhuman primates have also shown significant improvements (50%-60%) in the unified Parkinson disease rating scale scores throughout a 12-month observation period (15). In our laboratory, behavioral improvements in bilaterally parkinsonian nonhuman primates are still observed 4 to 5 years after unilateral hRPE implants. A positron emission tomography study in these animals demonstrated both a significant increase in [18F]dopa accumulation and a concomitant decrease in [11C]raclopride binding (a surrogate marker of synaptic DA release) at the implant site 2 months postimplant (16). To date, there is still a continued increase in [18F]dopa uptake 5 years after implant in a subset of these animals (D.J. Doudet, unpublished observations, 2006). Apart from sustained clinical improvement seen in primates, there has only been 1 histologic report suggesting continued survival of the hRPE cells after implant. Using the unilateral 6-hydroxydopamine (6-OHDA) rodent model of PD, Subramanian et al (17) demonstrated significant apomorphine-induced behavioral improvements in hemiparkinsonian rats after hRPE cell implants. A postmortem analysis of these rats showed the presence of hRPE-like cells using TH immunohistochemistry (IHC). However, TH is not a specific hRPE marker, and more specific evidence of long-term survival of hRPE cells is needed.

The goal of this study was to characterize and validate a variety of markers to specifically identify hRPE cells in vitro and to use these markers to qualitatively examine the long-term survival of hRPE cells in vivo. For our investigatation we implanted nonimmunosuppressed, 6-OHDA-lesioned rats with hRPE cells attached to gelatin microcarriers (hRPE-GM) and immunohistochemically examined the survival of hRPE-GM implants at various time points. The relative advantages of each marker for use in future quantitative assessments of hRPE-GM implant survival are discussed.

Materials and Methods

Sprague-Dawley Rats (University of British Columbia Animal Facility) weighing 275 to 300 g were used for these experiments. All animals were housed either in pairs or in groups of 3 in Plexiglas cages with ad libitum access to food and water. A 12-hour light/dark cycle was maintained (for which the animals' dark [active] cycle was 12:00 noon to 12:00 midnight). Animal housing, animal care, and all experimental procedures were approved by the University of British Columbia Animal Care Committee.

6-Hydroxydopamine Lesions

The surgical procedure for the 6-OHDA lesions is described in detail in the companion article (18). Briefly, rats were anesthetized with isoflurane and received a unilateral stereotaxic 6-OHDA (Sigma Aldrich, St. Louis, MO) injection (10 μg/4 μL in 0.05% ascorbic acid in saline infused at 1 μl/min) into the SNc and medial forebrain bundle, at the following coordinates: anteroposterior (AP) −2.8, mediolateral (ML) −1.8, and dorsoventral (DV) −8.0; and AP −4.7, ML −1.5, and DV −7.9.

Human Retinal Pigment Epithelial Cell Culturing

hRPE cells were isolated and expanded at Titan Pharmaceuticals, Inc. (Somerville, NJ) and frozen after 2 to 3 passages; each passage represents the subdivision of hRPE cells into 2 new culture flasks with fresh culture medium. We received hRPE cells frozen in storage medium containing 7.5% dimethyl sulfoxide, 20% fetal calf serum, and Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Carlsbad, CA). Upon arrival, hRPE cells were stored in vapor phase liquid nitrogen until use. Approximately 4 to 6 days before implantation, cryopreserved hRPE cells were rapidly thawed in a 37°C water bath and resuspended in high-glucose DMEM (with L-glutamine, 25 mM HEPES buffer, and pyridoxine hydrochloride; Invitrogen) containing 10% heat inactivated fetal bovine serum (Biomeda, Foster City, CA). This medium will be referred to hereafter as "complete hRPE medium." The cell suspension was centrifuged, and the remaining cell pellet was seeded in nonlaminated polystyrene T-25 cell culture flasks containing complete hRPE medium. Initial seeding densities ranged from 200,000 to 300,000 cells per flask with greater than 85% viability. hRPE cells were grown to 100% confluence in an incubator (37°C, 5% CO2) before GM attachment and implantation. HRPE medium was exchanged every 48 hours or if a color change was noted in the hRPE medium, and hRPE cells were examined regularly for proper cell growth and health.

Human Retinal Pigment Epithelial Cell Attachment to Microcarriers

Twenty-four hours before intended use, 40- to 60-μm-diameter dry GMs (proprietary; Theracell Inc., Menlo Park, CA) were hydrated in a 1.5-ml microcentrifuge tube with calcium/magnesium-free PBS (Invitrogen) for a minimum of 1.5 hours and autoclaved (121°C, 15 psi, 15 minutes) for sterility. Sterile GMs were resuspended and washed in fresh PBS and stored in complete hRPE medium until cell attachment time.

The 100% confluent hRPE cells were removed from the incubator, washed with PBS, and harvested by trypsinization and mechanical agitation. The hRPE cell suspension was transferred to a sterile polypropylene-coated 15-ml tube containing complete medium and centrifuged. After centrifugation, the hRPE cell pellet was resuspended with fresh complete medium, and a small (10 μl) sample of hRPE cells was taken and assessed for cell viability and vial concentration using the trypan blue exclusion method (minimal criteria: ≥80% cell viability and ≥1 × 106 cells/ml solution) to stain dead or dying hRPE cells blue. Cell viability (calculated as live cells [no trypan blue inclusion]/total number of hRPE cells × 100%) and the calculated mean number of hRPE cells per vial was then used to determine the volume of cell suspension needed for a final concentration of 1 × 106 cells/10 mg of GM suspension. hRPE cells were passively adsorbed or adhered to GMs, and the hRPE-GM suspension was stored in complete medium in a 1.7-ml tube, which was placed on its side in a 37°C incubator for 15 to 18 hours until immediately before implant.

On the morning of implantation, the hRPE-GM suspension was gently washed with sterile Hanks' balanced salt solution (HBSS) (without phenol red; Invitrogen) to remove any unattached hRPE cells. Before use, a slurry was created by removing the HBSS until a liquid meniscus formed just above the hRPE-GM suspension. A small (10 μl) sample was taken, treated with a Dispase solution to break down GMs, and assessed for cell viability and dose concentration using the trypan blue exclusion method. hRPE-GM suspensions not meeting the minimum required cell viability (≥80%) or dose concentration (≥2000 cells/μl hRPE-GM suspension) were reassessed and/or discarded. GM suspensions were treated in a similar manner. All preparations were kept cool in a beaker surrounded by shaved ice during the implantation procedure for a maximum storage time of 6 hours.

Details of the surgical implantation are described in the companion article (18). Briefly, before each implant, 6 μl of the hRPE-GM slurry was drawn into a sterile 20-μl Hamilton syringe (preloaded with sterile HBSS) attached to a 22-gauge micropolished beveled needle to facilitate the passage of beads. Each rat received 2 striatal hRPE-GM (or GM-alone) implants on the same side as the lesion at the following coordinates: AP +1.6, ML −2.5, and DV −6.0, −4.0; and AP −0.4, ML −3.5, and DV −6.0, −4.0. At each AP coordinate, using a "pulse-like" injection technique, the 6 μl of hRPE-GM slurry was divided between the 2 DV locations (DV −6.0, −4.0). Each animal was injected with a set volume of hRPE-GM suspension with hRPE-cell concentrations dependent on preimplant calculations (see Results section). GM-alone implants were performed following the same procedure.


Rats were killed at several time points postimplant: 48 hours (n = 4), 1 week (n = 4), 4 weeks (n = 4), and 5 months (n = 4). Rats from the 5-month group underwent behavioral testing as described elsewhere (18). Rats were euthanized by administration of a ketamine HCl-xylazine overdose, and perfused with 1 of 2 fixatives, depending on the antibody used. Some of the rats were perfused through the ascending aorta under systolic pressure with 0.9% cold saline followed by 70% EtOH in distilled water or 4% paraformaldehyde in 0.1 M phosphate buffer. The brain was removed, cut into 4- to 5-mm-thick coronal blocks and immersed in additional 70% EtOH (at 4°C) for 24 to 48 hours. The remaining brains, fixed with 4% paraformaldehyde, were immersed in additional fixation medium for 24 hours, transferred to PBS for 2 days, and then stored in 70% EtOH until paraffin processing. Coronal blocks were paraffin-embedded, and coronal sections (4 μm) were made through the implant site, placed on Superfrost-plus charged slides (Fisher Scientific, Pittsburgh, PA), and placed in a 37°C oven overnight.

To identify the implant site, representative coronal sections through the striatum were deparaffinized and stained with Gill's hematoxylin solution #2 (Fisher Scientific) and ethanolic eosin (hematoxylin and eosin [H&E]), dehydrated in ascending grades of EtOH, cleared in xylene, and mounted with Permount. Adjacent sections were processed for IHC.

In vitro hRPE-GM pellets for histologic analyses were prepared in a similar manner. After the 15- to 18-hour attachment time, the hRPE-GM suspension was washed in HBSS and a small sample hRPE-GM was Dispase-treated and assessed for cell viability and attachment efficiency. The hRPE-GM suspension was then fixed with a sterile 10% formalin solution, washed in additional HBSS, and stored in 70% EtOH until the pellet was paraffin embedded. Sections (4 μm) were made through the hRPE-GM pellet for IHC.


Coronal sections were deparaffinized in 3 washes of xylene and hydrated in descending concentrations of EtOH. After deparaffinization, in vitro hRPE-GM sections and in vivo sections fixed with 4% paraformaldehyde underwent 20 minutes of microwave heat-induced antigen retrieval in 10 mM citric acid buffer (pH 6.0; Vector Laboratories, Burlingame, CA). EtOH-fixed sections did not require antigen retrieval. After cooling, in vitro sections and 70% EtOH-fixed brain sections were treated with a blocking solution (for 1 hour at room temperature) containing 10% normal goat serum, 2% bovine serum albumin, and 0.25% Triton X-100 (Sigma-Aldrich). EtOH-fixed sections were incubated overnight at 4°C with the following primary antibodies diluted in blocking solution: 1) nuclear mitotic apparatus protein (NuMA-Ab2) (1:500; Oncogene Research Products, San Diego, CA), a mouse monoclonal antibody (Ab) against a 240-kDa nuclear matrix protein derived from the human cervical carcinoma cell line; or 2) extracellular matrix metalloproteinase inducer (EMMPRIN) (1:500; Zymed Laboratories, South San Francisco, CA), a rabbit polyclonal Ab against the cell membrane of normal epithelium.

Sections treated with anti-NuMA mouse primary Ab were washed with 0.1 M PBS (pH 7.4) and incubated with the fluorochrome-conjugated secondary Ab Alexa Fluor 488 goat anti-mouse IgG (1:500; Molecular Probes, Eugene, OR). Polyclonal EMMPRIN-treated tissue was incubated with either Alexa Fluor 488 or Alexa Fluor 555 goat anti-rabbit IgG (1:500, Molecular Probes) fluorescent secondary antibodies. Secondary Ab incubation was for 1 hour at room temperature. Sections were cover slipped and sealed using an antifade mounting medium (Molecular Probes). Sections were examined using an Olympus Fluoview 500 confocal laser scanning microscope. Differential interference contrast images of sections were taken with each scan and viewed using Olympus Fluoview software. Negative control sections were incubated with no primary Ab.

Tissue sections from a subsequent set of animals (4% paraformaldehyde fixed) at 3 and 5 months (n = 5) underwent IHC using RPE65 (Chemicon, Temecula, CA), a mouse monoclonal Ab that specifically reacts to a 65-kDA protein on retinal pigment epithelial (RPE) cell membranes, or DA-specific mouse monoclonal TH (Sigma Aldrich). RPE65 IHC has been previously described for flat-mounted, retinal tissue sections (19). Therefore, we describe here our development of RPE65 IHC for fixed, coronal, paraffin-embedded brain sections. Paraformaldehyde-fixed sections were subjected to antigen retrieval, were quenched in 3% hydrogen peroxide in PBS for 15 minutes, and were subsequently treated with a blocking solution containing 0.5% Tween-20 and 10% normal goat serum. Sections were incubated overnight at 4°C with anti-RPE65 (1:10) diluted in blocking solution. After primary Ab incubation, sections were washed and treated using a goat anti-mouse secondary Ab preadsorbed to rat IgG (1:300; Jackson Immunoresearch, West Grove, PA), and then visualized using a Vector elite ABC kit (Vector Laboratories) and 3,3′-diaminobenzidine (DAB) (MP Biomedicals, Solon, OH). In addition, to confirm TH-positive findings in previous studies (17), sections taken from a set of animals at 12 weeks (3 months; an intermediate time point postimplant in the behavioral studies described in Reference 18) were blocked with a solution containing 10% normal horse serum, 2% bovine serum albumin, and 0.1% Triton X-100 and incubated overnight with anti-TH (1:400). After primary Ab incubation, sections were treated with a horse anti-mouse secondary Ab, visualized using a Vector elite ABC kit and Vector red (Vector Laboratories), and counterstained with hematoxylin. Sections were then dehydrated, coverslipped, and examined by light microscopy.

Additional tissue sections were immunohistochemically stained for glial fibrillary acidic protein (GFAP) and ED1 (rat CD68) to identify astrocytes and macrophages/activated microglia, respectively. Briefly, tissue undergoing ED1 IHC were digested with proteinase-K (5 μg/ml, Sigma-Aldrich) at 37°C for 15 minutes. Sections were then incubated overnight at 4°C with either 1) rabbit polyclonal anti-GFAP (1:1000; Chemicon) or 2) mouse monoclonal anti-ED1 (1:500; Serotec, Raleigh, NC). Tissue sections were then processed and visualized using a Vector elite ABC kit and 3,3′-DAB, dehydrated, coverslipped and examined.

Electron Microscopy

For ultrastructural analyses, rats were euthanized and perfused through the ascending aorta with a fixative solution containing 4% paraformaldehyde and 1% glutaraldehyde in 0.1 M phosphate buffer fixative for 60 minutes. Afterwards, the brain was removed and placed in additional fixative solution at 4°C for 24 to 48 hours. Ultrastructural analyses were performed on in vitro hRPE-GM and in vivo implanted hRPE-GM at 48 hours (n = 2) and 5 months (n = 1) postimplant. Four to 6 blocks of whole brain tissue (measuring approximately 5.0 × 5.0 × 10.0 mm) through the implant site were sectioned, and each block was cut with the long axis oriented in the dorsal-ventral direction. Individual tissue blocks were washed in 0.1 M PBS and postfixed overnight in 1% buffered osmium tetroxide. Blocks were then washed in acetate buffer (pH 3.7) and stained with 2% aqueous uranyl acetate. Tissue blocks were dehydrated in ascending grades of EtOH, equilibrated in propylene oxide, and embedded in Epon. Each block was trimmed, and ultrathin (65-70 nm) sections were collected, individually mounted on Formvar-coated slot grids, and stained with lead citrate. Electron micrographs were photographed with 3.25 × 4.00-inch Kodak 4489 plate film, and negatives were scanned on an Epson Expression 1680 scanner using Epson Twain-Pro software. All images were taken as TIFF files and edited with Adobe Photoshop.


All lesioned rats demonstrated behavioral deficits manifested by a lack of use of the contralateral forelimb and a significant increase in hindlimb contralateral step errors during the ledged beam test (see companion article 18). With the use of previously described phosphor-imaging autoradiography (20), a random subset of animals (n = 4) underwent [3H]WIN 35,428 (PerkinElmer, Boston, MA) binding to the striatal DA transporter to quantify 6-OHDA lesion severity. [3H]WIN 35,428 binding (using a minimum of six 20-μm coronal sections per animal) showed a >97% loss of striatal DA terminals compared with the unlesioned striatum.

All hRPE cell cultures appeared healthy and presented unaltered morphology during the 5-day incubation period before cell attachment. The cells displayed typical elliptical-shaped nuclei and elongated cell shape. They reached 100% confluence in T-25 flasks within 4 to 6 days after seeding, arranging in a monolayer with a characteristic cobblestone appearance. Confluent hRPE cells produced consistent results: cell numbers ranged from 2 to 2.5 × 106 cells per flask, with cell viabilities between 80% and 92%. Only after cells were 100% confluent (and before they started to proliferate beyond a monolayer) were they trypsinized and prepared for GM attachment. hRPE-GM dose estimates before implantation consistently resulted in greater than 83% (89 ± 4.5%) viability with a mean calculated dose concentration of 2,453 ± 369 cells/μl of hRPE-GM solution (total of 24,000-30,000 hRPE cells injected). At the end of each surgical day, cell assessments were made to determine changes in cell viability and concentration during storage. Cell viability slightly decreased to 83 ± 3.2%.

In Vitro Characterization of Human Retinal Pigment Epithelial Cells

All primary antibodies were previously assessed for hRPE cell specificity in human RPE cultured monolayer and RPE-choroid eye tissue, and for nonspecificity to striatal rat tissue (Dr. S. Bernstein, QualTek Molecular Laboratories, Santa Barbara, CA, personal communication, 2003). These results were confirmed in our laboratory. NuMA-Ab2 demonstrated good staining of hRPE cells, with minimal or no staining of GMs or rat tissue. EMMPRIN and RPE65 also demonstrated good specificity to human epithelial structures in hRPE cells but not to rat tissue or GM components. GFAP and ED1 (CD68) showed typical staining in vivo in rat tissue but were negative for hRPE cells in vitro.

The morphologic features of in vitro hRPE-GM are illustrated in Figure 1. GM profiles were spherical or ovoid in shape with characteristic surface grooves and pores, with a maximum diameter of approximately 60 μm (Fig. 1A, B). GMs were moderately eosinophilic (Fig. 1A). The majority of hRPE cells were attached to the outer edges of the GMs, exhibiting an elongated cell profile with large, irregularly shaped, eosinophilic cytoplasm. In some cases, hRPE cells appeared to have penetrated the GM grooves and pores. Relatively few hRPE cells were observed to be isolated, not attached to GMs. The morphology of attached hRPE cells did not resemble the typical cobblestone-like appearance characteristic of hRPE cells in monolayer during cell culture. RPE cells attached to the GMs were immunoreactive for NuMA-Ab2 (Fig. 1C), EMMPRIN (Fig. 1D), and RPE65 (Fig. 1E) but not for GFAP (Fig. 1F). The GMs displayed little or no immunoreactivity to any of the antisera.


In vitro characterization of human retinal pigment epithelial cells attached to gelatin microcarriers (hRPE-GM). Hematoxylin and eosin-stained (A) and differential interference contrast photomicrograph (B) of hRPE cells attached to GM. HRPE cells are characterized in vitro by NuMA-Ab2 (C), EMMPRIN (D), and RPE65 (E). No GFAP-positive reactivity was seen hRPE-cell cultures (F). The DIC photomicrograph in (B) corresponds to the EMMPRIN fluorescence shown in (D). Scale bars = 25 μm. (Note: scale bar in F applies to B-E.)

In Vivo Characterization of Gelatin Microcarrier-Attached Human Retinal Pigment Epithelial Cell Implants

Histologic sections stained with H&E were used to verify the striatal placement of the hRPE-GM implants (Fig. 2). All implants were localized within the striatum. Individual implants contained a variable number of GMs at all postsurgical time points examined. In a few rats, some of the GMs were observed in the subcortical white matter and corpus callosum, adjacent to the injection tract. At 48 hours and 1 week after surgery, the implant sites were characterized by edema, focal hemorrhage, and an accumulation of inflammatory infiltrates, within and immediately surrounding the implant site (Fig. 2A, B, E). Electron microscopy confirmed that the inflammatory infiltrates consisted predominantly of polymorphonuclear leukocytes. At 1 month and 5 months postimplant, the edema and hemorrhage had resolved as the adjacent neuropil of the striatum gradually closed over the implant site (Fig. 2C, D, F). Gliosis was largely limited to the neuropil immediately surrounding the implant, and macrophages and microglia were observed in and around the implant site.


Representative photomicrographs of the intrastriatal implant sites in rats 48 hours (A), 1 week (B), 1 month (C) and 5 months (D) after implantation. At higher magnification, edema, hemorrhage, and acute inflammatory infiltrates were observed at 48 hours postimplant (E). By 5 months postimplant (F) macrophages and activated microglia were observed, whereas the surrounding neuropil had gradually closed over the implantation site. Sections are stained with hematoxylin and eosin. Scale bars = (A-D) 200 μm; (E, F) 30 μm.

IHC using a number of different antibodies was used to demonstrate the survival of implanted hRPE cells. NuMA-positive cells were found attached to the implanted GMs at all time points in the experiment (Fig. 3). Although there was some heterogeneity among animals, there appeared to be a gradual loss of immunoreactive hRPE cells within the implant. NuMA-immunoreactive cells were most numerous at 48 hours postimplant, but there was a marked decrease in their numbers at later time points. Positive staining was more homogeneous at late time points, which allowed us to make gross visual estimates of the number of cells per bead of 4 to 5 cells/bead (images not shown). This estimate was consistent with the calculated hRPE cell/bead ratio detailed in a previous study (17). No NuMA-immunoreactive cells were observed outside of the implant.


In vivo human retinal pigment epithelial cells attached to gelatin microcarriers implants demonstrating NuMA-Ab2 fluorescence at 48 hours (A), 1 week (B), 4 weeks (C), and 5 months (D) postimplant. NuMA-Ab2-positive cells are seen both infiltrating and on the outer edges of GMs. Differential interference contrast NuMA fluorescence overlay images (not shown) indicate that all NuMA-Ab2 cells are attached to variable-sized GMs. Scale bar = 25 μm.

Figure 4 illustrates EMMPRIN-immunoreactive cells at the different time points postimplant. EMMPRIN-positive cells were observed at all time points in the experiment, attached to the GM surfaces and infiltrating the surface pores. The morphology of EMMPRIN-positive cells was similar to that observed in hRPE-GM in vitro. At 48 hours postimplant, EMMPRIN-positive cells were most numerous, although smaller in size than at later time points. The number of immunoreactive cells decreased noticeably from 48 hours to 1 week after implantation and then decreased more gradually at later time points. Implanted hRPE cells with confluent cytoplasm were observed attached to the GMs as long as 5 months after the surgery. EMMPRIN-immunoreactive cells were seen only in the implant site, with no staining observed in other brain regions or surrounding neuropil.


In vivo human retinal pigment epithelial cells attached to gelatin microcarriers implants demonstrating EMMPRIN fluorescence at 48 hours (A), 1 week (B), 4 weeks (C), and 5 months (D) postimplant. Scale = 25 μm.

IHC using antibodies against RPE65 confirmed the survival of implanted hRPE cells as late as 5 months postimplant (Fig. 5). RPE65-immunoreactive cells were observed attached to the surface and within the pores of GMs, although occasional isolated cells could be seen in the space between the GMs (Fig. 5A). These cells may have been attached to GMs beneath or above the plane of section. No detached immunoreactive cells or cells outside the implant area were observed.


(A) RPE65 immunohistochemistry at 5 months postimplant, illustrating the distribution of human retinal pigment epithelial (hRPE) cells attached to gelatin microcarriers. Scale bar = 50 μm. (B) At higher magnification, immunoreactivity can be seen localized to the cytoplasm of hRPE cells. Scale bar = 30 μm.

TH IHC revealed a prominent unilateral striatal lesion in all animals studied (Fig. 6A). Striatal sections from a subgroup of rats 3 months postimplant revealed TH-immunoreactive cells within the GMs (Fig. 6B, C) on the lesioned side. TH immunoreactivity was largely cytoplasmic surrounding distinct, round nuclei (stained with hematoxylin) and was predominantly found within the grooves (not the outer surface) of GMs. This was presumed to be TH-positive presumptive hRPE cells as no other TH immunoreactivity was found anywhere else in the lesioned striatum (Fig. 6A).


Tyrosine hydroxylase (TH) immunohistochemistry at 3 months postimplant. (A) Low-magnification photomicrograph showing prominent asymmetric TH staining of the left striatum with a near complete depletion of the contralateral, human retinal pigment epithelial cells attached to gelatin microcarriers (hRPE-GM) implanted, striatum. Scale bar = 1 mm. Note the presence of the implant (containing visible GMs) near the center of the lesioned striatum. (B, C) At higher magnification, TH immunoreactivity, localized to the cytoplasm surrounding presumptive (hematoxylin-stained) hRPE cell nuclei, can be seen within the pores of GMs. Scale bar = 20 μm.

Electron Microscopy

The identity of implanted hRPE cells and their survival up to 5 months postimplant were confirmed by electron microscopy (Fig. 7). In vitro preparations of hRPE-GM displayed the ultrastructural characteristics of hRPE cells (Fig. 7A), including evenly distributed dense nuclear chromatin and dark cytoplasm containing an abundance of rough endoplasmic reticulum. Electron-dense pigment granules and melanosomes were observed in the cytoplasm of these cells, although the relative size and number of these inclusions varied considerably among individual cells (Fig. 7A). At 48 hours postimplant, numerous cells with similar morphology were observed attached to the GMs (Fig. 7B). These cells displayed the characteristic dark cytoplasm with abundant rough endoplasmic reticulum and occasional pigment granules. Relatively fewer hRPE cells were observed at 5 months postimplant, but they contained a greater volume of confluent cytoplasm and more abundant pigment granules and melanosomes (Fig. 7C).


Electron micrographs of human retinal pigment epithelial cells from an in vitro preparation (A) and at 48 hours (B) and 5 months (C) after implantation. Note the relative electron density of the gelatin microcarriers in (B) and (C). Scale bars = 5 μm.

Immune Response to Gelatin Microcarrier-Attached Human Retinal Pigment Epithelial Cell Implants

Photomicrographs of ED1 and GFAP immunoreactivity at the implant site at 5 months postimplant are illustrated in Figure 8. ED1-positive macrophages and activated microglia were observed most frequently in the neuropil at the edge of the implant site (Fig. 8A), although occasional immunoreactive cells were observed within the implant. GFAP immunoreactivity revealed hypertrophied astrocytes adjacent to the implant site (Fig. 8B), which encapsulated the implant to a depth of approximately 50 to 100 μm. This observation was confirmed by electron microscopy. Interestingly, the neuropil of the striatum immediately adjacent to the layer of hypertrophied astrocyte processes appeared relatively normal.


(A) Macrophages and activated microglia, identified by their immunoreactivity to ED1 (black arrows), in and around the implant site at 5 months after implantation. Scale bar = 100 μm. (B) Glial fibrillary acidic protein immunohistochemistry illustrating gliosis and hypertrophied astrocyte processes encapsulating the implant site at 5 months postimplant. Scale bar = 50 μm.


This is the first study to assess the specificity of a number of immunohistochemical markers for identifying hRPE cells both in vitro and implanted into the rat brain. Our qualitative evaluation demonstrates the presence of hRPE cells when implanted into parkinsonian brains. hRPE cells attached to GMs and implanted into the lesioned striatum of rats can be identified up to 5 months postimplant as shown by NuMA-Ab2, EMMPRIN, RPE65, and TH immunoreactivity. This finding indicates the relatively long-term survival of hRPE cells despite the absence of immunosuppression and is consistent with previous studies in which the potential survival of presumptive hRPE cell xenotransplants in hemiparkinsonian rats were described (17).

In Vitro Human Retinal Pigment Epithelial Cell Characterization

The immunomarkers used, NuMA-Ab2, EMMPRIN, and RPE65, which are human nuclear, epithelial, and RPE-specific markers, respectively, were chosen strictly for their potential specificity. Previous studies have shown specific NuMA-Ab2 and EMMPRIN immunoreactivity in hRPE cell monolayers and RPE/choroid tissue, but not in the rat striatum (Dr. S. Bernstein, QualTek Molecular Laboratories, personal communication, 2003). Our in vitro data validated these results: immunopositive NuMA-Ab2 and EMMPRIN staining was seen in hRPE cells attached to GMs in vitro. In addition, GFAP expression was negative in hRPE-GM cultures, and there was minimal antibody absorption by the microcarriers. The morphologic features of NuMA- and EMMPRIN-positive cells were identical to those of hRPE cells along the GM surface, stained with H&E or examined by differential interference contrast microscopy. NuMA and EMMPRIN staining showed both the characteristic stretching and elongation of hRPE cells with attachment to the GM surface, with some cells penetrating within the GM grooves and pores.

Similarly, antibodies against RPE65 showed good specific reactivity to hRPE cells with negative reactivity to rat brain tissue. RPE65 staining also showed characteristic stretching of hRPE cells along the outer surface of GMs, indicative of RPE65 protein localized to the RPE cell membrane. RPE65 immunoreactivity has been shown to be present in RPE/choroid, normal human epidermis, and other whole tissue preparations (21, 22). Interestingly, our preliminary results from an unpublished study demonstrated that RPE65 immunoreactivity is negative in isolated hRPE cells grown into a monolayer in culture (data not shown). This is consistent with previous hypotheses that some of the properties expressed by RPE cells in situ can be lost by these cells in culture, including the expression of proteins such as RPE65 (23). However, the RPE65 antigen can reexpress itself once hRPE cells are attached to GMs. Studies are underway to ascertain any phenotypic differences between hRPE cells in normal host versus culture conditions.

Human Retinal Pigment Epithelial Cell Implantation

hRPE cells are epithelial cells and, as such, can be grown in culture and survive well for extended periods of time (24), allowing for 1 donor to provide sufficient tissue for implantation into multiple recipients. Furthermore, hRPE cell culture utilizes routine culturing procedures, making screening for contaminants such as viruses, endotoxins, or mycoplasms feasible. hRPE cells can be cryopreserved for long-term storage and/or shipping without losing their viability (24). This is an important issue to consider when one is developing a new cell therapy. For instance, the cryopreservation techniques used to store fetal ventral mesencephalic tissue before transplantation (5, 25) were associated with significant decreases in cell viability (26).

We observed consistent viability between experiments. All cell viabilities were in the clinically acceptable range (>80%) with dose concentrations within the normal range of 2,000 to 3,000 cells/μl of hRPE-GM solution. Postsurgery viability counts were taken to determine viability changes between the beginning and end of a routine surgical day (i.e. 4-6 hours). During surgery, viability was consistent and never dropped below 80%. These viability counts indicate that the majority of implanted cells were healthy, living hRPE cells.

The results of the present study demonstrate that an appreciable number of hRPE cells implanted into the rat brain survived the 5-month experimental period. All antibodies used, NuMA-Ab2, EMMPRIN, RPE65, and TH, showed numerous immunopositive presumptive hRPE cells within the implant attached to GMs. These hRPE cells were restricted to the implant site with no observable migration to the surrounding striatal neuropil. Detached hRPE cells were rarely observed, which is expected as most detached hRPE cells will undergo immediate apoptosis if there is no extracellular matrix support (13). Of particular importance was the RPE65 immunoreactivity. Numerous RPE65-immunoreactive cells were observed within the implant site. The antigenic profile of RPE65 was more favorable than that for either NuMA or EMMPRIN immunofluorescence. RPE65 appeared more compartmentalized, showing distinct immunoreactivity to the cytoplasm of presumed RPE cells. As observed in many cell cultures, when hRPE cells are isolated and grown in culture, they appear to lose some of their phenotypic properties. Previously, it was uncertain whether or not hRPE cells on GMs would continue to express their typical hRPE phenotype when implanted in a foreign tissue environment such as the brain. Our results demonstrate RPE65 staining within the implant site, which suggests that hRPE cells, when attached to GMs, express at least some of their typical hRPE characteristics.

Each primary Ab we assessed possesses certain advantages and disadvantages with respect to future use. hRPE cell identification using antibodies against NuMA, which is human-specific, may be beneficial for use in rodent models because of its specificity to human antigens in an animal brain. However, this feature will hamper its use in human and nonhuman primate studies because of its relative nonspecificity to RPE cells and cross-reactivity to primate and human antigens. Furthermore, our inability to restrict NuMA immunofluorescence to only label cell nuclei may restrict its use in any quantitative procedure. Similarly, TH IHC is an indirect measure of hRPE cell survival, as it may cross-react with surviving striatal DA terminals. It is also unclear whether hRPE cells can constitutively express TH: hRPE cells have been reported to contain TH-like activity (12, 27), but whether this expression continues when hRPE cells are implanted into a novel environment still needs to be studied. Thus, although TH-immunoreactivity was observed in presumptive hRPE cells at the implant site and has been previously used to show hRPE cell survival (17), it is unlikely to be of use for quantifying cell survival because of its poor hRPE cell specificity. In contrast, the use of antibodies against RPE65 and EMMPRIN provides many advantages, including robust specificity for RPE and epithelial cells and minimal cross reactivity in all species.

Electron microscopy confirmed our immunohistochemical findings. hRPE cells exhibited a characteristic morphology when attached to the GMs. hRPE cells displayed large cytoplasmic volumes containing abundant rough endoplastic reticulum and granular inclusions. The ultrastructure at 5 months was similar to that seen in vitro. Five months after implantation at least some of the hRPE cells contained dense granular inclusions and melanosomes, a sign of healthy hRPE cells actively producing melanin.

Immune Response to Gelatin Microcarrier-Attached Human Retinal Pigment Epithelial Cell Implants

The commonly observed host immune response to a neural xenotransplant in a nonimmunosuppressed rat is initiated within 48 hours of the transplant, and total graft destruction is reported to occur within 3 to 5 weeks postimplant (28). In the present study, we observed signs of acute inflammation at 48 hours and 1 week postimplant and chronic inflammation up to 5 months after implantation. IHC using antibodies against GFAP revealed encapsulation of the implant site by hypertrophied astrocyte processes. Macrophages and activated microglia, revealed by their immunoreactivity to ED1, were most frequently observed in the surrounding neuropil, but also within the implant in proximity to GMs and the implanted hRPE cells. However, despite the presence of chronic inflammation, a substantial number of hRPE cells survived in these rat brains when attached to GMs. In addition, electron microscopy revealed apparently normal neuropil in the surrounding striatum external to the glial encapsulation.

This observation is consistent with a previous study that demonstrated the amelioration of behavioral deficits after hRPE-GM implants with only a minimal host inflammatory response at 18 weeks postimplant (17). Certain hRPE characteristics may increase cell survival in nonimmunosuppressed rats. RPE cells express CD95 antigens and release transforming growth factor-β, both of which may be responsible for local immunosuppression (29). The passive attachment of hRPE cells to GMs can also contribute to their survival postimplant by changing their morphologic features to make the cells more stretched and flattened, a phenomenon that has been suggested to increase cell survival (30). Nevertheless, the presence of an acute inflammatory response is probably responsible for the substantial hRPE cell loss noted in the present study between 48 hours and 1 week postimplant. Alternative transplantation procedures have also shown significant decreases in cell concentrations after transplant (5). Studies in rodents involving fetal mesencephalic xenotransplants demonstrated that only a small fraction of fetal DA neurons survived after transplant (31). Moreover, fetal ventral mesencephalic cell survival was only apparent in immunosuppressed rats. In our study, it is remarkable that human RPE cell survival was apparent as late as 5 months after implantation despite the lack of immunosuppression. Although it was possible to immunosuppress rats with cyclosporin, we elected to reproduce the human protocol in which patients did not receive immunosuppression in our studies, recognizing that we were placing ourselves in a worst case scenario (xenograft versus allograft rejection). Indeed, using RPE65 IHC and a variety of inflammatory markers, a preliminary postmortem examination of an hRPE-GM implanted monkey 4 months postimplant demonstrated relatively reduced inflammatory response compared to that seen in these rat studies (J. Flores and J.R. O'Kusky, unpublished observations, 2006).

A subset of the rats used in this study were tested in parallel for hRPE-GM induced behavioral improvements (18). hRPE-GM striatal implants have been previously shown to ameliorate apomorphine-induced rotations in unilaterally 6-OHDA-lesioned rats (17). Our own studies exhibited similar behavioral improvements in spontaneous motor function in hRPE-GM implanted rats but not in animals implanted with the GMs alone. Although we have not performed a formal stereologic study and cannot firmly comment on possible correlations between behavioral improvement and cell survival, these results suggest a potential relationship between hRPE cell survival and hRPE-GM implant efficacy. Additional studies combining behavioral investigations and hRPE cell counts by morphometric and stereologic analyses are underway to confirm this hypothesis.


We thank Titan Pharmaceuticals, Inc. for their gift of the hRPE cells and gelatin microcarriers. This study would not have been possible without the technical assistance of Jessica Grant and Rick Kornelsen. Special thanks are due to the personnel of the University of British Columbia Animal Care Facilities for their outstanding care of the animals.


  • This work was supported by a grant-in-aid from Titan Pharmaceuticals, Inc. and Schering AG.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
View Abstract