Survival and Long Distance Migration of Brain-Derived Precursor Cells Transplanted to Adult Rat Retina



Neural precursor cells transplanted to adult retina can integrate into the host. This is especially true when the neural precursor rat cell line RN33B is used. This cell line carries the reporter genes LacZ and green fluorescent protein (GFP). In grafted rat eyes, RN33B cells are localized from one eccentricity to the other of the host retina. In the present study, whole-mounted retinas were analyzed to obtain a more appropriate evaluation of the amount of transgene-expressing cells and the migratory capacity of these cells 3 and 8 weeks post-transplantation. Quantification was made of the number of β-galactosidase- and GFP-expressing cells with a semiautomatized stereological cell counting system. With the same system, delineation of the distribution area of the grafted cells was also performed. At 3 weeks, 68% of the grafted eyes contained marker-expressing cells, whereas at 8 weeks only 35% of the eyes contained such cells. Counting of marker-expressing cells demonstrated a lower number of transgene-expressing cells at 3 weeks compared with 8 weeks post-transplantation. The distribution pattern of marker gene-expressing cells revealed cells occupying up to 21% at 3 weeks and up to 68% at 8 weeks of the entire host retina post-grafting. The precursor cells survived well in the adult retina although the most striking feature of the RN33B cell line was its extraordinary migratory capacity. This capability could be useful if precursor cells are used to deliver necessary genes or gene products that need to be distributed over a large diseased area.


The two major groups of retinal degenerations in humans are known as retinitis pigmentosa and age-related macular degeneration. The diseases within these two groups are very heterogeneous, with varying onsets, severity, and other general clinical manifestations, but they all have two things in common: A) they manifest degeneration of photoreceptor cells, and B) there is no effective treatment available.

During the second half of the 1980s, the field of experimental retinal tissue transplantation expanded greatly [1, 2]. The method aims at reconstructing the degenerating retina by subretinal grafting of healthy pieces of retinal tissue, which, upon integration with the host retina, will form a functional unit. A few clinical trials using such retinal transplants have been performed, but so far only limited improvement of visual function has been noted [38].

Instead of using retinal tissue transplants, in vitro expanded multipotent neural precursor cells could serve as donor cells. Currently, protocols are well established for genetic and epigenetic expansion of rodent and human embryonic and adult neural precursor cells. In several experimental intracerebral transplantation studies, such cells have shown a great capacity to survive, migrate, and differentiate into region-specific neurons with appropriate axonal projections as well as to form astrocytes and oligodendrocytes [911].

Retinal transplantation of precursor cells has been performed in order to investigate the possibilities to replace diseased retinal cells and/or to rescue diseased retinal cells in situ by cell-mediated delivery of survival factors. Indeed, transplantation of various neural precursor cells into the retina has been carried out with good cell survival in the normal neonatal and adult host retina [1218], degenerating retina [1923], and injured retina [2426].

To some extent, migration of grafted precursor cells, such as adult rat hippocampal progenitor cells and neonatal rat retinal precursor cells, into the host retina has been observed [12, 13, 19]. We recently described extensive migration in a long-term transplantation experiment using the cell line RN33B, revealing in retinal sections the distribution of RN33B cells from one eccentricity of the adult host retina to the other [17]. Although good retinal graft survival has been estimated by several investigators, no exact number of transplanted cells has been presented, probably due to technical limitations. However, attempts have been made to quantify the survival of grafted cells on cryoprotected sections of the host [12, 13, 19]. In the present study, the brain-derived precursor cell line RN33B was further studied. The cells are genetically immortalized with the temperature-sensitive simian virus (SV40) large T antigen. In order to detect the cells post-grafting, the RN33B cell line is genetically labeled with the reporter genes LacZ and the green fluorescent protein (GFP) [2729].

The purpose of the present study was primarily to estimate survival and assess migration, but also to study integration and differentiation of grafted RN33B cells in whole-mounted retinas.

Materials and Methods

Neural Precursor Cell Line

The RN33B cell line (kindly provided by Dr. S.R. Whittemore) was generated from an embryonic day-12.5 Sprague Dawley (SD) rat medullary raphé nucleus using retroviral transduction of the temperature sensitive mutant of the SV40 large T antigen [27, 30]. When these cells are grown in permissive temperature (33°C) they divide continuously, and when the temperature is raised to nonpermissive levels (37–39°C) the cell division ceases and the differentiation process begins. In brief, the RN33B cells were propagated in Dulbecco's modified Eagle's medium/F12 medium (cat. no 32500, GIBCO, Life Technologies; Grand Island, NY; with 1% glutamine and 0.2% NaHCO3, and supplemented with 10% fetal bovine serum. The cells were cultured on polyornithine-coated flasks (Nalge Nunc; Roskilde, Denmark; with 5% CO2/95% air at 33°C.

The cell line previously had been retrovirally transduced with the reporter gene LacZ [27] and was additionally transduced with the GFP gene using a lentiviral vector [28, 29].


On the day of transplantation, the cell culture medium was replaced with Hank's balanced salt solution (HBSS; GIBCO, Life Technologies) containing 0.1% trypsin. When the cells were detached, serum-containing medium was added. The cells were then centrifuged, resuspended, and counted in a hemocytometer and finally prepared into a suspension of 100,000 cells/μl in HBSS. At three different transplantation time points, adult SD rats (n = 59; BK Universal; Uppsala, Sweden) were grafted (Table 1). The left eyes of the animals were injected according to a previously described method [16, 31] with 2 μl of RN33B cells. A 23-gauge needle (Becton Dickinson; Franklin Lakes, NJ; with the tip bent in a 105-degree angle was used to gently open the sclera and the choroid on the superotemporal side of the left eye, between the ora serrata and the equator. Thereafter, a paracentesis was made at the limbus at the nine o'clock position to lower intraocular pressure. The cell suspension was slowly injected using a GELoader Tip (1–10 μl, Eppendorf; Hamburg, Germany; through the sclerotomy. The injection was completed with a 0.1–0.3-μl air bubble to prevent reflux of the cells. A localized retinal detachment could be seen during the injection through the almost transparent sclera and afterwards through the pupil, which indicated that the cell suspension was delivered subretinally. Additionally, seven control rats were subretinally injected with the HBSS medium only. At surgery, the animals were anesthetized using equithesin (pentobarbital 9.72 mg/ml, chloral hydrate 42.5 mg/ml, magnesium sulphate 86.25 mM, 10% vol/vol ethanol, 40% vol/vol propylene glycol, 3 ml/kg body weight i.p.). All animal-related work was carried out according to local ethical guidelines and approved animal care protocols.

Table Table 1.. Number of animals grafted with RN33B cells and number of animals with surviving cells and survival times. Adult Sprague Dawley rats (n = 59) were transplanted with the neural cell line RN33B and sacrificed 3 (n = 28) and 8 (n = 31) weeks post-transplantation. The β-Gal and GFP expression was examined. Seven rats were injected with the HBSS medium alone, and sacrificed 3 (n = 3) and 8 (n = 4) weeks later.
  1. a

    Transplantation was performed at three different time points: (a) 25 animals with RN33B cells and 2 HBSS controls (3-week survival), (b) animals with RN33B cells (3 weeks, n = 3 and 8 weeks, n = 13) and animals with the HBSS medium (3 weeks, n = 1 and 8 weeks, n = 2), (c) 18 animals with RN33B cells and 2 with HBSS medium (8-week survival).

Animal survival timen of transplanted animalsn of retinas with β-Gal and GFP expressionn of HBSS-injected animals
3 weeks28 (25a+3b)193 (2a+1b)
8 weeks31 (13b+18c)114 (2b+2c)

A few microliters of the cell suspension were spread on a glass slide for epifluorescence microscope analysis of the endogenous GFP expression of the cells prior to transplantation (Fig. 1).

Figure Figure 1..

Smears of the RN33B precursor cells taken from the prepared cell suspension prior to transplantation revealed that almost 100% of the cells expressed endogenous GFP.Scale bar = 60 μm.


Different regimens of immunosuppression previously have been tested on normal rats [16], and daily injections of cyclosporin A resulted in healthy looking retinas; this regimen was therefore applied in the present study.

The graft recipients were immunosuppressed with daily injections of cyclosporin A (10 mg/kg i.p.; Sandimmun Neoral, Novartis; Täby, Sweden; The immunosuppressed rats were also given antibiotics (2 ml/l in drinking water; Borgal, Hoechst Roussel Vet; Münich, Germany;

Whole-Mount Preparation

Three and 8 weeks post-transplantation, 31 rats (28 RN33B transplanted rats and 3 HBSS controls) and 35 rats (31 RN33B transplanted and 4 HBSS controls), respectively, were sacrificed with an overdose of CO2 (Table 1). The eyes were enucleated and placed in 4% paraformaldehyde (PF) for 10–20 minutes. The anterior segment and the lens were then removed; the sensory retina was subsequently dissected, placed on a filter paper (type HA, Millipore; Molsheim, Germany;, and post-fixed in 4% PF for 2–3 hours. Retinas were processed in rising concentrations of sucrose containing Sörensen's phosphate buffer (pH 7.2) and mounted onto glass slides.


The whole-mounted retinas were exposed either to a polyclonal rabbit primary anti-serum against β-galactosidase (β-Gal; 1:500, 5Prime-3Prime, Inc.; Boulder, CO) in a moist chamber (16–18 hours at 4°C) or to a polyclonal chicken primary anti-serum against GFP (1:5,000, Chemicon; Temecula, CA; in a moist chamber for 48 hours at 4°C, followed by rinsing in 0.1 M phosphate-buffered saline with 0.25% Triton-X-100 (Pharmacia Biotech; Uppsala, Sweden; code No 17-1315-01). The retinas were incubated with anti-rabbit Texas Red-conjugated or anti-rabbit fluorescein isothiocyanate (FITC)-conjugated, and/or anti-chicken FITC-conjugated secondary antibodies (1:200, Jackson Immunoresearch; West Grove, PA; for 1–2 hours in room temperature in the dark.

Ten of the retinas transplanted with RN33B cells (two from the 3-week group and eight from the 8-week group), pretreated in 2 M HCl for 10 minutes at 37°C, were used for SV40 large T antigen immunohistochemistry (1:100, Santa Cruz, CA). In addition, two retinas injected with the HBSS medium alone were subjected to the same treatment.

The specimens were examined in an epifluorescence microscope. Distribution patterns of β-Gal- and GFP-expressing cells were examined by superimposing separate digital images of the fluorochromes. As similar results were obtained with confocal microscope analysis (Bio-Rad 1024 UV laser scanning), we found it more convenient to use the superimposed images.

Cell Counting

Retinas with the most intensive endogenous GFP fluorescence and the widest distribution area of grafted cells were chosen for cell counting (3 weeks, n = 6 and 8 weeks, n = 6). A high fluorescence intensity was required since the counting method exposes the retinas to intense light, which causes fading of the fluorescence. Each of the 12 whole-mounted slides was counted using a semiautomatized stereological cell counting system (CAST-Grid software version 1:10; Olympus; Albertslund, Denmark; composed of an Olympus BX50 microscope connected to an X-Y-Z-step motor stage.

The retinal area covered with transplanted GFP-expressing cells was delineated in the whole-mounted retina, and a counting frame was randomly placed in the delineated area to mark the first sampling point before being systematically and randomly moved through the entire area. The procedure was then repeated using the Texas Red filter to count the β-Gal-expressing cells within the delineated area. Data were then extrapolated according to a stereological algorithm: N = SQ × F1 × F2 × F3 (N = total number of marker-expressing cells; SQ = sum of cells counted; F1 = fraction of sections used; F2 = fraction of tissue depth used to collect data; F3 = fraction of tissue area used to collect data) [32]. Delineation of the entire retina was also performed in order to achieve the total retinal area (Table 2).

Table Table 2.. Number of surviving marker expressing RN33B cells and distribution area. Retinas with the most intense endogenous GFP fluorescence and the widest distribution area of grafted cells were chosen for cell counting. A semiautomatized stereological cell counting system was used. With this system, delineation of the retinal area with marker-expressing cells was performed. Delineation of the entire retina was also performed in order to achieve the total retinal area.
  1. a

    (a) The different total retinal area is explained by asymmetric dissection of neuroretina from the scleral cup. (b) The retina was not measured due to retinal fragility. (c) The β-Gal immunoreactivity of the retina was too weak to constitute a reliable counting.

Retina, 3 weeksRetinal area (mm2)aDistribution area with GFP-expressing RN33B cells (mm2)Total n of GFP-expressing cellsTotal n of β-Gal-expressing cells
1not doneb2.33287355
3not doneb3.231,0151,228
Mean 5.55 (± 3.75)1,351 (± 1,045)2,838 (± 2,625)
167.3319.0110,493Not donec
245.6225.6414,673Not donec
3not doneb12.1815,804Not donec
462.1936.7026,203Not donec
556.1239.4533,098Not donec
Mean 28.96 (± 11.85)22,359 (± 10,058) 


Similar results in the 3-week and 8-week animals on survival, migration, integration, and differentiation were obtained independent of the different transplantation sessions (Table 1). Therefore, the 3-week and the 8-week groups with the same survival time were pooled.

Survival of the Transplanted RN33B Cells

Post-transplantation β-Gal- and GFP-expressing cells were found in 68% (19/28) and 35% (11/31) of the grafted animals at 3 and 8 weeks, respectively (Table 1). A large fraction of the transplanted cells coexpressed β-Gal and GFP (Fig. 2A and 2B). Images superimposed on each other showed β-Gal/GFP double-labeling of transplanted cells, as well as subpopulations expressing either β-Gal or GFP (Fig. 2C and 2D). Smears of the RN33B cells taken prior to transplantation revealed that close to 100% of the cells expressed endogenous GFP (Fig. 1). In control animals, untreated or injected with HBSS medium alone, neither β-Gal nor GFP immunoreactivity were found.

Figure Figure 2..

RN33B cells 3 weeks post-transplantation demonstrating immunoreactivity to (A)β-Gal (red) and (B) GFP (green).Both A and B are from the same retinal area. The precursor cells have differentiated with variable morphologies and size. Most of the grafted cells have developed long, slender cell somas and processes. The processes have branched and beaded processes. A few cells only express GFP (arrows). C) A retina with double labeling 3 weeks post-transplantation. Cells expressing eitherβ-Gal (small arrow) or GFP (large arrow), as well as cells expressing both markers (arrowhead), are demonstrated. D) A retina with double labeling 8 weeks post-grafting. Cells expressingβ-Gal (small arrow) or GFP (large arrow), as well as cells expressing both markers (arrowhead), are observed. E) RN33B cells immunoreactive to anti-SV40 (red) 8 weeks post-transplantation. The poor quality of the retina is caused by the pretreatment in 2M HCl before anti-SV40 was used. Both cell markers of RN33B, GFP, andβ-Gal are labeled green. More SV40-expressing cells than marker-expressing cells are revealed, indicating a downregulation of the marker genes. Scale bars = 30 μm.

With the stereological method of analysis, cell counting of six retinas with the most intensive endogenous GFP fluorescence and the widest distribution area of grafted cells from the 3-week group revealed an average of 2,838 (± 2,625) β-Gal-immunoreactive cells and of 1,351 (± 1,045) GFP-positive cells (Table 2). At 8 weeks, the β-Gal expression was too weak for an appropriate analysis, and quantification of β-Gal-expressing cells was only possible in one retina out of six, which displayed 31,985 β-Gal-expressing cells. The average number of GFP-expressing cells in retinas with the most intense fluorescence was estimated as 22,359 (± 10,058) at 8 weeks (n = 6), which indicates that from 3 weeks to 8 weeks post-grafting, the average number of transplanted GFP-expressing cells had increased about 16 times.

Four of the grafted retinas (3 weeks, n = 2 and 8 weeks, n = 2) were exposed to anti-β-Gal and FITC-conjugated secondary antibodies in order to reveal the total population of grafted cells expressing one or both transgenes. The retinas were then further immunoprocessed using anti-SV40 and Texas Red secondary antibody and analyzed for coexpression of β-Gal/GFP and SV40. Cells were observed immunoreactive to β-Gal/GFP and SV40, as well as cells only immunoreactive to SV40 (Fig. 2E). Six of the 20 retinas with no detectable β-Gal or GFP immunoreactivity from the 8-week group were exposed to anti-SV40; no SV40 immunoreactivity was found. Control retinas (n = 2) injected with the HBSS medium alone and untreated retinas did not express SV40 antigen.

Distribution and Migration of the Transplanted RN33B Cells

With the semiautomatized cell counting system, the distribution area of GFP-positive grafted cells could be delineated. The population of GFP-positive cells was easier to distinguish, thus the distribution area of grafted cells was delineated according to these cells. Since the distribution area of the β-Gal-expressing cells overlapped the area of GFP-expressing cells at both survival times, this was not considered to introduce a significant bias. The total adult retinal area was estimated to an average of 60.17 (± 7.84) mm2 using values obtained from the nine retinas measured (Table 2). At 3 weeks, RN33B cells were found on a restricted retinal area of, at most, 12.57 mm2 (mean, 5.55 ± 3.75) and on an area of up to 40.8 mm2 (mean, 28.96 ± 11.85) 8 weeks following grafting (Table 2). This correlates to a migratory capacity of the RN33B cells to be distributed throughout a maximum of 21% and 68% of the entire host retina at 3 and 8 weeks, respectively.

Although the RN33B cells migrated over long distances, a large fraction of the grafted cells was found at or close to the implantation site at both 3 and 8 weeks post-grafting. Aggregates of marker-expressing cells were found in the surrounding retina and single scattered cells were seen far from the implantation site (Fig. 3).

Figure Figure 3..

Whole-mounted retina 8 weeks post-transplantation with GFP-expressing RN33B cells (white).A-J represent parts of the schematic drawing of a whole-mounted retina (the boxes in the drawing are larger than the representing digital pictures). Close to the implantation site (black arrow) many cells are demonstrated (C, D, E, and G), and far from the implantation site single scattered cells are revealed (A and H, white arrow). Scale bar = 120 μm.

A distinctive migration pattern was found in four of the grafted retinas at 8 weeks, with RN33B cells migrating as chains forming tubule-like structures (Fig. 4). These cells had morphologies of migrating cells, with elongated cell bodies and an apical process in the direction of the migration [33]. A leading process was demonstrated on both single migrating cells as well as on the cell tubules. The chains of transplanted cells were often arranged radial to the optic nerve head but not along the great retinal vessels (Fig. 5).

Figure Figure 4..

Chains of migrating RN33B cells at 8 weeks after grafting.These long, slender GFP-expressing cells seemed to climb along each other. A leading process of the whole tubule can be seen. Scale bar = 30 μm.

Figure Figure 5..

Higher magnification of Figure3D. Near the optic nerve head (ON), single cells and chains of grafted cells are arranged radially (arrows) and oriented towards the optic nerve. Scale bar = 120 μm.

Differentiation and Integration of the Transplanted RN33B Cells

Previously, both the GFP and β-Gal reporter genes have been reported to be highly useful for morphological analysis of grafted neural precursors [29, 34, 35]. However, in this cell line, GFP allows for a more fine-detailed detection of structural elements such as dendritic spines, terminal boutons, and axonal projections than β-Gal [29]. In agreement, also in the present study, GFP was found to be more informative concerning cell morphologies than β-Gal (Fig. 2A).

Both at 3 weeks and 8 weeks post-transplantation, the RN33B cells displayed highly variable morphologies and sizes (Figs. 2C and 2D, 6A and 6B). Differentiated RN33B cells did not differ in morphology between the 3- and 8-week retinas (Fig. 2C and 2D). The grafted cells adapted both neuronal and glial-like morphologies. The cellular profiles included bipolar-like cells with small round or long slender somas and multipolar cells with different sizes of the cell somas. Most of the cells had varying numbers of processes, many of them being branched with multiple neuronal-like spines (Figs. 2B, 6C, and 6D). The cell processes were often in close contact with each other, indicating a cell-to-cell contact (Figs. 2C, 2D, and 6C).

Figure Figure 6..

RN33B cells 3 weeks post-transplantation on two different retinal areas.A)β-Gal-expressing cells (red); B) GFP-expressing cells (green). Different morphologies can be detected, from rounded to long, slender cell somas. Most of the cells have developed long branched processes. Scale bars = 60 μm. C) RN33B cells 3 weeks post-transplantation. The branched processes of the grafted cells are often in close contact with each other (arrow). The processes are beaded or have small spines. Scale bar = 30 μm. D) RN33B cells 3 weeks post-grafting demonstratingβ-Gal (red) and one double-labeled cell (yellow) with four branched and beaded processes. Scale bar = 20 μm. E) RN33B cells expressingβ-Gal (red) 8 weeks post-transplantation on a whole-mounted retina that has been sectioned. Most of the cells are integrated into the OPL and IPL. Scale bar = 60 μm.

In addition, the grafted cells were observed in different layers of the retina, which was revealed using varying focus levels in the microscope. From previous studies [16, 17, 22], it is known that grafted RN33B cells migrate and integrate mainly into the outer plexiform layer (OPL) and the inner plexiform layer (IPL) of the retina. The same distribution pattern was observed when one whole-mounted retina was sectioned (Fig. 6E).


The present results demonstrate on whole-mount preparations the survival (3 or 8 weeks post-transplantation), migration, integration, and differentiation of the brain-derived precursor cell line RN33B after subretinal transplantation to the immunosuppressed adult rat. The findings are in accordance with previous studies from our laboratory on subretinal transplantation of this cell line [16, 17, 22].

Survival of the Transplanted RN33B Cells

In the present study, the survival and number of grafted RN33B cells was based on the expression of the reporter genes for GFP and β-Gal. The GFP marker has proven to possess a long-term and stable expression after transplantation, to be specific for the transplanted cells, and to not be transported to host cells [28, 29]. Here, expression of both the GFP and β-Gal transgenes was observed up to 8 weeks post-transplantation on whole-mounted retinas. Significant numbers of GFP- and β-Gal-positive cells were demonstrated on 68% of the transplanted eyes at 3 weeks and 35% at 8 weeks post-transplantation. The mechanism(s) behind the major drop in survival following transplantation, which is as low as zero survival in 32% (at 3 weeks) and 65% (at 8 weeks) of the eyes, is unclear. However, the decline in survival rate at longer survival times might be due to slow rejection of the RN33B cells.

Our recent studies demonstrate that RN33B cells continue to proliferate after transplantation, judging from the expression of the proliferation markers proliferating cell nuclear antigen and SV40 large T antigen [17, 22, 29]. In the brain, only immature cell profiles express SV40 [29]. It might therefore be suggested that RN33B cells expressing only SV40, and not β-Gal or GFP, are newly formed precursors that have downregulated the reporter genes. However, it cannot be excluded that the persistence of SV40 expression in vivo reflects a loss of the temperature sensitivity of the RN33B cells. The larger number of GFP-positive cells (average, 22,359) estimated by stereological counting at 8 weeks compared with 3 weeks (average, 1,351), further confirms the mitotic activity of RN33B cells after implantation into the adult retina.

The variation of expression of reporter genes (as used here) depends on, for example, vector construct, donor cell type used, age of recipient, and survival time [28, 36]. A study by Lundberg et al. [37], with β-Gal-expressing RN33B cells labeled with the nuclear marker [3H]thymidine prior to transplantation to adult brain, demonstrated fewer cells that expressed β-Gal compared with [3H]thymidine post-grafting. Taken together, in order for the most accurate estimation of the number of surviving grafts and cell numbers, a marker(s) independent of the reporter gene is required.

Several intracerebral experimental transplantation studies report overall high cell survival rates after grafting of different in vitro expanded neural precursor types to various regions and ages of the central nervous system [911]. Results from retinal transplantation of similar cell types are sparse. Young and coworkers [19] report that at least 50% of the GFP-transduced adult rat hippocampal progenitor cells (AHPCs) transplanted intravitreally to dystrophic rats, aged 1–18 weeks, survived 4 weeks in 80% of the animals. Takahashi et al. [13] also describe good survival after intravitreal transplantation of LacZ-transduced AHPCs to healthy immature (P0–P2) rat retinas. The present study shows a modest survival of transplanted cells compared with the results obtained from transplantation of AHPCs [13, 19].

Instead of an intravitreal grafting technique, we use a subretinal transplantation approach in order to obtain retinal integration of the cells into the adult retina. The surgical method, per se, causes a break of the outer blood-retinal barrier, the retinal pigment epithelium. This break triggers an inflammatory response towards the host retina as well as towards the engrafted precursor cells. From earlier experiments it is documented that transplantation with the RN33B cell line to adult retina only results in surviving cells if immunosuppression is used [17, 22].

Migration of Grafted Cells

An extensive migratory capacity of the RN33B cells has been shown after transplantation to different regions of the neonatal and adult rodent brain [29, 3739]. As we reported recently, when RN33B cells are transplanted to the adult retina, they migrate radially from the subretinal space into the host retina as well as laterally within the retina. In addition, grafted cells are found from one eccentricity to the other in cryoprotected sections of the adult SD rat retinas up to 16 weeks post-transplantation [17]. A similar distribution pattern was observed in adult Royal College of Surgeons rats 6 weeks post-grafting [22]. In the present study, we show on whole-mounted retinas that the grafted cells are distributed over a large area (up to 68% of the entire retinal area) at 8 weeks post-transplantation. This again indicates that RN33B cells migrate to regions distant from the implantation site in the host retina, although a large fraction of the grafted cells were found at or close to the implantation site.

Two types of neural migration patterns for immature cells have been described in the brain: the radial glial and the tangential pattern [40]. The first type uses the columnar architecture of the brain as a scaffold. The second, also described as the chain migration type, is explained as immature neuronal cells climbing on each other forming cell tubules [33]. In the present study, an additional finding on four of the retinas 8 weeks post-grafting were the tubules of cells close to the implantation site, resembling the chain migration formation of immature cells found in the rostral migratory stream of the adult brain [33].

The retina is highly columnar, organized by the radial glial Müller cells [41]. One might speculate that the precursor cells climb on the radial glial cells when entering the retina from the subretinal space, and then within the retina using an unknown mechanism(s) for lateral migration. In the two main neurogenic regions (hippocampus and subventricular zone) in the adult rodent brain, and especially in the subventricular zone and rostral migratory stream, neuroblasts migrating in chain-like structures express specific markers, such as doublecortin, which is a microtubule-associated protein important for neuronal migration [33, 42]. Another marker for migrating neuronal progenitors is collapsing-response-mediated protein 4 (CRMP-4/Tuc-4), a molecule involved in axonal guidance [33, 43]. In the present study, by using GFP/doublecortin or GFP/CRMP-4 double immunohistochemistry we did not observe immunoreactivity to doublecortin or CRMP-4 on the retinal chain migration tubules or in other grafted cells (Anita Blixt Wojciechowski and Karin Warfvinge, unpublished observations).

Differentiation of Grafted Cells

The grafted cells had the capacity to differentiate into complex morphologies within the adult retina. Several studies on brain transplantation of the currently used cells have addressed the issue on the differentiation process, and different grafting paradigms have shown that they differentiate into region-specific neurons, such as hippocampal, cortical pyramidal, and striatum-specific medium-spiny projection neurons [29, 37, 39, 44]. In addition, RN33B cells have been found to extend appropriate axonal projections as revealed by retrograde tracing studies [37, 44]. Moreover, RN33B cells exhibit multipotential differentiation, i.e., into neurons, astrocytes, and oligodendrocytes, when transplanted into the neonatal rat brain [29]. Recent data using the GFP-marker gene and electrophysiological recordings demonstrate that grafted RN33B cells can become functionally integrated within host cortical neural circuits [44].

Our recent results have demonstrated that these cells do not differentiate into retinal cells, judging from the lack of expression of retinal neuronal markers, but into glial cells, mainly oligodendrocytes [16, 17, 22]. Here, GFP-expressing cells at 3 and 8 weeks displayed a wide range of morphologies, including glial- and neuronal-like profiles. The glial-like cells had slender cell bodies with long processes, while the neuronal-like cells were multipolar with branched processes exhibiting multiple spines (Fig. 6A and 6B). Comparison between the retinas 3 and 8 weeks after grafting demonstrated similar types of differentiated cells.


The present study demonstrates that the precursor cell line RN33B transplanted to adult intact retina survives at 3 weeks in 68% and at 8 weeks in 35% of the animals. However, more cells were found on each retina after 8 weeks compared with 3 weeks. The actual number of grafted cells is most likely not demonstrated because of downregulation of the two marker genes soon after grafting. The amount of SV40 large T antigen-expressing cells in the host retina gives an impression of a far greater survival of cells than revealed by the β-Gal and GFP expression. It is also demonstrated that the grafted cells display a widespread migration from the implantation site at longer survival times and that the grafted cells differentiated to neuronal- and glial-like phenotypes.

Still little is known about the factors that influence precursor cell survival, proliferation, and differentiation upon retinal and brain transplantation, therefore multipotent precursor cell transplantation to the retina is a valuable tool in investigating these issues. The possibility to introduce genes ex vivo to modify the precursor cells to produce different factors necessary for the diseased host retina gives an additional perspective. This emerging science will clearly be useful to the obstacles involved in retinal reconstruction.


The Crown Princess Margareta's Committee for the Blind, the Swedish Association of the Visually Impaired, the Foundation Fighting Blindness, the Swedish Science Council (Medicine), Second ONCE International Award for New Technologies for the Blind, and the Lund Faculty of Medicine are gratefully acknowledged for financial support. We thank Katarzyna Said and Karin Arnér for excellent technical assistance.