GDNF Regulates Chicken Rod Photoreceptor Development and Survival in Reaggregated Histotypic Retinal Spheres

Abstract

purpose. To investigate the role of glial-cell-line-derived neurotrophic factor (GDNF) on proliferation, differentiation, and apoptosis of different retinal cell types—in particular, photoreceptor cells.

methods. Reaggregated histotypic spheres, derived from retinal cells of the E6 chicken embryo were used. Under rotation, so-called rosetted spheroids arose by aggregation of dissociated retinal cells, followed by the proliferation, migration, differentiation and programmed cell death of particular cell types. Rosetted spheroids were cultured under serum-reduced conditions, either in the absence or presence of 50 ng/mL GDNF. At appropriate stages, rosetted spheroids were analyzed by using conventional staining and immunolabeling with antibodies against different retinal cell types.

results. At early stages of culture, the application of GDNF to rosetted spheroids significantly increased and sustained the rate of proliferation. In particular, a de novo production of rod photoreceptors was observed, whereas cone photoreceptors and amacrine, horizontal, ganglion, and Müller cells were not affected. In addition, in GDNF-treated cultures, rod photoreceptors differentiated earlier than in nontreated cultures. In older rosetted spheroids raised in absence of GDNF, rod but not cone photoreceptors underwent apoptosis. By supplementation with GDNF, the percentage of dying rod photoreceptors was dramatically reduced (31%–6% at 8 days in culture, 71%–3% at 10 days in culture). Both the mitogenic and survival promoting effect of GDNF were dose dependent.

conclusions. The results strongly suggest that GDNF, at least in vitro, affects rod photoreceptors. Depending on the developmental stage, GDNF regulates their proliferation, differentiation, and survival.

Glial cell line-derived neurotrophic factor (GDNF) is a distant member of the transforming growth factor (TGF)-β superfamily. GDNF and the related polypeptides neurturin (NRTN), artemin (ARTN), and persephin (PSPN) belong to the GDNF family ligands (GFLs). Each of the GFLs has a preference for one of four GPI-linked coreceptors. ^{1 2 3 4} GDNF has the highest affinity for GFRα1, whereas NRTN preferentially binds to GFRα2, ARTN to GFRα3, and PSPN to GFRα4. 

Originally, GDNF as the first known member of the GFLs, was identified as a neurotrophic factor that prevents dopaminergic neurons from cell death. ⁵ Then it became clear that GDNF can also act as a potent trophic factor for developing enteric, sympathetic, parasympathetic, sensory, and motor neurons. ^{6 7 8 9 10 11 12 13 14 15 16 17} In addition, GDNF attracts attention because of its possible therapeutic application for the treatment of various neuronal degenerations, such as Parkinson’s, Alzheimer’s, and Hirschsprung’s diseases. ^{18 19 20 21 22 23}  

However, as in other parts of the nervous system, the role of GDNF in the embryonic retina is not well understood. Recently, it has been shown that GDNF and its receptors GFRα1 and GFRα2 are expressed in an overlapping and specific pattern in the developing chicken retina. ²⁴ GDNF has been found throughout retinogenesis in all retinal cell layers, whereas the expression of GFRα1 and -2 is restricted to particular cell types. GFRα1 has been detected in amacrine and horizontal cells, whereas GFRα2 expression has been observed in amacrine, ganglion, and photoreceptor cells. This spatial expression pattern may explain the results of previous studies that have shown that GDNF can act as a trophic factor for both retinal ganglion cells and photoreceptors and therefore serve as a useful therapeutic tool to restore deficient cell types caused by ophthalmic disorders. ^{25 26 27 28 29 30} Furthermore, treatment of photoreceptor-enriched rat monolayer cultures with GDNF results in an increase of photoreceptor precursors within the first hours in culture, and delays the onset of programmed cell death at a later stage. ³¹ Although several studies have shown an effect of GDNF on photoreceptors, it is not clear whether GDNF acts on all photoreceptors in a general manner or is restricted to either rod or cone photoreceptors. Likewise, little is known about the effect of GDNF on the other retinal cell types, such as bipolar, amacrine, horizontal, and Müller cells. It is now accepted that a series of soluble factors can influence the fate of retinal precursor cells (RPCs). ³² These factors are mostly derived from retinal cells or adjacent tissue in close approximation to uncommitted RPCs. In this way, RPCs receive information from the environment (extrinsic signals), and this in turn determines the fate of specific cell types. However, in addition to extrinsic factors it has been postulated that intrinsic properties of cells also contribute to cell fate decisions. This means that the three-dimensional environment is essential for proper retinal development. 

In our experiments we used reaggregated organotypic retinal spheres, called rosetted spheroids, with the advantage that they imitate a de novo retinal development in three-dimensional in vitro conditions and can be easily manipulated by the addition of GDNF. Under rotation conditions, rosetted spheroids arise through aggregation of dispersed retinal cells of the chicken embryo, followed by the proliferation, migration, differentiation, and programmed cell death of particular cell types. After 2 weeks in culture, mature rosetted spheroids represent composites of morphologic structures homologous to all retinal layers (Fig. 1F ; for detail see Refs. ^{33 34 35} ). First, rosetted areas consist of rosettes, their internal lumen lined with photoreceptors, thus corresponding to the outer nuclear layer (ONL) of a normal retina. Each rosette is surrounded by a circular outer plexiform layer (OPL), followed by sections of an inner nuclear layer (INL). Second, inner plexiform layer (IPL)-like areas resemble an in vivo IPL, because they are mainly cell-free areas consisting of fibers composed of bipolar, ganglion, and amacrine cells. Within these circular IPLs, only a few displaced ganglion cells were detectable. Moreover, there were nonorganized areas consisting of different retinal cell types and their corresponding fibers. 

In the current study, we applied GDNF to rosetted spheroid cultures to investigate its role in proliferation, differentiation, and survival during development of the chicken retina in vitro. GDNF increased and sustained proliferation of retinal cells, promoted the onset of differentiation of rod photoreceptors, increased their number, and effectively prevented programmed cell death. In contrast, it did not induce increases the number of cone photoreceptors or ganglion, amacrine, horizontal, or Müller cells. 

Methods

Tissue Culture

To produce rosetted spheroids, ^{34 36} central parts of the retinas of 6-day-old chicken embryos (white leghorn) were isolated and collected in F12 medium on ice. The tissues were fully dissociated by tryptic digestion in Hanks’ balanced salt solution (HBSS) containing 0.05 mg/mL trypsin (Worthington Biochemicals/Cell Systems, Remagen, Germany) for 5 minutes at 37°C. After two washes in F12 medium, including tissue sedimentation, the remaining cell clusters were mechanically dissociated in F12 medium containing 0.5 mg/mL DNase I (Worthington Biochemicals/Cell Systems) by 30 to 40 gentle strokes with a round-bored Pasteur pipette. For the generation of rosetted spheroids, 2 × 10⁶ cells/mL were cultivated, either in the presence or absence of 50 ng/mL rat GDNF (Sigma, Deisenhofen, Germany) in 35-mm dishes containing 2 mL aggregation medium (DMEM, 2% FCS, 1% l-glutamine, and 0.15% penicillin/streptomycin, all from Gibco, Berlin, Germany) on a gyratory shaker in a Heraeus incubator (37°C, 95% air, 5% CO₂). Fresh medium and GDNF were added on alternate days. 

Cryosections

For cryosections, rosetted spheroids were harvested at appropriate stages and fixed in 4% formaldehyde (Merck, Darmstadt, Germany) for 30 minutes at room temperature. After the fixative was removed and two washes performed in PBS, rosetted spheroids were soaked in 25% sucrose (Merck, Darmstadt, Germany) and stored at 4°C. Frozen sections of 10 to 12 μm thickness were cut on a cryostat (Microm, Heidelberg, Germany), mounted on gelatin-coated slides, and stored for immunostaining at −20°C. For BrdU-incorporation experiments, BrdU was added 16 hours before the spheroids were harvested and fixed as just described. 

Primary Antibodies

The polyclonal antiserum CERN 901 was raised against purified chicken rhodopsin, and the antiserum CERN 906 was raised against purified chicken red and green pigments. The CERN antibodies (a generous gift of Willem DeGrip, University of Nijmegen, Nijmegen, The Netherlands) were used at a dilution of 1:2000 in PBST (PBS, 0.1% Triton-X-100) and were incubated for 1 hour at room temperature. BrdU antibodies (Sigma, Deisenhofen, Germany) were used 1:100, whereas Pax-6 hybridoma supernatant (specific for amacrine, ganglion and horizontal cells in differentiated retinal cells of the chick ^{37 38 39} ; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA) was added undiluted. 

Staining Procedures

For immunostaining, sections were dried at 37°C and preincubated in blocking solution (3% BSA, 0.1% Triton-X-100 in PBS) for 30 minutes at room temperature. The tissues were then incubated with the appropriate primary antibodies, followed by three washes in PBS. For detection of the primary antibodies, either goat-anti-rabbit-Cy2 or donkey anti-mouse-conjugated-Cy3 secondary antibodies were applied for 1 hour at a dilution of 1:100 in PBST. Between the last two washes, cell nuclei were stained with DAPI (0.1 mg/mL 4′,6-diamidine-2-phenylindol-dihydrochloride in PBS). Finally, sections were dried and embedded in Kaiser’s glycerin gelatin (Merck, Darmstadt, Germany). For double-labeling experiments, sections were stained with TUNEL red (according to the manufacturer’s instructions; Roche Molecular Biochemicals, Mannheim, Germany), followed by immunostaining with the primary antibodies described. For the detection of BrdU, we used the same procedures as just described, with the exception that sections were treated first with 4 N HCl for 5 minutes before the BSA-containing blocking solution was added. 

Cell Counting and Statistical Analysis

To determine the number of immuno- and TUNEL-positive cells, frozen sections (each containing 30–40 spheroids) were stained with DAPI and the corresponding antibodies or staining reagents, respectively. The percentage of immunolabeled cells per cryosection of a single spheroid were calculated in relation to DAPI-positive cells of the same spheroid section. At least nine cryosections of different spheroids derived from two individual experiments were analyzed (nine spheroids investigated). This corresponds to counting of 4000 to 5000 cells per stage and staining. Data are presented as the mean ± SD and compared by a two-tailed, paired Student’s t-test. Note that the standard deviation represents the differences between the calculation of individual spheroids. 

Microscopy and Photography

Photomicrographs of sections were taken with a microscope (Axiophot; Carl Zeiss, Jena, Germany) combined with a charge-coupled device, three-color (CCD-3) digital camera (Intas, Göttingen, Germany) and processed on computer (focus imager model 4000; Intas, Photoshop 5.0; Adobe, Mountain View, CA; and Powerpoint, Microsoft, Redmond, WA). 

Results

Effect of GDNF on Proliferation of Retinal Cells at Early Stages of Spheroid Development

Exogenous application of GDNF stimulated the proliferation of retinal cells in rosetted spheroids (Figs. 1A 1B 1C 1D) . The quantitative data obtained at appropriate stages are shown in Figure 1E . GDNF-treated spheroids showed a significantly increased BrdU uptake during the first 6 days in culture. After 2 days in culture, when the rate of proliferation is still very high, 17% more BrdU-positive cells were detected in GDNF-treated spheroids (Figs. 1A 1B 1E) . After 4 days in culture, the rate had significantly dropped, but the effect became much more pronounced (Figs. 1C 1D 1E) ; now, 257% more BrdU-positive cells were detectable in the presence of GDNF. At day 6, GDNF increased the rate of proliferation up to 185%. At later stages (days 8–10) proliferation had become very low and did not differ significantly when compared with control spheroids (Fig. 1E) . 

Effect of GDNF on the Number of Rod Photoreceptors and the Onset of Differentiation

Analyzing the course of spheroid development, a continuously increasing number of rods was revealed within the first 8 days of culture in both treated and nontreated spheroids (Fig. 2A , cf. Figs. 4A 4B ). However, at all stages the number of rods was significantly increased in the presence of GDNF (Fig. 2A) . Remarkably, from days 8 to 10, the number of rod photoreceptors in nontreated spheroids decreased by 41%. A striking opposite effect was detected in GDNF-treated spheroids. Here, the number of rods was further increased from days 8 to 10, resulting in a dramatic increase of rod photoreceptors, when compared with control spheroids at stage 10 (Fig. 2A ; green cells in Figs. 4A 4B ). GDNF affected rod photoreceptor differentiation at a very early stage of culture. In GDNF-treated spheroids, a small number of rods was detectable as early as 2 days (Figs. 3A 3B) , whereas at this time in nontreated spheroids, rod photoreceptors were not yet detectable (Figs. 3C 3D) . It is remarkable that these early rods achieved an advanced state of maturation, as indicated by opsin expression on their cell surfaces, extending into processes (Fig. 3A , inset). 

Protection of Rod Photoreceptors from Programmed Cell Death

In the presence of GDNF, the number of rod photoreceptors did not decrease from days 8 to 10 (cf. Fig. 2A ), as observed in the absence of GDNF, indicating a specific effect of GDNF on survival of rods. Therefore, we used either a rod/TUNEL or a cone/TUNEL double-labeling technique in combination with DAPI staining to visualize apoptotic photoreceptors. Rod (Figs. 4A 4B) and cone (Figs. 4C 4D) photoreceptors appear green, apoptotic cells red, and apoptotic photoreceptors yellow in Figure 4 , because opsins and defragmented DNA are colocalized in the same cell. Cone photoreceptors did not undergo apoptosis, regardless of whether spheroids were treated with or without GDNF (Figs. 4C 4D) . The situation was dramatically different for rod photoreceptors (Figs. 4A 4B) . After 10 days in culture in the absence of GDNF, there was a significant number of apoptotic rods (Fig. 4A , yellow), which, remarkably, in GDNF-treated spheroids was dramatically decreased (Fig. 4B) . 

For quantification of apoptotic rod photoreceptors, we determined the number of rod/TUNEL-positive cells in both GDNF and nontreated cultures in relation to the total number of rod photoreceptors (Fig. 4E) . The first apoptotic rods were detected at 7 days in culture (data not shown). After 8 days in culture, 31% of all rods were apoptotic in the absence of GDNF. In contrast, only 6.3% of apoptotic rods occurred in GDNF-treated cultures. Two days later (day 10), apoptotic rod photoreceptors treated with GDNF decreased from 71% to negligible 2.8%. Moreover, in GDNF-treated spheroids, we detected a small fraction of apoptotic cells that belonged to cell types other than photoreceptors (Figs. 4B 4C 4D , arrowheads). Counterstaining with the Pax-6 antibody identified these cells predominantly as amacrine and ganglion cells. The quantification of apoptotic amacrine, ganglion, and horizontal cells by Pax-6/TUNEL double labeling showed that only one fifth of all Pax-6-positive cells were apoptotic (Figs. 5E 5F) ; more important, GDNF did not decrease programmed cell death in the populations of amacrine, ganglion, and horizontal cells (Fig. 5F) . 

Concentration Dependent Effect on Proliferation of Retinal Cells and on Survival of Rod Photoreceptors

To examine whether the observed increase in proliferation and survival of rod photoreceptors was specifically induced by GDNF, spheroids were grown in the presence of 0, 0.5, 5, 25, 50, and 100 ng/mL GDNF. As shown in Figure 4F , both the increase in the rate of proliferating cells at stage 4 (diamonds) and the decrease in the number of apoptotic rod photoreceptors at stage 10 (squares) followed a dose-response curve characteristic of growth factors and neurotrophins. In both cases, the effect on proliferation and on survival was enhanced with increasing concentrations of GDNF, reaching saturation at approximately 50 ng/mL. The EC₅₀s of GDNF for proliferation and for survival of rods both amounted to 2.5 ng/mL. 

Effect of GDNF on Proliferation of Various Cell Types

To investigate whether the proliferative effect of GDNF is exclusively restricted to the rod cell lineage, we used several specific antibodies to quantify the number of cone photoreceptors, amacrine cells, retinal ganglion cells, horizontal cells, and Müller cells in treated and nontreated spheroids. In contrast to rod photoreceptors, no significant elevation of cone photoreceptors was achieved in GDNF-treated spheroids (Figs. 2A 2B) , and the temporal expression of cone photoreceptors resembled that of rod photoreceptors; there was no decrease in cones after day 8. Apart from differences in number of cells, the local distribution of rod and cone photoreceptors was identical within treated and nontreated spheroids (compare green cells in Figs. 4A 4B with Figs. 4C 4D ). Most of the photoreceptors were localized in nonorganized areas (arrows); only a smaller number occurred in rosettes (stars). 

In comparison with DAPI staining, most Pax-6-positive cells were detected in INLs surrounding a circular IPL (Figs. 5A 5C) , whereas only a few Pax-6-positive cells were detectable within the IPL (Figs. 5A 5C , arrows) and nonorganized areas (Figs. 5A 5C , arrowheads). The number of Pax-6-positive cells was nearly identical in nontreated and treated spheroids. At days 8 and 10 of culture, we found no quantitative differences in the number of Pax-6-positive cells between treated and nontreated spheres (Fig. 5E) . Similarly, based on glutamine synthetase staining, the number of Müller cells did not change after application of GDNF (data not shown). 

Discussion

This study demonstrates for the first time that GDNF has multiple effects on rod photoreceptors during in vitro development of the chicken retina. Depending on the developmental stage, GDNF affected proliferation, onset of differentiation, and survival of rod photoreceptors. 

Effect of GDNF on Proliferation and Onset of Rod Photoreceptor Differentiation

In a series of previously published reports, a proliferative effect of GDNF has been described in nonretinal tissue. In particular, in the enteric nervous system, it has been shown that GDNF is essential for the proliferation of enteric precursorcells. ^{11 12 13 15 40 41} Similarly, GDNF has been found to stimulate proliferation during development of kidneys in vivo and in vitro. ^{42 43 44} Beyond this, a regulatory mitogenic effect has been demonstrated for rat glioma cells by adding exogenous GDNF ⁴⁵ or by using antisense oligonucleotides for the suppression of endogenous GDNF. ⁴⁶ A proliferative effect of GDNF has been established in photoreceptor-enriched rat monolayer cultures. ³¹ Our data profoundly extend these findings. We showed a specific function of GDNF on rods within a histotypic three-dimensional tissue context consisting of all retinal cell types. GDNF not only increased proliferation but also sustained the phase of proliferation up to day 6 in culture. Moreover, GDNF increased proliferation in a dose-dependent fashion, reflecting the specificity of GDNF as a mitogenic factor. The expanded period of proliferation resulted mainly in an increased number of rod photoreceptors, whereas other retinal cell types like cones and Müller, amacrine, ganglion, and horizontal cells did not respond to GDNF. This indicates that rod photoreceptor precursors are positively affected by GDNF, because at a very early stage of culture (up to day 6), the number of differentiated rod photoreceptors was always higher than in control spheroids. It is important to note that the increasing number of rods at early stages was not caused by the survival-promoting effect of GDNF (discussed later), because photoreceptor cell death did not occur until day 7 of culture. Therefore, we postulate that GDNF either increases the number of rod precursors, or it promotes the proliferation of uncommitted precursors that will predominantly differentiate into the rod phenotype, provided GDNF is available in sufficient concentration. 

In this context, it is remarkable that the development of cones is probably not affected by GDNF. At each stage of culture the number of cones is nearly identical, regardless of whether GDNF was added or not. Ciliary neurotrophic factor (CNTF) also increased the number of rod precursors in chick retina monolayer cultures, whereas cones were unaffected. ⁴⁷ We have shown that the number of rod photoreceptors depends on microenvironmental changes—in particular, rods developed in spatial proximity to preexisting cones. Their development, in turn, was autonomous. ⁴⁸ These data support the idea that the development of rods in contrast to cones is dependent on various extrinsic signals, one of these factors presumably being GDNF. An alternative, yet at the most only partial, explanation for the missing effect of GDNF on cone photoreceptors could be related to the age of the retinal tissue (embryonic day 6) that is used for the production of rosetted spheroids. In contrast to rods which appeared late during retinogenesis, a proportion of cones were already postmitotic by embryonic day 6. Therefore, such cone precursors may be incapable of responding to GDNF at that point. Further investigations with rosetted spheroids derived from retinal tissues younger than embryonic day 6 are needed. 

Furthermore, our study showed that GDNF promotes the onset of rhodopsin expression in spheroids within 2 days. Even though they appeared early in spheroid development, these rods expressed high levels of opsin, therefore representing an advanced stage of photoreceptor differentiation. This means that the action of GDNF was restricted to rod photoreceptors, and it not only induced the proliferation of this cell type, it also accelerated its differentiation. 

Prevention of Apoptosis in Rod Photoreceptors at Late Stages of Retinal Development In Vitro

Programmed cell death is a typical feature of retinal development, occurring in different retinal cell types of the chicken embryo but not in photoreceptors. ⁴⁹ To induce cell death of photoreceptors, we produced spheroids under reduced serum conditions. Although the concentration of serum was low, spheroids showed an intact and well-organized morphology. This means that apoptosis in spheroids occurred, but took place in a proper cellular environment and was not induced by the necrotic processes. Therefore, the use of spheroids represents a suitable culture system to investigate the survival effect of GDNF on different retinal cell types, particularly rod photoreceptors. A survival-promoting effect of GDNF has been reported in certain cell types of the retina. ^{25 26 27 28 30 50} Photoreceptors and ganglion cells seem to be especially sensitive to GDNF. In this context, it has been shown that GDNF increased the survival time of rod outer segments in vitro. ²⁶ Moreover, subretinal injection of GDNF into the eyes of rd/rd mice prevents photoreceptors from cell death. ²⁸ In organ cultures of rd mice, GDNF alone is unable to prevent photoreceptor cell death, but in combination with CNTF, a partial rescue of photoreceptors has been observed. ³⁰ Therefore, the authors speculated that GDNF and other growth factors act synergistically rather than individually. In contrast to this, it has been shown that in rat retinal cultures GDNF alone is able to increase the survival of rods, but in combination with docosahexaenoic acid, this effect is dramatically enhanced. ³¹ Nevertheless, our study clearly showed that GDNF can effectively protect rod photoreceptors from cell death without addition of any substance. In the presence of GDNF only 2.8% of all rods were apoptotic, whereas in untreated cultures 71% of rods underwent programmed cell death after 10 days in culture. This means that GDNF prevented cell death of rod photoreceptors by a factor of 25. Therefore, GDNF could become a potential therapeutic tool for the treatment of a series of retinal degenerations that are primarily characterized by the loss of rod photoreceptors. ⁵¹  

Moreover, we found that the survival-promoting effect of GDNF acted in a dose-dependent manner, reflecting the specific action of GDNF on the survival of rods. Our results also showed that cone photoreceptors did not undergo apoptosis under serum-reduced conditions. This strongly suggests that the survival of rod and cone photoreceptors is regulated by two independent mechanisms. A GDNF-dependent mechanism rescues rods from programmed cell death, whereas the survival of cone photoreceptors is GDNF independent, or probably does not require any additional signal for survival. In contrast to cones, the populations of retinal ganglion and amacrine cells undergo programmed cell death during spheroid development, and their loss is not counteracted by GDNF. This observation is surprising, because GFRα1 is expressed in ganglion cells and GFRα2 is expressed in both amacrine and ganglion cells in the chicken retina. ²⁴ Therefore, we postulate that GFRα/GDNF signaling is not responsible for survival of these retinal cell types, or alternatively, additional factors are needed to rescue them from programmed cell death. In striking contrast, the survival of rod photoreceptors was positively affected, but GFRα1, which preferentially binds GDNF, is not expressed in the ONL. Instead, only GFRα2 is exclusively expressed in the photoreceptor layer. ²⁴ This indicates that at least in the chicken retina, survival of rod photoreceptors is probably regulated through the interaction of GDNF and GFRα2, but not by the GFRα1/GDNF signaling pathway. 

In conclusion, in our study GDNF showed different functions during retinal development in vitro: It acted as a mitogenic factor for rod photoreceptors at early stages; influenced the onset of differentiation of rod photoreceptors; and supported photoreceptor survival at later stages. Moreover, in contrast to rods, a de novo production of other retinal cell types was not stimulated by GDNF. For further understanding of the role of GDNF in rod photoreceptor regulation, we have begun to investigate the temporal and spatial expression of GDNF and its receptors during retinal development. 

 Supported by German Research Foundation Grants SFB 269/B3 and La 379/12, Human Capital and Mobility, and German Israeli Foundation Grant 512/96.

 Submitted for publication September 9, 2002; revised November 6 and November 21, 2002; accepted November 26, 2002.

 Disclosure: A. Rothermel, None; P.G. Layer, 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.

 

Figure 1.

 

GDNF effected an increase in the number of proliferating cells. Cryosections of nontreated (A, C) and GDNF-treated (B, D) rosetted spheroids were stained with anti-BrdU antibodies to detect proliferating cells. After 2 (A, B) and 4 (C, D) days in culture, a significant increase of BrdU-positive cells was detectable in rosetted spheroids grown in the presence of GDNF. (E) Quantification of proliferating cells in rosetted spheroids. GDNF treatment increased proliferation during the first 6 days in culture when compared with nontreated spheroids. (F) Structure of reaggregated rosetted spheroids, consisting of rosettes, IPL-like areas and nonorganized areas. MCs, Müller cells; PRs, photoreceptors. (E) The percentage of BrdU-positive cells per cryosection of a single spheroid was calculated in relation to DAPI-positive cells of the same spheroid section. Each data point represents the mean ± SD of multiple spheroid cryosections (n = 9). *P < 0.01, **P < 0.001. Scale bar in (D): (A, B) 50 μm; (C, D) 100 μm.

Figure 2.

 

Development of rod (A) and cone (B) photoreceptors in GDNF-treated and nontreated rosetted spheroids. For quantification, the percentages of immunolabeled rod and cone photoreceptor per spheroid section were calculated in relation to DAPI-positive cells. In GDNF-treated cultures, rods developed 2 days earlier than in control cultures (cf. Fig. 3 ). Then, at each stage of culture, the number of rods was higher when compared with growth in absence of GDNF. From 8 to 10 days, the number of rods significantly decreased in the absence of GDNF. Note that the number of cones was not affected by the application of GDNF. Each data point represents the mean ± SD of multiple spheroid sections (n = 9). * P < 0.01, ** P < 0.001.

Figure 3.

 

GDNF promoted the onset of photoreceptor differentiation. Cryosections of 2-day-old spheroids were double stained with CERN901 (A, C) and DAPI (B, D). In GDNF-treated spheroids (A, B) the first rods appeared after 2 days in culture, whereas in control cultures (C, D) opsin expression was not yet detectable. Prominent opsin expression (A, inset) indicates the maturing state of these early photoreceptors. Scale bar, 50 μm.

Figure 4.

 

GDNF treatment promoted survival of rod photoreceptors in rosetted spheroids. Cryosections of 10-day-old rosetted spheroids were double stained, either with TUNEL (red) and the rod-specific antibody CERN 901 (A, B, green), or with TUNEL and the cone-specific antibody CERN 906 (C, D, green). Most of the photoreceptors were localized in nonorganized areas (arrows), with only a small number detected in rosettes (∗). (A) In nontreated spheroids, the number of apoptotic rods (yellow, colocalization of TUNEL and rod staining) was much higher than in cultures treated with GDNF (B). Cone photoreceptors in both non-(C) and GDNF-treated spheroids (D) showed many fewer TUNEL-positive cells. Note that, in addition to rods, other retinal cell types also underwent apoptosis (red). (E) The percentage of apoptotic rods was quantified by the rod/TUNEL double-labeling technique in relation to the total number of rods per spheroid section. At days 8 and 10 in culture, the ratio of TUNEL-positive rods was dramatically reduced in the presence of GDNF. (F) Rosetted spheroids were grown in the presence of 0, 0.5, 5, 25, 50, and 100 ng/mL GDNF and were harvested either after 4 days (for calculation of BrdU-positive cells) or after 10 days (for quantification of apoptotic rods) in culture. GDNF acted in a dose-dependent manner on proliferation (EC₅₀ = 2.5 ng/mL) and survival of rod photoreceptors (EC₅₀ = 2.5 ng/mL). Each data point represents the mean ± SD of multiple spheroids sections (n = 9).

Figure 5.

 

The de novo production and survival of Pax-6-positive cells were not affected by GDNF. After 10 days in culture, cryosections of GDNF-treated (A, B) and nontreated (C, D) spheroids were double stained with the Pax-6 antibody (red) and DAPI (blue). Pax-6-positive cells surrounded IPL-like areas and rosettes (R). Only a few immunolabeled cells were localized within the IPLs (A, C, arrows) and in nonorganized areas (A, C, arrowheads). (E) Quantification of Pax-6-positive cells in relation to DAPI-positive cells revealed that in each investigated spheroid section the number of Pax-6-positive cells after 8 and 10 days in culture was not altered. (F) Quantification of TUNEL/Pax-6-positive cells in cryosections of rosetted spheroids grown either in the presence or absence of GDNF. Survival of Pax-6-positive cells was not affected by GDNF. (E, F) Each data point represents the mean ± SD of multiple spheroid sections (n = 9). Scale bar, 100 μm.

The authors thank Elmar Willbold and Hans-Dieter Hofmann for helpful discussions, Kathryn Nixdorff for reading of the manuscript and making suggestions, Jutta Huhn-Smidek and Meike Stotz-Reimers for expert technical assistance, and Katja Volpert, Vanessa Jacob, and Sandra Schlosser for valuable assistance.