Ciliary Neurotrophic Factor as a Transient Negative Regulator of Rod Development in Rat Retina

Abstract

purpose. Ciliary neurotrophic factor (CNTF) has been shown to inhibit the developmental expression of rod differentiation markers in rat retinal cultures. The present study was undertaken to investigate whether CNTF transiently inhibits rod differentiation or induces irreversible changes in the developmental fate of rod precursors.

methods. The effects of CNTF on rod differentiation were monitored in organotypic slice cultures from early postnatal rat retinas by quantification of opsin-immunoreactive cells. The in vitro formation of the photoreceptor layer was analyzed by light and electron microscopy. The developmental expression of the CNTF receptor in photoreceptors was determined by immunoblot analysis.

results. CNTF did not interfere with the generation of rod precursors, their morphologic differentiation, or the formation of the outer nuclear layer. Inhibition of rod differentiation was reversible. In the continuous presence of CNTF the number of opsin-positive cells increased at a normal rate but with a delay of 3 to 4 days. Developing rods became resistant to CNTF, both in vivo and in vitro, and this correlated temporally with the downregulation of CNTF receptor expression. Receptor downregulation was inhibited by CNTF in a dose-dependent manner. At higher CNTF concentrations with sustained receptor expression, the CNTF-induced decrease in opsin expression was accompanied by an increase in the expression of a bipolar cell marker in rod precursors located in the photoreceptor layer.

conclusions. These data indicate that, although rod precursors exhibit some phenotypic plasticity in the presence of CNTF, the factor does not induce a switch in the developmental fate of rod precursors but plays a role as a transient and reversible negative regulator of rod differentiation.

Several lines of evidence suggest that the determination of progenitor cell fate and the subsequent differentiation of neuronal phenotypes in the vertebrate retina are controlled by developmentally regulated extrinsic signals. In vitro studies focusing on rod photoreceptors, which constitute more than 70% of all cells in the mature rodent retina, ¹ have demonstrated that the differentiation of these cells depends on influences in their microenvironment. ^{2 3} In vitro, a variety of molecules have been identified that influence the generation and differentiation of rods. ^{4 5} Although the steps by which terminally differentiated photoreceptors develop from neuroepithelial precursors are still unknown, there is evidence for a sequence of developmental events, each of which depends on the action of appropriate environmental signals. 

Ciliary neurotrophic factor (CNTF) is one of the molecules that have been implicated in the regulation of photoreceptor development. The effects of CNTF are mediated by a tripartite receptor complex, ⁶ consisting of two signal transducing subunits (gp130, leukemia inhibitory factor [LIF] receptor β) and a CNTF-specific, ligand-binding α-component (CNTFRα). In the retina, CNTF promotes the survival and differentiation of various neuronal cell types, ^{7 8 9} and this diversity of biological effects correlates with the differentially regulated expression of CNTFRα in the developing retina. ^{10 11 12} Several studies have shown that CNTF effectively blocks the production of differentiated opsin-positive rods in rat retinal cultures. ^{13 14 15 16} However, conclusions concerning the functional implications of this atypical inhibitory effect of the neurotrophic factor on rodent photoreceptors are controversial. In explant cultures, the inhibition of opsin expression was accompanied by an increase in the number of cells expressing bipolar cell markers, raising the interesting possibility that CNTF may induce a phenotypic switch in progenitor cells. ^{15 17} In contrast, results from studies in monolayer cultures prompted the conclusion that CNTF does not influence the phenotypic choice but retards the terminal differentiation of rods. ^{14 16} In this way, CNTF may contribute to the temporal coordination of photoreceptor development, which is characterized by a delay of several days between terminal mitosis of progenitors and the appearance of differentiation markers. ¹⁸ This seemed to be supported by the observation that cultures from newborn mice without CNTFRα exhibit enhanced production of rods at early postnatal stages without concomitant changes in the number of bipolar cells. ¹⁵  

To distinguish between the two alternatives and to further clarify the potential role of CNTF during photoreceptor development, we examined the effects of this protein on photoreceptor development in organotypic slice cultures from newborn rat retina. In this in vitro model the laminar structure of the retina is largely preserved, allowing the study of the influence of CNTF on the formation of the photoreceptor layer and the phenotypic differentiation of rods from proliferating neuroepithelial cells that develop in an environment resembling the in vivo situation. 

Materials and Methods

Organotypic Cultures

Cultures were prepared from retinas of Sprague-Dawley rats, which were killed by decapitation during the first 24 hours after birth (postnatal day [PD]0). Eyes were enucleated and collected in ice-cold Hanks’ buffered saline solution (HBSS; Gibco, Eggenstein, Germany). A central piece of the retina (approximately 5 mm² in size) was separated from the pigment epithelium and sclera, and blood vessels were carefully removed from the retinal surface. The tissue was transferred to a Teflon disc and mounted with the ganglion cells pointing upward. After 100-μm-thick vertical sections were cut on a McIlwain tissue chopper (Vibratome, St. Louis, MO), the retinal slices were rinsed off the disc into a Petri dish containing HBSS and 10% horse serum and kept at room temperature for 15 to 30 minutes. Slices from several retinas were pooled. Intact, even slices were selected for culturing. 

Retinal slices were cultured by a method adapted from Stoppini et al., ¹⁹ either on the membranes of inserts for 6- or 12-well culture plates (0.4 μm pore size; Falcon; BD Biosciences, Heidelberg, Germany) or on glass coverslips (14 mm in diameter) placed in 24-well culture plates. Membranes and glass coverslips had been sequentially pretreated with poly-l-lysine (0.1 mg/mL) and laminin (4 μg/mL). Culture medium contained 73 mL Dulbecco’s modified Eagle’s medium (DMEM; Gibco), 10 mL heat-inactivated horse serum (Sigma, Diesenhofen, Germany), 22.5 mL H₂O, 1 mL HEPES (1 M; pH 7.2), 2.5 mL HBSS, 1 mL glutamine (200 mM), 100 U/mL penicillin, and 100 U/mL streptomycin. Routinely, every third day 50% of the culture medium was replaced. Recombinant rat CNTF (Peprotech, London, UK) was added at a concentration of 5 ng/mL at the beginning of the experiments and with each medium change. In addition, fresh factor was added every second day at a concentration of 2.5 ng/mL CNTF. AADH-CNTF was used to block endogenous CNTF activity. AADH-CNTF is a recombinant protein variant with four amino acid substitutions (F152A, K155A, S166D, and Q167H) that acts as a potent competitive CNTF receptor antagonist. ²⁰ It was added to the culture medium at a concentration of 50 μg/mL at the beginning of the experiments and with each medium change. This concentration is sufficient to block the effect of 50 ng/mL or more of CNTF, as determined by choline acetyltransferase–stimulating activity in IMR-32 neuroblastoma cells. ²⁰  

Immunocytochemistry

Organotypic cultures were fixed in 4% paraformaldehyde dissolved in 0.1 M phosphate buffer (PB; pH 7.35) for 20 to 30 minutes. The tissue was washed twice and then treated for 60 minutes with PB containing 1.0% Triton X-100. Incubation with primary antibodies was performed overnight at 4°C using the following dilutions: mouse monoclonal antibody to bovine rhodopsin, 1:250 (rho-4D2; kindly provided by Robert Molday, University of British Columbia, Vancouver ²¹ ); rabbit antibody to recoverin, 1:3000 (kindly provided by Ann Milam, University of Pennsylvania, Philadelphia, PA ²² ); mouse monoclonal antibody to protein kinase C, ²³ 1:500 (clone MC5; Sigma); mouse monoclonal antibody 115A10, which specifically stains bipolar cells in vertebrate retinas, 1:200 (the generous gift of Shinobu C. Fujita, Mitsubishi Kasei Institute of Life Sciences, Tokyo ²⁴ ). Antibody binding was visualized with the appropriate biotinylated secondary antibodies, avidin peroxidase (BioTrend, Köln, Germany), and diaminobenzidine as substrate. 

Quantification of Retinal Cell Types

Cell numbers in organotypic cultures were determined after dissociation of the tissue into single cells. For this, the medium was removed from the culture dish, and the cultures were incubated in Ca²⁺/Mg²⁺-free HBSS (CMF) for 15 minutes. They were then treated with 0.25%-trypsin dissolved in CMF for 15 minutes at 37°C. The enzyme was inactivated by the addition of medium containing 20% horse serum, and the detached slices were transferred to a tube. The tissue was washed twice, and the slices were dissociated into single cells by gentle trituration with a flame-narrowed glass pipette. The dissociated cells were seeded at a density of 50.000 cells/cm² on poly-l-lysine/laminin-coated coverslips placed in 48-well culture plates. They were incubated for 2 to 4 hours, fixed with 4% paraformaldehyde after attachment to the substrate and subjected to immunocytochemical staining for cell type–specific markers, as described earlier. For quantification, at least 500 cells per coverslip in randomly selected fields were evaluated, and each determination was performed at least in triplicate. 

For comparison with in vivo development, retinas of the appropriate age were dissected and dissociated into single cells, as described previously. ¹¹ Immunocytochemistry and quantification were performed exactly as detailed earlier for the organotypic cultures. 

Immunoblot Analysis

Expression of CNTFRα protein in vivo and in culture was analyzed by immunoblot analysis. Retinas were carefully dissected from rats of different ages (5, 8, and 13 days and 6 weeks) which had been killed by exposure to CO₂. For separate determination of CNTFRα protein in the outer retina (OR; photoreceptor layer) and inner retina (IR; all other retinal cell layers), the isolated retinas were cut horizontally according to a recently described method ²⁵ with minor modifications. Briefly, retinas were flatmounted in 4% gelatin on a 20% gelatin block with the vitreal surface up. Two horizontal cuts were made on a vibratome: one (140 μm thick at PD5 and PD8 and 130 μm thick at later stages) to isolate the IR and a second (150 μm) to obtain the OR. Correct sectioning was controlled by inspection of vertical sections that were cut from the two horizontally separated retinal compartments and stained with cresyl violet or by opsin immunocytochemistry. Retinal tissues were homogenized in lysis buffer (10% glycerol, 20 mM HEPES [pH 7.9], 10 mM KCl, 400 mM NaCl, 0.2% Nonidet P-40, 2 mM dithiothreitol, and protease inhibitors). Equal amounts of protein from each sample were separated by SDS-PAGE and blotted on PVDF membranes (Millipore, Eschborn, Germany). Rabbit anti-CNTFRα antibodies used for immunodetection were the generous gift of Ralph Laufer and Isabella Saggio (IRBM, Rome, Italy). Signals were visualized by using alkaline phosphatase-conjugated goat anti-rabbit IgG (1:10,000; PE-Applied Biosystems, Weiterstadt, Germany) as secondary antibodies and a chemiluminescence kit (CDP-Star; Tropix, Inc., Bedford, MA.) and recorded on autoradiograph film (XOMAT/LS; Eastman Kodak, Rochester, NY). Slice cultures grown for the indicated times were washed three times with warm HBSS and then mechanically detached from the membranes in a small volume (100 μL) of lysis buffer. Further treatment was exactly as described for freshly dissected retinas. 

Scanning Electron Microscopy

Retinal slices obtained from the same piece of rat central retina were cultured on glass coverslips, with or without CNTF (2.5 ng/mL) for 4 days. The coverslips were immersed in 4% paraformaldehyde and 2% glutaraldehyde in PB for 30 minutes. After dehydration with acetone, the tissue was subjected to critical-point drying, mounted for spattering with gold, and studied with a scanning electron microscope (JEOL, London, UK). For comparison with in vivo development, retinas were isolated from rats of the appropriate age, as described earlier. The central part of the retina was carefully dissected from the outer ocular sheaths and pigment epithelium to expose the immature photoreceptor layer. The tissue was immersed in fixative (4% formaldehyde and 2% glutaraldehyde) and processed for scanning electron microscopy (SEM). 

For quantitative evaluation of photoreceptor density by SEM, photomicrographs showing the outer aspect of the photoreceptor layer were made at ×11,000 (see Fig. 3 ). Immature photoreceptors were clearly identifiable by a sclerally oriented, 1- to 3-μm-long process that represents their rudimentary outer segments, as has been shown in organotypic retinal cultures. ²⁶ These processes were counted in three randomly chosen 100-μm² areas of each specimen. Three specimens were evaluated for each experimental condition, and the mean density ± SD was calculated from the nine density measurements obtained in this way. 

Results

Retinal slices from newborn or early postnatal rat retina growing under appropriate conditions largely retained their laminar structure, in particular in the inner nuclear and photoreceptor layers (Fig. 1) . However, there was one remarkable difference compared with the in vivo situation. Within hours after being placed onto membranes or coverslips, the slices underwent a rapid rearrangement that resulted in a horizontal doubling of the retinal layers (Fig. 1A) . This reorganization, which may be related to the formation of rosettes in other types of retinal cultures, ²⁷ did not interfere with the formation of an ordered outer nuclear layer (Fig. 1) . In cultures grown for 1 day from newborn animals, only a few relatively small cells located at the outer margin of the unlaminated neuroblastic cell layer were identifiable as rod photoreceptors, based on their opsin immunoreactivity (Fig. 1A) . Starting at 3 to 4 days in vitro (DIV) opsin-positive photoreceptors markedly increased in number, density, and size (Figs. 1B 1D) . Photoreceptor cells were quantified after dissociation of the cultured retinal slices and immunocytochemical staining of the dissociated cells with antibodies to opsin. The proportion of opsin-positive cells was 4% of all cells at 3 DIV, approximately 30% at 7 DIV, and reaching a maximum of 70% to 75% at approximately 12 DIV. This time course of rod differentiation very closely correlated with the in vivo situation, as determined by the same method. ^{18 28}  

After 1 week in culture, photoreceptors and their proximal processes formed a distinct border adjacent to the outer plexiform layer (Fig. 1C , arrows) that contained the processes of horizontal and bipolar cells (not shown). Short, rudimentary outer segments protruded distally between the two ONLs (Fig. 1C , arrowheads). These observations show that the cellular and laminar differentiation of the ONL in vitro proceeded similar to the in vivo situation. 

When the organotypic cultures were grown in the presence of CNTF (5 ng/mL), the number of opsin-positive cells in the ONL was strongly reduced, compared with control cultures (Figs. 1D 1E) . However, the localization and the orientation of the remaining immunoreactive rods indicated that the ONL had formed normally. Quantitative evaluation showed that the number of opsin-expressing cells at 7 DIV was reduced by approximately 75% in CNTF-treated cultures (Fig. 1F) . Thus, as in monolayer cultures, CNTF effectively inhibited the generation of opsin-positive cells during the first week in vitro. 

To study the in vitro differentiation of rods in the ONL and the effect of CNTF more closely, we took advantage of the fact that the photoreceptor layer could be directly inspected from its scleral aspect in the area between the two ONLs forming in the slice cultures (Figs. 1B 1C) . This is illustrated on a light microscopic level in Figures 2A and 2B , after immunocytochemical staining for opsin. After 4 DIV, the developing ONL in control cultures appeared as a densely packed cellular layer and already contained many opsin-positive photoreceptors with short, immunoreactive apical processes. The ONL of CNTF-treated cultures looked similar, except that only a small number of cells exhibited opsin immunoreactivity. When inspected by SEM, the ONL in cultures grown in the presence of CNTF was indistinguishable from that in the control, in the density and morphology of immature rods identified by their rudimentary outer segments (Figs. 2C 2D) . The in vivo development of the ONL between PD0 and PD4, resulting in a marked increase in the density of identifiable rods, was clearly demonstrated by SEM (Figs. 2E 2F) and confirmed quantitatively by determining the density of outer segment–like processes (Fig. 3) . Comparison of SEM micrographs of the P4 retina and slice cultures of the same age, in combination with the quantitative evaluation (Fig. 3) , showed that the development of the ONL proceeded very similarly in vivo and in vitro. Treatment of the cultures with CNTF had no effect on rod morphology and cell density (Fig. 2) in the ONL. 

Thus, exogenous CNTF did not influence the in vitro formation of the outer nuclear layer, but prevented the terminal differentiation of photoreceptors indicated by the expression of opsin. These results support the hypothesis that, in vivo, CNTF may function to regulate the time course of rod differentiation. ¹⁴ Further evidence for such a function was obtained by experiments that demonstrated that the inhibitory effect of CNTF in the organotypic cultures was reversible and transient. First, we examined the generation of opsin-positive rods in CNTF-treated cultures after withdrawal of the factor (Fig. 4A) . When CNTF was removed from the culture medium after 6 or 8 days of continuous treatment, a marked increase occurred after a reproducible lag phase of 2 days. The rate at which opsin-expressing rods were generated after withdrawal of the factor was very similar to the maximal rate observed in control cultures showing that the action of exogenous CNTF was completely reversible. 

There are two possible ways in which the action of CNTF could be terminated during in vivo development to permit the further differentiation of precursor cells. Either the availability of the factor or the responsiveness of the developing rods toward CNTF could decrease with postnatal age. The first possibility seems to be unlikely, because CNTF expression has been shown to increase during early postnatal development of the rat retina. ¹¹ Testing the second possibility, we prepared retinal slice cultures from older animals and cultured them in the presence or absence of CNTF (Fig. 4B) . Under control conditions, cultures from PD3 and PD6 retinas showed the expected increase in opsin expression from a basal level characteristic of their developmental stage. In the presence of CNTF, the number of opsin-positive cells remained almost constant in PD3 cultures and increased with a rate not much different from the control in PD6 cultures. Obviously, CNTF could not suppress opsin production in those cells that had already reached an opsin-positive state in vivo. Rather, even opsin-negative precursors seemed to become CNTF-insensitive at later stages, as indicated by the significant increase in opsin expression in PD6 cultures, despite the presence of CNTF. 

Corresponding experiments were performed in cultures from PD0 retinas to study the CNTF responsiveness of opsin-positive rods that had differentiated in vitro (Fig. 4C) . When CNTF was added to PD0 control cultures after 3 or 6 DIV, the further generation of opsin-positive rods was inhibited; but, again, the number of immunoreactive cells did not decrease below the level reached before the addition of CNTF. This indicated that in vitro differentiating rods also lost their sensitivity to the opsin-suppressing action of CNTF. 

Having demonstrated that developing rods become resistant to CNTF, both in vivo and in organotypic cultures under control conditions, we asked whether inhibition of opsin expression is also transient in the presence of exogenous CNTF. This was investigated by analyzing the time course of opsin expression in CNTF-treated cultures (5 ng/mL CNTF) for up to 14 days, which turned out to be the maximum culture period we could use (Fig 5A) . Thereafter, the cultured retinal slices started to disorganize, resulting in inconsistent findings, due to varying cellular composition. Under these conditions, the number of opsin-positive rods started to increase significantly after 8 to 10 DIV and reached almost 50% of all cells after 14 DIV. As shown in Figure 5A , opsin was generated in the presence of exogenous CNTF at the same rate as in control cultures, but with a delay of 4 days. A similar result was obtained when we studied the expression of recoverin, another marker for differentiated rods (Fig. 5B) . This protein has been reported to be expressed in rods slightly earlier than opsin. This was confirmed in the organotypic cultures where the number of recoverin-expressing cells increased rapidly during the first week in vitro and reached its maximum by day 8. The maximum number of recoverin-positive cells was slightly higher than that of the opsin-positive cells, because a subpopulation of bipolar cells (cone bipolar cells) also expresses this protein. ²² As with opsin, the expression of recoverin was retarded in the presence of exogenous CNTF, but showed a delay of only 2 days compared with the control. However, after 14 DIV, recoverin expression in CNTF-treated cultures reached the same level as in control cultures, indicating complete reversibility of the inhibitory effect. 

If the inhibitory effects that occurred after application of CNTF reflect the enhancement of the regular function of endogenous CNTF or a related molecule, it would be expected that blockade of the CNTF receptor in slice cultures would result in an acceleration of opsin expression. As shown in Figure 5C , such a stimulatory effect occurred when the cultures were grown in the presence of AADH-CNTF, which binds to CNTFRα with high affinity, but does not activate intracellular signal transduction. ²⁹ The antagonist-induced increase in the number of opsin-positive cells was first detected after 4 DIV, it reached a maximum (180% of control) at 8 DIV and declined again thereafter. These data indicate that AADH-CNTF blocked the action of an endogenous factor that retarded the expression of opsin in control cultures. 

The observed loss of sensitivity to CNTF at later stages of photoreceptor differentiation would be most easily explained by a downregulation of the CNTF receptor. To study the expression of CNTFRα in developing photoreceptors, we cut the retinas horizontally (see the Methods section), and the inner and outer retinas were analyzed separately by immunoblot. At PD5, strong signals for CNTFRα protein were observed in both retinal compartments (Fig. 6A) . By in situ hybridization, expression of CNTFRα mRNA has been demonstrated in ganglion cells, in the amacrine cell layer and in horizontal cells of the developing postnatal retina, but not in the ONL. ¹¹ The discrepancy between the latter observation and the presence of significant amounts of CNTFRα protein in the outer retina determined by immunoblot analysis may be due to low mRNA levels in individual ONL cells combined with high cell density in the ONL, which makes CNTFRα protein levels detectable biochemically. After PD5, CNTFRα levels in the outer retina decreased dramatically and were barely detectable at PD13 and later. In the inner retina, expression of the receptor protein appeared to be relatively constant throughout this developmental period. These results show that CNTFRα is expressed in early postnatal outer retina, containing almost exclusively immature photoreceptors, and is downregulated during the second postnatal week in temporal correlation with the massive generation of opsin-positive photoreceptors. 

Expression of CNTFRα protein was also studied in organotypic cultures (Fig. 6B) . Because ganglion cells, the major CNTFRα-expressing neurons of the inner retina, ¹¹ die within hours in culture, it is likely that a substantial proportion of the CNTFRα protein found in the cultures was produced by immature photoreceptors. Under control conditions, receptor levels in extracts from the cultured retinas decreased markedly during development, similar to the ONL in vivo. In the presence of CNTF, however, this developmental decrease was inhibited in a dose-dependent manner (Fig. 6B) . At a CNTF concentration of 20 ng/mL, expression of CNTFRα remained virtually constant between 1 and 9 DIV. At the concentration (5 ng/mL) used for studying CNTF’s effects on opsin expression (see e.g., Figs. 5A 5B ), receptor protein levels were significantly downregulated during this period, but less than in untreated cultures. The downregulation of CNTFRα in vivo and in control cultures as well as in the presence of moderate levels of exogenous CNTF explains why differentiating rods become unresponsive to CNTF. The prolonged expression of the receptor by rod precursors in the presence of increased CNTF concentrations would provide an explanation for the delay in rod differentiation with increased concentrations of the factor. 

The results presented so far support the conclusion that CNTF acts transiently and reversibly on rod precursor cells, resulting in a temporal retardation of terminal rod differentiation. In explant and reaggregate cultures of newborn rat retinas, it has been shown, however, that CNTF-induced reduction of the generation of opsin-positive cells is accompanied by a substantial increase in the number of cells that express immunocytochemical markers of retinal bipolar cells. ¹⁴ In organotypic cultures, a slight but statistically nonsignificant increase in bipolar cell–like neurons was observed, when organotypic cultures were grown in the presence of 5 ng/mL CNTF. ¹⁶ At higher CNTF concentrations, the observations from other culture systems were confirmed (Fig. 7) . Addition of 20 ng/mL CNTF to the culture medium resulted in a 71% reduction of opsin-positive cells, whereas the number of cells stained with the bipolar cell–specific monoclonal antibody 115A10 showed a 3.5-fold increase compared with the control after 10 DIV (Fig. 7A) . Between 10 DIV and 14 DIV the number of opsin-positive cells in CNTF-treated cultures increased from 20% to 43% of all cells, whereas the number of 115A10-positive cells decreased from 50% to 35%. This indicated that, even at the higher CNTF concentration used in these experiments, the effect of the factor was at least partially reversible, with prolonged culture periods resulting not only in a late increase of opsin expression but also in a concomitant decrease in the number of cells with bipolar cell–like immunoreactivity. 

Dissociates of slice cultures used for cell counting (see the Methods section) contained two types of 115A10-positive cells: relatively large cells that were heavily stained, and smaller cells with weaker labeling that frequently showed a polarized distribution over the plasma membrane (Fig. 7B) . The latter type constituted most of the labeled cells in CNTF-treated cultures. To assess the identity of the 115A10-positive cells appearing in the presence of excess CNTF, immunocytochemistry was applied to slice cultures after 10 DIV. In control cultures, typical bipolar cells were strongly labeled, forming a layer adjacent to the outer plexiform layer and clearly separated from the ONL (Fig. 7C) . A few weakly immunoreactive cells were located in the ONL but were never present in rat retinas developing in vivo (data not shown). In the presence of CNTF, numerous cells of the ONL became immunoreactive (Fig. 7D) . These cells appeared smaller than the labeled bipolar cells in the inner nuclear layer. They were less intensely stained and were similar to immature rods (see Fig. 1D ) in morphology and arrangement. Thus, CNTF induced the expression of the bipolar cell marker in ONL cells, but obviously did not change the migratory behavior and the morphologic differentiation of responsive precursor cells. This is in agreement with the results presented in Figures 1 and 2 . 

Discussion

Results and conclusions from previous studies on the inhibitory effect of CNTF on rod development have been controversial. From experiments in monolayer cultures of both mouse and rat retina it has been concluded that exogenous CNTF arrests rod precursor cells at an immature stage. ^{14 16} Observations in rat retinas cultured as explants or as dissociated cells in collagen gels, however, have indicated that CNTF reduces the number of differentiated rods by inducing a switch in the bipolar cell’s fate in uncommitted precursor cells. ¹⁴ In these studies, expression of photoreceptor-specific proteins involved in the phototransduction process were used as markers of rod differentiation. No information was provided about how the massively reduced generation of opsin-positive rods would influence the development of the retinal layers and how the inhibitory action of CNTF would affect the behavior of rod precursors in their normal environment. Results obtained in the current study in organotypic cultures provide strong evidence that the inhibitory effect of CNTF is restricted to certain phenotypic features and results in a transient retardation of terminal rod differentiation, but does not lead to permanent alterations of the structure or cellular composition of the rat retina. 

Effect of CNTF on Formation of the ONL In Vitro

In the rodent retina, a substantial portion of the rod precursors are generated during the first postnatal days, ³⁰ accumulate in the presumptive ONL, and start to differentiate morphologically. There is a delay of several days between birth and terminal differentiation of rod photoreceptors, indicated by the expression of opsin and the beginning formation of the rod outer segments. ^{14 31} From PD5 onward, the ONL is progressively separated from the other retinal layers by the outer plexiform layer formed by the proximal photoreceptor processes that synapse onto dendrites of bipolar and horizontal cells. ³² All these developmental steps apparently proceeded normally in the organotypic cultures, except for the differentiation of mature outer segments, which is known to depend on the interaction with the pigment epithelium. ²⁶ Application of excess CNTF did not interfere with the generation of rod precursors, with migratory processes, or with the morphologic differentiation of rods in the ONL. This shows that the suppression of opsin and other differentiation markers such as recoverin, which has also been observed in CNTF-treated retinal monolayer and explant cultures, ^{14 15 16} does not reflect a total blockade of rod development. 

Retardation of Rod Differentiation Versus Determination of a Cell’s Fate

CNTF has been shown to influence the phenotypic choice in neuronal precursors and neural stem cells. ^{33 34} Therefore, it was an attractive hypothesis that the reduction of rod numbers caused by CNTF could be due to a switch in the developmental fate resulting in the overproduction of other cells types. However, our results provide several arguments against this hypothesis. First, if CNTF induces a nonphotoreceptor phenotype, the responding precursors would be expected to migrate to the appropriate retinal cell layer and to acquire the respective morphology. This did not happen in the organotypic cultures, as discussed herein. Rather, the ONL contained the normal density of immature, morphologically identifiable photoreceptors. In line with these results, it has been shown that rod precursors identified on the basis of their characteristic nuclear morphology did not change in number in CNTF-treated monolayer cultures under conditions resulting in an almost complete inhibition of opsin expression. ¹⁴ Second, the inhibitory effect of CNTF on the expression of the differentiation markers opsin and recoverin was reversible after withdrawal of the factor and turned out to be transient, even in the permanent presence of excess exogenous factor. This does not seem to be compatible with the assumption that CNTF functions to specify another phenotype in rod precursor cells, which then would have to switch back to the fate of a rod. Third and most important, the cells that became immunoreactive for the bipolar cell marker 115A10 in the presence of high CNTF concentrations were located in the photoreceptor layer. In addition, they displayed a morphology that resembled that of neighboring opsin-positive rods but was different from that of normal bipolar cells in the inner nuclear layer. A very similar observation was made by Ezzedine et al. ¹⁵ in explants where the supernumerary cells with bipolar cell–like immunoreactivity in CNTF-treated cultures also were found to exhibited an untypical morphology reminiscent of immature rods. Taken together, these results demonstrate that the rod precursor does not adopt a true bipolar cell phenotype under the influence of CNTF. However, the expression of bipolar cell–specific proteins by cells of the photoreceptor layer, which was also seen in a few cells of control cultures, indicates that the rod precursors possess a certain degree of phenotypic plasticity. It can be speculated that in the in vitro situation, rod precursors that are kept in an immature state for extended periods at high concentrations of CNTF respond to bipolar cell–promoting signals—the timing, composition, and concentration of which is not sufficient to induce a complete switch in cell fate. Thus, our results do not exclude that postmitotic rod precursors in the appropriate environment can be switched to a bipolar cell fate, as has been concluded in a variety of in vitro studies. ^{3 15 17 35} The action of CNTF by itself, however, obviously does not lead to such a switch in cell fate, as supported by the observation that in dissociated cultures, CNTF completely blocks opsin expression without enhancing the expression of bipolar cell markers. ^{14 16}  

Transient Inhibition of the Terminal Differentiation of Postmitotic Rod Precursors

Previous studies in different culture systems have shown that the inhibitory effect of CNTF on rod development is not due to interference with progenitor cell proliferation and the initiation of cell differentiation or cell survival. ^{16 35} The data indicate that CNTF acts on postmitotic rod precursor cells that are generated in normal numbers in the presence of exogenous CNTF. ¹⁴ This is in agreement with the present findings. The reversibility of the effect of CNTF on expression of recoverin and opsin showed that normal numbers of immature rods were present in the CNTF-treated cultures. In addition, CNTFRα was expressed in the forming ONL during the postnatal phase, when this layer consists of newly generated, opsin-negative photoreceptor precursors. Together, these findings show that CNTF acts directly on postmitotic precursor cells committed to the rod fate and exerts its inhibitory effect during the phase of phenotypic differentiation. 

The developmental downregulation of CNTFRα expression in culture and in the ONL of the intact retina provides an explanation of why the differentiating rods become resistant to CNTF and how the inhibitory effect is terminated. In vivo, this downregulation between PD5 and PD8 exactly coincides with the rapid increase in the number of opsin-positive cells. ¹⁸ Different from other CNTF target cells, ³⁶ CNTF enhanced expression levels of its receptor in a dose-dependent manner in photoreceptor cells. This resulted in a delayed downregulation of CNTFRα in CNTF-treated slice cultures and thus could explain the prolonged responsiveness of precursor cells and the delay in opsin expression under these conditions. The timing of opsin expression apparently depends on the balance between the inhibitory action of CNTF and differentiation-promoting signals that induce the downregulation of CNTFRα. In dissociated cultures of low cell density where the number of opsin-positive cells remains low and precursors seem to be permanently arrested at an immature stage, ^{2 3 16} we observed a sustained expression of CNTFRα (data not shown). These cultures seem not to have signals that are necessary for CNTFRα downregulation and rod differentiation. It is noteworthy that conditioned media of high-density cultures that support the differentiation of opsin-positive rods possess both stimulatory and inhibitory activities. ³⁷  

The conclusion that CNTF acts as a transient negative regulator of rod development is further supported by the observed stimulation of rod development by the CNTF antagonist and by a series of in vivo observations. CNTF is produced by Müller glia and is present in the rat retina during early postnatal stages. ¹¹ CNTFRα (this study) as well as the signal transducing receptor component LIF receptor β (Hofmann H.-D., unpublished data, 2001) are expressed in the photoreceptor layer with the appropriate time course. Retinal explants of mice without the CNTFRα contained higher numbers of opsin-positive rods, whereas the number of other cell types was unchanged, ¹⁵ indicating that rod differentiation was accelerated in the absence of CNTF/LIF receptor signaling (as in our experiments with AADH-CNTF). Transgenic mice with LIF overexpression in the lens epithelium exhibited reduced numbers of photoreceptors at early postnatal stages compared with wild-type animals of the same age, but normal numbers at later stages. ³⁸ This confirms the transient sensitivity of immature rods to the inhibitory action of CNTF/LIF receptor–mediated signals. 

In summary, our results provide strong evidence for a novel regulatory function of the neuropoietic cytokine CNTF in neuronal development. In concert with stimulatory signals, the negative influence of CNTF on rod differentiation may serve to control the relative long delay of 5 to 12 days between the last mitosis and the terminal differentiation of rod progenitors, which is typical of this cell type. ¹⁸ As proposed by Cepko et al., ¹⁷ this delay may be necessary, because, otherwise, the formation of photoreceptor outer segments would interfere with ongoing cell division occurring in the same retinal layer. Several molecules have been described to exert stimulatory effects on photoreceptor differentiation. ⁴ Thus, the identification of a photoreceptor-specific inhibitory signal indicates that neuronal differentiation, similar to other developmental processes like neurogenesis or axonal pathfinding, are under the control of antagonistic influences. 

 Supported by Grant SFB 505/A4 from the German Research Foundation.

 Submitted for publication February 19, 2002; accepted May 7, 2002.

 Commercial relationships policy: N.

 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 3.

 

Density of rod-like cells in the outer nuclear layer developing in vivo and in vitro. SEM micrographs, as shown in Figure 3 , were used to quantify the density of rodlike cells with apical processes that represent rudimentary outer segments. P0 and P4: retinas from newborn and 4-day-old rats, respectively; P0+4, organotypic cultures from newborn animals grown for 4 days, in the absence or presence of CNTF.

Figure 1.

 

Development of opsin-immunoreactive cells in the photoreceptor layer of organotypic slice cultures and the effect of CNTF. (A–C) Cultures prepared from PD0 rat retinas were grown under control conditions for 1 DIV (A), 2 DIV (B), and 6 DIV and then stained with antibodies to rhodopsin. The micrographs demonstrate the doubling of the retinal layers, the increase in number and density of opsin-immunoreactive rods, and the formation of the ONL during the first week in culture. Arrows: sharp border of the ONL toward the outer plexiform layer present after 6 DIV. (D) Higher magnification of the ONL of a control culture stained for opsin after 7 DIV. (E) Sister culture grown in the presence of CNTF (5 ng/mL) shows normal orientation and layering of the rods. (F) Quantification of opsin expression in control and CNTF-treated cultures. Data are expressed as the mean percentage ± SD of total cells (n = 5). Bar: (C) 50 μm (valid for A–C); (D) 20 μm (valid for D and E).

Figure 2.

 

In vitro and in vivo development of cell density and morphology in the ONL and the effect of CNTF. Photomicrographs show light microscopic (A, B) or SEM (C–F) views of the scleral surface of the ONL. (A) Slice culture (prepared from PD0 retinas) grown under control conditions after 4 DIV. Immunostaining showed the presence of many labeled rods with rudimentary outer segments. (B) Sister culture grown in the presence of CNTF (5 ng/mL). Total cell density appeared to be similar, but opsin expression was markedly reduced. (C) Control culture (as in A) viewed by SEM. (D) CNTF-treated culture (as in B). (C, D, Insets) Same cultures at lower magnification. Cell morphology and cell density were not detectably different in (C) and (D). (E) ONL of a PD0 retina. (F) ONL of a PD4 retina shows the developmental changes between PD0 and PD4 and the similarity of the PD4 retina with retinas developing in culture (C, D). Bar: (B) 10 μm (valid for A and B); (D): 0.5 μm (inset: 5 μm; valid for C and D); (F) 1 μm (valid for E and F).

Figure 4.

 

Inhibition of opsin by CNTF in organotypic retinal cultures was reversible and was restricted to precursor cells. (A) Reversibility of the CNTF effect: cultures were grown for 4 DIV (filled diamonds) or 6 DIV (filled squares) in the presence of CNTF (5 ng/mL) before the factor was removed by careful washing. Open squares: cultures grown in the continuous presence of CNTF. (B) Effect of CNTF in slice cultures prepared from PD3 (squares) or PD6 (circles) retinas. Cultures were grown until postnatal day 9 in the absence (open symbols) or presence (filled symbols) of 5 ng/mL CNTF. (C) Effect of delayed addition of CNTF to PD0 slice cultures. Open squares: control cultures grown without CNTF; filled diamonds: control conditions 0 to 3 DIV, addition of CNTF at 3 DIV; filled circles: control conditions 0 to 6 DIV, addition of at 6 DIV.

Figure 5.

 

In vitro time course of the expression of opsin and recoverin in the presence of CNTF and the effect of CNTF receptor blockade. (A, B) Organotypic cultures from PD0 retinas were grown for different periods in the absence (open symbols) or presence (closed symbols) of CNTF (5 ng/mL) and the number of cells (percentage of total cells) immunoreactive for opsin (A) or recoverin (B) was determined. (C) Slice cultures were grown in the presence of AADH-CNTF added at a concentration of 50 μg/mL, which had been determined to have saturating effects.

Figure 6.

 

Expression of CNTFα protein during in vivo development of the rat retina and in organotypic slice cultures. (A) Rat retinas of the indicated postnatal stage (PD5, PD8, PD13, and postnatal week 6) were sectioned horizontally to obtain the inner retinal layers (IR) and the photoreceptor-containing outer retina (OR). CNTFRα protein was analyzed by immunoblot. (B) Immunoblots for CNTFRα of organotypic cultures grown for 1, 5, and 9 days under control conditions (lanes 1, 2, and 8), in the presence of CNTF (2.5 ng/mL, lanes 3 and 9; and 5 ng/mL, lanes 4 and 10; 10 ng/mL, lanes 5 and 11; or 20 ng/mL, lanes 6 and 12) or in the presence of LIF (10 ng/mL, lane 7).

Figure 7.

 

Effects of CNTF on the expression of the bipolar cell antigen 115A10. (A) Percentage of cells stained with the monoclonal antibody 115A10 or with antibodies to opsin in organotypic cultures grown in the (□) or presence (20 ng/mL; ▪) of CNTF. (B) Dissociate from a CNTF-treated culture stained for 115A10 immunoreactivity. Note the presence of a larger heavily stained cell and of smaller, less intensely labeled cells with inhomogeneously distributed membrane labeling (arrowheads). (C, D) Immunostaining with monoclonal antibody 115A10 of organotypic slice cultures grown for 10 DIV under control conditions (C) or in the presence of 20 ng/mL CNTF (D). Bar: (B) 10 μm; (D) 20 μm (valid for C and D).

The authors thank Gabriele Kaiser for excellent technical assistance, Bernd Heimrich for advice with the slice cultures, and Jörg Wilting for assistance with the scanning electron microscopic studies.