SEARCH

SEARCH BY CITATION

Keywords:

  • Ciliary neurotrophic factor;
  • Notch;
  • Retina;
  • Stem cells/progenitors;
  • Neurons;
  • Glia;
  • Müller cells

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

In the retina, as elsewhere in the central nervous system, neurogenesis precedes gliogenesis; that is, the only glia in the retina, Müller cells, are born when the majority of neurons have already been generated. However, our understanding of how the multipotent retinal stem cells/progenitors choose to differentiate along neuronal and glial lineages is unclear. This information is important in promoting directed differentiation of retinal stem cells/progenitors in an ex vivo or in vivo stem cell approach to treating degenerative retinal diseases. Here, using the neurosphere assay, we demonstrate that ciliary neurotrophic factor (CNTF), acting in a concentration-dependent manner, influences the simultaneous differentiation of retinal stem cells/progenitors into neurons or glia. At low CNTF concentrations differentiation of bipolar cells is promoted, whereas high CNTF concentrations facilitate Müller cell differentiation. The two concentrations of CNTF lead to differential activation of mitogen-activated protein kinase and Janus kinase-signal transducer and activator of transcription (Jak-STAT) pathways, with recruitment of the former and the latter for the differentiation of bipolar and Müller cells, respectively. The concentration-dependent recruitment of two disparate pathways toward neurogenesis and gliogenesis occurs in concert with Notch signaling. Furthermore, we demonstrate that the attenuation of Jak-STAT signaling along with Notch signaling facilitates the differentiation of retinal stem cells/progenitors along the rod photoreceptor lineage in vivo. Our observations posit CNTF-mediated signaling as a molecular switch for neuronal versus glial differentiation of retinal stem cells/progenitors and a molecular target for directed neuronal differentiation of retinal stem cells/progenitors as an approach to addressing degenerative changes in the retina.

Disclosure of potential conflicts of interest is found at the end of this article.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Author contributions: S.B.: collection and/or assembly of data, data analysis and interpretation, manuscript writing; A.V.D. and K.B.M.: collection and/or assembly of data; I.A.: conception and design, data analysis and interpretation, manuscript writing.

Evidence emerging from a variety of approaches has shown that different types of retinal neurons and the sole retinal glia, Müller cells, arise from common multipotent stem cells/progenitors in response to interplay between cell-intrinsic and cell-extrinsic factors [1]. Among intrinsic factors, transcription factors belonging to basic helix-loop-helix (bHLH) and homeodomain classes have emerged as key regulators of neuronal and glial specification [2]. Cell-cell interactions mediated by membrane bound factors, as exemplified by Notch signaling [3, [4], [5], [6]7] and known diffusible factors such as fibroblast growth factor 2 (FGF2) [8], ciliary neurotrophic factor (CNTF) [9, [10]11], and Sonic hedgehog [12], have been demonstrated to play an important role in the differentiation of specific neuronal types and Müller cells. Although these observations have shed valuable light on the mechanism of retinal histogenesis, our understanding of how retinal progenitors differentiate along neuronal and glial lineages remains unclear. This information, besides being valuable in understanding the generation of neural diversity, is important in promoting directed differentiation of retinal stem cells/progenitors to treat degenerative retinal diseases, such as age-related macular degeneration (AMD), retinitis pigmentosa (RP), and glaucoma, in an ex vivo or in vivo stem cell approach.

Since CNTF influences differentiation of both bipolar cells [9, 10] and Müller glia [11], we were interested in knowing whether signaling mediated by CNTF represents an underlying mechanism for neuronal versus glial differentiation of late retinal stem cells/progenitors. CNTF is a member of the cytokine family that includes interleukin 6 (IL-6), IL-11, leukemia inhibitory factor (LIF), oncostatin M, and cardiotrophin 1 [13]. These cytokines mediate their signal through the gp130 and leukemia inhibitory factor receptor β (LIFRβ) complex, except CNTF, which requires CNTF receptor α (CNTFRα) to confer CNTF specificity. CNTF can activate multiple signaling pathways, including the Janus kinase-signal transducers and activators of transcription (Jak-STAT) [14, 15], mitogen-activated protein kinase (MAPK) [16], phosphatidylinositol 3-kinase [17], and mammalian target of rapamycin [18, 19] pathways.

Here, using the neurosphere assay, we demonstrate that CNTF has concentration-dependent effects on the differentiation of late retinal stem cells/progenitors; at a low concentration (50 ng/ml) the differentiation of bipolar cells is promoted, whereas at a high concentration (100 ng/ml) neuronal differentiation is suppressed and glial differentiation is facilitated. The two concentrations of CNTF lead to differential activation of the MAPK and Jak-STAT pathways in late retinal stem cells/progenitors, with recruitment of the former for bipolar cell differentiation and use of the latter for Müller cell differentiation. We also demonstrate that CNTF-mediated recruitment of the MAPK/Jak-STAT pathway is influenced by Notch signaling. Our observations suggest that a differential recruitment of CNTF-mediated intracellular pathways serves as a molecular switch for neuronal versus glial differentiation of late retinal stem cells/progenitors and that the differentiating Müller cells may set the temporal CNTF gradient underlying the molecular switch. Furthermore, we demonstrate that a selective decrease in Jak-STAT signaling along with Notch signaling facilitates the differentiation of the late retinal stem cells/progenitors along the rod photoreceptor lineage in vitro and in vivo. Thus, CNTF-mediated signaling may serve as a molecular target for facilitating directed neuronal differentiation in an ex vivo or in vivo stem cell approach to treating retinal degeneration.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Neurosphere Assay of Retinal Stem Cells/Progenitors

Neurosphere assay of late (postnatal day 1 [PN1]) retinal stem cells/progenitors from Sprague-Dawley rats (Sasco, Wilmington, MA, http://www.sasco.com) was carried out as previously described [9]. Briefly, retinal cell dissociates were cultured (105 cells per cm2) in retinal culture medium (RCM) containing 20 ng/ml epidermal growth factor (EGF) for 5 days to generate neurospheres. To promote differentiation in specific conditions, neurospheres were cultured in RCM and 1% fetal bovine serum containing different concentrations of recombinant human CNTF (R&D Systems Inc., Minneapolis, http://www.rndsystems.com), in the presence or absence of the drugs PD98059 (20 μM), AG490 (25 μM), and N-(N-[3,5-difluorophenacetyl-l-alanyl])-S-phenylglycine t-butyl ester (DAPT) (3 μM); neutralizing antibody (anti-CNTFRα [5 μg/ml]); or enriched Müller cells [20] for 5 days. In some experiments neurospheres were transduced/transfected with Notch intracellular domain (NICD) retrovirus [21] or dominant-negative mitogen-activated protein kinase (dnMAPK)/dominant-negative signal transducer and activator of transcription (dnSTAT3) expression constructs to perturb Notch or MAPK/Jak-STAT pathways. Cells were either transferred to poly-d-lysine and laminin-coated glass coverslips for immunocytochemical analysis or frozen for reverse transcription (RT) polymerase chain reaction (PCR) analysis. Each experiment was done twice in triplicate.

Intravitreal Injection of Drugs

Intravitreal injection was carried out as previously described [22]. Briefly, a mixture of DAPT (30 μM) and AG490 (40 μg) in a 2-μl volume was intravitreally injected in the right eyes of anesthetized PN1 pups (n = 6) using a glass micropipette. The left eyes received vehicle injection. Eyes were enucleated at the PN8 stage. Retina was dissected, dissociated into single cells, plated on glass coverslips, and processed for immunocytochemical analysis.

Transfection and Luciferase Reporter Assay

PN1 neurospheres were transfected with A1 Mut glial fibrillary acidic protein (GFAP) (STAT binding site mutation)-Luc/STAT3F (dominant interfering forms of STAT3)/MEK1KA97 (dominant interfering forms of MAPK)/(G1–G8)-Luc constructs [23] using Fugene (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com). The empty vector was used as a negative control. Transfection efficiency was examined by cotransfecting cells with pGFP-C3 (Clontech, Palo Alto, CA, http://www.clontech.com). For luciferase reporter assay (G1–G8 Luc constructs containing GFAP promoter sequences), cells were lysed in 1× reporter lysis buffer (Promega, Madison, WI, http://www.promega.com), and 100 μl of lysate was diluted five times using assay reagent. Diluted samples (100 μl) were analyzed for luciferase activities using a luminometer (BD Pharmingen, San Diego, http://www.bdbiosciences.com/index_us.shtml) [21].

Retrovirus Transduction

Recombinant retrovirus was produced as previously described [24]. Briefly, 293T cells were transiently transfected with murine stem cell virus (MSCV)-NICD-IRES-green fluorescent protein (GFP), MSCV-Gagpol, and MSCV-vesicular stomatitis virus (VSV).G constructs using Lipofectamine (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) for 2 days at 35°C. The virus-containing supernatant was harvested after 2 days and stored at −80°C until use. Retinal dissociates were transduced with this retrovirus-containing supernatant in the presence of polybrene (4 μg/ml). Supernatant was aspirated after 24 hours, replaced with fresh medium (RCM), and cultured in the presence of EGF.

Immunofluorescence Analysis

Detection of cell-specific markers was performed as previously described [9]. Briefly, paraformaldehyde-fixed cells were incubated in phosphate-buffered saline (PBS) containing 5% NGS and 0.2% Triton X-100, followed by an overnight incubation in antibodies against protein kinase C (PKC) (1:100 dilution), β-tubulin (1:2,500 dilution), GFAP (1:100 dilution), glutamine synthetase (GS) (1:100 dilution), rhodopsin (ρ-4D2; 1:50 dilution), or rhodopsin kinase (1:100 dilution) at 4°C. Cells were examined for epifluorescence after incubation in IgG conjugated to Cy3 using Openlab software (Improvision, Waltham, MA, http://www.improvision.com). Proportions of different cell types in particular conditions were determined by counting cells with epifluorescence for 4,6-diamidino-2-phenylindole (total number of cells) and cells expressing specific immunoreactivities in four to six randomly selected fields in three different coverslips, with experiments repeated three times. Values are expressed as the mean ± SEM. Student's t test was used to determine the significance of the differences between groups.

Enzyme-Linked Immunosorbent Assay

CNTF levels in culture supernatants were determined using enzyme-linked immunosorbent assay (ELISA)-paired antibodies. Briefly, 50 μl of culture supernatant or recombinant CNTF at different concentrations and 50 μl of biotinylated CNTFRα antibody were added to each well of Nunc Maxisorp plates (Nunc, Rochester, NY, http://www.nuncbrand.com), preadsorbed with CNTFRα (2 μg/ml). After 2 hours of incubation, the plates were washed, and the immunoreactivity was determined using the avidin-horseradish peroxidase-tetramethyl benzidine (HRP-TMB) detection system (Dako, Glostrup, Denmark, http://www.dako.com) and an ELISA microtiter plate reader (Bio-Tek Instruments, Winooski, VT, http://www.biotek.com) at 450 nm. A standard curve of the absorbance versus CNTF concentrations in the standard wells was plotted.

RT-PCR Analysis

RNA was isolated from frozen cells using a Qiagen RNA isolation kit (Qiagen, Hilden, Germany, http://www1.qiagen.com), and cDNA synthesis was performed as previously described [25]. Briefly, ∼5 μg of RNA was transcribed to cDNA in a 50-μl volume. Specific transcripts were amplified with gene-specific forward and reverse primers [9, 20, 26] using a step-cycle program on a Robocycler (Stratagene, La Jolla, CA, http://www.stratagene.com).

Western Analysis

Western analysis of extracellular signal-regulated kinase (ERK)1/2 and STAT3 was carried out as previously described [27]. Briefly, ∼15 μg of the total protein, extracted from neurospheres, was resolved on SDS gel (9%) and transferred onto polyvinylidene difluoride membrane (GE Water and Process Technologies, Trevose, PA, http://www.gewater.com). Specific immunoreactivities were examined using phosphorylated ERK 1/2 kinase (pERK1/2), phosphorylated signal transducer and activator of transcription 3 (pSTAT3), ERK1/2, STAT3, and glyceraldehyde-3-phosphate dehydrogenase antibodies (Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com) followed by detection using HRP-conjugated secondary antibody and the enhanced chemiluminescence kit (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com).

Chromatin Immunoprecipitation Assay

Chromatin immunoprecipitation (ChIP) assay was performed on PN1 neurospheres as previously described [27]. Briefly, PN1 neurospheres were treated with 1% formaldehyde in PBS for 10 minutes at room temperature. Cells were rinsed with ice-cold PBS, harvested, and resuspended in lysis buffer. Cell lysates were sonicated 10 times for 10 seconds and centrifuged at 10,000g for 10 minutes at 4°C. Collected supernatants were precleared by incubating them with salmon sperm DNA/protein A-agarose slurry (1 hour, 4°C). Following immunoprecipitation using antibody against pSTAT3, DNA was extracted from the immune complexes using a desalting column. GFAP promoter sequences were identified by PCR using promoter-specific primers (5′-CCTTTTGTGCCCAACGAGTG-3′, 5′-CTTGCTGAATAGAGCCTTGTCCTC-3′).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Concentration-Dependent Influence of CNTF on Neuronal Versus Glial Differentiation

We tested the hypothesis that CNTF acts as a molecular switch for neuronal versus glial differentiation of retinal stem cells/progenitors using the neurosphere assay [28, [29]30]. Since neuronal versus glial differentiation occurs during the late stage of retinal histogenesis (embryonic day 18 to PN6) [9, 31], we examined the effects of CNTF on bipolar versus Müller cell differentiation in neurospheres generated by PN1 retinal stem cells/progenitors. When PN1 neurospheres were cultured in increasing concentrations of CNTF (0–200 ng/ml), levels of the bipolar cell-specific transcripts Ath3, Chx10, and mGluR6 increased until a CNTF concentration of 50 ng/ml was reached, and they decreased following a further increase in CNTF concentrations (Fig. 1A). Immunocytochemical analyses of bipolar cell differentiation using antibodies against bipolar cell-specific marker PKC revealed an approximately twofold increase (p < .001) in the proportion of PKC+ cells in a low (50 ng/ml) CNTF concentration, which was abrogated to a proportion less than that of controls (p < .05) when neurospheres were exposed to a high (100 ng/ml) CNTF concentration (Fig. 1B–Fig. 1G, Fig. 1H). In contrast, levels of Müller cell-specific transcripts Vimentin, GS, and GFAP increased when neurospheres were cultured in CNTF concentrations >50 ng/ml (Fig. 1I). The increase in levels of Müller cell-specific transcripts was accompanied by an approximately twofold increase (p < .0001) in the proportion of GS+ cells in neurospheres that were exposed to a high CNTF concentration, compared with controls (Fig. 1J–Fig. 1O, Fig. 1P). We observed that the proportion of GS+ cells was slightly but significantly less (p < .05) in neurospheres exposed to a low CNTF concentration compared with controls. Together, these results suggested that CNTF influenced neuronal versus glial differentiation of retinal stem cells/progenitors in a concentration-dependent manner.

thumbnail image

Figure Figure 1.. Concentration-dependent effects of CNTF on neuronal versus glial differentiation. PN1 neurospheres were cultured in the presence of increasing concentrations of CNTF (0–200 ng/ml) for 5 days, and at the end of the culture bipolar or Müller cell differentiation was examined by reverse transcription polymerase chain reaction analyses of cell type-specific transcripts. Levels of bipolar cell (Ath3, Chx10, and mGluR6)-specific transcripts increased until a CNTF concentration of 50 ng/ml was reached and decreased thereafter (A), whereas the level of Müller cell (Vimentin, GS, and GFAP)-specific transcripts increased maximally at a CNTF concentration of 100 ng/ml (I). PN1 neurospheres were cultured in the presence of low (50 ng/ml) and high (100 ng/ml) CNTF concentrations for 5 days, followed by immunocytochemical analysis of cell type-specific markers. Controls represent neurospheres cultured in 0 ng/ml CNTF. The proportions of cells expressing bipolar cell-specific (PKC) (B–H) and Müller cell-specific (GS) (J–P) markers increased in low and high CNTF concentrations, respectively, compared with controls. A small but significant decrease in the proportion of PKC+ and GS+ cells was observed in high and low CNTF concentrations, respectively. Data are expressed as the mean ± SEM obtained from triplicate cultures of two different experiments. ***, p < .001; **, p < .01. Abbreviations: BF, bright field; bp, base pairs; CNTF, ciliary neurotrophic factor; GFAP, glial fibrillary acidic protein; GS, glutamine synthetase; M, marker; PKC, protein kinase C.

Download figure to PowerPoint

Müller Cell-Derived CNTF Influences Neuronal Versus Glial Differentiation

Since the differentiation of Müller cells overlaps that of bipolar cells [31] and Müller cells are the source of CNTF in the retina [32, [33], [34]35], we hypothesized that a developmental increase in the number of Müller cells will lead to an increase in the concentration of CNTF, shifting the balance from neuronal to glial differentiation. To test this hypothesis PN1 neurospheres were cocultured in increasing numbers of Müller cells, and the effects on differentiation were examined. We observed a progressive increase in Vimentin and GS transcript levels with an increase in the number of Müller cells in culture (Fig. 2A). The levels of Ath3 and mGluR6 transcripts increased until the number of Müller cells reached 10 × 104 and decreased thereafter with further increase in Müller cell number. A close correlation between Müller cell number and CNTF concentrations in culture medium was observed; levels of CNTF increased in parallel with an increase in the number of Müller cells (Fig. 2B). Next, to determine that the observed effects on neuronal and glial differentiation were due to Müller cell-derived CNTF, we cocultured PN1 neurospheres with 5 × 104 (low CNTF concentration) and 4 × 105 (high CNTF concentration) Müller cells in the presence and absence of CNTFRα neutralizing antibodies. Immunocytochemical analysis revealed an approximately twofold increase (p < .001) in the proportion of PKC+ cells (Fig. 2C) and GS+ cells (Fig. 2E) when PN1 neurospheres were cocultured with low and high numbers of Müller cells, respectively; this increase was abrogated in the presence of CNTFRα antibodies. The abrogation of the increase in Chx10 and mGluR6 (Fig. 2D) and Vimentin and GS (Fig. 2F) transcript levels in the presence of CNTFRα antibody, as examined by RT-PCR, corroborated results obtained by immunocytochemical analyses. Together, these results suggested that a change in the concentration of Müller cell-derived CNTF signals a shift from neuronal to glial differentiation.

thumbnail image

Figure Figure 2.. Influence of Müller cell-derived CNTF on neuronal versus glial differentiation. Postnatal day 1 (PN1) neurospheres were cocultured in an increasing number of enriched Müller cells for 5 days, and at the end of the culture bipolar and Müller cell differentiation was examined by reverse transcription polymerase chain reaction (RT-PCR) analysis of cell type-specific transcripts (A). Levels of bipolar cell (Ath3 and mGluR6)-specific transcripts increased progressively when Müller cell density in coculture reached 5 × 104 cells and decreased thereafter. Levels of Müller cell (Vimentin and GS)-specific transcripts increased remarkably when Müller cell density in coculture reached 2 × 105 cells and remained stable thereafter. Enzyme-linked immunosorbent assay demonstrated CNTF concentration in the media with different densities of Müller cells (B). PN1 neurospheres were cocultured in low (5 × 104) and high (4 × 105) densities of enriched Müller cells, followed by immunocytochemical and RT-PCR analyses of cell type-specific markers. The proportions of PKC+ (C) and GS+ (D) cells increased in low- and high-density Müller cell coculture, respectively. The increase in the proportion of specific cell-types was abrogated when coculture was carried out in the presence of CNTFRα Ab. RT-PCR analysis of transcripts corresponding to Chx10, mGluR6 ([C], lower panel), Vimentin, and GS ([D], lower panel) corroborated immunocytochemical results. Data are expressed as the mean ± SEM obtained from triplicate cultures of two different experiments. ***, p < .001; **, p < .01. Abbreviations: Ab, antibody; bp, base pairs; CNTF, ciliary neurotrophic factor; CNTFR, ciliary neurotrophic factor receptor; GS, glutamine synthetase; M, marker; PKC, protein kinase C.

Download figure to PowerPoint

CNTF Influences MAPK and Jak-STAT Pathways in a Concentration-Dependent Manner

Next, we hypothesized that CNTF influences neurogenesis and gliogenesis by concentration-dependent recruitment of specific intracellular pathways, namely the MAPK [36] and Jak-STAT [23] pathways, respectively. To test this premise, PN1 neurospheres were cultured in low and high CNTF concentrations in the presence and absence of MAPK inhibitor (PD98059) and Jak inhibitor (AG490), and the effects of these drugs on differentiation of bipolar and Müller cells were examined. We observed an increase in Ath3, Chx10, and mGlur6 transcript levels in a low CNTF concentration, which was abrogated in the presence of PD98059 and not AG490 (Fig. 3A). Similarly, immunocytochemical analyses revealed an increase in the proportion of PKC+ cells, compared with controls (26.11 ± 3.2 vs. 15.13 ± 2.8; p < .001), which was abrogated in the presence of PD98059 (26.11 ± 3.2 vs. 11.8 ± 2.6; p < .001) and not AG490 (26.11 ± 3.2 vs. 27.1 ± 2.1; p > .001) (Fig. 3B–3J). The decrease in the proportion of PKC+ cells was significant relative to controls (15.13 ± 3.2 vs. 11.8 ± 2.6; p < .01), suggesting inhibition of the normally active MAPK pathway. In contrast, the increase in Vimentin, GS, and GFAP transcript levels observed in the presence of a high CNTF concentration was abrogated in the presence of AG490 and not PD98059 (Fig. 3K). Similarly, the increase in proportion of GS+ cells in the high CNTF concentration, compared with controls (58.51 ± 3.7 vs. 30.7 ± 2.3; p < .001), was abrogated in the presence of AG490 (58.51 ± 3.7 vs. 25.3 ± 1.9; p < .001) and not PD98059 (58.51 ± 3.7 vs. 59.42 ± 3.2; p > .001) (Fig. 4L–4T). The decrease in the proportion of PKC+ cells was significant relative to controls (30.7 ± 2.3 vs. 25.3 ± 1.9; p < .01), suggesting inhibition of the normally active Jak-STAT pathway. That the MAPK and Jak-STAT pathways are involved in neurogenesis/gliogenesis in the absence of CNTF was demonstrated by a decrease in bipolar/Müller cell differentiation, as ascertained by the expression of cell type-specific markers, when PD98059 or AG490 was added to control culture conditions, respectively (supplemental online Fig. 1). Next, we monitored the respective pathways in late retinal stem cells/progenitors, in response to different concentrations of CNTF. The activation of MAPK and Jak-STAT pathways was determined by an increase in the levels of pERK1/2 and STAT3 (pSTAT3), respectively. Western analysis of PN1 neurospheres revealed pERK1/2 (Fig. 3U) and pSTAT3 (Fig. 3V) in the presence of both low and high CNTF concentrations. However, pERK1/2 levels were relatively higher in a low CNTF concentration than in a high CNTF concentration and abrogated in the presence of PD98059 and not AG490 (Fig. 3U). Conversely, pSTAT3 levels were relatively higher in a high CNTF concentration than in a low CNTF concentration and abrogated in the presence of AG490 and not PD98059 (Fig. 3V).

thumbnail image

Figure Figure 3.. Influence of inhibitors of specific intracellular signaling pathways on concentration-dependent effects of CNTF on neuronal and glial differentiation. Postnatal day 1 neurospheres were cultured in low (50 ng/ml) (A–J) and high (100 ng/ml) (K–T) CNTF concentrations in the presence or absence of the MAPK inhibitor PD and the JAK inhibitor AG for 5 days followed by the examination of bipolar/Müller cell differentiation and levels of p-ERK1/2 and p-STAT3. Reverse transcription polymerase chain reaction analyses demonstrated changes in levels of bipolar cell (Ath3, Chx10, and mGluR6) (A) and Müller cell (Vimentin, GS, and GFAP) (K)-specific transcripts in low and high CNTF concentrations, respectively, in the presence of specific drugs. Immunocytochemical analyses revealed changes in the proportion of protein kinase C+ (B–J) and GS+ (L–T) cells in neurospheres, cultured in low and high CNTF concentrations, respectively, in the presence of specific drugs. Controls consisted of neurospheres cultured without CNTF. Western analysis revealed the relative levels of p-ERK1/2 versus ERK1/2 (U) and p-STAT3 versus STAT3 (V) in neurospheres cultured in low and high CNTF concentrations, in the presence or absence of specific drugs. Levels of GAPDH represented controls for loading. Data are expressed as the mean ± SEM obtained from triplicate cultures of two different experiments. ***, p < .001; **, p < .01. Abbreviations: AG, AG490; bp, base pairs; c, control; CNTF, ciliary neurotrophic factor; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GFAP, glial fibrillary acidic protein; GS, glutamine synthetase; M, marker; PD, PD98059; p-ERK1/2, phosphorylated ERK 1/2 kinase; p-STAT3, phosphorylated signal transducer and activator of transcription 3; STAT3, signal transducer and activator of transcription 3.

Download figure to PowerPoint

thumbnail image

Figure Figure 4.. Perturbations of CNTF-dependent MAPK and Janus kinase-STAT pathways influence neuronal and glial differentiation. Postnatal day 1 neurospheres, transfected with dnMAPK or dnSTAT3 expression constructs, were cultured in low (50 ng/ml) and high (100 ng/ml) CNTF concentrations followed by the examination of bipolar/Müller cell differentiation and levels of p-ERK1/2 and p-STAT3. Reverse transcription polymerase chain reaction analyses demonstrated changes in Ath3, Chx10, and mGluR6 transcript levels (A, B); proportion of PKC+ cells (C); Vimentin, GS, and GFAP transcript levels (E, F); and proportion of GS+ cells (G) in low and high CNTF concentrations, respectively, in neurospheres transfected with dnMAPK/dnSTAT3 constructs. Controls consisted of neurospheres transfected with empty constructs and cultured without CNTF. Western analyses reveal changes in levels of p-ERK1/2 versus ERK1/2 (D) and p-STAT3 versus STAT3 (H) in neurospheres, transfected with dnMAPK/dnSTAT3 constructs and cultured in low or high CNTF concentrations. ***, p < .001; **, p < .01. Abbreviations: bp, base pairs; CNTF, ciliary neurotrophic factor; dnMAPK, dominant-negative mitogen-activated protein kinase; dnSTAT3, dominant-negative signal transducer and activator of transcription 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GS, glutamine synthetase; M, marker; p-ERK1/2, phosphorylated ERK 1/2 kinase; PKC, protein kinase C; p-STAT3, phosphorylated signal transducer and activator of transcription 3; STAT3, signal transducer and activator of transcription 3.

Download figure to PowerPoint

To further test that CNTF-mediated activation of ERK1/2 and STAT3 augments the neurogenic and gliogenic machineries, respectively, we examined the effects of the loss of function on neuronal versus glial differentiation by ectopic expression of dn forms of MEK1 (MEK1KA97 is the same as dnMAPK) and STAT3 (STAT3D is the same as dnSTAT3) [23] in retinal stem cells/progenitors. The dnMAPK carries a mutation in the ATP binding site of protein kinase, thus rendering it nonfunctional. dnSTAT3 carries a mutation that prevents its phosphorylation by Jak1. As a consequence, it cannot dissociate from the receptor, blocking endogenous STAT3 from docking and being phosphorylated by Jak1. PN1 neurospheres, transfected with dnMAPK or dnSTAT3 constructs, were cultured in the presence of low and high CNTF concentrations, followed by the examination of bipolar and Müller cell differentiation, respectively. We observed that the increases in Ath3, Chx10, and mGluR6 transcript levels (Fig. 4A, 4B) and the proportion of PKC+ cells (Fig. 4C) in a low CNTF concentration were abrogated in dnMAPK- and not dnSTAT3-expressing neurospheres. In contrast, the increases in GS and Vimentin transcript levels (Fig. 4E, 4F) and the proportion of GS+ cells (Fig. 4G) in a high CNTF concentration were abrogated in dnSTAT3- and not in dnMAPK-expressing neurospheres. That the perturbation of MAPK and Jak-STAT attenuates neurogenesis/gliogenesis in the absence of CNTF was demonstrated by a decrease in bipolar/Müller cell differentiation when the expression of cell type-specific markers was compared in dnMAPK/dnSTAT3-transfected and control neurospheres in differentiation conditions without CNTF (supplemental online Fig. 2).

To demonstrate that the neurogenic and gliogenic effects of dnMAPK and dnSTAT3 involve their respective pathways, we carried out Western analysis on PN1 neurospheres, transfected and cultured as described above. We observed a significant reduction in levels of pERK1/2 in dnMAPK-expressing cells but not in those transfected with dnSTAT3 constructs (Fig. 4D). Conversely, levels of pSTAT3 were reduced in dnSTAT3-expressing cells and not in those transfected with dnMAPK (Fig. 4H). Together, these observations suggested that neuronal and glial differentiation in PN1 retinal progenitors were facilitated by CNTF-dependent MAPK and Jak-STAT pathways, respectively.

High CNTF Concentration Activates Jak-STAT-Dependent GFAP Promoter Activities in Late Retinal Stem Cells/Progenitors

Next, we wanted to know whether the expression of Müller cell-specific genes in retinal stem cells/progenitors was directly influenced by CNTF-mediated Jak-STAT pathway in a concentration-dependent manner. To address this issue, we transiently transfected PN1 neurospheres with GFAP promoter-reporter constructs (G1–G8) and monitored their activities in the presence of low and high CNTF concentrations. It has previously been demonstrated that GFAP-promoter activities are regulated by the STAT-binding site [23]. We observed a significant increase in levels of luciferase activities in the presence of a high CNTF concentration in PN1 neurospheres, which were transfected with G1 and G8 constructs, containing the STAT-binding site (Fig. 5A, 5B). There was a significant decrease in luciferase activities with shorter GFAP promoter-reporter constructs lacking STAT-binding site. The specificity of STAT-mediated GFAP promoter activities was further demonstrated by a significant decrease in luciferase activities with the GFAP promoter-reporter constructs (A1 Mut) containing a point mutation in the STAT-binding sites. A significant decrease in luciferase activities was observed when PN1 neurospheres were cotransfected with dnSTAT3 constructs, suggesting that the activation of the GFAP promoter is directly related to CNTF-mediated activation of the Jak-STAT pathway. CNTF at a low concentration failed to elicit luciferase activities above the background with G1 constructs, suggesting a concentration-dependent influence of CNTF on the GFAP promoter. Next, we determined whether CNTF had similar influence on the endogenous GFAP promoter. First, we carried out ChIP analysis on nuclear extracts of PN1 neurospheres, cultured in low and high CNTF concentrations, using antibodies against pSTAT3 to precipitate the DNA protein complex. Amplification of precipitated DNA revealed sequences corresponding to GFAP promoter in both low and high CNTF concentrations (Fig. 5C). However, the intensity of amplification of the GFAP promoter sequence was severalfold higher in the latter than in the former. The intensity of amplification of GFAP promoter sequences, precipitated from neurospheres and exposed to a high CNTF concentration, was abrogated when these cells were transiently transfected with dnSTAT3 constructs, suggesting that the activation of native GFAP promoter in response to CNTF signaling was mediated through the Jak-STAT pathway. Second, we cultured PN1 neurospheres from GFAP-GFP mice [34] in a high CNTF concentration in the presence of either PD98059 or AG490 and examined the cells for expression of GFP. We observed GFP+ cells in neurospheres that were cultured in a high CNTF concentration (Fig. 5G–5I), and the proportion of these cells decreased when neurospheres were exposed to AG490 (Fig. 5M–5O) compared with those exposed to PD98059 (Fig. 5J–5L). GFP+ cells were not detected in control neurospheres (Fig. 5D–5F) or those cultured in a low CNTF concentration (data not shown). Together, these observations suggested that CNTF in a high concentration targets the GFAP promoter in late retinal stem cells/progenitors via the Jak-STAT pathway.

thumbnail image

Figure Figure 5.. Activation of Jak-STAT-dependent GFAP promoter activities in a high CNTF concentration. Postnatal day 1 (PN1) neurospheres, transfected with GFAP promoter-reporter constructs (G1–G8), were cultured in low (50 ng/ml) and high (100 ng/ml) CNTF concentrations followed by examination of luciferase activities (A, B). G1 and G8 constructs contain STAT-binding sites that were absent in shorter GFAP promoter reporter constructs (G2–G6). The construct A1 Mut contained a point mutation in the STAT-binding sites. Chromatin immunoprecipitation (ChIP) analysis was carried out on nuclear extracts of PN1 neurospheres cultured in low and high CNTF concentrations, using antibodies against phosphorylated STAT3 to precipitate DNA protein complex. Amplification of precipitated DNA revealed sequences corresponding to GFAP promoter in both low and high CNTF concentrations; however, levels of precipitated sequences were higher in latter than former and absent in neurospheres transfected with dnSTAT3 constructs (C). Controls included neurospheres cultured without CNTF and those on which ChIP analysis was carried out using IgG. PN1 neurospheres from GFAP-GFP mice were cultured in a high CNTF concentration (G, H, I) in the presence of either PD (J, K, L) or AG (M, N, O) for 5 days and examined for the expression of GFP. GFP+ cells were detected in a high CNTF concentration, compared with controls (D, E, F), and their proportions decreased in the presence of AG and not PD. Abbreviations: AG, AG490; BF, bright field; bp, base pairs; CNTF, ciliary neurotrophic factor; DAPI, 4,6-diamidino-2-phenylindole; dnSTAT3, dominant-negative signal transducer and activator of transcription 3; GFP, green fluorescent protein; Mut, mutant; PD, PD98059; WT, wild-type.

Download figure to PowerPoint

Interactions Between CNTF and Notch Pathways During Neuronal and Glial Differentiation

Next, we examined whether Notch signaling, which is known to influence neuronal [4, 5] and glial [6, 7] differentiation, interacted with CNTF-mediated signaling during neural versus glial differentiation of retinal stem cells/progenitors. We first examined the effects of Notch signaling on CNTF-mediated neuronal differentiation. We cultured PN1 neurospheres, transfected with NICD or empty retrovirus, in a low CNTF concentration, in the presence and absence of the γ-secretase inhibitor DAPT. As expected, we observed an increase in Chx10 and mGluR6 transcript levels in the CNTF-treated group, compared with controls (Fig. 6A, 6B). Levels of these transcripts were decreased and increased in the CNTF+NICD and CNTF+DAPT groups, respectively, suggesting that Notch signaling influenced CNTF-mediated neurogenesis. The specificity of the perturbation in Notch signaling was demonstrated by changes in levels of the Notch target genes Hes1 and Hes5 in response to NICD overexpression or the presence of DAPT. Since Notch signaling is known to suppress the MAPK pathway [37], we examined whether its influence on CNTF-mediated neurogenesis was mediated via MAPK. In an experiment performed as described above and carried out in a low CNTF concentration, we observed a decrease and an increase in levels of pERK1/2 kinase in the CNTF+NICD and CNTF+DAPT groups, respectively, compared with those in CNTF-only or control groups, suggesting that Notch signaling adversely affected CNTF-mediated activation of MAPK (Fig. 6E1). Since Notch signaling plays an instructive role in glial differentiation in the retina [7], we examined whether CNTF and Notch signaling interacted during Müller cell differentiation. In an experiment similar to that described above but carried out in a high CNTF concentration, we observed an increase in Vimentin and GS transcript levels in the CNTF-treated group, compared with controls (Fig. 6C, 6D). However, levels of these transcripts increased and decreased in the CNTF+NICD and CNTF+DAPT groups, respectively, compared with the CNTF-only and control groups, suggesting that Notch signaling was required for CNTF-mediated glial differentiation of retinal stem cells/progenitors. To know whether Notch signaling affected CNTF-mediated gliogenesis through the Jak-STAT pathway, we examined pSTAT3 levels in response to perturbation in Notch signaling in PN1 neurospheres, cultured as described above in a high CNTF concentration. We observed an increase in pSTAT3 levels in CNTF+NICD group, compared with those in CNTF-only or control group (Fig. 6E2). In contrast, pSTAT3 levels were undetectable in the CNTF+DAPT group, suggesting that interactions with Notch signaling were necessary for CNTF-mediated gliogenesis. It followed from these observations that Notch signaling is coincidental with CNTF signaling for gliogenesis and that the latter may sustain the former. To test this premise, we examined the expression of Notch1 and Hes5 transcripts in PN1 neurospheres, exposed to either increasing CNTF concentrations or Müller cells. We observed a progressive increase in Notch1 and Hes5 transcript levels with increase in CNTF concentrations (Fig. 6F) and Müller cell numbers (Fig. 6G). This increase in the expression of components of the canonical Notch pathway was accompanied by an increase in CNTFRα transcript levels, corroborating the feed-forward influence of CNTF signaling observed during gliogenesis [38].

thumbnail image

Figure Figure 6.. Interactions between CNTF and Notch pathways during neuronal and glial differentiation. Postnatal day 1 neurospheres, transduced with NICD or empty retrovirus, were cultured in low (50 ng/ml) or high (100 ng/ml) CNTF concentrations, in the presence or absence of DAPT, followed by the examination of bipolar/Müller cell differentiation and levels of pERK1/2 and pSTAT3. Reverse transcription polymerase chain reaction (RT-PCR) analyses demonstrated levels of bipolar cell (Chx10 and mGlur6) (A, B) and Müller cell (Vimentin, GS) (C, D)-specific transcripts in neurospheres transduced with NICD/empty retrovirus and cultured in low and high CNTF concentrations, respectively, in the presence or absence of DAPT. Hes1 and Hes5 transcript levels were examined as a measure of the perturbation of Notch signaling. Western analyses revealed changes in levels of pERK1/2 versus ERK1/2 (E1) and pSTAT3 versus STAT3 (E2) in neurospheres transduced and treated as above. RT-PCR analyses revealed levels of Notch1, Hes5, and CNTFRα transcripts in different CNTF concentrations (F) and numbers of enriched Müller cells (G). Controls included neurospheres cultured without CNTF. Abbreviations: bp, base pairs; CNTF, ciliary neurotrophic factor; CNTFR, ciliary neurotrophic factor receptor; DAPT, N-[N-(3,5-difluorophenacetyl-l-alanyl)]-S-phenylglycine t-butyl ester; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GS, glutamine synthetase; M, marker; NICD, Notch intracellular domain; pERK1/2, phosphorylated ERK 1/2 kinase; pSTAT3, phosphorylated signal transducer and activator of transcription 3.

Download figure to PowerPoint

Inhibition of Jak-STAT and Notch Signaling Promote Neuronal Differentiation in Müller Stem Cells

Our observations suggested that perturbation of CNTF and Notch signaling could be used to preferentially direct neurogenesis. We tested the premise in Müller cells, which were recently demonstrated to be the latent stem cells in the mammalian retina [20]. When exposed to FGF2+EGF, Müller cells generate clonal neurospheres containing self-renewing and multipotent cells. Müller cell-derived neurospheres were cultured in a high CNTF concentration, with AG490, DAPT, AG490+DAPT, PD98059, or NICD+PD98059. The NICD group consisted of neurospheres transduced with NICD retrovirus. Controls consisted of neurospheres transduced with an empty retrovirus cultured without drugs and with CNTF. We observed that neuronal differentiation, as judged by the relative levels of β-tubulin and mGluR6 transcripts, increased in the CNTF+AG940/CNTF+DAPT group (Fig. 7A). There was a synergistic increase in the levels of β-tubulin and mGluR6 transcripts in the CNTF+AG490+DAPT group. In contrast, glial differentiation, as judged by the relative levels of Vimentin and GS transcripts, was suppressed in these groups. However, levels of Vimentin and GS transcripts and those corresponding to Hes1 and Hes5 increased in the CNTF+PD98059/CNTF+NICD/CNTF+PD98059+NICD group. In contrast, β-tubulin and mGluR6 transcripts were undetectable in all these groups. Together, these observations suggested that the attenuation of CNTF and Notch signaling by STAT3 and γ-secretase inhibitors, respectively, could selectively promote neuronal differentiation. Next, to determine that the perturbation of CNTF and Notch pathways could constitute an approach to increasing the efficiency of directed differentiation of Müller cells along a specific neuronal type (i.e., the rod photoreceptors), we cocultured Müller cell-derived neurospheres with PN1 retinal cells across a membrane, in the presence of AG490/DAPT/AG490+DAPT. Controls included neurospheres cocultured without drugs. It has been demonstrated previously that PN1 retinal cells have elaborate rod photoreceptor-promoting activities and influence rod photoreceptor differentiation [20, 26, 39]. We observed an increase in levels of transcripts corresponding to regulators (e.g., Nrl, Otx) and markers (e.g., opsin, rhodopsin kinase) of rod photoreceptors in neurospheres cultured in the presence of AG409/DAPT (Fig. 7B, 7C). Those cultured with AG409+DAPT displayed a synergistic increase in levels of these transcripts. In contrast, Vimentin and GS transcript levels decreased in the presence of drugs. To ensure that the result was not a function of culture conditions and ascertain that the perturbation of Jak-STAT and Notch pathways affect normal rod photoreceptor differentiation, we exposed retinal stem cells/progenitors to AG490+DAPT in vivo by intravitreal injection of drugs in the eyes of PN1 rats and examined the effects on rod photoreceptor differentiation a week later. We observed a significant increase (p < .01) in the proportion of retinal cells expressing rhodopsin (Fig. 7D, 7E, 7H) and rhodopsin kinase (Fig. 7F, 7G, 7H) obtained from the retina of treated eyes, compared with that from control eyes. Together, these observations suggested that both CNTF signaling and Notch signaling negatively regulate neural differentiation in general and rod photoreceptor differentiation in particular.

thumbnail image

Figure Figure 7.. Regulation of neuronal differentiation by CNTF and Notch signaling in Müller stem cells. Müller cell-derived neurospheres, transduced with NICD or empty retrovirus, were cultured in high CNTF concentrations and subdivided into the following groups, based on drug treatments (AG, DAPT, AG+DAPT, PD) or retrovirus transduction (NICD, NICD+PD). Controls consisted of untransduced neurospheres, cultured in CNTF and without drugs. Reverse transcription polymerase chain reaction (RT-PCR) analysis demonstrated changes in levels of neural (β-tubulin and mGluR6) and glial (Vimentin and GS)-specific transcripts when Jak-STAT (AG), mitogen-activated protein kinase (MAPK) (PD), and Notch (DAPT and NICD) pathways are perturbed alone or together (AG+DAPT; PD+NICD) (A). Hes1 and Hes5 transcript levels were examined as a measure of the perturbation of Notch pathway. To examine the potential of Müller stem cells to differentiate along photoreceptor lineage, Müller glia-derived neurospheres were cocultured with postnatal day 1 (PN1) retinal cells across a membrane, in the presence of AG, DAPT, and AG+DAPT (B, C). Controls included neurospheres cocultured in the absence of drugs. RT-PCR analysis demonstrated changes in levels of transcripts corresponding to regulators (e.g., Nrl, Otx) and markers (e.g., opsin, rhodopsin kinase) of rod photoreceptors in neurospheres when Jak-STAT (AG) or Notch (DAPT) pathways or both (AG+DAPT) were inhibited. Vimentin and GS transcript levels were examined as controls for glial phenotype of enriched Müller cells. To demonstrate that normal photoreceptor generation can be influenced by perturbing Jak-STAT and Notch pathways, AG+DAPT/vehicle was injected intravitreally in PN1 rat eyes and retinal cells were examined a week later (PN8) by immunocytochemical analysis. The proportion of cells expressing the rod photoreceptor-specific markers rhodopsin (D, E, H) and RK (F–H) was significantly increased in treated group, compared with controls. The differentiation of late retinal stem cells/progenitors along neuronal (bipolar cells) and glial (Müller cells) lineages is facilitated in response to cell-cell interactions via membrane-bound factors (i.e., Delta, in the case of Notch signaling) and diffusible factors such as cytokines (i.e., CNTF) (I). When the concentration of CNTF in the environment is low, the MAPK pathway, which facilitates the differentiation of stem cells/progenitors into bipolar cells, is preferentially recruited. As CNTF concentrations increase in the milieu, presumably due increase in the number of Müller cells, the Jak-STAT pathway overtakes it, promoting glial (Müller cells) differentiation. Attenuation and strengthening of Notch signaling inhibits and promotes neuronal and glial differentiation, respectively, in concert with CNTF signaling. Müller cells activate their latent stem cell potential upon injury or exposure to mitogens and/or Wnt ligands and differentiate along photoreceptor lineage [20], the efficiency of which can be enhanced by inhibiting Jak-STAT and Notch pathways. **, p < .01. Abbreviations: AG, AG490; bp, base pairs; CNTF, ciliary neurotrophic factor; DAPT, N-[N-(3,5-difluorophenacetyl-l-alanyl)]-S-phenylglycine t-butyl ester; GFAP, glial fibrillary acidic protein; GS, glutamine synthetase; JAK-STAT, Janus kinase-signal transducer and activator of transcription; M, marker; NICD, Notch intracellular domain; PD, PD98059; RK, rhodopsin kinase.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

The simultaneous generation of neurons and glia from a similar pool of stem cells/progenitors is not well understood. Here, we have examined the role of CNTF in differentiation of late retinal stem cells/progenitors into bipolar and Müller cells. The two cell types become postmitotic within hours of each other in the developing retina. For example, 95% of bipolar cells and Müller glia are generated within a difference of 2.4 hours [31]. Our study demonstrates that single factors in the microenvironment can influence the differentiation of both bipolar cells and Müller glia. The key to the differential effects of single factors, exemplified by CNTF, is the concentration of the cytokine in the microenvironment to which differentiating progenitors are exposed. For example, a low concentration of CNTF favors the differentiation of bipolar cells, and a high CNTF concentration facilitates Müller cell differentiation. The concentration-dependent effects of CNTF on neurogenesis and gliogenesis appear to be due to a selective engagement of disparate intercellular pathways. Our study suggests that CNTF signaling involves both Jak-STAT and MAPK pathways in differentiating retinal stem cells/progenitors. However, at a low CNTF concentration the MAPK pathway is preferred over the Jak-STAT pathway, which facilitates the differentiation of bipolar cells. At this concentration, the effect of CNTF is akin to that of growth factors on cortical neurogenesis [36]. The activation of the MAPK pathway promotes neural differentiation of cortical progenitors, which may involve activation of neurogenic and neuronal phenotype-specific genes by the CCAAT/enhancer-binding protein family of transcription factors [40, [41]42]. It is possible that the MAPK pathway, preferentially activated by low CNTF concentrations, similarly facilitates the expression of the neurogenic bHLH transcription factor gene Ath3, a known regulator of bipolar cells [9, 43]. This premise is supported by observations that perturbation of MAPK pathways by either drugs or dnMAPK constructs led to a decrease in the expression of Ath3, accompanied by a compromised bipolar cell differentiation.

CNTF mediates the differentiation of neural progenitors along the astrocytic lineage by activating the Jak-STAT pathway [23, 44, [45]46]. Our study suggests that CNTF, at a high concentration, similarly influences the differentiation of late retinal stem cells/progenitors along Müller cell lineage by directly influencing Müller cell-specific genes such as GFAP. Unlike the previous study [11], our results suggest that the gliogenic effect may not be mediated by the MAPK pathway. A similar mechanism of CNTF-mediated gliogenesis, which involves the Jak-STAT and not the MAPK pathway, has been observed during differentiation of cortical progenitors into astrocytes [23]. We observed that CNTF, at concentrations above 50 ng/ml, was gliogenic. However, there was a slight but significant decrease in the number of GS+ cells when progenitors were cultured in the presence of 50 ng/ml CNTF, despite the evidence of an activated Jak-STAT pathway. This intriguing observation may indicate the concentration-dependent recruitment of Jak-STAT and MAPK pathways by CNTF. Both pathways may be activated in neural progenitors/precursors when exposed to low CNTF concentrations. However, the gradient of the activated MAPK pathways reaches its crest at 50 ng/ml, where it promotes neurogenesis maximally and suppresses gliogenesis, presumably by negatively regulating the marginally activated Jak-STAT pathway. The MAPK pathway is known for simultaneously promoting neurogenesis and inhibiting gliogenesis, by inhibiting Jak-STAT [47] and/or Notch pathways (described below).

We observed interactions between CNTF-mediated signaling and Notch pathway during differentiation of retinal stem cells/progenitors. Notch signaling is associated with diverse developmental processes in the retina; acting in a context-dependent manner it regulates the maintenance of retinal stem cells/progenitors [48] and the differentiation of specific neurons [4, 5, 49, 50] and glia [6, 7]. Our observations suggest that the differentiation of Müller cells in the presence of a high CNTF concentration is orchestrated by interactions between the Notch and CNTF-activated Jak-STAT pathways. Such interactions may involve CSL (CBF1, Su [H], Lag-1)-mediated activation of glial-specific genes such as GFAP [51] and/or phosphorylation of STAT3 during astrogliogenesis [52, 53]. Observations that the instructive influence of Notch signaling requires coincidental activation of the Jak-STAT pathway [51] and that Notch signaling can activate the Jak-STAT pathway suggest the likelihood of a positive feedback loop between these two pathways. This may be the case as CNTF, besides activating the expression of its own receptor, positively influences the expression of Notch1 and Hes5 in a concentration-dependent manner. Therefore, it can be surmised that the positive feedback loop progressively strengthens as concentrations of cytokines increase with development, shifting the balance more toward gliogenesis. At low CNTF concentrations, when the MAPK pathway is involved and neurogenesis is promoted, Notch signaling is kept low for two reasons. A decrease in Notch signaling is essential in (a) promoting neurogenesis [4, 5] and (b) suppressing errant gliogenesis in late retinal stem cells/progenitors, which display proclivity toward glial differentiation [6]. Such an interactive modulation of neurogenesis and gliogenesis presents the Jak-STAT and Notch pathways as molecular targets for induced neurogenesis for regeneration purposes, particularly in light of recent evidence that adult Müller cells possess stem cell properties [20, 54] and participate in neurogenesis [20, 55, 56]. Our findings suggest that the inhibition of the Jak-STAT and Notch pathways facilitates rod differentiation of enriched Müller cells in vitro and during late histogenesis in vivo (Fig. 7), lending credence to this premise.

A question remains as to how a difference in the concentration of a particular cytokine leads to the engagement of two different intracellular pathways, delineating neuronal versus glial differentiation. One can consider the possibility that the different concentrations of CNTF engage different receptor complexes. For example, although CNTF is known to act through the CNTFRα-LIFR-gp130 trimeric receptor complex with half-maximal stimulation at ∼1 ng/ml, it can directly induce signaling through the LIFR-gp130 heterodimer complex in the absence of CNTFRα, albeit at a high concentration [57]. Also, at high concentrations, human CNTF has been observed to deliver signaling via the IL-6R-LIFR-gp130 complex [58]. Therefore, it can be suggested that the neuronal versus glial differentiation at different CNTF concentrations is a reflection of the engagement of different receptor complexes on similar or different precursor populations within neurospheres. This notion is supported by the observation that LIF can facilitate neurogenesis/gliogenesis in PN1 neurospheres, but at a concentration of five times lower than CNTF (supplemental online Fig. 3).

The other possibility is that cells in neurospheres interpret CNTF concentrations on the basis of the levels of the expression and occupancy of receptors, as proposed for morphogens [59]. In this situation, when the CNTF concentration is low, not all receptors expressed on retinal stem cells/progenitors and precursors are occupied. The partial occupation of receptors is interpreted as neurogenic signal via the recruitment of the MAPK pathway. The MAPK pathway further potentiates neurogenesis by inhibiting Notch signaling, presumably through the inhibition of γ-secretase, the enzyme responsible for the cleavage of the activated Notch receptor and/or the phosphorylation-based inhibition of Groucho, the global corepressor involved in mediating the repression by transcription inhibitors, including those that belong to the Hes class [60, 61]. At a high CNTF concentration, significantly more CNTF receptors are occupied in retinal stem cells/progenitors and precursors, which translates into a gliogenic signal via the recruitment of the Jak-STAT pathway. In both situations, the role of intracellular regulatory proteins, particularly that of suppressor of cytokine signaling (SOCS) proteins, may be considered for modulating disparate intracellular signaling pathways in different cytokine concentrations. SOCS belong to a family of intracellular proteins, some of which act in a classic negative feedback loop to regulate the Jak-STAT pathway by ubiquitination-mediated degradation of the signaling complex and inhibiting Jak tyrosine kinase [62]. Evidence suggests that in addition to regulating the Jak-STAT pathway, SOCS proteins may interact with the MAPK pathway. For example, it has been observed that phosphorylated SOCS-3, although it inhibits the Jak-STAT pathway, potentiates the MAPK pathway by maintaining the phosphorylation of ERK [63]. Also, evidence has emerged that expression of SOCS proteins can be induced by growth factors [64]. Against the backdrop of these observations an explanation can be tendered for the potentiation of the MAPK pathway vis-à-vis the Jak-STAT pathway in low CNTF concentrations, where both pathways are activated but the induction and phosphorylation of SOCS proteins potentiate the former and attenuate the latter. Although additional investigations are needed to test these premises, our observations identify a mechanism in which concentration-dependent modulation of disparate cytokine-mediated pathways constitutes a molecular switch for neuronal versus glial differentiation of retinal stem cells/progenitors.

Summary

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Our in vitro observation posits a model in which single cytokines, elaborated by differentiating cells, can act as a molecular switch to generate neurons or glia (Fig. 7I). The source of the cytokines is the differentiating Müller cells. The late retinal stem cells/progenitors and precursors are exposed to a low concentration of the cytokine that preferentially activates the MAPK over the Jak-STAT pathways, facilitating neuronal differentiation. The activated MAPK pathway keeps the Jak-STAT and Notch signaling inhibited, thus accentuating neurogenesis. As Müller cells accumulate and/or begin to elaborate more cytokine over time, the microenvironment changes quantitatively, and as a result retinal progenitors and precursors are progressively exposed to higher concentrations of cytokines, activating the Jak-STAT pathway in preference to the MAPK pathway. This pathway has a feed-forward loop and positively regulates Notch signaling. Differentiation is shifted toward gliogenesis as these pathways are progressively strengthened. These observations suggest that the inhibition of the Jak-STAT and Notch pathways may constitute an approach to enhancing the efficiency of directed differentiation of retinal stem cells/progenitors or a subset of Müller cells with latent stem cell properties into specific neuronal types, namely photoreceptors, cells vulnerable in AMD and RP.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Thanks are due to Drs. Anirvan Ghosh for GFAP promoter reporter constructs, Yi Sun for dnMAPK and dnSTAT3 expression constructs, Vijay Sarthy for GFAP-GFP mice, Robert Molday for rho 4D2 antibody, Rakesh Singh for help with ELISA, Anathbandhu Chaudhuri for help with microinjection, and Katie Brown for technical help. This work was supported by the Lincy Foundation, the Pearson Foundation, and the Nebraska Tobacco Settlement Biomedical Research Development.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
SC-08-0222_Suppl_Figure_1.pdf121KSupplemental Figure 1
SC-08-0222_Suppl_Figure_2.pdf152KSupplemental Figure 2
SC-08-0222_Suppl_Figure_3.pdf65KSupplemental Figure 3
SC-08-0222_Suppl_Figure_Legends.pdf15KSupplemental Figure Legends

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.