Notch Signaling Activates Stem Cell Properties of Müller Glia through Transcriptional Regulation and Skp2-mediated Degradation of p27Kip1

Müller glia (MG), the sole glial cells generated by retinal progenitors, have emerged as a viable cellular target for therapeutic regeneration in degenerative blinding diseases, as they possess dormant stem cell properties. However, the mammalian MG does not display the neurogenic potential of their lower vertebrate counterparts, precluding their practical clinical use. The answer to this barrier may be found in two interlinked processes underlying the neurogenic potential, i.e., the activation of the dormant stem cell properties of MG and their differentiation along the neuronal lineage. Here, we have focused on the former and examined Notch signaling-mediated activation of MG. We demonstrate that one of the targets of Notch signaling is the cyclin-dependent kinase inhibitor (CKI), p27Kip1, which is highly expressed in quiescent MG. Notch signaling facilitates the activation of MG by inhibiting p27Kip1 expression. This is likely achieved through the Notch- p27Kip1 and Notch-Skp2-p27Kip1 axes, the former inhibiting the expression of p27Kip1 transcripts and the latter levels of p27Kip1 proteins by Skp2-mediated proteasomal degradation. Thus, Notch signaling may facilitate re-entry of MG into the cell cycle by inhibiting p27Kip1 expression both transcriptionally and post-translationally.


Introduction
The retina, an integral part of the central nervous system (CNS), does not display active neurogenesis under normal conditions in adult mammals. However, when injured it displays neurogenic potential, which can be traced to the major glial cell type of the retina, MG [1][2][3].

Materials and Methods Animals
All experiments were conducted in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research, and were approved by the Institutional Animal Care and Use Committee (IACUC) at University of Nebraska Medical Center (protocols: # 95-005-09 and # 97-100-08). Various ages of Sprague-Dawley (SD) rats, beginning from embryonic day 18 (E18) to postnatal day 21 (PN21), were purchased from Charles River Laboratories (Wilmington, MA). Animals were housed in the Department of Comparative Medicine at the University of Nebraska Medical Center. Rats were euthanized by CO 2 exposure followed by decapitation using sterile surgical scissors to ensure death.

Viral vectors and lentivirus production
The p27 Kip1 shRNA lentivirus construct was obtained from Thermo Scientific (Waltham, MA). The Skp2 siRNA lentivirus construct was obtained from Applied Biological Materials Inc. (ABM, Richmond, BC, Canada). The control GFP lentivirus construct was obtained from Addgene (Cambridge, MA). In the p27 Kip1 shRNA and Skp2 siRNA lentivirus constructs, the expression of sh/siRNA sequences were driven by U6 promoters. In GFP lentivirus constructs, the CMV promoter drove expression of the GFP sequence. Lentivirus preparation and transduction were as previously described [27]. Briefly, the recombinant lentiviral particles were generated using the ABM lentivirus packaging system, followed by transient transfection of 293T cells, and subsequently concentrated using BioVision PEG lentivirus precipitation kit (Milpitas, CA). Virus titer was determined using the ABM lentivirus titration kit. Enriched MG and retinal explants were transduced with lentiviruses with the multiplicity of infection (MOI) of 2-4, overnight. Viruses were removed the following morning. The specificity of LOF assays by virus-mediated sh/siRNA expression was determined in C6 glioma cells by Western and Q-PCR analyses. The efficiency of transduction in MG was ascertained by simultaneously cotransducing sh/siRNA and GFP lentivirus, where the GFP + cells should reflect the sh/siRNA transduction efficiency.

Explant assays
The explant assays were carried out as previously described [14]. Briefly, retinas removed from PN21 SD rats were placed on a semi-permeable membrane (0.4 mm pore size; Millipore, Temecula, California, USA), with the ganglion cell layer (GCL) oriented upward, and were maintained in RCM with 5% fetal bovine serum (FBS; Hyclone, Canada). Explants were transduced with p27 Kip1 shRNA or Skp2 siRNA or control lentivirus (overnight, with removal of viral media the following morning) and cultured in the presence of Jag1 (20μM) or Jag1+MG132 (2μM) for 5 days. Before explants were collected for analyses on day 5, they were exposed to 5-Bromo-2-deoxyuridine (BrdU, Sigma Aldrich, St Louis, MO, USA, 10μM) for 8 hours. To determine the efficiency of the LOF approach, explants were simultaneously cotransduced with a GFP lentivirus and either p27 Kip1 shRNA or Skp2 siRNA lentivirus.

Cell cycle analysis by flow cytometry
Cell cycle analysis of activated MG was carried out as previously described [29]. Briefly, cells were fixed in 70% ethanol at 4°C for 30 minutes, followed by incubations at 4°C for 30 minutes in Telford reagent containing the DNA-intercalating dye, propidium iodide. Data were acquired by flow cytometry and analyzed by ModFitLT V3.3.11 (Windows).

Quantitative polymerase chain reaction (qPCR)
Total RNA was isolated using Mini RNeasy Kit (Qiagen, Hilden, Germany), and cDNA was synthesized as previously described [30]. Specific transcripts were amplified with gene-specific forward and reverse primers (Table 1) by using Quantifast SYBR Green PCR kit (Qiagen, Hilden, Germany) and a RotorGene 6000 (Corbett Robotics, San Francisco, CA, USA). Results obtained from qPCR were generated by measuring each sample in triplicate; no-template blanks were used as negative controls. Amplification curves and gene expressions were normalized to the housekeeping gene, glyceraldehyde phosphate dehydrogenase (GAPDH).

Chromatin immunoprecipitation (ChIP) analysis
ChIP analysis was carried out as previously described [31]. Antibody-specific information for ChIP (as well as Western and immunofluorescence analyses) can be found in Table 2. Briefly, cells were first cross-linked and serially quenched with 1% formaldehyde and glycine, respectively, followed by immunoprecipitation using anti-Hes1/anti-CSL antibodies and Chromatin Immunoprecipitation kit (Magna CHIP HiSens, Millipore, Germany) as per manufacturer's instruction. For controls, immunoprecipitation was carried out with specific immunoglobulin G (IgG) antibodies. After proteinase and RNAse A digestion, the precipitated DNA was purified using a Qiaquick PCR purification kit (Qiagen Inc). Regular PCR was carried out using promoter-specific primers (Table 1), followed by the resolution of the PCR products by gel electrophoresis.

Immunofluorescence analysis
Immunofluorescence analysis was performed as previously described [5]. Briefly, 4% paraformaldehyde-fixed cells or explant cryosections (12μm thickness) were blocked in blocking solution, followed by an overnight incubation with specific antibodies (Table 2). Cells were examined for epifluorescence after incubation with anti-species-specific IgG conjugated to Cy3 or FITC (Life Technologies). Samples were mounted using VectaShield (Vector Laboratories, Burlingame, CA) and images were taken using a Zeiss AX10 fluorescence microscope and AxioVision Rel. 4.8 software. For quantification of the percentage of specific cell types in each experiment, the numbers of cell-specific antigen-positive cells were counted in 15 randomly selected fields (3 cover slips, 5 fields each), in triplicate. Adobe Photoshop CS4 was used for image presentation. Manipulation of the images was restricted to threshold and brightness adjustments to the whole image. Controls for the experiments consisted of the omission of primary antibodies. No staining was observed in these cases.

Hoechst dye efflux assay (HDEA)
Neurospheres were dissociated into single cells and Hoechst dye efflux assays (HDEA) were carried out as previously described [5,31]. Neurosphere dissociates were suspended in Hoechst Iscove's modified Dulbecco's medium (IMDM) supplemented with 2% FBS and 1mM HEPES at 4°C overnight. Cells were stained with Hoechst 33342 (3.0 μg/ml) for 30 min at 37°C. Verapamil (100 μM), an inhibitor of the efflux pump called the ABCG2 transporter, and propidium iodide were used as negative and dead cell controls, respectively. The SP and non-SP regions were defined on Hoechst-low and Hoechst-bright, based on the Hoechst blue and red axes, respectively.

Statistical analysis
Cell type-specific antigen-positive cells were counted in 10-15 randomly selected fields in three to five different coverslips (cell culture) or sections (explants). Each experiment was repeated at least three times. Values were expressed as mean ± SEM. Data were analyzed using the Student's t-test or ANOVA to determine the significance of the differences between treatment and control groups. QPCR results were based on the individual gene expression in comparison to the respective control of the experimental group.

Evidence of Notch-dependent regulatory axes during late retinal histogenesis and MG activation
To determine whether or not the proposed Notch-dependent regulatory axes exist during late retinal histogenesis, when MG are generated, we examined the temporal patterns of expression of the axes' components against those encoding the regulators of progenitors and MG. During late histogenesis, retinal progenitors are progressively depleted as they differentiate into late born retinal cells. Accordingly, levels of transcripts corresponding to Nestin, a neural progenitor marker, and Hes1, the Notch-dependent regulator of retinal progenitors, decreased with time, compared to those at the beginning of late histogenesis, i.e., the embryonic day 18 (E18) stage ( Fig 1A and 1B). In contrast, a reciprocal temporal increase in levels of MG-specific transcripts, glutamine synthetase (GS), was observed, demonstrating the expected inverse temporal relationship between the loss and gain of progenitor and MG phenotypes, respectively ( Fig 1C). Next, we examined the temporal expression patterns of p27 Kip1 and Sox9, both associated with retinal progenitors during retinal histogenesis and matured MG [28,[32][33][34][35][36]. Levels of p27 Kip1 and Sox9 transcripts decreased with time, however, the trend in temporal decrease plateaued around PN10 and increased there after such that significant transcript levels were detected at PN20, compared to those at PN10 (Fig 1D and 1E). Levels of transcripts corresponding to Skp2, a direct target of Notch signaling [24], temporally decreased in parallel with the attenuation in the levels of Hes1 (Fig 1F). These observations suggested that the components of Notch-dependent regulatory axes, i.e., Notch-p27 Kip1 and Notch-Skp2-p27 Kip1 , were expressed during late retinal histogenesis, which may be recruited for MG activation. We tested this hypothesis in enriched MG culture, a robust model for testing the activation and progenitor properties of MG when coupled with neurosphere assays [5,14]. We first examined the expression of the components of the proposed axes in enriched MG; immunocytochemical analysis revealed glutamine synthetase (GS) + MG co-expressing immunoreactivities corresponding to Sox9, p27 Kip1 , Hes1, Skp2, and NICD (activated Notch1) ( Fig 1G). Next, we examined the perturbation of Notch signaling in FGF2-mediated neurosphere assays [5]. To activate Notch signaling, we cultured MG in the presence of Jag1, a 17-aminoacid peptide, corresponding to a region in the CSL domain of Notch ligand, Jagged1 (Jag1) [14]. To attenuate Notch signaling, MG were cultured in the presence of DAPT, a small molecule that compromises the generation of NICD by inhibiting gamma-secretase [14]. By the end of five days in control culture conditions, a subset of MG generated neurospheres (Fig 2). Cells in neurospheres expressed Nestin and Sox9, and incorporated BrdU, thus displaying their progenitor nature (Fig 2A). However, when MG were cultured in the presence of Jag1, they generated neurospheres which were significantly larger and more numerous, compared to controls (Fig 2B, 2C and 2D). In contrast, when DAPT was added to the culture medium, MG gave rise to significantly smaller and fewer neurospheres, compared to those in either Jag1 group or controls (Fig 2B, 2C and 2D). Quantification of neurosphere cells after 8 hour exposure to BrdU in the culture medium revealed a significant increase and decrease in the number of BrdU + cells in the Jag1 and DAPT groups, respectively, versus controls (Fig 2E), corroborating effects of Jag1/DAPT on MG proliferation as ascertained by changes in the frequency and size of neurospheres. That the effects on MG proliferation were due to the perturbation in Notch signaling was demonstrated by a relative increase and decrease in the levels of Hes1 transcripts in the presence of Jag1 and DAPT, respectively, compared to controls ( Fig  2F). The effects of Notch signaling on MG proliferation is likely due to its influence on cell cycle regulation [24,25]. To test this premise, we examined the effects of perturbation in Notch signaling on cell cycle by flow cytometry. We observed a relative increase in the proportions of cells in the S phase in Jag1 group versus controls, which decreased remarkably when cells were exposed to DAPT (Fig 2G). The increase in the ratio of the proportion of cells in G1/G2 and S phase between Jag1 and DAPT groups revealed that DAPT led to the arrest of MG at both G1 and G2 phases, suggesting the influence of Notch signaling on cell cycle regulation during the activation of MG ( Fig 2H). Next, we were interested in knowing whether or not Notch signaling influenced the dormant stem cell properties of the activated MG. Another attribute, besides the capacity to generate clonal neurospheres of activated MG, is their ability to emerge as side population (SP) cells when subjected to the Hoechst dye efflux assay (HDEA) [5]. SP cell phenotype is a universal characteristic of stem cells/progenitors [37][38][39]. We observed a relative increase in the proportion of MG-SP cells in the presence of Jag1 versus controls, which decreased in the presence of DAPT ( Fig 2G). Together, these results suggested that Notch signaling influenced the proliferation, cell cycle regulation, and stem cell properties of the activated MG.
Influence of Notch signaling on p27 Kip1 and Skp2 expressions during MG activation Next, we examined whether or not Notch signaling influenced the expression of p27 Kip1 and Skp2 during the activation of MG. Since the premise required Notch signaling to influence their expression at transcriptional and post-translational levels, we examined both their transcript and protein expressions. We first determined the effects of Notch signaling on levels of p27 Kip1 and Skp2 transcripts. We observed that the levels of transcripts corresponding to p27 Kip1 and Skp2 decreased and increased significantly in MG-derived neurospheres exposed to Jag1, respectively, compared to controls (Fig 3A and 3B). Exposure of neurospheres to DAPT abrogated the effects of Jag1 on the expression of p27 Kip1 and Skp2; DAPT significantly increased and decreased p27 Kip1 and Skp2 transcript levels, respectively, compared to those in  Notch-Mediated Muller Glia Activation Jag1-treated groups, thus restoring to those in controls. The lack of significant difference in p27 Kip1 and Skp2 transcript levels, when compared between controls and DAPT groups, is likely due to the stability of NICD in glial cells (see below) and the fact that p27 Kip1 and Skp2 are regulated by other mechanisms in addition to Notch signaling. Next, to test the premise that p27 Kip1 and Skp2 expression are under Notch-mediated transcriptional repression and activation, respectively, we examined their proximal promoters for the presence of Hes1/CSL binding sites. Genes targeted by Notch signaling for repression, such as p27 Kip1 , contains Cbox or class C site (CACGCG) elements where Hes1 forms transcriptional repression complexes (Fig 3B) [23]. Conversely, the direct target of Notch genes, such as Hes1/Skp2, contains CSL binding sites (TGGGAA) where NICD forms a complex for transcriptional activation ( Fig  3E) [24]. To demonstrate that these sites were occupied by Hes1/CSL in MG, and therefore were targets of Notch-mediated transcriptional regulation, we carried out ChIP analyses of p27 Kip1 /Skp2 promoters using Hes1/CSL antibodies (Fig 3D and 3F). We observed that Hes1 and CSL antibodies precipitated complexes from MG nuclei, containing DNA sequences corresponding to p27 Kip1 (Fig 3D) and Skp2 (Fig 3F) proximal promoters, the class C site and CSL binding sites, respectively. These DNA sequences with specific binding sites were absent in the ChIP assays carried out using IgG. Together, these observations suggested that both p27 Kip1 and Skp2 are targeted by Notch signaling during MG activation.
Next, we examined whether or not Notch signaling influenced p27 Kip1 and Skp2 expression post-translationally. We examined levels of these proteins in response to perturbation of Notch signaling against the backdrop of their levels in enriched MG by Western analyses, carried out on pooled samples of cells/neurospheres because of the cell number limitations (Fig  4). A semi-quantitative analysis of the intensity of immunoreactive bands corresponding to specific protein levels in different conditions with respect to those at E18 revealed the following: robust levels of p27 Kip1 were observed in enriched MG, which decreased by more than 3 fold in FGF2 and Jag-1-treated groups (p27 Kip1 /GAPDH: 1:00 versus 0.30 and 0.25, respectively) (Fig 4A). p27 Kip1 levels remained decreased in DAPT-treated group relative to that in E18. However, when compared to Jag-1-treated groups, p27 Kip1 levels were observed to be relatively higher (p27 Kip1 /GAPDH: 0.25 versus 0.35). In contrast, levels of Skp2 increased in FGF2 and Jag-1 groups, compared to that in E18 (Skp2/GAPDH: 1:00 versus 1.30 and 2.93, respectively) ( Fig 4B). Skp2 levels in DAPT-treated groups remained lower versus all groups; they were more than 3 fold lower when compared to that in the FGF2 and Jag-1-treated groups (Skp2/GAPDH: 0.42 versus 1.30 and 2.93, respectively). That Notch signaling was perturbed in the experimental paradigm was demonstrated by an increase in NICD levels in FGF2/Jag-1-treated groups, compared to those in MG (Fig 4C). The decrease in NICD levels in response to DAPT treatment was observed in relation to FGF2/Jag-1 treatment. The detection of NICD in the presence of DAPT suggested an altered post-translational regulation for persistent Notch signaling, presumably required for the maintenance of glial phenotype. For example, the loss of function of Fbxw7 (F-box-and WD-repeat domain containing protein 7), which is required for targeted degradation of NICD, leads to accumulation of NICD with skewed differentiation of neural stem cells into astrocytes [40]. The accumulation of NICD may directly affect Skp2 through transcriptional regulation and maintain its levels ( Fig 3A) but not Skp2 protein, which might undergo normal degradation to have lower levels (Fig 4B,  graph). To address whether or not the lack of difference in relative levels of p27 Kip1 and Skp2 proteins in FGF2 and Jag-1 groups were bonafide or due to the pooling of cells, we carried out quantitative Western analysis. Because of cell limitations, particularly in the DAPT group, where size and numbers of neurospheres were low, the analysis included only two groups. We observed a significant decrease and increase in levels of immunoreactivities corresponding to p27 Kip1 (Fig 4D) and Skp2 (Fig 4E) in the Jag1 group, respectively, compared to the FGF2 group. Together, these observations suggested that Notch signaling-mediated transcriptional and post-translational regulation of p27 Kip1 and Skp2 might play a role in the activation of MG.

Involvement of p27 Kip1 in Notch-dependent activation of MG
To test whether or not p27 Kip1 is involved in Notch-dependent activation of MG, we examined the effects of Jag1 on the activation of MG where levels of p27 Kip1 were experimentally attenuated. The loss of function (LOF) approach was preferred over the gain of function (GOF) approach, as the latter may lead to non-specific effects. Given the cell number limitation that precluded Western analysis in MG, we tested the efficiency and specificity of p27 Kip1 LOF in C6 glioma cells. We observed that p27 Kip1 shRNA lentivirus-mediated transduction caused a reproducible and significant decrease in levels of both proteins and transcripts corresponding to p27 Kip1 , compared to controls (Fig 5A and 5B). Next, enriched MG were transduced with p27 Kip1 shRNA or control lentivirus and subjected to neurosphere assays in the presence of Jag1 peptide. We measured the generation frequency of neurospheres and the proportion of SP cells to evaluate the effects of the p27 Kip1 LOF on proliferation and stem cell property of the activated MG, respectively. We observed a significant decrease in p27 Kip1 transcripts levels in the p27 Kip1 LOF group, compared to controls (Fig 5C), which was correlated with a significant increase in the number of neurospheres (Fig 5D), suggesting that the loss of p27 Kip1 expression further facilitated Notch-mediated generation of neurospheres. The neurospheres were subjected to HDEA to measure the proportion of SP cells by flow cytometry. The proportion of MG-SP cells was observed to have almost doubled in the p27 Kip1 LOF group, compared to controls (0.26% versus 0.15%), suggesting an increase in the number of MG cells with stem cell property when the function of p27 Kip1 was compromised ( Fig 5E).
Next, to determine that the Notch-dependent effects on the activation of enriched MG is not a function of the enrichment protocol that entails the exposure of MG to hypoxia and prolonged culture conditions, we carried out p27 Kip1 LOF experiments in adult retinal explants, where in vivo cell-cell interactions are maintained and functional [14,39]. Adult SD rat retinal explants were transduced with p27 Kip1 shRNA or control lentivirus and cultured in retinal culture medium. BrdU was added in the last 8 hours before the termination of the culture [14]. Immunohistochemical analyses of retinal explants revealed a decrease in immunoreactivities corresponding to p27 Kip1 in the p27 Kip1 LOF group, compared to controls (Fig 6A). In contrast, we observed an increase in immunoreactivities corresponding to Sox9 and those detecting the incorporated BrdU, compared to controls (Fig 6B). These immunoreactivities, representing MG and proliferating cells, respectively, were largely confined to the inner nuclear layer and were co-localized. Since proliferating MG tend to migrate in retinal explant or in response to injury in vivo, some of these cells were also localized in the ONL [14]. Quantification of Sox9 + BrdU + cells (= activated MG), revealed their numbers to be significantly increased in the p27 Kip1 LOF group versus controls (Fig 6C). To further determine the specificity of the LOF approach, we co-transduced retinal explants with p27 Kip1 shRNA lentivirus+GFP lentivirus/ GFP lentivirus and cultured them as above. Explants were dissociated, transduced MG were identified as GFP epifluorescent cells, which co-expressed immunoreactivities corresponding to BrdU and glutamate/aspartate transporter (GLAST), an MG-specific marker (Fig 6D). Quantification of GFP + GLAST + BrdU + cells revealed a significantly higher proportion in the p27 Kip1 LOF group versus controls, suggesting along with the above results, that the loss of p27 Kip1 function facilitates Notch signaling-mediated activation of MG (Fig 6E). Rare BrdU + cells that were negative for MG markers, were excluded from quantification as they could represent microglia.

Involvement of Skp2 in Notch-dependent activation of MG
Next, we used the LOF approach, as described above, to determine whether or not Skp2 was involved in Notch-dependent activation of MG. First, we examined the efficiency and specificity of the Skp2 LOF approach in C6 glioma cells, where Skp2 siRNA lentivirus-mediated transduction led to a reproducible decrease in Skp2 protein ( Fig 7A) and transcript (Fig 7B) levels, compared to controls. Second, when enriched MG were transduced with Skp2 siRNA or control lentivirus and subjected to neurospheres assay in the presence of Jag1, a significant decrease in Skp2 transcripts levels was observed, compared to controls (Fig 7C). Correspondingly, we observed an increase in the levels of p27 kip1 protein in Skp2 siRNA-treated neurospheres, compared to controls, demonstrating the regulatory influence of Skp2 on p27 kip1 expression (Fig 7D). The decrease in Skp2 expression was correlated with a significant decrease in numbers of neurospheres (Fig 7E), suggesting that the reduction of Skp2 expression abrogated Notch-mediated generation of neurospheres. Since Skp2 mediates proteasomal degradation of target proteins such as p27 Kip1 [24], it was expected that the pharmacological inhibition of proteasomes would have similar effects as the inhibition of Skp2 expression. To test this To determine the specificity of the p27 Kip1 LOF approach, retinal explants were co-transduced with p27 Kip1 shRNA and GFP lentiviruses/GFP lentivirus. Explants were dissociated, and transduced MG were identified as GFP + cells which had incorporated BrdU and displayed immunoreactivities to GLAST. The lentivirus transduction efficiency is illustrated by transduced GFP + cells (arrows), co-localized with immunoreactivities corresponding to BrdU and GLAST. (E) Immunocytochemical analysis and quantification of GFP + BrdU + GLAST + cells revealed their proportion to be significantly higher in the p27 Kip1 LOF group versus controls. p<0.005. Scale bar = 25μm. Data are mean ± SEM. Experiments were carried out three times in triplicates with 4-6 retina per group for ex-vivo perturbation. Broken lines distinguish the outer nuclear layer from inner nuclear layer. Notch-Mediated Muller Glia Activation premise, enriched MG were subjected to neurosphere assay in the presence of proteasomal inhibitor MG132 [41]. The presence of MG132 led to a significant decrease in the generation of neurospheres, compared to controls, suggesting that the effects of Notch signaling at the posttranscriptional levels involved proteasomal degradation of proteins ( Fig 7F). Third, we carried out HDEA on Skp2/control siRNA lentivirus-transduced MG-derived neurospheres. We observed 0.15% of MG-SP cells in Jag1-treated neurospheres, the proportion of which decreased to 0.07% and 0.08% when MG were pre-transduced with Skp2 siRNA lentivirus or cultured in the presence of MG132, respectively ( Fig 7G).
Lastly, to rule out that the observed effects were peculiar to enriched MG, Skp2 LOF experiments were carried out on retinal explants, as described above, where the transduction with Skp2 siRNA led to a decrease in Skp2 immunoreactivities in retinal explants compared to controls (Fig 8A). In contrast to the results obtained by p27 Kip1 LOF approach (Fig 6A), immunohistochemical analysis of transduced explants revealed a significant decrease in the number of Sox9 + BrdU + cells in Skp2 LOF groups, compared to controls, revealing the negative influence of Skp2 LOF on the Notch-mediated activation of MG (Fig 8B and 8C). That the stabilized p27 Kip1 might have mediated the inhibitory effects was revealed by the increase in p27 Kip1 immunoreactivities in Skp2 siRNA lentivirus transduced explants versus controls (Fig 8D). Skp2 LOF experiments involving co-transduction with GFP lentivirus to determine the specificity of the approach as above (Fig 8E) revealed a significant decrease in the number of GFP + GLAST + BrdU + cells, versus controls ( Fig 8F). Together, these observations suggested that the Notch signaling-mediated accentuation of Skp2 expression was involved in MG activation.

Discussion
A variety of approaches involving different species suggests that MG, besides maintaining retinal homeostasis, may act as progenitors, capable of supporting neurogenesis in adults [1][2][3]. The evidence emerging from teleosts, particularly transgenic zebrafish, is robust for the neurogenic potential of MG; they serve as progenitors for photoreceptors during normal development and are recruited in response to injury for regeneration [2,3]. While such regenerative properties of MG were not observed in the mammalian retina, they have been observed to respond to injury by re-entering the cell cycle [42,43]. Given their regenerative capacity in lower vertebrates and biochemical resemblance with radial glia, examination of MG enriched from rodent retina revealed that a subset of these cells possessed the ability to self renew and differentiate along the neuronal lineage [5]. It was observed that the display of stem cell properties by MG in vitro was accompanied by the activation of Notch signaling, and that its inhibition significantly compromised the abilities of these cells to generate clonal neurospheres in culture [5] or incorporate BrdU in vivo in response to injury [5,14]. The involvement of Notch signaling in the regulation of MG is also observed in zebrafish regeneration, where the temporal sequence of the activation of Notch signaling is better understood than in mammals [2]. For example, molecular examination of gfap-EGFP-labeled MG, enriched after different time intervals following a light induced lesion in transgenic zebrafish, revealed that during the initial stages of the injury, when MG have not yet divided, the indices of Notch signaling were decreased [44]. However, following injury, a stage associated with MG division, the expression of the components of Notch signaling were up-regulated, suggesting an association of Notch signaling with MG proliferation, as observed in mammals.
Among the multiple signaling pathways that are recruited for the development and differentiation of the retina, the one that is more appropriately placed, enabling MG to sense and respond to injury, is the Notch signaling. It is used in its classical function to regulate cell commitment and also to promote proliferation of progenitor/precursor populations, as observed in a variety of tissues [45]. In the developing retina, Notch receptor activation ensures that a subset of progenitors remain uncommitted during early histogenesis, presumably to sustain the prolonged cell type-specific differentiation. During late histogenesis, Notch receptors are recruited to promote differentiation of MG, and therefore, when Notch signaling is experimentally perturbed, MG generation is adversely impacted [46]. How does Notch function change To determine the specificity of from maintaining progenitor populations during early histogenesis to promoting differentiation of glia in late histogenesis remains poorly understood. It is thought that the strengthening of the JAK-STAT pathway in retinal progenitors, at the behest of cytokines secreted by differentiating cells during late histogenesis, may underlie the gliogenic effects of Notch signaling [47,48]. Evidence based on the spatial localization of the expression of Notch receptor in adult retina suggests that Notch signaling may remain active in MG or at least in a subset of these cells [49][50][51][52]. The role of Notch signaling in adult MG remains unexplored, however, current observations suggest that it is involved in cell cycle regulation to sustain the regenerative role of MG [1,2]. For example, experimental accentuation of Notch signaling in S334 rats in the beginning of rod photoreceptor degeneration results in MG proliferation and migration to the ONL where a minor subset of these cells begin to express markers corresponding to degenerated neurons [14]. Similar influence of Notch signaling on MG proliferation and regenerative potential has been observed in chicks [51][52][53]. Here, we have demonstrated the possible mechanism by which the activation of Notch signaling is linked to MG proliferation. While myc and cyclins have been demonstrated to be the direct targets of Notch signaling for G1-S transition in Notch-activated cells [54], it is likely that the proliferative influence in MG is mediated through p27 Kip1 (Fig 9). p27 Kip1 is up regulated during histogenesis to promote committed progenitors/precursors exit from the cell cycle [28,36]. Its expression attenuates in differentiated cells but persists in MG as one of its characteristic molecular features [5,55]. While the maintenance of high levels of p27 Kip1 expression in terminally differentiated cells is rather enigmatic, it could be suggested that in MG, it is employed to keep their evolutionarily conserved propensity to re-enter the cell cycle [1]. In this model, Notch signaling, influenced by adjacent cells and/or by Mash-1 based cell intrinsic machinery observed in zebrafish [56], abrogates the levels of p27 Kip1 proteins by acting through the Notch-p27 Kip1 and Notch-Skp2-p27 Kip1 axes. The recruitment of these axes, one inhibiting the transcription of p27 Kip1 and the other leading to the Skp2-mediated degradation of p27 Kip1 proteins, may have evolved for a rapid and sustained proliferative response of MG to injury.
Our observations shed light on the possible mechanism of Notch signaling-mediated activation of MG, and not when it is employed or how it is linked to the neural conversion of MG. If one leans upon the evidence emerging from zebrafish [2,20,21,56], it might not be implausible to consider that sometime during the early response of MG to injury, Notch signaling is recruited to facilitate a cell proliferation response. This window is likely to be narrow and with the change in context, spatial (migration out of the INL) and/or temporal, the effects of Notch signaling would also change. If Notch signaling persists when MG are poised for neuronal conversion, it might be deleterious for the regenerative process, given the inhibitory influence of Notch signaling on the expression of pro-neural genes. Therefore, inhibiting Notch signaling at this stage has been observed to expand the regenerative field [56] as well as neuronal conversion of MG [1,2]. Similar temporal manipulation of Notch signaling, directly or indirectly, by targeting the Notch-dependent molecular axes, may help recruit MG for regeneration in the mammalian retina.
the Skp2 LOF approach, retinal explants were co-transduced with Skp2 shRNA and GFP lentiviruses/GFP lentivirus. Explants were dissociated, and transduced MG were identified as GFP + cells which had incorporated BrdU and displayed immunoreactivities to GLAST. The lentivirus transduction efficiency is illustrated by transduced GFP + cells (arrows), co-localized with immunoreactivities corresponding to BrdU and GLAST. (E) Immunocytochemical analysis and quantification of GFP + BrdU + GLAST + cells revealed their proportion to be significantly higher in the Skp2 LOF group versus controls. p<0.005. Scale bar = 25μm. Data are mean ± SEM. Experiments were carried out three times in triplicates with 4-6 retina per group for ex-vivo perturbation. Broken lines distinguish the outer nuclear layer from inner nuclear layer. doi:10.1371/journal.pone.0152025.g008 Notch-Mediated Muller Glia Activation