SWI/SNF Chromatin Remodeling ATPase Brm Regulates the Differentiation of Early Retinal Stem Cells/Progenitors by Influencing Brn3b Expression and Notch Signaling*

Based on a variety of approaches, evidence suggests that different cell types in the vertebrate retina are generated by multipotential progenitors in response to interactions between cell intrinsic and cell extrinsic factors. The identity of some of the cellular determinants that mediate such interactions has emerged, shedding light on mechanisms underlying cell differentiation. For example, we know now that Notch signaling mediates the influence of the microenvironment on states of commitment of the progenitors by activating transcriptional repressors. Cell intrinsic factors such as the proneural basic helix-loop-helix and homeodomain transcription factors regulate a network of genes necessary for cell differentiation and maturation. What is missing from this picture is the role of developmental chromatin remodeling in coordinating the expression of disparate classes of genes for the differentiation of retinal progenitors. Here we describe the role of Brm, an ATPase in the SWI/SNF chromatin remodeling complex, in the differentiation of retinal progenitors into retinal ganglion cells. Using the perturbation of expression and function analyses, we demonstrate that Brm promotes retinal ganglion cell differentiation by facilitating the expression and function of a key regulator of retinal ganglion cells, Brn3b, and the inhibition of Notch signaling. In addition, we demonstrate that Brm promotes cell cycle exit during retinal ganglion cell differentiation. Together, our results suggest that Brm represents one of the nexus where diverse information of cell differentiation is integrated during cell differentiation.

Cell fate specification and subsequent cell differentiation in the nervous system are orchestrated and finessed by interplay between cell intrinsic and cell extrinsic factors. This process is exemplified during the development of the retina, an excellent model of the central nervous system. Recent evidence suggests that disparate transcription factors belonging to basic helixloop-helix, homeodomain, and zinc finger classes cooperate toward lineage specification and differentiation (1)(2)(3)(4)(5)(6). For example, the basic helix-loop-helix transcription factor, Math5, and the homeodomain transcription factor, Pax6, have been shown to cooperate during the specification of retinal progenitors into retinal ganglion cells (RGCs) 2 (2,(7)(8)(9). As progenitors are committed along RGC lineage, Wt1, a zinc finger transcription factor, and Brn3b, a homeodomain transcription factor of POU class, are expressed and ultimately promote the differentiation and maturation of the specified progenitors into RGCs (10). The progress in the identification of intrinsic factors has been paralleled by the characterization of extrinsic factors and intercellular signaling pathways, e.g. Notch pathway, that mediates the regulatory influence of the microenvironment on retinal progenitors' maintenance and their differentiation into RGCs (11)(12)(13)(14)(15).
Although the identification of cell intrinsic and cell extrinsic factors has helped our understanding of the mechanisms that regulate the differentiation of retinal cell types, particularly that of RGCs, it is not known as to how the remodeling of chromatin, which is necessary for eukaryotic gene expression, is recruited toward the coordinated regulation of genes during RGC differentiation. There are two major classes of chromatin remodeling complexes, that are associated with specific enzymes that remodel chromatin by changing nucleosome structure or position (16,17). The first class includes complexes consisting of histone acetyltransferase or histone deacetylase that covalently modify chromatin by adding and removing acetyl groups from the amino terminus of the four core histones, respectively. The second class includes SWI/SNF complexes that utilize the energy obtained from ATP hydrolysis to disrupt nucleosomal structure or position. The mammalian SWI/SNF complexes consist of 10 -12 proteins, including the homologous but mutually exclusive ATPases, Brahma (Brm) and Brm related gene 1 (Brg1) (16,18).
Although both major classes of chromatin remodeling complexes are likely to contribute to developmental processes, including those in the central nervous system (19 -21), evidence is emerging that the SWI/SNF chromatin remodeling complexes play an important role in the differentiation of specific cell types (22). For example, these complexes have been shown to facilitate the differentiation of a variety of cell types such as erythrocytes (7,23), macrophages (24), myeloid cells (25), adipocytes (26), myoblasts (27), osteoblasts (28), neurons (29,30), and glia (31). Here we demonstrate the role of Brm in the differentiation of retinal progenitors into RGCs. Brm is expressed in the developing rat retina, and its temporal and spatial patterns of expression correlate with retinal histogenesis. Using perturbation of expression and function analyses, we demonstrate that Brm influences the differentiation of retinal progenitors into RGCs by facilitating Brn3b expression and function and inhibiting Notch signaling. In addition, we demonstrate that Brm may influence differentiation generically by promoting cell cycle exit. Together, our results suggest that chromatin remodeling by Brm may represent one of the nexus where cell intrinsic and cell extrinsic influence may be integrated toward the differentiation of retinal progenitors.

EXPERIMENTAL PROCEDURES
Progenitor Cell Culture-Timed-pregnant (E14) Sprague-Dawley rats were obtained from Sasco (Wilmington, MA). The gestation day was confirmed by the morphological examination of embryos (32). Fertilized hen eggs were incubated in a humidified chamber at 38°C, and embryos were staged according to Hamburger and Hamilton (33). Embryos were harvested at appropriate gestation periods, and eyes were enucleated. Retina were dissected out and dissociated as described previously (34). Cells were cultured in RCM (Dulbecco's modified Eagle's medium/F-12, 1ϫ N2 supplement (Invitrogen), 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin) containing FGF2 (10 ng/ml), epidermal growth factor (20 ng/ml), and 0.1% fetal bovine serum for 5 days to generate clonal neurospheres. 5-Bromo-2-deoxyuridine (BrdUrd) (10 M) was added to the culture for the final 24 h. For co-culture, neurospheres were collected, washed extensively to remove BrdUrd, and plated on poly-D-lysineand laminin-coated glass coverslips with E3 chick/PN1 rat retinal cells in 1% fetal bovine serum. For RT-PCR analysis co-culture was performed across a 0.4-m membrane (Millipore, Bedford, MA). Cells were either frozen for RNA extraction or fixed with 4% paraformaldehyde for 15 min at 4°C for immunofluorescence analysis.
Cell Cycle Analysis-The DNA content of the retinal cells was measured by flow cytometry using propidium iodide as described (37). Cell cycle analysis was done using a FACStar flow cytometer (BD Biosciences).
Semi-quantitative RT-PCR-Total RNA was isolated from frozen cells or explants using Qiagen RNA isolation kit (Valencia, CA), and cDNA synthesis was performed as described previously using ϳ5 g of total RNA (38). Specific transcripts were amplified with gene-specific forward and reverse primers using a step-cycle program on a Robocycler (Stratagene, La Jolla, CA) for less than 30 cycles to keep the amplification within the range of linearity. The gene-specific primers used for RT-PCR are described in Table 1.
Northern Analysis-Northern analysis was carried out to detect rBrm transcripts as described previously (39). Briefly, ϳ2 g of poly(A) RNA, isolated from adult retina, was electrophoresed on a 1.2% formaldehyde gel and transferred to Nytran Plus. Hybridization was carried out using a 32 P-labeled rBrm cDNA probe overnight at 65°C. Blot was washed sequentially with 2ϫ SSC, 0.1% SDS at room temperature for 20 min, then twice with 1ϫ SSC, 0.1% SDS at 65°C for 15 min, and finally with 0.1ϫ SSC, 0.1% SDS at 65°C for 10 min, followed by autoradiography.
siRNA Electroporation-Brm siRNA sequence was cloned into pSuper vector (Oligogene) according to the vendor's protocol. Electroporation was carried out according to a modified protocol of Matsuda and Cepko (40). Briefly, E14 retinas were dissected out and collected in Hanks' balanced salt solution containing pSuper-Brm and pEGFP-C3. The explants were placed in wells made in a 2% agarose gel. Electroporation in agarose gel reduced the shock, and thereby cell death. It also prevented aggregation of explants with one another. Electroporation was carried out by applying 5 pulses at 40 V for 50-ms durations with 950-ms intervals using gold-plated electrodes and an Electro Square Porator (BTX Inc.). The efficiency of electroporation was monitored by green fluorescent protein epifluorescence, and the explants were cultured in RCM containing 5% fetal bovine serum for 4 days. The effect of siRNA was ascertained by analyzing the Brm/␤-actin protein and mRNA levels by Western and RT-PCR analyses, respectively.
Co-immunoprecipitation-Total protein was extracted from RGC-5 cells using a M-PER protein extraction kit (Pierce). Five hundred micrograms of total protein in 1 ml of RIPA buffer (50 mM Tris-HCl, pH 7.4, 15 mM NaCl, 1% Triton X-100, 0.1% SDS, 1 mM EDTA, 1% sodium deoxycholate) was incubated with 5 g of antibody overnight at 4°C. Protein-antibody complex was precipitated by incubating with protein A/G-Sepharose for 1 h at 4°C. The protein-antibody-Sepharose complex was precipitated by centrifuging at 1000 rpm for 1 min, and precipitates were washed three times with RIPA buffer, and the final pellet was resuspended in appropriate volume of loading buffer. The mixture was boiled to dissociate the complex and electrophoresed in a 7-9% denaturing polyacrylamide gel. Negative controls included reactions carried out without the antibody and with IgG.
Western Blot Analysis-Western blot analysis was done as described previously (39). Samples from co-immunoprecipitation or protein isolated from siRNA-transfected retinas were electroblotted onto polyvinylidene difluoride membranes, following electrophoresis. Membranes were blocked for 1 h in 5% nonfat dry milk in TBST and incubated with anti-Brm/anti-Notch (Santa Cruz Biotechnology), diluted 1:500 in TBST overnight at 4°C with shaking. After incubating with anti-mouse horseradish peroxidase, the blots were washed with TBST, and immunoreactive bands were detected using ECL Western blotting detection reagents (RPN 2108, Amersham Biosciences). The blots were then exposed to x-ray film to visualize immunoreactive bands.
Chromatin Immunoprecipitation-Chromatin immunoprecipitation (ChIP) assay was done using a modified procedure from Upstate Biotechnology, Inc. Briefly, transduced (pBabe/ pBabe-Brm) or transfected (pSuper-Brm siRNa/Wt1 (KTS Ϫ )) RGC5 cells were grown until they reached confluency, and histones were cross-linked to DNA by adding formaldehyde directly to culture medium to a final concentration of 1% and incubating for 10 min at room temperature on a rocking platform. Cells were washed three times with ice-cold phosphatebuffered saline containing protease inhibitors. Cell pellets were resuspended in prewarmed SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.1). To reduce the nonspecific background, the samples were pre-cleared using 80 l of salmon sperm DNA/protein A-agarose slurry at 4°C for 30 min. Samples were centrifuged at 100 rpm for 1 min at 4°C. Supernatants were transferred to a new tube, and the immunoprecipitating antibody was added, and incubation was carried out overnight at 4°C on a rocking platform. For a negative control, we used no antibody or IgG. The histone-antibody complex was precipitated using 60 l of salmon sperm DNA and protein A-Sepharose (Upstate) for 1 h at 4°C. Precipitates were washed sequentially at room temperature for 5 min, once with low salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-Cl, pH 8.1, 150 mM NaCl), high salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-Cl, pH 8.1, 500 mM NaCl), and lithium salt immune complex wash buffer (2.25 M LiCl, 1% IGEPAL-CA630, 1% deoxycholic acid-sodium salt, 1 mM EDTA, 10 mM Tris, pH 8.1), and twice with TE buffer (10 mM Tris-Cl, 1 mM EDTA, pH 8.0). After completely removing the TE buffer, the precipitate was resuspended and extracted twice in 250 l of freshly prepared elution buffer (1% SDS, 0.1 M NaHCO 3 ). To reverse the histone-DNA cross-linking, samples were heated at 65°C for 4 h. 200 l of initial sonicated sample was reverse cross-linked and used as an input. After removing the antibodies by protease digestion of the samples, DNA was recovered and column-purified. PCRs were performed using gene-specific primers.
Restriction Enzyme Accessibility Assay-Restriction enzyme accessibility using LM-PCR assay was carried out as described previously (41). Briefly, 5 ϫ 10 6 nuclei from pBabe/pBabe-Brm transduced RGC5 cells were subjected to BglII (75-100 units) restriction enzyme digest for 30 min at 37°C. First strand synthesis of genomic DNA (100 ng) was carried out using Pfu DNA polymerase (Stratagene) and gene-specific forward primer (5Јacccacgtctttctgcactag-3Ј). Following linker ligation, using T4 DNA ligase overnight at 17°C, PCR was carried out on the ligated products using the gene-specific forward primer and linker-specific reverse primer (5Ј-ccgggagatctgaattcgat-3Ј) for 25 cycles at an annealing temperature at 65°C. The PCR products were resolved on 1.2% agarose gel followed by Southern analysis using a PCR product corresponding to a sequence between the second TaqI site and the BgllII site in proximal Brn3b promoter as a probe (Fig. 6D).
Statistics-Results were expressed as mean Ϯ S.E. of at least three separate experiments. Statistical analyses were done using Student's t test to determine the significance of the differences between the various conditions.

Expression of Brm in the Developing
Retina-Emerging evidence suggests that Brm plays an important role in cell differentiation (27,42). To ascertain whether or not Brm has a similar role in retinal development, we examined the temporal patterns of Brm expression during early (E12-E18) and late (E18-PN6) retinal histogenesis by RT-PCR analysis. Retinal cells are born in an evolutionarily conserved temporal sequence; RGCs, cone photoreceptors, horizontal cells, and a majority of amacrine cells are born during early histogenesis, whereas rod photoreceptors, bipolar cells, and Müller glia are generated during late histogenesis (43,44). We observed that levels of Brm transcripts increased at E14 stage, compared with those at the E12 stage ( Fig. 1, A and B). The stages between E12 and E14 represent the period of active neurogenesis when the majority of RGCs, horizontal cells, and cone photoreceptors are born (45). The decrease in the expression of Brm at stage E16 is likely because . Northern analysis, carried out using rat Brm cDNA as a probe, on mRNA isolated from adult retina shows 5.8-kb transcripts corresponding to full-length Brm mRNA (C). Immunohistochemical analysis, carried out on E14 (D-G) and E18 (H-K) retina, obtained from embryos in utero treated with BrdUrd, reveals Brm immunoreactivities (arrows) in E14 retina, distributed both toward scleral (s) and ventricular (v) sides. In contrast, in E18 retina, Brm immunoreactivities (arrows) are predominantly localized toward the scleral side in the inner retina (IR). As expected, BrdUrd immunoreactivities are distributed throughout the width of E14 retina, whereas they are confined in the outer neuroblastic retina (OR) in E18 retina, where proliferating progenitors are distributed. Immunocytochemical analysis, carried out on cell dissociates from BrdUrd-exposed E14 (L-P) and E18 (P) retina, demonstrates that Brm immunoreactivities are associated with both BrdU ϩ and BrdU Ϫ cells (O). However, the proportion of BrdU Ϫ and Brm ϩ cells is significantly (p Ͻ .001) higher, compared with that of BrdU ϩ and Brm ϩ cells, suggesting a correlation of Brm expression with the process of retinal differentiation. Magnification, ϫ200. of the fact that E16 represents a stage of relative quiescence during early histogenesis (12). Similarly, during late histogenesis, there was a significant increase in levels of Brm transcripts at PN1 and PN3, compared with those at E18 when the majority of rod photoreceptors, which constitute ϳ80% of total cells in rodent retina, are born (45). The Brm expression persisted through postnatal stages and its levels were highest in adult retina, compared with all stages examined. In addition, the specificity of Brm expression was determined by sequencing PCR products and Northern analysis of RNA from the adult retina, which revealed a 5.8-kb band corresponding to fulllength Brm transcripts (46) (Fig. 1C). Together, these results suggested that Brm expression is associated with retinal histogenesis. To obtain further insight into the involvement of Brm with the process of cell differentiation in retina, we examined the cell-specific expression of Brm in the retina of E14 and E18 embryos. We observed that Brm immunoreactivities were distributed toward the scleral and ventricular sides in the developing retina ( Fig. 1, D-I). In both E14 and E18 retina, Brm immunoreactivities were predominantly localized toward the scleral side where differentiating precursors are located. As reported previously, Brm immunoreactivities had nuclear as well as cytoplasmic localization (47). Next, to test whether or not Brm expression is associated with differentiating precursor population, we examined the cellular distribution of Brm immunoreativities in the retina of E14 and E18 embryos, pre-exposed to BrdUrd in utero to identify proliferating (BrdU ϩ ) cells (Fig.  1J). We observed that Brm immunoreactivities were associated with both BrdU ϩ and BrdU Ϫ cells. However, in the total cell population at E14, the proportion of BrdU Ϫ Brm ϩ cells (6.8 Ϯ 0.23) was significantly higher than those of BrdU ϩ Brm ϩ cells (4.09 Ϯ 0.23), suggesting that Brm expression was predominantly associated with post-mitotic cells (Fig. 1L). A comparison of different classes of cells at the two different stages revealed that the proportion of BrdU Ϫ Brm ϩ cells increased ϳ2-fold in E18 retina relative to that in E14 retina (12.43 Ϯ 0.75 versus 6.8 Ϯ 0.23, p Ͻ 0.05), demonstrating a progressive association of Brm with post-mitotic cells as development progressed. Together, these observations suggested that the expression of Brm correlated with the process of differentiation of retinal stem cells/progenitors. Next, we examined the involvement of Brm expression and function in the differentiation of stem cells/progenitors along a specific lineage. Although it is likely that Brm is involved in the differentiation of multiple retinal cell types, we studied Brm in the context of RGC differentiation because these are the first cells born (15), and their regulators and markers (10) are well characterized, thus affording an unambiguous interpretation of results. To know whether or not there is a correlation between Brm expression and RGC differentiation, we examined the spatial and cellular distribution of Brm immunoreactivities in the developing retina. Immunohistochemical analysis of E14 retina revealed a spatial overlap of Brm immunoreactivities with that of Brn3b (Fig. 2, A-D) and RPF1 (Fig. 2, E-H). Immunocytochemical analysis of freshly dissociated E14 retinal cells revealed that Brm immunoreactivities are co-localized with that of regulators of RGCs, Brn3b, Wt1, RPF1, and Shh in a subset of cells (Fig. 2, I-L). Together these observations dem-onstrated the association of Brm expression with nascent RGCs in E14 retina. To determine the temporal association of Brm expression with RGC differentiation, we carried out immunohistochemical analysis of retinal cell dissociates from embryos in different stages (E12, E14, E16, and E18 stages) of early histogenesis. We observed that a subset of Brm ϩ cells expressed Brn3b at all stages of early histogenesis, albeit at different proportions (Fig. 2M). For example, compared with those at E12 stage, the proportion of Brm ϩ Brn3b ϩ cells increased by ϳ2-fold at E14, the peak of RGC generation (23.9 Ϯ 3.0 versus 46.0 Ϯ 2.6, p Ͻ 0.001). Subsequently, the number of Brm ϩ Brn3b ϩ cells decreased with the time; compared with those at E14, the proportion of such cells decreased at E18 stage, by which time the majority of RGCs is differentiated (46.0 Ϯ 2.6 versus 4.2 Ϯ 0.89, p Ͻ 0.001). These results are indicated toward the association of Brm with RGC differentiation in vivo; Brm is expressed in differentiating RGCs, and the expression is progressively extinguished from these cells as they complete their differentiation. To know whether Brm had a specific role in RGC differentiation required information regarding the dynamic patterns of Brm expression during RGC differentiation. Does Brm expression correlate with the process of RGC differentiation? This issue was addressed using neurosphere assay, which serves as a robust model of RGC differentiation (48). Retinal stem cells/progenitors in neurospheres utilize normal mechanism of RGC differentiation (38), and the resulting RGCs show target selectivity in terms of establishing contacts with cells in superior colliculus and not with those in inferior colliculus (49). Proliferating progenitors in E14 neurospheres were tagged with BrdUrd, before shifting them to conditions that promote RGC differentiation. This was to ensure that we accounted for the differentiation of only those cells that were proliferating before the onset of differentiation, and not of those that had spontaneously differentiated by asymmetrical division. Samples of neurospheres were assayed at specified times (0, 6, 12, 24, and 48 h) to determine the temporal proportion of Brm ϩ cells that were BrdUrd-tagged (BrdU ϩ ) or expressed Brn3b immunoreactivities. We observed that more Brm ϩ cells were BrdUrd-tagged in differentiating conditions, and their proportions increased with the time, compared with those in proliferating conditions (0 h), corroborating our earlier observation that the expression of Brm is tightly correlated with the process of differentiation (Fig. 2N). That the increase in Brm expression is specifically correlated with the differentiation of RGCs was demonstrated by a coincidental increase in the proportion of Brm ϩ cells expressing Brn3b immunoreactivities, compared with those in proliferating conditions (0 h). The difference in the proportion of Brm ϩ Brn3b ϩ and Brm ϩ BrdU ϩ cells is likely because of the presence of post-mitotic precursors, which did not incorporate BrdUrd, and proliferating progenitors, which failed to incorporate BrdUrd because they were not in S phase of the cell cycle when exposed to BrdUrd.
Functional Involvement of Brm in RGC Differentiation-The correlation of Brm expression with RGC differentiation suggested that Brm influences the process of RGC differentiation. This premise was examined further by perturbing Brm expression and function in retinal stem cells/progenitors and examin- NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48

Brm-mediated Differentiation of Retinal Stem Cells/Progenitors
ing the effects on RGC differentiation. First, we used the siRNA-mediated gene silencing approach to attenuate Brm expression. pSuper vector (Oligogene) containing Brm siRNA was co-electroporated with pEGFP-C3 plasmid in E14 rat retinal explants. Controls included explants, similarly electroporated with siRNA corresponding to scrambled Brm sequence. Because ϳ90% of cells in E14 retina are mitotic (50), the major-ity of cells electroporated with siRNA was likely to be proliferating progenitors. The efficiency of siRNA electroporation was demonstrated by green fluorescent protein epifluorescence in explant sections (Fig. 3A). The specificity of gene silencing was demonstrated by a decrease in levels of Brm protein and its corresponding transcripts in Brm siRNA-treated explants, compared with controls (Fig. 3, B and H). As expected, the expression levels of nontargeted ␤-actin protein and its corresponding transcripts were similar in both groups. The effects on RGC differentiation were examined by changes in the proportion of cells expressing immunoreactivities corresponding to the RGC regulators and markers, Brn3b and RPF1 (36). Immunoreactivities corresponding to both RPF1 (Fig. 3, C-F) and Brn3b (G-J) were decreased in siRNA-treated explants, compared with controls. The RPF1 immunoreactivities, displaying relatively discrete nuclear localization in explants than Brn3b, were used for the quantitation of the number of nascent RGCs. We observed an ϳ3-fold decrease in the proportion of RPF1 ϩ cells in Brm siRNA-treated retinal explants, compared with controls (RPF1 ϩ cells, 70.0 Ϯ 12.41 versus 20 Ϯ 4.08, p Ͻ 0.001), suggesting a decrease in RGC differentiation when Brm expression is attenuated (Fig. 3K). RT-PCR analysis revealed a decrease in levels of regulators of RGCs, RPF1, Brn3b, and Shh transcripts in Brm siRNA-treated explants, compared with controls, thus corroborating our immunohistochemical results. Together, these results demonstrated that the attenuation of Brm expression adversely affected the number of cells expressing RGC-specific markers. Second, we determined whether or not Brm chromatin remodeling function was required for RGC differentiation. We overexpressed dominant negative Brm in retinal progenitors through retrovirus transduction (pBabe-dnBrm) of neurospheres culture. The dominant negative Brm contains a mutation in the ATPase domain, and when overexpressed it is expected to participate in the formation of the nonfunctional ATPase chromatin remodeling complex, thus impairing activities of genes that require Brm (46,51). Controls included neurospheres transduced with wild type Brm (pBabe-Brm) and empty (pBabe) retrovirus. Following transduction, neurospheres were shifted from proliferating to differentiating conditions and scored for the proportion of cells expressing RGCspecific markers. Because retroviral infection of retinal explants does not lead to a uniform transduction of retinal progenitors, which might lead to ambiguous results, we used E14 neurospheres instead. We observed a significant increase in the proportion of both Brn3b ϩ and RPF1 ϩ cells in neurospheres infected with wild type Brm, compared with those infected with empty retrovirus (Brn3b (Fig. 4, A-M). In contrast, there were significantly fewer cells expressing RGC markers in neurospheres infected with dominant negative Brm, compared with those infected with empty retrovirus (Brn3b ϩ cells, 38.7 Ϯ 2.2 versus 23.5 Ϯ 4.5, p Ͻ 0.05; RPF1 ϩ cells, 37.3 Ϯ 4.2 versus 11.7 Ϯ 0.3, p Ͻ 0.005) (Fig. 4M) suggesting the involvement of Brm-mediated chromatin remodeling in RGC differentiation. RT-PCR analysis showed an increase and a decrease in levels of transcripts corresponding to Brn3b, RPF1, and Shh in neurospheres infected with wild type Brm retrovirus and dominant negative Brm retrovirus, respectively, compared with controls (Fig. 4N), corroborating the immunocytochemical results. There was no difference in Tunel-positive cells in control and experimental groups demonstrating that the observed difference in the number of RGCs was not because of Brminduced survival of nascent RGCs (Fig. 4, O-T). Third, we were interested in knowing whether the lack of Brm could lead to RGC phenotype in vivo. We carried out immunohistochemical analysis of retinal sections obtained from adult Brm Ϫ/Ϫ and wild type mice. The retina of Brm Ϫ/Ϫ mice was comparable with that of the wild type in terms of lamination and thickness. However, there were significantly fewer RPF1 ϩ and Brn3b ϩ cells, observed more in the periphery than in the center, in Brm Ϫ/Ϫ retina, compared with wild controls (Fig. 5, A-E). Together, these observations suggested the

involvement of Brm expression and its function in RGC differentiation.
Influence of Brm on the Regulation of Brn3b Expression and Function-Next, we examined the mechanism of Brm-mediated RGC differentiation. Brn3b is a prominent RGC regulatory gene whose activation heralds phenotypic differentiation of RGC precursors through subsequent activation of downstream target genes such as Shh (10,52). We were interested in knowing whether Brm regulated RGC differentiation by facilitating the activation of Brn3b and Shh. The decrease in levels of Brn3b and Shh transcripts in response to siRNA-mediated attenua-tion of Brm expression (Fig. 3) and overexpression of dominant negative Brm (Fig. 4) suggested that Brm positively regulated Brn3b and Shh expression in retinal progenitors. This notion was further supported by observations that the overexpression of wild type Brm led to an increase in levels of Brn3b and Shh transcripts (Fig. 4). To know if Brm directly activated the Brn3b promoter, RGC-5 cells, a transformed RGC cell line (53,54), transduced with wild type Brm or empty retrovirus, were transiently transfected with Brn3b-Luc constructs (24). A 4.5-fold increase in reporter activities was observed in cells transduced with wild type Brm retrovirus, compared with those in controls, suggesting a direct influence of Brm on Brn3b promoter activities (Fig. 6A). It has been demonstrated that ATPase chromatin remodeling complexes are recruited to specific promoters by cell-specific transcription factors where it remodels chromatin to facilitate the activities of the promoters (55-57).
To know if such a mechanism was involved in Brm-mediated facilitation of Brn3b expression, we examined whether or not Brm interacted with one of the upstream regulators of Brn3b, Wt1 (4, 5). Co-immunoprecipitation analysis carried out on nuclear extracts from RGC5 cells revealed that Wt1 antibody precipitated protein complexes from the nuclear extract, which were immunoreactive to Brm antibody (Fig.  6B). This observation suggested that Brm and Wt1 co-existed in protein complexes in RGC5 nuclei. To know whether Brm was recruited to endogenous Brn3b promoter and if the recruitment was influenced by Wt1, we carried out ChIP analysis on RGC5 cells. We observed that DNA-protein complex precipitated by Brm antibody contained sequences corresponding to Brn3b promoter, confirming the presence of Brm on Brn3b promoter (Fig. 6C). The specificity of Brm interactions with the Brn3b promoter was demonstrated by an increase and decrease in the levels of PCR products in pBabe-Brm-transduced and Brm siRNA-treated RGC5 cells, respectively, compared with those in respective controls. To know whether or not Wt1 influenced interactions of Brm with the Brn3b promoter, we carried out ChIP assay on RGC5 cells that were transfected with Wt1 expression con- structs, following transduction with pBabe/pBabe-Brm retrovirus. We observed that the levels of amplified Brn3b promoter sequence increased when RGC5 cells overexpressed Wt1, compared with controls. Together, these observations suggested that Brm was recruited to the Brn3b promoter by Wt1, and this step may constitute a mechanism to facilitate Brn3b expression. To know whether or not Brm expression is associated with nuclease accessibility at the endogenous Brn3b promoter, we examined the restriction enzyme accessibility at the BglII site, Ϫ613 bp upstream of the first ATG site in Brn3b promoter in RGC5 cells transduced with pBabe/pBabe-Brm retrovirus. In an LM-PCR assay, we observed a significant increase in the accessibility to the BglII sites in RGC5 cells transduced with pBabe-Brm, compared with those transduced with empty retrovirus (pBabe), suggesting a Brm-mediated change in chromatin configuration at the Brn3b promoter (Fig. 6D).
A similar mechanism could be invoked where the function of Brn3b in activating Shh is facilitated by Brm. To test this notion, RGC5 cells, transduced with wild type Brm or empty retrovirus, were transiently transfected with luciferase reporter constructs in which a highly conserved Shh regulatory sequence containing a Brn3b-binding site was cloned upstream of a prolactin minimal promoter (58). Controls included transfection of transduced RGC5 cells with reporter constructs containing only the minimal promoter. We observed an ϳ6-fold increase in reporter activities in Brm-transduced cells, transfected with reporter constructs containing the Shh regulatory sequence, compared with activities in those that were transduced with empty retrovirus (Fig. 6E). Reporter activities were minimal in transduced cells, transfected with reporter constructs containing the minimal prolactin promoter, demonstrating the specificity of the Brn3b-binding site in Brm-mediated activation of reporter. To know if Brm interacted with Brn3b, the upstream regulator of Shh, we carried out co-immunoprecipitation analysis on nuclear extracts from RGC5 cells. We observed Brn3b antibody-precipitated protein complexes that were immunoreactive to Brm antibody, suggesting interactions between Brm and Brn3b (Fig. 6F). Next, to know whether Brm was recruited to endogenous Shh promoter, we carried out ChIP analysis on RGC5 cells. We observed that the DNA-protein complex precipitated by Brm antibody contained Shh regulatory sequences with Brn3b-binding site (Fig. 6G). The specificity of Brm interactions with Shh enhancer was demonstrated by increased and decreased levels of PCR products in pBabe-Brm-transduced and Brm siRNA-treated RGC5 cells, respectively, compared with those in respective controls. Taken together, these results demonstrated that Brm facilitated the expression and function of Brn3b, a key regulator of RGC differentiation.
Influence of Brm on Notch Signaling-One of the key regulators of retinal progenitors during RGC differentiation is Notch signaling. It is thought that attenuation of Notch signaling allows progenitors to commit along RGC lineage (11,59). Because it has been shown previously that Brm might interfere with Notch signaling by forming complexes with CSL (55), we were interested in knowing if Brm influenced Notch signaling during the differentiation of retinal progenitors into RGCs. For Brm to regulate Notch signaling, it should be co-expressed with Notch1 receptor in retinal progenitors. Immunocytochemical analysis of retinal progenitors in differentiating conditions revealed the co-localization of Brm and Notch1 immunoreactivities, suggesting their possible interactions (Fig. 7, A and B). To understand the nature of interactions between Brm on Notch signaling, we first wanted to know if Brm affected the expression of Notch target genes, Hes1/Hes5, in retinal progenitors. We observed that, in response to overexpression of dominant negative Brm, there was an increase in levels of transcripts corresponding to both Hes1 and Hes5, compared with those in control neurospheres, suggesting that Brm negatively regulated Notch signaling (Fig. 7C). Next, we wanted to know the mechanism of Brm-mediated inhibition of Hes1 gene. Did Brm directly influence the activities of Hes1 promoter? To address this question, we transiently transfected Hes1-luc vectors in Notch intracellular domain (NICD) expressing 293T cells that were transduced with either wild type Brm retrovirus or empty retrovirus. Reporter activities were easily detected in cells transduced with Hes1-luc vectors and served as positive controls (Fig. 7D). We observed an ϳ3-fold decrease in Hes1 promoter activities in cells transduced wild type Brm retrovirus, compared with controls, suggesting a direct negative effect of Brm on Hes1 promoter activities (Fig. 7D). Next, we wanted to know whether or not these effects on promoter activities involve interactions between CSL/NICD and Brm. CSL is a transcriptional repressor, which is converted to a transcriptional activator when NICD, cleaved upon the activation of Notch receptor, binds to it (60). To address this question, we carried out a co-immunoprecipitation assay, performed on nuclear extracts from cells overexpressing Brm. We observed that protein complexes immunoprecipitated by CSL and NICD antibodies were immunoreactive to Brm, suggesting interactions between Brm and CSL (Fig. 7E). To address the possibility that the faint the Brm immunoreactivity in the complex pulled down by NICD antibody was because of low levels Notch signaling, we carried out co-immunoprecipitation on 293T cells overexpressing NICD. There was an increase in Brm immunoreactivity in a complex precipitated by NICD antibody from cells overexpressing NICD, compared with controls, confirming the specificity of interactions between Brm and NICD. Next, to test whether or not Brm is recruited to the Hes1 promoter, we carried out ChIP analysis in RGC5 cell lines. We observed that DNA-protein precipitated by Brm contained Hes1 promoter sequences, suggesting the presence of Brm as a part of Hes1 transcriptional complex (Fig. 7F). The specificity of Brm interactions with the Hes1 promoter was demonstrated by increase and decrease in the levels of PCR products in pBabe-Brm-transduced and Brm siRNA-treated RGC5 cells, respectively, compared with those in respective controls. Taken together, these observations suggested that Brm negatively regulated Notch signaling by facilitating the repression of Hes1.
Influence of Brm on Cell Cycle-Recent observations have demonstrated that ATPase chromatin remodeling complexes are involved in the facilitation of cell cycle exit (61)(62)(63)(64). We were interested in knowing if Brm-mediated cell cycle  1) (D). The levels of PCR products, amplified with gene-specific (forward arrow) and LM-PCR (reverse arrow) primers, were higher in cells overexpressing Brm, compared with those transduced with empty retrovirus. LM-PCR assay was carried out on BglII-digested RGC5 cell genomic DNA-constituted controls (lane 3). To determine that Brm facilitates the activation of Shh, RGC5 cells, transduced with wild type Brm (pBabe-Brm) (ϩ)/empty (pBabe) (Ϫ) retrovirus, were transiently transfected with luciferase reporter vectors (E). The luciferase reporter vectors contained the rat minimal prolactin promoter (A) or conserved 67-bp Shh sequence containing the Brn3b-binding site, upstream of the minimal prolactin promoter (B). There was a significant increase in luciferase activities in Brm-transduced cells, transfected with vector b compared with those transfected with vector a. Co-immunoprecipitation analyses were carried out to determine the interactions between Brm and Brn3b. Complexes immunoprecipitated with Brn3b antibody from the nuclear extract of RGC5 cells were immunoreactive to Brm antibody, suggesting a complex formation between Brm and Brn3b (F). DNA-protein complexes, immunoprecipitated with Brm antibody, in a ChIP assay on RGC5 cells, contained sequence corresponding to the Shh enhancer as demonstrated by the size-and sequence-specific PCR products, the levels of which were increased and decreased when RGC5 overexpressed Brm (lane 4) and treated with Brm-siRNA (lane 6), respectively (G). Controls are same as described in C. * ϭ p Ͻ 0.05; ** ϭ p Ͻ 0.001. exit could constitute another mechanism by which Brm promotes the differentiation of E14 progenitors into RGCs. We examined this possibility from different angles. We first determined if the proportion of proliferating cells changed in response to perturbations in Brm expression and function in E14 neurospheres. Because retroviral infection of retinal explants does not lead to a uniform transduction of retinal progenitors, which might lead to ambiguous results, we used E14 neurospheres, which are enriched for progenitors and transduced uniformly. E14 neurospheres were transduced with wild type Brm or dominant negative Brm retrovirus in proliferating conditions. Neurospheres were shifted to differentiating conditions and collected at 4-, 12-, 24-, and 48-h time points, after a 4-h pulse with BrdUrd to gauge the temporal proportion of proliferating cells. We observed that in comparison with that at 4 h, the proportion of BrdU ϩ cells decreased with the time in neurospheres, transduced with wild type Brm retrovirus, compared with controls (Fig. 8A). In contrast, the proportion of BrdU ϩ cells increased with the time in neurospheres, transduced with dominant negative Brm retrovirus, suggesting a role of the endogenous Brm and its chromatin remodeling function in cell proliferation (Fig. 8B). Next, we wanted to know at which stage of the cell cycle Brm exerted its influence, so that we could speculate about the mechanism by which Brm promotes cell cycle exit. E14 neurospheres were transduced with wild type Brm retrovirus and maintained in proliferating condition for 48 h. Cells were dissociated, stained with propidium iodide, and subjected to cell cycle analysis by FACS. We observed that E14 neurospheres, transduced with the empty retrovirus, were enriched for cells that were in G 2 /M phase (Fig. 8C). In contrast, the proportion of cells in G 2 /M phase had decreased, and more cells were shifted to G 1 phase in neurospheres transduced with wild type Brm retrovirus. These observations suggested that Brm influenced the G 1 -S transition. One of the mechanisms by which Brm may influence G 1 -S transition is by influencing the expression of cyclin E, the G 1 phase cyclin that regulates G 1 checkpoint (61). Therefore, we argued that the facilitation of RB-E2F-mediated repression of cyclin E, in particular, would ensure that committed precursors do not escape into S phase. We also examined the expression of cyclin A because, like cyclin E, it is also regulated via the RB-E2F complex and is therefore a likely target of Brm-mediated repression. To test the notion, E14 neurospheres were transduced with wild type Brm/dominant negative Brm/empty retrovirus as described above, and levels of transcripts corresponding to cell cycle-related genes were examined by RT-PCR analysis. We observed that levels of transcripts corresponding to cyclin E, cyclin A, and Ki67, a cell cycle marker, were decreased in neurospheres transduced with wild type Brm retrovirus, compared with controls (Fig. 8D). In contrast, their levels increased in neurospheres transduced with dominant negative Brm retrovirus, compared with controls. Similarly, in a separate experiment that involved siRNA-mediated silencing of Brm in neurospheres, we observed an increase in levels of these transcripts in  A and B). Examination of the expression of Hes1 and Hes5 in E14 neurospheres transduced with dominant negative Brm (pBabe-dnBrm) and empty (pBabe) retrovirus revealed increase in levels of Hes1/ Hes5 transcripts in the former, compared with latter, suggesting that Brm has a negative influence on Notch signaling (C). To determine the mode of the regulation of Hes1 by Brm, NICD expressing 293T cells, transduced with Brm (pBabe-Brm)/empty (pBabe) retrovirus, were transfected with luciferase reporter vectors, pGV-BHes1-Luc (positive control)/pGV-B-Luc (negative control) driven by Hes1 promoter (D). There was a significant decrease in luciferase activities in Brm-transduced cells, compared with positive controls, suggesting a direct effect of Brm on Hes1 promoter activities (D). Complexes, immunoprecipitated with CSL and NICD antibodies from the nuclear extracts of cells transduced with wild type Brm, were immunoreactive for Brm, suggesting interactions between Brm and CSL/NICD (E1). There was an increase in Brm immunoreactivity in a complex immunoprecipitated by NICD antibody from cells transfected with NICD expression constructs (lane 2), compared with those transfected with empty expression constructs (lane 1) (E2). DNA-protein complexes, immunoprecipitated with Brm antibody, in a ChIP assay on RGC5 cells, contained sequence corresponding to Hes1 promoter as demonstrated by the size-and sequence-specific PCR products, the levels of which were increased and decreased when RGC5 overexpressed Brm (lane 4) and treated with Brm-siRNA (lane 6), respectively (F). Controls are same as described in Fig. 6C. Magnification, ϫ200. * ϭ p Ͻ 0.05; ** ϭ p Ͻ 0.001. neurospheres treated with Brm siRNA, compared with controls ( Fig. 8E). Together, these observations suggested that Brm negatively regulated cell cycle during RGC differentiation, and the mechanism may involve inhibition of cyclin E and cyclin A expression.
Because the expression of cyclin A is required for the entry into M phase, we surmise that a decrease in cyclin A expression may explain the persistence of cells in the S phase (61,64).

DISCUSSION
The existence of eukaryotic DNA as chromatin renders constraints on the accessibility of the regulatory sequence of genes to tissue-specific and basic transcription factors. The chromatin remodeling complexes, by relaxing (euchromatin) or compacting (heterochromatin) the chromatin organization, modulate the accessibility of these sequences to transcription factors and therefore facilitate gene activation or repression. That SWI/SNF chromatin remodeling complexes have roles to play in neurogenesis was apparent from long known expression of Brg1 and Brm in the developing brain and retina (65,66). Brm and Brg1 are highly homologous ATPases; however, several lines of evidence suggest that they have different functions and they remodel chromatin in different cellular contexts (67).
First, the expression of Brg1 is constitutive and associated with proliferating cells since early embryonic stages, whereas that of Brm strongly correlates with cell differentiation in vivo (68,69) and in vitro (70). Second, mice lacking Brg1 die at the pre-implantation stage (71), and those without Brm survive with an overt phenotype of weight gain (72). Third, SWI/SNF complexes containing Brg1 or Brm have different subunit compositions, and Brmcontaining complexes appear to have lower chromatin remodeling activities than those with Brg1 (67,73,74). These distinct roles of Brg1 and Brm have emerged from the study of higher vertebrates, particularly mammals. In nonamniotic vertebrates, like frogs and fish, Brg1 has been observed to promote differentiation rather than the maintenance of stem cells/progenitors (29,30). A more recent study, using the conditional knock-out strategy, has reaffirmed the earlier observations that Brg1 maintains stem cells, whereas Brm promotes their differentiation in mammals (31). Our study demonstrates a similar role for Brm in retinal progenitors.
Our results suggest that Brm-mediated chromatin remodeling affects three overlapping steps in RGC differentiation. First, Brm may promote RGC differentiation by facilitating transcriptional activation and function of Brn3b. Brn3b is a key RGC regulatory gene. Unlike Math5, it is not required for RGC specification, but it is essential for the normal differentiation and survival of RGCs (75)(76)(77). Therefore, it occupies a lower position than Math5 in the hierarchical regulatory gene network of RGC differentiation and is thought to be under the regulation of Math5 (76,78). Another important upstream regulator of Brn3b during RGC differentiation is Wt1. A recent study has demonstrated that Brn3b is a direct target of Wt1 because its proximal promoter contains a Wt1binding site, WRE, and Wt1 can directly activate Brn3b promoter (5). Our results suggest that Brm interacts with Wt1 and gets recruited to the Brn3b promoter, where it may relax nucleosomal structure, facilitating transcriptional activities of Wt1. In addition, we have demonstrated that a similar mechanism may be involved in promoting the function of Brn3b in activating Shh expression, thus facilitating a cascade of transcriptional activities needed for RGC differentiation. Our observations add to the evidence, emerging from other systems, that the recruitment of Brm to specific promoters by cell-specific transcription factors is a mechanism that provides cell specificity to chromatin remodeling during differentiation (55). To determine the influence of Brm on cell proliferation during RGC differentiation, temporal analysis of cell proliferation was carried out in response to perturbation of Brm expression and function. When E14 neurospheres were transduced with wild type Brm (pBabe-Brm) retrovirus and shifted to RGC-differentiating conditions, the proportion of BrdU ϩ cells decreased and increased with the time, compared with those transduced with empty retrovirus (A). In contrast, the proportion of BrdU ϩ cells increased with the time in neurospheres transduced with the dominant negative Brm (pBabe-dnBrm), compared with those transduced with the empty (pBabe) retrovirus (B). To know the influence of Brm on cell cycle regulation during RGC differentiation, E14 neurospheres, transduced with wild type Brm (pBabe-Brm)/empty (pBabe) retrovirus, were dissociated, labeled with propidium iodide, and subjected to FACS analysis. The proportion of cells in G 2 /M phase decreased significantly (S phase, p Ͻ 0.001; G 2 /M phase, p Ͻ 0.05) and those in G 1 /G 0 phase increased (p Ͻ 0.05) in neurospheres transduced with wild type Brm retrovirus, compared with those in controls, suggesting the influence of Brm on G 1 -S check point (C). RT-PCR analysis revealed a decrease in the levels of transcripts corresponding to cyclin A, cyclin E, and Ki67 in neurospheres transduced with wild type Brm, compared with controls (D). In contrast, levels of these transcripts were increased in neurospheres electroporated with Brm siRNA, compared with controls, suggesting a negative influence of Brm on cell cycle regulators during RGC differentiation (E). * ϭ p Ͻ 0.05; ** ϭ p Ͻ 0.001.
However, the observation that Brm interacts with Wt1 is at odds with a report that Brg1, and not Brm, interacts with zinc finger transcription factors (55). This discrepancy could be reconciled by the facts that SWI/SNF complexes consist of different subunits, which have tissue-specific isoforms and act in a tissue-specific manner (72,74,79). Such a complex nature of interactions, with temporal and cellular contexts, may be the reason why interactions between Brg1 and ␤-catenin (80), and Brg1 and basic helix-loop-helix transcription factors (30) observed by others in different systems, were not detected by Kadam and Emerson (55).
Second, Brm may influence RGC differentiation by attenuating Notch signaling, thereby promoting cell commitment. This notion is supported by the observation that levels of Hes1 and Hes5 increase when Brm expression and function are compromised. We are proposing that Brm influences Notch signaling by interacting with CSL (55). One of the consequences of such interactions could be the repression of Hes1/Hes5. For example, Brm may prevent NICD-CSL interactions by binding CSL and gets recruited to the Hes1/Hes5 promoter, where it could participate in CSL-mediated suppression of promoter activities as CSL without NICD acts as a transcriptional repressor. Additionally or alternatively, Brm may interact with NICD, accentuating the repressor function of CSL. In either case, Brm will inhibit Notch signaling by repressing Hes1/Hes5, thus promoting RGC differentiation. Repression of genes through Brm/Brg1mediated chromatin remodeling complexes is not without precedence. For example, Brm/Brg1 forms a repressor complex with Rb to inhibit the expression of E2F-mediated expression of cyclin genes (77).
Third, Brm facilitates differentiation by ensuring that committed precursors do not make the G 1 -S transition. The mammalian somatic cell cycle alternates between the S phase and the M phase with gaps, G 1 and G 2 , between them (81,82). From the viewpoint of the maintenance of retinal progenitors, the G 1 phase has a specific significance. A point that comes late in the G 1 phase, when crossed, progenitors irreversibly enter the S phase (G 1 -S transition). That point is the G 1 restriction/checkpoint. The G 1 checkpoint is regulated by two types of cyclins, cyclin D and cyclin E, that regulate activities of their respective cyclin-dependent kinases. D-type cyclins (D1, D2, and D3) are sensitive to growth factors (i.e. FGF2 and Wnts). The growth factor-mediated activation and accumulation of cyclin D-dependent kinases phosphorylate Rb, rendering it incapable of forming repression complex with histone deacetylase and Brm complex. In the absence of the Rb repression complex, E2F is able to activate the cyclin E gene. The activation and accumulation of cyclin E-dependent kinases complete Rb phosphorylation and, in addition, inactivate its own repressor, p27 Kip1 , a member of Cip/Kip family of peptides that inhibit cyclin E-and cyclin A-dependent kinases. These two processes ensure G 1 -S transition, at which time cyclin E is degraded and replaced by cyclin A, the S phase cyclin, whose expression is also E2F-dependent. Retinal neurogenesis is intricately linked to the cell cycle (83,84). Different cell cycle regulators, cyclin A, cyclin E, cyclin D1, cdk2, p27 Kip1 , and Rb, for example, are expressed in the developing retina. In addition, levels of their expression are different in stem cells, progenitors, and precursor populations suggesting their modulation during retinal neurogenesis (83,84). Based on conditional and classical knock-out experiments and perturbation of expression/function approaches, the roles some of these regulators play in retinal neurogenesis have begun to emerge. For example, cyclin D1 plays an essential role in progenitor populations because there is a remarkable decrease in retinal thickness in cyclin D1 knock-out mice, because of compromised progenitor proliferation (85,86). Both Rb and p27 Kip1 play context-dependent roles in the developing retina, i.e. depending on the cellular context they regulate either proliferation or differentiation, the latter of Müller cells and the former of rod photoreceptors (87,88). We are proposing that Brm antagonizes G 1 -S transition by facilitating the inhibition of cyclin E and cyclin A. This notion is supported by our observations that both the position of cells in different phases of the cell cycle and the expression levels of cyclin E and cyclin A change in experiments involving the perturbation of expression and function of Brm. In one of the emerging mechanisms, based on a variety of approaches in transformed cell lines, Brm/Brg1 constitute an integral part of the repressor complex consisting of the retinoblastoma protein and histone deacetylase. This complex inhibits the E2F-mediated expression of cyclin E and cyclin A, thus preventing G 1 -S transition (61)(62)(63)(64). A similar mechanism may be involved Brm-mediated repression of cyclin E and cyclin A during RGC differentiation. In summary, our observations suggest that developmental chromatin remodeling, mediated by Brm, may serve as the hub where diverse information for cell differentiation is integrated during neurogenesis. This mechanism is likely to be recruited reiteratively for temporal differentiation of retinal progenitors into different retinal cell types.