Developmental Biology 349 (2011) 90–99

Contents lists available at ScienceDirect

Developmental Biology

journal homepage: www.elsevier.com/developmentalbiology

Neuronal transcriptional repressor REST suppresses an Atoh7-independent program for initiating retinal ganglion cell development

Chai-An Mao a,⁎, Wen-Wei Tsai a, Jang-Hyeon Cho a, Ping Pan a, Michelle Craig Barton a, William H. Klein a,b,⁎ a Department of Biochemistry and Molecular Biology, The University of Texas M. D. Anderson Cancer Center, Houston, TX, 77030, USA b Training Program in Genes and Development, The University of Texas Graduate School of Biomedical Sciences at Houston, Houston, TX, 77030, USA article info abstract

Article history: As neuronal progenitors differentiate into neurons, they acquire a unique set of transcription factors. The Received for publication 16 June 2010 transcriptional repressor REST prevents progenitors from undergoing differentiation. Notably, REST binding Revised 27 September 2010 sites are often associated with retinal ganglion cell (RGC) genes whose expression in the retina is positively Accepted 12 October 2010 controlled by Atoh7, a factor essential for RGC formation. The key regulators that enable a retinal progenitor Available online 20 October 2010 cell (RPC) to commit to an RGC fate have not been identiﬁed. We show here that REST suppresses RGC gene expression in RPCs. REST inactivation causes aberrant expression of RGC transcription factors in proliferating Keywords: Retinal ganglion cells RPCs, independent of Atoh7, resulting in increased RGC formation. Strikingly, inactivating REST in Atoh7-null Retinal progenitor cells retinas restores transcription factor expression, which partially activates downstream RGC genes but is REST (NRSF) insufﬁcient to prevent RGC loss. Our results demonstrate an Atoh7-independent program for initial activation Atoh7 (Math5) of RGC genes and suggest a novel role for REST in preventing premature expression in RPCs. Neurod1 © 2010 Elsevier Inc. All rights reserved.

Introduction proposed to explain virtually all aspects of this intricate process (for recent reviews, see Cayouette et al., 2006; Harada et al., 2007; Lamba The vertebrate retina is an accessible, well-described sensory et al., 2009; Mu and Klein, 2008; Agathocleous and Harris, 2009; tissue that continuously provides valuable insights into neurogenesis, Jadhav et al., 2009). This is especially true for RGCs, for which nerve structure, and nervous system circuitry (Chalupa and Williams, extensive analysis has revealed the transcriptional network circuitry 2008). In particular, major advances have been made using the retina critical for RGC specification and differentiation (Mu et al., 2004, 2005, as a model for central nervous system development. In the developing 2008; Hernandez et al., 2007; Agathocleous and Harris, 2009; Souren retina, neurogenesis begins in the central retina as newly differenti- et al., 2009). A critical gap in our knowledge, however, is the ated neurons signal to the adjacent retinal progenitor cells (RPCs) to identification of the key regulators that enable an RGC-competent RPC initiate a central-to-peripheral wave of neurogenesis. During this to alter its genetic program and advance to a committed RGC fate. In process, multipotent RPCs must decide whether to continue to divide this study, we provide new insight into this issue by identifying REST or to exit the cell cycle and commit to a more restricted lineage- (also called NRSF) as one of these key regulators. competent state. Once committed, RPCs undergo differentiation into The proneural bHLH transcription factor Atoh7 (also called Math5 one of seven cell types in an evolutionarily conserved temporal order in the mouse and Ath5 in other vertebrates) determines the and distinct laminated pattern (Livesey and Cepko, 2001; Agathocl- competency state for RPCs by providing a favorable intrinsic eous and Harris, 2009). Retinal ganglion cells (RGCs) are invariably environment for advancement to an RGC fate (Fig. 1A; reviewed in the first cell type to differentiate followed immediately by amacrine Mu and Klein, 2008). RGC competency occurs when a subpopulation cells, horizontal cells and cone photoreceptor cells in a highly of proliferating RPCs lose their ability to respond to Notch signaling overlapping manner. Subsequently, the later cell types, bipolar cells, and exit the cell cycle (Perron and Harris, 2000; Nelson et al., 2006; Le rod photoreceptor cells, and Müller glial cells, are produced. Much has et al., 2006; Riesenberg et al., 2009). Atoh7 begins to be expressed at been learned about the genetic regulatory mechanisms that are approximately the same time and in the same cells where the responsible for retinal development and several models have been mediators of Notch signaling, Delta and Hes1/5 are upregulated (Yang et al., 2003; Willardsen et al., 2009). But Atoh7 alone is not sufficient to specify the RGC lineage. Once RPCs exit the cell cycle, Atoh7- expressing RPCs give rise to multiple retinal cell types (Fig. 1A; Yang ⁎ Corresponding authors. Department of Biochemistry and Molecular Biology, The et al., 2003). Commitment to the RGC lineage is marked by University of Texas M. D. Anderson Cancer Center, Unit 1000, 1515 Holcombe Blvd., downregulation of Atoh7 and onset of expression of the POU domain Houston, TX, 77030, USA. E-mail addresses: [email protected] (C.-A. Mao), [email protected] transcription factor Pou4f2 and the LIM-homeodomain transcription (W.H. Klein). factor Isl1 (Figs. 1A, B; Gan et al., 1999; Mu et al., 2008; Pan et al.,

et al. (2007) using genome-wide mapping of in vivo protein–DNA interactions, identified REST-RE1 occupancy sites at 1946 loci in the human genome, many of which were close to genes known to be associated with neuronal differentiation. Intriguingly, several of these RE1 sites were the same as those we identified as Atoh7-dependent genes, including Pou4f2. On the basis of the results of Mu et al. (2005) and Johnson et al. (2007), we reasoned that release of REST-mediated repression might play an important role in activating RGC genes. If so, REST could influence RPCs in their decision whether to commit to an RGC fate. To determine the role of REST in the developing retina of mice, we deleted a floxed allele of REST using a Six3-Cre transgene (Furuta et al., 2000). We found that REST played a critical role in suppressing RGC gene expression in proliferating RPCs. Most strikingly, deletion of REST partially restored the RGC gene expression program that is normally lost in Atoh7-null retinas. In addition, we present evidence to show that another proneural bHLH competency factor, Neurod1, is upregulated in REST-deleted retinas and plays an unexpected role in RGC development that is independent on the presence of Atoh7.

Materials and methods

Gene targeting and animal breeding

A gene targeting vector was made that contained a floxed REST Fig. 1. Schematic illustration of RGC production from RPCs. (A) REST is active in allele in which exon 2 of REST was flanked by two loxC2 sites and proliferating RPCs and is degraded upon cell cycle exit/differentiation. Atoh7 begins to could be deleted by Cre-mediated recombination. To construct this be expressed at the G2/M phase before RPCs exit the cell cycle to commence differentiation. Neighboring RPCs respond to the local environment to express different vector, we used genomic DNA from G4 ES cells to PCR-amplify 1.07-, proneural bHLH genes. The onset of Pou4f2 and Isl1 expression marks the initial 2.21-, and 5.4-kb fragments from the REST locus (see Fig. 4A) and commitment to RGC differentiation. Note that Atoh7 is required for expression of Pou4f2 subsequently cloned them into a knockout vector. The resulting and Isl1 and for RGC formation. Atoh7-expressing cells give rise to multiple cell types, constructs were linearized and electroporated into G4 ES cells including RGC, amacrine cells (AC), horizontal cells (HC), and photoreceptor cells (George et al., 2007), after which G418-resistant ES cells were (PhR). (B) REST gene targets include Pou4f2 but not Isl1. BC: bipolar cells and MGC: Müller glial cells. selected to identify homologous recombination events. A 5' probe from outside the homologous recombination region was used to detect 11-kb wild-type and 7.6-kb targeted fragments produced by 2008). At present, little is known about how Atoh7-expressing RPCs EcoRV/Nhe1 digestion of ES cell DNA (Figs. 4A, B). Two targeted ES cell respond to local environmental signals to activate Pou4f2 and Isl1. lines were identified, expanded and injected into B6(GC)-Tyrc-2J/J REST, a zinc-finger transcription factor, offers a possible route in blastocysts, and the injected blastocysts were transferred into the solving the problem of RPC commitment to an RGC fate. REST was uteri of pseudopregnant C57/BL/6J female mice. Chimeric males initially identified as a master repressor of neuronal gene expression resulting from the injected blastocysts were bred to B6(GC)-Tyrc-2J/J in non-neuronal cell types (Chong et al., 1995; Schoenherr and females (Jackson Laboratory) to generate the targeted floxed REST Anderson, 1995; Bruce et al., 2004, 2009). REST binds to a conserved allele. The targeted allele was further bred to a Rosa26-FlPeR line to 21-bp motif termed repressor element 1 (RE1), which is found within remove the FRT-flanked Neo cassette (Farley et al., 2000). The the transcriptional regulatory regions of hundreds of neuronal genes resulting line was assigned as RESTfx, which was distinguished from (Mortazavi et al., 2006; Johnson et al., 2007; Otto et al., 2007). REST the wild-type allele by PCR genotyping using primers re08 (5- mediates active repression via recruitment of histone deacetylases by CATGCGAGTACTGCCATACCCAAC-3), re09 (5-GTGATGGGGCAGTC- its corepressors mSin3 and CoREST (Ballas et al., 2005; Lunyak and TTCTGGAGG-3), and re11 (5-GGGCACACCTTTAATCCTAGCTTC-3) Rosenfeld, 2005). In neuronal progenitor cells, REST is expressed at (Figs. 4C–E). Atoh7 (Math5) knockout mice were genotyped as levels that are sufficiently high enough to maintain neuronal genes in described in Wang et al. (2001). Neurod1 knockout mice were a chromatin-inactive state poised for activation (Ballas et al., 2005). genotyped as described in Pennesi et al. (2003). Embryos were Upon neuronal differentiation, REST is degraded through ubiquitin- designated as E0.5 at noon on the day in which vaginal plugs were mediated proteolysis (Guardavaccaro et al., 2008). A wealth of observed. information exits in the literature on the molecular mechanisms by All animal procedures in this study followed the United States which REST functions as a transcriptional repressor. Most investiga- Public Health Service Policy on Humane Care and Use of Laboratory tions addressing the biological role of REST use mammalian tissue Animals and were approved by the Institutional Animal Care and Use culture cell systems (Ballas et al., 2005; Su et al., 2004; Watanabe Committee at The University of Texas M. D. Anderson Cancer Center. et al., 2004). Although tissue culture cells are more amenable to mechanistic analysis, they may not accurately reflect the complexity Histology, in situ RNA hybridization, and immunohistochemical analysis of the in vivo environment where neurogenesis occurs. REST-null mice have been generated but exhibit embryonic lethality (Chen et al., Embryos or eyes dissected from embryos and adults were fixed, 1998), limiting their use in studying REST's role in neurogenesis. cryo- or paraffin-embedded, and sectioned into 7- or 12-μm slices for In a previous study, where we identified genes whose expression histology, in situ hybridization, or immunohistochemical analysis. For was downstream of Atoh7, we discovered that many Atoh7- histological analysis, the sections were stained with hematoxylin and dependent genes harbor RE1 elements within 100 kb of their coding eosin. In situ RNA hybridization on frozen or paraffin-embedded sequences (Mu et al., 2005). Most notable among these genes was sections was performed as described previously (Smith et al., 2000) Pou4f2, which contains two RE1 sites. In a separate study, Johnson with the following modifications. After de-waxing and rehydration, 92 C.-A. Mao et al. / Developmental Biology 349 (2011) 90–99 sections were fixed for 30 min at room temperature in 4% parafor- regions of flat-mounted retinas. Values from three littermate pairs of maldehyde in PBS and treated for 10 min with 8 μg/ml proteinase K in P20 mice were used for statistical analysis using a simple t-test 50 mM Tris–HCl and 5 mM EDTA (pH 7.0). Prior to hybridization, the (STATISTICA 6). For estimating RGC number in developing retinas, sections were washed twice in 2XSSC for 15 min, and incubated in retinal sections representing the same place in the retina from 0.1 M Tris and 0.1 M glycine for 30 min. The hybridization solution littermates of different genotypes were stained with anti-Pou4f2 or (100 μl/slide) contained 50% deionized formamide, 5× SSC (pH anti-Isl1 antibodies and the number of Pou4f2/Isl1 positive cells was adjusted with citric acid to pH 6.0), 10% dextran sulfate, 1 mg/ml determined by counting. yeast tRNA and 10 to 20 ng/μl of the riboprobes, and was performed overnight at 65 °C–68 °C under coverslips. Next, the sections were TUNEL assays and BrdU labeling washed for 1–2 h in 0.5× SSC, 20% formamide at 65 °C. Subsequently, they were treated with 10 μg/ml RNaseA for 30 min at 37 °C in NTE, TUNEL assays on retinas were conducted using an in situ cell death then washed for 4 h in 0.5× SSC, 20% formamide at 65 °C for 30 min in detection kit (Roche Applied Science) following the manufacturer's 2× SSC, and blocked for 1 h at room temperature in 1% blocking instructions. For pulse labeling with BrdU to detect S-phase RPCs, reagent (Roche) in MABT. A 1:400 to 1:1000 dilution of anti- 100 μg of BrdU (Upstate Biotechnology) per gram of body weight was digoxigenin-AP conjugated antibody (Roche) was preincubated for intraperitoneally injected into pregnant females 30 min before at least 1 h in 1% blocking reagent and 10% normal sheep serum in euthanization. The sections were processed using the microwave MABT at 4 °C. The sections were incubated with the antibody retrieval technique described earlier in the section on immunohisto- overnight at 4 °C, washed for 6 h in PBST, and for 30 min, in NTMT, chemical analysis. and stained using centrifuged BM purple AP substrate (Roche) in 0.1% Tween 20 for 12–36 h at 4 °C or room temperature. They were washed Chromatin immunoprecipitation analysis (ChIP) in NTMT, dehydrated, and then mounted in Aquamount (Poly- sciences). Images were collected using an Olympus X70 microscope Retinas were isolated from E14.5 wild-type embryos of C57BL/ with an Olympus UPlanApo 10/0.40 objective lens and Olympus DP71 6J:129sv mixed background and were cross-linked with 1% formal- camera with Olympus DP-Controller software. dehyde for 10 min. ChIP assays were performed as previously Immunohistochemical analysis was performed as described described (Tsai et al., 2008) with minimal modifications. Briefly, the previously (Mao et al., 2008a,b). Briefly, frozen or paraffin-embedded fragmented, precleared chromatin lysate was incubated overnight sections were placed in a microwave oven at 600 W in 10 mM sodium with specific antibodies: anti-REST (Upstate/Millipore) and normal citrate for 18 min to expose the antigen epitopes, and then blocked in rabbit IgG (Upstate/Millipore). Quantitative (q)PCR was conducted in 10% normal serum and 0.1% Tween 20 for 1 h at room temperature. a 7500 FAST ABI instrument. Incubation with primary antibodies was performed at 4 °C over 1 or 2 nights. Secondary antibodies were applied to sections for 2 h at room RT-PCR analysis temperature. The primary antibodies used were: goat anti-Pou4f2/ Brn3b (1:200, Santa Cruz Biotechnology), rabbit anti-calbindin Total RNA was collected from two E14.5 retinas using Trizol (1:500, Swant), sheep anti-Chx 10 (1:1000, Covance), rabbit anti- reagent (Invitrogen). RNAs were reversed transcribed using the Eomes (1:200 Chemicon), mouse monoclonal ant-GFAP (1:300, SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen) Sigma), rabbit anti-GSK3β (1:200, Cell Signalling), mouse anti-Isl1 following the manufacturer's instructions. A twentieth of the total (1:250 DSHB, University of Iowa), Goat anti-Neurod1 (Santa Cruz cDNA was used for PCR. For quantitative reverse transcriptase-PCR Biotechnology), mouse anti-NFL (1:200 InVitrogen), rabbit anti-B- analysis, cDNAs were amplified using SYBR green PCR master mix opsin and anti-R-opsin (1:500, Chemicon), mouse monoclonal anti- (Applied Biosystems, CA). The relative expression levels were Pax6 (1:200, DSHB, University of Iowa), rabbit anti-Sox9 (1:200, normalized to that of GAPDH and calculated using the comparative

Chemicon), and chicken anti-TUJ-1 (1:200, Chemicon). Secondary Ct method (7500 Fast Real-time PCR systems SDS software, Applied antibodies were conjugates of Alexa Fluor 488 and Alexa Fluor 555 Biosystems). DNA sequences of PCR primers were as follows: REST (Invitrogen). DAPI (4, 6-diamidino-2-phenylindole) was used as a rrt01 (5-GTGCGAACTCACACAGGAGA-3), REST rrt02 (5-AAGAGGTT- nuclear counterstain. Finally, slides were washed and mounted in TAGGCCCGTTGT-3), GAPDH forward (5-AGGTCGGTGTGAACG- fluoromount G (EMS). GATTTG-3), GAPDH reverse (5-TGTAGACCATGTAGTTGAGGTCA-3), Stmn2 forward (5-CTGAAGTTGTTGTTCTCCTCC-3), Stmn2 reverse Retinal flat-mount analysis (5-CTCCACGAACTCTAGCTTCTC-3), GAP43 forward (5-GTGCT- GCTAAAGCTACCACT-3), GAP43 reverse (5-GTACAAAGTGTCACCT- To detect RGC number, eyes were removed from postnatal mice CAGT-3), Persyn forward (5-GTACAAAGTGTCACCTCAGT-3), Persyn and fixed with 4% paraformaldehyde for 30 min. The cornea, ciliary reverse (5-CAGCAGCATCTGATTGGTGA-3), and Atoh7 forward (5- band, and lens were removed using a pair of iris scissors. The CAGGACAAGAAGCTGTCCAAG-3), Atoh7 reverse (5-GGTCTACCTG- remaining retinal tissue and attached pigmented epithelium were GAGCCTAGCA-3). fixed for 1 h and then washed four times in PBS saline and 0.1% Triton X-100 at room temperature. The retinas were then incubated in Results blocking solution (PBST plus 5% fetal bovine serum) for 1 h, and incubated with anti-Pou4f2/Brn3b (1:100) antibodies for 48 h at 4 °C. REST is expressed in the embryonic and postnatal retina Retinas were washed four times with PBST and stained with Alexa488-conjugated donkey anti-goat secondary antibody (Invitro- Since REST was discovered as a major repressor of neuronal genes, gen). After the retinas were washed thoroughly with PBS, the five additional alternatively spliced isoforms, namely REST1 to REST5, pigmented epithelium was removed from the retinas and four or have been shown to be expressed in the differentiated neurons of the five symmetrical cuts were made halfway from the peripheral rim to rat brain (Palm et al). In mice, REST1, REST3, and REST4 isoforms are the central optic disk. Retinas were then flat-mounted onto glass also expressed in neuroblastoma cell lines. Among these, REST4, slides and analyzed using an Olympus FluoView1000 confocal encoding a truncated protein (Palm et al., 1998, Fig. 2B), is the major microscope. To avoid the skewed distribution of Pou4f2-expressing isoform acts as a de-repressor to modulate REST-mediated repressing RGCs along the nasal–temporal axis, Pou4f2-positive RGCs were activity (Shimojo et al., 1999; Lee et al., 2000). To determine when counted in three randomly chosen areas of the dorsal temporal REST and REST4 were expressed in the mouse retina, we used semi- C.-A. Mao et al. / Developmental Biology 349 (2011) 90–99 93

Fig. 2. The expression of alternative spliced REST isoforms in retinas. (A) Semi- quantitative RT-PCR for REST expression in developing and adult retinas. The primers used in the PCR for REST, rrt01 and rrt02, are indicated as arrows in B. Note that the ′ forward primer rrt01 aligns to 3 end of exon 2 and extends to exon 3. One-fourth of Fig. 3. REST binds to an RE1 site downstream of Pou4f2 gene. (A) Two RE1 sites are – GAPDH primers were used in the same reaction. M: 1 kb ladder; E12.5R E18.5R: wild- found within the Pou4f2 locus. Site A is located 1.7-kb upstream from the – type embryonic retinas; P10R P60R: postnatal retinas; E14.5B: E14.5 forebrain; E14.5R- transcriptional start site, and site B is located 1.4-kb downstream from the fx/fx mut: E14.5 REST : Six3-Cre retinas. The alternatively spiced isoform, RERST4, was transcriptional stop site. Sites A and B are highly similar to the canonical RE1 (cRE1) found mainly in postnatal retinas. (B) A schematic illustration and partial sequences of site. (B) Chip analysis of E14.5 wild-type retinas. qPCR was performed to detect REST exons. The location of exons 2, 3, N, and 4 are indicated in Fig. 4. The sequence of antibody-bound DNA fragments encompassing sites B and a non-specific region 10-kb exon N is identical to that in Lee et al. (2000). downstream of Pou4f2 (ns). REST-bound levels were normalized to input DNA. DNA sequences of qPCR primers were RE1 site B, RE1b1 (5-GTTAGCTGTTGTAGCGCTCCCTG-3), RE1b2 (5-CTGTCCCCATCCTAGGTTTCAGG-3), and non-specific downstream region, ns1 quantitative RT-PCR with embryonic and adult retinas. Between E12.5 (5-CCACTTATCCACTGAGTCATCTC-3), ns2 (5-GTACCCTACGAGATAGCACCATC-3). and E18.5, REST was the major isoform expressed. This was also the case in the E14.5 developing cortex (Fig. 2A, lanes E12.5R to E18.5R and E14.5B). In contrast, REST4 was expressed weakly in the significant differences in the formation of non-RGC retinal cell types developing embryonic retina and cortex, but it was the predominant (Fig. S1, Table S1). To quantify the number of RGCs in REST mutant isoform in postnatal retinas (Fig. 2A, lanes P10R–P60R). The retinas, we immunostained flat-mounted retinas from P20 mice with embryonic retina largely consists of proliferating progenitor cells, an anti-Pou4f2 antibody. We randomly selected three areas within the whereas the postnatal retina is composed of differentiated neurons dorsal temporal regions of the retinas from three pairs of P20 and Müller glial cells. Thus, the temporal expression profile of REST littermates and observed 18.2% more RGCs in RESTfx/fx:Six3-Cre retinas isoforms in retinas supports the notion that REST is expressed in than in heterozygous controls (Figs. 5C, D). These data suggested that neural progenitor cells and REST4 in differentiated neurons. removing REST from RPCs during early retinal development substan- tially increases RGC production. REST binds to an RE1 site downstream of Pou4f2 The increase in numbers of RGCs could arise from two possible mechanisms. First, REST is expressed in neural stem cells and neuronal Two conserved RE1 sites are found in close proximity to Pou4f2. progenitors (Ballas et al., 2005; Sun et al., 2008; Westbrook et al., Site A is 1.7-kb upstream and site B is 1.4-kb downstream (Fig. 3A). To 2008). In the absence of REST, proliferating RPCs might exit the cell determine if REST occupies these sites in the developing retina, we cycle prematurely, as has been shown for neural stem cells (Sun et al., performed ChIP assays with ten E14.5 retinas using an anti-REST 2008; Westbrook et al., 2008). Because RGCs are the first retinal cells antibody. The immunoprecipitated DNA fragments were amplified by to differentiate, this would cause more RPCs to adopt an RGC fate. qPCR using primers to detect site A and site B (Fig. 3A). REST bound to However, using BrdU labeling as an indicator of S-phase, we detected site B with a 2.4-fold enrichment compared to non-specific occupancy a 40.2% increase of mitotic RPCs in retinal sections of RESTfx/fx: Six3-Cre by rabbit IgG (Fig. 3B). Moreover, no REST or rabbit IgG binding was mice versus RESTfx/+ littermate controls (Brdu+/total cell in NBL: observed within a randomly selected sequence 10-kb downstream of 41.464.15% vs. 29.576.13%; n=3, Pb0.01). A second possibility is that Pou4f2 (Fig. 3B). In contrast to site B, we detected only weak REST removing REST in mitotically active RPCs might result in aberrant binding to site A (data not shown). Site B's location was consistent activation of RGC genes. If this were the case, we would expect to see with it being a cis-regulatory element within a transcriptional greater numbers of RPCs giving rise to RGCs. To determine whether regulatory region downstream of Pou4f2. RGC genes were expressed in proliferating RPCs, we looked for Isl1 and Pou4f2 expression in the neuroblast layer of E15.5 retinas. We Aberrant RGC gene expression and increase RGC number in REST-deleted observed many more Isl1- and Pou4f2-positive cells in the neuroblast retinas layer of RESTfx/fx;Six3-Cre retinas than in that of RESTfx/+ littermates (Figs. 6A–D). To compare the number of RGCs in wild-type and REST To determine REST's function during retinogenesis, we generated mutant retinas, we counted the number of Pou4f2-positive and Isl1- mice with a floxed allele of REST (Fig. 4A) and bred them with a Six3- positive RGCs at E14 and E15.5. We found a 12–25% increase in Cre transgenic line to delete exon 2 of REST in the developing retina Pou4f2-positive cells and a 6–17% increase in Isl1-positive cells in beginning at E10 (Figs. 4A–D; Furuta et al., 2000). We first examined REST mutant retinas compared to wild-type littermates (Table S2). histological sections from control and RESTfx/fx;Six3-Cre retinas twenty Furthermore, we did not detect an increase in Pou4f2-positive or Isl1- days after birth (P20) and found a significant increase in the thickness positive cells in the neuroblast layer between E11 and E13. The results of the inner plexiform and ganglion cell layers (Figs. 5A,B, arrow). In suggested that loss of REST did not cause premature cell cycle exit in addition, many cell clumps protruded toward the retinal pigment mitotic RPCs but rather caused an upregulation in the expression of epithelia (Fig. 5B, arrowheads), resulting in a mis-patterned retina. key transcription factors that are required for RGC differentiation to Aside from these histological abnormalities, we did not detect commence during in a very narrow developmental window. 94 C.-A. Mao et al. / Developmental Biology 349 (2011) 90–99

Fig. 4. Generation of REST conditional mutant mice. (A) Genomic structure for REST and the targeting vector, and predicted structure of the targeted floxed REST allele. Exon 2 encoding the N-terminal repressor domain and part of the zinc-finger DNA-binding domain, is flanked by loxC2 recombination sites. The black bars underneath indicate the DNA fragments amplified from genomic DNA making the targeting vector. Arrows indicate the PCR primers re08, re09, and re11 that were used for PCR genotyping of wild-type, floxed and deleted REST alleles. (B) Southern blot analysis using a 5 probe to distinguish wild-type and REST-targeted alleles from genomic DNA of targeted ES cells digested with EcoRV/ Nhe1. Arrow indicates a targeted ES cell clone. (C) Representative PCR genotyping using re08 and re09 for wild-type and floxed REST alleles, and representative PCR genotyping of the Six3-Cre transgene using c01 and c02. Arrows indicate the RESTfx/fx;Six3-Cre allele. (D) The RESTdel/+ allele was generated by breeding mice containing the RESTfx allele with a CMV-Cre mouse line to remove exon 2 from the germline. PCR genotyping with re08 and re11 was used to distinguish the wild-type and deleted REST alleles of DNA from E10.5 embryos from interbred RESTdel/+ mice. Arrows indicate dying E10.5 RESTdel/del embryos with phenotypes identical to that described previously (Chen et al., 1998).

We always observed more Pou4f2-positive cells than Isl1-positive cells in the neuroblast layer of RESTfx/fx;Six3-Cre retinas (Figs. 6B,D). Notably, Pou4f2, but not Isl1, has been shown to be a direct target of REST (Mortazavi et al., 2006; Johnson et al., 2007). At E15.5, the neuroblast layer contains both proliferating RPCs and committed RGCs that have exited the cell cycle and are expressing Isl1 and Pou4f2 (Mu et al., 2008; Pan et al., 2008). To distinguish proliferating from nonproliferating cells, E15.5 retinas were co-labeled with anti-Pou4f2 and anti-BrdU antibodies. We found a substantial number of Pou4f2- positive cells were co-labeled with BrdU in RESTfx/fx:Six3-Cre retinas whereas fewer were co-labeled in control retinas (Figs. 6E–J; Pou4f2- positive/Brdu-positive, 8.18%1.50% versus 4.04%0.86%, n=3, Pb0.01). Besides Pou4f2, other REST target genes known to be expressed in the retina, including Pou4f1 and Rtn1, were not upregulated in the neuroblast layer of RESTfx/fx;Six3-Cre retinas (data not shown). Aberrant expression of Pou4f2 and Isl1 in retinas of REST mutants did not cause increased cell death (data not shown).

Aberrant expression of Pou4f2 and Isl1 in proliferating RPCs of REST mutant retinas does not require Atoh7

Atoh7 is required for the expression of Pou4f2 and Isl1 and for RGCs to commit to an RGC fate (Brown et al., 2001; Wang et al., 2001; Mu et al., 2005, 2008; Pan et al., 2008). However, Atoh7 is not a target of REST, and Atoh7 expression was not upregulated in RPCs of REST mutant retinas (Fig. S2). Because Pou4f2 is a direct target of REST, its aberrant expression in REST mutants might be the direct result of chromatin de-repression at the Pou4f2 locus. The aberrant upregula-

Fig. 5. Increased RGC formation in retinas of REST mutant mice. (A, B) Histological sections tion of Isl1, however, must occur through a less direct mechanism of P20 eyes from Restfx/+ (A) and RESTfx/fx;Six3-Cre (B) littermates. Arrow points to the because Isl1 is not known to be a direct target of REST. In both cases, ganglion cell layer. Arrowheads point to the cell clumps protruding toward the retinal we expected that aberrant expression would be independent of the pigmented epithelia. (C, D) Immunostaining of Pou4f2-positive RGCs on ﬂat-mounted presence of Atoh7 and that REST would act as a repressor in RPCs to RESTfx/+ (C) and RESTfx/fx;Six3-Cre (D) retinas. (E) A Box-Whisker plot of a t-test (n=3, prevent the Atoh7-independent expression of Pou4f2 and Isl1. pb0.05) reveals a signiﬁcant difference in RGC numbers between control and REST mutant G/G retinas. GCL: ganglion cell layer; INL: inner nuclear layer; IPL: inner plexiform layer; ONL: To determine whether this was the case, we generated Atoh7 ; fx/fx outer nuclear layer. REST ;Six3-Cre mice and examined Pou4f2 and Isl1 expression in C.-A. Mao et al. / Developmental Biology 349 (2011) 90–99 95

regions. We compared the relative expression levels of these genes in E14.5 retinas of REST mutant embryos and heterozygous controls, and in Atoh7-REST double mutant embryos and Atoh7 mutant embryos. The expression levels of all the genes were upregulated to varying degrees (Fig. 8). These data supported the hypothesis that removing REST in developing retinas caused an upregulation of RGC genes and triggered an Atoh7-independent program to initiate RGC gene expression.

Expression of RGC genes downstream of Pou4f2 and Isl1 in Atoh7-REST double mutant retinas

We have proposed a gene regulatory network for RGC development that has as its central feature four hierarchical tiers of transcription factors (Mu et al., 2008). The most downstream of these tiers includes the T-box-containing transcription factor Eomes (also called Tbr2) along with several proteins associated with the differentiation and maintenance of RGCs and their axons (Mu et al., 2005; Mao et al., 2008a). To determine the extent to which this gene regulatory network was operable in Atoh7-REST double mutant retinas, we examined the expression of a set of RGC genes downstream of Pou4f2 and Isl1 in Atoh7G/G;RESTfx/fx;Six3-Cre retinas from E15.5 embryos. Eomes, whose expression depends on both Pou4f2 and Isl1, was expressed in a subset of RGCs located in the ganglion cell layer of RESTfx/+ wild-type retinas (Fig. 9A). In Atoh7G/G; RESTfx/fx;Six3-Cre retinas, many Eomes-positive cells were detected in the neuroblast layer in a dorsal-to-ventral descending gradient similar to what was observed for Pou4f2 and Isl1 expression (Fig. 9A1). Likewise, GAP43, another gene downstream of Pou4f2 and Isl1,was activated in the neuroblast layer of Atoh7G/G;RESTfx/fx;Six3-Cre retinas, although at less than wild-type levels (Figs. 9D, D1). Neither Eomes nor Gap43 was expressed at significant levels in Atoh7-mutant retinas. However, other RGC-expressed genes, including Stmn2, Persyn, TUJ1, and GSK3β, were not noticeably upregulated in the neuroblast layer of Fig. 6. Upregulation of Pou4f2 and Isl1 in proliferating RPCs of retinas from REST mutant G/G fx/fx – – mice. Images of E15.5 Restfx/+ and RESTfx/fx;Six3-Cre retinas. Pou4f2 expression (A, B) Atoh7 ;REST ;Six3-Cre retinas (Figs. 9B C1, E F1). Together, these and Isl1 expression (C, D). Arrowheads in panels B and D point out the boundaries of results indicated that some but not all of the genes downstream of dorsal region within the neuroblast layer where Pou4f2 (B), and to a lesser extent, Isl1 Pou4f2 and Isl1 are activated in the NBL in Atoh7-REST double mutant (D) are upregulated relative to Pou4f2 and Isl1 expression in the neuroblast layer of retinas. This suggests that the RGC gene regulatory network is Restfx/+ control retinas (A and C, respectively). Arrowheads in panels B and D point to the region within the neuroblast layer where Pou4f2 and Isl1 are upregulated. (E–J) Co- partially but not fully restored in Atoh7-mutant retinas when the expression of Pou4f2 and BrdU-positive RPCs. Scale bars in panels C and I: 100 μm. repressive functions of REST are alleviated. DbNV, dorsal–ventral axis; GCL, ganglion cell layer; NBL, neuroblast layer. Upregulation of Neurod1 in REST-deleted retinas and fewer RGCs specified in Atoh7-Neurod1-REST triple-mutant retinas than in Atoh7-REST double developing retinas at different embryonic times. At E13.5, many mutant retinas Pou4f2- and Isl1-positive RPCs were detected in the neuroblast layer of Atoh7G/G;RESTfx/fx;Six3-Cre retinas, while, as expected, very few Previous investigations have shown that a few RGCs still remain in Pou4f2- and Isl1-positive cells were seen in Atoh7G/G retinas (Table S2; Atoh7-null adult retinas (Lin et al., 2004; Moshiri et al., 2008). These Figs. 7A–E). A small fraction of Pou4f2- and Isl1-positive RPCs in RGCs form in retinal peripheral rim of the developing retina by Atoh7-REST double mutants were mitotically active as shown by co- unknown mechanisms. In earlier work, we replaced Atoh7 with another labeling with BrdU (Figs. 7C1, F1, arrowheads). Despite their proneural bHLH gene, Neurod1, and found that the Atoh7Neurod1 allele significant upregulation in the neuroblast layer, only a few Pou4f2- replaced Atoh7's function in restoring RGC formation, albeit not positive cells were detected in the nascent ganglion cell layer of the completely (Mao et al., 2008b). This was a somewhat surprising result retina, suggesting that these cells were defective in their ability to since endogenous Neurod1 hasnotbeenshowntofunctioninRGC migrate and differentiate into mature RGCs. Consistent with this formation in mice (Liu et al., 2008; Pennesi et al., 2003; Morrow et al., observation, many dying cells were observed in the ganglion cell layer 1999). It is therefore possible that Neurod1 might play a role in RGC of Atoh7G/G;RESTfx/fx;Six3-Cre retinas (Fig. 7I, arrowheads). The results formation during retinal development, independent of Atoh7.Moreover, suggested that Pou4f2- and Isl1-positive cells were defective in their Neurod1 is a direct target of REST (Mortazavi et al., 2006; Johnson et al., ability to differentiate into RGCs and could not survive. In addition, 2007; Otto et al., 2007), suggesting that REST inactivation in proliferating retinas from adult Atoh7 G/G;RESTfx/fx;Six3-Cre mice displayed more RPCs of Atoh7-null retinas will result in upregulation of Neurod1 severe mis-patterned retinal than that seen in RESTfx/fx;Six3-Cre expression. If so, this could subsequently contribute to the aberrant retinas (Fig. S3), but unlike RESTfx/fx;Six3-Cre retinas, the double activation of Pou4f2 and Isl1. mutants did not have more RGC cells (data not shown). We detected a 60.4% increase in Neurod1-expressing RPCs in We used qRT-PCR analysis to compare the expression levels of RESTfx/fx: Six3-Cre retinas compared with RPCs of heterozygous other Atoh7-dependent genes, including Pou4f2, Isl1, Stmn2, GAP43, controls, and a 15.3% increase in RPCs of Atoh7G/G: RESTfx/fx: Six3-Cre and Persyn. Among these, Isl1 and GAP43 are not known targets of retinas compared with RPCs of Atoh7G/G retinas (Fig. S4). This REST but Pou4f2, Stmn2, and Persyn harbor RE1 in their regulatory upregulation might contribute to the observed increases in Pou4f2 96 C.-A. Mao et al. / Developmental Biology 349 (2011) 90–99

Fig. 7. Restored expression of Atoh7-downstream RGC genes Pou4f2 and Isl1 in Atoh7G/G;RESTfx/fx;Six3-Cre double mutant retinas. (A–C) Pou4f2 expression in E13.5 RESTfx/+ (A), RESTfx/+;Atoh7G/G (B), and RESTfx/fx;Atoh7G/G;Six3-Cre retinas (C), and its co-expression with BrdU-labeled RPCs in E14.5 retinas (A1–C1). (D–F) Isl1 expression in E13.5 RESTfx/+ (D), RESTfx/+;Atoh7G/G (E), and RESTfx/fx;Atoh7G/G;Six3-Cre (F) retinas, and its co-expression with BrdU-labeled RPCs in E14.5 retinas (D1–F1). Note that a dorsal to ventral descending gradient is seen in panels C, F, C1, and F1. Arrowheads in panels A1–F1 point to mitotically active RPCs. (G–I) TUNEL assay for cell death. Arrowheads in panels G–I point to dying cells. Scale bars in panels A and A1: 100 μm. VbND, ventral–dorsal axis.

and Isl1 expression, and it further suggested a role for Neurod1 in RGC results were consistent with the hypothesis that Neurod1 contributes formation. The results prompted us to determine whether Atoh7- to the Atoh7-independent program for RGC development and may be Neurod1-REST triple-mutant retinas would lead to a loss of the required for the formation of the small population of RGCs that form upregulation of Pou4f2 that are observed in Atoh7-REST double in the absence of Atoh7. mutant retinas. We interbred Atoh7G/+, Neurod1+/−, and RESTfx/+ − − mice to generate Atoh7G/G: Neurod1 / : RESTfx/fx; Six3-Cre embryos. To Discussion determine the number of speciﬁed RGCs, we compared Pou4f2 expression between Atoh7G/G;RESTfx/+, Atoh7G/G;RESTfx/fx; Six3-Cre, Investigations of the biological role of REST in neurogenesis have − − and Atoh7G/G;Neurod1 / ;RESTfx/fx; Six3-Cre retinas. We found that been hampered by a lack of readily accessible in vivo models. The the aberrantly upregulated Pou4f2 in Atoh7G/G: RESTfx/fx; Six3-Cre mouse retina is particularly suited for use in addressing the role of retinas was reduced in the triple-mutant retinas (Fig. 10). These REST in neurogenesis and sensory neuron development in the context of an intact animal. Extensive knowledge of the major regulatory processes that control neural development in the retina provides a framework into which information gained on REST can be integrated. Our study establishes for the ﬁrst time an important in vivo role for REST in mammalian retinogenesis. By conditionally deleting REST in the developing retina, we uncovered the existence of a novel RGC gene expression program that operates independently of the proneural bHLH gene Atoh7. The Atoh7-independent program activates the expression of Pou4f2 and Isl1, two early expressing transcription factors required for commitment to an RGC fate. Remarkably, the expression of these regulatory genes, which is normally restricted to differentiating RGCs that have exited the cell cycle, could be induced in mitotically active progenitors once REST suppression was relieved. Moreover, RGC numbers increased signif- icantly in the REST-deleted retinas of developing embryos and postnatal mice further suggests that REST plays an important role in Fig. 8. Upregulation of Atoh7-downstream RGC genes in REST mutant retinas. qRT-PCR suppressing RGC differentiation in actively dividing RPCs. In Atoh7- analysis of Atoh7-dependent genes Pou4f2, Isl1, Stmn2, GAP43, and Persyn was REST double mutant retinas, Pou4f2, Isl1, and some but not all performed using retinas from 4 different genotypes. The y-axis is the relative transcript downstream genes in the RGC gene regulatory network were level using GAPDH as internal control. Note that Isl1 and GAP43 are not direct targets of REST. The sequences of the PCR primers are shown in Materials and methods (n=3, *: activated. However, RGCs that formed in the Atoh7-REST double pb0.05; **: pb0.01). mutant retinas were abnormal and did not survive. This indicates that C.-A. Mao et al. / Developmental Biology 349 (2011) 90–99 97

Fig. 9. Upregulation of RGC genes downstream of Pou4f2 and Isl1 in Atoh7-REST double mutant retinas. (A–C1) Immunostaining. (D–F1) In situ hybridization. Eomes (A, A1) and GAP43 (D, D1) expression was detected in the neuroblast layer in Atoh7G/G;RESTfx/fx;Six3-Cre retinas. The expression of other downstream RGC genes, including TUJ1 (B, B1), GSK-3® (C, C1), Persyn (E, E1), and Stmn2 (F, F1), was not detectable in the neuroblast layer but was detectable in the ganglion cell layer. Scale bars in panels A1 and D: 200 μm. NBL, neuroblast layer.

Atoh7 is required for normal RGC development even when REST Neurod1 has been shown to function in the formation of amacrine suppression is relieved. As is the case in other neuronal progenitors, cells and photoreceptor cells, and in the maintenance and survival of REST-mediated gene repression in RPCs most likely functions to photoreceptor cells in postnatal life (Inoue et al., 2002; Liu et al., 2008; prevent premature RGC differentiation and ensure that proper Ochocinska and Hitchcock, 2009). Conversely, Atoh7 is likely to have spatiotemporal expression patterns are maintained in both mitoti- other roles in the developing retina besides specifying RGC fate cally active RPCs and newly committed RGCs (Su et al., 2004; Ballas because many postmitotic Atoh7-expressing RPCs give rise to other et al., 2005). De-repression of REST likely relieves the repressive retinal cell types (Yang et al., 2003). Atoh7, Neurod1, and other chromatin in RGC genes, which subsequently allows the access of proneural bHLH genes are thus likely to have multiple functions in the Atoh7 and other transcriptional activators to initiate the expression of developing retina that cannot be identified by simple analysis of RGC genes. single-gene knockouts. Residual RGCs are always observed in Atoh7-mutant retinas, REST-mediated suppression appears to be a highly conserved indicating that a few RGCs can form independently of Atoh7 (Lin mechanism that has evolved in vertebrates to repress the expression et al., 2004; Moshiri et al., 2008). Our analysis of Atoh7-Neurod1-Rest of genes associated with neuronal differentiation (Chen et al., 1998; triple-mutant retinas suggests that Neurod1 regulates an Atoh7- Coulson, 2005; Majumder, 2006). Because it is a potent transcriptional independent pathway for RGC fate specification. Because only a few repressor, REST might have originally functioned more broadly to RGCs appeared to develop by this pathway, their loss would be repress terminal differentiation genes in actively dividing cells difficult to detect in Neurod1-null retinas (Pennesi et al., 2003). (Majumder, 2006). In neural progenitors, REST's suppressive function

Fig. 10. Atoh7-Neurod1-REST triple-mutant retinas contain fewer Pou4f2-positive RGCs than do Atoh7-REST double mutant retinas. (A–C) Pou4f2 expression in E13.5 Atoh7G/G RESTfx/+ (A), Atoh7G/G;RESTfx/fx;Six3-Cre (B), and Atoh7G/G ; Neurod1lz/lz;RESTfx/fx;Six3-Cre retinas (C). Scale bars in panel C: 100μm. 98 C.-A. Mao et al. / Developmental Biology 349 (2011) 90–99

Fig. 11. Model for REST function in retinal cell fate determination. (A) In the early stages of retinal development (E12–E16), REST suppresses RGC genes in RPCs to prevent premature activation and maintain an appropriate balance of proliferating RPCs and differentiating RGCs. REST also regulates the expression of some proneural bHLH genes, such as Neurod1,to fine-tune the expression of downstream target genes. Neurod1, and perhaps other bHLH factors, appear to have supplementary roles in regulating RGC gene expression. (B) In the absence of REST, additional RGCs are formed as the result of de-repression of RGC gene expression in RPCs, as well as de-repression of Neurod1, which then supplements Atoh7 in upregulating RGC genes. (C) In later stages of retinal development (E16 onwards), the retinal environment is no longer conducive to production of RGCs. REST may respond to local environmental signals dynamically to differentially bind to distinct RE1 sites within different bHLH genes and repress expression levels in a lineage- or cell-specific manner. The expression level of a particular bHLH gene within a particular RPC will influence the competency state and the decision about when to commence differentiation. Gray arrow on the left represents the developmental progression.

is likely to be intertwined with gene regulatory networks involved in cy states (Mu et al., 2005; Le et al., 2006). REST and the proneural neuronal differentiation. This appears to be the case in RPCs, where bHLH genes that REST represses may constitute a double-negative RGC genes repressed by REST are activated by Atoh7-independent and gate to regulate downstream outputs in different RPC lineages as has Atoh7-dependent mechanisms. The existence of the Atoh7-indepen- been observed for other developmental gene regulatory networks dent process was unexpected but it might be necessary for fine-tuning (Fig. 11; Revilla-i-Domingo et al., 2007). This model could be tested by the production of RGCs from RPCs. REST is an important epigenetic using lineage-specific promoter-Cre constructs or by specifically factor, suggesting that RGC genes are dynamically regulated through deleting RE1 sites associated with the different proneural bHLH REST-mediated epigenetic mechanisms. However, it is presently genes that control the sequential formation of the retinal cell types. unclear how REST responds to the local environment to engage Our study highlights the complexity associated with gene with or disengage itself from RE1 elements within RGC gene regulatory networks operating in the developing retina. We predict regulatory regions. that other transcriptional regulators involved in RGC formation will be REST clearly has more general functions in regulating neuronal identified as downstream targets of REST, thereby reinforcing REST's gene expression. Upon neuronal differentiation, REST is released from place at a distinct node in the gene regulatory network controlling chromatin-repressed RE1 sites in the regulatory regions of neuronal RGC development. Our study raises questions as to how redundant genes and is rapidly degraded at the G2 phase (Ballas et al., 2005; mechanisms for RGC formation arose and why they persist. The Guardavaccaro et al., 2008; Westbrook et al., 2008), a time when conditional floxed REST allele will be a valuable tool for addressing several proneural bHLH genes begin to be expressed and when these questions, and more generally, for investigating the role of REST subpopulations of progenitors, including RPCs, make the decision to in neurogenesis. exit the cell cycle and adopt a competency state (Ballas et al., 2005; Supplementary materials related to this article can be found online Lunyak and Rosenfeld, 2005). Several proneural bHLH genes, at doi: 10.1016/j.ydbio.2010.10.008. including Math3, Neurod1, and Ngn2, contain distinct RE1 sites and are direct targets of REST (Mortazavi et al., 2006; Johnson et al., 2007; Otto et al., 2007). In retinogenesis, each of these proneural bHLH Acknowledgments genes has evolved specialized functions that are essential for regulating the specification of individual retinal cell fates (Hata- We especially thank our colleague, Xiuqian Mu, for providing his keyama and Kageyama, 2004). REST is known to occupy distinct RE1 expertise on REST RE1 sites in RGC genes. We thank Yas Furuta for elements and repress downstream targets in a cell-type-specific providing the Six3-Cre transgenic mice, Jan Parker-Thornburg and the manner (Bruce et al., 2009). The release of REST-mediated repression Genetically Engineered Mouse Facility at The University of Texas M. D. and the subsequent degradation of REST at the G2 phase must relieve Anderson Cancer Center for generating the floxed REST-mouse, and the chromatin-repressed state of different proneural bHLH genes at Ming-Jer Tsai (Baylor College of Medicine) for Neurod1 mice. We different times and in different subpopulations of RPCs as retinogen- acknowledge the M. D. Anderson Cancer Center DNA Analysis Facility esis proceeds, hence enabling those RPCs to adopt a particular for DNA sequencing. The Genetically Engineered Mouse Facility and competency state (Fig. 11). This model implies that REST functions as a DNA Analysis Facility are supported in part by a National Cancer key regulatory link between extrinsic signals and intrinsic gene Institute Cancer Center Core Grant (CA016672). The work was regulatory programs to advance RPCs to particular cell fates (Fig. 11). supported by grants to W.H.K. from the National Eye Institute Besides REST repression, proneural bHLH genes have been shown to (EY011930 and EY010608-139005) and from the Robert A. Welch repress one another's expression, thus adding another layer of Foundation (G-0010), and grant to M.C.B. from the National Institutes complexity to the regulatory mechanisms controlling RPC competen- of Health (GM081627). C.-A. Mao et al. / Developmental Biology 349 (2011) 90–99 99

References Mortazavi, A., Chen, E., Thompson, L., Garcia, S.T., Myers, R.M., Wold, B., 2006. Comparative genomics modeling of the NRSF/REST repressor network: from single conserved sites to genome-wide repertoire. Genome Res. 16, 1208–1221. Agathocleous, M., Harris, W.A., 2009. From progenitors to differentiated cells in the Moshiri, A., Gonzalez, E., Tagawa, K., Maeda, H., Wang, M., Frishman, L.J., Wang, S.W., vertebrate retina. Annu. Rev. Cell Dev. Biol. 25, 45–69. 2008. Near complete loss of retinal ganglion cells in the math5/brn3b double Ballas, N., Grunseich, C., Lu, D.D., Speh, J.C., Mandel, G., 2005. REST and its corepressors knockout elicits severe reductions of other cell types during retinal development. mediate plasticity of neuronal gene chromatin throughout neurogenesis. Cell 121, Dev. Biol. 316, 214–227. 645–657. Mu, X., Klein, W.H., 2008. Gene regulatory networks and the development of retinal Brown, N.L., Patel, S., Brzezinski, J., Glaser, T., 2001. Math5 is required for retinal ganglion cells. In: Chalupa, L.M., Williams, R.W. (Eds.), Eye, Retina, and Visual ganglion cell and optic nerve formation. Development 128, 2497–2508. System of the Mouse. MIT Press, Cambridge, MA, pp. 321–332. Bruce, A.W., Donaldson, I.J., Wood, I.C., Yerbury, S.A., Sadowski, M.I., Chapman, M., Mu, X., Beremand, P.D., Zhao, S., Pershad, R., Sun, H., Scarpa, A., Liang, S., Thomas, T.L., Gottgens, B., Buckley, N.J., 2004. Genome-wide analysis of repressor element 1 Klein, W.H., 2004. Discrete gene sets depend on POU domain transcription factor silencing transcription factor/neuron-restrictive silencing factor (REST/NRSF) Brn3b/Brn-3.2/POU4f2 for their expression in the mouse embryonic retina. target genes. Proc. Natl Acad. Sci. USA 101, 10458–10463. Development 131, 1197–1210. Bruce, A.W., López-Contreras, A.J., Flicek, P., Down, T.A., Dhami, P., Dillon, S.C., Koch, C.M., Mu, X., Fu, X., Sun, H., Beremand, P.D., Thomas, T.L., Klein, W.H., 2005. A gene network Langford, C.F., Dunham, I., Andrews, R.M., Vetrie, D., 2009. Functional diversity for REST downstream of transcription factor Math5 regulates retinal progenitor cell (NRSF) is defined by in vivo binding affinity hierarchies at the DNA sequence level. competence and ganglion cell fate. Dev. Biol. 280, 467–481. Genome Res. 19, 994–1005. Mu, X., Fu, X., Beremand, P.D., Thomas, T.L., Klein, W.H., 2008. Gene regulatory logic in Cayouette, M., Poggi, L., Harris, W.A., 2006. Lineage in the vertebrate retina. Trends retinal ganglion cell development: Isl1 defines a critical branch distinct from but Neurosci. 29, 563–570. overlapping with Pou4f2. Proc. Natl Acad. Sci. USA 105, 6942–6947. Chalupa, L.M., Williams, R.W. (Eds.), 2008. Eye, Retina, and Visual System of the Mouse. Nelson, B.R., Gumuscu, B., Hartman, B.H., Reh, T.A., 2006. Notch activity is down- MIT Press, Cambridge, MA. regulated just prior to retinal ganglion cell differentiation. Dev. Neurosci. 281, Chen, Z.F., Paquette, A.J., Anderson, D.J., 1998. NRSF/REST is required in vivo for repression 128–141. of multiple neuronal target genes during embryogenesis. Nat. Genet. 20, 136–142. Ochocinska, M.J., Hitchcock, P.F., 2009. NeuroD regulated proliferation of photoreceptor Chong,J.A.,Tapia-Ramirez,J.,Kim,S.,Toledo-Aral, J.J., Zheng, Y., Boutros, M.C., Altshuller, Y.M., progenitors in the retina of zebrafish. Mech. Dev. 126, 128–141. Frohman, M.F., Kraner, S.D., Mandel, G., 1995. REST: a mammalian silencer protein that Otto, S.J., McCorkle, S.R., Hover, J., Conaco, C., Han, J.J., Impey, S., Yochum, G.S., Dunn, J.J., restricts sodium channel gene expression to neurons. Cell 80, 949–957. Goodman, R.H., Mandel, G., 2007. A new binding motif for the transcriptional Coulson, J.M., 2005. Transcriptional regulation: cancer, neurons, and the REST. Curr. repressor REST uncovers large gene networks devoted to neuronal functions. J. Biol. 15, R665–668. Neurosci. 27, 6729–6739. Farley, F.W., Sorioano, P., Steffen, L.S., Dymecki, S.M., 2000. Widespread recombinase Palm, K., Belluardo, N., Metsis, M., Timmusk, T., 1998. Neuronal expression of zinc finger expression using FLPeR (Flipper) mice. Genesis 28, 106–110. transcription factor REST/NRSF/XBR gene. J. Neurosci. 18, 1280–1296. Furuta, Y., Lagutin, O., Hogan, B.L., Oliver, G.C., 2000. Retina- and ventral forebrain- Pan, L., Deng, M., Xie, X., Gan, L., 2008. ISL1 and BRN3B co-regulate the differentiation of specific Cre recombinase activity in transgenic mice. Genesis 26, 130–132. murine retinal ganglion cells. Development 135, 1981–1990. Gan, L., Wang, S.W., Huang, Z., Klein, W.H., 1999. POU domain factor Brn-3b is essential Pennesi, M.E., Cho, J.H., Yang, Z., Wu, S.H., Zhang, J., Wu, S.M., Tsai, M.J., 2003. BETA2/ for retinal ganglion cell differentiation and survival but not for initial cell fate NeuroD1 null mice: a new model for transcription factor-dependent photoreceptor specification. Dev. Biol. 210, 469–480. degeneration. J. Neurosci. 15, 453–461. George, S.H.L., Gertsenstein, M., Vintersten, K., Korets-Smith, E., Murphy, J., Stevens, M.E., Perron, M., Harris, W.A., 2000. Determination of vertebrate retinal progenitor cell fate Haigh, J.J., Nagy, A., 2007. Developmental and adult phenotyping directly from mutant by the Notch pathway and basic helix–loop–helix transcription factors. Cell. Mol. embryonic stem cells. Proc. Natl Acad. Sci. USA 104, 4455–4460. Life Sci. 57, 215–223. Guardavaccaro, D., Frescas, D., Dorrello, N.V., Peschiaroli, A., Multani, A.S., Cardozo, T., Revilla-i-Domingo, R., Oliveri, P., Davidson, E.H., 2007. A missing link in the sea urchin Lasorella, A., Iavarone, A., Chang, S., Hernando, E., Michele Pagano, M., 2008. Control embryo gene regulatory network: hesC and the double-negative specification of of chromosome stability by the ®-TrCP–REST–Mad2 axis. Nature 452, 365–369. micromeres. Proc. Natl Acad. Sci. USA 104, 12383–12388. Harada, T., Harada, C., Parada, L.F., 2007. Molecular regulation of visual system Riesenberg, A.N., Liu, Z., Kopan, R., Brown, N.L., 2009. Rbpj cell autonomous regulation development: more than meets the eye. Genes Dev. 21, 367–378. of retinal ganglion cell and cone photoreceptor fates in the mouse retina. J. Hatakeyama, J., Kageyama, R., 2004. Retinal cell fate determination and bHLH factors. Neurosci. 29, 12865–12877. Semin. Cell Dev. Biol. 15, 83–89. Schoenherr, C.J., Anderson, D.J., 1995. The neuron-restrictive silencer factor (NRSF): Hernandez, J., Matter-Sadzinski, L., Skowronska-Krawczyk, D., Chiodini, F., Alliod, C., a coordinate repressor of multiple neuron-specific genes. Science 267, Ballivet, M., Matter, J.M., 2007. Highly conserved sequences mediate the dynamic 1360–1363. interplay of basic helix–loop–helix proteins regulating retinogenesis. J. Biol. Chem. Shimojo, M., Paquette, A.J., Anderson, D.J., Hersh, L.B., 1999. Protein kinase A regulates 282, 37894–37905. cholinergic gene expression in PC12 cells: REST4 silences the silencing activity of Inoue, T., Honjo, M., Bessho, Y., Tano, Y., Lee, J.E., Kageyama, R., 2002. Math3 and NeuroD neuron-restrictive silencer factor/REST. Mol. Cell. Biol. 19, 6788–6795. regulate amacrine cell fate specification in the retina. Development 129, 831–842. Smith, M.D., Parker, A., Wikaningrum, R., Coleman, M., 2000. Combined immunohis- Jadhav, A.P., Roesch, K., Cepko, C.L., 2009. Development and neurogenic potential of tochemical labeling and in situ hybridization to colocalize mRNA and protein in Müller glial cells in the vertebrate retina. Prog. Retin. Eye Res. 28, 249–262. tissue sections. In: Darby, I.A. (Ed.), Methods in Molecular Biology, Vol. 123: In Situ Johnson, D.S., Mortazavi, A., Myers, R.M., Wold, B., 2007. Genome-wide mapping of in Hybridization Protocols. Humana Press, Totowa, NJ, pp. 165–175. vivo protein–DNA interactions. Science 316, 1497–1502. Souren, M., Martinez-Morales, J.R., Makri, P., Wittbrodt, B., Wittbrodt, J., 2009. A global Lamba, D.A., Karl, M.O., Reh, R.A., 2009. Strategies for retinal repair: cell replacement survey identifies novel upstream components of the Ath5 neurogenic network. and regeneration. Prog. Brain Res. 175, 23–31. Genome Biol. 10, R92. Le, T.T., Wroblewski, E., Patel, S., Riesenberg, A.N., Brown, N.L., 2006. Math5 is required for Su, X., Kameoka, S., Lentz, S., Majumder, S., 2004. Activation of REST/NRSF target genes both early retinal differentiation and cell cycle progression. Dev. Biol. 295, 764–768. in neural stem cells is sufficient to cause neuronal differentiation. Mol. Cell. Biol. 24, Lee, J.-H., Chai, Y.-G., Hersh, L.B., 2000. Expression patterns of mouse repressor element- 8018–8025. 1 silencing transcription factor 4 (REST4) and its possible function in neuroblas- Sun, Y.M., Cooper, M., Finch, S., Lin, H.H., Chen, Z.F., Williams, B.P., Buckley, N.J., 2008. toma. J. Mol. Neurosci. 15, 205–214. Rest-mediated regulation of extracellular matrix is crucial for neural development. Lin, B., Wang, S.W., Masland, R.H., 2004. Retinal ganglion cell type, size, and spacing can PLoS ONE 3, e3656. be specified independent of homotypic dendritic contacts. Neuron 43, 475–485. Tsai, W.W., Nguyen, T.T., Shi, Y., Barton, M.C., 2008. p53-targeted LSD1 functions in Liu, H., Etter, P., Hayes, S., Jones, I., Nelson, B., Hartman, B., Forrest, D., Reh, T.A., 2008. repression of chromatin structure and transcription in vivo. Mol. Cell. Biol. 28, NeuroD1 regulates expression of thyroid hormone receptor 2 and cone opsins in 5139–5146. the developing mouse retina. J. Neurosci. 16, 749–756. Wang, S.W., Kim, B.S., Ding, K., Wang, H., Sun, D., Johnson, R.L., Klein, W.H., Gan, L., 2001. Livesey, F.J., Cepko, C.L., 2001. Vertebrate neural cell-fate determination: lessons from Requirement for math5 in the development of retinal ganglion cells. Genes Dev. 15, the retina. Nat. Rev. Neurosci. 2, 109–118. 24–29. Lunyak, V.V., Rosenfeld, M.G., 2005. No rest for REST: REST/NRSF regulation of Watanabe, Y., Kameoka, S., Gopalakrishnan, V., Aldape, K.D., Pan, Z.Z., Lang, F.F., neurogenesis. Cell 121, 499–501. Majumder, S., 2004. Conversion of myoblasts to physiologically active neuronal Majumder, S., 2006. REST in good times and bad: roles in tumor suppressor and phenotype. Genes Dev. 18, 889–900. oncogenic activates. Cell Cycle 5, 1929–1935. Westbrook, T.F., Hu, G., Ang, X.L., Mulligan, P., Pavlova, N.N., Liang, A., Leng, Y., Maehr, R., Mao, C.A., Kiyama, T., Pan, P., Furuta, Y., Hadjantonakis, A.K., Klein, W.H., 2008a. She, Y., Harper, J.W., Elledge, S.J., 2008. SCFbeta-TRCP controls oncogenic Eomesodermin, a target gene of Pou4f2, is required for retinal ganglion cell and transformation and neural differentiation through REST degradation. Nature 452, optic nerve development in the mouse. Development 135, 271–280. 370–374. Mao, C.A., Wang, S.W., Pan, P., Klein, W.H., 2008b. Rewiring the retinal ganglion cell Willardsen, M.I., Suli, A., Pan, Y., March-Armstrong, N., Chien, C.B., El-Hodiri, H., Brown, gene regulatory network: Neurod1 promotes retinal ganglion cell fate in the N.L., Moors, K.B., Vetter, M.L., 2009. Temporal regulation of Ath5 gene expression absence of Math5. Development 135, 3379–3388. during eye development. Dev. Biol. 326, 471–481. Morrow, E.M., Furukawa, T., Lee, J.E., Cepko, C.L., 1999. NeuroD regulates multiple Yang, Z., Ding, K., Pan, L., Deng, M., Gan, L., 2003. Math5 determines the competence functions in the developing neural retina in rodent. Development 126, 23–36. state of retinal ganglion cell progenitors. Dev. Biol. 264, 240–254.