Positional cloning of the young mutation identifies an essential role for the Brahma chromatin remodeling complex in mediating retinal cell differentiation

Abstract

Zebrafish with the young (yng) mutation show a defect in retinal cell differentiation. Here we demonstrate that a mutation in a brahma-related gene (brg1) is responsible for the yng phenotype. Brahma homologues function as essential subunits for SWI/SNF-type chromatin remodeling complexes. Our analysis indicates that brg1 is required for the wave of mitogen-activated protein kinase activity that precedes retinal cell differentiation. Using specific inhibitors of the mitogen-activated protein kinase pathway we show this signal has a direct role in retinal cell differentiation. Lastly, through investigations of mutants in other chromatin remodeling subunits, we provide genetic evidence for gene and tissue specificity of the Brahma chromatin remodeling complex.

The vertebrate retina develops from a morphologically uniform layer of proliferating neuroepithelial cells (1). These optic cup cells will produce a mature retina composed of three cellular layers. Each layer contains neurons of distinct cell classes that can be distinguished based on gene expression, morphology, and function. Retinogenesis, like development of other complex tissues, can be divided into several broad stages. These steps include cell cycle withdrawal of the multipotent precursor cells, cell type fate commitment, and then cellular differentiation where genetic pathways are initiated to promote the unique biochemical and morphological specializations for each cell type. Although many of the molecules that coordinate early events have been identified, genes that mediate retinal cell differentiation pathways have remained elusive (2).

In the developing zebrafish retina, the stages of retinogenesis progress as waves, moving first in a circular fashion from the ventronasal to the ventrotemporal retina, and then outward in a central to peripheral fashion (3). Cell cycle withdrawal initiates at the point where the optic stalk contacts the optic cup (4). Once the first cells begin to leave the cell cycle, a wave of terminal mitosis sweeps concentrically around the optic cup. The first cells to withdraw from the cell cycle are fated to become ganglion cells, the inner-most cell type. Successive waves of terminal mitosis precede the commitment of cells to inner nuclear layer fates and then, finally, to outer retinal cell fates (5).

Some of the molecules that coordinate these early developmental events within the retina have been identified. For example, Ath5, a basic helix–loop–helix transcription factor, is expressed in a wave at the time of cell cycle withdrawal (4, 6). Disruption of this gene delays cell cycle withdrawal and promotes amacrine cell fates at the expense of ganglion cells (7, 8). At the time of ath5 expression and cell cycle withdrawal, waves of hedgehog gene expression sweep through the retina. Hedgehog activity has been shown to regulate cell fate decisions and laminar organization (9–11, 12). After the waves of hedgehog, ath5, and cell cycle withdrawal, a wave of mitogen-activated protein (MAP) kinase activity progresses through the retina (9). This activity is thought to depend on hedgehog gene expression, but the significance of the MAP kinase activity is unknown. Finally, cellular morphogenesis and the organization of plexiform layers also occur in a concentric wave starting from the optic stalk (13, 14). Little is known about the regulation of morphogenesis, and the molecules that facilitate cellular differentiation have not been identified.

From genetic studies in invertebrates and in vitro experiments in mammalian cells, chromatin remodeling complexes have been implicated in coordinating differentiation. Chromatin remodeling complexes are large macromolecular protein complexes (>2 MDa) that use the energy of ATP to slide nucleosomes along DNA and create local alterations in chromatin structure (15). Chromatin remodeling complexes can be grouped into three classes based on the type of ATPase subunit present: SWI2/SNF2, ISWI, or Mi-2 (16). Within each of these classes, specific complexes are defined by their exact subunit composition. For example in mammals, two SWI2/SNF2 ATPases exist: Brg1 and Brm. The temporal and spatial expression pattern for each subclass of chromatin remodeling complex is differentially regulated, suggesting cell- and tissue type-specific functions. In addition to remodeling activity, proteins within these large complexes have been shown to interact directly with both basal transcriptional machinery and gene-specific DNA-binding factors (17). By opening chromatin and recruiting transcription regulators, these complexes are believed to be key regulators of gene transcription during development.

An example of a chromatin remodeling complex with developmental significance comes from work in Drosophila on the brahma gene. The brahma gene encodes a SWI2/SNF2 ATPase and was identified as a suppressor of polycomb mutations and subsequently shown to interact with other developmentally significant proteins (18). Analysis of brahma mutations demonstrated roles during development of the fly nervous system, mechanosensory bristles, and the wings. In vitro experiments have suggested the Brahma complex may be important for differentiation events within mammalian cells. For example in muscle precursor cells in which dominant negative versions of brahma related 1 (brg1) or brm were induced, cell cycle arrest occurred, but activation of myoD and other differentiation events could not proceed (19, 20). Similarly, a human pheochromocytoma-derived cell line (SW-13), which is deficient in brg1, will not undergo differentiation until cDNA encoding WT Brg1 is transfected into the cells (21, 22).

Through forward genetic analysis, we have identified several zebrafish mutations that are deficient in retinal cell differentiation. One mutation, young (yng), shows the normal pattern of cell cycle withdrawal and cell fate determination, but fails to undergo cellular differentiation and morphogenesis (23). In the present study, we used a combination of positional cloning and bacterial artificial chromosome (BAC) rescue experiments to identify brg1 as the gene disrupted in yng mutants. In our analysis, we address the mechanism by which the Brahma chromatin remodeling complex facilitates retinal cell differentiation. Similar to yng/brg1 mutants, we found retinal dismorphogenesis in mutations for another subunit of the SWI/SNF Brahma complex (baf53). In contrast, mutations in the ATPase for an ISWI chromatin remodeling complex (snf2h) did not cause defects in retinal cell differentiation. These data indicate that ATPase-dependent chromatin remodeling complexes have tissue specificity and the Brahma complex is required for retinal cell differentiation.

Materials and Methods

Positional Cloning. For all experiments the yng^a8 allele was used. Genomic DNA was isolated from homozygous yng mutant embryos, and simple sequence length polymorphism markers from Research Genetics (Huntsville, AL) were used to establish linkage to LG3 (24). To obtain high-resolution mapping data 1,012 embryos were genotyped, and the critical region containing the yng mutation was refined to between markers z1140 and z5623. Single-nucleotide polymorphisms (SNPs) were identified from ESTs that had been mapped to the critical region by amplifying a fragment of each EST from both parental fish and cloning and sequencing the fragment. Embryos showing recombination in the critical region were then genotyped for each SNP by using either primer extension or when possible a restriction fragment length polymorphism. EST fb79a02 showed no recombination with yng and it was used to screen zebrafish BAC (PCR-based) and P1-derived artificial chromosome (PAC) (filter hybridization-based) libraries from Incyte Genomics (Palo Alto, CA) according to protocols suggested by the manufacturer. Positive clones were purified by using a PSIΨClone BAC DNA kit (Princeton Separations, Adelphia, NJ) and end-sequenced. Primers designed against the termini were used to reconfirm the positions of the BAC/PAC ends on LG3 and to extend the contig. BAC/PAC ends were placed on the genetic map by using SNPs with the same strategy as was used for the ESTs.

BAC/PAC/cDNA Phenotype Rescue. BAC and PAC clones that comprised the critical region to which yng mapped were injected into one- or two-cell stage embryos from yng heterozygous crosses (25). Equivalent numbers of embryos were injected with buffer as controls. Mutant embryos were identified at 34 h postfertilization (hpf) based on decreased pigmentation and the heart phenotype. At 84 hpf yng and WT embryos were processed for cryosectioning and scored for retinal tyrosine hydroxylase (TH) immunoreactivity. Only WT and mutant embryos injected with PAC 81j4 (and later with brg1 cDNA) showed retinal expression of the differentiated interplexiform cell marker.

Mutation Detection. EcoRI libraries were made from BAC 38j4 and PACs 81j4 and 182p8 by cloning digested fragments into pZErO-2 (Invitrogen). Shotgun libraries were constructed from PACs 81j4 and 182p8 by using the Invitrogen TOPO Shotgun Subcloning kit following the manufacturer's instructions. The ends of all representative fragments from EcoRI clones along with ≈200 shotgun clones from each BAC/PAC were sequenced and assembled by using SEQMAN software (DNASTAR, Madison, WI). Remaining sequence gaps were filled in by primer walking. BLAST searches of GENSCAN-predicted proteins from PAC 81j4 showed sequence identity to two known genes, ldlr and brg1. Searches against the zebrafish EST database identified multiple clones from the 5′ and 3′ ends of brg1. These sequences were used to design primers (forward, 5′-ACTCGCTGAAGCTGCCCTTT-3′; and reverse, 5′-CTCCCCCATTCCTCCTGACT-3′) that encompassed the entire brg1 ORF. Total RNA was extracted from a pool of 75 yng embryos by using TRIzol Reagent (Invitrogen). cDNA was synthesized by using Super-Script II RnaseH^– Reverse Transcriptase (Invitrogen), and a full-length fragment containing the brg1 ORF was amplified by using PfuTurbo hotstart DNA polymerase (Stratagene) from WT and yng embryos. Amplified products were cloned, sequenced, and analyzed for potential mutations. The identified C→A transversion that introduces a premature stop codon (Y390X) into the yng allele also destroys a RsaI restriction enzyme site, which was subsequently used to confirm the mutation was present in genomic DNA from yng embryos.

Morpholino Antisense Knockdown. We designed two nonoverlapping morpholino oligonucleotides (Gene Tools, Philomath, OR) antisense to the predicted translational start site of brg1. Brg1 MO1 bound–27 to –3, whereas Brg1 MO2 bound–2 to +24. These oligonucleotides, 15–30 μM, were injected into one- or two-cell stage WT eggs, and these embryos were screened at 2 and 3 days postfertilization for the yng phenotype.

Histology. Semithin retinal sections were obtained after fixing embryos overnight at 4°C in 2.5% gluteraldehyde/1% paraformaldehyde/phosphate-buffered sucrose, pH 7.4. Embryos were dehydrated and infiltrated with Epon/Araldite. Transverse sections, 1–2 μm, were heat-mounted and stained with 1% methylene blue in 1% borax (14).

Riboprobes and in Situ Hybridization. To assess brg1 expression, parallel samples were hybridized with two nonoverlapping probes, one within the coding sequences and one within the 3′ UTR. Each gave equivalent expression patterns. Sonic hedgehog, tiggywinkle hedgehog, patched-2, and ath5 probes (gift from D. L. Stenkamp, University of Idaho, Moscow) were hybridized as described (11).

Detection of Activated Extracellular Signal-Regulated Kinase (ERK) 1/2. To detect activated ERK 1/2, embryos were pretreated 30 min with a mixture of tyrosine and serine/threonine phosphatase inhibitors (Sigma P2850 and P5726), each diluted 10^–2 in fish water. After phosphatase inhibitor treatment, embryos were fixed for 1 h at room temperature with 4% paraformaldehyde that included phosphatase inhibitors. Embryos were washed in PBS and processed for cryosectioning. Two independent antibodies that detect diphosphorylated ERK 1/2 were used: Sigma monoclonal MAPK-YT (M819) at 1:800 and Cell Signaling Technology (Beverly, MA) polyclonal phospho-p44/42 MAP kinase (9101) at 1:200. Each antibody showed equivalent staining patterns. Western blot analysis using protein extracts from zebrafish embryos was preformed to confirm antibody specificity with zebrafish.

MAP Kinase Inhibitor Experiments. MAP kinase pathway inhibitors U0126 (25 nM) and PD98059 (5 nM) were obtained from Cell Signaling Technology. Inhibitors or control dilutions of DMSO were microinjected into the eyes of anesthetized embryos at 42 hpf. Phenol red (0.1%) was included to ensure the injected solution bathed the retina. To control for efficacy, a subset of U0126- and PD98059-injected embryos were assessed for dually phosphoylated ERK 1/2 immunoreactivity at 43 hpf. Analysis indicated a loss in MAP kinase activity within 1 h after injections. At 60 hpf, control and inhibitor-injected embryos were processed for zn8 (ganglion cell differentiation marker) and zpr1 (photoreceptor differentiation marker) cryosection immunoreactivity. Each antibody was obtained from the Zebrafish International Resource Center Monoclonal Antibody Bank at the University of Oregon, Eugene. The zn8 marker was included as an internal control to assess the survival of cells that had already differentiated and to identify central retina sections by the presence of the optic nerve and maximum lens diameter. Central retina sections were used to measure photoreceptor differentiation as scored by the number of zpr1 immunoreactive cells. Results are represented as the proportion of zpr1-positive cells in inhibitor-injected eyes to those injected with DMSO (n = 8 independent embryos for each condition).

Accession Numbers. GenBank accession nos. for full-length cDNA are AY205256 for zebrafish brg1 and AY218841 for PAC 81j4.

Results

Zebrafish embryos with the yng mutation show arrested retinal development. Their retinas express early cell-type markers indicating all major retinal cell types are specified; however, mutant retinas fail to complete cellular differentiation and morphogenesis (23). To identify the gene disrupted in yng mutants, 2,024 meioses were used to restrict the location of the mutant gene to a ≈0.2-centimorgan interval on linkage group 3. A BAC/PAC physical map encompassing the critical region was constructed (Fig. 1a), and each PAC was then injected into embryos from yng heterozygote intercrosses. Retinas from PAC-injected mutant embryos were then examined for expression of TH, which is a late marker for interplexiform cells and is not expressed in yng embryos. One PAC, 81j4, yielded yng embryos with a subset of retinal cells rescued for TH expression (Fig. 1b). PAC 81j4 was sequenced and GENSCAN analyses predicted the presence of two genes, a low-density lipoprotein receptor, ldlr, and a brahma homologue, brg1. In situ mRNA hybridization and RT-PCR showed that brg1 was maternally expressed and later found throughout the anterior portion of the embryo during ocular morphogenesis. Within the retina, brg1 was highly expressed in the differentiating neuroepithelium and at later stages restricted to the ganglion cell and inner nuclear layers (Fig. 1c). cDNA clones representing brg1 from yng embryos were obtained, and sequence analysis identified a C→A transversion at position 1170. This mutation introduced a premature stop codon (Y390X) into this 1,627-aa protein that truncates all predicted functional domains of brg1, including the chromodomain, SNF2 ATPase/helicase, and the bromodomain (Fig. 1d).

Fig. 1.

yng is caused by a mutation in brg1. (a) Genetic and physical map. Linkage was initially discovered between simple sequence length polymorphism markers z1140 and z5623. Polymorphisms within ESTs mapped to this region on the T51 radiation hybrid panel and were then used to refine the interval. EST fb79a02 was found to contain no recombinants of 2,024 meioses, and BAC and PAC clones were used to build a contig around this locus. PACs 182p8 and 81j4 and BAC 38j4 were sequenced to reveal a high level of synteny with human chromosome 19 based on blast results. Positive rescue experiments using PAC 81j4 limited the search to either zebrafish ldlr or brg1. Vertical lines are exons determined by PCR of embryonic cDNA and based on genscan predictions. (b) BAC/PAC rescue analysis. Transverse retinal sections showing the presence of TH (a late interplexiform cell marker, arrows) in a WT embryo injected with control buffer, a yng mutant injected with BAC 93m1 (which does not contain brg1 and is negative for TH expression), and a yng mutant injected with PAC 81j4 (which does contain brg1 and rescued TH expression). Arrows indicate retinal TH-positive neurons. TH is also normally expressed in the ventral forebrain of both yng and WT embryos (Right). (c) In situ mRNA analysis of brg1. Sixteen-cell stage embryos, a time before zygotic transcription begins, shows that brg1 transcript is provided maternally. Results with sense control and antisense riboprobes are shown. Whole-mount analysis at 48 hpf reveals strong anterior brg1 expression. Transverse retinal sections at 24, 48, and 60 hpf show robust brg1 expression during differentiation. (d) Sequence analysis and predicted protein structure. A C→ A transversion was identified in cDNA from yng homozygous mutants and later confirmed through genomic sequence analysis (red arrow). This base substitution changes a tyrosine codon to a stop codon very early with in the coding sequence of brg1 (red arrowhead). This truncation deletes all of the identified functional domains.

To confirm that the mutation in brg1 was responsible for the yng phenotype, we rescued the phenotype by expressing a full-length brg1 cDNA under the control of the EF1α promoter. Similar to results obtained with PAC rescue, injection of brg1 cDNA rescued markers for retinal differentiation within the mutant eyes (Table 1). Overexpression of brg1 in WT embryos did not result in any noticeable phenotype. We also used morpholino antisense techniques to knock down Brg1 translation in WT embryos (26). Either of two nonoverlapping morpholinos phenocopied the yng mutation when injected into newly fertilized WT embryos (Fig. 2). This included reduced numbers of melanocytes, disrupted heart, ear, and fin morphogenesis, and a lack of retinal lamination. Cumulatively, these experiments demonstrate that the identified mutation in brg1 is responsible for the yng phenotype.

Table 1. Phenotype rescue by injected DNA.

DNA	Total injected	Embryos at 24 hpf	yng mutants	Rescued mutants*
PAC81j4	293	265	58	15
BAC93m1	190	185	46	0
EF1 αbrgl	211	181	44	11
EF1 αnucGFP	163	139	34	0
Buffer alone	126	119	33	0

^*Rescue was assessed by scoring retinal expression of tyrosine hydroxylase.

Fig. 2.

Morpholinos directed against brg1 phenocopy yng. (a and b) Standard control oligo. (c and d) Brg1MO1. (e and f) yng mutant, uninjected. Morphological inspection (a, c, and e) at 3 days postfertilization of Brg1MO1 and yng mutant embryos reveals similar defects in heart, ear, fin, and pigment cell development. Retinal histology (b, d, and f) shows that brg1 morpholinos also phenocopy the yng retinal lamination phenotype. The nonoverlapping oligo Brg1MO2 showed equivalent results to Brg1MO1.

To examine the mechanisms by which the loss of brg1 contributes to the yng ocular phenotype, we investigated several potential regulators of retinal development in yng mutants. The transcription factor Ath5 is expressed in a wave throughout the retina just preceding cell cycle withdrawal and initiation of cell fate determination (4). After this wave of expression, ath5 becomes restricted to the ciliary margin where in zebrafish cell fate determination and differentiation is ongoing. In both invertebrate and vertebrate retinas, Ath5 is thought to regulate the expression of hedgehog genes, which in turn results in a wave of MAP kinase activity (9, 27, 28). Although loss-of-function studies with hedgehog genes suggest this pathway has a role in cell fate determination and patterning within the retina, the significance of MAP kinase activity in retinogenesis has not been tested (9–11, 29). In situ hybridization experiments showed similar expression in yng and WT retinas for ath5 and components of the hedgehog pathway (sonic hh, tiggy-winkle hh, and the retinal receptor patched-2). Furthermore, the dynamic wave-like pattern of ath5 and its restriction to the ciliary margin zone occurred in yng mutants (Fig. 3). The robust expression of these genes shows that an absence of brg1 does not simply result in a general suppression of transcription and, importantly, within the retina brg1 is essential for cellular differentiation events.

Fig. 3.

Expression of ath5 and shh is normal in yng retinas. In situ mRNA hybridizations of ath5 at 34, 72, and 96 hpf (a–f) and shh at 54 hpf (g and h) show similar patterns in yng (a, c, e, and g) and WT (b, d, f, and h) retinas. For each, whole-mount in situ hybridizations were performed and then transverse retinal sections were obtained.

In contrast to the above observations, the wave of MAP kinase activation that follows the wave of hedgehog gene expression is disrupted within yng retinas. Specifically, the activated form of ERK 1/2 is absent from yng mutant retinas (Fig. 4 a and b). These data indicate that Brg1 is required for MAP kinase signaling and suggest that MAP kinases are critical for retinal cell differentiation. To examine this directly we injected specific inhibitors of MAP kinase kinase (MEK) 1/2 (30, 31), which is the kinase that activates ERK 1/2, and examined the effects on retinal differentiation. Fig. 4 shows that retinal cell differentiation is disrupted by these inhibitors and supports our hypothesis that the MAP kinase pathway is critical for this process. Interestingly, MEK 1/2 inhibitor injections at earlier time points severely disrupted lens differentiation consistent with previous in vitro studies investigating the role of MAP kinase signaling on lens differentiation (32, 33).

Fig. 4.

Disruption of MAP kinase activity blocks retinal differentiation. (a and b) Transverse retinal sections showing dual phosphorylated ERK 1/2inWT (a) and the lack of retinal activation of MAP kinase in yng mutant (b) eyes. (c) Example of focal intraocular injection. MAP kinase kinase (MEK) 1/2 inhibitors were injected intraocularly at 42 hpf, a time when ganglion cells have differentiated, but before photoreceptor differentiation. Ganglion cell and photoreceptor markers were assessed at 60 hpf. Phenol red is used as a visible marker to monitor injections. (d–f) Retinal sections of embryos injected with DMSO control (d) or MEK 1/2 inhibitors U0126 (e) and PD98059 (f) and then processed for a mixture of markers that indicate ganglion cell differentiation (zn8, arrows) and photoreceptor differentiation (zpr1, *). Note the absence of photoreceptor differentiation in embryos treated with either U0126 or PD98059. Ganglion cell markers were not affected, demonstrating the inhibitors did not alter retinal cell survival. (g) Quantitative comparison of photoreceptor differentiation between embryos with intraocular injections of DMSO, U0126, or PD98059. After statistical comparison, data were transformed to percent of control. *, P < 0.001, two-tailed Student's t test; n = 8.

Chromatin remodeling complexes are thought to modulate the expression of many genes and might have overlapping functions (16, 34). To address specificity of the Brahma complex with regard to retinal differentiation, we analyzed the eyes of zebrafish with mutations in other chromatin remodeling genes (35). We found that mutations in baf53, another Brahma complex subunit that has been shown to bind with Brg1, also disrupted retinal histogenesis (Fig. 5) (36). However, mutations in snf2h, the ATPase of an ISWI chromatin remodeling complex (37), did not affect retinal cell differentiation (Fig. 5). Cumulatively, these results provide in vivo evidence for the role of distinct vertebrate chromatin remodeling complexes in tissue-specific differentiation processes. In particular, our analysis shows that brg1 function is essential for the wave of MAP kinase activation within the retina and this signaling event is necessary for retinal cell differentiation.

Fig. 5.

Analysis of additional chromatin remodeling mutants reveals specificity for function in retinal development. (a–c) Transverse sections through the central retina of WT (a), baf53 mutant hi550 (b), and snf2h mutant hi1072 (c). Baf53, which has been shown to complex with Brg1, has a retinal lamination phenotype similar to yng when mutated. snf2h, which encodes the ATPase for ISWI-type chromatin remodeling complexes, shows normal retinal differentiation. (d) Summary model: Ath5 and Hedgehog (Hh) are required for retinal cell type specification, whereas the Brahma complex is required for MAP kinase activation, which is necessary for retinal differentiation and morphogenesis.

Discussion

Studies in multiple vertebrate species have implicated Ath5 and hedgehog signals as important regulators of retinogenesis. In particular, Ath5 and hedgehog activity have been shown to regulate cell cycle withdrawal and cell-type fate decisions. For example, Ath5 is necessary for promoting specification of ganglion cells (7, 8), whereas hedgehog activity has been shown to be repressive for this cell fate decision (12). Sonic hedgehog expression has been linked to a wave of MAP kinase activity that progresses through the developing retina (9). The effect of Sonic hedgehog on MAP kinase activity is unlikely to be direct as hedgehog ligands are thought to signal through distinct kinases such as Fused and protein kinase A (38). We found that both Sonic and Tiggywinkle hedgehog, and the hedgehog receptor Patched-2, are expressed in brg1 mutants. By inhibiting MAP kinase activation independent from mutations in the Brahma complex, we show that MAP kinase signaling is critical for retinal cell differentiation. Fig. 5d summarizes these results and places Brahma and MAP kinase as differentiation factors downstream of Ath5 and hedgehog signals.

An intriguing question raised from our results is how does the absence of brg1 adversely affect the MAP kinase pathway? One possibility is that the Brahma complex regulates expression of a secreted factor that induces a localized MAP kinase signaling cascade and promotes retinal cell differentiation. This hypothesis is consistent with the observation that differentiation markers and overt morphogenesis proceeds in a localized wave throughout the retinal epithelium. This envisioned role of Brg1 is also consistent with results from genetic mosaic experiments that demonstrated the yng mutation acts non-cell-autonomously for retinal cell differentiation (23). Interestingly, transcript profile experiments in cell culture have identified several secreted factors regulated by Brg1 (CD44, SPARC, IGF1) that are important for MAP kinase regulation (21, 39). Alternatively, the Brahma complex may intrinsically regulate cytoplasmic activators of MAP kinases. In this view, MAP kinase activity would then promote gene transcription of a secreted differentiation signal. Experiments such as chromatin immunoprecipitations are necessary to define precisely which genes the Brahma complex regulates within the retina.

An important feature of the yng mutant is its ability to progress through early development. This is in striking contrast to brg1 knockout mice, which display early blastula lethality (40). One possible reason zebrafish are able to progress through early development in the absence of brg1 is because of the availability of maternally supplied transcripts coupled with rapid development. This explanation has also been put forward for brahma mutants in Drosophila (41). Alternatively, zebrafish may express another brahma-related gene that can compensate for early events. Regardless of why zebrafish brg1 mutants progress past gastrulation, the yng mutation provides a unique opportunity to analyze, in vivo, the role of chromatin remodeling during retinal cell differentiation and other organogenesis events.

Two major conclusions can be made from our studies. First, the Brahma chromatin remodeling complex is required for MAP kinase activation within the zebrafish retina and this signaling cascade is essential for cellular differentiation. On the other hand, cell type specification is not affected by mutations in the Brahma chromatin remodeling complex. Second, the Brahma chromatin remodeling complex not only shows tissue specificity, but also regulates specific genes within the cells in which they are expressed. Ath5 and other genes are regulated normally in yng/brg1 mutant retinas, but MAP kinase activity and late differentiation genes are not expressed in these same cells.