Rybp, a polycomb complex-associated protein, is required for mouse eye development

Background Rybp (Ring1 and YY1 binding protein) is a zinc finger protein which interacts with the members of the mammalian polycomb complexes. Previously we have shown that Rybp is critical for early embryogenesis and that haploinsufficiency of Rybp in a subset of embryos causes failure of neural tube closure. Here we investigated the requirement for Rybp in ocular development using four in vivo mouse models which resulted in either the ablation or overexpression of Rybp. Results Our results demonstrate that loss of a single Rybp allele in conventional knockout mice often resulted in retinal coloboma, an incomplete closure of the optic fissure, characterized by perturbed localization of Pax6 but not of Pax2. In addition, about one half of Rybp-/- <-> Rybp+/+ chimeric embryos also developed retinal colobomas and malformed lenses. Tissue-specific transgenic overexpression of Rybp in the lens resulted in abnormal fiber cell differentiation and severe lens opacification with increased levels of AP-2α and Sox2, and reduced levels of βA4-crystallin gene expression. Ubiquitous transgenic overexpression of Rybp in the entire eye caused abnormal retinal folds, corneal neovascularization, and lens opacification. Additional changes included defects in anterior eye development. Conclusion These studies establish Rybp as a novel gene that has been associated with coloboma. Other genes linked to coloboma encode various classes of transcription factors such as BCOR, CBP, Chx10, Pax2, Pax6, Six3, Ski, Vax1 and Vax2. We propose that the multiple functions for Rybp in regulating mouse retinal and lens development are mediated by genetic, epigenetic and physical interactions between these genes and proteins.


Background
The vertebrate eye is a complex neurosensory organ. Normal function of the eye requires precise spatial organization and interaction between individual tissues to respect the laws of optics. During embryonic development, reciprocal inducing events result in the formation of the lens and the retina that originate from progenitor cells located in the head surface ectoderm and neuroepithelium of the ventral diencephalon, respectively (see reviews: [1,2]). Abnormal lens and retinal development can cause isolated or widespread ocular abnormalities that can obstruct vision at different levels and lead to blindness (reviewed in [3]).
Between embryonic day E9.5 and E11.5 of mouse development, the optic vesicle undergoes dorso-ventral patterning of the neuroepithelium followed by its invagination to form the bilayered optic cup [4]. This process gives rise to the optic fissure, allowing blood vessels originating from the vascular mesoderm to enter the developing eye. By E13.5, the nasal and temporal retina on either side of the choroid fissures fuses around the optic nerve axons and blood vessels. Failure of this closure results in a specific developmental abnormality, called ocular coloboma. Ocular coloboma is often seen in association with severe neurological and/or craniofacial abnormalities [5] or can develop as an isolated condition [6].
Lens differentiation at E12.5 is marked by cellular elongation of the lens cells forming the posterior part of the lens vesicle [7]. Differentiating lens fiber cells upregulate expression of various classes of structural proteins including crystallins, intermediate filament bead proteins, and cytoskeletal, membrane, and channel proteins. Abnormal lens fiber cell differentiation disrupts lens homeostasis leading to precipitation of lens proteins resulting in lens opacification and perturbed vision (see review [8]).
Rybp (RING1 and YY1 binding protein) is an evolutionarily conserved protein with a zinc-finger motive that was identified first as an interacting partner for the Polycomb group protein Ring1A [23]. Polycomb group (PcG) proteins function as transcriptional repressors acting part through histone modification, and are believed to be important regulators of organogenesis and cell lineage specification [24,25]. Recent studies in Drosophila have shown that Drosophila RYBP depends upon PcG proteins to repress transcription, suggesting that Rybp could be classified as a PcG protein itself [26]. Other studies have demonstrated Rybp's interaction with DNA binding transcription factors [27][28][29] as well as with apoptotic [30] and ubiquitinated proteins [31]. Although the precise molecular function of Rybp is not yet known, these interactions suggest that Rybp may be a multifunctional developmental regulator. Indeed, we have shown recently that the Rybp is required both for early mouse development and for proper brain formation. Specifically a dose-dependent role in the central nervous system (CNS) for Rybp was uncovered: haploinsufficiency in the subset of embryos caused an exencephalic phenotype due to imperfect closure of the neural tube [32].
Given established parallels between brain and eye development, we hypothesized that Rybp may also play a role in ocular development. Here we determined Rybp protein localization patterns in the murine eye and analyzed the function of Rybp in four mouse models representing reduced or increased Rybp gene dosage. These studies have shown that aberration in the normal protein levels of Rybp can result in retinal coloboma, abnormal lens and anterior eye development, and corneal neovascularization.

Rybp is expressed in multiple tissues of the mouse embryonic eye
In previous work, we reported the localization of Rybp in the developing CNS [32]. The common embryonic origin of the brain, lens, and retina from the primitive ectoderm prompted us to determine the protein localization of Rybp during mouse ocular development ( Fig. 1). At E10.5, Rybp staining first strongly marked the head surface ectoderm surrounding the invaginating lens placode (Fig. 1A). Weak, speckled expression was also seen beneath the surface ectoderm, in the periocular mesenchyme, and throughout the emerging optic cup (Fig. 1A). From E11.5, Rybp appeared both in the anterior cells of the lens vesicle and in the nuclei of the elongating primary lens fiber cells (Fig. 1B). Similar to the lens, speckled expression was also evident in the optic cup, with a higher density of positive cells towards the ventral and marginal portion (Fig. 1B). Expression of Rybp in the hyaloid plexus was also detected (Fig. 1B). With progressive development, expression of Rybp persisted in the cells of the hyaloid cavity (Fig. 1C, D) and the head surface ectoderm. Between E14.5 and E16.5, Rybp localized in the differentiating secondary fiber cells of lens (E16.5; Fig. 1C and 1E) and in the ventral part of the neuroretina (E16.5; Fig. 1C, F). At E16.5 Rybp was also detected in the lens epithelium ( Fig. 1C, G). In the cornea, Rybp was expressed in the corneal epithelium and some cells of the corneal stroma (Fig.  1G). Rybp displayed non-uniform weak staining in the cells of the optic nerve and stronger staining in the peri-optic mesenchyme (Fig. 1H). One marked change in the expression pattern of Rybp during mouse eye development is that its widespread expression in undifferentiated cells of the retina becomes restricted to a layer-specific Rybp localization during prenatal mouse ocular development expression profile marking the gradually emerging ganglion (GCL) and inner nuclear (INL) cell layers ( Fig. 1I-L). Rybp's expression is robust in the ganglion cell layer of the differentiating retina, but strong staining is also visible in the INL and in a few cells of the future photoreceptor layer of the neuroretina (Fig. 1L).
Next, we analyzed the localization of Rybp in two day (P2, Figs. 2A,C,E,G), and twenty one day old (P21, Figs. 2B,D) mouse eyes. In the postnatal P2 retina, Rybp showed intense staining in the GCL and in the differentiating INL of the retina ( Fig. 2A). At P21, Rybp still was expressed in the GCL and in the INL, specifically in its dorsal part likely coincident with the bipolar and horizontal cell layers (Fig.  2B). Although Rybp was detected in the early stages of primary lens fiber cell development (Fig. 1B), in the more mature lens its expression was attenuated both in fully elongated primary lens fiber cells and in the lens epithelium (Figs. 2C and 2D). However, strong expression of Rybp in the lens was seen in the transitional zone where the secondary lens fiber cells are formed (Figs. 2C and 2D). At P2.0, Rybp was expressed in the majority of lens epithelial cells, and sporadic expression also was observed in the corneal epithelium, stroma, and basal membrane (Fig. 2E). By P21.0, Rybp is no longer expressed in the lens epithelium but still is weakly expressed in the corneal epithelium (Fig. 2F). Finally, Rybp was strongly expressed postnatally in the connectiva (Fig. 2G), and uniform localization of Rybp was observed around the newborn optic nerve (Fig. 2H). This dynamic and cell-type restricted expression pattern of Rybp (Figs. 1 and 2) raises the possibility that Rybp may have specific functional roles in the generation/maintenance of particular cell types during mammalian eye development.

A subset of the Rybp+/-embryos exhibits retinal coloboma
Previously we reported that a subset of Rybp heterozygous null embryos exhibited perturbed brain development including forebrain overgrowth and exencephaly [32]. As the retina forms from the forebrain, we next examined the possibility of aberrant eye development in these animals. We found that 32% (6/19) of Rybp+/-exencephalic mice examined at several stages of development (from E12.5 to postnatal stages) showed retinal/optic nerve coloboma (compare Fig. 3A and 3B). Colobomas of this type often are caused by an incomplete closure of the optic fissure Rybp localization in the postnatal mouse eye that occurs normally at E13.5 [33]. Rybp+/-colobomas were observed both bilaterally and unilaterally. In addition, Rybp+/-eyes had thickened neuroretinas, their lenses were ventrally rotated and misplaced within the eyeball (see Fig. 3B in comparison to Fig. 3A). The optic nerve was also frequently regressed (Fig. 3F). One possible explanation for the development of colobomas is that the decreased level of Rybp in the mutant retinas influences the normal distribution of regulatory proteins such as Pax6 [34][35][36] or Pax2 [33,37,38]. Both Pax6 and Pax2 have been shown to be essential for proper closure of the optic fissure. Accordingly, we tested whether immunolocalization of Pax6 and Pax2 was affected in the mutant Rybp eyes. Normally, Pax6 protein is localized in the ventral side of the retina but disappears from the developing optic nerve after E12.0 in wild-type animals [39]. In the Rybp mutant embryos, Pax6 expression spreads across the entire thickness of the retina, expanding to its margin (Figs. 3C, D). In contrast, the localization of Pax2 in mutant eyes were unchanged ( Fig. 3E compare to 3F).
The proper ratio between neural progenitor and postmitotic neuronal cell types is important for normal retinal development and disturbed morphogenesis of this process can lead to colobomas. [40]. Accordingly, we investigated whether the ratio between early-and late-born neurons changed in the mutant retinas exhibiting the coloboma phenotype. In the prenatal mutant retina, expression of specific neuronal cell fate markers was similar to the expression of these markers in the retina of control mice. These included Tuj1 (marks early neural cell types; Fig. 4A,B), NeuN (postmitotic, marks late neuronal cell types; Fig. 4C, D) and nestin (marks neural progenitor cells; Fig. 4E, F). The apparent normal distribution of these neuronal markers shows that the neuronal cell differentiation is not affected in the Rybp mutant retinas and probably is not the direct cause of the failure of optic fissure closure.

Rybp-/-<-> Rybp+/+ chimeric embryos show a series of eye defects
Early lethality of Rybp -/-embryos [32] prevented our analysis of the effect of a complete loss of Rybp during eye development. However, chimeric mice (n = 90) have been generated from Rybp homozygous null -/-and wild type (+/+) ES cells, and 20% of them showed low overall contribution of Rybp -/-ES cells and displayed forebrain abnormalities [32]. Half of the chimeric embryos (examined between E9.5 and E14.5) with brain abnormalities also showed eye defects similar to those described above for the Rybp+/-embryos. These included retinal colobomas (Figs. 5A and 5B), and defects in lens formation (compare Figs. 5D to 5C and 5F to 5E). In addition, in an E13.5 chimera, the separation between the lens epithelium and the surface ectoderm was compromised (com-

Overexpression of Rybp results in lens, retinal and corneal defects
Next we employed a conditional ectopic overexpression strategy [41] to assess the effects of Rybp overexpression in the lens. bind to Ring1A as described earlier [23], the excised cells (ROSA26-RYBP/EGFP; Cre cell line) were transiently transfected with Flag-tagged Ring1A, and lysates were immunoprecipitated with either RYBP or GFP antibodies and blotted with a Flag antibody. As expected, the RYBP/ EGFP fusion protein was found together with Ring1A in vivo (Fig. 6C).
The ROSA26-RYBP/EGFP mice were crossed with two different reporter mouse lines and the proper expression of Lens-specific and ubiquitous expression of RYBP/EGFP fusion protein the fusion protein was confirmed by fluorescent microscopy: lens specific expression was seen for the ROSA26-RYBP/EGFP;αA-crystallin/Cre double transgenics (Fig. 6D) and ubiquitous expression for the ROSA26-RYBP/EGFP;β-Actin/Cre double transgenics (Fig 6F). Aberrant lens morphology of the ROSA26-RYBP/EGFP; αA-crystallin/Cre mice is shown in Fig. 7. Although the P2 embryonic lenses showed only subtle abnormalities in fiber cell morphology (7A-B), older mice developed severe opacities of the lens resulting in a collapse of lens fiber mass (Figs. 7C-D).
Eyes from ROSA26-RYBP/EGFP; β-Actin/Cre mice were examined at embryonic (E16.5, E18.5), postnatal (P1-P4, P7, P14, P21) and adult stages (2, 3, 6 month) (Fig. 8, and data not shown). This ubiquitous overexpression of RYBP/EGFP led to the abnormal formation of a number of ocular tissues. The most frequent phenotype seen (in 35% of the mice) was neovascularization of the corneas (compare Fig. 8C-D to 8A-B). These structural changes were visible at postnatal stages (P7-21) during which the stromal layer thickened. When hemizygous mice were mated to obtain mice homozygous for the RYBP/EGFP transgene, the penetrance of the phenotype increased to 80%. Small vessels were apparent in the stroma of the cornea under bright field microscopy, and by 2-4 months postnatal, gross corneal neovascularization was visible (data not shown). Electron microscopy studies showed that the capillaries are already present at birth in the transgenic corneas (Fig. 8E, F). This was further supported by immunohistochemistry using an anti-CD34 antibody which marked the newly formed vessels of the mutant corneas (data not shown). Aside from neovascularization, other observed ocular phenotypes included irregular folding of the retina, retinal coloboma, defects in anterior eye development (absence of the vitreous body, absence of the anterior chamber), and lens opacification during later adulthood (Fig. 8G,H and data not shown). In summary we have developed transgenic mouse models in which expression of Rybp in the lens disrupted normal fiber cell differentiation and ubiquitous expression of Rybp resulted in corneal neovascularization and defects in the anterior eye development. These models suggest that normal eye development is sensitive for the proper dose of Rybp and that improper dosage of Rybp causes multiple eye abnormalities.

Genes with altered expression level in the Rybp transgenic lenses
As a first attempt to elucidate the molecular basis of abnormal lens fiber cell differentiation of the ROSA26-RYBP/EGFP; αA-crystallin/Cre mice (Figs 7, 8), we performed expression analysis of the genes involved in this process including major lens structural proteins, selected cell adhesion molecules, and major transcription factors implicated in lens development (Fig. 9). Total RNA from P1 transgenic and control lenses was used to synthesize cDNA, and quantitative RT-PCR was performed. The expression level of both Rybp and EGFP mRNA was increased 16-times compared to the non-transgenic control as expected. From the major lens structural proteins, the Cryba4 transcript showed a seven-fold reduction in transgenic lenses, while the expression of other crystallins and filensin was unchanged. From the tested cell adhesion molecules, α6-integrin mRNA level was five-times reduced while other integrins (α5-and β1-integrins) remained unchanged. Among tested transcription factors, AP-2α mRNA level was 30-times and Sox2 was 12-times elevated. The level of transcription factors Pax6, Prox1, MafA, MafB and c-Maf did not change significantly in the ROSA26-RYBP/EGFP; αA-crystallin/Cre lenses. These results further confirm that normal lens development depend on the Rybp gene dosage. RNA microarray studies may need to be conducted in the future to provide further insight into which membrane proteins, gap junctions and β/γ-crystallins are being affected.

Discussion
Rybp encodes an essential regulatory protein involved in early post-implantation development of the mouse embryo. In addition, Rybp plays important roles in orga- nogenesis as evidenced by disrupted development of the forebrain found both in Rybp heterozygous null and chimeric embryos [32]. Here we characterized the protein localization and in vivo function of Rybp in another organ -the developing mouse eye. We show that Rybp is required for normal retinal and lens development, and may also be involved in the formation of the anterior eye segment.

Abnormal lens development in the lens-specific Rybp trans-genic mice
Our previous work revealed that during CNS development, Rybp is predominantly localized in postmitotic neurons and differentiated cell types of the developing mouse embryo [32]. In the present study, we show that in the developing mouse eye, Rybp also becomes robustly made in the differentiating layers of the neuroretina including the ganglion and inner nuclear layer cells. This suggests once again a possible role for Rybp in cell cycle exit or the commitment to differentiation. Gaining further insight into the precise cellular role of Rybp during mouse ocular development likely requires a conditional mutant version of the gene where Rybp can be eliminated in specific cell lineages or at specific time points of embryonic development.
The major phenotypic malformation observed in all Rybp mouse models (heterozygous, chimeric and transgenic, see Table 1) was the failure of the closure of the optic fissure, retinal coloboma. Morphological analysis of embryonic eyes showed that 32% of Rybp deficient (heterozygous) and 50% of chimeric mice have eye coloboma in conjunction with brain defects, but coloboma was also seen in the Rybp transgenic mice that did not display obvious brain defects. A number of mammalian genes encoding transcription factors (BCOR, CBP, Chx10, Cited2, c-Maf, Foxg1, Pax2, Pax6, Ptch, Six3, Ski, Vax1 and Vax2), signaling molecules (Jnk1, Jnk2 and Shh), or members of the retinoic acid pathway have been associated with coloboma [5]. Our study describes Rybp as a novel gene associated with coloboma. While the molecular and cellular mechanisms underlying this condition are still poorly understood, they likely involve perturbations in cell adhesion, cell shape, cell proliferation, cell death, and/or the extracellular matrix.
Aside from the retinal colobomas, a wide range of lens abnormalities were found in the Rybp mouse models. In both Rybp heterozygous and chimeric eyes, ventral rota- tion of the lens was found in association with coloboma (see Fig. 3 and 5). A similar phenotype has been reported for Otx mutant mice [42]. Lenses of Rybp chimeric embryos showed abnormal separation of the lens vesicle from the surface ectoderm (Fig. 5), a condition found in Pax6 heterozygous [43] and Foxe3 homozygous [44] mouse embryos. Transgenic lenses overexpressing Rybp developed abnormally differentiated lenses at birth (if not earlier), and this condition further deteriorated with age resulting in a total collapse of the lens fiber mass (see Fig.  7). This demonstrates that high level of Rybp disrupts normal fiber cell differentiation and maturation. Similar phenotype was observed when AP-2α transcription factor, individual E2Fs or growth factors were overexpressed in lens specific manner ( [19,45], see review [46]). Whether excess of Rybp interferes with adhesion and migration (like AP-2α), cell cycle (like E2Fs), terminal differentia-tion (like TGFβ) of secondary fiber cells or acts via completely different mechanisms, requires further studies. Indeed, RYBP was shown to be involved in specification of function within the family of E2F transcription factors in cell culture systems [28] however this has not been investigated in vivo during lens fiber cell differentiation.

Corneal neovascularization and irregular retinal development in ubiquitous Rybp transgenic mice
Ubiquitous overexpression of Rybp caused similar lens defects and additionally led to corneal neovascularization (see Fig. 8). Genes such as VEGF, FGF2 and MMP-2 have already been associated with corneal neovascularization [47][48][49][50]. A possible connection between Rybp and angiogenesis warrants further investigation, especially in light of the observation that during embryonic development Rybp is expressed in hyaloid vessels and in the endothelium of blood vessels outside of the eye ( Fig. 1; Pirity and Schreiber-Agus, unpublished data).
The variable penetrance of the phenotypes may indicate that expression and/or function of the retained wild type Rybp allele is being modified differentially in affected or non-affected animals. This is also supported by our previous observation that the penetrance of the exencephaly, a previously described heterozygous semipenetrant phenotype, was influenced by genetic background [32]. Indeed the phenomenon "semi-penetrance" is a common observation in genetically-engineered mice.
One potential molecular mechanism for abnormal eye development in Rybp mice could relate to Rybp's role in Polycomb group protein transcriptional regulation [23]. Polycomb proteins function in multiprotein complexes to regulate expression of homeotic and other regulatory genes during embryonic development [51][52][53][54][55][56][57] and also in eye development [58,59]. Notably, in our limited gene expression studies using a candidate approach (Figure 9), several key eye regulators were found to be significantly affected in the Rybp transgenic lenses examined, including Sox2 and AP-2α (see also [60,18,19,61]) as well as were some of the crystallins [62]. Moreover, Pax6 localization in retina and lens appears affected in the Rybp heterozygous null mice (Figure 3). Taken together, our findings

Conclusion
Collectively, the present work provides the first in vivo evidence for a role of the Polycomb group binding protein Rybp in mouse eye development and disease. Further studies need to address the molecular basis of this role and to determine how Rybp functionally relates to other known regulators of ocular processes.
Mice were housed in a 12-h light, 12-h dark cycle and maintained in the animal facility at AECOM in accordance with institutional guidelines. Noon of the day the vaginal plug was observed was considered E0.5 of embryogenesis. For genotyping targeted ES cell colonies/mice, EGFP Primers: A, (5'-aagttcatctgcaccaccg-3') and B, (5'tgctcaggtagtggttgtcg-3') were used as described. The genotypes were determined by PCR analysis on DNA extracted from the tail or yolk sac.

Rybp conditional transgenics
The strategy used to generate a Cre recombinase-mediated conditional ectopic Rybp allele by targeting the ubiquitous ROSA26 locus is shown on Figure 6. The ROSA26 locus (referred to as an R26 knock-in) was targeted with a floxed neo cassette followed by an RYBP/EGFP fusion. To generate the RYBP/EGFP fusion, the Rybp open reading frame (ORF) was amplified with primers A (5'-gcacgtcgaccagcccgtccatgaccatgg-3') and B (5'-ctctggatccgaaagattcatcattcactgc 3') and cloned into pEGFP-N3 with NheI/SalI. The Rybp/EGFP fusion was transferred to pBigT [41] with NheI/Not1, and then the Rybp/EGFP together with the floxed neomycin cassette was cloned into the pROSA26 vector [41] with AscI/PacI. Rybp transgenic ES cell lines were generated by electroporating linearized targeting vector into R1 ES cells [63] as described in 96 clones resistant to G418 (Gibco, 300 μg/ml) were selected, and screened by genomic Southern blot hybridization on DNA digested with EcoRV as previously described [41]. Two targeted clones were injected into C57BL/6 blastocysts and produced germ-line chimeras (ROSA26-RYBP/EGFP mice) [67]; mice carrying the targeted allele were genotyped by the PCR as described [41]. Male chimeras were mated with ICR females, and their agouti offspring were tested for transmission by tail PCR and blotting. Animals heterozygous for the mutation were bred with corresponding Cre transgenic lines and analyzed for the phenotype. Nontransgenic littermates were used as controls for all experiments. Heterozygous transgenic progeny were mated to maintain the allele. All analyses were performed on a mixed (129 × ICR) background and mutant mice were analyzed in comparison to their wild-type littermates.

Histology and immunohistochemistry
All embryos were dissected in PBS, fixed overnight in PBS buffered 4% paraformaldehyde, and paraffin-embedded sections (

Electron microscopy
Dissected P1 and P21 day old eyes were placed in Ito's fixative [68] for 24 h after the cornea had been pierced with a fine needle. The eyes were washed overnight in cacodylate buffer, postfixed with OsO4, dehydrated, and embedded in Epon (Roth, Karlsruhe, Germany). Semithin sections were stained with toluidine blue. Ultrathin sections were stained with uranyl acetate and lead citrate and viewed with a Zeiss (Oberkochen, Germany) EM 902 electron microscope.