High‐throughput custom capture sequencing identifies novel mutations in coloboma‐associated genes: Mutation in DNA‐binding domain of retinoic acid receptor beta affects nuclear localization causing ocular coloboma

Abstract Uveal coloboma is a potentially blinding congenital ocular malformation caused by the failure of optic fissure closure during the fifth week of human gestation. We performed custom capture high‐throughput screening of 38 known coloboma‐associated genes in 66 families. Suspected causative novel variants were identified in TFAP2A and CHD7, as well as two previously reported variants of uncertain significance in RARB and BMP7. The variant in RARB, unlike previously reported disease mutations in the ligand‐binding domain, was a missense change in the highly conserved DNA‐binding domain predicted to affect the protein's DNA‐binding ability. In vitro studies revealed lower steady‐state protein levels, reduced transcriptional activity, and incomplete nuclear localization of the mutant RARB protein compared with wild‐type. Zebrafish studies showed that human RARB messenger RNA partially reduced the ocular phenotype caused by morpholino knockdown of rarga gene, a zebrafish homolog of human RARB. Our study indicates that sequence alterations in known coloboma genes account for a small percentage of coloboma cases and that mutations in the RARB DNA‐binding domain could result in human disease.

Uveal coloboma is believed to have a significant genetic component; however, the genes and mutations behind the majority of coloboma cases currently remain unknown (Gregory-Evans, Vieira et al., 2004;Morrison et al., 2002;L. Wang et al., 2012). The presence of multiple patterns of Mendelian inheritance, incomplete penetrance, variable expressivity, genetic heterogeneity, and environmental factors make the genetic investigation of this disease particularly challenging (Chang et al., 2006;Skalicky et al., 2013;. Although several genes identified in animal studies are critical for optic fissure closure during eye development, very few are known to cause coloboma in humans (Brown et al., 2009;. On the other hand, reports of human mutations are often not complemented by experimental mechanistic evidence either in animal models or in cell culture (Bourchany et al., 2015;Chassaing et al., 2016;Graham et al., 2003;Kahrizi et al., 2011;Martinez-Garay et al., 2007;Ng et al., 2012;Wenger et al., 2014). In many cases, studies on coloboma genetics have focused on a select number of genes, and many have included syndromic conditions with severe ocular phenotypes (e.g., severe microphthalmia and anophthalmia; Gonzalez-Rodriguez et al., 2010;Guo, Dai, Huang, Liao, & Bai, 2013;Mihelec et al., 2009;Morrison et al., 2002;Schimmenti et al., 2003;X. Zhang et al., 2009;J. Zhou et al., 2008).
Transcription factors such as RAX, SOX2, PAX6, PAX2, MITF, and OTX2 are expressed very early during development and are involved in eye field specification and emergence of the optic vesicle, cell lineage specification, and development of the neural retina and RPE (Avilion et al., 2003;Chow & Lang, 2001;Levine & Brivanlou, 2007;Zuber, Gestri, Viczian, Barsacchi, & Harris, 2003). Secreted signaling molecules regulate the spatio-temporal expression of multiple transcription factors and play a critical role in the morphogenesis of the optic cup and the establishment of RPE and retinal domains (Behesti, Holt, & Sowden, 2006;Kobayashi, Yasuda, & Araki, 2010;Zhao, Saitsu, Sun, Shiota, & Ishibashi, 2010). Members of TGFB/BMP, WNT, and Sonic hedgehog (SHH) signaling are involved in dorsal-ventral patterning of the retina and RPE by regulating the expression of T-box (Tbx5, Tbx3, andTbx2), Vax1/2, Msx1, Pax6, and Pax2 genes (Behesti et al., 2006;Kobayashi et al., 2010;Zhao et al., 2010). WNT, TGFB/BMP, and Hippo signaling pathways have been shown to regulate RPE specification, development, and choroid fissure closure (Fossdal et al., 2004;Fuhrmann, Levine, & Reh, 2000;Miesfeld et al., 2015;Muller, Rohrer, & Vogel-Hopker, 2007;Steinfeld et al., 2013;. Fibroblast growth factor signaling has been shown to induce and maintain the neural identity of the retina (Hyer, Mima, & Mikawa, 1998;Pittack, Grunwald, & Reh, 1997). Genes associated with SHH signaling have been shown to be expressed in the neural retina and other ocular tissues during development (Amato, Boy, & Perron, 2004;Wallace, 2008). Mutations in humans and studies with specific misexpression and loss-of-function of SHH genes have been associated with coloboma and other severe ocular defects (Schimmenti et al., 2003; X. M. Zhang & Yang, 2001). TGFB/BMP signaling genes have been shown to be expressed in all ocular tissues including the periocular mesenchyme and reports in humans (e.g., BMP4) and animal studies suggest a role in ocular, brain, craniofacial, and skeletal abnormalities (Bakrania et al., 2008;Furuta & Hogan, 1998). Mutations in the WNT receptor, FZD5 have been associated with optic fissure closure defects and were shown to disrupt the apical junctions of the retinal neural epithelium leading to coloboma (C. Liu et al., 2016). The complexity of the interactions between the signaling pathways and the transcription factors involved in patterning and development of the eye, however, remains only in part understood.
Among the signaling pathways, RA signaling has been well characterized for its role in ocular development, and mutations in several genes associated with RA signaling have been reported to cause coloboma in human and animal models (Jakubiuk-Tomaszuk et al., 2019;Lupo et al., 2011;Matt et al., 2005;Matt, Ghyselinck, Pellerin, & Dupe, 2008;Srour et al., 2013Srour et al., , 2016. RA temporal and spatial expression in the developing eye is tightly regulated through the balance of synthesis, controlled by retinol and retinaldehyde dehydrogenases, and degradation by cytochrome P450 (Rhinn & Dolle, 2012). In the cytoplasm, RA binds to cellular RA-binding proteins (CRABP1/2) and upon entering the nucleus, RA binds to heterodimer receptor complexes comprising nuclear RA receptors (RARA/B/G) and retinoid X receptors (RXRA/B/G) to regulate the expression of target genes (Cvekl & Wang, 2009;Kastner et al., 1997;Kedishvili, 2013).
During embryonic development, RA signaling has been shown to have a predominant role in eye morphogenesis and optic fissure KALASKAR ET AL. closure (Gestri, Bazin-Lopez, Scholes, & Wilson, 2018;Matt et al., 2008). Several genes associated with RA signaling, such as retinoid receptors, RARs (A, B, and G) and RXRs (A, B, and G), STRA6, and RA-synthesizing enzymes, RALDH1 and RALDH3, are expressed in the developing eye and the surrounding periocular mesenchyme (Bouillet et al., 1997;Ghyselinck et al., 1997;Matt et al., 2005;Mori, Ghyselinck, Chambon, & Mark, 2001). Retinoid nuclear receptors are the key transducers of RA signaling and have been shown to heterodimerize to regulate their target genes (Kastner et al., 1997).
The role of retinoid receptors has been investigated in single, double, and multigene knockout animal models Grondona et al., 1996;Kastner et al., 1997;Lohnes et al., 1994;Mendelsohn et al., 1994). Rarb knockout mice look overall morphologically normal; however, they show ocular defects including retrolenticular membrane with persistent hyaloid vasculature, congenital retinal folds, cataracts, reduced eye size and weight, and decreased retinal ganglion cell number G. Zhou, Strom, Giguere, & Williams, 2001). In contrast, compound mutants of Rarb gene with other receptor genes, Rara or Rarg, exhibit developmental defects in several organ systems leading to embryonic lethality during late gestation. Rarb and Rara or Rarg compound mutant mouse embryos display severe ocular defects including retinal dysplasia and degeneration, shortening of the ventral retina, absence of the anterior chamber, and ventral rotation of the lens Grondona et al., 1996;Lohnes et al., 1994;Mendelsohn et al., 1994). These studies underscore the importance of RARB and its heterodimerization partners in correctly transducing the RA signaling during ocular morphogenesis.
In recent reports, patients carrying RARB mutations are described with a severe syndromic phenotype affecting several organ systems and including coloboma, microphthalmia, anophthalmia, cardiac defects, progressive motor impairment, pulmonary hypoplasia, and diaphragmatic hernia. The severity of these phenotypes is reported to be a consequence of either a total loss-of-function or two-to threefold gain-of-function of the RARB gene (Srour et al., 2013(Srour et al., , 2016. The molecular mechanisms underlying the multitude of phenotypes observed in individuals with RARB mutations are not clear.
In the present study, we report on the screening of a large cohort of syndromic and nonsyndromic uveal coloboma patients using custom capture sequencing and on the identification of novel mutations in genes previously associated with coloboma. We further explore the mechanisms underlying the mutation in RARB DNAbinding domain that resulted in structural and functional changes within the RARB protein ultimately affecting RA signaling.

| Subjects
Two hundred twenty-eight study subjects (99 affected) from 66 families, including probands with primarily syndromic or nonsyndromic coloboma and their first-degree relatives, were examined by a single ophthalmologist (BPB) at the National Eye Institute (NEI) Ophthalmic

| Custom capture sequencing and mutation detection
Genomic DNA was extracted from whole blood or saliva. Two custom capture designs (CC-1 and CC-2) were used for targeted sequencing covering a total of 193 genes (Table S1). The first (CC-1) was created   (Teer, Green, Mullikin, & Biesecker, 2012). The first-pass filter included the following criteria for variant quality screening and filtering: ClinSeq reference allele is the major allele and present at >50% frequency in the population, ClinSeq variant (minor) allele frequency ≤0.02 is considered and genotypes called when >25 samples in the cohort covered for the location (Biesecker et al., 2009 Adzhubei et al., 2010;Kircher et al., 2014;Kumar, Henikoff, & Ng, 2009;Schwarz, Rodelsperger, Schuelke, & Seelow, 2010

| Cell cultures and plasmid DNA constructs
Human embryonic kidney 293 (HEK 293) cells were cultured in Dulbecco's modified Eagle's medium containing 1% each of L-glutamine, sodium pyruvate, and antibiotics (penicillin/streptomycin: 10,000 units/ml/10,000 µg/ml) and 10% fetal bovine serum (FBS) as previously described (George et al., 2016). For experiments involving RA treatment, cells were cultured in medium containing 10% FBS or charcoal-stripped serum for about 24 hr before RA treatment. pcDNA3.1+-DYK-tagged RARB plasmid DNA (DYK-tag sequence at C-terminus; Clone ID: OHu15136) was obtained from GenScript, site-directed mutagenesis was performed to generate the RARB-mutant plasmid DNA and verified by sequencing. RARB and RARB-mutant complementary DNA (cDNA) were cloned in-frame with a green fluorescent protein (GFP) into a pAcGFP1-C3 plasmid vector (GFP at C-terminus). RXRA, RXRG, and RARG cDNA were cloned in-frame with RFP into a pAcRFP1-C3 plasmid vector (RFP at C-terminus) and used in cotransfection studies. Cells were transfected at 60-70% confluency with plasmid DNA (9 nM) using Extreme HP transfection reagent (Cat # 6366236001; Sigma-Aldrich) followed by RA treatment (10 µM) after 12 hr. Cells were fixed or harvested 12 hr after RA treatment for further analysis.

| Immunofluorescence and imaging
Immunofluorescence staining was performed following a standard protocol as previously described (George et al., 2016). HEK 293 cells on slides (Nunc Lab-Tek II Chamber slide system, Cat # 154526; MG Scientific) were washed with 1X phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde for 10 min followed by three 5-min washes with 1X PBS and stored at −30°C until used. Slides were blocked in 1X PBS/10% donkey serum/1% bovine serum albumin (BSA)/0.1% Tween20 for 1 hr followed by overnight incubation at 4°C with primary antibody diluted in blocking solution. Slides were washed five to six times after primary antibody incubation with wash buffer containing 1X PBS/1% BSA/0.1% Tween20. Slides were then incubated in secondary antibody with Hoechst (1:2500, Cat # H3570; Invitrogen) for 1 hr at room temperature followed by washings and mounting in Fluormount-G (Cat # 00100-01; Southern Biotech). The following antibodies and dilutions were used: rabbit anti-DYK Flag (1:100, Cat # 14793; Cell Signaling), rabbit polyclonal anti-RARB (1 µg/ml, Cat # ab53161; Abcam), donkey anti-rabbit Alexa 555 (1:1,000). Images were taken on Zeiss LSM 700 and LSM 800 confocal microscopes. Z-stack images were taken with ×63/1.4 Oil Plan-Apochromat objective with a scan area of X: 0.5, Y: 0.5 and the following parameters: Z-stack range 6-14 µm; image scaling (X) 0.198 × (Y)0.198 × (Z)0.30 µm; bit depth −16 bit; image size 1,024 × 1,024 pixels; bidirectional scanning with averaging at 2. At least three to five different single plane fields for each specific staining were used for statistical analysis and single-plane images were used to generate figure panels. Statistical significance for the cell counts was determined by one-way analysis of variance test followed by Tukey's multiple comparison tests. Li-COR Bio). Images were captured on a Bio-Rad ChemiDoc imaging system (Cat # 17001402). Western blots were performed on at least six biological replicates of whole-cell protein extracts and at least three biological replicates of nuclear-cytoplasmic protein fractions.

| Western blot analysis
At least two technical repeats were performed for each experiment.
Densitometric analysis was performed using the Bio-Rad's Image Lab 6.0.1 software.

| Protein degradation assay
Protein degradation assay was performed as previously described (Alur et al., 2010). HEK 293 cells in culture were treated with 100 µg/ml of cycloheximide (dissolved in 100% ethanol) after 24 hr of transfection.
Cells were harvested before and at 3, 6, and 12 hr after treatment. For control, transfected cells were treated with an equal volume of 100% ethanol (vehicle). At least three replicate protein samples were collected for each of the time points and western blot analysis repeated two times.

| Luciferase assays
Renilla luciferase reporter plasmid DNA (pRL-TK) and CYP26A1 promoter region cloned firefly luciferase reporter plasmid DNA (pGL4.10) were cotransfected with pcDNA3.1+-DYK-tagged RARB or RARB-mutant plasmids into HEK 293 cells in culture. Genomatix genome analyzer software was used to identify the binding sites for RARB protein in the CYP26A1 promoter region. The cultured cells were lysed in passive lysis buffer and samples prepared as per manufacturer protocol using the Dual-Luciferase Reporter assay system (Cat # E1960; Promega). Luminescence was measured using a Modulus microplate multimode reader (Model # 9300-010; Turner Biosystems). The experiments were performed at least six times with three replicates for each sample luminescence reading. Statistical significance was determined by paired t test. Error bars represent standard error of the mean (SEM). 2.9 | Quantitative real-time polymerase chain reaction RNA was prepared from zebrafish embryos or HEK 293 cells using NucleoSpin RNA isolation kit (Cat # 740955.50; Macherey-Nagel) and cDNA synthesized using SuperScript First-strand synthesis system for RT-PCR (Cat # 11904018; Thermo Fisher Scientific).  (Table S1). These genes have been reported to cause human coloboma, microphthalmia, and/or anophthalmia (Brown et al., 2009;Chow & Lang, 2001;Cvekl & Tamm, 2004;Fuhrmann, 2010;Lang, 2004;Martinez-Morales, Rodrigo, & Bovolenta, 2004;Zuber et al., 2003). The first-pass filter    Table 1). The proband was a 38-year-old male with bilateral coloboma inherited from the mother, indicating complete penetrance   deleterious, nine heterozygotes are present in the gnomAD database, likely too frequent to cause fully penetrant uveal coloboma. As such, we classified this as a variant of unknown significance and did not pursue it further.

| Mutation in RARB DNA-binding domain results in decreased protein levels and affects nuclear localization and transcriptional activity
We focused specifically on the RARB mutation for further investigation as this mutation has been recently reported in a patient with syndromic microphthalmia (Nykamp et al., 2017). RARB protein comprises six distinct regions: two major domains (Figure 2a), a DNA-binding domain (DBD) and a ligand-binding domain (LBD) connected by a coiled-coil hinge region, along with two N-terminal low complexity domains and a C-terminal activation domain (Alvarez et al., 2007;Brand et al., 1988;Letunic et al., 2004;Schultz, Milpetz, Bork, & Ponting, 1998). The Arg144Gln mutation is in the DBD, which is a highly conserved region across different vertebrate species from lamprey-zebrafish to rhesus monkey-human (Figure 2a,b and Table S2). On the basis of structural modeling of the wild-type and mutant protein (Figure 2c,d), the substitution of Arg144 with Gln in the DBD of RARB caused a change in the orientation of a 3-carbon aliphatic straight chain of Arg137, the distal end of which is capped by a guanidium group. This change is predicted to alter the location of the α-helix interacting with the DNA in the mutant and it is expected to decrease the DNA-binding strength of the domain. Protein stability was predicted in silico using the FoldX software (Schymkowitz et al., 2005). The Arg144Gln mutation was predicted to result in improved protein stability by −1.19 kcal/mol compared with the native protein, which suggests a potential loss of native interaction within RARB protein. In addition, the interaction energy between the protein and DNA was predicted to increase by +0.024 kcal/mol, suggesting a complete loss of interaction and thus instability of the protein-DNA complex.
Because mutations in the DBD may also affect the nuclear localization of a protein (Bunn et al., 2001), we investigated cellular localization and transcriptional activity of RARB in vitro. We predicted that the mutant protein would respond to RA treatment like wild-type as the LBD appeared unaffected by the mutation. We transfected HEK 293 cells in culture with wild-type and mutant RARB expression constructs, followed by RA treatment. By western blot analysis, cells transfected with the wild-type RARB expression construct showed higher steady-state levels of exogenous RARB protein compared with cells transfected with RARB-mutant expression construct (Figures 3a   and 3a′). Immunofluorescence showed that most of the RARB-mutant protein was retained in the cytoplasm while the wild-type protein was mostly detected in the nucleus (Figures 3b-e and 3b′). We obtained comparable results when cells were cultured in charcoal-stripped serum-containing medium or in regular serum-containing medium and when transfected with either DYK-tagged or GFP-tagged expression constructs ( Figure S3). We confirmed the localization of mutant and wild-type RARB proteins by separating the nuclear and cytoplasmic fractions and showed that the mutant protein was equally distributed between the cytoplasm and the nucleus whereas the highest proportion of wild-type protein was present in the nuclear fraction in the presence of ligand (Figures 3f and 3f′). Translation blocking with cycloheximide treatment of transfected cells in culture confirmed the reduced expression of the mutant protein at T 0 and complete degradation over a period of 12 hr compared to only partial degradation of the wild-type protein over the same time period (Figures 3g and 3g′). Gene expression quantification with qRT-PCR revealed significantly less mutant RARB transcripts compared with the wild-type (Figure 3h). Taken together, these results suggest that the reduced steady-state levels of mutant RARB are likely due to a combination of reduction in both mRNA and protein stability.
We then tested the transcriptional activity of the wild-type and mutant RARB proteins using a dual luciferase assay and the ratio of Firefly/Renilla luciferase intensity. The promoter region of the human CYP26A1 gene that contains the binding sites for RARB protein was PCR amplified and cloned into the pGL4.10 (luc2) vector to study the response to RA treatment. Wild-type RARB transcriptional activity was higher than baseline and control empty vector and was greatly increased by RA treatment. On the contrary, transcriptional activity of the RARB-mutant protein was reduced at baseline and only slightly increased upon RA treatment (Figure 3i). Transcriptional activity of the RARB-mutant protein was not significantly different from that of the control empty vector.

| RARB-mutant protein does not alter the localization of RXR proteins
RA receptors have been shown to form heterodimers with RXRs to transduce the RA signal (Chandra et al., 2017;Glass, 1996;. We In summary, the Arg144Gln mutation in RARB identified in Family 1 is localized to the DNA-binding region of the protein and appears to partially interfere with the nuclear localization and transcriptional activity of RARB.
3.5 | Human RARB mRNA partially reduces the ocular phenotype caused by rarga morpholino knockdown in zebrafish embryos To understand the role of RARB in optic fissure closure, we performed functional studies in the developing zebrafish. The closest homolog to human RARB in zebrafish is rarga, with 79% identity on phylogenetic analysis ( Figure S5). rarga is expressed in the cranial mesoderm and periocular mesenchyme and represents one of the most abundantly expressed retinoic receptor genes during the early   Table 2). The expression levels of rarga mRNA were significantly reduced in the rarga SB morpholino-injected embryos compared with the uninjected and standard morpholinoinjected control embryos (Figure 5f). The ocular phenotype included microphthalmia, and a range of mild (e.g., Grade 2, Panel 5d) to severe (e.g., Grade 3, Panel 5e) forms of coloboma depending on the dose of morpholino. We then used the human RARB wild-type and mutant mRNA to rescue the ocular phenotype caused by rarga gene knockdown. Upon coinjection of rarga SB morpholino with RARB wildtype mRNA, 39% of embryos displayed coloboma compared with 65% of embryos injected with morpholino alone. Furthermore, the percent of more severe (Grade 3) coloboma dropped to 19% from 34%. Conversely, 56% of embryos coinjected with rarga SB morpholino and the RARB-mutant mRNA showed a coloboma phenotype, 33% of which were Grade 3 ( Figure 5g and Table S3).
The results were consistently replicated during multiple injections and the direction of change in the proportion of embryos with coloboma suggests an effect of RARB wild-type mRNA but not of RARB-mutant mRNA. Thus, knockdown of the zebrafish equivalent to RARB gene, rarga, results in a phenotype reminiscent of human coloboma.

| DISCUSSION
Custom capture or targeted gene sequencing has been a cost-effective approach to detect mutations in genes known to be associated with a specific disease condition (Aparisi et al., 2014;Bonachea et al., 2014;Patel et al., 2018;Rehm, 2013;Shang et al., 2018;Trujillano et al., 2014;X. Wang et al., 2016). Previous studies exploring mutations in colobomaassociated genes have been limited to a subset of genes in affected families (Gonzalez-Rodriguez et al., 2010;). In the present study, we performed targeted sequencing of 37 genes known to be associated with human coloboma in a large cohort of  Perez Millan et al., 2018;Redin et al., 2014). A recent report on diagnosing genetic developmental eye disorders indicated a low diagnostic yield of 8% for microphthalmia, anophthalmia, and coloboma disease, similar to the yield reported here (Patel, Hayward et al., 2019).
Custom capture or targeted next-generation sequencing arrays (Illumina HiSeq 2000 and 2500) have been widely used for mutation detection of several genetic disease conditions, although, with variable yield depending on the genetic disorder and the precise population studied.
Our patient population largely excluded probands where a clear, syndromic diagnosis might be made and molecular testing serves more for confirmation of a clinical diagnosis (e.g., patients who easily meet diagnostic criteria for CHARGE syndrome); had more of such patients been included in this cohort, we posit our yield would have been higher.
Our US-based population was also largely devoid of known consanguinity. In fact, recent studies using next-generation sequencing panels have reported 39% mutation detection in orodental genetic disorders and 61% mutation detection in a cohort of highly consanguineous individuals with microphthalmia, supporting these hypotheses (Patel et al., 2018;Prasad et al., 2016 The previously reported mutation in the DBD of RARB (Arg119*) caused a truncation and total loss of function (Srour et al., 2013).
However, the Arg144Gln mutant described in this study is a viable protein that is mostly retained in the cytoplasm. We hypothesized that mutant RARB that had lost its ability to translocate to the nucleus would "trap" its heterodimeric partners in the cytoplasm. RA and other nuclear receptors such as thyroid hormone and vitamin D3 exert their function by heterodimerizing with RXRs (Germain & Bourguet, 2013;Khorasanizadeh & Rastinejad, 2001). While ligands (at-RA for RARs and cis-RA for RXRs) are important for translocation to the nucleus, RAR and RXR receptors have also been shown to translocate to the nucleus independently of the presence of the ligand and their heterodimerization was shown to play a role in nuclear localization (Baumann, Maruvada, Hager, & Yen, 2001;Mavinakere et al., 2012). RARB/RXRA and RARG/ RXRA heterodimers are instrumental for ocular morphogenesis and RXRA is the main RXR dimerizing with RA receptors during development (Kastner et al., 1997;Krezel et al., 1996;Mark, Ghyselinck, & Chambon, 2006;Mascrez et al., 2001). By confocal imaging, it appeared that RXRA protein was retained in the cytoplasm when cotransfected with RARBmutant but not with wild-type RARB. However, cellular fractionation and quantification did not reveal any difference in RXRA-RFP protein nuclear localization when cotransfected with either wild-type or mutant RARB. Therefore, RARB-mutant protein does not seem to affect the localization of at least two members of the RXR family.
The clinical manifestation in the proband and the mother carrying the Arg144Gln RARB mutation were mainly bilateral uveal coloboma.
These findings represent a system-restricted phenotype compared with previous reports of RARB mutations in humans with either total loss-offunction or twofold to threefold gain-of-function showing involvement of several organs, including eye, heart, diaphragm, lungs, brain, and locomotor system (Srour et al., 2013(Srour et al., , 2016. Srour et al., showed that stop-gain mutation in the DBD and frameshift mutation in LBD caused a total loss-of-function while missense mutations in the LBD resulted in gain-of-function with all the affected showing multisystemic phenotypes. Taken together, these reports and our findings indicate that the RARB protein levels are critical for normal organogenesis and RAsensitive organ systems such as the eye are highly vulnerable to subtle changes in RARB levels. The zebrafish homolog of the human RARB gene, rarga, is expressed in the cranial mesoderm and periocular mesenchyme and is reported to be the most abundant RA receptor expressed during early stages of zebrafish embryonic development (Linville et al., 2009;Oliveira et al., 2013). Previous reports in zebrafish indicate that RA receptor signaling is required for morphogenesis of the ventral optic cup and closure of the optic fissure (Lupo et al., 2011). We reasoned that by knocking down rarga we could recapitulate coloboma in zebrafish and use this as a developmental system to attempt to rescue the phenotype. Indeed, morpholino knockdown of rarga caused dose-dependent ocular coloboma; a comparatively mild and partial rescue of the phenotype was observed only with coinjection of human wild-type RARB mRNA. We posit that the rescue would have been more robust had there been an equivalent of RARB in zebrafish. Nonetheless, this trend was reproducible in three separate experiments, as was our observation that the mutant RARB mRNA did not rescue as well.
This study was initiated with the goal to apply custom capture targeted sequencing for the identification of potential diseasecausing variants in known coloboma-associated genes. The custom capture screening contributed to the identification of mutations in some of the syndromic cases, while the nonsyndromic cases remain elusive. Because of the high genetic heterogeneity, we conclude that broader sequencing approaches, such as whole-exome or genome sequencing, will be required for gene discovery in this disease. The mutation we identified in the DBD of RARB prompted us to further investigate its consequences at the cellular and molecular level. Our data suggest that RARB plays an essential role in RA signaling during optic fissure closure in eye development. Further work on RARB and its downstream targets may help elucidate the RA-induced transcriptional network involved in optic fissure closure. We are also thankful to Dr Sunit Dutta for his suggestions and help with the zebrafish work.