Functional Role of the RNA-Binding Protein Rbm24a and Its Target sox2 in Microphthalmia

Congenital eye defects represent a large class of disorders affecting roughly 21 million children worldwide. Microphthalmia and anophthalmia are relatively common congenital defects, with approximately 20% of human cases caused by mutations in SOX2. Recently, we identified the RNA-binding motif protein 24a (Rbm24a) which binds to and regulates sox2 in zebrafish and mice. Here we show that morpholino knockdown of rbm24a leads to microphthalmia and visual impairment. By utilizing sequential injections, we demonstrate that addition of exogenous sox2 RNA to rbm24a-deplete embryos is sufficient to suppress morphological and visual defects. This research demonstrates a critical role for understanding the post-transcriptional regulation of genes needed for development.


Introduction
Congenital eye defects represent a large class of disorders affecting roughly 21 million children worldwide [1]. Defects can affect every part of the eye from the retina to the lens, but can also include the eye as a whole. Microphthalmia, a smaller than normal eye(s), and anophthalmia, a lack of an eye(s), are both relatively common congenital defects affecting between 1 in 7000 and 1 in 30,000 live births, respectively [2][3][4][5].
Mutations in SRY (sex determining region Y)-box2 (SOX2) account for approximately 20% of human anophthalmia cases [6]. As a transcription factor, SOX2, often in cooperation with a partner transcription factor, is responsible for the regulation of many genes and is one of the Yamanaka reprogramming factors that regulates pluripotency [7][8][9]. Thus, understanding the regulation of SOX2 is crucial. While the transcriptional regulation of SOX2 has been explored, only recently has the post-transcriptional aspect of its regulation been appreciated [10].
Previously, we identified the RNA-binding protein, RBM24, as a post-transcriptional regulator of Sox2 [11,12]. Consistent with SOX2 s known role in microphthalmia and anophthalmia in human patients, knockdown and knockout Rbm24-deficient mice and zebrafish often displayed eye defects including microphthalmia and/or anophthalmia [12,13]. In addition to its role in eye development, rbm24a knockdown, mutation, and overexpression are also associated with cardiac defects. Previous inquiries found Rbm24 mutant mice to be embryonic lethal, dying between E7.5 and E14.5. The embryonic lethality observed is thought to be cardiomyopathy due to numerous cardiovascular malformations, which included ventricular septum defects, reduced trabeculation and compaction, dilated atria and ventricle chambers, thinner atrioventricular endocardial cushions, sarcomere disarray, and fibrosis [12,[14][15][16]. Most previous research focused on the cardiac phenotypes associated with rbm24a depletion.
It is challenging to study global factors such as Rbm24a due to their wide-ranging impacts on a multitude of target RNAs. We chose to focus on a single proposed RNA target, sox2, to perform more in-depth analyses as to the relationship between Rbm24a and sox2. To do so, we analyzed the expression pattern of rbm24a and found, consistent with previous studies in zebrafish, Xenopus, mice, and chick, its expression in the somites, heart, and lens in a spatiotemporal manner [11,14,[17][18][19][20][21][22][23]. To study the role of rbm24a in zebrafish development, we performed morpholino knockdown and CRISPR mutagenesis which phenocopied previously reported morphants and mutants [19,20,23]. We next performed sequential injection of rbm24a morpholino and exogenous sox2 RNA to determine the extent to which sox2 can suppress the rbm24a-depleted microphthalmia. We found phenotypic suppression of the rbm24a-induced microphthalmia. We also performed a visual assay which tests for light/dark detection on these embryos and found partial visually functional rescue by sox2 RNA. This work highlights the study of post-transcriptional modification and their target RNAs during eye development.

Animal Care
Zebrafish are maintained in standard conditions under the approval of the University of Iowa Institutional Animal Care and Use Committee (#8071513, 13 August 2018). Embryos are collected from natural spawning and raised between 28 and 30 • C. No more than 50 embryos are kept per 100 mm plate. Embryo plates are cleaned of dead daily and water changes are made as needed.

Automated Startle Response
The automated vision startle response, VIZN, was performed on 5 or 6 days postfertilization (dpf) larvae as previously described [26]. Phenotypically normal larvae were first tested for the ability to swim by being prodded to ensure touch responsiveness and swimming ability, before being sorted and placed in 48-well plates.

Statistical Analysis
Data in tables were analyzed statistically with Fisher's exact test implement in R [27]. VIZN assays were statistically analyzed by Mann-Whitney.

Mutation of rbm24a Leads to Microphthalmia and Cardiomyopathy
Zebrafish rbm24a is expressed in the developing heart, somites, and lens [19,20,22]. Translation and splice-blocking morpholino knockdown leads to dose-dependent defects, with lower doses exhibiting microphthalmia, while higher doses display microphthalmia and cardiac defects leading to edema (Supplemental Figure S1). We previously analyzed the microphthalmia phenotype both morphologically and histologically [13]. To validate the usage of the morpholino, we aimed to generate an rbm24a mutant. Similar to other organisms, including mice and humans, the zebrafish Rbm24a contains two domains: the RNA recognition motif (RRM) and an Alanine-rich region. The RRM domain spans exons 1 and 2 and is the domain responsible for binding to target RNAs. This domain is highly conserved, with 96.2% identity shared between zebrafish and humans. The Alanine-rich region is in exon 4 and has an unknown function in Rbm24a ( Figure 1). Due to the key role of the RRM domain, we designed a CRISPR site in exon 1 which was a predicted null mutation ( Figure 1). We performed CRISPR knockout and found phenotypes similar to those of morpholino knockdown. Most CRISPR-injected F0 embryos displayed microphthalmia (small eye(s)) and cardiac edema (fluid around the heart) which is consistent with the expression pattern of rbm24a, previously reported morphant phenotypes, and previously reported mutant phenotypes ( Figure 1B-E) [19,20,23]. We extracted genomic DNA (gDNA) from both uninjected and CRISPR-injected F0, amplified the region flanking the target cut site, and sequenced the products. We next used Synthego ICE to determine the nature and frequency of indels [25]. Analysis of the CRISPR-injected F0 mutants showed an indel frequency between 20% and 24% when compared to the uninjected control embryos ( Figure 1F). There is a statistically significant difference between the phenotypes of uninjected control and CRISPR-injected F0 embryos (p < 1 × 10 −16 ) ( Table 1).

rbm24a RNA Suppression of rbm24a Morpholino Knockdown
Due to the high number of CRISPR-injected F0 embryos exhibiting both microphthalmia and cardiac edema, we chose to utilize a low dose of translation-blocking morpholino in order to generate microphthalmia-only rbm24a knockdown phenotypes. When compared to uninjected embryos, morphant embryos injected with a low dose (1-1.3 ng) displayed only microphthalmia, while embryos injected at a higher dose (1.7-2 ng) phenocopied the CRISPR-injected F0 embryos with both microphthalmia and cardiac edema (      As an additional confirmation of the rbm24a knockdown phenotype, we next determined the extent to which exogenous rbm24a RNA can suppress the knockdown defects. We modified the rbm24a RNA construct so the morpholino could not bind to this exogenous rbm24a RNA. When injected alone, rbm24a RNA led to an overexpression phenotype very similar to the rbm24a knockdown phenotype (100-400 pg; Figure 2D-D'). The similarity of the rbm24a knockdown and overexpression phenotypes poses a challenge for phenotypic suppression studies, yet when we performed sequential injections of rbm24a morpholino and rbm24a RNA, we demonstrated phenotypic suppression ( Figure 2E-E'). In morpholino knockdown alone, approximately 20% of the embryos were phenotypically normal, while the remaining 80% of embryos displayed defects. When embryos were sequentially injected with rbm24a morpholino and rbm24a RNA, the number with normal morphology doubled to nearly 40%, resulting in partial suppression of the knockdown microphthalmia defect (p = 0.0745) ( Figure 2F, Table 2). The overlapping phenotypes generated by two different morpholinos, the suppression of the knockdown phenotype by exogenous rbm24a RNA, and the phenotypic similarity between the CRISPR-injected and morphant embryos indicate that the phenotypes observed are specific to rbm24a.

EGFP RNA Does Not Suppress rbm24a Morpholino Knockdown Phenotypes
Rbm24a is an RNA-binding protein. Therefore, we sought to determine if any exogenous RNA was sufficient to suppress the rbm24a knockdown phenotype. We utilized EGFP RNA as a control for RNA injection. EGFP is not naturally found in zebrafish and should have no function. Additionally, we can check for successful injection by screening embryos for EGFP fluorescence ( Figure 3A). When compared to uninjected control embryos, both embryos injected with rbm24a morpholino (1.7-2 ng) and embryos injected with both rbm24a morpholino and EGFP RNA (200-300 pg) showed similar phenotypes, including many with microphthalmia and microphthalmia with cardiac edema ( Figure 3B-D'). Injection of EGFP RNA alone yielded a phenotype similar to the uninjected control group ( Figure 3E-E'). The number of normal embryos in both the uninjected group and EGFP RNA group is almost identical (p = 0.1050), while the number and nature of the affected embryos in both the rbm24a knockdown and rbm24a knockdown with EGFP RNA are also very similar (p = 0.5884) ( Figure 3F, Table 3). Taken together, these data suggest that rbm24a knockdown defects cannot be suppressed by sequential injection of an unrelated exogenous RNA.

sox2 RNA Phenotypically Suppresses rbm24a Morpholino-Induced Microphthalmia
We previously identified sox2 as a target of Rbm24a in both zebrafish and mouse models [12]. In that study, we demonstrated that loss of rbm24a led to decreased levels of sox2. We hypothesize that supplying exogenous sox2 RNA would supplement the reduced endogenous sox2 RNA levels. We previously identified the binding site in the 3 UTR of Sox2 for RBM24 via mouse cell culture studies [12]. Thus, we generated a zebrafish sox2 RNA construct which lacked the 3 UTR to prevent Rbm24a binding (referred to as sox2 RNA). To test the functional role of sox2 as a target of Rbm24a, we performed sequential injection of rbm24a morpholino followed by sox2 RNA. Knockdown of rbm24a (1.7-2 ng) generated embryos with microphthalmia or microphthalmia with cardiac edema when compared against the uninjected control ( Figure 4A-B').   In contrast, embryos sequentially injected with both rbm24a morpholino and sox2 RNA (200-300 pg) were often normal in morphology ( Figure 4C-C'). Injection of sox2 RNA alone also generated mostly phenotypically normal embryos ( Figure 4D-D'). A majority of uninjected, rbm24a morpholino and sox2 RNA-injected, and sox2 RNA-injected embryos were morphologically normal with some exceptions in both injection groups. The rbm24a knockdown embryos displayed more embryos with microphthalmia or microphthalmia with cardiac edema when compared to the rbm24a knockdown with sox2 RNA, indicating that injection of exogenous sox2 RNA can partially suppress the rbm24a knockdown phenotype ( Figure 4E, Table 4).

EGFP RNA Does Not Functionally Suppress rbm24a-Induced Visual Defects
We next wanted to determine the functional role of rbm24a in vision. Previously, we demonstrated that morpholino knockdown of rbm24a impacts the startle response in a dose-dependent manner [13]. When 5 to 6 days post-fertilization zebrafish larvae are exposed to an interruption of a constant light source, they perform a characteristic escape response which can be tracked with motion detection cameras and software. We utilized the automated startle response assay (VIZN) which generates five interruptions in light and records the zebrafish movement in response to the interruption in light [26]. Uninjected control larvae responded roughly four out of five times, indicating that they are visually responsive to the stark change in light in the startle response assay. Our previous studies indicated that low-dose knockdown of rbm24a maintained light detection [13]. For this study, we utilized a mid-range dose of rbm24a morpholino to better evaluate the effectiveness of RNA suppression. For all of our vision assays, we only utilized larvae that were touch responsive with a characteristic swimming response and had no cardiac edema. These rbm24a morphant larvae responded to the automated startle response assay (VIZN) significantly fewer times (approximately two out of five times) than the uninjected control ( Figure 5). When sequentially injected with both rbm24a morpholino and EGFP RNA, larvae had a similar response to the rbm24a morpholino-alone larvae and responded about two out of five times. The difference between the rbm24a morpholino-only and rbm24a morpholino with EGFP RNA was not significantly different. The larvae injected with only EGFP RNA responded roughly four out of five times. There was no significant difference between the uninjected control and EGFP RNA-alone groups ( Figure 5).

sox2 RNA Partially Functionally Suppresses rbm24a-Induced Visual Defects
We next investigated the extent to which sox2 RNA can restore visual functionality to rbm24a knockdown larvae. We performed VIZN on the uninjected control and rbm24a knockdown larvae. Uninjected control larvae responded approximately four out of five times, indicating that they are visually responsive to the startle response assay ( Figure 6). As stated previously, we purposely utilized a dose of rbm24a morpholino which resulted in larvae responding statistically significantly fewer times (roughly two out of five) than the uninjected control. Sequential injection of rbm24a morpholino and sox2 RNA larvae responded significantly more times (about three out of five) than rbm24a morpholino-injected larvae. However, the rbm24a morpholino and sox2 RNA larvae were also statistically significantly different from the uninjected control larvae. Larvae injected with only sox2 RNA responded similarly (about four out of five times) to the uninjected control ( Figure 6). Taken together, these data indicate sox2 partially restores visual function in rbm24a morpholino knockdown larvae.

sox2 RNA Partially Functionally Suppresses rbm24a-Induced Visual Defects
We next investigated the extent to which sox2 RNA can restore visual functionality to rbm24a knockdown larvae. We performed VIZN on the uninjected control and rbm24a knockdown larvae. Uninjected control larvae responded approximately four out of five times, indicating that they are visually responsive to the startle response assay ( Figure 6). As stated previously, we purposely utilized a dose of rbm24a morpholino which resulted in larvae responding statistically significantly fewer times (roughly two out of five) than the uninjected control. Sequential injection of rbm24a morpholino and sox2 RNA larvae responded significantly more times (about three out of five) than rbm24a morpholino-injected larvae. However, the rbm24a morpholino and sox2 RNA larvae were also statistically significantly different from the uninjected control larvae. Larvae injected with only sox2 RNA responded similarly (about four out of five times) to the uninjected control (Figure 6). Taken together, these data indicate sox2 partially restores visual function in rbm24a morpholino knockdown larvae. Figure 5. EGFP RNA does not improve visual function of rbm24a knockdown embryos. Automated startle response assay (VIZN) was performed on uninjected, rbm24a knockdown, rbm24a knockdown with EGFP RNA, and EGFP RNA larvae. Knockdown of rbm24a inhibits visual function, which was statistically significant compared to uninjected. Addition of EGFP RNA to rbm24a knockdown does not statistically significantly increase visual function when compared to rbm24a knockdown alone. Injection of EGFP RNA alone does not statistically significantly alter visual function from uninjected. Mann-Whitney *** p < 0.001, **** p < 0.0001. Nonsignificant interactions not shown.
Biomedicines 2021, 9, x FOR PEER REVIEW 13 of 16 Figure 6. Injection of sox2 RNA can partially restore visual function by VIZN to rbm24a knockdown embryos. VIZN was performed on uninjected, rbm24a knockdown, rbm24a knockdown with sox2 RNA, and sox2 RNA larvae. Knockdown of rbm24a statistically significantly inhibits visual function when compared to uninjected. Addition of sox2 RNA to rbm24a knockdown statistically significantly improves visual function when compared to rbm24a knockdown, but not to the same level as uninjected. Injection of sox2 RNA alone does not statistically significantly alter visual function from uninjected. Mann-Whitney *p < 0.05, **p < 0.01, ****p < 0.0001. Nonsignificant interactions not shown.

Discussion
Our previous research indicated sox2 as a target for Rbm24a and suggests that Rbm24a binds to and stabilizes the sox2 mRNA transcript [12]. Localization data indicate that rbm24a is expressed exclusively in the lens. This suggests that Rbm24a binds to and regulates lens-expressed sox2 in the lens [17][18][19][20][21]28]. When knocked down via morpholino or mutated via CRISPR, the levels of Rbm24a are depleted which leads to a decrease in the stability and amount of sox2 mRNA. One possibility is that the reduced levels of sox2, a proliferative factor, lead to decreased lens vesicle size. Due to coordinated development Figure 6. Injection of sox2 RNA can partially restore visual function by VIZN to rbm24a knockdown embryos. VIZN was performed on uninjected, rbm24a knockdown, rbm24a knockdown with sox2 RNA, and sox2 RNA larvae. Knockdown of rbm24a statistically significantly inhibits visual function when compared to uninjected. Addition of sox2 RNA to rbm24a knockdown statistically significantly improves visual function when compared to rbm24a knockdown, but not to the same level as uninjected. Injection of sox2 RNA alone does not statistically significantly alter visual function from uninjected. Mann-Whitney * p < 0.05, ** p < 0.01, **** p < 0.0001. Nonsignificant interactions not shown.

Discussion
Our previous research indicated sox2 as a target for Rbm24a and suggests that Rbm24a binds to and stabilizes the sox2 mRNA transcript [12]. Localization data indicate that rbm24a is expressed exclusively in the lens. This suggests that Rbm24a binds to and regulates lens-expressed sox2 in the lens [17][18][19][20][21]28]. When knocked down via morpholino or mutated via CRISPR, the levels of Rbm24a are depleted which leads to a decrease in the stability and amount of sox2 mRNA. One possibility is that the reduced levels of sox2, a proliferative factor, lead to decreased lens vesicle size. Due to coordinated development between the lens vesicle and optic cup via reciprocal induction, the optic cup develops in tandem with the lens vesicle and results in microphthalmia (Figure 7). There, the protein acts to bind to and stabilize sox2 mRNA. This interaction leads to the development of a normal sized lens and, with reciprocal induction signaling between the lens and retina, the retina also develops normally to size-match the lens. (B) With either knockdown or mutation of rbm24a, there is lessened Rbm24a protein in the lens which in turn cannot stabilize as many sox2 mRNA molecules (represented as transparent shapes). The lack of the correct amount of sox2, a proliferative factor, causes the lens to develop smaller than normal. However, reciprocal induction has not been affected, which leads to the lens and retina developing small in tandem.
Recently, a zebrafish TALEN mutant for rbm24a has been published which phenocopies both our morpholino knockdown and CRISPR-injected F0 defects [23]. In the Shao et al. 2020 study, the authors suggest that microphthalmia is the result of a cardiac morphological defect leading to poor blood flow to the eye. We hypothesized that the microphthalmia was due to target RNA disruption by decreased rbm24a expression. In our previously published studies, we identified that RBM24 binds to Sox2 in mouse cell cultures [12]. From this, we hypothesized that knockdown of rbm24a with the addition of exogenous sox2 would suppress the microphthalmia phenotype. We found that sequential knockdown of rbm24a and the addition of exogenous sox2 RNA generated a phenotype more similar to that of uninjected control embryos than rbm24a knockdown ( Figure 4A-C'). Our data indicate sox2 is a main contributor to the rbm24a-induced microphthalmia. In terms of the previous study by Shao et al., it is possible that the TALEN mutant was not a null mutation which could have allowed for some function in the lens. Additionally, the utilization of heterotypic parabiosis allows for the potential transfer of more than just blood (including hormones, signaling molecules, etc.) between embryos. It is possible that the transfer of these biomolecules could have played a role in suppressing the microphthalmia phenotype [23]. Follow-up studies utilizing an organ-specific driver would resolve this contrast.
We occasionally observed cardiovascular defects in the sox2 overexpression embryos, perhaps stemming from the ubiquitous overexpression of sox2. Ubiquitously expressed There, the protein acts to bind to and stabilize sox2 mRNA. This interaction leads to the development of a normal sized lens and, with reciprocal induction signaling between the lens and retina, the retina also develops normally to size-match the lens. (B) With either knockdown or mutation of rbm24a, there is lessened Rbm24a protein in the lens which in turn cannot stabilize as many sox2 mRNA molecules (represented as transparent shapes). The lack of the correct amount of sox2, a proliferative factor, causes the lens to develop smaller than normal. However, reciprocal induction has not been affected, which leads to the lens and retina developing small in tandem.
Recently, a zebrafish TALEN mutant for rbm24a has been published which phenocopies both our morpholino knockdown and CRISPR-injected F0 defects [23]. In the Shao et al. 2020 study, the authors suggest that microphthalmia is the result of a cardiac morphological defect leading to poor blood flow to the eye. We hypothesized that the microphthalmia was due to target RNA disruption by decreased rbm24a expression. In our previously published studies, we identified that RBM24 binds to Sox2 in mouse cell cultures [12]. From this, we hypothesized that knockdown of rbm24a with the addition of exogenous sox2 would suppress the microphthalmia phenotype. We found that sequential knockdown of rbm24a and the addition of exogenous sox2 RNA generated a phenotype more similar to that of uninjected control embryos than rbm24a knockdown ( Figure 4A-C'). Our data indicate sox2 is a main contributor to the rbm24a-induced microphthalmia. In terms of the previous study by Shao et al., it is possible that the TALEN mutant was not a null mutation which could have allowed for some function in the lens. Additionally, the utilization of heterotypic parabiosis allows for the potential transfer of more than just blood (including hormones, signaling molecules, etc.) between embryos. It is possible that the transfer of these biomolecules could have played a role in suppressing the microphthalmia phenotype [23]. Follow-up studies utilizing an organ-specific driver would resolve this contrast.
We occasionally observed cardiovascular defects in the sox2 overexpression embryos, perhaps stemming from the ubiquitous overexpression of sox2. Ubiquitously expressed exogenous sox2 may be able to substitute for cardiac members of the SOX family including Sox6 which has been shown to be a cardiomyocyte regulator in murine cells, Sox9 which is involved in heart valve development in mice, and Sox17 which has been shown to be essential for specification of cardiac mesoderm in mouse cell cultures [29][30][31]. In the future, an eye-specific promoter could be utilized to allow for the study of rbm24a/sox2 interactions with specificity to the eye.
In this study, we demonstrated that knockdown and mutation of the RNA-binding motif protein 24a gene, rbm24a, leads to microphthalmia at low-dose knockdown and microphthalmia with cardiac edema at increased knockdown doses. Additionally, we demonstrated that the microphthalmia induced by rbm24a functionally impacted the visual capabilities of the larvae, as knockdown fish performed worse in a visual behavior study. Previously, we identified sox2 as a target of Rbm24a [12]. When sequentially injected, exogenous sox2 RNA is able to phenotypically and partially functionally suppress the rbm24a knockdown larvae. This shows functional validation of a suspected target RNA by an RNA-binding protein. This work broadly demonstrates the key role RNA-binding proteins and post-transcriptional modification plays during development. It also highlights that genes not associated with disease, such as rbm24a, can impact a well-studied diseaseassociated gene (sox2), making it critical to better understand post-transcriptional regulation and the potential impacts of RNA-binding proteins and their associated target RNAs.