Abstract
Mutations in the key transcription factor, SOX2, alone account for 20% of anophthalmia (no eye) and microphthalmia (small eye) birth defects in humans—yet its regulation is not well understood, especially on the post-transcription level. We report the unprecedented finding that the conserved RNA-binding motif protein, RBM24, positively controls Sox2 mRNA stability and is necessary for optimal SOX2 mRNA and protein levels in development, perturbation of which causes ocular defects, including microphthalmia and anophthalmia. RNA immunoprecipitation assay indicates that RBM24 protein interacts with Sox2 mRNA in mouse embryonic eye tissue. and electrophoretic mobility shift assay shows that RBM24 directly binds to the Sox2 mRNA 3’UTR, which is dependent on AU-rich elements (ARE) present in the Sox2 mRNA 3’UTR. Further, we demonstrate that Sox2 3’UTR AREs are necessary for RBM24-based elevation of Sox2 mRNA half-life. We find that this novel RBM24–Sox2 regulatory module is essential for early eye development in vertebrates. We show that Rbm24-targeted deletion using a constitutive CMV-driven Cre in mouse, and rbm24a-CRISPR/Cas9-targeted mutation or morpholino knockdown in zebrafish, results in Sox2 downregulation and causes the developmental defects anophthalmia or microphthalmia, similar to human SOX2-deficiency defects. We further show that Rbm24 deficiency leads to apoptotic defects in mouse ocular tissue and downregulation of eye development markers Lhx2, Pax6, Jag1, E-cadherin and gamma-crystallins. These data highlight the exquisite specificity that conserved RNA-binding proteins like RBM24 mediate in the post-transcriptional control of key transcription factors, namely, SOX2, associated with organogenesis and human developmental defects.
Introduction
Several key transcription factors, including LHX2, OTX2, PAX6, RAX, SIX3, SIX6 and SOX2, among others, are involved in early stages of vertebrate eye development, in the specification of an eye field in the anterior neural plate (1–3). Subsequently, Sonic hedgehog signaling bifurcates the single eye field into two structures termed optic sulci that interact with their opposing surface ectoderm tissue, to eventually form the retina and lens, respectively (4). Thus far, precise spatiotemporal expression of key transcription and signaling factors are shown to be necessary for these processes, disruption of which causes developmental ocular defects.
Indeed, developmental eye disorders such as microphthalmia (small eye) and anophthalmia (no eye) affect 1 in 7000 and 1 in 30 000 live human births, respectively (5,6). About 20% of human anophthalmic cases are linked to mutations in the SOX2 gene alone, which, in addition to its critical function in development of eye (7–9) and other organs, is also one of the Yamanaka cellular reprogramming factors that controls pluripotency (9–12). While transcriptional regulation of Sox2 is characterized in detail in various developing tissues (13), the impact of post-transcriptional regulatory mechanisms, particularly those regulated by RNA-binding proteins (RBPs), on the fate of its transcribed mRNA is less clear. This reflects a larger knowledge gap, as there are similar numbers of transcription factors and RBPs encoded by the human genome, but the functional significance of the latter in development is not well described (14–17).
We recently published an update to the database ‘iSyTE’ (integrated Systems Tool for Eye gene discovery) which allows in silico prediction of genes linked to eye development and disease (18–20). However, such iSyTE gene predictions need to be tested experimentally to conclusively establish their role in eye development and disease. Here, we applied iSyTE to identify a new regulator of eye development—the RNA-binding motif family protein, RBM24. RBM24 is a highly conserved RBP in vertebrates with a single RNA recognition motif domain located near its N-terminus (21,22). Thus far, RBM24 has been largely studied in the context of alternative splicing in cardiac development and differentiation (23–25). Further, RBM24 has been implicated in controlling mRNA decay of the cell cycle regulators p63 and p21 and the mRNA stability of the myogenic factor MyoD and translation of p53 (26–28).
Here we report the unprecedented finding that RBM24 binds to the AU-rich elements (AREs) within the 3’UTR of Sox2 mRNA and leads to its stabilization, which is necessary for normal expression levels of SOX2 protein in eye development. We find that Rbm24 deficiency in mouse and fish results in downregulation of Sox2 and causes the eye defects anophthalmia and microphthalmia. Rbm24 knockout mice show elevated apoptosis in developing ocular tissue and abnormal expression of eye development regulators and markers such as LHX2, PAX6, JAG1, E-cadherin and γ-crystallin. These data provide a transformational understanding of RBP-mediated post-transcriptional control of key transcriptional regulators, particularly SOX2, in organogenesis.
Results
RBM24 deficiency causes microphthalmia and anophthalmia in mouse
Using the bioinformatics tool iSyTE (integrated Systems Tool for Eye gene discovery), which prioritizes genes based on eye tissue enrichment scores (18–20,29–32), we identified Rbm24 as a potential regulator of eye development (Fig. 1A). In mouse, RBM24 is expressed in the presumptive lens ectoderm and the optic vesicle at embryonic day (E) 9.5 (33) (Fig. 1B) and later in the lens and the retina (Fig. 1C), suggesting that RBM24 functions in eye development from early stages.
RBM24 is expressed during mouse eye development. (A) iSyTE prioritizes Rbm24 as a gene with potential function in the eye based on high lens-enriched expression. Heat map red intensity indicates high lens enrichment. (B) During mouse development, Rbm24 protein is expressed in the presumptive lens ectoderm (*) and optic vesicle (**) at embryonic day (E)9.5 and (C) in the cornea, lens anterior epithelial cells and fiber cells at E14.5. (D) Generation of Rbm24−/− mice using floxed Rbm24 alleles and CMV promoter-driven Cre recombinase. (E) RT-qPCR demonstrates Rbm24 downregulation in Rbm24−/− embryos at E7.5. (F) Western blot analysis confirms absence of RBM24 protein in Rbm24−/− mouse embryonic tissue at E14.5. (G, H) Immunostaining shows absence of Rbm24 protein in Rbm24−/− mouse eye tissue at E14.5. Abbr.: ple, presumptive lens ectoderm; ov, optic vesicle; e, anterior lens epithelium; f, lens fiber cells; r, retina; f.c., fold change. Asterisks in E indicate P-value < 0.005. Scale bar for B, C, J, K is 70 μm, for G, H is 140 μm.
To investigate RBM24 function in development, we generated a new Rbm24 constitutive knockout mouse line, Rbm24−/−, and confirmed the absence of RBM24 at both RNA and protein levels in mutant tissue (Fig. 1D-H). Rbm24−/− mice die at variable frequency after E7.5 and do not survive after E14.5, and further, after stage E9.5, they are smaller in contrast to control (Table 1). The embryonic lethality defect is observed in an independently generated Rbm24 null mouse line and is attributed to cardiac abnormalities (23).
Embryonic lethality and ocular defects in Rbm24−/− mice
| Embryonic stage | Number of litters dissected | Number of litters with no Rbm24−/− embryos | Number of Rbm24−/− embryos | Unilateral anophthalmia Rbm24−/− embryos | Bilateral microphthalmia Rbm24−/− embryos |
|---|---|---|---|---|---|
| E7.5 | 6 | 1 | 11 | na | na |
| E9.5 | 12 | 5 | 14 | na | na |
| E10.5 | 10 | 6 | 5 | 2 | 3 |
| E11.5 | 5 | 2 | 6 | 3 | 3 |
| E12.5 | 9 | 5 | 5 | 2 | 3 |
| E14.5 | 7 | 4 | 6 | 2 | 4 |
| Embryonic stage | Number of litters dissected | Number of litters with no Rbm24−/− embryos | Number of Rbm24−/− embryos | Unilateral anophthalmia Rbm24−/− embryos | Bilateral microphthalmia Rbm24−/− embryos |
|---|---|---|---|---|---|
| E7.5 | 6 | 1 | 11 | na | na |
| E9.5 | 12 | 5 | 14 | na | na |
| E10.5 | 10 | 6 | 5 | 2 | 3 |
| E11.5 | 5 | 2 | 6 | 3 | 3 |
| E12.5 | 9 | 5 | 5 | 2 | 3 |
| E14.5 | 7 | 4 | 6 | 2 | 4 |
Embryonic lethality and ocular defects in Rbm24−/− mice
| Embryonic stage | Number of litters dissected | Number of litters with no Rbm24−/− embryos | Number of Rbm24−/− embryos | Unilateral anophthalmia Rbm24−/− embryos | Bilateral microphthalmia Rbm24−/− embryos |
|---|---|---|---|---|---|
| E7.5 | 6 | 1 | 11 | na | na |
| E9.5 | 12 | 5 | 14 | na | na |
| E10.5 | 10 | 6 | 5 | 2 | 3 |
| E11.5 | 5 | 2 | 6 | 3 | 3 |
| E12.5 | 9 | 5 | 5 | 2 | 3 |
| E14.5 | 7 | 4 | 6 | 2 | 4 |
| Embryonic stage | Number of litters dissected | Number of litters with no Rbm24−/− embryos | Number of Rbm24−/− embryos | Unilateral anophthalmia Rbm24−/− embryos | Bilateral microphthalmia Rbm24−/− embryos |
|---|---|---|---|---|---|
| E7.5 | 6 | 1 | 11 | na | na |
| E9.5 | 12 | 5 | 14 | na | na |
| E10.5 | 10 | 6 | 5 | 2 | 3 |
| E11.5 | 5 | 2 | 6 | 3 | 3 |
| E12.5 | 9 | 5 | 5 | 2 | 3 |
| E14.5 | 7 | 4 | 6 | 2 | 4 |
Here, we demonstrate that both Rbm24−/− mice exhibit severe ocular defects at 100% penetrance. At E11.5, approximately half of the Rbm24−/− mice exhibit anophthalmia (absent/rudimentary ocular tissue due to arrested eye development) on one side and microphthalmia (small eye) on the other, while the remainder of the mutants exhibit bilateral microphthalmia (Fig. 2A-D’).
Rbm24 is required for mouse eye development. (A-D’) At E11.5, Rbm24−/− mice exhibit anophthalmia (*) on one side and microphthalmia on the other side (**). Abbr.: l, lens; r, retina.
rbm24a deficiency causes eye defects in zebrafish
Zebrafish have two Rbm24 orthologs (rbm24a and rbm24b), and we examined the expression of both orthologs by in situ hybridization (ISH). rbm24a is expressed early in the eye, heart and somites (Fig. 3A). And, analogous to mouse, rbm24a becomes enriched in the lens at 2-day post fertilization (dpf) (Fig. 3B). In contrast, rbm24b exhibits high expression in the heart and somites with little or no expression in the developing eye tissue (Fig. 3C, C’). The expression patterns of rbm24a and rbm24b are supported by the zebrafish single-cell RNA sequencing database, SPRING (34). Additionally, we previously demonstrated that rbm24a morphant zebrafish have visual defects (35). These data suggest that rbm24a likely functions in the eye, and therefore we used morpholinos and CRISPR/Cas9 to generate rbm24a knockdown and deletion, respectively, in zebrafish. In a dose-dependent manner, rbm24a morphants exhibit progressively severe microphthalmia at high frequency (1 ng, 76%; 2 ng, 92%) or anophthalmia (rudimentary or diffused ocular tissue) at 1.5% penetrance at 3 dpf stage (Fig. 3D-E’, Table 2). To confirm the zebrafish rbm24a knockdown phenotypes, CRISPR/Cas9 was used to generate deletions in rbm24a, which resulted in microphthalmia in F0 embryos at 3 dpf stage (Fig. 3F-G’).
rbm24a is required for zebrafish eye development. (A) ISH shows rbm24a transcript expression in the eye (arrowhead), heart and somites at 20 somite stage (~19hpf). (B) At stage 2 dpf rbm24a transcript expression is strong in the eye region (arrowhead). (C-C’) At stage 2 dpf rbm24b transcript expression is expressed in the somites, but expression in the eye tissue is not detected. (D-E’) While control eyes are normal, rbm24a morpholino injected embryos exhibit microphthalmia (asterisk). (F-G’) CRISPR/Cas9-based targeted rbm24a deletion zebrafish exhibit microphthalmia (asterisk). Abbr.: MO, morpholino; KO, knockout; l, lens; r, retina.
Ocular defects in rbm24a morphant and F01 mutant zebrafish
| Treatment | Microphthalmia/anophthalmia | Normal | Total |
|---|---|---|---|
| Uninjected | 1/0 | 220 | 221 |
| Control MO (1–2 ng) | 2/0 | 238 | 240 |
| rbm24a AUG MO (1–1.5 ng) | 109/0 | 32 | 141 |
| rbm24a AUG MO (2–2.5 ng) | 125/2 | 11 | 138 |
| rbm24a F0 CRISPR injected | 49/0 | 24 | 73 |
| Treatment | Microphthalmia/anophthalmia | Normal | Total |
|---|---|---|---|
| Uninjected | 1/0 | 220 | 221 |
| Control MO (1–2 ng) | 2/0 | 238 | 240 |
| rbm24a AUG MO (1–1.5 ng) | 109/0 | 32 | 141 |
| rbm24a AUG MO (2–2.5 ng) | 125/2 | 11 | 138 |
| rbm24a F0 CRISPR injected | 49/0 | 24 | 73 |
Ocular defects in rbm24a morphant and F01 mutant zebrafish
| Treatment | Microphthalmia/anophthalmia | Normal | Total |
|---|---|---|---|
| Uninjected | 1/0 | 220 | 221 |
| Control MO (1–2 ng) | 2/0 | 238 | 240 |
| rbm24a AUG MO (1–1.5 ng) | 109/0 | 32 | 141 |
| rbm24a AUG MO (2–2.5 ng) | 125/2 | 11 | 138 |
| rbm24a F0 CRISPR injected | 49/0 | 24 | 73 |
| Treatment | Microphthalmia/anophthalmia | Normal | Total |
|---|---|---|---|
| Uninjected | 1/0 | 220 | 221 |
| Control MO (1–2 ng) | 2/0 | 238 | 240 |
| rbm24a AUG MO (1–1.5 ng) | 109/0 | 32 | 141 |
| rbm24a AUG MO (2–2.5 ng) | 125/2 | 11 | 138 |
| rbm24a F0 CRISPR injected | 49/0 | 24 | 73 |
Together, these data identify RBM24 as a new post-transcriptional regulator necessary for eye development in vertebrates.
Rbm24-deficient mouse eye tissue exhibits apoptotic defects
Because microphthalmia/anophthalmia can result from elevated apoptosis in development (36–38), we performed cell death analysis, which demonstrated Rbm24−/− ocular tissue to have high number of TUNEL-positive cells (Fig. 4A-E). We find that the Notch pathway effector JAG1, which positively controls apoptosis inhibitors and functions in eye development (39,40), is downregulated at both the mRNA and protein levels in Rbm24−/− mice (Fig. 4F-H). These data suggest that RBM24-mediated positive control of JAG1 is necessary for cell viability in eye development and the disruption of this regulatory module contributes to the pathology of the ocular defects in Rbm24 mutants. Further, one of the apoptosis inhibitors regulated by JAG1, namely, BIRC2 (41), is severely downregulated in Rbm24−/− mouse ocular tissue at the mRNA level (Fig. 4I). Interestingly, RBM24 binds to Birc2 mRNA directly as shown by RNA immunoprecipitation (RIP) assay suggesting that RBM24-mediated regulation of apoptotic inhibitor is required for cell viability in the ocular tissue (Fig. 4J).
RBM24 deficiency causes apoptotic defects. (A-D) In contrast to control, Rbm24−/− mice have high number of TUNEL-positive nuclei (arrowhead) in the ocular region at E10.5 and E12.5. (E) Quantification of TUNEL immunofluorescence indicates significantly higher number of apoptotic nuclei in Rbm24−/− ocular region. (F) RT-qPCR analysis indicates significant Jag1 downregulation in Rbm24−/− in contrast to controls. (G, H) Immunostaining analysis indicates Jag1 is downregulated in E12.5 Rbm24−/− lenses in contrast to control. (I) RT-qPCR analysis indicates significant Birc2 downregulation in Rbm24−/− in contrast to controls. (J) RNA immunoprecipitation with RBM24 antibody on mouse embryonic eye tissue shows enrichment of Birc2 mRNA. Abbr.: lpt, lens pit; oc, optic cup; e, lens epithelial cells; f, lens fiber cells; r, retina; f.c., fold change. Asterisks in E, F, I, J indicate a P-value < 0.05. Scale bar for A-D, G, H is 70 μm; # indicates whole-body tissue.
RBM24 deficiency causes downregulation of key eye development markers
To gain further insight into the molecular basis of the ocular phenotype in Rbm24−/− mice, we examined the expression of lens development markers PAX6, E-cad (E-cadherin) and γ-cry (gamma-crystallins). In the mildly affected Rbm24−/− embryos, PAX6 is undetected in the anterior cells of the lens vesicle. Further, Pax6 was detectable in only a few cells in the posterior region of the lens vesicle. On the other hand, in the severely affected Rbm24−/− embryos, PAX6 is undetected in the region where lens vesicle is normally formed. Pax6 expression in the retina of Rbm24−/− embryos is unaltered (Fig. 5A-C). While E-cad and γ-cry are significantly reduced in less-severely affected mutant lenses, E-cad is downregulated, and γ-cry is absent in severely affected mutant lenses, suggesting abnormalities in lens epithelial and fiber cells, and therefore lens differentiation, in Rbm24−/− embryos (Fig. 5D-I).
Rbm24−/− mice exhibit downregulation of eye development markers. (A-C) PAX6 is significantly downregulated in the lens vesicle, but not in the retina tissue of Rbm24−/− at E11.5 in contrast to control. (D-I) E-cad and gamma-cry are severely downregulated in lens epithelial and lens fiber cells, respectively, at E12.5 and E14.5 in Rbm24−/− in contrast to control. Abbr.: c, cornea; e, lens epithelial cells; f, lens fiber cells; l, lens; r, retina; se, surface ectoderm; ov, optic vesicle; lv, lens vesicle; oc, optic cup. Scale bar for A-I is 70 μm.
Next, we investigated early eye development markers, namely, SOX2 and LHX2, which are linked to anophthalmia and/or microphthalmia.
LHX2 is a direct target of RBM24
Because Rbm24 is expressed in the optic vesicle, we next examined whether its deficiency impacted LHX2 expression. We find LHX2 is downregulated in the optic vesicle of Rbm24−/− mice at E9.5 (Fig. 6A and B) and consistently, Lhx2 mRNA is severely downregulated in Rbm24−/− eye tissue (Fig. 6C). Furthermore, we performed RIP assay using Rbm24 antibody on E14.5 wild-type whole eye lysates and find enrichment of Lhx2 mRNA in the Rbm24 antibody pulldown (Fig. 6D). As confirmation, we performed Rbm24 knockdown in NIH3T3 cells and found significant downregulation of Lhx2 mRNA (Fig. 6D). Further, Lhx2 is significantly upregulated when RBM24 is overexpressed in 21EM15 lens epithelial cell lines (Fig. 6F). To address mRNA stability, 21EM15 cells transfected with the RBM24 overexpression vector were treated with the general transcriptional inhibitor actinomycin D, and the rate of Lhx2 mRNA decay was measured. We observed an increase in the half-life of Lhx2 transcript in RBM24 overexpressing cells in contrast to the controls (Fig. 6G), indicating that RBM24 stabilizes Lhx2 transcript.
RBM24-deficient mouse exhibit downregulation of LHX2 expression. (A, B) Lhx2 is significantly downregulated in the optic vesicle of Rbm24−/− at E9.5 in contrast to control. (C) RT-qPCR analysis confirms significant Lhx2 downregulation in Rbm24−/− mice. (D) RNA immunoprecipitation with Rbm24 antibody on mouse embryonic eye tissue shows enrichment of Lhx2 mRNA. (E) Rbm24 KD in NIH3T3 cells results in downregulation of Lhx2 transcript. (F) Overexpression of Rbm24 in 21EM15 cells causes upregulation of Lhx2 transcript. (G) Rbm24 overexpression in 21EM15 cells results in significant increase in the half-life of Lhx2 mRNA. Abbr.: se, surface ectoderm; ov, optic vesicle; f.c., fold change. Asterisks in C-F indicate P-value < 0.05. Scale bar in A, B is 70 μm.
SOX2 is a direct target of RBM24
Because, SOX2 is required for the sustained expression of PAX6 in early lens development (42) and is linked to human anophthalmia/microphthalmia, we next investigated the expression of SOX2 in Rbm24−/− mice. Interestingly, SOX2 protein is significantly downregulated in Rbm24−/− mice during early eye development at E9.5 in the presumptive lens ectoderm and optic vesicle (Fig. 7A-B’) and is undetected in Rbm24−/− ocular tissue at later developmental stages (Fig. 7C-D’). Sox2 mRNA is expectedly found to be severely downregulated in Rbm24−/− eye tissue (Fig. 7E). Moreover, sox2 mRNA is also found to be downregulated in rbm24a morphants zebrafish (Fig. 7F), suggesting that the function of RBM24 in controlling SOX2 is conserved during development in vertebrates. Interestingly, Sox2 mRNA 3’UTR has three AREs (henceforth referred as A1, A2 and A3) (Fig. 8A), which is significant because RBM24 can bind AREs in other mRNAs (namely, p21 and p63) in human colorectal carcinoma cells (26,27). Therefore, we hypothesized that RBM24 regulates Sox2 mRNA through the AREs and first tested the possibility that RBM24 directly binds to Sox2 ARE by performing electrophoretic mobility shift assay (EMSA). We find that the full-length RBM24 protein binds, in a sequence-specific manner, to a 20 bp biotin-labeled RNA oligomer that contains the Sox2 3’UTR ARE site A1, A2 and A3 (Fig. 8B-D). The RBM24 protein–Sox2 3’UTR RNA complex can be competed out with a 1000-fold excess of the non-biotinylated wild-type ARE oligomer, but not with the non-biotinylated mutant ARE oligomer (Fig. 8B-D). Furthermore, we performed RIP assay that shows enrichment of Sox2 mRNA in the Rbm24 antibody pulldown (Fig. 8E). Together, these findings demonstrate that Rbm24 directly binds to Sox2 mRNA through interaction with its 3’UTR ARE.
RBM24-deficient mouse and zebrafish exhibit downregulation of SOX2 expression. (A-B’) At early eye development stage E9.5, immunostaining shows that Rbm24−/− mice exhibit significant reduction of SOX2 protein in the surface ectoderm (future lens) and the optic vesicle (future retina). (C-D’) By E14.5, even the Rbm24−/− mice with less severe eye phenotype exhibit severe reduction of SOX2 protein in the lens and retina. High-magnification of dotted line areas in A, B, C, D are shown in A’, B’, C’, D’, respectively. (E) RT-qPCR analysis confirms significant Sox2 downregulation in Rbm24−/− ocular tissue and (F) rbm24a morphant zebrafish. Abbr.: se, surface ectoderm; ov, optic vesicle; e, anterior epithelium of the lens; f, lens fiber cells; f.c., fold change. Asterisks in E and F indicate P-value < 0.05. Scale bar in A, B, C’, D’ is 70 μm, C-D is 140 μm, A’, B’ is 12 μm; # indicates whole embryonic tissue.
RBM24 regulates the half-life of Sox2 mRNA. (A) Schematic representation of human and mouse SOX2 mRNA, which has a single exon and three AREs (A1, A2 and A3) in the 3’UTR. Relative positions of the human and mouse ARE sites are indicated. (B-D) RNA EMSA demonstrates that RBM24 protein binds to Sox2 3’UTR by directly interacting with its AU-rich element (A1-A3) to form a protein–RNA complex that can be competed away by un-labeled wild-type ARE containing oligomer but not by un-labeled mutant ARE. (E) RNA immunoprecipitation with RBM24 antibody on mouse embryonic eye tissue shows enrichment of Sox2 mRNA. (F) Rbm24 KD in NIH3T3 cells results in downregulation of Sox2 transcript. (G) RBM24 overexpression in NIH3T3 cells results in significant increase in the half-life of luciferase mRNA fused to Sox2 3’UTR. (H-I) Upon RBM24 overexpression and actinomycin D treatment, luciferase reporter fused to wild-type Sox2 3’UTR mRNA exhibits elevated stability. However, under these conditions, the luciferase reporters fused to Sox2 3’UTR mRNA containing individual mutations in the AREs A1, A2 and A3 (*) or combined mutations in all three AREs (A1*2*3*) do not exhibit elevated stability. Luciferase mRNA levels after 2 h of actinomycin D treatment is shown as bar graph in I. Abbr.: IP, immunoprecipitate; ActD, actinomycin D; f.c., fold change. Asterisk in C indicates P-value < 0.05.
RBM24 positively controls Sox2 mRNA stability
To understand the biological significance of RBM24 binding with Sox2 mRNA, and thereby investigate the molecular mechanism by which it regulates SOX2 expression, we next performed reporter analysis. First, we confirmed that in NIH3T3 fibroblast cells, Rbm24 knockdown using siRNA results in downregulation of Sox2 mRNA similar to that of Rbm24−/− ocular tissue (Fig. 8F). We co-transfected NIH3T3 cells with an Rbm24 overexpression vector and a Sox2 reporter vector, which carries the Renilla luciferase open reading frame (ORF) fused to Sox2 3’UTR. After treating co-transfected cells with actinomycin D, the rate of luciferase mRNA decay was measured. We find that RBM24 overexpression causes an increase in the half-life of luciferase–Sox2 3’UTR fusion reporter transcript (Fig. 8G), indicating that RBM24 stabilizes Sox2 transcript through interactions with its 3’UTR. Next, we performed mRNA decay assay on RBM24 overexpressing NIH3T3 cells co-transfected with reporter vector wherein luciferase is fused with various mutated versions of Sox2 3’UTR. Unlike wild-type Sox2 3’UTR that contains intact ARE sites, mutation of the ARE sites A1, A2 or A3 individually, or all three sites concomitantly, did not result in a similar increase in the half-life of luciferase-Sox2 reporter vector (Fig. 8H and I). These findings demonstrate that Rbm24 directly controls Sox2 mRNA stability, which requires the presence of all three ARE sites in the Sox2 3’UTR.
Discussion
While signaling, transcriptional, epigenetic and non-coding RNA-mediated control of gene expression is well understood, the significance of RBP-mediated post-transcriptional control mechanisms in vertebrate organ development remains less clear. This represents a substantial knowledge gap considering that (1) RBPs are involved in every aspect of mRNA regulation such as capping, intracellular localization, stability/decay or translation (43), (2) there are similar number of RBPs encoded by the human genome as transcription factors (14–17) and (3) RBP function in RNA processing is increasingly recognized as a critical factor for understanding human disease phenotypes (15,44,45). Thus, our findings in this report have the following novelty and broad impact: (1) identification of SOX2 and LHX2 as a novel target of RBM24, (2) identification of an ARE-based mechanism for positively controlling Sox2 mRNA stability, (3) identification of RBM24 as the pioneering example of a conserved RBP that mediates post-transcriptional gene expression control necessary in vertebrate early eye development and (4) Rbm24 as a new gene linked to anophthalmia and microphthalmia eye defects. These data present a new model for defining the function of RBM24 in eye development (Fig. 9).
Model for RBM24-mediated post-transcriptional gene expression control in the early eye development. In normal eye development, RBM24 is required for positively regulating several key transcription factors in both the optic vesicle and the lens placode lineage. In the optic vesicle cells, RBM24 positively regulates LHX2 and SOX2 that in turn regulate the gene regulatory network of optic vesicle. In the lens placode, RBM24 regulates PAX6 and SOX2, thus regulating the gene regulatory network of the lens placode. Further it regulates JAG1 and BIRC2, which along with SOX2 regulates cell survival in both the lens and the retinal lineages.
Previously, RBM24 has been linked to cardiac defects in vertebrates (24,25,46–49). In zebrafish, rbm24a and rbm24b knockdown results in craniofacial defects and microcephaly. While these studies also cursorily mention microphthalmia, the mechanism of these pathologies or the universal requirement of RBM24 in vertebrate eye development remained undefined. We show that in addition to the expression defects in key transcription factors, Rbm24−/− mice exhibit reduced Birc2 and JAG1, which offers an explanation for the elevated apoptosis in these knockout mice as JAG1 downregulation is linked to increased apoptosis (40). Further, reduced SOX2 levels in Rbm24−/− mice also likely contribute to the increased cell death observed in the developing eye tissue, as SOX2 has been shown to negatively regulate apoptosis through the MAP4K4/survivin pathway in human lung cancer cells (50). In addition, misexpression of Pax6 in Rbm24−/− mice could result in downregulation of Pitx3 in the anterior region of the lens vesicle, which may impact apoptosis (51,52).
In only one other context, in human mesenchymal stem cells, another RBP—HuR—has been shown to control SOX2 mRNA decay via the ARE sites, but even here, a direct interaction of the RBP–mRNA by EMSA is not described (53). But more importantly, the outcome of this HuR-based control results in the negative regulation of the Sox2 mRNA. In contrast, here we give the first evidence that RBM24 directly binds to the ARE sites of Sox2 mRNA 3’UTR and stabilizes it, perhaps by preventing the recruitment of RISC RNA decay complex to the mRNA by other RBPs such as HuR and thus protecting it from degradation. These findings inform on the complexity of combinatorial control mediated by distinct RBPs on the Sox2 mRNA and more generally on their exquisite specificity for key regulatory factors in development and disease. Moreover, the post-transcriptional control of SOX2 may represent an important regulatory mechanism as this gene is among the rare 3% genes encoded by just one exon (54) and thus the Sox2 transcript—devoid of binding with proteins such as the exon–junction complex—may have evolved distinct mechanisms for controlling mRNA stability and half-life. Thus, we present a new molecular mechanism of RBM24-mediated post-transcriptional control of gene expression of the Sox2 mRNA, which is required for proper development of the eye, and deficiency of which causes the ocular developmental defects anophthalmia and microphthalmia. Further, our new evidence of direct interactions between RBM24 and the ARE and the outcome of this interaction for the mRNA have broader impact, as 8% of mRNAs encoded by the human genome contain AREs in their 3’UTRs (55). Thus, these data point to new directions for investigating the relationship of Rbm24 with its target mRNAs in other cell and tissue types.
Finally, taken in the context of our recent discovery that the Tudor domain protein TDRD7 is linked to cataracts in human, mouse and chicken (29) and CELF1 is linked to cataract and/or lens defects in mouse, frog and zebrafish (31), these new findings on RBM24 firmly establish that post-transcriptional regulatory networks—like signaling and transcription networks—have evolved conserved functions to control eye development in vertebrates.
Materials and Methods
Animal husbandry
Mice used in this study were housed in the animal facility at the University of Delaware. The Institutional Animal Care and Use Committee (IACUC) at the University of Delaware approved animal experimental protocols, and all animal experiments were conducted in accordance with the Association for Research in Vision and Ophthalmology (ARVO) statement for the use of animals in ophthalmic and vision research (AUP#1226).
Generation of Rbm24-targeted deletion mouse mutants
A new conditional knockout mouse model was generated in this study wherein exon 1, 2 and 3 of Rbm24 is flanked by loxP sites (Rbm24flox/flox) (Genoway, Lyon, France). Rbm24flox/flox mice were crossed with B6-Tg(CMV-Cre)1Cgn mice carrying the Cre recombinase gene under the influence of the CMV promoter to generate constitutive knockout alleles of Rbm24 (Rbm24−/−). Embryos were staged by designating the day of observation of the vaginal plug as E0.5. Genotyping was performed on tail DNA prepared from post-natal or embryonic tissue using the Direct PCR Lysis reagent (Viagen Biotech, Los Angeles, CA). Animals were genotyped by using the following primers. Rbm24 germline deletion allele is amplified by primers: 5’-CAAGGACAGCCTGGGATACACAATACC-3′ and 5’-GCTATGTCCATCTTGGTTCAGGATTCAG-3′.
Zebrafish maintenance
All work involving zebrafish was approved by the University of Iowa’s IACUC, PHS Assurance No. A3021-01, under animal protocol No. 5091513. Zebrafish embryos were raised between 28.5 and 30°C with no more than 50 embryos per 100 mm plate.
Knockdown of rbm24a in zebrafish by morpholinos
Zebrafish embryos collected from natural mating were pressure-injected at the 1–4 cell stage with 1–3 nanogram/embryo of translation blocking morpholino with the following sequence: rbm24a, 5′-GCATCCTCACGAAACGCTCAAGTGC-3′ (Gene Tools LLC, Philomath, OR). Standard control MO (5′-CCTCTTACCTCAGTTACAATTTATA-3′) was injected to generate ‘control’ zebrafish at an equal quantity as the experimental MO for each experiment. Microinjection volume was measured in a 1 μl capillary tube and calculated using the Microinjection Calculator Android app (available from Google Play Store).
CRISPR/Cas9 gene targeting
A vector containing the coding sequence for Cas9 protein flanked by two nuclear localization signals (nls) (pT3TS) was used to generate nls-Cas9-nls mRNA (56). Two sgRNAs were designed and constructed to target 20 bp sequences starting with guanine and preceding the PAM motif (5′-NGG-3′) to create a 252 bp deletion in rbm24a (57). Potential off-target effects of sgRNA candidates were analyzed using the online tool CRISPR Design developed by the Zhang laboratory (http://crispr.mit.edu/). An injection mixture was made containing 300 ng/μl Cas9 mRNA, 12.5 ng/μl of each guide RNA 5’-TAATACGACTCACTATAGGGGGAGATTGAAGAAGCTGGTTTTAGAGCTAGAA-3′ and 5’-TAATACGACTCACTATAGGCCTTCATGCAACCAAGTGGTTTTAGAGCTAGAA-3′.
ISH
Total RNA was isolated from 20 2-dpf zebrafish embryos. This total RNA was reverse transcribed into cDNA using SMART MMLV Reverse Transcriptase (Clontech) primed with oligo-dT primers. cDNA obtained was then used to clone a portion of rbm24a and rbm24b using the following primers: rbm24a-Forward-5’-CCAGGGGTTATGGATTTGTG-3′, rbm24a-Reverse-5’-TGCAGTTGTTGGGGTTGATA-3′, rbm24b-Forward-5’-CGGAGGTCTTCCCTATCACA-3′ and rbm24b-Reverse-5’-CCAAACGCACACAAGAGCTA-3′. The products were ligated into a TOPO-TA PCR-II vector (Thermo Fisher Scientific, # K461020) following the manufacturers protocol. DIG-labeled RNA probes (DIG labeling Mix, Roche, #11277073910) were synthesized using T7 and SP6 Maxiscript kits (Thermo Fisher Scientific, #AM1320) following the manufacturers protocol. Embryos at 20 hpf and 2 dpf were fixed with 4% PFA, and ISH was performed as described previously (58,59). Post ISH, embryos underwent cryoprotection by sequential transfer from 1X PBST, to 15, 30% sucrose, then to OCT (optimal cutting temperature) embedding medium overnight. The following day, embryos were placed in blocks containing fresh OCT, aligned as desired and frozen. 8 μm sections were collected and imaged.
Western blot
Whole-body tissue from control (Rbm24+/+) and Rbm24−/− mice at E14.5 was collected and lysed in RIPA buffer on ice. Western blot was performed as described previously (30) and probed with RBM24 antibody (diluted 1:200 in 5% milk in TBST) (Abcam, #ab174919) or beta-actin antibody (diluted 1:500 in 5% milk in TBST) (Abcam, #ab8227).
Immunofluorescence
Mouse head tissue from control (Rbm24+/+) and Rbm24−/− mice at developmental stages E9.5, E10.5, E12.5 and E14.5 were flash frozen in OCT (Tissuetech, # 14-373-65). Frozen sections (16 μm thickness) were fixed with 4% PFA for 15 min at room temperature followed by three 1X PBS washes for 10 min each. The sections were then blocked in either 5% chicken serum (Abcam, # ab7477) or 10% BSA (Sigma-Aldrich, A2153) along with 0.3% Triton X-100 for 1 h at room temperature. NIH3T3 cells were fixed at 70% confluency with methanol for 10 min. The following primary antibodies were used in the given dilutions in the respective blocking buffers: RBM24 (Abcam, #ab94567, 1:100), SOX2 (EMD Millipore, #AB5603, 1:100), LHX2 (Santa Cruz Biotechnology, #sc-19 344, 1:100), SOX2 (Thermo Fisher Scientific, #48-1400, 1:200), OCT4 (Thermo Fisher Scientific, #PA5-27438, 1:200), PAX6 (EMD Millipore, #AB2237, 1:200), JAG1 (Santa Cruz Biotechnology, #sc8303, 1:50), E-cad (Cell Signaling Technology, #3195S, 1:100) and gamma-crystallin (Santa Cruz Biotechnology, # sc-22 415, 1:100). After 1-h blocking, the sections were incubated with the primary antibody overnight at 4°C. Slides were washed and incubated with the appropriate secondary antibody conjugated to Alexa Fluor 488 or 594 (Life Technologies, 1:200) and the nuclear stain DAPI (Thermo Fisher Scientific, #D21490, 1:1000). Slides were imaged using the Zeiss LSM 780 confocal microscope (Carl Zeiss Inc.).
Dark-field microscopy
Control (Rbm24+/+) and Rbm24−/− mouse embryos at stage E11.5 were dissected in 1X PBS and imaged by light microscopy (Zeiss Stemi SV dissecting microscope).
RNA isolation and RT-qPCR in mouse
Rbm24−/− and Rbm24+/+ ocular tissue at E14.5 was microdissected in 1X PBS for RNA isolation and flash frozen on dry ice and stored in −80°C until RNA isolation. Four E14.5 eyes were treated as one biological replicate, and total RNA was extracted from three biological replicates. Total RNA was extracted using RNeasy Mini Kit (Qiagen, #74104) with no modifications. On-column DNase treatment was performed using RNase-Free DNase (Qiagen, # 79254). RNA concentration was determined by using ND-1000 UV/Vis Spectrophotometer (NanoDrop Technologies; Software V3.8.1), and RNA was stored at −80°C until ready to use. cDNA synthesis was performed with 200 ng of RNA per reaction using manufacturer’s instructions, and RT-qPCR was performed as described (32) on ABI7300 Real-Time PCR system (Applied Biosystems) using Fast SYBR Green PCR master mix by the investigator’s laboratory (Invitrogen Life Technologies, # 4367659). Hprt was used as a control housekeeping gene to normalize transcript levels. The experiment was performed three times with three biological replicates. Differential expression of each transcript was determined using 2-ΔΔCT method. Statistical significance for RT-qPCR data was determined using nested ANOVA (60). The following primers were used for qRT-PCR, Rbm24-Forward-5′-TCTTCGGAGACATCGAGGAA-3′ and Rbm24-Reverse-5′- AAAACCTGGCTGCATGATTC-3′, amplicon length, 200; Sox2-Forward-5′-TTAACGCAAAAACCGTGATG-3′ and Sox2-Reverse-5′-GAAGCGCCTAACGTACCACT-3′, amplicon length, 194; Jag1-Forward-5′-TAGCTGCCTGCCGAACCCCT-3′ and Jag1-Reverse-5′-GCTGGAGGCTGGAGGACCGA-3′, amplicon length, 590; and Pax6-Forward-5′- AGTTCTTCGCAACCTGGCTA-3′ and Pax6-Reverse-5′-ACTTGGACGGGAACTGACAC-3′, amplicon length, 846.
RNA isolation and RT-qPCR in zebrafish
RNA was isolated from 20 control (injected with control morpholino) and 20 rbm24a knockdown 1 dpf zebrafish embryos per biological replicate and used to synthesize cDNA as described above. Samples were prepared in a 96-well reaction plate with four biological replicates. qRT-PCR was performed with the LightCycler 480 System (Roche Applied Science) using SYBR Green I master mix (Roche, # 04707516001). Samples were normalized to elongation factor 1 alpha (elfa). Log fold change was determined using 2-ΔΔCT, method and statistical significance was calculated using nested ANOVA. The following primers were used for RT-qPCR: sox2-Forward-5’-AACCAGAAAAACAGCCCGGA-3′, sox2-Reverse-5′-AATGGTCGCTTCTCGCTCTC-3′, elfa-Forward-5′-GATGCACCACGAGTCTCTGA-3′ and elfa-Reverse-5′-TGATGACCTGAGCGTTGAAG-3′.
RNA immunoprecipitation
Wild-type E14.5 mouse eye lysates were used (n = 6 eyes per biological replicate; total three biological replicates) for pre-conjugation with Rbm24 antibody (Abcam, #ab174919) and IgG with magnetic beads. Manufacturer’s instructions were followed for the EZ-Nuclear RIP Kit (EMD Millipore, #17-10 523) for the immunoprecipitation, followed by cDNA synthesis and qRT-PCR for Sox2 and Hprt.
RNA EMSA
Full-length Rbm24 CDS fused with GST was cloned in PGEX-6P3 vector (GE Healthcare Life Sciences, #28-9546-51) and transformed in competent BL21 bacterial cells (Thermo Fisher Scientific, #C600003). Glutathione Sepharose bead (Sigma-Aldrich, #GE17-0756-01) was used to affinity purify Rbm24 protein from BL21 lysate, following which GST was cleaved from the fused protein using PreScissionTM Protease (GE Healthcare Life Sciences, #27-0843-01). RNA EMSA was performed using manufacturer’s instruction for Chemiluminescent RNA EMSA (Thermo Fisher Scientific, #20158). A1 of Sox2 3′UTR were heated at 80°C for 5 min to relax RNA folding. Following RNA oligos were used: ARE1, 5′-ATTTGAACATTTTAGTTTTA-3′-biotin, 5′-ATTTGAACATTTTAGTTTTA-3′, 5′-ATTTGAACAGGGGAGTTTTA-3′-biotin; ARE2, 5′-ATTGTGATATTTTAAGGTTT-3′-biotin, 5′-ATTGTGATATTTTAAGGTTT-3′, 5′-ATTGTGATAGGGGAAGGTTT-3′-biotin; ARE3, 5′-TAGTTGTATTTTAAAAGATT-3′-biotin, 5′-TAGTTGTATTTTAAAAGATT-3′ and 5′-TAGTTGTAGGGGAAAAGATT-3′. Rbm24 purified protein (0.5 mg/ml) was allowed to bind to the relaxed oligo for 30 min at room temperature. The samples were run on a 6% native gel polyacrylamide gel made in 0.5X TBE and then transferred to a nylon membrane. RNA was crosslinked to the membrane by irradiating the membrane with 254 nm UV lamp for 3 min. The membrane was then blocked; washed and biotinylated RNA was detected with chemiluminescent substrate provided in the kit.
TUNEL assay
The in situ cell death kit (Roche, #11684795910) was used for labeling cells undergoing apoptosis following the manufacturer instructions. The percent of TUNEL-positive nuclei was calculated as the ratio of TUNEL-positive nuclei and all nuclei multiplied by a factor of 100 (61,62). Apoptotic index was calculated from three biological replicates and three technical replicates.
Cell culture
The mouse fibroblast NIH3T3 cells and lens epithelial cell line 21EM15 were cultured under standard conditions (10 ml of DMEM with 4.5 g/l glucose and L-glutamine (Corning Cellgro, #10-017-CV)), 10% Fetal Bovine Serum (Gemini Bio Products, # 900-208) and 1% Penicillin-Streptomycin (GE Healthcare Life Sciences, # SV30010)). Cells were incubated at 37°C in a humid chamber with 5% CO2 as described previously (29,63).
Generation of Sox2 3′UTR reporter vector
To generate wild-type Sox2 reporter vector wherein Renilla luciferase is fused to 3′UTR of Sox2 mRNA, wild-type Sox2 3′UTR was cloned downstream of the firefly luciferase gene in the pmirGLO vector (Promega, #E1330) with the following primers: 5′-GACGAGCTCGGGCTGGACTGCGAACTGGAGA-3′ and 5′-AGGCTCTAGATTCAGTGTCCATATTTCAAA-3′. Mutated Sox2 reporter vector were generated following manufacturer’s instructions for QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies, #200523) using the below mentioned HPLC-purified primers: A1*-Forward-5′-TGTTTTCCTTTTGTACAATTTTAAAACTCCCCTGTTCAAATCCGAATAAACTCCTTCCTTGTTTGTAACGG-3′, A1*-Reverse-5′- CCGTTACAAACAAGGAAGGAGTTTATTCGGATTTGAACAGGGGAGTTTTAAAATTGTACAAAAGGAAAACA-3′, A2*-Forward-5′-GAAATATTTTCTTATGGTTTGTAATATTTCTGTAAATTGTGATACCCCAAGGTTTTTCCCCCCTTTTATTTTCCG-3′, A2*-Reverse-5′-CGGAAAATAAAAGGGGGGAAAAACCTTGGGGTATCACAATTTACAGAAATATTACAAACCATAAGAAAATATTTC-3′, A3*-Forward-5′-CTGATTCCAATAACAGAGCCGAATCTTTTGGGGTACAACTACGGAAAATAAAAGGGGGGAAA-3′ and A3*-Reverse-5′-TTTCCCCCCTTTTATTTTCCGTAGTTGTACCCCAAAAGATTCGGCTCTGTTATTGGAATCAG-3′. pmirGLO dual luciferase vector with wild-type 3′UTR of Sox2 vector was used as a template for site-directed mutagenesis. Cloning was confirmed by Sanger sequencing.
mRNA decay assay
To generate the RBM24 overexpression plasmid, RBM24 ORF was cloned to pcDNA3.1 vector (Thermo Fisher Scientific, # V79020) using the following primers, Forward-5′-ATTAGAATTCATGCACACCACCCAGAA-3′ and Reverse-5′- CTTCTCGAGCTACTGCATTCGGTCTGTCT-3′, and was confirmed by Sanger sequencing. For Lhx2 decay assays, RBM24 overexpression vector was transiently transfected into 21EM15 cells for 48 h using the Lipofectamine® 3000 Transfection Reagent (Thermo Fisher Scientific, #L3000015). Control cells were transfected with empty pcDNA3.1 vector. Transfected cells were treated with 5 μg/ml actinomycin D to halt general transcription. Cells were collected every 2 h up to after 10 h of actinomycin D treatment, total RNA was isolated and cDNA were synthesized, and RT-qPCRs were performed as described above. For Sox2 decay assays, both RBM24 overexpression plasmid and the dual luciferase-Sox2 3’UTR plasmid were transiently transfected into NIH3T3 cells for 48 h using the Lipofectamine® 3000 Transfection Reagent (Thermo Fisher Scientific, #L3000015). Control cells were transfected with empty pcDNA3.1 vector and dual luciferase-Sox2 3’UTR plasmid. Transfected cells were treated with 5 μg/ml actinomycin D to halt general transcription. Cells were collected every hour up to after 4 h of actinomycin D treatment, total RNA was isolated and cDNA were synthesized, and RT-qPCRs were performed as described above using the following primers to amplify the luciferase product: 5′-GCTCAGCAAGGAGGTAGGTG-3′ and 5′- TCTTACCGGTGTCCAAGTCC-3′. The assay was performed multiple times with three biological replicates and three technical replicates.
Funding
This work was supported by National Institutes of Health’s National Eye Institute [grant number R01EY029770 to S.L.]. Support from the University of Delaware Core Imaging facility was made possible through the Institutional Development Award (IDeA) from the National Institutes of Health’s National Institute of General Medical Sciences INBRE Program Grant [grant number P20GM103446]. Acquisition of the confocal microscope used in this study was funded by the National Institutes of Health/National Center for Research Resources grant [1S10 RR027273]. S.D. received a Summer Scholarship from Fight For Sight, Inc. S.L. is a Pew Scholar in the Biomedical Sciences.
Conflict of interest statement
None declared.
