TAR syndrome causal gene RBM8A is critical for embryonic bone development and proper Hedgehog signaling

Abstract

RNA-Binding Motif Protein 8A (RBM8A) is a core component of the exon junction complex (EJC), which plays a fundamental role in post-transcriptional RNA regulation. Mutations in the RBM8A gene cause thrombocytopenia-absent radius (TAR) syndrome, characterized by reduced platelet number and radial defects. However, how its function contributes to bone development remains poorly understood. We found that RBM8A is highly expressed in human bone marrow-derived mesenchymal stem cells (MSCs) and mouse embryonic limb buds. To further investigate the role of RBM8A in bone development, we generated an Rbm8a conditional knockout mouse (cKO) model in the early limb bud and craniofacial mesenchyme using the Prx1-cre line, which recapitulates key features of TAR syndrome. Rbm8a deficiency in this mouse model resulted in significant defects in limb and craniofacial development, including shorter limb bones across developmental timepoints, delayed ossification, altered cranial morphogenesis, along with impairments in motor function and reduced brain size. To elucidate the molecular basis of these phenotypes, we performed RNA immunoprecipitation sequencing (RIP-seq) in embryonic limb tissue. Our analysis revealed that RBM8A preferentially associates with long, exon-rich transcripts with high isoform complexity. These transcripts are enriched in pathways related to primary cilia, RNA processing, and developmental signaling. Consistent with this, transcriptomic and enrichment analyses identified strong associations with ciliopathy-related and congenital skeletal disease gene sets. At the cellular level, Rbm8a deficiency resulted in altered expression of key chondrogenic and ossification markers in embryonic limb bone tissue, such as reduced SOX9 and increased IHH in the primary ossification center, suggesting impaired progression of endochondral ossification and disrupted Hedgehog pathway activity. Quantitative Reverse Transcription–PCR (qRT-PCR) analysis further validated selective downregulation of Hedgehog pathway targets, including Gli3 and Hhip, while other developmental genes remained unchanged, indicating pathway-specific attenuation. Together, these findings support a model in which Rbm8a regulates structurally complex transcripts that are enriched in ciliary and developmental signaling pathways, and that disruption of this process leads to selective impairment of Hedgehog signaling and skeletal development. These results provide a novel mechanistic link between RNA processing and developmental signaling and offer new insight into how defects in RNA regulation contribute to congenital skeletal abnormalities.

Introduction

Birth defects impacting skeletal and craniofacial development are among the most common congenital diseases, representing some of the urgent unmet medical needs across the world¹. The prevalence of major structural birth defects, including skeletal and craniofacial abnormalities, is alarmingly high, affecting 3% of newborns in the US, with an annual hospitalization cost of $22.6 billion in 2013². Limb reduction defects (LRD) and craniofacial anomalies (CFA) are particularly common, affecting 1 in 1900 and 1 in 1600 babies, respectively³. Of note, congenital bone disorders represent a major group of developmental conditions that cause significant lifelong structural and functional health complications⁴. Moreover, these conditions are often associated with nervous system defects, such as autism spectrum disorders (ASD) and intellectual disability (ID)^5–8. Therefore, there is a significant medical need for individuals affected by these debilitating disorders.

These diseases are caused by genetic or environmental factors, and identifying the direct causal factor(s) for each disorder is critical to understanding fundamental mechanisms of bone development. Skeletal development is a highly coordinated process that begins during early embryogenesis, progressing through tissue patterning, mesenchymal condensation, chondrogenesis, endochondral ossification, growth plate maturation, and continuing into postnatal remodeling⁹. Because skeletal fate is established early during embryogenesis, it is uniquely sensitive to disruptions in developmental signaling¹⁰. Bone morphogenesis is precisely arranged by interconnected signaling pathways that control cellular proliferation, lineage determination, and maturation timing¹¹. Hedgehog signaling is one such major pathway, as Sonic Hedgehog (SHH) governs early limb patterning and Indian Hedgehog (IHH) controls endochondral ossification¹². As such, IHH mutations cause congenital skeletal disorders such as Brachydactyly Type A1 or Acrocapitofemoral dysplasia^13,14. Moreover, mutations in GLI3, a central regulator of Hedgehog signaling, result in congenital limb and craniofacial disorders such as Pallister–Hall syndrome or Greig cephalopolysyndactyly syndrome^15,16. Investigating the developmental mechanisms that give rise to specific bone phenotypes can provide insights into core signaling pathways that are critical during early development.

Although more genetic associations are being identified, the mechanisms by which these genes affect bone development remain poorly understood^4,11,17. One example is TAR syndrome, which is a rare congenital disorder characterized by the absence of radius in each forearm (with the presence of both thumbs) along with a deficiency in platelet counts (less than 50 platelets/nL, where the normal range is 150–400 platelets/nL)¹⁸. In addition to these anomalies, the severity of other skeletal abnormalities varies among those with this syndrome¹⁹. Craniofacial and other upper- and lower-limb bones can also be underdeveloped, affecting mobility and other motor functions involved in everyday tasks for TAR patients²⁰. Moreover, some TAR patients have hypoplasia of the cerebellar vermis and the corpus callosum²¹. This brain underdevelopment, along with the prevalence of cranial defects and intellectual disability¹⁹, suggests there may be distinct changes in the crania and brains of affected TAR patients.

The causal factors of TAR syndrome are compound mutations of a 1q21.1 null deletion and rare single-nucleotide polymorphisms in the regulatory region of RBM8A²². RBM8A, also known as Y14, is a core EJC factor, which is critical for splicing, translation, and mRNA degradation²³. EJC plays a major role in nonsense-mediated mRNA decay (NMD), which controls RNA stability by tagging mRNAs with premature termination codons for decay²³. RBM8A has previously been reported to regulate cell cycle progression and is implicated in cancer, neurodegenerative diseases, and neurodevelopmental disorders²⁴. Previous works from our group^25–28 and others²⁹ have demonstrated the critical role of RBM8A in brain development. However, the role of RBM8A in bone development and the mechanisms by which its mutations cause the specific phenotypic anomalies in TAR syndrome remain unclear³⁰. Understanding how RBM8A gene mediates embryonic bone development is critical not only for advancing knowledge of congenital skeletal disorders, but also for identifying potential therapeutic targets and uncovering new roles these genes may play in key developmental signaling pathways.

To address this knowledge gap, we present a novel Rbm8a cKO mouse model, which resembles the TAR phenotype. We demonstrate that these mice exhibit structural skeletal defects across developmental timepoints, functional impairments, and changes in the brain. Importantly, we identify the altered Hedgehog signaling pathway as the mechanism by which loss of RBM8A leads to limb maldevelopment in TAR syndrome. These findings will serve as a basis for understanding the novel role of RBM8A in skeletal development. Insights into RBM8A’s functions in long bone and cranial development, and its implication in brain development, will be the first critical step in therapeutic development for this disorder.

Results

Expression of RBM8A in developing bones

As RBM8A deficiency leads to TAR syndrome with well-defined bone defects, we first examined the RBM8A level in bone-related cells and tissues. MSCs, characterized by surface markers such as CD90, can differentiate into osteogenic cell types³¹. We cultured human bone marrow-derived MSCs and found that RBM8A is highly expressed in CD90-positive cells (Fig. 1A). Moreover, to determine whether RBM8A is expressed in the early embryonic limb bud, we performed co-immunostaining for RBM8A and SOX9, a key regulator of pre-chondrogenic specification³², in embryonic day 10.5 (E10.5) mouse sections. We found that RBM8A is ubiquitously expressed in most cells, with higher expression in the limb bud (Fig. 1B). Interestingly, RBM8A expression is decreased in SOX9-positive cells compared with SOX9-negative neural crest cells. We further examined RBM8A and SOX9 expression in both embryonic autopods and zeugopods (radius and ulna) at E14.5 (Fig. 1C–D). SOX9-positive chondroprogenitors and chondrocytes are localized in the digital cartilage condensations within the autopod but are absent in the interdigital mesenchymal cells (Fig. 1C). In contrast, RBM8A expression is higher in the interdigital mesenchymal cells that are SOX9-negative but lower in SOX9-positive cells. Consistently, in both the radius and ulna, RBM8A expression is higher in SOX9-negative hypertrophic chondrocytes in the center of the diaphysis than in SOX9-positive chondrocytes at the sides of the diaphysis (Fig. 1D), suggesting an inverse relationship between SOX9 and RBM8A. These data collectively suggest that RBM8A could play an essential role in early forelimb development.

Figure 1. RBM8A is expressed in human bone marrow-derived mesenchymal stem cells and mouse embryonic skeletal tissues.

A. Immunofluorescence staining of cultured human bone marrow-derived mesenchymal stem cells showing CD90 (green) and RBM8A (red). Merged image (right) demonstrates RBM8A localization in CD90-positive cells. Scale bar = 50 μm.

B. Whole-mount immunofluorescence staining of E10.5 mouse embryo section showing RBM8A (red), SOX9 (green), and merged image with DAPI (blue). RBM8A is broadly expressed, with high expression in the limb bud. In the neural crest, RBM8A expression is decreased in SOX9-positive regions. Scale bar = 200 μm.

C. Whole-mount immunofluorescence staining of E14.5 mouse forelimb autopod sections showing RBM8A (red), SOX9 (green), and merged image with DAPI (blue). RBM8A is broadly distributed throughout the autopod, including SOX9-positive cartilage condensation. Scale bar = 200 μm.

D. Whole-mount immunofluorescence staining of E14.5 mouse forelimb zeugopod longitudinal tissue section showing RBM8A (red), SOX9 (green), and merged image with DAPI (blue). RBM8A is detected in and around SOX9-positive skeletal condensations, including the central region of the developing cartilage element. Scale bar = 200 μm.

Generation of Prx1-Rbm8a^f/+ mice

To study the role of RBM8A in embryonic limb development, we generated an Rbm8a limb cKO mouse model. We used Prx1-Cre mice to drive the expression of Cre recombinase throughout the early limb bud mesenchyme and a subset of craniofacial mesenchyme³³ (Fig. 2A; Supplementary Fig. 1A). The resulting Prx1-Rbm8a^f/+ mice have noticeably shorter forelimbs and forepaws, as well as significantly smaller overall body length compared to their littermate controls (Fig. 2B). Quantification of Alizarin Red and Alcian Blue staining revealed significantly shorter digit, radius, and humerus lengths in Prx1-Rbm8a^f/+ mice compared to their littermate controls (Fig. 2C, 2D). This phenotype is observable as early as E15 (Fig. 2E), in which the Prx1-Rbm8a^f/+ embryos show notably shorter limbs and altered cranial shape. Forelimb and cranial Ai14 reporter signal intensities of optically cleared Prx1-Rbm8a ^f/+; Ai14 ^f/+ E15 embryos were significantly reduced than those of Prx1; Ai14 ^f/+ (control) (Supplementary Fig. 1). Interestingly, we observed embryonic lethality in homozygous deletion Prx1-Rbm8a^f/f mice. At E11, 42.9% of Prx1-Rbm8a^f/f embryos were alive. The survival rate decreased sharply to 16.7% at E12, then to 11.1% at E13, eventually reaching 0% from E14 onward (Fig. 2F). This indicates that Rbm8a is required for the Prx1-cre mesenchymal lineage during early development. In all, our Prx1-Rbm8a^f/+ mouse model captures the limb and craniofacial malformation observed in patients with TAR syndrome.

Figure 2. Prx1-Rbm8a^f/+ mice show limb and craniofacial malformation.

A. Schematic of the breeding strategy used to generate Prx1-Rbm8a^f/+ mice. Rbm8a^f/f mice were crossed with Prx1-Cre mice to generate Prx1-Cre; Rbm8a^f/+ offspring, which have selective deletion of Rbm8a in limb bud mesenchyme and some craniofacial mesenchyme. Created with BioRender.com.

B. Overall phenotype of Prx1-Rbm8a^f/+ mice (right) compared to sex-matched littermate control (left). Prx1-Rbm8a^f/+ mice have substantially shorter forelimbs and forepaws, and significantly smaller overall body length. Mean ± SEM (n=6 per group). Unpaired t-test. **, p < 0.01.

C. Representative images of Alizarin Red and Alcian Blue staining of forelimb bones (left: forepaws, center: radius and ulna, right: humerus) of adult Prx1-Rbm8a^f/+ mice (bottom) and sex-matched littermate control (top).

D. Quantification and statistical analysis of Alizarin Red and Alcian Blue staining. Left: all digits measured (II to V) of Prx1-Rbm8a^f/+ mice were significantly shorter than those of their littermate controls. Middle: radii of Prx1-Rbm8a^f/+ mice were significantly shorter than those of their littermate controls. Right: humeri of Prx1-Rbm8a^f/+ mice were significantly shorter than those of their littermate controls. Mean ± SEM (n=4 per group). Unpaired t-test. ***, p < 0.001.

E. Comparison of Prx1-Rbm8a^f/+ (right) and control (left) E15 embryos collected from the same litter. The Prx1-Rbm8a^f/+ embryo shows noticeably smaller body, shorter limbs, and an altered craniofacial region compared to the control.

F. Survival curve of homozygous deletion Prx1-Rbm8a^f/f embryos. At E11, 42.9% of Prx1-Rbm8a^f/f embryos were viable, decreasing to 16.7% at E12 and 11.1% at E13, with no viable embryos detected from E14 onward.

Prx1-Rbm8a ^f/+ mice show significant bone defects across developmental timepoints

To further investigate the alterations in bone development, we utilized micro-computed tomography (micro-CT) to measure the lengths of the limb bones in Prx1-Rbm8a^f/+ mice and control littermates at postnatal days 5, 15, and 30 (P5, P15, and P30). The stark difference in bone phenotype can be observed via micro-CT as early as at P5 and becomes increasingly noticeable as they develop (Fig. 3). At P5, control mice show expected early postnatal mineralization^34,35, whereas Prx1-Rbm8a^f/+ mice already show shorter long bones and delayed cranial growth (Fig. 3A). At P15, which is an active skeletal growth phase, control mice demonstrate an overall increase in bone length and stronger cranial mineralization than at P5³⁴ (Fig. 3B). Contrastingly, Prx1-Rbm8a^f/+ mice continue to display shortened long bones and delayed cranial maturation (Fig. 3B). At P30, control mice show expected near-mature skeletal growth^34,36, but Prx1-Rbm8a^f/+ mice have persistently reduced overall bone length, indicating long-term structural developmental consequences (Fig. 3C). When quantified, we saw statistically significantly shorter lengths of radius, ulna, humerus, tibia, and femur bones in the Prx1-Rbm8a^f/+ mice compared to their littermate controls across all three ages (Fig. 3D–3H). Interestingly, the radius and ulna lengths of Prx1-Rbm8a^f/+ mice appear to halt between P15 and P30 (Fig. 3D, 3E), suggesting that there is a defect in postnatal longitudinal bone growth between these developmental timepoints.

Figure 3. Micro-CT of Prx1-Rbm8a^f/+ mice shows significant bone defects across developmental timepoints.

A-C. Representative whole body images of micro-CT scans of (A) P5, (B) P15, (C) P30 Prx1-Rbm8a^f/+ mice (right) and control mice (left). Scale bar = 7mm.

D-H. Quantification and statistical analyses of (D) radius, (E) ulna, (F) humerus, (G) tibia, (H) femur lengths of Prx1-Rbm8a^f/+ mice (red) and control mice (black) across developmental timepoints (P5, P15, P30). Mean ± SEM (n=8 per group). Unpaired t-test. ****, p < 0.0001.

I-K. Representative micro-CT scan image of crania of (I) P5, (J) P15, (K) P30 Prx1-Rbm8a^f/+ mice (right) and control mice (left). Scale bar: 6mm.

L. Cranial landmarks used for quantification. NW: nasal bone width, NL: nasal bone length, LR: anterolateral distance, FW: frontal bone width, FL: frontal bone length, PL: parietal bone length, IW: interparietal bone width, IL: interparietal bone length.

M-S. Quantification and statistical analyses of (M) nasal bone widths, (N) nasal bone lengths, (O) orbital socket area, (P) parietal bone length, (Q) interparietal bone width, (R) interparietal bone width, (S) sagittal cranial length of Prx1-Rbm8a^f/+ mice (red) and control mice (black) across developmental timepoints (P5, P15, P30). Mean ± SEM (n=3 per group). Unpaired t-test. **, p < 0.01.

T. Representative micro-CT scan image of first molars (left) and quantification and statistical analyses (right) of first molar length of Prx1-Rbm8a^f/+ mice and control mice. Mean ± SEM (n=7). Unpaired t-test. *, p < 0.05.

U. Representative micro-CT scan image of incisors (left) and quantification and statistical analyses (right) of incisor widths of Prx1-Rbm8a^f/+ mice and control mice. Mean ± SEM (n=7). Unpaired t-test. *, p < 0.05.

In addition to the limb bones, we also scanned the crania of Prx1-Rbm8a^f/+ mice and control littermates at P5, P15, and P30 (Fig. 3I–3K). At P5, all mice show partially mineralized frontal and parietal bones and open sutures³⁷, but Prx1-Rbm8a^f/+ mice show more drastic underdevelopment of nasal, frontal, parietal, and interparietal bones compared to those of littermate mice (Fig. 3I). At P15, the control mice show expected increased mineralization of frontal and parietal bones and narrowing sutures than at P5³⁷ (Fig. 3J). Conversely, the crania of Prx1-Rbm8a^f/+ mice continue to show less mineralization and underdevelopment of sutures, especially in the frontal and parietal region, compared to those of littermate mice (Fig. 3J). At P30, the control mice show structurally adult-like crania³⁷, but Prx1-Rbm8a^f/+ mice persistently display reduced mineralization, as well as crooked sutures compared to those of littermate mice (Fig. 3K). This misalignment may indicate the loss of coordination and balance between osteoblast and suture mesenchyme activity^38,39.

As patients with TAR syndrome are often present with abnormal head size⁴⁰, we next measured specific cranial bones of Prx1-Rbm8a^f/+ mice and control littermates at P5, P15, and P30 using established cranial landmarks^37,41,42 (Fig. 3L). Interestingly, Prx1-Rbm8a^f/+ mice had significantly wider nasal bones than littermates at P5, but showed no statistical difference at P15, and then showed a trend toward being narrower at P30 (Fig. 3M). Similarly, they had significantly shorter nasal bones at P5, showed no statistical difference at P15, and exhibited a trend toward being slightly longer at P30 (Fig. 3N). These results seem to suggest a dynamic bone growth event that is defective in Prx1-Rbm8a^f/+ mice. Orbital socket area, interparietal bone width, and sagittal cranial length were significantly smaller in Prx1-Rbm8a^f/+ mice across all three ages (Fig. 3O–3Q; Supplementary Fig. 2E, 2F). Notably, Prx1-Rbm8a^f/+ mice had significantly longer parietal bones than littermates at P5 and P15 but had no statistical difference at P30 (Fig. 3R). These mice also had significantly shorter interparietal bones than their littermates at P5, but there was no statistical difference at P15 or P30 (Fig. 3S). Additionally, their frontal bones were significantly narrower than those of their littermates at P30 (Supplementary Fig. 2A). Intriguingly, these bones were significantly shorter than those of their littermates at P5, but caught up by P15, only to be significantly shorter comparatively again at P30 (Supplementary Fig. 2B). The distance between the anterolateral corners of the frontal bone and the height of cranial cavity between bregma and intersphenoidal synchondrosis were comparable between Prx1-Rbm8a^f/+ and control mice across all three timepoints (Supplementary Fig. 2C, 2D). To further examine the craniofacial differences, we also measured the teeth of these mice⁴³. Interestingly, Prx1-Rbm8a^f/+ mice had significantly shorter first molars and significantly narrower incisors (with no difference in length) compared to littermate controls (Fig. 3T, 3U; Supplementary Fig. 2G). Taken together, these measurements reflect an altered craniofacial development in Prx1-Rbm8a^f/+ mice.

Prx1-Rbm8a cKO mice show significant motor function deficits and potentially altered psychiatric states

Building on our observations of the striking structural phenotype, we wanted to test whether the structural limb defects also affect functional abilities in Prx1-Rbm8a^f/+ mice. First, we used a grip-strength meter to measure the maximum force exerted by the forelimbs of the mice^44,45. Prx1-Rbm8a^f/+ mice showed significantly weaker grip strength (normalized by body weight) compared to control mice (Fig. 4A). Next, we used a slightly elevated apparatus with a wire grid to test the number of paw slips through the wire^46,47. Prx1-Rbm8a^f/+ mice had a significantly higher number of foot slips compared to their counterparts, indicating a motor coordination deficit (Fig. 4B). Similarly, Prx1-Rbm8a^f/+ mice fell from the rotarod significantly faster than control mice across all three days of testing (Fig. 4C). Notably, both control and Prx1-Rbm8a^f/+ mice took significantly longer to fall across the days, indicating that motor skill learning and memory were not affected^48,49 (Fig. 4C). Additionally, Prx1-Rbm8a^f/+ mice buried significantly fewer marbles than control mice in the marble burying test (Fig. 4D), further suggesting fine motor impairment⁵⁰. In the open field test⁵¹, control and Prx1-Rbm8a^f/+ mice traveled similar total distances, but Prx1-Rbm8a^f/+ mice spent significantly less time not moving than control mice (Fig. 4E). This is interesting because it suggests that Prx1-Rbm8a^f/+ mice exhibited higher locomotor activity despite their shortened limb phenotype and apparent motor impairment. This can be interpreted as Prx1-Rbm8a^f/+ mice exhibiting hyperactivity or heightened arousal⁵². In addition, Prx1-Rbm8a^f/+ mice spent significantly more time in the border than control mice (Fig. 4E), which can also suggest increased anxiety-like behavior⁵². There was no significant difference observed in the average velocity, latency to the center, or frequency in the center (Supplementary Fig. 3A). In the elevated plus maze⁵³, Prx1-Rbm8a^f/+ mice spent significantly less time not moving than control mice, although there was no statistically significant difference in the total distance traveled, time spent in or latency to open arms, frequency in open arms, or average velocity (Fig. 4F; Supplementary Fig. 3B). Interestingly, in the nesting test, Prx1-Rbm8a^f/+ mice had significantly lower nesting scores at earlier timepoints but eventually caught up by later timepoints, indicating delayed nest-building behavior compared to controls⁵⁴ (Supplementary Fig. 3D). Results from the forced swimming test⁵⁵ and cylinder test^56,57 also did not show significant differences between control and Prx1-Rbm8a^f/+ mice (Supplementary Fig. 3C, 3E). These results indicate that there are potential differences in the brains of Prx1-Rbm8a^f/+ mice that affect anxiety levels in novel environments (open field⁵² and elevated plus maze⁵⁸) while still having normal coping behavior to acute stress (forced swimming⁵⁹).

Figure 4. Prx1-Rbm8a^f/+ mice show significant motor function deficits and potentially altered psychiatric states.

A. Grip strength test to measure the maximum force exerted by the forelimbs of the mice; normalized by body weight. Prx1-Rbm8a^f/+ mice show significantly weaker grip strength compared to control mice. Mean ± SEM (n=30 total; control: n=18 (9 males, 9 females); mutant: n=12 (5 males, 7 females)). Unpaired t-test. ****, p < 0.0001. Created with BioRender.com.

B. Grid walking test to measure motor coordination function. Mean ± SEM (n=30 total; control: n=18 (9 males, 9 females); mutant: n=12 (5 males, 7 females)). Unpaired t-test. ****, p < 0.0001. Created with BioRender.com.

C. Rotarod test to measure motor coordination and motor learning across three days. Both control and Prx1-Rbm8a^f/+ mice took significantly longer to fall from the rotarod by day 3 compared to day 1. Mean ± SEM (n=30 total; control: n=18 (9 males, 9 females); mutant: n=12 (5 males, 7 females)). Unpaired t-test. *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. Created with BioRender.com.

D. Marble burying test to measure fine motor coordination of the forepaw and forelimb. Mean ± SEM (n=30 total; control: n=18 (9 males, 9 females); mutant: n=12 (5 males, 7 females)). Unpaired t-test. **, p < 0.01. Created with BioRender.com.

E. Open field test to measure locomotor activity and exploratory behavior. Representative EthoVision-generated movement tracks. Mean ± SEM (n=30 total; control: n=18 (9 males, 9 females); mutant: n=12 (5 males, 7 females)). Unpaired t-test. ns, not significant, *, p < 0.05. Created with BioRender.com.

F. Elevated plus maze test to measure anxiety-like behavior and locomotor activity.

Representative EthoVision-generated movement tracks. Mean ± SEM (n=30 total; control: n=18 (9 males, 9 females); mutant: n=12 (5 males, 7 females)). Unpaired t-test. ns, not significant, *, p < 0.05. Created with BioRender.com.

G. Representative images of P15 Prx1-Rbm8a^f/+ (right) and control (left) mouse brains. The brains of Prx1-Rbm8a^f/+ mice are noticeably smaller and differ in shape from those of littermate controls at P15.

H. Quantification and statistical analysis of P15 brain weights of Prx1-Rbm8a^f/+ and control. Mean ± SEM (control: n=6; mutant: n=4). Unpaired t-test. **, p < 0.01.

I. Representative coronal whole-brain MRI slices and fractional anisotropy maps from P15 control and Prx1-Rbm8a^f/+ mice.

J. Quantification and statistical analyses of whole-brain volume and cortical volume from MRI analysis of P15 control and Prx1-Rbm8a^f/+ mice. Mean ± SEM (control: n=5; mutant: n=5). Unpaired t-test. ***, p < 0.001, ****, p < 0.0001.

We next investigated whether these potential altered psychiatric states are consequences of changes in the brain structure and function. Given the significantly altered cranial development in Prx1-Rbm8a^f/+ mice (Fig. 3I–3K), we hypothesized that the brains growing inside the cranium could be restricted and compressed, as in the case of craniosynostosis^60,61. Remarkably, the brains of Prx1-Rbm8a^f/+ mice were indeed noticeably smaller than those of littermate controls at P15 (Fig. 4G). They were distinctly different in size and shape and significantly lighter than controls (Fig. 4H). Magnetic resonance imaging (MRI) analysis confirmed statistically reduced whole-brain and cortical volumes in Prx1-Rbm8a^f/+ mice (Fig. 4I, 4J). Taken together, these findings suggest that the severe skeletal defects in Prx1-Rbm8a^f/+ mice not only affect motor function but also brain shape and size, thereby potentially altering neural function.

RBM8A-associated transcripts are significantly enriched in ciliary processes and developmental signaling pathways

RBM8A is an RNA-binding protein regulating multiple steps of mRNA processing, including mRNA splicing, export, degradation, and translation⁶². However, tissue-specific mRNAs that are bound to the EJC have yet to be identified. To investigate the downstream pathways controlled by RBM8A, we performed RIP-seq^63,64 using embryonic mouse limb tissue to identify RBM8A-bound transcripts and to assess how that binding relates to our observed skeletal phenotype. We used an anti-RBM8A antibody to immunoprecipitate protein-RNA complexes from E15 mouse forelimbs from wild-type mice, and isolated RNA from the co-precipitation for high-throughput sequencing. Our RIP-seq dataset identified a subset of transcripts with significant enrichment in RBM8A IP relative to input controls. A volcano plot of differential enrichment identified significantly enriched mRNAs linked to developmental regulators (Fig. 5A). To elucidate the biological functions associated with the RBM8A-associated transcripts, we performed Gene Ontology (GO) Biological Process enrichment analysis⁶⁵ on the significantly enriched genes. Enriched GO terms revealed strong representation of pathways related to cell cycle process, cilium organization and assembly, and neural tube development (Fig. 5B). Because these processes are associated with developmental and skeletal pathways, we next explored whether RBM8A-associated transcripts were enriched for genes implicated in human disease pathways using DisGeNET⁶⁶. Notably, the top-most enriched disease terms were related to intellectual disability, general developmental conditions, and craniofacial/skeletal abnormalities (Fig. 5C). Specifically, microcephaly, polydactyly, and craniofacial abnormalities are of meaningful relevance to TAR syndrome¹⁹. Additionally, enrichment of gene sets associated with ciliopathies is notable, as it is consistent with GO biological process enrichment for cilium organization and assembly (Fig. 5B). This is further supported by the enrichment of genes involved in ciliary function and Hedgehog signaling (Fig. 5D, 5E), suggesting that RBM8A-associated transcripts overlap with pathways regulating primary cilia function. To assess the consistency of pathway enrichment across multiple databases, we compared gene sets identified from GO, KEGG, and Reactome analyses, which revealed a shared subset of enriched transcripts of RBM8A in Hedgehog signaling pathways across all three databases (Supplementary Fig. 4F).

Figure 5. RIP-seq reveals enrichment of RBM8A-associated transcripts in ciliary processes and developmental signaling pathways.

A. Volcano plot showing differentially enriched transcripts associated with RBM8A, identified by RIP-seq. The x-axis indicates log₂ fold change, and the y-axis represents −log₁₀ adjusted P value (q value). Significantly enriched transcripts are highlighted based on adjusted P value and fold-change thresholds. Transcripts meeting significance thresholds (adjusted P < 0.05 and log₂|FC| ≥ 0.5) are highlighted in red. The gene names of the top 15 enriched transcripts are labeled.

B. Selected GO biological process enrichment analysis of transcripts enriched in RBM8A immunoprecipitates. Enriched terms include cell cycle process, cilium organization, cilium assembly, neural tube development, neural tube formation, and regulation of cilium assembly.

C. DisGeNET disease enrichment analysis of RIP-seq–identified transcripts enriched in RBM8A immunoprecipitates. Enriched disease terms were identified based on adjusted P value. Bars are color-coded to highlight major functional groupings: cognitive disability (yellow), general developmental conditions (blue), and craniofacial/skeletal abnormalities (black).

D. Volcano plot showing differentially enriched transcripts associated with RBM8A, highlighting top 20 enrichment of transcripts involved in ciliary function and Hedgehog signaling pathways (e.g., Ift172, Dync2h1, Ptch1, Gli1, Kif3a, and Mks1) from GO biological process, KEGG and Reactome databases.

E. Heat map of transcripts enriched in RBM8A immunoprecipitates associated with primary cilia function showing relative expression across samples. Values represent normalized expressions and are displayed as row-wise z-scores. 3 replicates of IgG (control) on the left and 3 replicates from Rbm8a immunoprecipitates on the right.

F. Transcript feature distribution plots show that RBM8A-associated transcripts (“list”; blue) were significantly enriched for longer transcript lengths and larger genomic span compared to background (“genome”; red). Chi-squared test.

G. Distribution of the number of transcript isoforms per protein-coding gene for RBM8A-associated transcripts (“list”; red) compared to the genomic background (“expected”; gray). RBM8A-associated genes were significantly enriched for higher isoform counts, indicating a preference for genes with increased transcript complexity. Chi-squared test (P = 9.4 × 10⁻³⁰⁰).

Following the identification of RBM8A-associated transcripts by RIP-seq, we next asked whether these transcripts exhibit distinct structural or genomic features that may explain their preferential binding with RBM8A. Transcript feature distribution plots show that RBM8A-associated transcripts were significantly enriched for longer transcript lengths compared to the background (Fig. 5F). Consistent with this, RBM8A-associated transcripts were significantly enriched for genes with larger genomic spans, indicating that RBM8A preferentially binds to structurally complex, intron-rich genes (Fig. 5F). Interestingly, RBM8A-associated transcripts show significant enrichment in a narrower and shorter 3’ UTR length range, suggesting that RBM8A binding is not driven by extended post-transcriptional regulatory regions (Supplementary Fig. 4A). Additionally, RBM8A-associated transcripts exhibited a modest but significant shift toward higher GC content, suggesting a mild sequence composition bias in those transcripts (Supplementary Fig. 4A). There was no significant difference in 5’ UTR length between RBM8A-associated transcripts and background, suggesting that RBM8A does not preferentially bind based on 5’ UTR structures (Supplementary Fig. 4A).

RBM8A-associated transcripts were strongly enriched for protein-coding genes, which is consistent with its established role in mRNA processing²³ (Supplementary Fig. 4B). Consistently, these transcripts were significantly enriched for genes with more exons and depleted in genes with fewer exons, indicating preferential binding to transcripts requiring extensive splicing and its role in NMD^67–71 (Supplementary Fig. 4C). Furthermore, these transcripts were significantly enriched for genes with more transcript isoforms, indicating preferential binding to structurally complex transcripts (Fig. 5G). The chromosomal distribution plot shows RBM8A-associated transcripts had a broad distribution without chromosomal bias, suggesting that binding is based on RNA features rather than position (Supplementary Fig. 4D).

Based on our finding that RBM8A-associated transcripts exhibit increased structural complexity, we next examined whether these genes were enriched for RNA processing pathways. As expected, heatmap analysis revealed enrichment of genes involved in RNA splicing and processing (Supplementary Fig. 4E), suggesting that Rbm8a preferentially regulates structurally complex transcripts within ciliary and developmental gene networks through RNA processing mechanisms.

Rbm8a deficiency alters cell fate and Hedgehog signaling in developing limbs

Given that RIP-seq analysis revealed enrichment of RBM8A-associated transcripts involved in cell cycle and ciliary function, we next asked whether disruption of Rbm8a causes early cellular defects relevant to limb development. To address this, we first performed immunocytochemistry on primary limb bud cells from Prx1; Ai14^f/+ mice (control) and Prx1-Rbm8a^f/+; Ai14^f/+ mice (Fig. 6A). Compared to control cells, we observed reduced fluorescence intensity of Ki67 (Fig. 6B) and of Fibronectin (Fig. 6C) in Prx1-Rbm8a^f/+; Ai14^f/+ cells, suggesting altered proliferation and extracellular matrix organization^72–74.

Figure 6. Prx1-Rbm8a^f/+ mice display disrupted ossification and Hedgehog signaling during embryonic limb development.

A. Representative images of primary limb bud culture of Prx1; Ai14^f/+ mice (control) and Prx1-Rbm8a^f/+; Ai14^f/+ mice showing Ai14 (tdTomato) reporter signal.

B. Representative immunofluorescence staining images of primary limb bud culture of Prx1; Ai14^f/+ and Prx1-Rbm8a^f/+; Ai14^f/+ mice showing Ki67 (green) merged with DAPI. Scale bar = 50 μm.

C. Representative immunofluorescence staining images of primary limb bud culture of Prx1; Ai14^f/+ and Prx1-Rbm8a^f/+; Ai14^f/+ mice showing Fibronectin (green) merged with DAPI. Scale bar = 50 μm.

D. Representative immunofluorescence staining images of E15 stylopods (humeri) from Rbm8a^f/+ and Prx1-Rbm8a^f/+ embryos showing COL2A1 (green), IHH (red), merged with DAPI (blue). The primary ossification center is outlined in yellow. Scale bar = 200 μm.

E. Quantification and statistical analysis of COL2A1 and IHH signal intensity of control and Prx1-Rbm8a^f/+ stylopods. Mean ± SEM (control: n=4; mutant: n=6). Unpaired t-test. *, p < 0.05.

F. Representative immunofluorescence staining images of E15 autopods (forepaws) from Rbm8a^f/+ and Prx1-Rbm8a^f/+ embryos showing COL2A1 (green), IHH (red), merged with DAPI (blue). Scale bar = 200 μm.

G. Quantification and statistical analysis of COL2A1 and IHH signal intensity of control and Prx1-Rbm8a^f/+ autopods. Mean ± SEM (control: n=6; mutant: n=8). Unpaired t-test. ns, not significant, **, p < 0.01.

H. qRT-PCR results of Hedgehog pathway targets using RNA isolated from E15 forelimbs of Prx1-Rbm8a^f/+ and control embryos. Relative gene expression was quantified using the ΔΔCt method and normalized to Actin. Mean ± SEM (n=3 per group). Unpaired t-test. ns, not significant, *, p < 0.05, **, p < 0.01.

To examine whether these cellular changes were associated with disrupted chondrogenic progression and developmental signaling, we performed immunohistochemistry on E15 forelimbs from Rbm8a^f/+ (control) and Prx1-Rbm8a^f/+ mice. Both COL2A1 and IHH expression were significantly higher in Prx1-Rbm8a^f/+ stylopods, although COL2A1-positive domain appeared spatially altered, suggesting disrupted chondrocyte differentiation^75,76 and a delayed or inefficient ossification center formation^9,77,78 (Fig. 6D, 6E). Additionally, Prx1-Rbm8a^f/+ mice had significantly higher COL2A1 expression (but not IHH) in the digits than control mice (Fig. 6F, 6G). These results indicate disrupted early ossification processes and altered Hedgehog signaling in Prx1-Rbm8a^f/+ mice^12,79,80.

To validate these findings at the mRNA level, we performed qRT-PCR analysis of Hedgehog pathway targets using RNA isolated from E15 forelimbs of Prx1-Rbm8a^f/+ and control mice. qRT-PCR revealed a significant reduction in Gli3 and Hhip in Prx1-Rbm8a^f/+ forelimbs, indicating reduced downstream Hedgehog signaling activity^81–83 (Fig. 6H). Moreover, genes associated with chondrogenic and osteogenic differentiation, such as Sox9 (Fig. 6H), Grem1, and Runx2, showed downward trends in Prx1-Rbm8a^f/+ forelimbs (Supplementary Fig. 4G). In contrast, expression levels of Foxf1 and Ccnd1 were not significantly different between control and Prx1-Rbm8a^f/+ forelimbs (Supplementary Fig. 4G), suggesting that Rbm8a deficiency selectively affects specific components of developmental signaling pathways. Together, these findings suggest that Rbm8a deficiency disrupts Hedgehog-dependent cell differentiation during early limb development^74,84,85.

Rbm8a plays a role in additional organogenesis and hematopoiesis

Alterations in ciliary processes and developmental signaling pathways observed in our bone-specific cKO mouse model provide potential mechanistic insights into the absence of radius (skeletal defects) observed in TAR. Primary cilia are also present on almost all hematopoietic cells and are critical for regulating key signaling pathways for their proper development. Therefore, RBM8A-related defects in primary cilia and associated signaling pathways may also contribute to the platelet abnormalities observed in TAR.

Although Prx1-Cre drives recombination in limb and craniofacial mesenchyme (Fig. 2A; Supplementary Fig. 1A), whether the defects in mesenchymal stromal cells in the bone marrow can cause indirect hematologic impairments is unknown. We examined whether thrombocytopenia is present in our Prx1-Rbm8a^f/+ mouse model. Prx1-Rbm8a^f/+ mice show no such effects on the complete blood count (CBC) test; their red blood cells, white blood cells, platelets, lymphocytes, neutrophils, and monocytes do not differ statistically from those of their littermate controls (Supplementary Fig. 5A). Their liver, kidney, and spleen are similar in size to those of their littermate controls, although their heart may be slightly smaller (Supplementary Fig. 5B).

To examine how mutations in RBM8A lead to platelet deficiency, we crossed Pf4-Cre mice with Rbm8a^f/f mice to selectively delete Rbm8a in megakaryocytes and platelets (Supplementary Fig. 6A). Interestingly, homozygous Pf4-Rbm8a^f/f embryos were viable and exhibited normal survival and development into adulthood. As expected, Pf4-Rbm8a^f/+ mice exhibit a significant reduction in platelet counts, with no differences in white blood cell, lymphocyte, neutrophil, or monocyte counts (Supplementary Fig. 6B). Strikingly, Pf4-Rbm8a^f/f mice showed an even more severe reduction in platelet counts (Supplementary Fig. 6B). The liver, kidney, and heart of Pf4-Rbm8a^f/+ mice were comparable to those of their littermate controls. Intriguingly, the spleen of homozygous Pf4-Rbm8a^f/f mice had noticeably enlarged spleens (splenomegaly) compared to those of their littermate controls and heterozygous Pf4-Rbm8a^f/+ mice (Supplementary Fig. 6C). This result further supports previous studies that RBM8A is required for normal hematopoietic processes^86,87. Further studies utilizing this blood-specific cKO mouse model can elucidate the role of Rbm8a in hematopoiesis.

Discussion

Here, we identify a novel role for Rbm8a in regulating skeletal development through its role in RNA processing of structurally complex transcripts. Because RBM8A mutations cause TAR syndrome²², we developed a mouse model to investigate how Rbm8a deficiency results in the TAR skeletal phenotype. We validate the first mouse model of TAR syndrome, Prx1-Rbm8a^f/+, by various phenotypic, behavioral, and molecular analyses. Our findings reveal new insights into structural and functional changes in these mouse brains, which are linked to neurological phenotypes reported in human TAR patients. Our RIP-seq analysis revealed that Rbm8a preferentially binds with long, exon-rich transcripts with more isoforms, which are enriched in ciliary and developmental signaling pathways. Immunostaining and qRT-PCR results showed the selective downregulation of Hedgehog target genes in our mutant mice. Together, our findings support a model in which Rbm8a regulates RNA processing of transcripts that are enriched in primary cilia and developmental signaling pathways, and pinpoints the disruption of this process as the mechanism by which Rbm8a deficiency leads to impaired Hedgehog signaling, contributing to skeletal maldevelopment in TAR syndrome.

Role of Rbm8a in primary cilia and Hedgehog signaling function

Primary cilia are non-motile sensory organelles that function as cellular antennae to sense and transduce extracellular signals and regulate multiple core signaling pathways^88,89. During osteogenesis, primary cilia are essential in coordinating osteoblast differentiation, chondrocyte proliferation, and bone matrix formation^88,90. Because primary cilia consist of many interdependent protein complexes that are tightly assembled, they are vulnerable to transcriptional changes⁸⁸. Notably, many ciliary genes have long introns, many exons, and complex splicing patterns, rendering them particularly sensitive to RNA processing defects^91,92. Consistent with previous findings demonstrating that alternative splicing programs regulate developmental processes^93–96 and that RNA surveillance mechanisms such as NMD maintain transcript integrity^97,98, these features make ciliary transcripts particularly susceptible to RBM8A deficiency. Given that RBM8A controls post-transcriptional RNA processing in its role in the EJC²³, its deficiency likely results in primary ciliary signaling defects (Fig. 5; Supplementary Fig. 4). These perturbations subsequently dysregulate core developmental signaling pathways (such as Hedgehog), leading to phenotypic alterations, as observed in this study. The skeletal phenotypes of Prx1-Rbm8a^f/+ mice align with those observed in human skeletal ciliopathies such as Jeune syndrome, Ellis–van Creveld syndrome, cranioectodermal dysplasia, and short-rib polydactyly syndromes^99,100, further suggesting that disrupted primary cilia-dependent signaling pathways result in congenital bone defects.

As such, our findings indicate that Rbm8a deficiency alters the expression of genes associated with the Hedgehog signaling pathway. Significantly reduced Gli3 and Hhip expression (Fig. 6H) in Prx1-Rbm8a^f/+ indicate weakened Hedgehog signaling^80–83. Downward trends, but not statistically significant attenuation of, downstream regulators such as Sox9, Grem1, and Runx2^74,84,85 (Fig. 6H; Supplementary Fig. 4G), suggest that the suppression of pathway activity is moderate. Surprisingly, increased IHH expression was observed in the pre-hypertrophic/hypertrophic chondrocyte region next to the primary ossification center in the humerus (Fig. 6D, 6E). IHH may be upregulated as a compensatory response to attenuated Hedgehog pathway activity^9,101,102. Another potential explanation is that Prx1-Rbm8a^f/+ mice produce more ligand (IHH) but respond less effectively (Gli3 and Hhip)^101,103. This would prevent proper progression of osteogenesis, resulting in reduced expansion of the primary ossification center (Fig. 6D) and a delay in ossification¹⁰⁴.

Role of Rbm8a in regulation of osteogenesis

Osteogenesis has two main processes: endochondral and intramembranous^9,11. Long bones, such as the limb bones, form through endochondral ossification, in which a cartilage template gets replaced by bone^9,11. Most cranial bones develop through intramembranous ossification, in which mesenchymal cells directly differentiate into osteoblasts, though some cranial bones form via endochondral ossification^9,11. Our study shows that in Prx1-Rbm8a^f/+ mice, there are significant alterations in long bones (digits, radii, ulnae, humeri, tibia, and femur) and various cranial landmarks (Fig. 2; Fig. 3; Supplementary Fig. 2). The fact that both ossification processes are disrupted by Rbm8a deficiency suggests that this gene is important in both mechanisms. Notably, both osteogenesis processes involve conserved developmental networks such as Hedgehog signaling^9,11,79, demonstrating how disruption of shared molecular regulators can yield context-dependent, tissue-specific phenotypes. Furthermore, embryonic lethality observed in homozygous Prx1-Rbm8a^f/f mice (Fig. 2F) indicates a critical role of Rbm8a in developmental processes required for fetal viability in mice, consistent with the key role of primary cilia-dependent Hedgehog signaling across multiple organ systems during development^81,88.

Notably, Prx1-Rbm8a^f/+ mice exhibit significant alterations in brain structure (Fig. 4G–4J). This is likely due to constraints imposed by the significant cranial abnormalities, which are present as early as P5 and continue to at least P30 (Fig. 2I–2S; Supplementary Fig. 2A-2F). This restriction would prevent the brain growing inside the cranium from developing properly, analogous to craniosynostosis^60,61. Subsequent changes in brain size and shape likely affect its function, as reflected in some of the behavioral tests we performed (Fig. 3E, 3F; Supplementary Fig. 3). This resembles the neurological findings reported in some TAR patients, such as intellectual disability and psychiatric disorders¹⁹. Due to the rarity of TAR syndrome, the neurological aspects of the condition are not fully defined. Our findings can serve as a foundation for further understanding of the distinct ways in which TAR brains may be altered. Moreover, microcephaly and intellectual disability were among the top-enriched disease terms for RBM8A-associated transcripts (Fig. 5C). These conditions are also observed in 1q21.1 deletion syndrome¹⁰⁵, highlighting that different deletions within the same chromosomal region yield overlapping phenotypes, though via distinct mechanisms.

Role of RNA regulation in ossification

Beyond the context of RBM8A and TAR syndrome, our findings illuminate how perturbations in RNA regulatory mechanisms can affect a developmental signaling program that guides embryonic bone formation by governing signaling robustness. As such, this study can serve as a potential framework for understanding the mechanisms of other congenital skeletal disorders. Our study supports the core concept that developmental signaling pathways function in a tightly coordinated regulatory network that combines transcriptional and post-transcriptional inputs²³, consistent with prior studies highlighting the role of post-transcriptional mechanisms in controlling developmental programs^97,106,107. Accordingly, it is through the RNA quality control function of RBM8A that the stability of transcripts necessary for proper Hedgehog signaling responses is affected. As a consequence of altered RNA regulatory function and subsequent changes in transcript stability, cellular signaling competence may be impacted^23,108. When cellular signaling competence is reduced, cells are less effective in properly interpreting and transducing signaling pathway cues¹⁰⁹. This explains the moderate (not fully abolished) molecular changes and variable skeletal phenotypes observed in this study. Our findings highlight the modulation of RNA processing mechanisms as a critical contributor to proper ossification processes during early skeletal development.

While this study provides novel insights into Rbm8a and bone development, some limitations should be considered. While our RIP-seq analysis identified RBM8A-associated transcripts, it cannot define individual targets that directly impact the observed skeletal phenotypes. Our qPCR analyses of several Hedgehog-related genes confirmed the attenuated Hedgehog pathway, but testing additional genes in the pathway is necessary to understand the broader scope. Future studies should include bulk RNA-sequencing comparing control and Prx1-Rbm8a^f/+ mouse limbs to investigate overall transcriptional alterations in relevant genes, thereby bridging RIP-seq and qPCR data. Moreover, examining the impact of Rbm8a deficiency in other primary cilia-dependent pathways involved in osteogenesis (Wnt, TGF-β/BMP, and PDGF)^88,90 can help determine the extent of primary ciliary signaling defects. Additionally, single-cell RNA-sequencing would help reveal potential cell-type-specific perturbations in the altered Hedgehog signaling pathway. Lastly, additional experiments to parse out the neurological aspects of Prx1-Rbm8a^f/+ mice can demonstrate a new understanding of TAR brains. Behavioral tests for learning and memory can reveal novel insights into how the distinct brain changes affect cognition, and examining cellular changes in cranial bones can help explain those behavioral outcomes.

In all, our present study establishes Rbm8a as a critical regulator of embryonic bone development by maintaining the integrity of structurally complex transcripts necessary for primary cilia and developmental signaling pathways. We demonstrate that disruption in RNA processing (i.e., Rbm8a deficiency) alters Hedgehog signaling, leading to significantly impaired ossification processes and subsequent skeletal maldevelopment. These findings provide new insight into the mechanisms underlying not only TAR syndrome, but also other congenital skeletal anomalies and other RBM8A-related disorders. Taken together, our study provides a framework for future research examining the impact of RNA processing pathways during early development, and for translational studies addressing human congenital skeletal disorders.

Methods

Human Bone Marrow-Derived MSC Culture

Human bone marrow-derived MSCs were obtained from ATCC (catalog #: PCS-500-012) and cultured using Bone Marrow-Mesenchymal Stem Cell Basal Medium (ATCC, catalog #: PCS-500-030) with 7% FBS and growth factor supplements (15 ng/ml IGF-1, 125 ng/ml FGF2).

Generation of the Prx1-Cre; Rbm8a^f/+ Mice

Rbm8a-floxed mice (Rbm8a^f/f) without the Neo cassette were previously generated in our lab^25,26,28. These mice were then crossed with Prx1-Cre (Jackson Laboratory, stock #: 005584) to selectively knock out Rbm8a in early limb bud mesenchyme and a subset of craniofacial mesenchyme. The resulting Prx1-Cre; Rbm8a^f/+ mice display significant bone development defects but are otherwise healthy. However, we cannot detect Prx1-Cre; Rbm8a^f/f mice at the postnatal stage, as they are embryonic lethal around E12–13.

Animals

All procedures involving mice were reviewed and approved by the Pennsylvania State University IACUC committee under IACUC protocols 44057 and 800335. Mice were housed by sex (2–5 mice per cage) in a room with a 12-hour light/dark cycle and provided ad libitum access to food and water.

Timed Mating and Embryo Collection

Timed matings were performed by housing a Prx1-Cre; Rbm8a^f/+ male and a Prx1-Cre; Rbm8a^f/+ female mouse together overnight, then designating the presence of a vaginal plug the following morning as E0.5. Pregnant female mice were euthanized at E15, and the embryos were harvested in cold PBS. Both forelimbs of each embryo were carefully isolated using forceps and a microscope, then stored immediately in −80°C.

Primary Limb Bud Cell Culture

Prx1-Cre; Rbm8a ^f/+ mice were crossed with Ai14 reporter mice (Jackson Laboratory, stock #: 007914) for the desired genotypes of Prx1-Cre; Ai14^f/+ mice (control) and Prx1-Cre; Rbm8a^f/+; Ai14^f/+ mice. Pregnant female mice were euthanized at E11-E12, and the embryos were harvested in a sterile Petri dish with ice-cold PBS. Both limb buds of each embryo were carefully isolated using forceps and a microscope, then transferred to a new dish with sterile PBS. The limb buds were minced using fine scissors, then incubated in 0.25% trypsin at 37°C for 15 minutes. The samples were further dissociated by gently pipetting up and down. The digestion was stopped by adding an equal volume of Gibco DMEM/F-12 medium (Thermo Fisher Scientific, catalog #: 11320033), supplemented with 10% FBS and 1% Penicillin-Streptomycin. This cell suspension mixture was passed through a 70 μm cell strainer to remove debris, then plated directly onto gelatin-coated glass coverslips. The cells were incubated at 37°C and 5% CO₂, with media changes every 2–3 days.

Immunocytochemistry

Sterile glass coverslips were incubated in EmbryoMax 0.1% gelatin solution (MilliporeSigma, catalog #: ES-006-B) for 2 hours at 37°C and 5% CO₂, then washed 3 times with sterile PBS and air-dried in the tissue culture hood. After the primary limb bud cells were plated and grown on these coverslips, the staining was performed as previously described²⁵. Briefly, they were fixed with 4% PFA for 15 minutes at room temperature, then permeabilized by incubation in 0.3% Triton X-100 solution for 1 hour at room temperature. Subsequently, they were incubated in 5% normal serum for 1 hour at room temperature, then incubated overnight at 4°C in rabbit anti-RBM8A (GeneTex, catalog #: GTX131387), mouse anti-Fibronectin (DSHB Hybridoma, catalog #: P1H11), mouse anti-N-Cadherin (DSHB Hybridoma, catalog #: MNCD2-s), rabbit anti-Ki67 (Invitrogen, catalog #: 701198), or goat anti-SOX9 (R&D Systems, catalog #: AF3075). After staining, coverslips were mounted onto microscope slides using ProLong Diamond Antifade Mountant with DAPI (Invitrogen, catalog #: P36971). Whole slide scanning (20x) was performed on the Zeiss LSM 800 Confocal Laser Scanning Microscope.

Immunohistochemistry

Wild-type E10.5 and E14.5 mouse embryos were fixed in 4% PFA in PBS for 1 hour, briefly washed with PBS, then left in 30% sucrose overnight, embedded in O.C.T freezing media, and stored at −80°C. Cryosections at 10 μm thickness were cut with a Leica LM1900 Cryostat. E15 Prx1-Rbm8a^f/+ and control embryos were collected from pregnant dams, and the forelimbs of those E15 mice were isolated using forceps and a microscope. The forelimbs were then postfixed for 24 hours in 4% PFA at 4°C, then stored in PBS at 4°C. Samples were processed and embedded in paraffin by HistoWiz Inc. (Long Island City, NY, USA) using their standard operating procedure. The paraffin-embedded samples were sectioned at 5 μm using the Thermo Shandon Finesse paraffin microtome, then dewaxed using the Leica Autostainer ST5010 XL at the Huck Institutes Microscopy Facility (University Park, PA, USA). The samples were then incubated in Pepsin-HCl (Mettler Toledo, catalog #: 01-912-587) for 30 minutes at 37°C for antigen retrieval. The staining was performed as previously described. Antibodies used include: CD-90 (Invitrogen, catalog #: 14-0909-82), rabbit anti-RBM8A (GeneTex, catalog #: GTX131387), goat anti-SOX9 (R&D Systems, catalog #: AF3075), rabbit anti-IHH (Proteintech, catalog #: 13388-1-AP), mouse anti-COL2A1 (DSHB Hybridoma, catalog #: II-II6B3). After staining, sections were coverslipped using ProLong Diamond Antifade Mountant with DAPI (Invitrogen, catalog #: P36971). Whole slide scanning (20x) was performed on the Zeiss LSM 800 Confocal Laser Scanning Microscope.

Behavioral Tests

All behavioral tests were performed in a sound-attenuated room. A cohort of 30 mice that are similar in age was used. They were 2–3 months old on the first behavioral test and 4–5 months old at the last test. The cohort consisted of the following: Rbm8a^f/+: n = 9 females, n = 9 males; Prx1-Cre; Rbm8a^f/+: n = 7 females, n = 5 males.

Open Field Test

Mice were placed in the center of the open field arena (40 cm × 40 cm) and allowed to move freely for 10 minutes while being tracked by the Noldus EthoVision XT video behavior recognition system. At the conclusion of each trial, the surface of the arena was cleaned with ethanol.

Elevated Plus Maze

Mice were placed in a four-arm maze that is elevated 50 cm above the floor. In this apparatus, the four arms intersect to form a plus shape. Two of the arms are closed on the sides, and the other two arms are open. Mice were placed in the center of the apparatus, facing an open arm, and allowed to freely explore the maze for 10 minutes while being tracked by the Noldus EthoVision XT video behavior recognition system. At the conclusion of each trial, the surface of the arena was cleaned with ethanol.

Rotarod

Mice were placed on a rotating drum with controlled, accelerating speed (4–40 rpm over 300 seconds). When the mice fall approximately 8 inches from the rotating drum onto the lever, their body weight presses the lever down, automatically pausing the test timer. The time each mouse took to fall was recorded. Each mouse was tested three times per day, and the latency to fall was averaged from those trials. All mice are tested for three consecutive days to test motor learning and memory.

Marble Burying Test

Each mouse was placed in a new cage with 3 cm of bedding material with five rows of four marbles on the surface of the bedding. The mice were allowed to move freely in the cage for 10 minutes. At the conclusion of each trial, the cages were photographed to document the status of the marbles. Three separate lab members blinded to genotype independently scored the number of marbles buried, and the results were averaged.

Grip Strength

Mice were handled by the tail and allowed to grasp the test bar with their forepaws, then gently pulled backward by the tail in a horizontal plane until they released the grip. The maximum force exerted before release was recorded using a grip strength meter. Each mouse underwent six consecutive trials with 1-minute rests between trials to minimize fatigue. Grip strength was normalized by the weight of each mouse, and the average of six trials was used for analysis.

Grid Walking Test

Mice were placed on a smooth, rounded wire grid (2 mm wire diameter) that was slightly elevated (10 cm) from the bottom of the apparatus. The mice were allowed to freely explore the arena for 5 minutes while being video-recorded from below the grid wire surface. Three separate lab members blinded to genotype independently counted the number of paw slips (paw fully going through the grid), and the results were averaged.

Micro-computed Tomography (micro-CT)

Postnatal mice were imaged ex vivo using SkyScan 1176 (Bruker, Billerica, MA, USA) in the Huck Institutes High Field Magnetic Resonance Imaging Core Facility (University Park, PA, USA). The micro-CT projection images were reconstructed into cross-section images using NRecon (Bruker, Billerica, MA, USA). The cross-section images were then analyzed using Avizo (Thermo Fisher Scientific, Waltham, MA, USA).

RNA Isolation

E15 forelimb tissues were homogenized using a mini handheld plastic tube pestle in 1.5 mL microcentrifuge tubes. A set of forelimbs per embryo was used as a single sample. RNA was isolated using the RNeasy Mini Kit (QIAGEN, catalog #: 74104) according to the manufacturer’s protocol. The RNA yields were assessed using the NanoDrop microvolume spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

Quantitative Reverse Transcription–PCR (qRT-PCR)

The total RNAs were converted into cDNA by reverse transcription with the oligo-dT primer, using Superscript reverse transcriptase III (Invitrogen, catalog #: 18080093). qRT-PCR was performed using PowerTrack SYBR Green Master Mix for qPCR (Applied Biosystems, catalog #: A46109) with the following primers (5′–3′): mouse RBM8A (F: ATTACGACAGTGTGGAGCAG, R1: TTGACCATTTAGTCCTTCCA), GLI3 (F: GGAGCAGGTTCTGGAACCTT, R: TCCATGTTCTGGGCAGTCTT), HHIP (F: TTGCCACTCAGGAGGTTGAA, R: TTCTGGTGCTCAGGACTGGT), SOX9 (F: CAGCAAGACTCTGGGCAAGC, R: GGTGGTCCTTTCCTCTTGTG). The expression of the β-Actin gene (F: CGTGGGCCGCCCTAGGCACCA, R: TTGGCCTTAGGGTTCAGG GGGG) was used as an internal control. Primers were designed using NCBI Primer-BLAST and synthesized by IDT. The qPCR reactions were performed on a StepOnePlus Real-Time PCR system (Applied Biosystems, Carlsbad, CA, USA), and the Ct calculations were performed using StepOne software version 2.3.

RNA Immunoprecipitation

Roughly 100 E15 embryos were collected from multiple pregnant dams. Both forelimbs from each E15 embryo were isolated and pooled to yield 3 total replicates. The pooled samples were lysed in homogenization buffer at 2% weight/volume using a mini handheld plastic tube pestle in 1.5 mL microcentrifuge tubes. Each homogenate was centrifuged at 10,000 rpm for 10 minutes at 4°C. Supernatants were transferred to fresh tubes, and either rabbit anti-RBM8A antibody (GeneTex, catalog #: GTX131387) or anti-IgG antibody was added to each tube, which were incubated in the cold room for 4 hours with rotation. Dynabeads Protein G for Immunoprecipitation (Invitrogen, catalog #: 10004D) was equilibrated to the homogenization buffer using DynaMag-2 Magnet (Invitrogen, catalog #: 12-321-D). The antibody-supernatant mixtures were added to pre-cleared Dynabeads Protein G and incubated overnight at 4°C with rotation. The next day, the antibody-supernatant-bead mixtures were placed on the DynaMag-2 Magnet to remove the supernatant. The beads were washed with a high-salt buffer three times, each for 10 minutes at 4°C with rotation. After the last wash, lysis buffer (10 μL β-mercaptoethanol per mL RLT lysis buffer) was immediately added to the beads, and the mixture was vortexed for 30 seconds. The samples were then placed on the DynaMag-2 Magnet, and the supernatant was collected and transferred to new tubes. RNA was isolated from these samples using the RNeasy Mini Kit (QIAGEN, catalog #: 74104) according to the manufacturer’s protocol. The RNA yields were assessed using the NanoDrop microvolume spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

RIP-seq – Library Preparation & Sequencing

RIP-seq library preparation and sequencing were performed by Azenta (Chelmsford, Massachusetts, USA). RNA samples were quantified using Qubit 4.0 Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA), and RNA integrity was assessed using TapeStation (Agilent Technologies, Palo Alto, CA, USA). The NEBNext Ultra II RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA, USA), including clustering and sequencing reagents, was used for library preparation without RNA enrichment or depletion, according to the manufacturer’s recommendations. Briefly, total RNA was directly fragmented and primed using random primers. RNA fragmentation was performed at 94°C for a duration selected based on RNA integrity to target an average insert size of approximately 200 nucleotides. First-strand and second-strand cDNA were subsequently synthesized. The resulting cDNA fragments were end-repaired and adenylated at 3’ ends, and universal adapters were ligated to the cDNA fragments, followed by index addition and library enrichment by PCR with a limited number of cycles. The sequencing library was validated on the Agilent TapeStation (Agilent Technologies, Palo Alto, CA, USA) and quantified using Qubit 4.0 Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA) and qPCR (KAPA Biosystems, Wilmington, MA, USA).

The sequencing libraries were multiplexed and clustered onto a flowcell on the Illumina NovaSeq instrument according to the manufacturer’s instructions. The samples were sequenced using a 2×150bp Paired End (PE) configuration. Image analysis and base calling were conducted by the NovaSeq Control Software (NCS). Raw sequence data (.bcl files) generated from Illumina NovaSeq were converted into fastq files and de-multiplexed using Illumina bcl2fastq 2.20 software. One mismatch was allowed for index sequence identification.

RIP-seq – Data Analysis

After assessing the quality of the raw data, sequence reads were trimmed to remove possible adapter sequences and nucleotides with poor quality. The trimmed reads were mapped to the reference genome available on ENSEMBL using the STAR aligner v.2.5.2b. The STAR aligner is a splice aligner that detects splice junctions and incorporates them to help align the entire read sequences. BAM files were generated from this step. Unique gene hit counts were calculated by using feature Counts from the Subread package v.1.5.2. Only unique reads that fell within exon regions were counted. After the extraction of gene hit counts, the gene hit counts table was used for downstream differential expression analysis. Using DESeq2, a comparison of gene expression between the sample groups was performed. The Wald test was used to generate p-values and log2 fold changes. Genes with adjusted p-values (Padj) < = 0.05 and log2 fold changes > = 0.5 were called as differentially expressed genes for each comparison. Differential exon usage between experimental groups was analyzed using DEXSeq v. 1.46.1, and exon bins with adjusted p-values (Padj) < = 0.05 and log2 fold changes > = 0.5 were considered significantly differentially used. Gene ontology and pathway enrichment analyses were performed on significantly differentially expressed genes and significantly differentially used exons using clusterProfiler v.4.10.0 together with org.Mm.eg.db v.3.17.0. Enrichment was assessed across GO, KEGG, and Reactome databases. Volcano plots were generated using EnhancedVolcano v. 1.18.0. Disease enrichment analysis was performed using enrichR v. 3.4. Mouse gene symbols were used as input for enrichment analysis against the DisGeNET disease annotation library. All downstream analyses were performed in R 4.3.1.

MRI - Acquisition

All samples were imaged using a 7T MRI system with a quadrature transmit volume coil (70 mm diameter) and a four-channel receive-only phased array cryocoil (Bruker Biospin, Billerica, MA, USA) at the Bernard and Irene Schwartz Center for Biomedical Imaging (New York, NY, USA). Co-registered T2-weighted images were acquired using a RARE sequence with the following parameters: TE/TR = 50/3000 ms, one signal average, echo train length = 8, the same FOV, matrix size, and slice thickness as the diffusion tensor images. Co-registered diffusion tensor images were acquired using a diffusion-weighted echo planar imaging (DW-EPI) sequence with the following parameters: TE/TR = 30/5000 ms, one signal average, four segments, diffusion gradient duration = 5 ms, diffusion time = 15 ms, 30 diffusion encoding directions, b = 1000, 2000 s/mm2, five non-diffusion-weighed images, FOV = 15 × 15 mm, matrix size = 128 × 128, slice thickness = 0.5 mm, and an in-plane resolution = 0.117 × 0.117 mm2.

MRI - Volume Measurement

Whole brain and cortex volumes were determined using segmentation analysis (AMIRA 3D, 2024.1; Thermo Fisher Scientific) software. The brain and cortex were manually outlined and segmented in AMIRA. The number of voxels from segmented tissue within each slice was added together to compute the total volume (= num of voxels * volume unit of voxels) for each brain region and the total volume. The computation was carried out in MATLAB R2025a.

Statistical Analysis

Statistical analyses were performed on GraphPad PRISM software (Boston, MA, USA), expressed as means +/− standard error of means (SEM). Unpaired t-tests and ANOVA tests were used.

Supplementary Material

Supplementary Files

This is a list of supplementary files associated with this preprint. Click to download.

• SupplementaryInformation.docx

Data availability:

All data necessary to support the conclusion of this study are available in the manuscript or the supplementary information. Raw data are deposited in GEO (Gene Expression Omnibus) associated with BioProject accession number PRJNA1401423.