Splicing factor RBM10 loss fuels thyroid cancer metastasis

RNA alternative splicing (AS) is finely orchestrated to gear organism development and contribute to tissue diversity and homeostasis, the dysregulation of which results in a pathological transcriptome and promotes tumorigenesis and development (Rahman et al., 2020; Bradley and Anczuków, 2023). The assembly of the spliceosome on pre-mRNA strictly depends on the presence of the 5′ splice site, branchpoint sequence, and 3′ splice site. The U2 small nuclear ribonucleoprotein (snRNP) plays a critical role in recognizing the branchpoint sequence, which is essential for the formation of the spliceosomal A complex and subsequent spliceosome activation. Mutations in SF3B1 (a component of the U2 snRNP), U2AF1 (which binds to the 3′ splice site), and U2AF2 (which recognizes the polypyrimidine tract)—all essential for proper spliceosome assembly and 3′ splice site selection—have been frequently implicated in various types of cancer (Rogalska et al., 2023).

RBM10 is an RNA-binding protein that interacts with SF1, U2AF, and U2 snRNA (Rodor et al., 2017; Inoue, 2021). Accordingly, it has been detected in the early spliceosomal A and B complexes (Behzadnia et al., 2007; Kuhn et al., 2009). In general, RBM10 binding near the splice sites of cassette exons interferes with the recognition of the 3′ and 5′ splice sites, thereby promoting the exclusion of these exons (Inoue, 2021). Generally regarded as a tumor suppressor, RBM10 mutations have been implicated in several cancers, most of which are frameshift or nonsense mutations, resulting in a loss of function (Hernández et al., 2016; Inoue, 2021).

Thyroid cancer, predominantly composed of indolent and well-differentiated papillary thyroid cancer, is typically associated with favorable clinical outcomes. However, a small percentage of patients (5%) exhibit distant metastasis, which is associated with lower 5-year survival rates (Boucai et al., 2024). Loss of RBM10 is associated with poor prognosis in patients with non-anaplastic thyroid cancer (Ibrahimpasic et al., 2017). However, the oncogenic role of RBM10 loss and its underlying mechanisms in thyroid cancer remain unexplored.

Krishnamoorthy et al. (2025) conducted a comprehensive analysis of multiple clinical cohorts to characterize the prevalence of splicing factor RBM10 mutations in thyroid cancer. They identified a significant enrichment of RBM10 alterations in non-anaplastic thyroid cancer with distant metastasis. RBM10 is highly conserved among mammals, with the human RBM10 protein sharing 96% amino acid sequence homology with the mouse protein (Inoue, 2021). To better understand how RBM10 loss contributes to thyroid tumorigenesis, the authors generated a thyroid-specific Rbm10 knockout in a genetically engineered mouse model expressing mutant Hras. No tumors were found at 12 mo in mice with either Rbm10 loss or Hras mutation. However, in mice with combined mutations, 97% of mice developed tumors between 10 and 12 mo, and 18% exhibited lung metastases. In addition, five isogenic human thyroid cancer cell lines were used to validate the oncogenic role of RBM10 loss. Knockdown of RBM10 promoted, while Dox-induced expression suppressed, cell proliferation.

Next, the authors analyzed how RBM10-regulated AS influences tumorigenesis. They employed two splicing analysis algorithms to comprehensively analyze splicing changes in the isogenic cell lines. Gene ontology analysis of the differentially spliced genes revealed a significant enrichment in extracellular matrix (ECM)/cytoskeleton remodeling pathways, with several ECM- and cytoskeleton-modulating genes ranking among the top splicing targets. Specifically, RBM10-null cells tended to express the exon 19–included isoform of VCL, thereby generating the protein meta-vinculin instead of vinculin. Two CD44 isoforms were also enriched: one with alternative exon 8 included and another with both exon 8 and exons 13–15 included. Additionally, the isoform of TNC-encoding tenascin C with exon 16 inclusion was enriched. To examine whether these AS events were directly regulated by RBM10, two crosslinking immunoprecipitation (CLIP) sequencing datasets were analyzed and suggested that VCL may be a direct binding target of RBM10, while CD44 splicing is regulated through indirect mechanisms, likely mediated by secondary effects of other splicing regulators whose expression and/or splicing is altered by RBM10 (Wang et al., 2013). While it may be context-dependent, RBM10 has been implicated in binding nearly all branch sites as part of an unusual U2 RNP particle, recently identified through an in vivo isolation method (Damianov et al., 2024). Therefore, understanding the splicing regulatory mechanisms of RBM10 loss in thyroid tumorigenesis is essential for uncovering its role in cancer progression and identifying potential therapeutic targets.

Further pathway analysis revealed a significant enrichment of AS events related to invasion, migration, ECM modulation, and cytoskeletal remodeling. This is supported by the observation that cells with endogenous loss or knockdown of RBM10 exhibit higher migration velocity and longer migration paths, while the re-expression of RBM10 reduced metastatic capability. Furthermore, isoform-specific knockdown revealed that all exon-included isoforms of VCL, CD44, and TNC confer varying degrees of pro-motility and pro-invasion properties. While individual knockdown of exon-included isoforms did not significantly affect the lung metastatic potential of mouse thyroid cancer cells, MVcl and Tnc double knockdown markedly suppressed metastasis. Triple knockdown of exon-included isoforms resulted in the most pronounced reduction. Mechanistically, since all three gene targets act upstream of RAC1, triple knockdown of exon-included isoforms significantly reduced RAC1-GTP levels. Further elucidation of the pro-metastatic oncogenicity of splicing switches, beyond exon-included isoforms, could provide critical insights for developing targeted therapeutic strategies. Approaches such as splice-switching antisense oligonucleotides, gapmers, or siRNAs can subsequently be evaluated in preclinical models.

In addition to its effects on cell motility and invasiveness, RBM10 loss was also found to be associated with increased proliferation and reduced apoptosis. A CRISPR-Cas9 dropout screen was conducted to identify potential genes required for viability of RBM10 loss thyroid cancer cells. Many of the genes depleted in the screen were associated with the NF-κB pathway, including RELA. Interestingly, CREBBP (CREB-binding protein), a critical transcriptional coactivator involved in NF-κB–mediated transcription, and a cassette splicing target of RBM10, also emerged as a significant hit in the screen. RBM10-deficient thyroid cancer cells with RELA knockdown exhibit increased vulnerability to TNFα-induced apoptosis. The synthetic lethality between RBM10 and NF-κB signaling is further evidenced by the higher sensitivity of RBM10-null cells to the NF-κB inhibitor.

RNA splicing is further complicated by the coregulation of individual splicing events by multiple splicing factors, some of which exert opposing effects (Rogalska et al., 2024). RBM5 and RBM10 are highly homologous splicing regulators (Inoue, 2021). Krishnamoorthy et al. show that the deletion of both genes is synthetic lethal, likely due to their shared RNA targets, consistent with RBM10’s ability to cross-regulate RBM5 expression through splicing-mediated nonsense-mediated RNA decay (Sun et al., 2017). In other contexts, RBM5 and RBM10 antagonistically regulate cancer cell proliferation and exhibit distinct positional effects on RNA splicing (Bechara et al., 2013). Exploring the interplay between RBM5 and RBM10 in thyroid cancer may facilitate the identification of key splicing targets. Additionally, while splicing regulation can be conserved across different biological contexts, it may lead to diverse physiological and pathological outcomes. A comprehensive evaluation of the clinical prognostic significance of dysregulated splicing events associated with RBM10 loss–driven metastasis—either individually or as part of an AS signature—could offer a strategy for patient stratification, potentially benefiting even those without RBM10 loss.

In summary, the study reveals the pro-metastatic role of RBM10 loss in thyroid cancer by dysregulating RNA splicing involved in cell motility and invasiveness. Loss of RBM10 leads to constitutive activation of RAC1 and confers synthetic lethality to NF-κB inhibition, shedding light on the complex biological pathways remodeled by splicing regulation. The specific insights into the mechanisms of RBM10 loss–driven metastasis and its associated vulnerabilities enhance our understanding of tumor progression and may open novel avenues for therapeutic strategies in thyroid cancer treatment.