Subtype-specific son isoform expression in Acute Myeloid Leukemia revealed by targeted long-read RNA sequencing

Abstract Background: The SON gene encodes a large nuclear protein that plays essential roles in RNA splicing, nuclear speckle integrity, and pluripotency. Dysregulation of SON has been implicated in various cancers, including leukemia. Despite increasing interest in its oncogenic potential, the isoform diversity of SON in acute myeloid leukemia (AML) remains poorly characterized. Given the potential for isoform-specific functions in splicing regulation and disease pathogenesis, especially in the context of fusion-driven and cytogenetically distinct AML subtypes, we aimed to systematically map and quantify SON isoforms in primary AML samples using targeted long-read RNA sequencing. Methods: To comprehensively characterize full-length SON transcripts, we performed targeted transcript enrichment followed by long-read RNA sequencing using Oxford Nanopore Technologies. Custom biotinylated DNA probes tiling the SON gene body and UTRs were designed and synthesized to capture SON-specific transcripts from total RNA. RNA samples were obtained from 24 specimens, including: CD34+ cells from primary AML patient bone marrow representing MLL-rearranged (n=6), inv(16) (n=2), inv(16)+MLLr (n=1), t(8;21) (n=2), and NPM1-mutated (n=1) subtypes. Normal CD34+ (n=3) and CD34− (n=2) bone marrow from healthy donors. AML cell lines including OCI-AML2, OCI-AML3, MOLM-13, MV4-11, SKNO-1, and THP-1 Following total RNA extraction and DNase treatment, cDNA was synthesized using strand-switching reverse transcription. Target capture was performed via hybridization of cDNA to SON-specific biotinylated probes, followed by magnetic bead purification. Enriched cDNA libraries were end-repaired, barcoded, and pooled for multiplexed Oxford Nanopore MinION sequencing. Reads were basecalled and demultiplexed using Guppy, and full-length isoforms were reconstructed using FLAIR and Mandalorion. Isoform abundance was quantified, and transcript novelty was assessed relative to GENCODE annotations. Differential isoform usage was analyzed across AML subtypes. Results: Initial analysis has revealed a surprisingly diverse repertoire of SON isoforms, including multiple unannotated transcripts. Subtype-associated isoform usage patterns are emerging, with preliminary evidence suggesting that certain truncated isoforms lacking RNA-binding domains are enriched in MLL-rearranged and inv(16) AML. Validation and quantification of these isoforms across subtypes are ongoing. Conclusions: This study is the first to define the SON isoform landscape in AML at full-length transcript resolution, uncovering subtype-specific splicing signatures and novel isoforms potentially involved in leukemogenesis. The differential expression of RNA-binding competent vs. truncated SON isoforms may reflect distinct nuclear architecture or splicing network rewiring in AML subtypes. These findings open new avenues for understanding isoform-level regulation of nuclear speckle dynamics and RNA processing in leukemia, with potential implications for biomarker development and isoform-specific targeting. Future Directions: We are currently validating the functional significance of key isoforms using isoform-specific overexpression and knockdown models in AML cell lines. Additionally, integration with rMATS-based splicing analyses and CLIP-seq will clarify how specific SON isoforms influence alternative splicing in a subtype-restricted manner.