SRSF2 mutation-induced CHD8 exon skipping represents a novel pathogenic mediator and therapeutic dependence in chronic myelomonocytic leukemia (CMML) Free

Introduction SRSF2, a splicing factor regulating both constitutive and alternative splicing (AS), is mutated in >40% of CMML pts. Proline-95 mutations promote widespread missplicing, including some well-characterised events previously linked to myeloid neoplasia, e.g. involving EZH2 and INTS3. While the functional consequences of SRSF2 mutations have been explored in MDS/AML models, their strong association with theCMML phenotype and precise roles in CMML pathobiology remain unclear.

Methods RNA-seq was performed on biobanked CMML CD34+ HSPCs and on a CRISPR-generated isogenic SRSF2^MT leukemia cell line (K562), seeking overlapping recurrent missplicing events. Meanwhile, we performed a whole-genome CRISPR^KO synthetic lethality screen in the isogenic K562s to identify genotype-specific dependencies. We then CRISPR-edited the THP1 monocytic leukemia cell line to model an identified CHD8 skipped exon (SE) event. Stable monoclones were subjected to in vitro proliferation and clonogenic assays, RIME (Rapid Immunoprecipitation Mass Spectrometry of Endogenous Proteins), RNA-seq and ChIP-seq.

Results Of 10,814 consecutive pts in the NW England myeloid malignancies database, 1,015 (9.4%) harbored ≥1 SRSF2 mutation/s. Despite only representing 7.5% of the cohort, CMML accounted for the largest proportion (37%) of SRSF2^MT disease. Conversely, 47% of CMMLs were SRSF2^MT, reaffirming CMML as the myeloid neoplasm most closely associated with SRSF2 mutations.

RNA-seq on CD34+ BM HSPCs from 33 CMML pts (SRSF2^MT n=15) revealed enrichment of inflammation-related pathways, increased senescence signatures and decreased S-phase activity in SRSF2^MT samples. In parallel, we performed RNA-seq on K562 cells carrying CRISPR-edited SRSF2^P95H, observing enrichment for DNA replication and drug resistance pathways in SRSF2^MTcells. rMATS revealed 3370 (SE 56%) and 3250 (SE 70%) missplicing events in CMML BM and isogenic K562s, respectively. Intersecting datasets revealed 373 shared events involving 281 genes, enriched in pathways related to mRNA processing/transport, protein transport between endoplasmic reticulum and cytosol, and telomere organization.

Seeking therapeutically tractable vulnerabilities in SRSF2^MTcells, a whole-genome CRISPR^KO synthetic lethality screen of the isogenic K562 model revealed 67 genes essential for survival of SRSF2^MT (vs wild-type) cells. These are involved in cellular processes such as RNA processing, splicing, chromatin remodelling and transcriptional regulation, revealing potential therapeutic targets.

The only gene common to all analyses was CHD8, a chromatin remodeller highly expressed in HSPCs with roles in maintaining stemness, also linked to roles in monocyte development/inflammation response. CHD8 was validated as essential in SRSF2^MT K562 cells, with knockout significantly reducing cell growth and clonogenic capacity vs wild-type cells.

The common CHD8 missplicing event in SRSF2^MT primary CMML and K562 cells results in exon skipping of ex36, which encodes a BRK domain integral to its chromatin remodelling function (likely mediating interactions with histones and CTCF). This SRSF2^MT-specific event was validated by RT-PCR in a panel of CMML BM samples.

We next modelled CHD8 ex36 skipping by CRISPR editing the splice acceptor site in ex36 of the THP1 monocytic leukemia cell line, generating 9 stable monoclones bearing: total exon retention (n=3), near-total SE (n=3) and a balanced/mixed pattern (n=3). Lines with prominent CHD8 ex36 skipping displayed enhanced proliferation and clonogenic capacity in vitro.

Transcriptomic profiling of these lines revealed that ex36 skipping dysregulates genes involved in chromatin remodelling, DNA replication, transcription-coupled replication and cell cycle pathways, reflecting defects evident in SRSF2^MTcells. In support, comparative RIME revealed differential binding of the CHD8 protein isoforms to the chromatin modifiers HDAC6 and UHRF2, and to the transcription elongation factor CDC73. Finally, ChIP-seq mapped comparative differences in genomic target profiles for the CHD8 isoforms.

Conclusion Our integrated transcriptomic and functional genomic approach identified novel recurrent AS events and potentially targetable vulnerabilities in SRSF2^MT CMML. We discovered a recurrent SRSF2^MT-induced CHD8 ex36 skipping event that contributes to the CMML phenotype, through altered recruitment of chromatin modifiers to target loci promoting defects in DNA replication and cell cycle progression.