2026 ASH®
Annual Meeting & Exposition December 12-15, 2026 ▪ New Orleans, Louisiana

Paper Details

Abstracts, authors, affiliations, and disclosures.

Monday, December 14, 2026
06:00 p.m. - 08:00 p.m.
MCCNO - Hall F
Technology and Procedures, Molecular Testing, Omics Technologies, Bioinformatics, Computational Biology, Emerging Technologies, Treatment Considerations, Biological Therapies, Immunotherapy, Non-Biological Therapies, Chemotherapy, Biological Processes, CHIP, Genomics, Hematopoiesis, Immune Mechanism, Molecular Biology, Diseases, Bone Marrow Failure Syndromes, Inherited Marrow Failure Syndromes, Genetic Disorders, Immunology, Autoimmune Disorders, Myeloid Malignancies, Acute Myeloid Malignancies, Chronic Myeloid Malignancies, Research, Fundamental Science, AML, Translational Research, MDS, MPN, CMML, Clinical Research, Study Population, Human, Adult, Elderly, Pediatric, Young Adult
PAPER

5588 Oncogenic SF3B1 mutations drive reproducible alterations of transcript 3' end formation distinct from their regulation of RNA missplicing

Agata Sekrecka, PhD, Michal Sekrecki, PhD2, Katelyn Yeh, N/A3, Ilya Kipkalov, N/A4, Shih-Han Lee, PhD5, Bita Fakhri, MD, MPH6, Sydney Xin-Li Lu, MD, PhD7
1 Stanford University, Stanford, California, United States
2 Stanford University, Stanford, California, United States
3 Stanford University, Stanford, California, United States
4 Stanford University, Stanford, California, United States
5 MD Anderson Cancer Center, Houston, Texas, United States
6 Stanford University, Stanford, California, United States
7 Stanford University, Stanford, California, United States
Recurrent mutations of SF3B1, an essential RNA splicing factor and constitutive U2 small nuclear ribonucleoprotein (snRNP) component, are established oncogenic drivers in myelodysplastic syndromes (MDS) and related disorders, chronic lymphocytic leukemia (CLL) and select solid tumors. Mutant SF3B1 is most studied for aberrant RNA splicing, which canonically results in the disruption of protein coding sequences (CDS) to cause loss of function. We recently showed that its missplicing of untranslated regions (UTRs) can instead yield gain of function (Sekrecki, Blood 2026). Further, coupling between U2 snRNP and the cleavage and polyadenylation (CPA) machinery is well documented: CPA subunits contact U2 snRNP directly (Kyburz, Mol Cell 2006), SF3B1 copurifies abundantly with the 3' end processing complex (Shi, Mol Cell 2009), and SF3B1 depletion drastically remodels RNA polyadenylation (Li, PLoS Genet 2015). Whether oncogenic SF3B1 alters 3' end formation (i.e. CPA) has nonetheless never been established. We hypothesized that SF3B1 mutations drive reproducible, stereotyped changes in CPA or 3' end formation in addition to their dysregulation of splicing catalysis.

We performed 3' end sequencing (3'-seq) on 161 libraries with matched RNA-seq: 28 SF3B1-mutant vs. 30 wildtype (WT) primary MDS/AML specimens, 4 mutant vs. 6 WT CLL, isogenic WT vs. SF3B1 mutant cells across 4 backgrounds (K562, NALM6, TF-1, MCF10A), and K562 SF3B1 knockdown (siSF3B1) as a loss-of-function (LOF) arm. Transcriptome-wide alternative polyadenylation (APA) geometry tracks cell identity rather than genotype, so no single contrast isolates SF3B1 mutations. Because each system carries independent confounds, we defined the SF3B1-attributable program as the alterations recurring across backgrounds, assessed by permutation.

Stringent cross-system recurrence for SF3B1 K700E over a 14,643-gene shared universe identified 19 genes altered in at least 3 of 4 backgrounds versus 3.7 expected (5.1-fold, p≤0.0005) and reproduced for SF3B1 K666N (1.6-fold, p=0.0025). The convergent set is AMD1, CAB39, CCDC154, CLCN7, DCTN5, DIP2A, GAA, GDI1, MAP4K2, NAE1, PINK1, PRDX6, PRPF4B, RIN1, SUN2, VEZF1, WDR33 and YWHAZ. Critically, these genes show zero overlap with ~30 of the most studied SF3B1 mis-splicing target genes, establishing that the 3' end program acts at least in part in trans and is not solely a byproduct of missplicing.

To test dose dependence, we built a per-sample splicing activity score on 30 mis-splicing target genes (WT 0.000±0.031, mutant 0.135-0.359; concordance between callers r=0.995). Within 28 MDS mutants, APA aberration scaled with expressed splicing dysfunction (r=+0.48, p=0.0095) but not with DNA variant allele frequency (r=-0.12, p=0.54), arguing the 3' end phenotype is mutation-driven rather than clonal. The underlying splicing ruler generalized to the rare SF3B1 G742D allele in a non-leukemic iPSC-erythroid lineage, where 10 of 30 targets were confirmed by 2 tools at matched coordinates across 3 differentiation stages.

Comparing oncogenic mutation with SF3B1 depletion (siSF3B1) revealed opposite architecture at the 2 layers. Against a gain-of-function ceiling of r=0.88 (K700E versus K666N), LOF-mutant concordance in splicing collapsed toward 0 as effect size increased, whereas 3' end concordance rose to r=+0.34 (65% same direction, n=383). Mutant SF3B1 therefore perturbs splicing predominantly by gain-of-function but 3' end formation with a substantial dosage/LOF component, explaining why the APA program is disjoint from the cis mis-splicing program.

All-peak analysis further uncovered a distinct, allele-general class of intronic polyadenylation events, recurrent at 8.9-fold over chance and non-overlapping with the 3'UTR program. These are coding-altering and coupled to intron retention at the same intron, exemplified by CCDC106 and by WDR33, itself a core CPA specificity factor. Separately, CPSF6 (CFIm68) shortened reproducibly in primary MDS (q=2e-3) and was unaffected by siSF3B1, marking it as a gain-of-function-specific candidate mediator.

Oncogenic SF3B1 thus reshapes transcript 3' end formation through a reproducible, dose-responsive and largely splicing-independent program, mechanistically split between a dominant dosage-like component and a small neomorphic set nominating CPSF6, WDR33 and CCDC106. This defines a layer of SF3B1 pathobiology invisible to splicing analysis and a candidate future therapeutic axis in SF3B1-mutant neoplasms.
Sydney Xin-Li Lu, MD, PhD: No relevant conflicts of interest to declare.