SF3B1 Mutation Alters The Selection Of 3' RNA Splice Sites In Chronic Lymphocytic Leukemia

Large-scale cancer sequencing efforts have enabled the discovery of previously unknown paths to carcinogenesis. In chronic lymphocytic leukemia (CLL), these initiatives have identified highly recurrent mutations in the core spliceosome component SF3B1 at restricted sites (i.e. 50% at K700E), implicating a role of altered RNA splicing in CLL. Mutated SF3B1 has been associated with adverse clinical outcome in CLL and a higher incidence in refractory/relapsed CLL. However, mechanistic insights into downstream paths affected by mutated SF3B1 and its role in the oncogenic phenotype are still lacking.

To this end, we first examined if mutated SF3B1 alters known disrupted pathways in CLL. We tested the effect of expressing constructs encoding full-length mutated vs. wild-type SF3B1 into hematopoietic cell lines on established pathway readout assays but observed no discernible changes in cell cycle, Notch or Wnt signaling. On the other hand, given the key role of SF3B1 in pre-mRNA splicing, we hypothesized that features of altered splicing associated with this driver might provide important mechanistic insights. We therefore generated RNA-seq data from 22 CLL cases (9 with mutant and 13 with wild-type SF3B1; with 45-128 million reads mapped per sample) and identified splicing effects associated with SF3B1 mutation using an update of the computational approach, JuncBASE (Junction Based Analysis of Spliced Events).

SF3B1 is the key catalytic component of the U2 snRNP, which is critical in the initial formation of the core spliceosome at 3’ exon/intron boundaries. Indeed, we identified pervasive changes in 3’ splice site selection in which 642 introns showed altered 3’ splice site selection associated with SF3B1 mutation (p < 0.05). This effect was relatively specific to 3’ splice sites as only 362 introns showed changes in 5’ splice site selection associated with SF3B1 mutation (p < 0.05). Alterations in 3’ splice site choice resulted in both partial exon skipping and partial intron retention, equivalently. Although splicing changes were significant and consistent amongst SF3B1 mutated CLLs, the majority of these splicing changes (63%) was subtle, such that there was a <10% change in the proportion of transcripts selecting a given alternative 3’ splice site. Additionally, we did not observe specific sequences at altered 3’ splice sites in association with SF3B1 mutation, in contrast to the 3’ splice sites motifs associated with mutation in the splicing factor U2AF1 (common to AML), suggesting different mechanisms of spliceosome alterations between the 2 spliceosomal component mutations.

Altered splicing might be expected to generate specific splice variants that could in turn affect cellular circuitry. To identify likely candidates, we detected 192 genes with splicing changes strongly associated with SF3B1 mutation, using a conservative statistical approach (q < 0.1). As expected, these genes were significantly enriched for 3’ splice-site variants (160 of 192, p < 0.00001), 25 of 192 altered splicing events involved upregulation of isoforms virtually undetectable in CLL samples without SF3B1 mutation. We validated these candidates using three complementary approaches. First, an analysis of an independent RNA-seq dataset (5 CLLs with SF3B1 mutation, 67 with wildtype SF3B1) confirmed the set of the altered splicing events identified from the discovery cohort. Second, we detected 5 of 5 tested candidate SF3B1 mutation-associated splice variants (in C11orf2, GCC2, MAP3K7, TPP2, ZNF91) by quantitative real-time RT-PCR in 6 of 6 CLLs with SF3B1 mutation, but in none of wild type SF3B1 CLLs (n=3) or normal B cell samples (n=3). Finally, K562 cells transfected with mutated (K700E) SF3B1 revealed a significant increase in the candidate splice variants (by ∼20-fold) compared to K562 cells transfected with the wild type variant.

A detailed examination of the candidate list revealed several genes with known roles in B cell biology, e.g., BLNK, a critical adaptor of B cell receptor function involved in lymphoproliferation; USP15, a modulator of NFkB signaling. Genes with significantly altered splicing were enriched with immune system gene sets (q <0.0001) and with genes unregulated in uveal melanoma (q < 10^-7), which carries a high rate of SF3B1 mutation. These analyses will be used to guide selection of candidates for downstream functional assessment and to identify variants with highest oncogenic potential.