Comprehensive Bulk and Single Cell Transcriptomic Characterization of SF3B1 Mutation Reveals Its Pleiotropic Effects in Chronic Lymphocytic Leukemia

Large-scale sequencing efforts have identified SF3B1 as arecurrently mutated gene in chronic lymphocytic leukemia (CLL). While SF3B1 mutations have been associated with adverse clinical outcome in CLL, mechanistic understanding of its role in the oncogenic phenotype remains lacking. We therefore undertook a comprehensive transcriptomic characterization of CLL in relation to SF3B1 mutation status at both bulk and single cell levels.

We first profiled bulk mature poly-A selected RNA by sequencing (RNA-seq) from 37 CLLs (13 SF3B1 wild-type, 24 mutated). After identifying and classifying splice alterations using the tool JuncBASE, we found SF3B1 mutation to be associated with increased alternative splicing, with the most pervasive changes in 3' splice site selection. 304 alternatively spliced events were significantly associated with SF3B1 mutation, 4 of which we validated by qRT-PCR in 20 independent CLL samples with known SF3B1 mutation status. We further identified 1963 differentially expressed genes (q < 0.2) associated with SF3B1 mutation. By gene set enrichment analysis, SF3B1 mutation appeared to impact a variety of cancer and CLL-associated gene pathways, including DNA damage response, apoptosis regulation, chromatin remodeling, RNA processing, and Notch activation (q < 0.01). ~20% of these gene sets were also found to be significantly enriched for genes exhibiting alternative splicing in association with SF3B1 mutation.

As SF3B1 acts at the level of pre-mRNA, we also performed bulk RNA-seq with total RNA libraries generated from 5 CLLs (2 SF3B1 wild-type, 3 with the common K700E mutation). We again observed an enrichment of 3' splice site changes, along with ~30% overlap of differentially expressed genes, and ~16% overlap of enriched gene sets with the aforementioned poly-A data analysis. One differentially over-expressed gene associated with SF3B1 mutation unique to this total RNA data analysis and validated by total RNA qPCR of independent CLL samples was TERC, an essential RNA component of telomerase that serves as a replication template during telomeric elongation. TERC is a non-polyadenylated transcript and thus was undetected by our previous poly-A selected RNA-seq and by targeted qRT-PCR of oligo dT-generated cDNA. Recent reports have highlighted the involvement of the spliceosome in telomerase RNA processing, and shorter telomere length of CLLs with SF3B1 mutation. Thus, although further investigation will be needed, our analyses suggest a potential mechanism by which SF3B1 mutation contributes to aberrant regulation of telomerase activity.

Since SF3B1 is commonly found as a subclonal mutation in CLL, and because signals obtained from bulk analyses reflect only the average characteristics of the population, we assessed the transcriptomic effects of SF3B1 mutation in single cells within a subset of CLL cases. We developed a novel and sensitive microfluidic approach that performs multiplexed targeted amplification of RNA to simultaneously detect somatic mutation status, gene expression (96 targets), and alternative splicing (45 targets) within the same individual cell for 96 to 288 cells from 5 patients with different SF3B1 mutations. From the same patient sample, single cells with SF3B1 mutation generally exhibited increased alternative splicing for events identified from the bulk analysis, thus confirming the association of SF3B1 mutation with altered splicing at the single cell level. Different SF3B1 hotspot mutations within the HEAT repeat domains exhibited similar patterns of alternative splicing while a mutation outside of the repeat domain did not. Furthermore, we confirmed significant changes in gene expression between SF3B1 wild-type and mutant cells of target genes involved in the Notch pathway (NCOR2), cell cycle (CDKN2A, CCND1) and apoptosis (TXNIP). Consistent with these analyses, functional studies with overexpression of full-length mutated SF3B1 in a hematopoietic cell lines confirmed the modulation of these pathways by this putative CLL driver. Our high-resolution single cell analysis further uncovered 2 transcription factors strongly associated with SF3B1 mutation but not previously appreciated (KLF3 and KLF8).

Our comprehensive transcriptomic analysis thus highlights SF3B1 mutation as an efficient mechanism by which a complex of changes relevant to CLL biology are generated that can contribute to disease progression.