Genome-Wide Analysis of a Novel Murine Model of Chronic Lymphocytic Leukemia

Large-scale cancer sequencing efforts worldwide have yielded numerous novel cancer drivers; however, how these genetic alterations functionally lead to cancer remains largely unknown. An indispensible approach for establishing the causal features of disease is through in vivo animal models. In chronic lymphocytic leukemia (CLL), only limited mouse models are currently available and most do not reflect the genetics of human CLL. Studies of whole-exome sequencing (WES) using CLL samples have consistently pointed to the common co-occurrence of mutations in the RNA splicing factor gene SF3B1 and mutations in the DNA damage response gene ATM or deletion of chromosome 11q (del(11q), whose minimally deleted region encompasses ATM). We therefore asked whether this combination of traits would be productive of CLL in mice.

To this end, we modeled the effects of these combined alterations by crossing mice with conditional knockout of Atm and mice with a conditional knock-in allele of SF3B1 mutation (Sf3b1-K700E). We achieved B cell-restricted expression of heterozygous Sf3b1 mutation and Atm deletion by breeding these mice with CD19-Cre homozygous transgenic mice. We found that in vivo co-expression of these two mutations in B cells, but not of either single lesion alone, led to clonal expansion of CD19+CD5+ B cells in blood, marrow and spleen (at low penetrance) in aged (18 to 24-month old) but not young mice. These malignant cells could be propagated by in vivo passaging, with detectable disease within 4 weeks following transfer, thus making this mouse line amenable to further drug discovery and biologic investigations.

To better understand how Sf3b1 mutation and Atm deletion synergistically contribute to CLL, we asked if RNA level changes are present in the double mutant mice. We performed transcriptome sequencing of splenic B cell RNA collected from age-matched mice that either express wild-type, or singly mutant alleles of Sf3b1 or Atm, or doubly mutant alleles with or without CLL-like disease (n=2-6 samples, per group). Using the tool JuncBASE, we classified and quantified splice variants associated with the different genetic alterations. Consistent with prior findings in human CLL, we observed that the splice variants in micewith mutated Sf3b1 alone (without CLL) were highly enriched at 3' splice sites (27 of 77 splice variants, t-test q<0.05, absolute ΔPSI >10%). On the other hand, mice with Atm single deletion displayed an RNA splicing pattern with enrichment of alternative first and last exons (11 and 12 of 52, chi-squared test, p=4.5 x 10^-4). B cells with the combined Sf3b1 and Atm mutations displayed a combination of splicing patterns that comprised of both alternative 3' splice variants, as well as alternative first and last exons. Moreover, we identified unique CLL splice variants in genes (Setdb2, Phf11c) previously demonstrated to be associated with CLL. We further investigated the differential gene expression between B cells from double mutant mice with and without CLL-like disease. We identified 1,875 CLL-specific genes (DESeq2, q<0.01). Gene set enrichment analysis (GSEA) of these genes indicated their involvement in cellular processes such as IL2-STAT5 signaling and the interferon gamma response, both pathways implicated in human CLL. In parallel, we asked if there are DNA level changes in the doubly mutant mice. We examined the mutation rate in DNA derived from splenic B cells collected from mice with a singly mutated allele of Sf3b1 or Atm, or with doubly mutated alleles with and without CLL-like disease through comparison against matched germline DNA from kidney by whole-genome sequencing. Preliminarily, we have observed that co-expression of Sf3b1 mutation and deletion of Atm results in a higher mutation rate compared to cells with only single mutation.

In summary, we have generated a genetically-engineered murine model that faithfully recapitulates human CLL genetics. This is the first demonstration that expression of putative CLL driver events identified from unbiased genome-wide sequencing indeed initiates CLL-like disease. Genome-wide DNA and RNA analysis using this model has revealed that altered RNA splicing, dysregulation of gene expression, and genomic instability all contribute to CLL leukemogenesis. We anticipate that further dissection of this murine model will shed light on mechanistic understanding of cooperation between Atm deletion and SF3B1 mutation in CLL.