Multiplexed CRISPR In Vivo Editing of CLL Loss-of-Function Lesions Models Transformation of Chronic Lymphocytic Leukemia into Richter's Syndrome

Although we have gained a wealth of knowledge from large-scale DNA sequencing studies across blood cancers, we still know little about the functional interplay of the discovered putative drivers in the generation of chronic lymphocytic leukemia (CLL) and its transformation into Richter's syndrome (RS). We have previously observed that CRISPR-Cas9 in vivo B-cell editing of common CLL loss-of-function (LOF) lesions (Atm, Trp53, Chd2, Birc3, Mga, Samhd1) can increase in vitro B cell fitness, but is not sufficient to sustain in vivo B cell survival after 12 months post-transplant.

We therefore asked whether combinatorial introduction of mutations was required for CLL development. To this end, we generated transplant lines by in vitro engineering of early stem and progenitor cells (Lineage^- Sca-1⁺ c-kit⁺ [LSK]) from MDR^-/-Cd19-Cas9 donor mice (animals expressing Cas9-GFP in a B-cell restricted fashion and the leukemogenic homozygous MDR lesion, mimicking del(13q)) with pooled lentivirus expressing sgRNAs against the 6 genes of interest and the mCherry marker. Engineered LSKs were then re-transplanted into sub-lethally irradiated immune-competent CD45.1 or immune-deficient NSG recipients (n=35/strain). Parallel control cohorts of equal size were generated by transducing LSKs with a pool of 6 non-targeting sgRNAs. Disease development (B220⁺CD5⁺Igk⁺ cells) was assessed by flow cytometric analysis of bi-monthly peripheral bleeds, starting at 4 months post-transplant, and flow cytometry/IHC were utilized to classify tumors at euthanasia. Analysis of PCR-based targeted NGS of peripheral blood edited tumor cells (GFP⁺mCherry⁺) was performed utilizing CRISPResso software to assess presence of the 6 LOF mutations.

We observed incidence of circulating CLL in 28/35 (80%) CD45.1 and 27/35 (77%) NSG mice, whereas only 5/35 (14%) CD45.1 and 4/35 (11.4%) NSG from the non-targeting control cohort developed CLL-like disease (P<0.0001, both strains), consistent with the expected penetrance of MDR. Analysis of lymphoid organs at euthanasia allowed identification of 3 disease presentations, namely 'pattern A' (CLL-like), 'pattern B' (co-presence of CLL and RS), and 'pattern C' (RS-like). Pattern C was predominantly of DLBCL histology (with 1 instance of HL in CD45.1), and characterized by circulating large cell disease and increased tumor cell infiltration of spleen, bone marrow and lymph nodes (P<0.001, all compartments), compared to pattern A. Disease onset (P=0.005) and overall survival (P<0.0001) was shorter in NSG recipients compared to CD45.1, suggesting a role for the immune-microenvironment in controlling progression in CD45.1 hosts.

To determine the genetic composition of the 55 leukemias/lymphomas (n=22, 11 and 22, with patterns A, B and C), we interrogated the LOF mutational burden at euthanasia. We observed a median number of 4 LOF mutations (range:1-6), and a high overall frequency of Trp53 lesions (58%). The other 5 drivers were less prevalent (Mga: 26%; Chd2: 21%; Samhd1: 17%; Birc3 and Atm: 13%). Trp53 mutations were predominantly clonal (≥90% indels, P<0.0001), while Birc3 and Atm were most commonly subclonal (<90%, P≤0.05). We observed increased numbers of clonal drivers in pattern B/C, as compared to pattern A (P≤0.01). Trp53 and Mga were enriched in B/C tumors (P<0.05) and frequently co-occurred, consistent with the recurrent TP53 losses and MYC gains observed in human RS (Mga is a negative regulator of MYC signaling). In 9 mice, either concomitant presence of large and small cells was identifiable in blood or sequential CLL and RS samples were available; we observed both linear (5/9) and branched (4/9) evolution, with RS cells acquiring mutations in DNA-damage response genes in addition to Trp53 (i.e. Atm and/or Samhd1), which may underlie increased genomic instability, a typical feature of human RS.

In conclusion, we demonstrate that combinatorial in vivo modeling of CLL-LOF mutations leads not only to CLL development, but also to RS, thus establishing a faithful framework for analysis of genetic and microenvironmental determinants of disease transformation. We are now interrogating genome-wide mutational patterns and clonal architecture of CLL vs. RS, while analyzing their microenvironmental composition. These novel models provide a unique platform to discern critical insights into RS pathogenesis and to discover RS-specific therapeutic vulnerabilities.