Identification of Genotype-Specific Therapeutic Vulnerabilities By Comparative Dynamic BH3 Profiling Analysis of Human and Murine CLL

Mouse models that reflect human cancer genetics are valuable tools for dissecting disease mechanisms and for discovering novel therapeutic vulnerabilities, although such models for chronic lymphocytic leukemia (CLL) have been largely lacking until now. The most commonly used CLL model, the TCL1 mouse, is based on constitutive B-cell restricted expression of the human TCL1 oncogene, but does not reflect CLL genetics per se. Novel mouse models, such as the MDR mice (modeling del[13q]) and our recently developed Sf3b1^mut/Atm^del double mutant mice, are able to recapitulate the expression of CLL driver mutations and develop CLL-like disease with a rather delayed onset (~18 months), mimicking the indolent nature of human CLL.

To understand the utility and limitations of these murine CLL models as preclinical platforms for drug screening, we compared BH3 profiles of murine and genetically-matched human CLLs in response to a panel of drugs including 3 BCL-2 signaling inhibitors (BCL-2i, including BCL-2 inhibitor ABT-199, BCL-2/XL inhibitor ABT-263, BCL-XL inhibitor A133 and MCL1 inhibitor S63845) and 7 BCR signaling inhibitors (BCRi, including PI3K inhibitors idelalisib, duvelisib, umbralisib, the mTOR kinase inhibitor AZD8055, the MEK1/MEK2 inhibitor trametinib, the BTK inhibitor ibrutinib and the BRAF inhibitor sorafenib). BH3 profiling is based on exposure of cellular mitochondria to synthetic peptides that mimic the BH3 domains of pro-apoptotic BCL-2 family members followed by measurement of mitochondrial permeabilization using flow-cytometry based detection of cytochrome-c loss. We measured apoptotic priming (BIM peptide), and the dependence on BCL-2 or MCL-1 (BAD and MS1 peptides, respectively) after drug exposure. The assay was miniaturized into 384 well plates, where OP9 stromal cells (used as a support to limit human and murine CLL spontaneous apoptosis) were coated at a 1:10 ratio compared to CLL cells, before treatment with drugs (experiments were performed in duplicate).

As a starting point, to determine whether primary CLLs with different mutated drivers have distinct

therapeutic vulnerabilities, we profiled 6 primary human CLLs (4 carrying del[13q] and 2 carrying del[13q], [11q] and mutant SF3B1). We observed a general increase in BCL-2 dependence in response to the aggregate set of 7 BCRi as compared to 3 BCL-2i (ABT-199, ABT-263 and A133) (p<2 ×10^-16, the Van Elteren test). The van Elteren test is a type of stratified Wilcoxon-Mann-Whitney test for comparing different treatments accounting for strata. S63845 functioned differentially from the other 3 BCL-2i such that it also induced strong BCL-2-dependent response in all patients (p<2 ×10^-16, the Van Elteren test). Notably, the 2 primary samples carrying del(13q)/(11q)-SF3B1 were more resistant to BCRi compared to those with del(13q), as evidenced by reduced BIM signals (p= 0.01, the Van Elteren test). In addition, they also exhibited reduced dependence on BCL-2 for survival (p=1.0 ×10^-13, the Van Elteren test). The results suggest the presence of a distinct genotype-specific mechanism of resistance to BCR signaling inhibition.

Similar analyses were performed using splenocytes from NSG recipient mice, transplanted with murine CLLs carrying either MDR or Sf3b1^mut/Atm^del mutations. Reduced activity of the MCL1 inhibitor S63845 was observed in murine CLLs compared to human CLLs (p=0.024, Mann-Whitney U test), consistent with the higher affinity of this compound for human, rather than murine MCL1. Notably, Sf3b1^mut/Atm^del CLL cells were more resistant to the panel of BCRi than MDR cells (p=0.046, the Van Elteren test) and their dependence on BCL-2 for survival after BCRi was likewise diminished (p=0.018, the Van Elteren test), similar to human CLLs carrying del(13q)/(11q)-SF3B1. These results support the use of the Sf3b1^mut/Atm^del line as a valuable platform for the exploration of therapeutic vulnerabilities of primary CLLs carrying the same genetic lesions.

Altogether, these results suggest that both MDR and Sf3b1^mut/Atm^del mouse lines are valuable tools for predicting therapeutic sensitivities of genetically matched primary CLLs. We are currently extending this approach to the testing of additional drugs, and to the validation of the most relevant compounds in larger cohorts of primary CLLs.