Quantitative modelling of B cell signaling in aggressive B cell lymphoma

The B cell receptor (BCR) signaling, required for the survival and maturation of B cells, is a major deregulated pathway in B cell lymphomas. Several mutations are known to enhance the tonic BCR signal in Burkitt lymphomas (BL) or to mimic an activated receptor in some diffuse large B cell lymphomas (DLBCL). While the proximal events and kinases of the BCR signaling are well studied, less is known about the interactions of downstream effector pathways. In order to reach a less abstract description of this pathway we decided to employ the Modular Response Analysis-based modelling approach STASNet on more directly connected phosphorylation data. Using a Luminex proteomics platform we measured the phosphorylation of fourteen signaling and effector proteins after activating the B cell receptor (BCR) signaling and/or using inhibitors directed against seven intracellular signaling molecules in two independent Burkitt lymphoma cells (BL-2, BL-41). Application of a novel two step STASNet modelling pipeline unveiled novel feedback and crosstalk structures in the BCR-driven signaling network, including a crosstalk from p38 which negatively regulates MEK/ERK-activity in the presence of active BCR. Using Western Blot measurements, we verified and further characterized the p38-MEK/ERK crosstalk and demonstrated that it is a general mechanism in BL cells. Global Tandem Mass Tag (TMT) phosphoproteomic analysis on BL-2 cells found PI3K to be a major mediator of B-cell receptor signaling and manifested the p38-ERK crosstalk directly on ERK phosphosites and indirectly on seven bona fide ERK targets. We also noticed that the BL-2-learned pathway network structure was transferable to perturbation data from closely related BL-41 cells. Moreover, - compared to a literature-derived network - the BL-2-developed predictive pathway proved also to be a better starting point for network development in cells with aberrant BCR signaling in two representative cell lines derived from Diffuse large B cells (HBL-1, OCI-LY3). This indicates that models trained on activated B-cells are highly informative to be transferred to closely related cancer signaling network.