Different Classes of ABL1 Fusions Activate Different Downstream Signalling Nodes

Background: Philadelphia-like acute lymphoblastic leukaemia (Ph-like ALL) is a high-risk subtype of ALL driven by a range of tyrosine kinase and cytokine receptor rearrangements. ABL1-class rearrangements (ABL1, ABL2, CSF1R and PDGFRB) account for 17% of Ph-like ALL cases in children, and are clinically important to identify as they can be therapeutically targeted with tyrosine kinase inhibitors (TKIs). While the p190 BCR-ABL1 fusion is well described, less is known about the function and downstream signalling by rare ABL1 fusions. We identified a rare ABL1 fusion, SFPQ-ABL1, in a paediatric B-ALL patient using RNA-sequencing. This fusion lacks the ABL1 Src-homology-3 (SH3) and part of the SH2 domain, which are retained in BCR-ABL1. Other ABL1 fusions, RCSD1-ABL1 and SNX2-ABL1, have a similar structure. In this work we have utilised phosphoproteomics and Stable Isotope Labelling by Amino Acids in Cell Culture (SILAC), as well as in vitro and in vivo models, to determine differential signalling pathways between SFPQ-ABL1 and BCR-ABL1.

Methods: We cloned SFPQ-ABL1 from patient cDNA, and engineered SFPQ-ABL1 and BCR-ABL1 fusions to include or delete the SH2 and SH3 domains. We performed proliferation and viability assays to assess the ability of these fusions to transform Ba/F3 cells and test sensitivity to TKIs. We performed total phosphopeptide and phosphotyrosine enrichments and utilised mass spectrometry to identify the phosphoproteome activated by canonical SFPQ-ABL1 and BCR-ABL1. Over representation analysis was performed on phosphopeptides significantly differing between BCR-ABL and SFPQ-ABL (Log fold change cut-off > 2.5) using the Gene Ontology (GO) knowledge base under the biological process category. Furthermore, we compared the phosphoproteome of canonical SFPQ-ABL1 to SFPQ-ABL1 with the SH2 and SH3 domains reintroduced (SFPQ-ABL1+SH). We have also developed novel mouse models, using syngeneic transplantation, of SFPQ-ABL1 and SNX2-ABL1 driven leukaemia.

Results: SFPQ-ABL1 expressing Ba/F3 cells are sensitive to cell death induced by TKIs that block ABL1. Interestingly, while SFPQ-ABL1 and BCR-ABL1 both effectively blocked apoptosis, SFPQ-ABL1 was less able to drive cytokine-independent proliferation. Phosphoproteomic analysis showed that BCR-ABL1 and SFPQ-ABL1 differentially activate downstream signalling pathways, including SH-binding proteins. Hierarchical clustering of phosphopeptides quantified from cells expressing canonical BCR-ABL1, SFPQ-ABL1, and SFPQ-ABL1+SH, demonstrated that BCR-ABL1 and SFPQ-ABL1+SH were more similar to each other than to SFPQ-ABL1. SFPQ-ABL1 expression resulted in phosphorylation of proteins involved in RNA processing, metabolism and splicing, suggesting that SFPQ region of SFPQ-ABL1 also contributes to signalling.

Conclusions: In this study, we have utilised phosphoproteomics for the unbiased identification of signalling nodes that are required for the function of different classes of ABL fusions. We have developed novel in vitro and in vivo models to further understand how these fusions function to drive leukaemia. Our data also suggests that ABL1 fusion partners play a role beyond dimerization and transphosphorylation of the kinase domains in oncogenic signalling, but further study is needed to establish the contribution to leukaemogenesis. Establishing signalling pathways that are critical to the function of rare ABL1 fusions may inform clinical approaches to treating this disease.