NOVEL THERAPEUTIC STRATEGIES IN LUNG CANCERSTRONG BIOLOGICAL BIAS FOR ALK INTRON 19 BREAKPOINTS IN NON-SMALL LUNG CANCER

Chromosomal rearrangements represent one of the most common genomic aberration in cancer and chromosomal translocation that lead to oncogenic kinase activation are observed in many solid tumors. These tumors express activated fusion kinases that drive the initiation and progression of the malignancy. In the last 15 years, several oncogenic drivers, including various oncogenic fusions, have been detected in non-small cell lung cancer (NSCLC) paving the way to targeted therapy with tyrosine kinase inhibitors. EML4-ALK translocations are detected in 4-8 % of NSCLC and define a specific subtype of oncogene-driven lung cancer. While different EML4-ALK variants are defined by different breakpoints in the EML4 gene, most frequently located in intron 6 or 13, ALK breakpoint is almost invariably in intron 19. Rare reports describe EML4-ALK translocations with breakpoints in intron 17 or intron 18 of the ALK gene. Despite all these ALK breakpoints potentially generate oncogenic fusions, the reasons of this strong imbalance toward intron 19 (exon 20) breakpoints in ALK positive NSCLC are currently unknown. The aim of this study is to investigate the mechanisms that underlie ALK translocation in NSCLC. Eight different EML4-ALK fusions were either overexpressed with a retroviral system or generated with CRISPR/Cas9 lentivirus from the endogenous loci. EML4 exons 1-6 (E6) or exons 1-13 (E13) with different ALK exons: E6;A18, E6;A19, E6;A20 or E6;A21 and E13;A18, E13;A19, E13;A20, E13;A21. The fusions E6;A21 and E13;A21 contained an early stop codon not producing functional proteins. These EML4-ALK fusions were expressed in NIH3T3 and PC9 by retroviral vectors or engineered in PC9 (an EGFR-dependent lung cancer cell line that is sensitive to inhibition with the EGFR inhibitor osimertinib) and BEAS-2B cells by CRISPR/Cas9. High-throughput genome-wide translocation sequencing was performed on engineered PC9 cells. NSG immunodeficient mice were used for in vivo experiments with NIH3T3 and BEAS-2B cells. The retroviral overexpression system showed that all EML4-ALK fusion proteins were expressed leading to an actively phosphorylated ALK in NIH3T3 and PC9 cells, except E6;A21 and E13;A21 fusions where no protein was detected, as expected. Accordingly, all active variants were able to induce NIH3T3 cell transformation and tumor formation in vivo and conferred resistance to osimertinib in PC9 cells. In contrast, when EML4-ALK translocations were induced by CRISPR/Cas9 in PC9 cells, only E6;A20 and E13;A20 variants rapidly overcame osimertinib inhibition while other variants rescued osimertinib-inihibited cells slowly and less efficiently. Surprisingly, the E6;A20 and E13;A20 fusions were invariably found in all resistant clones independently of the translocation induced by CRISPR/Cas9. Similarly, all tumors formed in vivo by BEAS-2B cells contained the E6;A20 or E13;A20 variants irrespective of the translocation induced originally. By cloning translocation junctions at large scale with HTGTS, we found that clones initially forced to break into ALK intron 17 or intron 18 eventually acquired a second breakpoint in ALK intron 19. These data show that all EML4-ALK fusion variants were equally oncogenic when overexpressed. In contrast, when EML4-ALK variants were generated from the endogenous loci there was a strong selection bias toward ALK fusions originating in intron 19 suggesting that intron 19 variants have the strongest oncogenic potential in lung epithelial cells.