Genomic and biological study of fusion genes as resistance mechanisms to EGFR inhibitors

Abstract

The clinical significance of gene fusions detected by DNA-based next generation sequencing remains unclear as resistance mechanisms to EGFR tyrosine kinase inhibitors in EGFR mutant non-small cell lung cancer. By studying EGFR inhibitor-resistant patients treated with a combination of an EGFR inhibitor and a drug targeting the putative resistance-causing fusion oncogene, we identify patients who benefit and those who do not from this treatment approach. Through evaluation including RNA-seq of potential drug resistance-imparting fusion oncogenes in 504 patients with EGFR mutant lung cancer, we identify only a minority of them as functional, potentially capable of imparting EGFR inhibitor resistance. We further functionally validate fusion oncogenes in vitro using CRISPR-based editing of EGFR mutant cell lines and use these models to identify known and unknown drug resistance mechanisms to combination therapies. Collectively, our results partially reveal the complex nature of fusion oncogenes as potential drug resistance mechanisms and highlight approaches that can be undertaken to determine their functional significance.

Fig. 1. Clinical response to combination therapy aimed at overcoming fusion-mediated drug resistance.

a EGFR L858R mutant adenocarcinoma acquired ALK fusion. Combination of erlotinib and alectinib evoked a response in lung and brain lesions. b Combined use of osimertinib and trametinib successfully shrank the EGFR exon 19 deletion (del19) lung adenocarcinoma, which had acquired an ESYT2-BRAF fusion. c Positron emission tomography (PET)-CT images show the progression of EGFR del19 adenocarcinoma with a putative GKAP1-NTRK2 fusion following treatment with osimertinib plus larotrectinib.

Fig. 2. Comprehensive analyses of all fusions in EGFR mutant lung cancer.

Prospectively collected genetic data on fusions from all cancer patients, and mutational data from patients with non-small cell lung cancer (NSCLC; detected by OncoPanel at Dana Farber Cancer Institute) were combined. Unique fusions identified in patients with EGFR L858R or deletion in exon 19 (del19) were classified into five groups.

Fig. 3. Clinical annotation of putative fusions in EGFR mutant lung cancer.

a Clinical course of patients with fusions, which were detected by DNA-based next generation sequencing (NGS) OncoPanel, following treatment with first or third generation (1G or 3G) EGFR tyrosine kinase inhibitors (TKIs). Asterisks indicate samples that did not undergo initial evaluation by OncoPanel. Bar charts show total durations of EGFR-TKI treatment before and after the fusions were detected. RNA from these samples were submitted for further RNA sequencing (blue). b Clinical course of patients with fusions that were detected prior to treatment with EGFR-TKI.

Fig. 4. Comparison of putative fusions by DNA-seq and RNA-seq.

a Schema of integrating data obtained by DNA-based next generation sequencing (NGS) OncoPanel, and RNA-sequencing (RNA-seq), to pick out expressed oncogenic fusions. b The candidate fusions detected by each fusion caller from RNA-seq data were aligned with putative oncogenic fusions detected by OncoPanel. Fusions detected by all three fusion callers were noted. Asterisk points to data obtained from the leftover RNA sample from the RNA-based anchored multiplex PCR for targeted NGS. c Circos plots showing discordant reads with ABL1, FGFR1, BRAF, DLG1, or NRG1 detected by RNA-seq.

Fig. 5. CRISPR-modified in vitro models, with EGFR mutation and oncogenic fusions.

a The structures of the fusion oncogene and the location of designed single guide RNAs (sgRNAs) are shown, with representative sequencing chromatograms of fusion cDNA derived from bulk CRISPR-modified EGFR mutant PC-9 cells. e: exon; UTR: untranslated region. b Colony formation assay after 1 week of treatment with osimertinib, using parental PC-9 cells or CRISPR-modified PC-9 cells that express fusion oncogenes. c Breakpoints of CCDC6-RET fusion in bulk CRISPR-modified PC-9 cells. d Expression of RET protein in permeabilized parental or CRISPR-modified PC-9 models, evaluated with use of flowcytometry. PC-9^CCDC6-RET bulk cells were selected with 100 nM osimertinib for 1 week, and then a single clone was picked. Ratio of RET-expressing cells in each of four categories are indicated. Pseudo-color represents cellular density. e Results of cell viability assay after 72 h of osimertinib treatment. Half maximal inhibitory concentrations (IC50s) are shown for parental PC-9 cells and for single clones from CRISPR-modified PC-9^CCDC6-RET models, selected with or without 1 week of exposure to 100 nM osimertinib (n = 3 biological replicates, mean ± s.d.). f Knockdown of RET, BRAF, FGFR3, or ALK genes in CRISPR-modified PC-9 clones after 48 hours of siRNA treatment, is shown by western blot analyses. WT: wild type. g Knockdown of RET, BRAF, FGFR3, or ALK genes by siRNA in CRISPR-modified PC-9 cells resensitized them to 1 μM osimertinib (n = 3 biological replicates, mean ± s.d., two-sided t test, **p < 0.01). Source data of e–g are provided as a Source Data file.

Fig. 6. Effective combination therapies in CRISPR-modified fusion models.

a The half maximal inhibitory concentrations (IC50s) of osimertinib and indicated drugs in parental PC-9 and in single clones from CRISPR-modified PC-9 models, after 72 h of treatment. For each fusion model, we used two single clones. Each dot indicates the mean of biological triplicate data. b Synergistic inhibitory effects of osimertinib and individual drugs in CRISPR-modified PC-9 models as shown by Combenefit (n = 2 biological replicates, mean). Pseudo-color represents synergy effects. c Western blot analyses following 48 h of treatment with 10 nM trametinib or 0.5 μM of other drugs, as indicated. d Proliferation of cells (top) and induction of caspase-3/7 (bottom) evaluated by Incucyte live-cell imaging (n = 3 biological replicates, mean ± s.d.). Source data of a–d are provided as a Source Data file.

Fig. 7. Mechanisms of acquired resistance to combination therapy in a patient and in in vitro models.

a Summary of mechanisms of acquired resistance to inhibition of EGFR and fusion genes in a patient, as well as in CRISPR-modified PC-9 models. osi: osimertinib. b Crystal structure of RET in a complex with ponatinib or selpercatinib. Modeling the G810S mutation shows the resulting clashes with selpercatinib, but not with ponatinib. c Western blot analyses following 48 h of treatment with 0.5 μM of the indicated drugs in PC-9^CCDC6-RET models that had acquired RET G810S mutation, after being exposed to osimertinib plus pralsetinib. d Cell viability assay after 72 h of treatment of PC-9 model harboring RET G810S mutation (n = 3 biological replicates, mean ± s.d.). e Copy number of YAP1 evaluated by quantitative PCR. RNaseP was used as an internal control (n = 4 biological replicates, mean ± s.d.). f Cell viability assay following 72 h of treatment with indicated drugs in PC-9 model harboring YAP1 amplification (n = 3 biological replicates, mean ± s.d.). g Expression of YAP1, CTGF, and CYR61 evaluated by qPCR; GUSB was used as an internal control. PC-9 models were treated for 48 hours with siRNA, 10 μM of MYF01-37, or 1 μM of other indicated drugs (n = 3 biological replicates, mean ± s.d.). h Western blot analyses following 48 h of treatment with siRNA, 10 μM of MYF01-37, or 1 μM of other drugs, as indicated. i Results of Autophagy Flux Assay, using 15 μM of chloroquine or 1 μM of indicated drugs in PC-9 model with YAP1 amplification. Ratios of LC3B-II to β-actin were quantified by Image J software (n = 1). Source data of c–i are provided as a Source Data file.