Induction of Acquired Resistance towards EGFR Inhibitor Gefitinib in a Patient-Derived Xenograft Model of Non-Small Cell Lung Cancer and Subsequent Molecular Characterization

Abstract

In up to 30% of non-small cell lung cancer (NSCLC) patients, the oncogenic driver of tumor growth is a constitutively activated epidermal growth factor receptor (EGFR). Although these patients gain great benefit from treatment with EGFR tyrosine kinase inhibitors, the development of resistance is inevitable. To model the emergence of drug resistance, an EGFR-driven, patient-derived xenograft (PDX) NSCLC model was treated continuously with Gefitinib in vivo. Over a period of more than three months, three separate clones developed and were subsequently analyzed: Whole exome sequencing and reverse phase protein arrays (RPPAs) were performed to identify the mechanism of resistance. In total, 13 genes were identified, which were mutated in all three resistant lines. Amongst them the mutations in NOMO2, ARHGEF5 and SMTNL2 were predicted as deleterious. The 53 mutated genes specific for at least two of the resistant lines were mainly involved in cell cycle activities or the Fanconi anemia pathway. On a protein level, total EGFR, total Axl, phospho-NFκB, and phospho-Stat1 were upregulated. Stat1, Stat3, MEK1/2, and NFκB displayed enhanced activation in the resistant clones determined by the phosphorylated vs. total protein ratio. In summary, we developed an NSCLC PDX line modelling possible escape mechanism under EGFR treatment. We identified three genes that have not been described before to be involved in an acquired EGFR resistance. Further functional studies are needed to decipher the underlying pathway regulation.

Keywords: EGFR inhibition; NSCLC; PDX; acquired resistance; reverse phase protein array; whole exome sequencing.

Figure 1

Quantification of EGFR and p-EGFR expression in a panel of NSCLC PDX models. (a) EGFR expression was determined by IHC on tissue micro arrays (TMAs) of the NSCLC PDX panel: sections were incubated with anti-human EGFR antibody (1:36) overnight at 4 °C, followed by DAB staining and hematoxylin counterstaining. Digitalized images of the IHC slides were evaluated to determine the percentage of EGFR positive areas using an inhouse software. The computerized analysis was used to quantify the EGFR expression using color classification and morphological image processing techniques. (b) The expression of EGFR and p-EGFR was determined by RPPA in a subset of the NSCLC PDX panel (n = 27). The samples were printed onto nitrocellulose covered microscope slides in five serial solutions and two replicates per dilution. Arrays were labeled with specific antibodies listed in Table 4. Median normalized data were used to compare expression levels between groups of samples. The ratio between phosphorylated and total protein was calculated by calculating the difference between the log-transformed phospho-protein expression and the log-transformed total protein expression. LXFA 677 showed high levels of EGFR and mean levels of p-EGFR. (c) LXFA 677 was implanted subcutaneously in NMRI nude mice and treatment started when a median tumor volume of 250 mm³ was achieved. Animals were assigned to the respective treatment arms. Dosing and schedule of the compounds are shown in Table 2. Treatment duration is indicated as lines on top of the figure. LXFA 677 showed EGFR dependent growth due to its sensitivity towards anti-EGFR monoclonal antibody, Cetuximab, as well as selective tyrosine kinase inhibitor (TKI), Erlotinib. Of note, the multi-TKI Sorafenib exhibited only marginal activity. (d) LXFA 677 was implanted subcutaneously in NMRI nude mice and treatment started when a median tumor volume of 250 mm³ was achieved. Animals were assigned to the respective treatment arms. Dosing and schedule of the compounds are shown in Table 2. Treatment duration is indicated as lines on top of the figure. LXFA 677 displayed a dose-dependent sensitivity towards treatment with Gefitinib.

Figure 2

Induction of resistance in NSCLC PDX LXFA 677 in vivo. (a): Tumor growth curves for individual tumors under different doses of Gefitinib. The orange line is indicating the duration of the treatment. The respective dose per day is shown above the line. (b): Tumor growth curves for individual tumors after re-implantation of tumors emerging under constant Gefitinib treatment. The orange line is indicating the duration of the treatment. The respective dose per day is shown above the line. (c): Determination of copy number variation (left diagram) and mRNA expression level (right diagram) was determined for three resistant sublines and the treatment naïve NSCLC PDX LXFA 677. The relative expression was determined by using 2.0 (copy number) and 1.0 (mRNA expression level) as default for the treatment naïve line.

Figure 3

Results of whole exome sequencing analysis of four different LXFA 677 sublines. (a): Mutational landscape of the four LXFA 677 sublines with more than 85% (n = 504) of genes with mutations being shared across clones. (b): Hierarchical clustering of genes with differing (=at least two; n = 53) mutation status. The whole exome sequencing was performed using the variant effect predictor (VEP). Candidate mutations were annotated and filtered considering only variants with moderate or high protein impact and those being rare in healthy populations (<1% in gnomAD). Mutations shared by all three resistant clones were annotated with protein functions using SIFT and Polyphen predictions from SNPnexus as depicted in detail in Table S2.

Figure 4

The proteomic landscape was determined by RPPA analysis of four different sublines of LXFA 677. The median corrected normalized values were used for all subsequent data analyses. The logarithmic fold change (logFC) was calculated comparing the sum of all resistant lines with the treatment naïve tumor. (a): Hierarchical clustering of average protein expression across five dilutions. (b): Hierarchical clustering of average phosphor- to total protein ratios across five dilutions.