Epithelial-to-Mesenchymal Transition Antagonizes Response to Targeted Therapies in Lung Cancer by Suppressing BIM

Abstract

Purpose: Epithelial-to-mesenchymal transition (EMT) confers resistance to a number of targeted therapies and chemotherapies. However, it has been unclear why EMT promotes resistance, thereby impairing progress to overcome it.Experimental Design: We have developed several models of EMT-mediated resistance to EGFR inhibitors (EGFRi) in EGFR-mutant lung cancers to evaluate a novel mechanism of EMT-mediated resistance.Results: We observed that mesenchymal EGFR-mutant lung cancers are resistant to EGFRi-induced apoptosis via insufficient expression of BIM, preventing cell death despite potent suppression of oncogenic signaling following EGFRi treatment. Mechanistically, we observed that the EMT transcription factor ZEB1 inhibits BIM expression by binding directly to the BIM promoter and repressing transcription. Derepression of BIM expression by depletion of ZEB1 or treatment with the BH3 mimetic ABT-263 to enhance "free" cellular BIM levels both led to resensitization of mesenchymal EGFR-mutant cancers to EGFRi. This relationship between EMT and loss of BIM is not restricted to EGFR-mutant lung cancers, as it was also observed in KRAS-mutant lung cancers and large datasets, including different cancer subtypes.Conclusions: Altogether, these data reveal a novel mechanistic link between EMT and resistance to lung cancer targeted therapies. Clin Cancer Res; 24(1); 197-208. ©2017 AACR.

Figure 1. Mesenchymal status and BIM (BCL2L11) levels are inversely correlated

(A) Dot plot of the top nine genes correlated to BIM: known mesenchymal related genes (red), epithelial related genes (blue) or other genes (black) were plotted against p values. (B–C) BIM mRNA levels were separated as low BIM and medium (med)/high BIM and levels of the mesenchymal marker Vimentin were plotted (each dot represents a unique solid tumor cancer cell line). For Fig. 1B left panel (low n=171, med/high n =686) and right panel (low n=114, med/high n =459). For Fig. 1C, left panel (low n=37, med/high n =150) and right panel (low n=26, med/high n =103). For Fig. 1B, P<0.0001 in all comparisons by Mann-Whitney U test, for Fig. 1C P<0.05 in all comparisons by Mann-Whitney U test. Data were obtained from the CCLE (Fig. 1B, left panel and Fig. 1C, left panel) and Oncomine® (Fig. 1B, right panel and Fig. 1C right panel). (D) BIM mRNA levels were separated as low BIM and medium (med)/high BIM and levels of the mesenchymal marker Vimentin were plotted (each dot represents a unique lung cancer patient sample). For both Fig. 1D, left panel (low n=20, med/high n=80), and Fig. 1D, right panel (low n=24, med/high n=97), the data were downloaded from the R2: Genomic Analysis and Visualization Platform and P<0.05 by Mann-Whitney U test. Please note: low BIM was considered the bottom 20% expressing cell lines and tumors.

Figure 2. EMT results in loss of BIM, loss of an apoptotic response to EGFR inhibitor, and loss of sensitivity to EGFR inhibitor

(A–C) HCC4006 and H1975 cells EGFR mutant NSCLC cell lines were transduced with ER-TWIST were either conditionally activated with 4-OHT incubation (ER-TWIST+4-OHT) or not (ER-TWIST) and (A) lysates from the cells were probed with the indicated antibodies, (B) RNA isolated and relative ZEB1 (left panel) and BIM (right panel) levels determined by qPCR (C) treated with 1µM gefitinib (HCC4006) or 1µM WZ4002 (H1975) for 72 hours and apoptosis quantified. (D and E) H1975 ER-TWIST and ER-TWIST+4-OHT cells were treated with 1µM WZ4002 for (D) 6 hours and lysates were probed with the indicated antibodies or (E) 24h and cell G₁ cell cycle increase quantified by FACS. (F) HCC4006 (4006) and H1975 (1975) ER-TWIST and ER-TWIST+4-OHT cells were treated with 1µM gefitinib (4006) or 1µM WZ4002 (1975) for 5 days and cells stained with crystal violet. (G and H) H1975 ER-TWIST+4-OHT cells were further transduced with lentiviral-particles containing a doxycycline-inducible BIM expression plasmid (pTREXBIM) and treated with 1µM WZ4002 with the indicated concentrations of doxycycline and (G) lysates were probed with the indicated antibodies or (H) apoptosis analysis performed by FACS. Error bars are +S.D. for (B), (E) and (H), and +S.E.M. for (C). For (B, C and H), the indicated data points were repeated in triplicate and the results are representative of three independent experiments; for (E), the indicated data points were repeated in triplicate and the results are representative of two independent experiments.

Figure 3. ZEB1 directly binds and suppresses the BIM (BCL2L11) promoter to promote resistance to EGFR inhibition

(A–C) Cells were treated with or without 1µM PF-299804 or 1µM WZ4002 for (A) crystal violet staining (B) Western blotting (C) RNA determination of H1975 parental cells (par), H1975 R1 cells (R1), and H1975 R2 cells (R2). The technical triplicates from duplicate qPCR run were used to determine the percent input of each amplicon. (D) ZEB1 binding profile within a 757 bp region encompassing the promoter and first exon of the BIM gene in wild type ("ZEB1 1975") and mutated ("ZEB1 1975R2") cells. Density and pileup of sequencing reads are shown. The Y-axes on both panels are equal scale. The "ZEB1 Hepg2" panel shows ZEB1 binding in Hepg2 cell line from ENCODE (separated by a dotted line to signify different source of data) (E) ChIP of ZEB1 at the BIM promoter. ChIP was performed against ZEB1 (grey bars) or a nonspecific IgG antibody (black bars). (left) Diagram of the first two exons of BCL2L11. E-Box motifs, each 250 nt upstream of their respective exons, are denoted by a black circle. Quantification of amplicons are shown for each E-Box for the (middle) residency of ZEB1 at the E-Box immediately upstream of the BCL2L11 transcriptional start site (Promoter E-Box) and (right) residency of ZEB1 at an E-Box sequence immediately upstream of Exon 2 of BCL2L11 (Intronic E-Box). (F–H) Transduction of virus containing ZEB1 CRISPR guide RNAs (CRISPR/Cas9 G1 and G2) in H1975 R2 cells followed by (F) Western blotting, (G) apoptosis induced by 1µM WZ4002 as measured by PI/Annexin staining and FACS and (H) 72 hours cell viability assay treated with the indicated doses of WZ4002. (I–K) Transduction of virus containing shSC or shZEB1 plasmids in H1975 ER-TWIST+4-OHT cells followed by (I) Western blotting, (J) apoptosis induced by 1µM WZ4002 as measured by PI/Annexin staining and FACS and (K) 72 hours cell viability assay treated with the indicated doses of WZ4002. Error bars are +S.D. for (C), (H) and (K), and +S.E.M. for (E), (G) and (J). For (C and G), the indicated data points were repeated in triplicate and the results are representative of three independent experiments. For (J), the indicated data points were repeated in triplicate and the results are representative of three independent experiments. For (H and K), the indicated data points were repeated in quadruplicate and the results are representative of at two independent experiments.

Figure 4. Pharmacological re-sensitization of EMT cells via ABT-263

(A) Apoptosis measured by PI/Annexin staining and FACS following treatment with 1µM of the indicated drugs and (B) 72 hours cell viability assay with the indicated doses of WZ4002 +/− 1µM ABT-263 of H1975 R1 cells and H1975 R2 cells. (C) Western blots of BIM immunoprecipitation and whole cell lysates of H1975 R2 cells. (D) Mice harboring H1975 ER-TWIST+4-OHT tumors were treated as indicated (** P<0.01, * P<0.05 compared to WZ4002). (E) Five epithelial (left) and five mesenchymal (right) KRAS mutant NSCLC cell lines were lysed and underwent Western blotting and probed with the indicated antibodies. (F) KRAS mutant NSCLCs were transfected with either a short-interfering (si) targeting a scramble (Sc) or ZEB1 sequence, lysed and underwent Western blotting for the indicated antibodies. For (B), the indicated data points were repeated in quadruplicate and the results are representative of at two independent experiments.