EGFR-TKI Resistance Due to BIM Polymorphism Can Be Circumvented in Combination with HDAC Inhibition

The EGF receptor (EGFR) tyrosine kinase inhibitors (TKI), gefitinib and erlotinib, have shown marked therapeutic effects against non–small cell lung cancer (NSCLC) with EGFR-activating mutations, such as exon 19 deletions and L858R point mutations (1). About 20% to 30% of patients, however, show intrinsic resistance to EGFR-TKIs despite having tumors harboring these EGFR mutations. In addition, patients who respond initially later develop acquired resistance to EGFR-TKIs after varying periods of time (2). Among the molecular mechanisms associated with acquired resistance to EGFR-TKIs are (i) gatekeeper mutations in EGFR (i.e., a T790M second mutation), (ii) activation of bypass signaling caused by Met amplification or hepatocyte growth factor overexpression, (iii) transformation to small-cell lung cancer, and (iv) epithelial-to-mesenchymal transition (3, 4). Several therapeutic strategies, including new generation EGFR-TKIs and the combination of an EGFR-TKI and a Met-TKI, have been evaluated clinically in patients with EGFR-mutant NSCLC who acquired resistance to EGFR-TKIs (2). The mechanisms of intrinsic resistance, however, remain poorly understood.

Recently, a BIM deletion polymorphism was reported to be a novel mechanism of intrinsic resistance to EGFR-TKIs (5). BIM, also called BCL2L11, is a proapoptotic protein and a member of the Bcl-2 family. Gene products (such as BIM_EL, BIM_L, and BIM_S) with a BH3 domain, which is essential for apoptosis induction, antagonize antiapoptotic proteins (such as Bcl-2, Bcl-X_L, and Mcl-1) and activate proapoptotic proteins (such as BAX and BAK), thereby inducing apoptosis (6, 7). Activation of BAX and BAK induce cytochrome c release into the cytoplasm and result in activation of the caspase cascade (8). BIM is pivotal in apoptosis induced by EGFR-TKIs in EGFR-mutant NSCLC cells (9). The expression and degradation of BIM is regulated mainly by the MEK-ERK pathway (10). The BIM deletion polymorphism is relatively common in East Asian populations (12.9%), with 0.5% of individuals being homozygous for this deletion. During the transcription of BIM, either exon 3 or exon 4, the latter of which encodes the BH3 domain, is spliced out due to the presence of a stop codon and a polyadenylation signal within exon 3 (11). The BIM deletion polymorphism involves the deletion of a 2903 bp fragment in intron 2 and results in the preferential splicing of exon 3 over exon 4, generating a BIM isoform that lacks the BH3 domain (5). A retrospective analysis in patients with EGFR-mutant NSCLC showed that progression-free survival (PFS) following EGFR-TKI treatment was significantly shorter in patients with the BIM polymorphism (6.6 months) than with wild-type BIM (11.9 months; ref.5). Another study in patients with EGFR-mutant NSCLC treated with EGFR-TKIs also reported that PFS was significantly shorter in patients with BIM-low (4.3 months) than BIM-high (11.3 months) expressing tumors (12), suggesting that reduced expression of BIM with a BH3 domain is associated with an unfavorable response to EGFR-TKIs. To date, however, no therapeutic strategy has yet been developed for patients with EGFR-mutant NSCLC with low BIM expression.

Histone deacetylase (HDAC) is an enzyme that regulates chromatin remodeling and is crucial in the epigenetic regulation of various genes (13). Many compounds targeting HDAC have been developed, including vorinostat, an HDAC inhibitor approved by the United States Food and Drug Administration (FDA) for the treatment of patients with cutaneous T-cell lymphoma (14). In mantle cell lymphoma (MCL) cell lines and in cells from patients with MCL, vorinostat induced histone hyperacetylation on promoter regions and consequent transcriptional activation of proapoptotic BH3-only genes, including BIM (15). Using in vitro and in vivo models, we assessed whether the combination of vorinostat and gefitinib restored the expression of BIM protein with a BH3 domain in EGFR-mutant NSCLC cells with the BIM polymorphism and overcame EGFR-TKI resistance associated with this polymorphism.

The NSCLC cell lines, PC-9, HCC827, and HCC2279, all of which have EGFR mutations, were obtained from Immuno-Biological Laboratories Co., ltd., the American Type Culture Collection (ATCC), and Dr. John Minna (University of Texas Southwestern Medical Center, Dallas, TX), respectively. PC-3 cells, established from a Japanese female patient with NSCLC and with an exon 19 deletion in EGFR, and differing from the prostate cancer cell line PC-3 (ATCC CRL1435), were purchased from Human Science Research Resource Bank (JCRB0077: http://cellbank.nibio.go.jp/~cellbank/cgi-bin/search_res_det.cgi?DB_NUM=1&ID=252). PC-3 and the other 3 cell lines were maintained in Dulbecco's Modified Eagle's Medium (DMEM) and RPMI-1640 medium, respectively, each supplemented with 10% FBS and antibiotics. All cells were passaged for less than 3 months before renewal from frozen, early-passage stocks. Cells were regularly screened for mycoplasma using a MycoAlert Mycoplasma Detection Kit (Lonza). The cell lines were authenticated at the laboratory of the National Institute of Biomedical Innovation (Osaka, Japan) by short tandem repeat analysis. Vorinostat and gefitinib were obtained from Selleck Chemicals and AstraZeneca, respectively.

Genomic DNA was extracted from cells using DNeasy Blood and Tissue Kits (Qiagen), according to the manufacturer's protocol. Total RNA was extracted from cells using RNeasy PLUS Mini kits (Qiagen). PCR methods were used to detect the BIM deletion polymorphism in the samples and the level of expression of BIM isoforms (5).

Cells (3 × 10³) were seeded into each well of 96-well, white-walled plates, incubated overnight, and treated with the indicated compounds or vehicle [dimethyl sulfoxide (DMSO)] for 48 hours. Cellular apoptosis was analyzed with Caspase-Glo 3/7 assay kits (Promega), which measure caspase-3/7 activity, and PE-Annexin V Apoptosis Detection Kits (BD Biosciences, in accordance with the manufacturers' directions.

Apoptotic cells in tumor xenografts were detected by terminal deoxynucleotidyl transferase–mediated nick end labeling (TUNEL) staining, using the DeadEnd Fluorometric TUNEL system (Promega), according to the manufacturer's protocol.

Duplexed Stealth RNAi (Invitrogen) against BIM and Stealth RNAi-negative control low GC Duplex #3 (Invitrogen) were used for RNA interference (RNAi) assays as described (4). The siRNA target sequences were 5′-CAUGAGUUGUGACAAAUCAACACAA-3′ and 5′-UUGUGUUGAUUUGUCACAACUCAUG-3′ for BIM #1, and 5′-UGAGUGUGACCGAGAAGGUAGACAA-3′ and 5′-UUGUCUACCUUCUCGGUCACACUCA-3′ for BIM #2.

Western blotting was conducted with antibodies against phospho-EGFR (Tyr1068), Akt, phospho-Akt (Ser473), cleaved PARP, cleaved caspase-3, histone H3, acetylated histone H3 (Lys27), BIM, and β-actin (Cell Signaling Technology); and against phospho-Erk1/2 (Thr202/Tyr204), Erk1/2, and EGFR (R&D Systems). Blots were subsequently incubated with horseradish peroxidase-conjugated secondary antibodies specific to mouse or rabbit immunoglobulin G, with signals detected by enhanced chemiluminescence (Pierce Biotechnology).

Male BALB/cAJcl-nu/nu mice, ages 5 to 6 weeks, were obtained from CLEA Japan Inc and injected subcutaneously into their flanks with cultured tumor cells (5 × 10⁶ cells/0.1 mL/mouse). When tumor volumes reached 100 to 200 mm³, the mice were randomized and treated once daily with gefitinib and/or vorinostat. Each tumor was measured in 2 dimensions, and the volume was calculated using the formula: tumor volume (mm³) = 1/2 × length (mm) × width (mm)². All animal experiments complied with the Guidelines for the Institute for Experimental Animals, Kanazawa University Advanced Science Research Center (approval No. AP-081088).

Between group differences were analyzed by one-way ANOVA. All statistical analyses were conducted using GraphPad Prism Ver. 4.01 (GraphPad Software, Inc.), with P < 0.05 considered statistically significant.

We first examined the BIM deletion polymorphism in EGFR-mutant NSCLC cell lines by PCR. PC-9 and HCC827 had wild-type alleles, with a PCR product 4.2 kb in size. Consistent with a previous report (5), HCC2279 cells were heterozygous for the BIM deletion polymorphism, with PCR products 4.2 kb (wild-type) and 1.3 kb (2.9 kb deletion polymorphism) in size. Among the 7 additional cell lines with EGFR mutations (Supplementary Table S1), PC-3 was heterozygous for the BIM deletion polymorphism (Fig. 1A). Western blot analyses reveal that the expression of the proapoptotic BIM protein was markedly lower in PC-3 and HCC2279 than in PC-9 and HCC827 cells. Analysis of BIM isoform transcripts showed that cells with the BIM polymorphism expressed more exon 3- than exon 4-containing transcripts (Supplementary Fig. S1A and S1B). Treatment with gefitinib enhanced BIM expression, caspase-3/7 activities, and apoptosis in PC-9 and HCC827 cells much more than in PC-3 and HCC2279 cells (Fig. 1B; Supplementary Fig. S1C, S1D, and S2). Moreover, gefitinib did not increase caspase-3/7 activity in PC-9 and HCC827 cells treated with BIM siRNA (Fig. 1C), indicating the crucial role of BIM in apoptosis induction in EGFR-mutant NSCLC cells treated with EGFR-TKI. These observations clearly showed that EGFR-mutant NSCLC cells with the BIM deletion polymorphism are much less sensitive to gefitinib, as shown by induction of apoptosis, than cells with wild-type BIM.

Because HDAC inhibition modulates the expression of various genes, including proapoptotic molecules (13), we hypothesized that the HDAC inhibitor, vorinostat, may sensitize EGFR-mutant NSCLC cells with the BIM polymorphism to gefitinib. In EGFR-mutated NSCLC cell lines, including PC-3 and HCC2279 cells, vorinostat dose dependently increased the expression of acetylated histone H3 and BIM with the BH3 domain (Fig. 2A, Supplementary Fig. S3A). We further explored whether the addition of vorinostat to gefitinib induced apoptosis in EGFR-mutant NSCLC cells with the BIM polymorphism (Fig. 2B and D). In HCC827 and PC-9 cells, which contain only wild-type BIM, gefitinib inhibited downstream signaling, including the phosphorylation of EGFR, Erk, and Akt, resulting in apoptosis, as shown by the expression of cleaved PARP and cleaved caspase-3. The further addition of vorinostat augmented BIM expression and caspase-3/7 activity. In PC-3 and HCC2279 cells, which contain the BIM polymorphism, however, treatment with gefitinib alone induced minimal apoptosis, although the phosphorylation of EGFR, Erk, and Akt was inhibited, whereas the combination of vorinostat and gefitinib markedly increased the expression of BIM, as well as of cleaved PARP and cleaved caspase-3 (Fig. 2B and Supplementary Fig. S3B). This combination also augmented caspase-3/7 activity compared with that of gefitinib or vorinostat alone (Fig. 2D and Supplementary Fig. S3C), but this activation of caspase-3/7 was inhibited by knockdown of BIM (Supplementary Fig. S4A and S4B). Conversely, overexpression of BIM_EL itself stimulated caspase-3/7 activities in cells with the BIM polymorphism, with these activities further enhanced by gefitinib treatment (Supplementary Fig. S4C and S4D). These results indicate that BIM mediates the activation of caspase-3/7 induced by gefitinib and vorinostat. Analysis of BIM transcripts revealed that vorinostat alone induced BIM mRNA, which was enhanced by the inclusion of gefitinib. Moreover, vorinostat treatment preferentially induced transcripts containing exon 4 over those containing exon 3 (Fig. 2C). These results indicate that the combination of vorinostat and gefitinib inhibits HDAC and increases the expression of BIM protein with the BH3 domain, thereby sensitizing EGFR-mutant NSCLC cells with the BIM polymorphism to apoptosis in vitro.

We next determined the in vivo efficacy of vorinostat and gefitinib. Gefitinib alone almost completely shrunk xenograft tumors induced by HCC827 cells (Fig. 3A). Although gefitinib monotherapy prevented the enlargement of tumors produced by PC-3 cells, which harbor the BIM polymorphism, it did not induce their complete regression, indicating that PC-3 cells remained less susceptible to gefitinib in vivo. Under these experimental conditions, vorinostat monotherapy inhibited tumor growth slightly, whereas the combination of vorinostat with gefitinib resulted in marked tumor shrinkage (Fig. 3B). None of the mice treated with these agents showed any macroscopic adverse effects, including loss of body weight (data not shown).

To clarify the mechanisms by which vorinostat and gefitinib act in vivo, we assessed tumor-cell apoptosis by TUNEL staining. Gefitinib treatment increased the number of apoptotic cells in HCC827 tumors but had little effect on PC-3 tumors (Fig. 4A and B), indicating that EGFR-mutant NSCLC cells with the BIM polymorphism are refractory to gefitinib-induced apoptosis in vivo as well as in vitro. Importantly, although vorinostat alone had little effect on apoptosis, the combination of vorinostat and gefitinib induced marked apoptosis in PC-3 tumors (Fig. 4A and B). Western blot analyses showed that gefitinib induced cleavage of caspase-3 in HCC827, but not in PC-3, tumors. In PC-3 tumors, treatment with gefitinib or vorinostat had little effect on caspase-3 cleavage, whereas their combination increased BIM expression and the cleavage of caspase-3 (Fig. 4C and D). These findings indicate that the combination of vorinostat and gefitinib increases BIM protein expression and induces tumor-cell apoptosis, thereby shrinking tumors produced by EGFR-mutant NSCLC cells with the BIM polymorphism.

EGFR-mutant NSCLC cells with the BIM deletion polymorphism show impaired generation of BIM with the proapoptotic BH3 domain, as well as resistance to EGFR-TKI–induced apoptosis (5). We have shown here that treatment of cells with the combination of vorinostat, a HDAC inhibitor, and gefitinib, an EGFR-TKI, restored the expression of BIM protein with a BH3 domain (predominantly BIM_EL), induced apoptosis, and overcame gefitinib resistance in vitro and in vivo.

Although vorinostat preferentially induced expression of BIM containing the BH3 domain, its exact mechanisms of action remain unclear. The wild-type allele may be more susceptible to the effects of HDAC inhibition than the deletion allele due to differences in the acetylation status of these alleles. Alternatively, vorinostat may affect the splicing process, resulting in the production of exon 4- rather than exon 3-containing transcripts from the deletion polymorphism allele as HDAC has been found to affect the splicing of RNA (16).

Vorinostat has been shown to induce the expression of several genes other than BIM (13). However, we found that BIM was pivotal not only for gefitinib-induced apoptosis but also when combined with vorinostat. Moreover, the combination of vorinostat and gefitinib increased BIM expression and markedly induced apoptosis in PC-3 and HCC2279 cells. Collectively, these findings strongly suggest that vorinostat promotes gefitinib-induced apoptosis in EGFR-mutant NSCLC cells with the BIM polymorphism, primarily by increasing BIM expression. Several other mechanisms, including inhibition of epigenetic modifications leading to a drug-tolerant state (17) and transition of cancer cells from a resistant mesenchymal state to an E-cadherin–expressing epithelial state (18) may be also involved.

Both the BIM polymorphism and EGFR mutations are more prevalent in East Asian than in Caucasian populations. Few East Asian patients with EGFR-mutant NSCLC show a complete response to EGFR-TKIs (1). This incomplete response, including intrinsic resistance, may be due, in part, to low BIM expression associated with the BIM polymorphism (6). Our preclinical data indicate that vorinostat increases BIM even in BIM-wild type EGFR–mutant NSCLC cells. However, a clinical trial with erlotinib and entinostat, an HDAC inhibitor, in unselected patients with NSCLC, more than 65% of whom were Caucasian, failed to show therapeutic benefits (19). These findings suggest that the combination of vorinostat and an EGFR-TKI should be tested in selected patients with NSCLC with EGFR mutations and the BIM polymorphism.

Resistance to EGFR-TKIs associated with the BIM deletion polymorphism may be overcome by treatment with BH3 mimetics, such as ABT-737 (5). Although ABT-737 antagonized antiapoptotic proteins, such as Bcl-2 and Bcl-X_L, it did not antagonize the antiapoptotic protein Mcl-1, which is overexpressed in NSCLC (20), suggesting that the effects of BH3 mimetics may be limited to overcoming EGFR-TKI resistance caused by the BIM polymorphism in NSCLC. BH3 mimetics are being evaluated in early-phase clinical trials but are not ready for use in clinical practice. In contrast, vorinostat has been approved by the FDA for the treatment of patients with advanced primary cutaneous T-cell lymphoma (15). Therefore, the combination of gefitinib and vorinostat could easily be tested clinically.

The BIM polymorphism can be detected in formalin-fixed paraffin-embedded tumor tissues and peripheral blood (5). Moreover, a convenient and easy access PCR screening method can detect this polymorphism in circulating DNA from serum (Supplementary Fig. S5A and S5B). As the BIM polymorphism is a germline alteration, it can be assayed in serum obtained at any time point. Collectively, our findings illustrate the importance of clinical trials testing the ability of combinations of vorinostat and EGFR-TKIs to overcome EGFR-TKI resistance associated with the BIM polymorphism in patients with EGFR—mutant NSCLC.

T. Nakagawa is an employee of Eisai Co., Ltd. for oncology research. Y. Hasegawa received research funding from Chugai Pharmaceutical Co., Ltd., Merck Sharp & Dohme Corp., AstraZeneca, and TAIHO Pharmaceutical Co., Ltd. S. Yano received honoraria from Chugai Pharmaceutical Co., Ltd. and AstraZeneca and received research funding from Chugai Pharmaceutical Co., Ltd., Kyowa Hakko Kirin Co., Ltd., and Eisai Co., Ltd. No potential conflicts of interest were disclosed by the other authors.

Conception and design: T. Nakagawa, S. Takeuchi, S. Nanjo, S. Yano

Development of methodology: T. Nakagawa, S. Takeuchi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Nakagawa, D. Ishikawa, Y. Hasegawa

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Nakagawa, S. Yano

Writing, review, and/or revision of the manuscript: T. Nakagawa, S. Takeuchi, H. Ebi, M. Sato, Y. Hasegawa, Y. Sekido, S. Yano

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Yamada, T. Sano, M. Sato, Y. Sekido

Study supervision: S. Takeuchi, Y. Sekido, S. Yano

The authors thank Dr. John Minna (University of Texas Southwestern Medical Center) for the HCC2279 cells.

This study was supported by Grants-in-Aid for Cancer Research (21390256 to S. Yano; 11019957 to S. Takeuchi), Scientific Research on Innovative Areas “Integrative Research on Cancer Microenvironment Network” (22112010A01 to S. Yano), and Grant-in-Aid for Project for Development of Innovative Research on Cancer Therapeutics (P-Direct) from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan.