Genome-Wide DNA Methylation Analysis Reveals GABBR2 As A

Published OnlineFirst May 10, 2017; DOI: 10.1158/1078-0432.CCR-16-2688

Cancer Therapy: Clinical Clinical Cancer Research Genome-wide DNA Methylation Analysis Reveals GABBR2 as a Novel Epigenetic Target for EGFR 19 Deletion Lung Adenocarcinoma with Induction Erlotinib Treatment Xiaomin Niu1, Fatao Liu2,3, Yi Zhou4, Zhen Zhou1, Daizhan Zhou5, Ting Wang2,3, Ziming Li1, Xiangyun Ye1, Yongfeng Yu1, Xiaoling Weng6,7, Hong Zhang6,7, Junyi Ye6,7, Meilin Liao1, Yun Liu6, Zhiwei Chen1, and Shun Lu1

Abstract

Purpose: The past decade has witnessed the rapid development CBFA2T3 and GABBR2, were clearly validated. A same dif- of personalized targeted therapies in lung cancer. It is still unclear ferential methylated region (DMR) between exon 2 and exon whetherepigeneticchangesareinvolvedintheresponsetotyrosine 3ofGABBR2 gene was confirmed consistently in both kinase inhibitor (TKI) treatment in epidermal growth factor recep- patients. GABBR2 was significantly downregulated in EGFR tor (EGFR)-mutated lung cancer. 19 deletion cells, HCC4006 and HCC827, but remained Experimental Design: Methyl-sensitive cut counting sequenc- conserved in EGFR wild-type A549 cells after erlotinib treating (MSCC) was applied to investigate the methylation changes in ment. Upregulation of GABBR2 expression significantly res- paired tissues before and after erlotinib treatment for 42 days with cued erlotinib-induced apoptosis in HCC827 cells. GABBR2 partial response (PR) from stage IIIa (N2) lung adenocarcinoma was significantly downregulated, along with the reduction of patients (N ¼ 2) with EGFR 19 deletion. The Sequenom EpiTYPER S6, p-p70 S6, and p-ERK1/2, demonstrating that GABBR2 assay was used to validate the changed methylated candidate may play an important role in EGFR signaling through the genes. Up- or downregulation of the candidate gene was per- ERK1/2 pathway. formed to elucidate the potential mechanism in the regulation of Conclusions: We demonstrated that GABBR2 gene might be a erlotinib treatment response. novel potential epigenetic treatment target with induction erlo- Results: Sixty aberrant methylated genes were screened tinib treatment for stage IIIa (N2) EGFR 19 deletion lung adeno- using MSCC sequencing. Two aberrant methylated genes, carcinoma. Clin Cancer Res; 23(17); 5003–14. 2017 AACR.

Introduction Lung cancer has been the leading life-threatening cancer in the world (1). With the development of molecular biology, targeted 1 Department of Shanghai Lung Cancer Center, Shanghai Chest Hospital, Shang- therapy has been more and more involved in anticancer treat- hai Jiao Tong University, Shanghai, P.R. China. 2Department of General Surgery, ment. In the clinical treatment of non–small cell lung cancer Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China. 3Institute of Biliary Tract Disease, Shanghai Jiao Tong Univer- (NSCLC), targeted therapy against epidermal growth factor recep- sity School of Medicine, Shanghai, P.R. China. 4Institute for Nutritional Sciences, tor (EGFR) has gradually matured and significantly contributed to Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, the improvement of patient outcomes and quality of life (2). The Shanghai, P.R. China. 5Bio-X Institutes, Key Laboratory for the Genetics of results of the IRESSA Pan Asia Study (IPASS) make it common Developmental and Neuropsychiatric Disorders (Ministry of Education), Shang- practice to screen for EGFR mutations before NSCLC treatment, as hai Jiao Tong University, Shanghai, P.R. China. 6Institutes of Biomedical 75% of NSCLC patients carrying EGFR mutation benefit from Sciences, Fudan University, Shanghai, P.R. China. 7Key Laboratory of Molecular Medicine, The Ministry of Education, Department of Biochemistry and Molecular tyrosine kinase inhibitor (TKI) treatment (3). However, the ben- Biology, Fudan University Shanghai Medical College, Shanghai, P.R. China. efit of EGFR-TKI regimens for patients with stage IIIa NSCLC Note: Supplementary data for this article are available at Clinical Cancer remains uncertain. The goal of the treatment in locally advanced Research Online (http://clincancerres.aacrjournals.org/). NSCLC is to cure without long-term therapy related complica- tions. With the successful integration of TKIs in the treatment of X. Niu and F. Liu contributed equally to this article and share lead authorship. advanced NSCLC with mutated EGFR, it is reasonable to wonder Y. Liu, Z. Chen, and S. Lu are co-senior authors of this article. whether we could extend this benefit to locally advanced NSCLC Corresponding Author: Shun Lu, Department of Shanghai Lung Cancer Center, and actually improve long-term survival rates for these patients. Shanghai Chest Hospital, Shanghai Jiao Tong University, 241 Huaihai West Road, There are some clinical trials (NCT01822496, NCT00600587) Shanghai 200030, P.R. China. Phone: 86-21-62821990; Fax: 86-21-62804970; evaluating the value of induction erlotinib therapy before thora- E-mail: [email protected] cotomy or radiotherapy in stage IIIa (N2) EGFR-mutated NSCLC. doi: 10.1158/1078-0432.CCR-16-2688 However, an important question is why the recurrence rates after 2017 American Association for Cancer Research. complete surgical resection still remain as high as 70% (4).

www.aacrjournals.org 5003

Niu et al.

help to discover how epigenetic changes are involved in the Translational Relevance response to TKI therapy and further indicate new potential com- The is the first study to investigate the changes in DNA bined induction therapy in EGFR-mutated NSCLC. methylation before and after induction tyrosine kinase inhibitor (TKI) treatment for epidermal growth factor receptor Materials and Methods (EGFR)-mutated lung adenocarcinoma at the whole-genome Patients and samples level. A differential methylated region (DMR) of GABBR2 gene The patients from Shanghai Chest Hospital were confirmed to was confirmed in the two patients involved in the present have locally advanced stage IIIa-N2 lung adenocarcinomas with study. Upregulation of GABBR2 expression significantly res- mediastinal lymphadenopathies confirmed by mediastinoscopy cued erlotinib-induced apoptosis, which indicated that that were unresectable at the first diagnosis, carrying EGFR 19 GABBR2 gene might be a novel potential epigenetic treatment deletion mutations by direct sequencing. After two cycles of target with induction erlotinib treatment for stage IIIa (N2) induction erlotinib treatment (21 days for one cycle of treatment, EGFR 19 deletion lung adenocarcinoma. Our research pro- a total of 42 days), the patients exhibited a partial response (PR) vides a new theoretical basis for the epigenetic study of EGFR- according to the Response Evaluation Criteria in Solid Tumors mutated lung adenocarcinoma treatment and suggests that (RECIST) version 1.1 (13) and then subsequently received a targeting GABBR2 together with EGFR inhibition may radical sleeve lobectomy after multidiscipline discussion on improve clinical outcomes in patients with EGFR-mutated tumor board. The current study conformed to the Declaration lung adenocarcinoma. of Helsinki and was performed after approval by the Institutional Review Board (IRB) of Shanghai Chest Hospital (No. of ethics approval: KS1307). The two patients signed the IRB-approved written informed Considering the current clinical issue, our research is to investigate consents and were further enrolled for this study, allowing for whether epigenetic changes such as DNA methylation are the collection and genomic analysis of archived tissue speci- involved in the response to induction TKI treatment in EGFR- mens. The metastatic mediastinal lymph nodes before TKI mutated lung cancer. treatment were collected using mediastinoscopy, and post-TKI Recent studies have demonstrated that DNA methylation plays treatment metastatic mediastinal lymph nodes were collected a role in the targeted therapy of NSCLC (5–7). DNA methylation duringthesleevelobectomyoperation(N ¼ 2, a total of 4 status of Wnt antagonist genes from 155 NSCLC patients who samples). After surgery, the samples were collected and stored received EGFR-TKI treatment were investigated using methyla- at 80C immediately. A total of 4 samples (2 matched pairs) tion-specific PCR, showing that DNA methylation status of Wnt were used for whole-methylome screening and were then antagonist SFRP5 can predict the response to EGFR-TKI therapy in verified by a Sequenom EpiTYPER assay to double confirm the NSCLC, and methylated SFRP5 may contribute to shorter pro- changed methylated candidate genes. gression-free survival (PFS; ref. 8). The EGFR 19 deletion cell line PC-9, with the unmethylated promoter region of EGFR gene, was more sensitive to gefitinib compared with the other EGFR 19 Genomic DNA extraction deletion cell line H1650 with the methylated promoter region, Genomic DNA was isolated from the four samples using the which was "resistant" to gefitinib, suggesting that EGFR gene QIAamp DNA Mini Kit (Qiagen). A Thermo NanoDrop 2000 fi fl promoter methylation may be a potential mechanism for (Thermo Scienti c) and a Qubit 2.0 uorometer (Life Technol- acquired resistance to gefitinib (9). Methylation of death-associ- ogies) were used to detect the DNA concentrations. An Agilent fi ated protein kinase (DAPK) is reported to be a novel target with 2100 Bioanalyzer (Thermo Scienti c) was used to determine the cetuximab and erlotinib resistance (10). DNA methylation reg- DNA length and ensure the integrity. ulates gene expression without altering the nucleotide sequence, thereby potentially influencing the sensitivity of EGFR-mutated MSCC library construction lung cancer to targeted drugs. Currently, there is an ongoing phase For each of the samples, two MSCC libraries were constructed I clinical trial testing the use of vorinostat–gefitinib combined according to the description of the procedure by Guo and col- therapy on resistance due to BIM polymorphism (BH3-only leagues published in Nature Neuroscience (14). A detailed exper- proapoptotic member of the Bcl-2 protein family) in EGFR- imental procedure is provided in the Supplementary Material mutant lung cancer (NCT02151721). Vorinostat, a small-mole- (Supplementary File S1). cule inhibitor of histone deacetylase (HDAC), can epigenetically restore BIM function and death sensitivity to EGFR-TKIs, which MSCC sequencing and data analysis may provide some direction for the epigenetic treatment of EGFR- After the MSCC libraries were pooled, an Illumina HiSeq2000 mutated lung cancer. sequencing system (Illumina) was used to perform the sequenc- Now the popularity of next-generation sequencing (NGS) ing. A detailed procedure of the NGS data analysis is provided in technology makes it possible to study the epigenetic mechanism the Supplementary Material (Supplementary File S1). of targeted therapy at the genome-wide level, while previous Basic units of 200-bp nonoverlapping windows were used to studies were limited to only one or several genes (11, 12). In look for differential methylated regions (DMR). For each of these this study, one of the most commonly used whole-genome DNA units (2 sequenced CpGs), the average methylation level of the methylation sequencing technologies, methyl-sensitive cut count- CpGs involved was calculated. A P value was also assigned using ing sequencing (MSCC), was used to investigate the changes in the Fisher exact test. Regions were identified as DMRs if their DNA methylation before and after induction TKI treatment for detection P value was <0.05 and the changed average methylation EGFR-mutated lung cancer at the whole-genome level, which may level (DMSCC, either increase or decrease) was 20%. Pathway

5004 Clin Cancer Res; 23(17) September 1, 2017 Clinical Cancer Research

DNA Methylome Screen for Lung Adenocarcinoma with Erlotinib

and gene ontology (GO) analyses were performed using Database Flow cytometric assays for apoptosis for Annotation, Visualization and Integrated Discovery (DAVID) After being transfected with plasmid and incubated with or v6.7 software (15, 16). Gene structure was predicted by Webscipio without erlotinib, cells were digested by trypsin/EDTA (Gibco) 2.0 (17). and washed twice with ice-cold PBS. Following ﬁxation with 70% ethanol overnight at 4C, the cells were washed twice with PBS Sequenom EpiTYPER assay and then digested by 50 mg/mL RNase A in 500 mL of PBS at 37 C After the MSCC screening sequencing, the methylation of for 30 minutes. Next, the cells were stained with 20 mg/mL GABBR2 and CBFA2T3 exhibited changes post-erlotinib treat- propidium iodide (PI) for 30 minutes at 37 C. For apoptosis, ment (please see the Results section), which needed further cells were harvested using trypsin/EDTA and washed with PBS, validation. A detailed experimental procedure of Sequenom Epi- and then binding buffer was added to resuspend the cells. Fol- TYPER assay and the primers of the two genes, GABBR2 and lowing incubation with Annexin-V and PI staining according to CBFA2T3, were provided in the Supplementary Material (Sup- the manufacturer's protocol (Bio-Vision), cells were analyzed by plementary File S1). FACSCanto II ﬂow cytometer (BD Biosciences).

Cell culture and treatment by erlotinib IHC A549 (EGFR wild-type), HCC4006 (EGFR 19 deletion, IHC staining was performed using freshly cut paraffin-embed- DelL747-E749, i.e., A750P), and HCC827 (EGFR 19 deletion, ded tissues. Slides were incubated for 12 hours at 4C with DelE746-A750, i.e., A750P) cell lines were purchased from the GABBR2 antibody (at a dilution of 1:10, Ab75838, Abcam). The American Type Culture Collection (ATCC) and maintained in resulting mixture were then washed by PBS and incubated with RPMI1640 medium (HyClone) supplemented with 10% FBS K5007 antibody (DAKO) for 50 minutes at room temperature. (HyClone) and 100 U/mL of penicillin and streptomycin (Gibco) Signals were then visualized (DAKO DAB detection kit, K5007) as fi ina37C incubator with a humidi ed atmosphere of 5% CO2. per the manufacturer's instructions. Erlotinib powder was purchased from Santa Cruz Biotechnology. The cells were then incubated with or without erlotinib for 48 hours. Results Patient characteristics and therapeutic effects of TKI treatment Plasmid construction and transfection These two patients both had locally advanced IIIa-N2 lung According to the cDNA fragments within the coding region of adenocarcinoma with mediastinal lymph node metastases the GABBR2 gene (GenBank: BC035071), the sequence was at diagnosis. The pathology and staging were diagnosed on cloned into a pCMV-HA plasmid to generate GABBR2 overexpres- the basis of the metastatic mediastinal lymph nodes using sion using Lipofectamine 2000 Transfection Reagent (Life Tech- mediastinoscopy at baseline, and EGFR 19 deletion in meta- nologies). Western blot analysis was applied to validate the gene static mediastinal lymphadenopathies was detected by direct expression level. sequencing for these two patients. The clinical characteristics of To knock out the GABBR2 gene, two shRNAs were designed and these two patients were summarized in Table 1. One patient, synthesized. The shRNAs were then inserted into the core plasmid patient A, was a 54-year-old nonsmoking male with a 7.6 5.6 of the FG12 lentiviral system labeled with green fluorescent cm mass in the right upper lobe (RUL) and an enlarged 1.8 cm protein (GFP). Along with the other three packaging plasmids, right lower paratracheal lymphadenopathy on the baseline the core plasmid was used to transfect cells and obtain a virus chest CT (Fig. 1A), who was diagnosed with EGFR exon 19 carrying the target sequence. Afterwards, the matured virus was delA750-E758 stage IIIa lung adenocarcinoma (Fig. 1C) by transferred into the three target cell lines used in this study to mediastinoscopy sampling with a right lower paratracheal knockout the corresponding candidate GABBR2 gene. Western lymphadenopathy. The RUL mass shrank significantly from blot analysis was applied to validate the gene expression level. 7.6 5.6 cm to 3.4 3.0 cm (Fig. 1A), with PR consisting of a 55.3% reduction after 42 days of induction erlotinib Cell lysis and Western blotting treatment. The other one, patient B, was a 59-year-old non- Cells were rinsed with ice-cold PBS and lysed in radioimmu- smokingmale,witha6.1 4.0 cm RUL consolidation with noprecipitation assay (RIPA) buffer (Cell Signaling Technology, greater involvement of the right lower paratracheal lymphade- #9806), containing complete protease inhibitors (Roche), nopathy measuring 1.5 cm on the baseline chest CT (Fig. 1B), phosphatase inhibitors (Roche), 5 mmol/L DTT (Sigma), and who was also diagnosed with EGFR exon 19 delE746-A750 1 mmol/L PMSF for 15 minutes on ice. Cells were then centrifuged stage IIIa lung adenocarcinoma (Fig. 1D) by mediastinoscopy at 15,000 g for 10 minutes at 4C. After the supernatant was sampling with a right lower paratracheal lymphadenopathy. collected, the protein concentration was measured using the Bio- The RUL mass shrank significantly from 6.1 4.0 cm to 2.2 Rad protein assay (Bio-Rad). Afterwards, the proteins were sep- 2.9 cm (Fig. 1B) with PR consisting of a 52.5% reduction after arated using SDS-PAGE and blotted onto a PVDF membrane 42 days of induction erlotinib treatment. These two patients (Millipore). The antibodies were obtained from Santa Cruz Bio- were further treated with a RUL sleeve lobectomy based on the technology, Inc. multidiscipline discussion on tumor board considering the

Table 1. Clinical characteristics of the two patients for the MSCC screening study Patient ID Age Gender Smoking status Primary tumor location N2 station Pathology EGFR exon 19 deletion status Patient A 54 Male Non-smoker RUL 4R Adenocarcinoma delA750-E758 Patient B 59 Male Non-smoker RUL 4R Adenocarcinoma delE746-A750 Abbreviations: RUL, right upper lobe; 4R, right lower paratracheal lymphadenopathy.

www.aacrjournals.org Clin Cancer Res; 23(17) September 1, 2017 5005

Niu et al.

Figure 1. A and B, The tumors shrank signiﬁcantly after 42 days of induction erlotinib treatment in these two lung adenocarcinoma patients with EGFR 19 deletion mutation. Patient A had a partial response (PR) consisting of a 55.3% reduction after 42 days of induction erlotinib treatment from a baseline of 7.6 5.6 cm to 3.4 3.0 cm on the right upper lobe (RUL; A). Patient B had a PR consisting of a 52.5% reduction from a baseline of 6.1 4.0 cm to 2.2 2.9 cm on the RUL (B). C and D, The pathologic images of these two lung adenocarcinoma patients (400). The left image showed the baseline metastatic mediastinal lymph nodes before erlotinib treatment by mediastinoscopy, and the right image showed the metastatic mediastinal lymph nodes collected during the sleeve lobectomy after 42 days of induction erlotinib treatment. C is for patient A, and D for patient B.

marked induction erlotinib response achieved. The four tissue (Supplementary Fig. S1A), showing that the CCGG data with high samplesofpre-andposterlotinib treatment from these two sequencing depth of 40þ reads could well reflect the data of all patients were received and stored. CCGG sites. Therefore, we chose CCGG sites with 40þ reads to proceed with the following analysis. After excluding CCGG sites Global properties of DNA methylation in lung cancer with a low sequencing depth (<40 reads), we profiled 265,020 We performed MSCC sequencing and obtained an average of and 314,366 CCGGs for the two patients, respectively, represent- 49.3 (44.1–52.9) million uniquely mapped reads for each sample ing approximately 1% of all CpGs in the whole genome. The (Supplementary Table S1). According to previous research, the moving average methylation levels of each region of genes were sequencing depth of every CCGG site has a significant effect on the calculated according to the relative location of the genes. Mapping accurate estimation of the methylation level, namely, the accuracy of the CCGG sites to their relative location to associated genes increases with sequencing depth (14). The distribution of CCGG showed that most regions of genes, including the gene body, had sites with high sequencing depth of 40þ reads in our study was similar methylation levels, while the transcription starting site very similar to that of all the CCGG sites in lung cancer tissues (TSS) region was hypomethylated (Supplementary Fig. S1B),

5006 Clin Cancer Res; 23(17) September 1, 2017 Clinical Cancer Research

DNA Methylome Screen for Lung Adenocarcinoma with Erlotinib

Figure 2. A and B, Using MSCC screening on tissue samples from patient A and patient B, there were 621 and 220 genes that acquired TKI-induced methylation changes for these two patients, respectively, according to the DMRs with different levels of methylation changes (DMSCC 20%). The results revealed that the genes were enriched in several different types of pathways, including endocytosis, long-term depression, Wnt signaling pathway, Cadherin signaling pathway, regulation of beta-cell development, and others, among which Wnt signaling pathway, Cadherin signaling pathway, and regulation of beta-cell development were enriched in both patients. C, The results of the gene ontology (GO) enrichment analysis of the 60 genes with aberrant methylation levels in both patients indicated that the 60 genes were over-represented (or under-represented) in pathways related to sequence-speciﬁc DNA binding, transcription factor activity, transcription regulator activity, DNA binding and RNA polymerase II transcription factor, and others.

which is similar to the ﬁndings from the human methylome (18, methylation sequencing results were accurate and reliable from 19). We meanwhile mapped the CCGG sites to the known CpG multiple aspects. islands (CGI), which were highly clustered CpGs located at gene promoter regions (20). The methylation level of the CCGG sites Using MSCC screening, 60 aberrant methylated genes were inside of CGIs was signiﬁcantly lower than those outside of CGIs demonstrated to be related to TKI treatment in both patients (Supplementary Fig. S1C), which is similar to previous reports Using MSCC screening on tissue samples from patient A and (21). All of these results showed that the whole-genome DNA patient B, there were 621 and 220 genes that acquired TKI-induced

www.aacrjournals.org Clin Cancer Res; 23(17) September 1, 2017 5007

Niu et al.

methylation changes for these two patients, respectively, accord- aberrant methylation levels revealed that these genes were ing to the DMRs with different levels of methylation changes enriched in pathways involving sequence-speciﬁc DNA binding, (DMSCC 20%). We conducted pathway analysis based on the transcription factor activity, transcription regulator activity, DNA two sets of genes with methylation changes for each patient, binding and RNA polymerase II transcription factor, etc (Fig. 2C). showing that the genes were enriched in several different types This indicated that the genes with changed methylation levels of pathways, among which genes in the Wnt signaling pathway, might have an effect on the expression of more genes by tran- Cadherin signaling pathway, and regulation of beta-cell devel- scription regulation. opment were enriched in both patients (Fig. 2A and B). Notably, there were 60 genes with aberrant methylation Two differential methylated genes were validated using the levels found in both patients using whole-genome methylation Sequenom EpiTYPER assay in both patients screening. Supplementary Table S2 lists these 60 aberrant genes By using the Sequenom EpiTYPER assay, the methylation level and the detailed information of the DMRs, including the chro- of the 60 genes (Supplementary Table S2) was further validated. mosomal locations and average methylation changes. The result The methylation state of the 60 genes from the MSCC sequencing of the GO enrichment analysis based on these 60 genes with was reanalyzed in both patients using the Sequenom EpiTYPER

Figure 3. A, Epigenetic validation results of CBFA2T3 and GABBR2, which were ﬁrst screened by MSCC sequencing and further conﬁrmed by a Sequenom EpiTYPER assay with hypermethylation in red and hypomethylation in yellow, demonstrating that the adjacent CG sites had increased methylation levels, which was consistent with the MSCC screening results. B, Genomic locations of the 60 genes screened with aberrant methylation levels in both patients indicated that the methylation changes were at the whole-genome level. The methylation changes of each gene in the two patients were represented by different shades of color, which can be referred to the scales on the right side, with the blue color at the top of histograms with 100% methylation change showing hypomethylation, and the red color at the bottom of histograms with 100% methylation change showing hypermethylation. The validated genes, GABBR2 and CBFA2T3, were marked by arrows. C, The DMR in the GABBR2 gene was located exactly in the same region between exon 2 and exon 3 in both patients. The DMRs of CBFA2T3 were located in the transcription starting site (TSS) region for one and intron 1 for the other patient. Note: The arrow in C indicated the start of transcription.

5008 Clin Cancer Res; 23(17) September 1, 2017 Clinical Cancer Research

DNA Methylome Screen for Lung Adenocarcinoma with Erlotinib

assay. Among the 60 genes screened, the methylation of GABBR2 On the basis of all these results, together with the fact that the and CBFA2T3 were clearly verified, with hypermethylation in red same DMR location and methylation change direction of GABBR2 and hypomethylation in yellow, demonstrating that the adjacent gene were exhibited in both patients, the GABBR2 gene was CG sites exhibited increased methylation levels consistent with selected as the targeted gene in the following studies. the MSCC screening results (Fig. 3A). The genomic locations of the 60 genes screened with aberrant methylation levels in both Knockdown of GABBR2 did not further increase erlotinib- patients indicated that the methylation changes were at the induced apoptosis in HCC4006 and HCC827 cells whole-genome level, with the two validated genes, GABBR2 and After treatment with erlotinib, the apoptosis percentages of CBFA2T3, marked by arrows in Fig. 3B. The relative locations of EGFR 19 deletion cells, HCC4006 and HCC827, were significantly the DMRs in the two genes were presented in Fig. 3C, showing that higher compared with that of EGFR wild-type cells A549. The the DMR of the GABBR2 gene was located exactly in the same apoptosis percentages were 21.9%, 32.7%, and 5.54% in location between exon 2 and exon 3 in both patients, while the HCC4006, HCC827 and A549 cells, respectively (P < 0.05 for DMRs of the CBFA2T3 gene were located in the TSS region in one A549 compared with HCC4006, P < 0.01 for A549 compared with patient and intron 1 in the other. Supplementary Table S2 showed HCC827). After knockdown of the GABBR2 gene, apoptosis did the detailed information about these two genes, illustrating the not increase further in the cells harboring EGFR 19 deletion exact same DMR location (chr_start: chr9_101338201) and the mutations, HCC4006 and HCC827 (P > 0.05, Fig. 4C and D). same direction of the methylation change for GABBR2 gene in both patients, with an average methylation change of the DMRs at GABBR2 overexpression rescued cleaved PARP and erlotinib- 42.35% for one and 23.50% for the other patient. The DMRs of induced apoptosis in HCC827 cells, and the effect of GABBR2 CBFA2T3 were located in the TSS region (chr16_89045001) in in combination with erlotinib may play an important role one patient and intron 1 (chr16_89035601) in the other patient, in EGFR signaling and the average methylation change of the DMRs was 26.88% and In our previous study, GABBR2 was found significantly decreased 25.95%, respectively. in HCC4006 and HCC827 cells when treated with erlotinib (Fig. 4A), and the apoptosis rate was the highest at 32.7% in HCC827 GABBR2 was verified to be a potential treatment target in vitro cells (P < 0.01 compared with A549) after treatment with erlotinib We validated the role of GABBR2 and CBFA2T3 by treating lung (Fig. 4C and D). So the role of GABBR2 was further investigated in cancer cells with erlotinib in vitro. After 1 mmol/L erlotinib HCC827 cells by cloning GABBR2 into the pCMV-HA plasmid to treatment for 48 hours, GABBR2 expression was significantly demonstrate whether overexpression of GABBR2 could rescue decreased in EGFR-mutated lung cancer cells, HCC4006 and erlotinib-induced apoptosis in EGFR 19 deletion cells. The level HCC827 cells, but remained conserved in EGFR wild-type of cleaved PARP, a marker of apoptosis, was obvious when treated A549 cells, while CBFA2T3 expression was not as significantly with erlotinib, while upregulation of GABBR2 significantly rescued reduced as GABBR2 expression in HCC827 cells (Fig. 4A). the apoptosis of HCC827 cells, reflected by the rescued level of We then studied whether a reduction in the expression of cleaved PARP (Fig. 5A). The apoptosis results showed that the protein GABBR2 is induced by methylation-related enzymes at apoptosis percentage of HCC827 cells after treatment with erlotinib the protein level. The expressions of DNMT1, TET2 and TET3 were significantly increased from 32.2% to 54.3% compared with the significantly downregulated in HCC4006 and HCC827 cells com- control group and was rescued from 54.3% back to 36.8% with pared to those in A549 cells, while DNTM3A expression did not overexpression of GABBR2 (P < 0.05 when compared with the show a significant change (Fig. 4A). The results illustrated that not "controlþerlotinib" group, and P > 0.05 when compared with the only erlotinib did induce a direct epigenetic effect on the GABBR2 "control" group, Fig. 5B and C). These results showed that GABBR2 gene, the downregulation of methylation-related enzymes may overexpression rescued erlotinib-induced apoptosis back to control also regulate the methylation of GABBR2, these two aspects levels in HCC827 cells, supporting the hypothesis that erlotinib- together leading to the reduction of GABBR2 expression. induced downregulation of GABBR2 which played a significant role We meanwhile detected the expression of GABBR2 in the in apoptosis of EGFR-mutated lung cancer cells. tissues before and after induction erlotinib treatment in these To characterize whether the apoptosis induced by GABBR2 in two patients by IHC. The IHC results showed that the expression combination with erlotinib is involved in EGFR signaling, we of GABBR2 was detected in the tissues before erlotinib treatment, examined the expression of several key regulators, p-ERK1/2, while decreased dramatically after erlotinib therapy (Fig. 4B). t-ERK1/2, S6, and p-p70 S6, which function within EGFR signal- Considering there are two classes of GABA receptors: GABAA ing pathway (22). The caspase family proteins and cleaved PARP and GABAB, with 19 subtypes of GABAA receptors in humans were also examined to investigate the underlying mechanism of and only 2 subtypes of GABAB receptors, GABBR1 and GABBR2, apoptosis induced by erlotinib. The lysates were analyzed using we reanalyzed erlotinib-related methylation changes of the antibodies against caspase-3 and its active cleaved form, cleaved whole GABA receptors families, 19 subtypes of GABAA receptors, caspase-3. As shown in Fig. 5D, after treatment with erlotinib, the and 2 subtypes of GABAB receptors, generated by MSCC expression of cleaved caspase-3 and cleaved PARP proteins, sequencing. The methylation status of GABBR1 was not screened together with the decreased GABBR2 expression, were observed by MSCC method because there were not enough CCGG sites in both EGFR 19 deletion cell lines, HCC4006 and HCC827, but in GABBR1 gene for analysis. Furthermore, the methylation not in EGFR wild-type A549 cells. These results verified that status of 19 subtypes of GABAA receptors were determined by GABBR2 was involved in the apoptotic effects of EGFR signaling MSCC, and the methylation of GABAA receptor families was when treated with erlotinib. Meanwhile erlotinib promoted relatively conservative before and after erlotinib treatment (no a reduction in the expression of the phosphorylated proteins 200-bp windows with >20% methylation change; Supplemen- p-ERK1/2 in HCC4006 and HCC827 cells, with a minimal effect tary Table S3). on the total protein levels of t-ERK1/2, and a reduction in the

www.aacrjournals.org Clin Cancer Res; 23(17) September 1, 2017 5009

Niu et al.

Figure 4. A, After treatment with 1 mmol/L erlotinib for 48 hours, GABBR2 expression was significantly decreased in HCC4006 and HCC827 cells (EGFR 19 deletion) but remained conserved in A549 cells (EGFR wild-type), while CBFA2T3 expression was not as significantly decreased as GABBR2 in HCC827 cells. Methylation- related enzymes, DNMT1, TET2 and TET3, were significantly downregulated in HCC4006 and HCC827 cells compared with those in A549 cells, while DNTM3A expression did not significantly change. B, IHC results of the tissues before and after induction erlotinib treatment in these two patients (400). Figures 1 and 2 represent the results of GABBR2 expression before and after induction erlotinib treatment of patient A, and Figures 3 and 4 represent the results of patient B. The IHC results showed that the expression of GABBR2 was detected in the tissues before erlotinib treatment, while decreased dramatically after erlotinib therapy. Note: Hematoxylin stained nuclei in blue; the positive GABBR2 expression of DAB showed brown in cytoplasm. C and D, After treatment with 2 mmol/L erlotinib for 48 hours, the apoptosis percentages in EGFR 19 deletion cells (HCC4006 and HCC827) were significantly higher compared with EGFR wild-type cells (A549). The apoptosis percentages were 21.9%, 32.7%, and 5.54% in HCC4006, HCC827, and A549 cells, respectively (P < 0.05 for A549 compared with HCC4006, P < 0.01 for A549 compared with HCC827). After knockdown of the GABBR2 gene by RNA interference using the siGABBR2 plasmid, the apoptosis percentages changed from 21.9% to 18.7% for HCC4006 cells, from 32.7% to 32.2% for HCC827 cells, and from 5.54% to 5.46% for A549 cells. The results showed that apoptosis did not increase in cells harboring EGFR 19 deletion mutations, HCC4006 and HCC827 (P > 0.05), when knockdown of GABBR2 gene by siGABBR2 plasmid transfection after incubation with 2 mmol/L erlotinib for 48 hours.

expression of S6 and p-p70 S6 which are the upstream of ERK1/2 to locally advanced stage NSCLC. On the basis of the current pathway, but not in A549 cells, suggesting that GABBR2 may play background, our research is to investigate whether epigenetic an important role in EGFR signaling through the ERK1/2 pathway. changes are involved in the response to induction TKI treatment in locally advanced EGFR-mutated lung cancer. Our research demonstrated that methylation of GABBR2 gene might play a Discussion role in the treatment of EGFR-mutated lung adenocarcinoma with Patients with advanced stage NSCLC harboring activating EGFR induction erlotinib, suggesting that targeting GABBR2 together mutations tend to respond well to EGFR-TKIs. However, the with EGFR inhibition may improve clinical outcome in patients beneﬁt of EGFR-TKI regimens for locally advanced NSCLC with locally advanced NSCLC. patients still remains uncertain, and there have been some clinical Changes in methylation patterns and levels have been associ- trials (NCT01822496, NCT00600587) trying to apply this beneﬁt ated with different types of cancers and could be used to predict

5010 Clin Cancer Res; 23(17) September 1, 2017 Clinical Cancer Research

DNA Methylome Screen for Lung Adenocarcinoma with Erlotinib

Figure 5. GABBR2 overexpression rescued cleaved PARP and erlotinib-induced apoptosis in HCC827 cells, and the effect of GABBR2 in combination with erlotinib may play an important role in EGFR signaling. A, The apoptosis represented by cleaved PARP was obvious when HCC827 cells were treated with 2 mmol/L erlotinib for 48 hours, while upregulation of GABBR2 significantly rescued the apoptosis of HCC827 cells, which was reflected by the rescued level of cleaved PARP. B and C, The apoptosis percentage of HCC827 cells after treatment with 2 mmol/L erlotinib for 48 hours significantly increased from 32.2% to 54.3% compared with the "control" group and was rescued from 54.3% back to 36.8% when the GABBR2 gene was overexpressed by cloning into a pCMV-HA plasmid (P < 0.05 when compared with the "controlþerlotinib" group, and P > 0.05 when compared with the "control" group), showing that GABBR2 overexpression rescued erlotinib-induced apoptosis, reverting to the level found in the "control" group in HCC827 cells. D, The expression of cleaved caspase-3 and cleaved PARP proteins, together with the decreased GABBR2 expression, was observed in both EGFR 19 deletion cell lines (HCC4006 and HCC827), but not in EGFR wild-type A549 cells, with treatment of 1 mmol/L erlotinib for 48 hours. At the same time, erlotinib promoted a reduction in the expression of the phosphorylated proteins p-ERK1/2 in HCC4006 and HCC827 cells, with minimal effect on the total protein levels of t-ERK1/2, and a reduction in the expression of S6 and p-p70 S6, which are the upstream of ERK1/2 pathway, but not in A549 cells, suggesting that GABBR2 may play an important role in EGFR signaling through the ERK1/2 pathway. treatment response or outcomes in malignancies (23, 24), includ- methylation status of the Wnt antagonist SFRP5 can predict the ing lung cancer (23). However the role of DNA methylation in the response to EGFR-TKI therapy in NSCLC (8). Wnt inhibitory response to TKI therapy for the treatment of lung cancer has factor 1 (Wif1) promoter methylation is an early and frequent remained unclear. In our study, two lung adenocarcinoma event as an epigenetic field manner, and Wif1 hypermethylation patients with EGFR 19 deletion showed a marked response to can be as an unfavorable prognosis marker of NSCLC with EGFR TKI treatment, indicating a good foundation for the following mutation (26). The cadherin signaling pathway has also been well whole-methylome screening. Hypomethylation of the TSS region studied in NSCLC with EGFR mutation. E-cadherin, a hallmark of and CGI region in our study was also consistent with previous epithelial–mesenchymal transition (EMT), plays a significant role studies (14, 18, 19). The pathway analysis of the differential in the sensitivity regulation of EGFR molecular targeting treat- methylated genes indicated that Wnt signaling pathway, Cadherin ment (27). EMT-induced cells by E-cadherin depletion facilitate signaling pathway, and other pathways enriched in both patients invasion in a matrix metalloproteinase-2 (MMP2)-dependent are related to the EGFR pathway. Many studies have proven that manner with aberrant activation of EGFR signaling (28). Wnt signaling pathway has a role in EGFR-mutated lung cancer In our study, genes with altered DNA methylation in lung (8, 25, 26). Inhibition of various components of Wnt pathway adenocarcinoma patients with EGFR 19 deletion before and after with shRNAs or small molecules significantly increased the effi- TKI treatment were first screened using an epigenetic genome- cacy of EGFR inhibitors both in vitro and in vivo (25). The DNA wide analysis and then verified by Sequenom EpiTYPER assay,

www.aacrjournals.org Clin Cancer Res; 23(17) September 1, 2017 5011

Niu et al.

with the result showing that the DMR of GABBR2 gene was located demonstrated that GABBR2 may play a role in EGFR signaling in the same region of the two patients. The IHC results showed through the ERK1/2 pathway and its cross-talk pathway, P70 S6 that the expression of GABBR2 was decreased dramatically after signaling. erlotinib therapy. The following in vitro experiments also sug- Although the epigenetic mechanism of TKI therapy in lung gested that TKI treatment induced downregulation of GABBR2 cancer was studied for the first time, there are still some short- and regulated the expression of methylation-related enzymes comings in our research. First, multiple tools and methods have DNMT and TET, which may, in turn, result in the aberrant been developed and used for differential methylation data pre- methylation and downregulation of GABBR2. There are three processing and data analysis, such as whole-genome bisulfite enzymatically active mammalian DNMTs-DNMT3A, DNMT3B sequencing (WGBS) and reduced representation bisulfite sequenc- and DNMT1. DNMT3A and DNMT3B are de novo methyltrans- ing (RRBS). Thus, the question that remains to be answered is ferases that establish DNA methylation patterns by targeting which platform should be used to better fit the needs of this study. unmethylated CpG sites. DNMT1 acts primarily as a maintenance MSCC sequencing technology could quantify approximately 1.1% methyltransferase. TET proteins are a-ketoglutarate-dependent of all CG sites in humans (14). The CCGG sites sequenced in the dioxygenases involved in the conversion of 5-methylcytosines present study could well reflect the methylation status of all CG (5-mC) to 5-hydroxymethylcytosine (5-hmC) (29, 30). DNMTs sites; and the number of CpGs with sufficient coverage from each and TETs play crucial roles in the cytosine methylation and appears more efficient for CpG capture, making sequence-based demethylation process. The intronic DMR may block transcrip- MSCC technology adequate and reliable for methylation experi- tion elongation of GABBR2 through affecting histone deacetyla- ments. Second, we investigated the possible function of GABBR2 in tion and chromatin structure (31, 32) and downregulate gene cell lines and noticed the relationship to the ERK signaling path- expression by regulating the methyl-sensitive binding of tran- way. However, a more detailed mechanism, including whether scription factors (33). The relationship between GABBR2 and overexpression of GABBR2 could rescue the expression of S6, cancer development and prognosis has been considered in several p-p70 S6, and p-ERK1/2, requires further study. In addition, studies (34–36). The possible role of GABBR2 in cancer progres- xenograft preclinical studies in vivo may be more helpful to sion was first reported by Stein and colleagues and increased elucidate the tumor-related function of GABBR2. Third, there were expression of GABBR2 was demonstrated in papillary thyroid only two patients in our study, which could limit our final results in carcinomas (PTC) using gene expression profiling (34). Similarly, some way. Considering this limitation, we first used MSCC to by using a whole-transcript expression array, Schulten and col- screen for changes in methylation, then used a Sequenom EpiTY- leagues found that expression of GABBR2 was significantly upre- PER assay to validate our screening results, and finally confirmed gulated in the follicular variant of PTC (35). Notably, the gene our results with the functional experiments in vitro. Importantly, we expression levels of GABBR2 were also significantly higher in used high quality, fresh surgical tumor specimens to provide NSCLC tissue than in the paired paracancerous tissue, and accurate and adequate molecular diagnoses, which are necessary patients with high expression of GABBR2 gene had a better for biomarker discovery and validation. Although it is difficult to prognosis (36). However, the relationship between the methyl- collect the tumor tissues before and after induction TKI treatment ation status of GABBR2 gene and EGFR-TKI response in lung with remarkable response, our report still describes our experience cancer has not been well investigated before. and challenges in obtaining fresh lung tumor specimens for the Our in vitro function study demonstrated that overexpression of utilization in EGFR-related biomarker discovery. In addition, the GABBR2 rescued the TKI-induced cell apoptosis, suggesting that ideal duration for induction treatment with EGFR-TKI drugs is still GABBR2 may be involved in EGFR -mutated lung cancer–targeted uncertain. Some clinical trials (NCT01822496, NCT00600587, therapy. Erlotinib downregulated GABBR2 expression in EGFR 19 NCT01407822) were performed to evaluate the value of induction deletion cells, along with a reduction in the expression of S6, p-p70 erlotinib therapy for uninterrupted 42 days before thoracotomy or S6, and p-ERK1/2, suggesting that GABBR2 may play an important radiotherapy in EGFR-mutated IIIa-N2 NSCLC, which were the role in EGFR signaling through the ERK1/2 pathway which may same design with our research. But cautions should be used when occur in two ways. First, GABBR2 may promote phosphorylation interpreting the epigenetic mechanism of TKI therapy for different of ERK directly. One research revealed that the GABA(B) receptor induction treatment durations and different generation TKI drugs. can induce ERK1/2 phosphorylation in the CA1 area (region I of In summary, we performed a genome-wide DNA methylation the hippocampus proper) of the mouse hippocampus (37). Acti- analysis in EGFR mutated lung adenocarcinoma with induction vation of the GABA(B) receptor by either GABA or the selective erlotinib treatment and found that the GABBR2 gene plays a role agonist baclofen can induce ERK(1/2) phosphorylation in cul- in targeted therapy. Our findings provide a new theoretical basis tured cerebellar granule neurons (38). Second, we observed the for the application of epigenetic studies in EGFR-mutated lung aberrant methylation of the cadherin signaling pathway, and as adenocarcinoma and suggest that targeting GABBR2 together with mentioned above, ERK is known to be involved in cadherin EGFR inhibition may improve clinical outcomes in patients with signaling (28, 39); therefore, the phosphorylated proteins, p- locally advanced NSCLC. ERK1/2, may be downregulated by the aberrant methylation of thecadherin signalingpathway.In addition,wealsodemonstrated Disclosure of Potential Conflicts of Interest the reduction of S6 and p-p70 S6 along with the downregulation of No potential conflicts of interest were disclosed. p-ERK1/2. Cross-talk between ERK and P70 S6 signaling has long been considered as early as in the year 2001 (40). In cancer, growth Authors' Contributions factors, including epidermal growth factors (EGF), can activate Conception and design: X. Niu, X. Ye, Y. Liu, Z. Chen, S. Lu ERK and PI3K pathways (22), while either activation of ERK or Development of methodology: X. Niu, F. Liu, Y. Zhou, D. Zhou, Y. Liu, S. Lu PI3K pathway can contribute to the activation of S6 and protein Acquisition of data (provided animals, acquired and managed patients, synthesis via upregulation of P70 S6 (22). Overall, our study provided facilities, etc.): X. Niu, Y. Zhou, Z. Zhou, Z. Li, Y. Liu, S. Lu

5012 Clin Cancer Res; 23(17) September 1, 2017 Clinical Cancer Research

DNA Methylome Screen for Lung Adenocarcinoma with Erlotinib

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, Science & Technology Commission (13ZR1438500), Youth Foundation of computational analysis): X. Niu, F. Liu, Y. Zhou, X. Weng, H. Zhang, S. Lu Shanghai Municipal Public Health Bureau (20124Y114), Interdisciplinary Writing, review, and/or revision of the manuscript: X. Niu, F. Liu, Y. Zhou, Program of Shanghai Jiao Tong University (YG2013MS10), "Shanghai D. Zhou, Z. Li, Y. Yu, H. Zhang, Z. Chen, S. Lu Young Physician Development Program" of Shanghai Municipal Public Administrative, technical, or material support (i.e., reporting or organizing Health Bureau, the second batch (2012(105)), "Shanghai Young Teacher data, constructing databases): X. Niu, F. Liu, T. Wang, X. Ye, X. Weng, H. Zhang, Education Program" of University Project in Shanghai, "Shanghai Jiao Tong J. Ye, S. Lu University SMC-Star Award for Young Scholar B program", Shanghai Jiao Study supervision: X. Niu, X. Ye, M. Liao, Y. Liu, Z. Chen, S. Lu Tong University Wang-Kuancheng Award, and the "1050 Talents Project" of Shanghai Chest Hospital, Project of Shanghai Chest Hospital (YZ13-16; all Acknowledgments to X. Niu); National Natural Science Foundation of China (NSFC; 81572250; to Z. Chen); National Natural Science Foundation of China We thank the patients who participated and all of the investigators and site (NSFC; 81401878; to Y. Liu); National Natural Science Foundation of China staff for their support. This research has been selected as a Poster Presentation (NSFC; 81302005; to X. Ye); and Project of Shanghai Chest Hospital (Abstract ID: 6219) in the 17th International Association for the Study of Lung (2014YZDC10600; to S. Lu). Cancer (IASLC) World Conference on Lung Cancer (Dec 7, 2016), and the ﬁrst The costs of publication of this article were defrayed in part by the author, Dr. Niu, has successfully won a 17th IASLC International Mentorship payment of page charges. This article must therefore be hereby marked Award by virtue of this research. advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Grant Support This work was supported by grants from the National Natural Science Received November 1, 2016; revised March 5, 2017; accepted May 2, 2017; Foundation of China (NSFC; 81302005), Project of Shanghai Municipality published OnlineFirst May 10, 2017.

References 1. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates of 15. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer of large gene lists using DAVID bioinformatics resources. Nat Protoc 2010;127:2893–917. 2009;4:44–57. 2. Chan BA, Hughes BG. Targeted therapy for non-small cell lung cancer: 16. Huang da W, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: current standards and the promise of the future. Transl Lung Cancer Res paths toward the comprehensive functional analysis of large gene lists. 2015;4:36–54. Nucleic Acids Res 2009;37:1–13. 3. Mok TS, Wu YL, Thongprasert S, Yang CH, Chu DT, Saijo N, et al. Gefitinib 17. Hatje K, Keller O, Hammesfahr B, Pillmann H, Waack S, Kollmar M. Cross- or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med species protein sequence and gene structure prediction with fine-tuned 2009;361:947–57. Webscipio 2.0 and Scipio. BMC Res Notes 2011;4:265. 4. Pignon JP, Tribodet H, Scagliotti GV, Douillard JY, Shepherd FA, Stephens 18. Laurent L, Wong E, Li G, Huynh T, Tsirigos A, Ong CT, et al. Dynamic RJ, et al. Lung adjuvant cisplatin evaluation: a pooled analysis by the LACE changes in the human methylome during differentiation. Genome Res Collaborative Group. J Clin Oncol 2008;26:3552–9. 2010;20:320–31. 5. Bjaanaes MM, Fleischer T, Halvorsen AR, Daunay A, Busato F, Solberg S, 19. Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, Tonti-Filippini J, et al. Genome-wide DNA methylation analyses in lung adenocarcinomas: et al. Human DNA methylomes at base resolution show widespread association with EGFR, KRAS and TP53 mutation status, gene expression epigenomic differences. Nature 2009;462:315–22. and prognosis. Mol Oncol 2016;10:330–43. 20. Suzuki MM, Bird A. DNA methylation landscapes: provocative insights 6. Greve G, Schiffmann I, Pfeifer D, Pantic M, Schuler J, Lubbert M. The pan- from epigenomics. Nat Rev Genet 2008;9:465–76. HDAC inhibitor panobinostat acts as a sensitizer for erlotinib activity in 21. Schubeler D. Function and information content of DNA methylation. EGFR-mutated and -wildtype non-small cell lung cancer cells. BMC Cancer Nature 2015;517:321–6. 2015;15:947. 22. Pearce LR, Komander D, Alessi DR. The nuts and bolts of AGC protein 7. Li J, Jia XF, Liu J, Liu JJ, Zhao HB. Relationship of EGFR DNA methylation kinases. Nat Rev Mol Cell Biol 2010;11:9–22. with the severity of non-small cell lung cancer. Genet Mol Res 2015;14: 23. Fleischhacker M, Dietrich D, Liebenberg V, Field JK, Schmidt B. The role of 11915–23. DNA methylation as biomarkers in the clinical management of lung cancer. 8. Zhu J, Wang Y, Duan J, Bai H, Wang Z, Wei L, et al. DNA Methylation status Expert Rev Respir Med 2013;7:363–83. of Wnt antagonist SFRP5 can predict the response to the EGFR-tyrosine 24. Shiovitz S, Bertagnolli MM, Renfro LA, Nam E, Foster NR, Dzieciatkowski S, kinase inhibitor therapy in non-small cell lung cancer. J Exp Clin Cancer Res et al. CpG island methylator phenotype is associated with response to 2012;31:80. adjuvant irinotecan-based therapy for stage III colon cancer. Gastroenter- 9. Li XY, Wu JZ, Cao HX, Ma R, Wu JQ, Zhong YJ, et al. Blockade of DNA ology 2014;147:637–45. methylation enhances the therapeutic effect of gefitinib in non-small cell 25. Casas-Selves M, Kim J, Zhang Z, Helfrich BA, Gao D, Porter CC, et al. lung cancer cells. Oncol Rep 2013;29:1975–82. Tankyrase and the canonical Wnt pathway protect lung cancer cells from 10. Ogawa T, Liggett TE, Melnikov AA, Monitto CL, Kusuke D, Shiga K, et al. EGFR inhibition. Cancer Res 2012;72:4154–64. Methylation of death-associated protein kinase is associated with cetux- 26. Lee SM, Park JY, Kim DS. Wif1 hypermethylation as unfavorable prognosis imab and erlotinib resistance. Cell Cycle 2012;11:1656–63. of non-small cell lung cancers with EGFR mutation. Mol Cells 2013;36: 11. Montero AJ, Diaz-Montero CM, Mao L, Youssef EM, Estecio M, Shen L, et al. 69–73. Epigenetic inactivation of EGFR by CpG island hypermethylation in 27. Xing RC, Zheng J, Zheng WH, Qin ZP, Liu W, Yao RC. Relevance of cancer. Cancer Biol Ther 2006;5:1494–501. E-cadherin expression to EGFR-TKI molecular targeted therapy sensitivi- 12. Scartozzi M, Bearzi I, Mandolesi A, Giampieri R, Faloppi L, Galizia E, et al. ty/resistance and its clinical significance. Genet Mol Res 2015;14:5785–92. Epidermal growth factor receptor (EGFR) gene promoter methylation and 28. Bae GY, Choi SJ, Lee JS, Jo J, Lee J, Kim J, et al. Loss of E-cadherin activates cetuximab treatment in colorectal cancer patients. Br J Cancer 2011;104: EGFR-MEK/ERK signaling, which promotes invasion via the ZEB1/MMP2 1786–90. axis in non-small cell lung cancer. Oncotarget 2013;4:2512–22. 13. Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, et al. 29. Scourzic L, Mouly E, Bernard OA. TET proteins and the control of cytosine New response evaluation criteria in solid tumours: revised RECIST guide- demethylation in cancer. Genome Med 2015;7:9. line (version 1.1). Eur J Cancer 2009;45:228–47. 30. Williams K, Christensen J, Helin K. DNA methylation: TET proteins- 14. Guo JU, Ma DK, Mo H, Ball MP, Jang MH, Bonaguidi MA, et al. Neuronal guardians of CpG islands? EMBO Rep 2011;13:28–35. activity modifies the DNA methylation landscape in the adult brain. Nat 31. Irvine RA, Lin IG, Hsieh CL. DNA methylation has a local effect on Neurosci 2011;14:1345–51. transcription and histone acetylation. Mol Cell Biol 2002;22:6689–96.

www.aacrjournals.org Clin Cancer Res; 23(17) September 1, 2017 5013

Niu et al.

32.NanX,NgHH,JohnsonCA,LahertyCD,TurnerBM,EisenmanRN, 36. Zhang X, Zhang R, Zheng Y, Shen J, Xiao D, Li J, et al. Expression of gamma- et al. Transcriptional repression by the methyl-CpG-binding protein aminobutyric acid receptors on neoplastic growth and prediction of MeCP2 involves a histone deacetylase complex. Nature 1998;393: prognosis in non-small cell lung cancer. J Transl Med 2013;11:102. 386–9. 37. Vanhoose AM, Emery M, Jimenez L, Winder DG. ERK activation by 33. Lai AY, Fatemi M, Dhasarathy A, Malone C, Sobol SE, Geigerman C, et al. G-protein-coupled receptors in mouse brain is receptor identity-speciﬁc. DNA methylation prevents CTCF-mediated silencing of the oncogene J Biol Chem 2002;277:9049–53. BCL6 in B cell lymphomas. J Exp Med 2010;207:1939–50. 38. Tu H, Rondard P, Xu C, Bertaso F, Cao F, Zhang X, et al. Dominant role of 34. Stein L, Rothschild J, Luce J, Cowell JK, Thomas G, Bogdanova TI, et al. GABAB2 and Gbetagamma for GABAB receptor-mediated-ERK1/2/CREB Copy number and gene expression alterations in radiation-induced pap- pathway in cerebellar neurons. Cell Signal 2007;19:1996–2002. illary thyroid carcinoma from chernobyl pediatric patients. Thyroid 39. Tashiro E, Henmi S, Odake H, Ino S, Imoto M. Involvement of the MEK/ 2010;20:475–87. ERK pathway in EGF-induced E-cadherin down-regulation. Biochem Bio- 35. Schulten HJ, Al-Mansouri Z, Baghallab I, Bagatian N, Subhi O, Karim S, phys Res Commun 2016;477:801–6. et al. Comparison of microarray expression proﬁles between follicular 40. Wang L, Gout I, Proud CG. Cross-talk between the ERK and p70 S6 kinase variant of papillary thyroid carcinomas and follicular adenomas of the (S6K) signaling pathways. MEK-dependent activation of S6K2 in cardio- thyroid. BMC Genomics 2015;16 Suppl 1:S7. myocytes. J Biol Chem 2001;276:32670–7.

5014 Clin Cancer Res; 23(17) September 1, 2017 Clinical Cancer Research

Genome-wide DNA Methylation Analysis Reveals GABBR2 as a Novel Epigenetic Target for EGFR 19 Deletion Lung Adenocarcinoma with Induction Erlotinib Treatment

Xiaomin Niu, Fatao Liu, Yi Zhou, et al.

Clin Cancer Res 2017;23:5003-5014. Published OnlineFirst May 10, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-16-2688

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2017/05/10/1078-0432.CCR-16-2688.DC1

Cited articles This article cites 40 articles, 7 of which you can access for free at: http://clincancerres.aacrjournals.org/content/23/17/5003.full#ref-list-1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/23/17/5003. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.