基因體甲基化圖譜與抑癌基因甲基化參與肺癌形成之機制及臨床應用探討

全文

(1)國立臺灣師範大學生命科學系博士論文. 基因體甲基化圖譜與抑癌基因甲基化參與肺癌形成之機制及臨床應用探討. Genome-Wide Methylation Profiling and Candidate Gene Methylation Spectrum in Lung Cancer: In Relation to Gene Expression and Clinical Significance 研究生：張哲維 Jer-Wei Chang 指導教授：王憶卿博士、蘇銘燦博士 Dr. Yi-Ching Wang, Dr. Ming-Tsan Su. 中華民國九十八年七月.

(2) 誌. 謝. 博士班的生涯即將告一段落，回想起這六年研究生的生活，心中真是百感交集；這些年因為實驗需求，經常往返於師大、中研院、長庚大學，只為尋求實驗突破，而且由於實驗室的搬遷，博五起就必須到台南成功大學進行實驗，經常往來於桃園及台南，心中疲憊不已；然而，卻也因為這樣，接觸了許多新的人、事、物，不論在學識上或生活上，皆受益良多。首先想要感謝的是恩師王憶卿教授，從碩士班開始，不管是在課業上或實驗上，總能一步步的帶領我學習，雖然我資質不佳，卻也能對這門科學有粗略的瞭解，實驗上有初步的成績，在此對老師表達由衷的謝意；另外還要感謝中研院統計所黃培瑛博士，感謝黃老師在實驗上的指導及在人生規劃的許多建議，讓我獲益匪淺。要感謝的人實在太多了，感謝蘇銘燦老師、李桂楨老師、孫智雯老師及生科系許多老師在實驗及課業上的指導，感謝實驗室學長姊：建智、慶孝、若嘉及若凱、在實驗上的幫助與建議，以及實驗室學弟妹：一泓、芳宜、信銘、家揚、懿瑩、宗翰、硯安、貝君、侑庭、久珊、宗瀚、子堂及助理宛靜在實驗及生活上的多方協助。另外，還要特別感謝師大生科的諸位學長：政光、瑞宏及同學：玄原、麗卿，承蒙你們在實驗過程中的幫助，實驗才能順利的完成，謝謝你們！! -1-.

(3) 而在實驗的過程中，感謝台中榮總許瀚水醫師提供外科檢體樣本；感謝長庚大學張玉生教授提供了甲基化晶片與儀器設備，和中研院統計研究所黃培瑛博士及其助理陳怡安在甲基化晶片數據統計分析上的幫助，讓實驗結果更加完善；由於你們的幫助，實驗才能順利完成，感激不盡！！同時在此，還要特別感謝我的口試委員：李德章教授、張玉生教授、陳全木教授、黃培瑛博士及蘇銘燦教授，撥冗參與本論文的審查及口試工作，謹此衷忱致謝。最後，感謝媽媽及二位姊姊對我的關懷、鼓勵與支持，讓我能順利的完成學業，以及老婆金祝的體諒與包容，總在我最沮喪低潮時，給我心靈上的支持與安慰。僅以此論文獻給我最親愛的家人及關心我的人。張哲維謹誌于國立台灣師範大學生命科學研究所中華民國九十八年七月. -2-.

(4) Content Chinese abstract ------------------------------------------------------------ 1 English abstract ------------------------------------------------------------- 3 Study rationale -------------------------------------------------------------- 5 Literature review ----------------------------------------------------------- 8 I. The classifications of lung cancer --------------------------------------- 8 II. Gene promoter hypermethylation and tumorigenesis --------------- 9 III. Chromatin structure and gene expression status --------------------11 IV. Alterations of genes in chromosome 3p21.3 involve in lung tumorigenesis ------------------------------------------------------------ 12 V. Methods for epigenetic alternation analysis --------------------------14. Specific aims -----------------------------------------------------------------19 AIM 1: Epigenome-wide detection for hypermethylated CpG islands and chromatin profiling and their clinical association in lung cancer ----------------------------21 Purpose ----------------------------------------------------------------------- 22 Materials and Methods ----------------------------------------------------- 23 Results ------------------------------------------------------------------------ 28 Discussion --------------------------------------------------------------------30. I.

(5) AIM 2: Validated the cancer related genes and TSGs from DMH and ChIP-on-chip analysis, and characterization of their transcriptional important CpG sites in primary non-small cell lung cancer --------- 32 Purpose ----------------------------------------------------------------------- 33 Materials and Methods ----------------------------------------------------- 34 Results ----------------------------------------------------------------------- 41 Discussion ------------------------------------------------------------------- 49. References ------------------------------------------------------------------- 54 Tables and Figures -------------------------------------------------------- 68 Appendices ------------------------------------------------------------------- 96. II.

(6) Table content Table 1. The primers used in the present study ------------------------- 69. Table 2. Clinical characterizations of 30 surgically resected NSCLC patients for DMH analysis ---------------------------- 73. Table 3. Selected hypermethylated genes from DMH methylation profiling of patients with AD or SQ type ------ 74. Table 4. Selected hypermethylated genes from DMH methylation profiling of patients with early stage or late stage ------------------------------------------------------- 75. Table 5. Selected hypermethylated genes from DMH methylation profiling of 15 patients with overexpression of DNMT1 --------------------------------------76. Table 6. Correlation analyses of COL14A1 DNA methylation, low mRNA and low protein expression, and clinicopathological parameters of 48 NSCLC Patients -------------------------------------------------------------77. III.

(7) Figure content Figure 1 Schematic flowchart for DMH and ChIP-on-chip ------------78 Figure 2 Determination of chromatin status of COL14A1 CpG island by ChIP-on-chip ------------------------------------------ 79 Figure 3 Determination of methylation status of RASSF1A and BLU CpG sites by MSO microarray --------------------------- 80 Figure 4 Distinct methylation and chromatin structure boundaries separated by the CTCF binding domain --------- 83 Figure 5 Methylation and protein/mRNA expression assays for COL14A1 genes in lung cancer cell lines and patients ------------------------------------------------------------- 85 Figure 6 Chromosomal location of RASSF1A and BLU gene at 3p21.3 region -------------------------------------------------- 87 Figure 7 RASSF1A chromatin status and gene expression in cell lines ----------------------------------------------------------- 88 Figure 8 E2F1 binding site analyses in RASSF1A promoter ----------- 89 Figure 9 Protein and mRNA expression and promoter methylation assays for RASSF1A and BLU genes ------------92 Figure 10 Correlation analysis of mRNA and protein expression between RASSF1A and BLU --------------------- 95. IV.

(8) Abbreviations AD. adenocarcinoma. BLU (ZMYND10). zinc finger, MYND-type containing 10. ChIP. chromatin immunoprecipitation. CGI. CpG island library. COL14A1. collagen, type XIV, alpha 1. CTCF. CCCTC-binding factor. DMH. differential methylation hybridization. DNMTs. DNA methyltransferases. EMSA. electrophoretic mobility shift assay. HATs. histone acetyltransferases. HDACs. histone deacetylases. HMTs. histone methyltransferases. IHC. immunohistochemistry. K. Lysine. LC. large cell carcinomas. LOH. loss of heterozygosity. MBD. methyl-CpG binding domain. MSO. methylation-specific oligonucleotide. MSP. methylation-specific PCR. NPC. nasopharyngeal carcinoma. NSCLC. non-small cell lung cancer. R. Arginine. RASSF1A. Ras association domain family member 1 isoform A. RLGS. restriction landmark genomic scanning. RT. reverse transcription. SCLC. small cell lung cancer. SQ. squamous carcinoma. TNM. tumor-node-metastasis. TSGs. tumor suppressor genes V.

(9) 中文摘要癌症一般認為與基因體與外顯基因體發生變異有關，而在外顯基因體研究中，基因啟動子過度甲基化是最主要造成基因不活化的原因之一。抑癌基因的啟動子過度甲基化，會造成抑癌基因不活化，進而導致癌症的發生。為了鑑定在癌症基因體中，過度甲基化的區域所包含的可能新穎抑癌基因，本研究利用差異甲基化雜交法（differential methylation hybridization）的微陣列分析及染色質免疫沈澱晶片分析法（chromatin immunoprecipitation -on -chip），針對 30 位非小細胞肺癌病人及數個肺癌細胞株進行基因體的過度甲基化區域及染色質鬆緊狀態研究。結果發現在不同的肺癌子類型及肺癌分期，有特定的基因被過度甲基化，這些過度甲基化基因也許可以作為早期偵測及預測癌症發展的生物指標。此外，在肺癌病人的差異甲基化雜交法的結果中，本研究發現一個與抗細胞增生、細胞靜止與細胞分化的 COL14A1 基因啟動子有過度甲基化的情形，而且在染色質免疫沈澱晶片分析法中，COL14A1 啟動子的染色質區域相較於正常肺細胞，肺癌細胞呈現較為緊密的狀態。此外，本研究發現有 60.4%的非小細胞肺癌病人有 COL14A1 基因啟動子過度甲基化的情形，而且其 mRNA 及蛋白質分別有 50.0%及 43.9%的低表達情形；另外本研究也發現 COL14A1 基因啟動子過度甲基化與晚期肺癌病人有統計相關。這些驗證實驗顯示外顯基因體研究是尋找癌症相關基因的有效工具，COL14A1 基因及其蛋白變異參與肺癌的分子機制將進一步由細胞及動物模式研究來鑑定。在本實驗室先前對基因體缺失的研究中，發現在染色體 3p21 的區域有高達 50%以上的基因座缺失情形。此外，在差異甲基化雜交法的結果中，也發現位於染色體 3p21.3 的 RASSF1 基因在肺癌早期 1.

(10) 的病人中有過度甲基化情形，因此染色體 3p21.3 區域的基因不活化對於台灣地區肺癌形成扮演一個非常重要的角色。而 RASSF1A 及 BLU 這二個頭尾相連的抑癌基因位於染色體 3p21.3 的區域，由於這二個基因位置非常靠近，因此本研究預測這二個基因的表達及啟動子過度甲基化具有區域效應，也就是此二基因的表達及基因甲基化具有一致性。如果沒有區域效應，可能是因為 RASSF1A 及 BLU 基因之間具有絕緣子（insulator）構造所導致。首先，本研究針對 32 位肺癌病人，利用特定序列甲基化微陣列分析法（methylationspecific oligonucleotide microarray）及反轉錄聚合連鎖反應，找出會影響 RASSF1A 及 BLU 基因 mRNA 表達的關鍵轉錄 CpG 位置。同時也發現在 RASSF1A 基因的關鍵轉錄 CpG 位置上，有 E2F1 這個轉錄因子的結合，當這些位置被過度甲基化時，會使 E2F1 無法結合在 RASSF1A 的啟動子上，導致 RASSF1A 基因表達下降。此外，本研究發現 RASSF1A 及 BLU 這二個基因各自的關鍵轉錄 CpG 位置的甲基化與各自基因的低轉錄與低轉譯有關；然而，這二個基因的甲基化狀態及基因表達卻沒有一致性，也就是沒有區域效應。利用免疫沈澱聚合連鎖反應（chromatin immunoprecipitation-PCR）證明 CTCF 蛋白結合在 RASSF1A 及 BLU 基因啟動子之間的絕緣子上，也利用亞硫酸鹽定序（bisulfite sequencing）發現在絕緣子兩端的甲基化不連續情形。所以 CTCF 也許提供了屏障效應導致這二個基因沒有所謂的區域效應。本研究找出了 RASSF1A 及 BLU 的關鍵轉錄 CpG 位置，這些位置的甲基化會影響基因的表達；同時也證明了 CTCF 結合在 RASSF1A 及 BLU 之間，使得這二個基因的表達沒有區域效應。本研究為首篇鑑定影響 RASSF1A 及 BLU 基因 mRNA 表達的關鍵轉錄 CpG 位置的報導，並提出絕緣子可以做為如染色體 3p21 基因群座（gene cluster）屏障效應的證據。. 2.

(11) Abstract Cancer is caused by the accumulation of both genetic and epigenetic changes. Promoter hypermethylation is one of the major epigenetic changes. that. cause. gene. inactivation.. Aberrant. promoter. hypermethylation of CpG islands associated with tumor suppressor genes (TSGs) can lead to transcriptional silencing and result in tumorigenesis. The genomic regions with hypermethylation status may possess. novel. microarray-based. candidate. TSGs.. The. epigenome-wide. present. methylation. study. used. analysis. a. called. differential methylation hybridization (DMH) to identify the regions of hypermethylation and a chromatin immunoprecipitation (ChIP)-on-chip analysis to identify the regions of condensed or open chromatin in 30 non-small cell lung cancer (NSCLC) patients and several lung cell lines, and. have. successfully. detected. several. cancer. subtype-. and. stage-specific hypermethylated genes. They may serve as biomarkers for the early detection or prognosis prediction of lung cancer. Using DMH, this study identified promoter hypermethylation of the COL14A1 (collagen, type XIV, alpha 1) gene, which has cell anti-proliferative activity and plays a role in cell quiescence and differentiation. Using ChIP-on-chip, COL14A1 promoter region was shown to be in compact chromatin structure in cancer cell lines compared to normal cell line. In addition, 60.4% of 48 NSCLC patients showed COL14A1 promoter hypermethylation and coincided with low mRNA and protein expression. Moreover, COL14A1 promoter hypermethylation was significantly associated with late stage lung cancer patients. The present study provided evidence that epigenomic tools such as DMH and ChIP-on-chip can be used for identifyication of cancer-related genes such as COL14A1. 3.

(12) In previous study of my laboratory, there was more than 50% of loss of heterozygosity in chromosome 3p21. In addition, RASSF1 promoter hypermethylation in chromosome 3p21.3 was shown in early stage patients of NSCLC in DMH data. Therefore, gene silencing in chromosome 3p21.3 is important for lung tumorigenesis in Taiwan. Tumor suppressor genes RASSF1A and BLU are two tandem head-to-tail genes located at 3p21.3. The current study hypothesized that there may be a regional effect on their gene expression and promoter methylation status. If not, then there may be an insulator between RASSF1A and BLU genes. This study first identified transcriptionally important CpG sites using the methylation-specific oligonucleotide microarray in relation to mRNA expression of RASSF1A and BLU genes in primary lung tumors. The data demonstrated that E2F1 bound to the transcriptionally important CpG sites in RASSF1A gene, and this transcriptional regulation was impaired when the targeted CpGs were hypermethylated. Both RASSF1A and BLU genes had their own transcriptionally important CpG regions. However, there was no correlation of methylation status between RASSF1A and BLU. Using chromatin immunoprecipitation-PCR (ChIP-PCR), CCCTC-binding factor (CTCF) was found to bind to insulator sequences located between these two genes. Bisulfite sequencing and ChIP-PCR revealed distinct methylation and chromatin boundaries separated by the CTCF binding domains. This study dissects for the first time the transcriptionally important CpG sites for both RASSF1A and BLU genes and demonstrates that CTCF binding to the insulator of BLU gene possesses a barrier activity within separate epigenetic domains of the juxtaposed BLU and RASSF1A loci in the 3p21.3 gene cluster region.. 4.

(13) Study rationale Lung cancer is one of the most common malignancies in the world and is the leading cause of cancer-related deaths in industrial countries (Jemal et al., 2004). In 2008, lung cancer is the leading and second-leading cause of cancer deaths in women and men in Taiwan, respectively (Department of Health, 2008). The prognosis of lung cancer is very poor for patients whose tumors cannot be completely resected, with almost 70% of the patients dying from the disease within 2 years of diagnosis (Bains, 1991). Therefore, improvements in early detection, couple with identification of a gene or genes that predispose individuals to lung cancer, could help reduce the death rate for lung cancer. Human cancers develop through multistep processes involving mutations in several types of genes, including those inactivating recessive tumor suppressor genes (TSGs), activation of dominant oncogenes, and inactivation of genes involved in DNA repair or replication (Fearon, 1997). Both copies of TSGs have to be inactivated for their function to be lost (Knudson, 1996). One allele may be inactivated by point mutation, methylation changes, or small deletions. The other allele is frequently inactivated by a large deletion involving the gene of interest as well as adjacent stretches of DNA (Kohno and Yokota, 1999). Thus, searches for genomic regions frequently deleted or hypermethylated in cancer have helped to identify or confirm the location of several TSGs. Hypermethylation of CpG islands is alternative mechanism to inactivate TSGs, and is a major epigenetic modification in mammalian genome that is not accompanied by changes in DNA sequence (Nephew 5.

(14) and. Huang,. 2003).. The. cancer. cells. undergo. changes. in. 5’-methylcytosine distribution including global DNA hypomethylation and the region-specific hypermethylation of promoter CpG islands associated with TSGs have long been known (Laird, 2005). Aberrant promoter hypermethylation of CpG islands associated with TSGs can lead to transcriptional silencing and result in tumorigenesis. DNA methylation is frequently not restricted to a single CpG island but affects multiple independent loci, reflective of a widespread deregulation of DNA methylation pattern in different types of tumors (Costello et al., 2000; Belinsky, 2004). Genomic screening of 98 different primary human tumors has revealed that on an average there exist about 600 aberrantly methylated CpG islands in each tumor (Costello et al., 2000). For example, silencing of TSGs are mediated by DNA methylation has been shown for glutathione S-transferase 1, O6-methylguanine-DNA methyltransferase. (MGMT),. p15INK4b,. RB,. E-cadherin. (CDH1),. H-cadherin (CDH13), death-associated protein kinase 1 (DAPK1) (Robertson, 2001). In addition, in previous publications of my laboratory, we show that silencing of TSGs is mediated by DNA methylation and is important in processes of lung tumorigenesis including hMLH1(Wang et al., 2003), hMSH2 (Hsu et al., 2005), FHIT (Tzao et al., 2004), BRCA1 (Lee et al., 2007), BRCA2 (Lee et al., 2007), XRCC5 (Lee et al., 2007), p16INK4a (Chen et al., 2002), p14ARF (Wang et al., 2005), RAR (Lin et al., 2007b), SLIT2 (Lin et al., 2007a), TIMP3 (Lin et al., 2007a), AXIN2 (Tseng et al., 2008), βTrCP (Tseng et al., 2008), hRAB37 (Wu et al., 2009), and HIC1 (Tseng et al., 2009). To search for specific genes with hypermethylated promoter and altered chromatin status in lung tumorigenesis, the present study performed two epigenomic analyses in lung tumor. First, differential methylation hybridization (DMH) was used to analyze the genome-wide 6.

(15) CpG islands methylation profile in pairs of matched tumor and normal lung tissues from 30 non-small cell lung cancer (NSCLC) patients, a major. subtype. of. lung. cancer. chromatin-immunoprecipitation-on-chip. in. Taiwan.. (ChIP-on-chip). Second, assay. was. performed using anti-acetylated histone H3, which recognizes open chromatin structure, in IMR90 normal lung cell line and three lung cancer cell lines (A549, H1299, and CL1-1) to detect the genome-wide chromatin status. In addition, gene methylation and chromatin status identified by DMH and ChIP-on-chip assay were validated in lung cancer patients. Furthermore, transcriptionally important CpG sites which regulate gene expression were examined in some candidate genes.. 7.

(16) Literature review. I. The classifications of lung cancer Lung cancer is the leading cause of cancer death in men and women in industrialized countries (Jemal et al., 2004). In 2008, an estimated 36.5 and 16.5 patients among 100,000 Taiwanese male and female individuals, respectively, were killed by lung cancer (Department of Health, 2008). Lung cancer is often classified into two major types: small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) (Minna et al., 1997). NSCLC, which accounts for 80-90% of lung cancers in Taiwan, includes two major histological subtypes, adenocarcinoma (AD) and squamous carcinoma (SQ) (Chen et al., 1990). Histology of tumor types was determined according to the World Health Organization classification method. The international tumor-node-metastasis (TNM) staging system (T= primary tumor, N= regional lymph nodes, M= distant metastasis) is the “international language” which was developed by American Joint Committee on Cancer for NSCLC, has provided specificity for diagnosis to determine prognosis and guide treatment recommendations. The present classifications of stage grouping are: T1N0M0, stage IA; T2N0M0, stage IB; T1N1M0, stage IIA; T2N1M0 and T3N0M0, stage IIB; and T3N1M0, T1N2M0, T2N2M0, T3N2M0, stage IIIA; the TNM subsets in stage IIIB- T4 any N M0, any T N3M0, and in stage IV- any T any N Ml, remain the same (Mountain, 1997; Rami-Porta et al., 2009). There are three standard ways to treat different stage lung cancer patients: surgery, radiotherapy, and chemotherapy treatment. However, the prognosis of lung cancer is very poor for patients, whose tumors cannot be completely resected, with almost 70% of the patients dying 8.

(17) from the disease within 2 years of diagnosis (Bains, 1991). A number of factors, such as tumor staging based on the results of physical and surgical examinations, are known to influence prognosis (Mountain, 2000). However, patients with the same stage often show varied prognoses. Furthermore, lung cancer classification based primarily on morphologic appearance of the tumor has limitations, because cancer is characterized by genetic, epigenetic, and phenotypic changes that result in a tremendous variability in clinical behavior. Therefore, epigenetic alterations detection methods could be used to identify patients at risk for disease metastasis and potentially improve survival of lung cancer. In addition, identification and characterization of the epigenetic changes that drive lung cancer development can shed light on the molecular mechanism involved in lung tumorigenesis.. II. Gene promoter hypermethylation and tumorigenesis Molecular changes in oncogenes and tumor suppressor genes (TSGs) are involved in lung tumorigenesis (Fong et al., 1999; Sekido et al., 1998). The TSGs are inactivated by genetic and epigenetic abnormalities, including deletion, mutation, loss of heterozygosity (LOH), promoter hypermethylation, and abnormal histone modifications (Agathanggelou et al., 2003b; Thiagalingam et al., 2002; Wang et al., 1998). Epigenetic modifications that do not involve changes in the DNA sequence, but result in changes in gene expression. Evidence is emerging that DNA methylation, histone modification, and alternative mRNA splicing are involved in various human epithelial cell cancers (Gronbaek et al., 2007; Sigalas et al., 1996). In mammals, DNA methylation usually occurs at cytosines located 5’ of guanines, known as CpG dinucleotides. The CpG islands contain cluster of CpG dinucleotides that are about 1-2 kb in. 9.

(18) length in or near the promoter and first exon regions of genes (Antequera et al., 1993). Hypermethylation of CpG islands is alternative mechanism to inactivate TSGs, and is a major epigenetic modification in mammalian genome (Baylin et al., 1998; De Smet et al., 1999; Jones et al., 1999). Promoter methylation may block the binding of transcription factors, or modify chromatin structure, which in turn blocks access by transcription factors (Antequera et al., 1990; Brandeis et al., 1994; Mummaneni et al., 1995). DNA methylation is frequently not restricted to a single CpG island but affects multiple independent loci, reflective of a widespread deregulation of DNA methylation pattern in different types of tumors (Costello et al., 2000; Belinsky, 2004). For example, silencing of TSGs are mediated by DNA methylation in different kind of cancers have been shown for APC, CASP8, CDKN2A, DAPK1, PTEN, RB, SMAD4, VHL, CDH13, and TP53 (Kurkjian et al., 2008; Luczak and Jagodzinski, 2006). In addition, in previous publications of my laboratory, we show that silencing of TSGs and repair genes are mediated by DNA methylation and is important in processes of lung tumorigenesis including hMLH1(Wang et al., 2003), hMSH2 (Hsu et al., 2005), FHIT (Tzao et al., 2004), BRCA1 (Lee et al., 2007), BRCA2 (Lee et al., 2007), XRCC5 (Lee et al., 2007), p16INK4a (Chen et al., 2002), p14ARF (Wang et al., 2005), RAR (Lin et al., 2007b), SLIT2 (Lin et al., 2007a), TIMP3 (Lin et al., 2007a), AXIN2 (Tseng et al., 2008), βTrCP (Tseng et al., 2008), hRAB37 (Wu et al., 2009), and HIC1 (Tseng et al., 2009). The enzymes that are responsible for DNA methylation are DNA methyltransferases (DNMTs). DNMTs methylate the cytosine residue of CpGs, and five types have been identified (DNMT1, 2, 3a, 3b, and 3L). DNMT1 has been shown in vitro to prefer hemimethylated over unmethylated DNA 30 to 40-fold (Fatemi et al., 2001; Goyal et al., 2006; 10.

(19) Pradhan et al., 1997). It is referred to as a “maintenance” methyltransferase and is the primary enzyme responsible for copying methylation patterns after DNA replication because it localizes to replication foci and interacts with PCNA (Chuang et al., 1997). However, recent evidences show that DNMT1 may also work together with DNMT3a and DNMT3b in “de novo” methyltransferase activity in certain genome in both embryonic cells and differentiated somatic cells (Ko et al., 2005; Ratnam et al., 2002). Overexpression of DNA methyltransferases DNMT1, DNMT3a, and DNMT3b has been reported in various malignancies including hepatomas, prostate, colorectal, breast, and lung tumors (Girault et al., 2003; Lin et al., 2007; Patra et al., 2002; Saito et al., 2003).. III. Chromatin structure and gene expression status The chromatin structure composed of DNA, histones, and nonhistone proteins. The basic unit of chromatin is the nucleosome, ~146bp of DNA wrapped around the histone octamer composed of two copies of each of four histones: H2A, H2B, H3, and H4. Posttranslational modifications of histones, including acetylation, phosphorylation, methylation, ubiquitination, and sumoylation, play important roles in regulating gene expression (Fischle et al., 2003; Witt et al., 2009). The histones are epigenetically modified by histone acetyltransferases (HATs), histone deacetylases (HDACs) and histone methyltransferases (HMTs). These enzymes acetylate, deacetylate or methylate at amine groups of histone Lys (K) or Arg (R) amino acid residues, respectively (Zhang and Reinberg, 2001). The amino acids at the N-terminal ends of histone which hyperacetylation is generally associated with loose chromatin, accessibility of DNA to binding. 11.

(20) proteins and increased transcriptional activity, whereas histone hypoacetylation. contributes. to. chromatin. condensation. and. transcriptional repression (Tse et al., 1998; Wang et al., 2001). Furthermore, a variety of posttranslational covalent modifications have been shown to specifically ‘‘mark’’ the transcriptional status of chromatin. The histone acetylation and histone methylation can mark both transcriptionally active and repressive chromatin depending on which amino-acid residues are involved: trimethylation of lysine 9, 27 and 36 at the N-terminal tail of histone H3 (H3-K9, H3-K27, H3-K36) and lysine 20 on histone H4 (H4-K20) are silencing signatures, while trimethylation of lysine 4 and 79 on histone H3 (H3-K4, H3-K79) has been associated with actively transcribed chromatin (Bannister and Kouzarides, 2005; Margueron et al., 2005). Although DNA methylation and histone modification are carried out by different chemical reactions and require different sets of enzymes, there seems to be a biological relationship between the two systems that plays a part in modulating gene repression programming in the organism (Meissner et al., 2008; Mohn et al., 2008).. IV. Alterations of genes in chromosome 3p21.3 involve in lung tumorigenesis Allelic loss of chromosome 3p21.3 are the most frequent genetic alterations in many sporadic cancers, including hepatocellular carcinomas, gallbladder carcinoma, breast cancer, and non-small cell lung cancer (NSCLC) (Ito et al., 2005; Martinez et al., 2001; Riquelme et al., 2007; Tischoff et al., 2005). In previous study of my laboratory, the region in 3p21 shows more than 50% of LOH in 71 microdissected samples of surgically resected primary NSCLC tumors compared to. 12.

(21) their matched normal lung tissues (Tseng et al., 2005), suggesting that allelic loss in 3p21.3 and the inactivation of genes in this locus are involved in lung tumorigenesis. The loss of function of a group of genes at 3p21.3 such as RASSF1A, BLU/ZMYND10 (BLU), and SEMA3B in different human carcinomas has been reported (Riquelme et al., 2007; Tischoff et al., 2005). RASSF1A (NM_007128) is a TSG that encodes a member of RAS effectors regulating cell proliferation and apoptosis (Donninger et al., 2007). RASSF1A and NORE1A can form complexes with MST1 or interact with CNK1 and MOAP1 to modulate apoptosis (Ortiz-Vega et al., 2002; Rabizadeh et al., 2004; Vos et al., 2006). Song et al. (2004) reported that RASSF1A regulated mitosis by inhibiting the APC-Cdc20 complex and mitotic arrest at pro-metaphase. However, Liu et al. (2007) demonstrated that there was no simple interaction between RASSF1A and Cdc20. Nevertheless, RASSF1A is involved in successful completion of cytokinesis. In addition, RASSF1A and RASSF1C were found to be involved in the control of microtubule polymerization and potentially in the maintenance of genomic stability (Vos et al., 2004). Overexpression of RASSF1A diminished the migration ability of lung cancer cell line, increased cell-cell adhesion and showed less refractive morphology (Dallol et al., 2005). A study demonstrated that re-expression of RASSF1A reduced colony-formation ability in glioma cell line (Hesson et al., 2004). Hypermethylation of RASSF1A promoter is frequently found in many human cancers such as lung, breast, kidney, gastric, bladder, neuroblastoma, medulloblastoma, and gliomas tumors (Agathanggelou et al., 2005). Furthermore, RASSF1A methylation correlated with adverse survival in lung cancer patients (Burbee et al., 2001; Kim et al., 2003). BLU (NM_015896) has homology to the MTG/ETO family of 13.

(22) transcription factors. A previous report showed that exogenous expression of BLU in lung cancer or neuroblastoma cell line resulted in reduced colony formation efficiency in vitro (Agathanggelou et al., 2003), and demonstrated that BLU could functionally suppress tumor formation in nude mice (Yau et al., 2006). Down-regulation of BLU RNA expression was observed in nasopharyngeal carcinoma (NPC) cell lines (83%) and NPC biopsies (80%) (Yau et al., 2006). BLU promoter is commonly found to be hypermethylated in cancers such as in glioma, cervical squamous cell carcinomas, NPC, neuroblastoma, and NSCLC (Hesson et al., 2007). A previous study showed that BLU was a stress-responsive gene and was regulated by E2F1, however, this response was impaired when the promoter was hypermethylated (Qiu et al., 2004). RASSF1A and BLU gene loci are located next to each other in the region 3p21.3. There may be a regional effect of their CpG island hypermethylation and expression status. For example, a study has shown a concordant hypermethylation in these two genes in NSCLC patients (Agathanggelou et al., 2003). Nevertheless, in other studies, RASSF1A and BLU promoters have no correlation of hypermethylation in NPC, gliomas, and lung cancer patients (Hesson et al., 2004; Marsit et al., 2005; Qiu et al., 2004). However, alterations of these two genes have never been examined in the same series of lung cancer patients to investigate whether the regional effect indeed exists in these two closely located genes at the protein/RNA expression levels and promoter methylation status.. V. Methods for epigenetic alteration analysis Traditionally, methylation analysis has been carried out using 14.

(23) Southern hybridization, which assesses a few methylation-sensitive restriction sites within CpG islands of known genes (Gonzalgo et al., 1997). Further development of sensitive assays, several techniques of analyzing. DNA. methylation,. such. as. bisulfite. conversion,. methylation-sensitive enzyme restriction, and affinity purification of methylation DNA, have been adapted (Beck and Rakyan, 2008). For example, the bisulfite DNA sequencing (Frommer et al., 1992) and methylation-specific PCR (MSP) (Herman et al., 1996), has allowed a detailed analysis of multiple CpG sites across a CpG island of interest. Using these methods, hypermethylation of promoter has been demonstrated in certain TSGs and repair genes such as cyclin-dependent kinase inhibitor p16INK4a, p53, hMLH1, von Hippppel-Lindau (VHL) (Cunningham et al., 1998; Gonzalez-Zulueta et al., 1995; Herman et al., 1994; Herman et al., 1995; Sakai et al., 1991). Until recently, several high-throughput technologies have been developed to determine the methylation profiles of thousands of CpG islands in several tumors (Eng et al., 2000; Huang et al., 1997; Ushijima et al., 1997). The restriction landmark genomic scanning (RLGS) was one of the earliest methods, which was based on the differential cleavage of isoschizomers with distinct methylation sensitivity, to be adapted for genome-wide methylation analysis in primary human tumors (Hatada et al., 1991). With this technique, DNA from normal and tumor samples is cleaved by a methylation-sensitive endonuclease, and DNA is radioactively labeled at cleaved sites. Following the digested DNA is separated in two-dimension gel, and normal samples are compared, more than 1000 CpG island methylation patterns can be analyzed. However, the limitations of RLGS include its reliance on specific digestion sites that are not present in all CpG islands and the fact that not all of the resulting fragments can be resolved in the 15.

(24) two-dimensional electrophoresis steps. In addition, bisulfite genomic sequencing is the best technique for mapping DNA methylation at single-base resolution, and provides a quantitative measure of methylation abundance, rather than relative measure from array-based methods. However, sequencing methods cost more than all the commonly used methods for whole-genome approaches, and direct bisulfite sequencing still cannot be carried out in human methylome analysis, because humans have a much larger genome. The DNA array-based method, called differential methylation hybridization (DMH), provides a tool that can efficiently scan the tumor genome for methylation alterations (Huang et al., 1999). The first part of DMH is the generation of CG-rich tags derived from a human CpG island genomic library, CGI (Cross et al., 1994). These tags are then arrayed onto solid supports (e.g., nylon membranes). The second part involves the preparation of amplicons, representing a pool of methylated CpG DNA from tumor or reference samples. These amplicons are used as probes for CpG island array hybridization. The differences in tumor and reference signal intensities on CpG island arrays tested reflect methylation alterations of corresponding sequences in the tumor DNA. Until now, genome-scanning approach of methylation profiles is only studied in ICF syndrome (Kondo et al., 2000), breast (Huang et al., 1997; Yan et al., 2000), and liver cancers (Ushijima et al., 1997). In breast cancer, close to 9% of the genomic tags of CGI exhibit extensive hypermethylation and a subset of CpG island methylation is associated with histological grades (Huang et al., 1997; Yan et al., 2000). Therefore, an essential next step toward a more comprehensive understanding of the underlying mechanisms of lung tumorigenesis would be a genome-wide analysis of hypermethylation in lung cancer. Such analyses can lead to the identification of previously uncharacterized 16.

(25) CpG islands associated with gene silencing and shed light on other yet unidentified factors governing aberrant methylation. In addition to genome-wide methylation analysis, single gene array-based DNA methylation profiling in cancer was also described in 2002 (Gitan et al., 2002). This methylation-specific oligonucleotide (MSO) microarray can be applied to map methylation CpG sites within the CpG island of a single gene in normal or tumor DNA. Essentially, this technique involves sodium bisulfite conversion of genomic DNA and PCR amplification of regions of interest. The fluorescently labeled PCR products are hybridization to arrayed oligonucleotides that can discriminate between methylated and unmethylated alleles in interest region, and provide readout of the original methylation state at that CpG site. The MSO microarray have been used for mapping methylation changes in CpG island loci of ERα, RASSF1A, p16INK4A, and hMLH1 genes in different cancers (Gitan et al., 2002; Mund et al., 2005; Yan et al., 2003; Zhang et al., 2006). To understand of the histone modifications correlated with transcriptional outcomes, the chromatin immunoprecipitation (ChIP) assay has been exploited to discriminate histone modification patterns in vivo. The approach focused on histone modifications profiling and used ChIP to pull down the histone methylation or acetylation region DNA. The commonly used ChIP assay includes a formaldehyde cross-linking step to capture the DNA sequence to the protein or modified protein of interest (Solomon and Varshavsky, 1985). Briefly, ChIP involves fixing cells using formaldehyde, thereby crosslinking DNA-binding proteins to the chromosome, followed by cell lysis and shearing of DNA by sonication. The protein of interest is then immunoprecipitated with specific antibodies together with any crosslinked DNA fragments. After reversal of the crosslinks and purification, DNA can be analyzed in 17.

(26) order to detect enrichment of the sequences bound by the protein of interest. Thus, using ChIP in conjunction with DNA microarray analysis (ChIP-on-chip) permits DNA binding to be measured on a genome-wide scale. In addition, direct sequencing analysis of ChIP DNA samples (termed ChIP-seq) was conducted using high-throughput short-read sequencer which provided high-resolution genome-wide maps of composition and structure of promoter nucleosomes facilitate or inhibit transcription (Jiang and Pugh, 2009). After data analysis, genome-wide distributions of histone modifications and chromatin protein target sites could be revealed.. 18.

(27) Specific aims. Lung cancer is the leading cause of cancer deaths in Taiwan. It has been shown that alterations of proto-oncogenes and TSGs are critical in the. multi-step. Down-regulation. development of. TSGs. and and. progression cancer. of. relative. lung. cancer.. genes. by. hypermethylation of 5’CpGs is one of the important events involved in tumor development. Therefore, studies association of hypermethylation with clinicopathological parameters and prognoses may shed lights on the molecular mechanisms of lung tumorigenesis.. Specific aim 1: Epigenome-wide detection for hypermethylated CpG islands and chromatin profiling and their clinical association in lung cancer.. Comparison of DMH and ChIP-on-chip to detect specific. methylation profile and chromatin structure profile was performed in lung cancer patients and cell lines. In addition, correlation of gene promoter hypermethylation / chromatin structure alterations with clinicopathological parameters (tumor type, stage, age, sex, and smoker) and DNMT1 expression status was analyzed in the lung cancer patients. The present study proposed that such epigenomics analyses can lead to the identification of previously uncharacterized CpG islands associated with gene silencing and shed light on other yet unidentified factors governing aberrant methylation.. Specific aim 2: Validated the cancer related genes and TSGs from DMH and ChIP-on-chip analysis, and characterization of their transcriptional important CpG sites in primary non-small cell lung 19.

(28) cancer.. Several hypermethylated cancer-related genes and TSGs in. patients were identified using the DMH analysis, such as COL14A1, RASSF1A, and BLU. The present study further used the RT-PCR, MSP, ChIP-PCR, and bisulfite sequencing to validate their potential role as hypermethylated genes in lung cancer patients and cell lines. In addition, MSO microarray was conducted to identify the CpG sites important for gene expression by comparing the methylation and chromatin structure profiles with the mRNA and protein expressions of the candidate gene. The present study aims to validate gene methylation and chromatin status identified by DMH and ChIP-on-chip assay such as COL14A1 in lung cancer patients. Furthermore, transcriptionally important CpG sites which regulate gene expression will be examined in some candidate genes such as RASSF1A and BLU.. 20.

(29) AIM 1. Epigenome-wide detection for hypermethylated CpG islands and chromatin profiling and their clinical association in lung cancer. 21.

(30) Purposes 1. Comparison. of. chromatin. immunoprecipitation-on-chip. (ChIP-on-chip) using the acetylated histone H3 antibody and differential methylation hybridization (DMH) data to detect specific chromatin structure profile and methylation profile in lung cancer. 2. Correlation of gene promoter hypermethylation / chromatin structure alterations with clinicopathological parameters (tumor type, stage, age, sex, smoker) and prognosis of the lung cancer patients. 3. Identification of the CpG sites important for gene expression by comparing the methylation and chromatin structure profiles with the mRNA and protein expressions of the candidate gene.. 22.

(31) Materials and Methods Tumor specimens and clinical characterizations of patients Tissues were collected after obtaining permission from the appropriate institutional review board and informed consents from the recruited patients. The cases include 30 surgically resected NSCLC patients admitted to Veterans General Hospital-Taipei, Taiwan between 1993 and 2004. The histologies of tumor types and stages are determined according to the WHO classification method and TNM system, respectively. There were 15 AD patients and 15 SQ patients. The patients consisted of 15 early stage patients (stages I and II) and 15 late stage patients (stages III and IV). Cell lines and culture conditions Human normal lung cell line IMR90 (obtained from the American Type Culture Collection, ATCC), human NSCLC carcinoma A549 and H1299 (obtained from ATCC), and CL1-1 (obtained from Dr. Pan-Chyr Yang, Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan) cell lines were cultured in DMEM medium (GIBCO, Grand Island, N.Y.) containing 10% fetal bovine serum (BIOCHROM AG, Leonorenstr, Berlin) and 1% penicillin-streptomycin (GIBCO), and incubated at 37°C in 5% CO2 atmosphere. Fractionation of DNA using an MBD column and preparation of a CpG island library DNA encoding the methyl-CpG binding domain (MBD) from MeCP2 was amplified by PCR and cloned into a bacterial expression vector containing a His-tag. The fusion histidine-MBD protein was attached to a nickel-agarose matrix column. Human DNA digested with 23.

(32) MseI (recognition site TTAA), which cuts bulk DNA into small fragments but leaves CpG island relatively intact, was loaded onto the MBD protein column. The DNA eluted in high salt fractions contained genomic regions of CpG islands, this DNA was collected for the CpG island DNA and construction of the CpG island library (CGI). Preparation of CpG island array slides A total of ~8000 CGI clones were individually organized in 96-well culture chambers as master plates. Clones of CGI were picked and placed into 96-well PCR microplates by MULTI-PRINT replicator. The insert from each clone was amplified by PCR. Prior to printing, spotting solution (Genetix, Hampshire, UK) was added to the PCR products for uniform dot morphology. The QArray2 arrayer (Genetix) was used to print PCR product on UltraGAPS coated slides (Corning, NY, USA). Printed slides were crosslinked by UV, post-processed, and stored in a desiccated chamber. Amplicon generation DMH method was performed as described previously (Yan et al., 2002). Briefly, DNA (2 μg) was extracted from surgically resected tumors and their matched normal tissues and then digested with MseI, which restricted DNA into small fragment but leaves CpG islands relatively intact. The cleaved ends were ligated with linkers, H-12/H-24 (linkers primer are shown in Table 1), in a buffer containing Fast Link Ligase (Epicentre, Madison, Wisconsin) at room temperature for 1 hour. After ligation, the subtracted DNA was digested with BstUI and HpaII (New. England. Biolabs,. Beverly,. MA). methylation. sensitive. endonucleases. The restricted DNA was purified in a QIAquick column (Qiagen, Valencia, CA) Linker-PCR reactions were performed with the BstUI and HpaII-resistant DNA. Reactions were carried out in a 24.

(33) volume of 300μl with digested DNA, 10 μM H-24 primers, 10x PCR buffer, 10 mM dNTP, and Deep Vent (exo-) DNA polymerase (New England Biolabs) on a DNA thermal cycler. PCR reaction condition was 5 min at 72oC, followed by 20 cycles of 1 min at 97oC, 3 min at 72oC, with a final extension of 10 min at 72oC. The amplified product was purified in a QIAquick column. At least 5-6μg of normal or tumor amplicons was used for fluorescence labeling to produce a strong array hybridization image. Cy dye coupling and array hybridization Incorporation of aminoallyl dUTP (Sigma, St. Louis, MO) into amplicons was performed using the BioPrime DNA labeling system (Invitrogen, Carlsbad, CA). Cy3 and Cy5 fluorescent dyes (Amersham, Buckinghamshire, UK) were coupled to aminoallyl dUTP-labeled normal and tumor amplicons, respectively. Both the Cy3-coupled normal and Cy5-coupled tumor amplicons were mixed with 20μg of human Cot-1 DNA (Invitrogen) and cohybridized to the CpG island array slide. Hybridization and post-hybridization washing protocols were according to previous study (Yan et al., 2002). Images of fluorescence intensities were generated by scanning array using Axon GenePix 4000B scanner (Molecular Devices, Sunnyvale, CA) Data analysis The fluorescence intensities of each spots were quantified by GenePix Pro 6.0 microarray analysis software (Molecular Devices) and the ratio was then determined as the intensity of the tumor sample vs. the normal sample (Cy5/Cy3). Raw intensity data of the ratio of the 30 arrays were normalized by LOWESS (limmaGUI package) (Wettenhall et al., 2004). After normalization, each array intensity data was ranked. 25.

(34) and selected for the top 200 spots, which were clustered and intersected with patients’ data by tumor type and tumor stage. Immunohistochemistry (IHC) assay Paraffin blocks of tumors were dissected into 5-m slices and then processed using standard deparaffinization and rehydration techniques. Polyclonal antibody for anti-DNMT1 (A22659, aa 177-550; 1:200; Asia Hepato Gene Co., Kaohsiung, Taiwan) was used as the primary antibodies to detect DNMT1 protein expression. The primary antibody was incubated for 4 oC overnight, then biotinylated secondary antibody for 30 min, streptavidin-horseradish peroxidase for 15 min (DAKO LSAB kit K0675, Dako, Glostrup, Denmark), 3,3’-diaminobenzidine tetrahydrochloride. as. the. chromagen.. The. evaluation. of. the. immunohistochemistry was conducted blindly without prior knowledge of the clinical and pathologic characteristics of the cases. DNMT1 was graded overexpression when tumor cell staining are >60%. The surrounding normal stroma and epithelial cells served as an internal positive control for each slide. Cell lines chromatin immunoprecipitation (ChIP)-on-chip assay Cells grown on a 100 x 20-mm culture dish to approximately 80–90% confluency were cross-linked with 1% formaldehyde for 10 min at 37oC and stopped by the addition of glycine to a final concentration of 0.125 M. Lysates were sonicated by BioruptorTM system (Diagenode, Liège, Belgium) to shear DNA to lengths between 200 and 800 bp. Subsequent steps were performed with the ChIP assay kit (Upstate Biotechnology, Lake Placid, NY, USA) according to the manufacturer’s instructions. Immunoprecipitation for 16 h at 4oC was performed. using. anti-acetylated. histone. 26. H3. (1:400;. Upstate.

(35) Biotechnology) in IMR90 normal lung cell line and 3 lung cancer cell lines (A549, H1299, and CL1-5). After de-crosslinking and purification of DNA, the immunoprecipitated and whole-cell extract DNA were amplified and labeled by ligation-mediated PCR with Cy5 and Cy3 fluorescent dyes, respectively. Both pools of labeled DNA were hybridized to the Agilent human CpG island 244k microarray chip (Agilent Technologies, Foster City, CA). Images of fluorescence intensities were generated by scanning array using GenePix 4000B scanner (Axon Instruments), and then data were extracted and normalized ChIP signals using Agilent Feature Extraction program.. 27.

(36) Results Epigenomic detection of hypermethylated CpG islands and their clinical association in lung cancer patients To search for the genomic regions containing high frequency of hypermethylation, which may possess cancer related genes or novel candidate TSGs, we established an epigenomic hypermethylation analysis using the microarray-based strategy called DMH (Figure 1a) and used it to identify the regions of hypermethylation in 30 lung cancer patients. Patients included 15 AD patients and 15 SQ patients consisting of 15 early stage patients (stages I and II) and 15 late stage patients (stages III and IV) (Table 2). Several cancer subtype- and stage-specific hypermethylated genes successfully identified by DMH are listed in Table 3 and Table 4. In addition, we had identified several hypermethylated TSGs in patients with overexpression of DNMT1 using the DMH and IHC analysis. They are listed in Table 5. Chromatin status was involved in cancer related genes expressions in lung cancer cell lines To investigate whether compact chromatin structure in addition to promoter hypermethylation of cancer related loci caused the loss of their gene expression, IMR90 normal lung cell line and three lung cancer cell lines (A549, H1299, and CL1-1) were analyzed by ChIP-on-chip (Figure 1b). ChIP was performed using antibody against acetylated histone H3, which recognizes loose chromatin structure. The immnoprecipitated DNA was then hybridized to the array containing oligonucleotide probes designed for genes’ promoters. The value of each probe was calculated as the ratio between ChIP-pulled down DNA and total input DNA from the same cell line. The higher the ratio, the looser. 28.

(37) the chromatin structure is. For example, among the gene analyzed, a group of 10 oligonucleotide probes was designed to test chromatin status of COL14A1 gene within the CpG island (Figure 2a). We found that the CpG island of COL14A1 gene showed looser chromatin structure in IMR90 normal cell line than did all cancer cell lines (Figure 2b).. 29.

(38) Discussion. Down-regulation. of. cancer-related. genes. and. TSGs. by. hypermethylation or chromatin alteration of promoter is one of the important events involved in tumor development. In this study, we have established DMH microarray for epigenome-wide hypermethylation analysis to identify the regions of hypermethylation and ChIP-on-chip analysis to identify the regions of condensed or open chromatin in 30 lung cancer patients and several lung cell lines. Some cancer subtypeand stage-specific hypermethylated genes are shown in Table 3 and Table 4. These genes may serve as biomarkers for early detection (for example, those hypermethylated genes in early-staged patients) or prognosis prediction (for example, those hypermethylated genes only in late-staged patients). In addition, using an antibody to acetylated histone H3 lysine 9, which indicates an open chromatin structure, several chromosomal regions containing potential TSGs such as COL14A1 and RASSF1 genes were revealed. DMH has a growing number of scanning methods for detecting hypermethylated genomic regions in cancer (Hatada et al., 1991; Ushijima et al., 1997; Gonzalgo et al., 1997). This approach has at least three unique features. First, DNA is more stabilization than mRNA and protein in experiment process, so that DNA biomarkers may be better used for clinical-detecting of cancers. Second, the high-density of DMH array-based technology has been used in differential screenings of thousands of cDNA sequences up- or down-regulated in cancer cells. Therefore, this DNA array chip could be used for dual analysis of DNA methylation and gene expression in cancer cells to provide an effective tool to elucidate the mechanisms of aberrant DNA methylation in cancer 30.

(39) (Shi et al., 2002). Third, the genomic fragments were derived from a library specifically constructed to contain highly enriched CpG island sequences (Cross et al., 1994). Therefore, DMH may lead to the identification of not only known expressed sequences but also novel TSGs down-regulated via methylation in cancer. For these reasons, DMH is useful for a genome-wide screening of methylation in cancer and can be converted into a high-throughput analysis by implementing the aforementioned microarray technologies. The multiple covalent modifications on the histone tail create specific. epigenetic. patterns. that. switch. genes. between. their. transcriptionally active and inactive stages (Moggs et al., 2004; Zhang and Reinberg, 2001). HDACs, HATs, HMTs and DNMTs play crucial roles in the epigenetic regulation of gene expression involved in carcinogenesis (Geiman et al., 2004; Sun et al., 1997). Early evidences suggest that histone methylation may even be an important determinant of DNA methylation patterns (Jackson et al., 2002; Tamaru and Selker, 2001). In previous studies, alterations in DNA methylation associated with gene expression have been shown to be intimately associated with changes in chromatin structure (Jones and Baylin, 2002). Thus, epigenetic gene regulation involves a complex interplay between DNA methylation and chromatin modification. Finally, we have shown that DMH and ChIP-on-chip methods are useful in identifying epigenetic alteration in cancer and have potential applications for cancer diagnostic and prognostic biomarkers. These biomarkers will be validated in a larger cohort with prospective study design. In addition, functional and mechanistic studies for the genes identified will be performed to reveal their roles in lung tumorigenesisis.. 31.

(40) AIM II. Validated the cancer related genes and TSGs from DMH and ChIP-on-chip analysis, and characterization of their transcriptional important CpG sites in primary non-small cell lung cancer. 32.

(41) Purposes. 1. Verification of the status of COL14A1, RASSF1A, and BLU identified from DMH and ChIP-on-chip analysis by comparing the alterations of methylation, mRNA, and protein expression data in NSCLC patients. 2. Identification of the methylated CpG sites of RASSF1A and BLU which are important to transcriptional regulation. 3. Investigation whether there was a regional effect on the gene expression and promoter methylation status of the juxtaposed RASSF1A and BLU genes. If not, then there may be an insulator between RASSF1A and BLU genes.. 33.

(42) Materials and methods Cell lines and culture conditions Human normal lung cell line IMR90 and MRC5 (obtained from the American Type Culture Collection, ATCC) and human lung cancer cell lines A549 and H1299 (obtained from ATCC), and CL1-0, CL1-1, CL1-3 and CL1-5 (obtained from Dr. Pan-Chyr Yang, Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan) were cultured in DMEM (IMR90, MRC5, A549, H1299, CL1-0, CL1-1, CL1-3, and CL1-5) medium (GIBCO, Grand Island, N.Y.) containing 10% fetal bovine serum (FBS) (BIOCHROM AG, Leonorenstr, Berlin) and 1% penicillin-streptomycin (GIBCO) and incubated at 37°C in 5% CO2 atmosphere. Clinical samples preparation and DNA/RNA extraction Tissues were collected after obtaining appropriate institutional review board permission and informed consent from the recruited patients. Surgically resected tumor tissue and corresponding normal tissue were collected from 76 patients diagnosed with primary non-small cell lung cancer admitted to Veterans General Hospital, Taipei. Of these patients, 55 had adenocarcinomas (AD), 18 had squamous cell carcinomas (SQ), and 3 had large cell carcinomas (LC). Histological classification was determined according to the WHO classification and the tumor-node-metastasis system. Information on the age, sex, and smoking history of the patients was obtained from hospital records. The genomic DNA was prepared using proteinase K digestion and phenol-chloroform extraction. Total RNA was prepared from tumors and normal lung tissues using TRIzol reagent (Invitrogen, Carlsbad, CA). cDNA was synthesized using SuperScript reverse transcriptase. 34.

(43) (Invitrogen). Semiquantitative multiplex reverse transcription-PCR (RT-PCR) assay cDNA was synthesized from the isolated RNA by reverse transcription. COL14A1, RASSF1A, and BLU mRNA expression was assayed in a multiplex RT-PCR analysis using the GAPDH gene as an internal control. The primer nucleotide sequences are shown in the Table 1. PCR products were separated in 1.5% agarose gels and visualized under UV illumination. To quantify the relative levels of mRNA expression in the multiplex RT-PCR assay, the value of the GAPDH in each reaction was used as the baseline gene expression of that sample and relative value was calculated for the COL14A1, RASSF1A, and BLU genes for each tumor and matched normal samples. Tumor cells that exhibited justified mRNA expression below that of normal cells were considered to have an abnormal pattern. Bisulfite modification and methylation-specific oligonucleotide (MSO) microarray Tumor DNA from 32 patients (23 were AD, 8 were SQ, and 1 was LC). was. analyzed. methylation-positive. by DNA,. MSO DNA. microarray. was. treated. To. prepare. with. M.. the SssI. methyltransferase (New England Biolabs, Beverly, MA) that methylates all cytosine residues of CpG dinucleotides in the genome. The test samples and SssI-treated DNA samples were treated with sodium bisulfite as previously described (Gitan et al., 2002). Preparation of MSO arrays was carried out essentially as described previously (Yan et al., 2004). The oligonucleotides and PCR primers for RASSF1A methylation assay were described in the previous study (Yan et al., 2003). Twenty sets of paired oligonucleotides were designed to include promoter and first exon of BLU CpG sites to be interrogated (Figure 3a). 35.

(44) For DNA target preparation, bisulfite-treated DNA was amplified from two regions (P1 as promoter region; E1 as exon 1 region) located in the BLU CpG island. Primers used for PCR to amplify bisulfite DNA are shown in the Table 1. To control the accuracy and reproducibility of the MSO probes, a series of MSO hybridization were performed with mixed samples. containing. 100,. 66,. 33,. and. 0%. SssI-treated. methylation-positive DNA (Yan et al., 2004). Standardization curves generated for these optimal probe sets were subsequently used to interrogate RASSF1A and BLU methylation in the test tumor DNA samples (data not shown). Slides hybridization and wash were described in the previous study (Gitan et al., 2002; Yan et al., 2004). The microarray slides were scanned with a GenePix 4000A scanner (Axon Instruments). MSO data were normalized according to a global ratio in each microarray image, and the intensity ratio of M/(M+U) (M: methylated probe; U: unmethylated probe) for each probe set was then derived. Methylation-specific PCR (MSP) assay MSP assay takes advantage of the fact that unmethylated cytosines are efficiently converted to uracil after sodium bisulfite treatment, whereas. methylated. cytosines. remain. unchanged.. EZ. DNA. Methylation-Gold™ Kit (Zymo Research, Orange, CA) was used for bisulfite conversion of DNA. After treatment, COL14A1 MSP analyses were conducted using the primers which are shown in the Table 1. In addition, the methylation status in the E2F1 binding sites of the RASSF1A and BLU promoters were determined by chemical treatment with sodium bisulfite and MSP analysis. MSP analyses were conducted using the primers for the –144 to –38 bp (MSP1) and -6 to +94 bp (MSP2) regions in RASSF1A promoter and exon1 and for -156 to +1 bp region in BLU promoter. Location of regions analyzed by MSP is shown 36.

(45) in Figure 3a and the primers used are listed in Table 1. Positive control samples of unmethylated IMR90 and MRC5 normal lung cell DNA and SssI methyltransferase treated methylated DNA were also included for each set of PCR. Bisulfite DNA sequencing analysis The methylation status of the upstream and downstream CTCF regions was examined by sequencing analysis after bisulfate modification. Bisulfite sequencing primers were designed for CpG sites -60 to +14 (corresponding to nucleotide numbers from -2380 to +161 bp) in which the first CpG site was defined as +1 corresponding to RASSF1A transcriptional start site (Figure 4b). Bisulfite-modified DNA was amplified using three pairs of primers: BS-P1, BS-P2, and BS-P3 shown in the Table 1. The products were isolated and sub-cloned into the pCR2.1-TOPO vector (Invitrogen), and five individual clones per each sample were sequenced. Immunohistochemistry (IHC) assay Paraffin blocks of tumors were dissected into 5-m slices and then processed using standard deparaffinization and rehydration techniques. Polyclonal antibody for anti-COL14A1 (1:200; Abnova, Taipei, Taiwan), monoclonal antibody for anti-RASSF1A (1:500; eBioscience, San Diego, CA), and polyclonal antibody for anti-BLU (1:100; Abcam, Cambridge, UK) were used as the primary antibodies to detect COL14A1, RASSF1A and BLU protein expression, respectively. The evaluation of the immunohistochemistry was conducted blindly without prior knowledge of the clinical and pathologic characteristics of the cases. COL14A1, RASSF1A and BLU were graded low or negative expression when tumor cell staining are <50%, <20% and <40%, respectively. The. 37.

(46) surrounding normal stroma and epithelial cells served as an internal positive control for each slide. Normal lung tissue slide from some patients were included to evaluate the IHC results. Electrophoretic mobility shift assay (EMSA) The E2F1 binding sites were identified using the transcription factor search program PROMO (http://alggen.lsi.upc.es/cgi-bin/promo_v3/ promo/promoinit.cgi?dirDB=TF_8.3) (Messeguer et al., 2002). The EMSA was performed using Raji nuclear extracts (Active Motif, Carlsbad, CA) and biotin end-labeled double-stranded DNA probes prepared by annealing complementary oligonucleotides. The probes of E2F1 binding sites for RASSF1A gene were as follows: probe 1, -169 to -138 bp; and probe 2, -7 to +25 bp. The probe sequences are shown in the Table 1. Probes were labeled in a reaction using terminal deoxynucleotide. transferase. (Promega,. Madison,. WI). and. biotin-14-dCTP (Invitrogen). The binding reaction was performed using the LightShift Chemiluminescent EMSA Kit (Pierce, Rockford, IL) and conducted according to the manufacturer’s protocol. Binding mixtures were loaded on a 6% polyacrylamide gel, and gels were run at 100 V for 1. hour. at. room. temperature.. Following. electrophoresis,. the. DNA-protein complexes were transferred onto nylon positively charged membranes and detected using chemiluminescence reagent (Pierce). Chromatin immunoprecipitation (ChIP) and target promoter ChIP-PCR Cells grown on a 100 x 20-mm culture dish to approximately 80–90% confluency were cross-linked with 1% formaldehyde for 10 min at 37oC and stopped by the addition of glycine to a final concentration of 0.125 M. Lysates were sonicated using BioruptorTM system (Diagenode, Liège, Belgium) to shear DNA to lengths between 200 and 800 bp. Subsequent steps were performed with the ChIP assay 38.

(47) kit (Upstate Biotechnology, Lake Placid, NY, USA) according to the manufacturer’s instructions. ChIP was performed using anti-acetylated histone H3 (Lys 9) (1:400; Upstate Biotechnology), anti-trimethylted histone H3 (Lys 27) (1:400; Upstate Biotechnology), anti-E2F1 (1:500; Active Motif), and anti-CTCF (1:500; Upstate Biotechnology) antibodies for 16 h at 4oC. De-crosslinks and purification of pulled down DNA were performed with the ChIP assay kit (Upstate Biotechnology). PCR analysis using the primer pairs for RASSF1A, BLU, and GAPDH promoters as well as CTCF binding domain and regions next to CTCF binding domain are shown in Table 1. Luciferase reporter gene analysis Promoter sequences of RASSF1A were cloned by PCR from IMR90 normal lung cell DNA. Primer pair used for PCR (-363 to +148 bp) are shown in Table 1 (Strunnikova et al., 2005). The fragment was cloned in to the pGL4 luciferase reporter vector (Promega). Mutation in E2F1 binding sites of RASSF1A gene were generated by QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) using the specific primers as: Mut-1, Mut-2 and Mut-3 (Table 1). A549 cell line was cotransfected with 1 μg of the constructs and 2 μg of the pCMV-SPORT6-E2F1 expression vector (Open Biosystems, Huntsville, AL) by using ExGen 500 in vitro transfection reagent (Fermentas, Glen Burnie, Maryland). After 24 hours the expression of luciferase gene was determined and normalized by Dual-Luciferase reporter assay (Promega). Statistical analysis The statistical analyses of promoter methylation, RNA/protein expression, patient survival and tumor characteristics were performed 39.

(48) using SPSS program (SPSS Inc., Headquarters Chicago, IL, USA). Comparisons were made using Chi-square test. P≤0.05 were considered to be statistically significant.. 40.

(49) Results Validation of promoter hypermethylation and mRNA/protein expression of COL14A1 in more lung cancer patients and cell lines COL14A1 was one of the hypermethylated genes identified in SQ patients by DMH, which has cell anti-proliferative activity and plays a role in cell quiescence and differentiation (Ehnis et al., 1996; Ruehl et al., 2005). To validate the methylation status of the COL14A1 promoter, this study used MSP to determine methylation pattern in lung cancer cell lines and 48 NSCLC primary tumors (including 30 patients analyzed by DMH). As demonstrated in Figure 5a, four cancer cell lines (CL1-0, CL1-5, A549 and H1299) showed methylation in COL14A1 promoter, and 60.4% (29/48) of lung tumors showed promoter hypermethylation of the COL14A1 gene. In addition, to determine whether there were decreased mRNA and protein expressions in COL14A1, multiplex semiquantitative RT-PCR and immunohistochemistry assays were performed in lung cancer patients whose RNA and tissue slide were available (Figure 5b and 5c). Decreased or absence of COL14A1 transcripts were shown by RT-PCR in 50.0% (23/46) tumors. Immunohistochemical staining data indicated that 43.9% (18/41) of tumors showed an absence or low expression of COL14A1 protein. The data indicated that COL14A1 had abnormal expression in lung cancer cell lines and patients. To investigate whether the COL14A1 promoter hypermethylation was associated with mRNA and protein expression, cross-tabulation analysis were done on the methylation, mRNA and protein expression data to examine their correlations, using the Pearson’s χ2 test. Aberrant promoter methylation was significantly associated with low mRNA transcript (P=0.012) (Table 6). However, there. was. no. correlations. of. protein. 41. expression. and. DNA.

(50) hypermethylation, and was no significantly associated with mRNA and low protein expression of COL14A1 in lung cancer patients. Correlation of COL14A1 alterations with the clinicopathological parameters in NSCLC patients To determine whether there were associations of COL14A1 alterations with the clinical characteristics of NSCLC patients, the DNA/RNA/protein alterations were compared with patient’s smoking status, tumor type, tumor stage and age using the Fisher’s exact test. The results showed that there are no correlations in smoking status, tumor type and age in mRNA and protein level. However, the COL14A1 promoter hypermethylation was significantly correlated with tumor late stage (P=0.017) (Table 6). The data indicated that COL14A1 promoter hypermethylation was not only associated with low mRNA transcript, but also correlated with tumor late stage, suggesting that COL14A1 hypermethylation may be involved in lung tumor progression. Chromatin status was involved in the COL14A1 expression in lung cancer cell lines. We found that the CpG island of COL14A1 gene showed looser chromatin structure in IMR90 normal cell line than did all cancer cell lines in ChIP-on-chip experiment (Figure 2b). To confirm the ChIP-on-chip data, expression of COL14A1 transcript was analyzed in IMR90 and seven lung cancer cell lines by RT-PCR. The results showed that the higher level of RNA expression was found in IMR90 than in other lung cancer cell lines (Figure 5b). Allelic loss at 3p21.3 in lung small cell lung cancer. In previous study of my laboratory, the region in 3p21 shows more than 50% of LOH in 71 microdissected samples of surgically resected 42.

(51) primary NSCLC tumors compared to their matched normal lung tissues (Tseng et al., 2005), suggesting that allelic loss in 3p21 and the inactivation of genes in this locus are involved in lung tumorigenesis. Therefore, to investigate the frequency of 3p21.3 deletion in primary lung tumors, LOH analysis was conducted using 4 microsatellite markers (D3S3604, D3S3667, D3S1568, and D3S1621) at region which nearby the RASSF1A and BLU genes (Figure 6). The data indicated that 44% and 46% of tumors were found to harbor LOH at the D3S3604-D3S3667. and. D3S1568-D3S1621. contiguous. markers,. respectively. In addition, we identified promoter hypermethylation of the. RASSF1. gene. by. DMH. method,. which. has. specific. hypermethylation in early stage patients, and plays a role in regulating cell proliferation and apoptosis (Donninger et al., 2007). Therefore, gene silencing in chromosome 3p21.3 is important for lung tumorigenesis in Taiwan. Chromatin status was involved in the RASSF1A expression in lung cancer cell lines. IMR90 normal lung cell line and three lung cancer cell lines (A549, H1299, and CL1-1) were analyzed by ChIP-on-chip. There are 19 oligonucleotide probes in the RASSF1 CpG island region, and the value of each probe was calculated by taking the ratio between ChIP DNA and total input DNA from the same cell line. ChIP was pulled down by antibody to acetylated histone H3, which recognizes open chromatin structure. The higher the ratio, the more open chromatin structure is. The RASSF1 CpG island showing less open in all cancer cell lines than IMR90 normal cell line (Figure 7a). To confirm the ChIP-on-chip data, expression of RASSF1A transcript was analyzed in IMR90 and 8 lung cancer cell lines by RT-PCR. The result showed that the higher level of RNA expression was found in IMR90 than in other lung cancer cell 43.

(52) lines (Figure 7b). Correlation of methylation spectrum and gene expression patterns in RASSF1A and BLU gene of lung cancer patients to identify the transcriptionally important CpGs Tumor suppressor genes RASSF1A and BLU gene loci are located next to each other in the region 3p21.3, and have been reported to be inactivated in a variety of human cancers by promoter hypermethylation. However, there is no data showing the transcriptionally important CpG sites for both genes. Therefore, the current study examined the association between CpG methylation and RNA expression of RASSF1A and BLU genes of 32 primary NSCLC tumor samples to identify the CpG sites that were hypermethylated and correlated with low transcriptional level. MSO microarray (Gitan et al., 2002) was used to assess methylation profiles of RASSF1A and BLU CpG island regions in lung cancer patients. A group of 20 oligonucleotide probes was designed to test 33 CpG sites of RASSF1A gene, and a group of 20 oligonucleotide probes was designed to test 38 CpG sites of BLU gene within the CpG island. Genomic maps of the RASSF1A and BLU CpG sites are shown in Figure 3a. The average of the methylation percentage in each CpG site was calculated for all patients. The CpG sites that were heavily methylated in all patients were excluded because they could not distinguish the transcriptional importance. The patients were then ranked by average of methylation percentage in the remaining non-excluded CpG sites. To establish the relationship between DNA methylation and gene expression, semiquantitative RT-PCR was conducted on 32 primary lung tumors. The results are shown in Figures 3b and 3c. For RASSF1A, the difference in mRNA expression pattern between the samples with heavily methylated CpG sites and samples with lightly methylated CpG sites was strikingly significant (t -test, 44.

(53) P=0.002). There was also a trend of significant difference (t -test, P=0.065) of BLU gene expression pattern between these heavily and lightly methylated groups. These transcriptionally important CpG sites were located in the proximal promoter and exon 1 regions for RASSF1A. For BLU, they were located in the proximal promoter and the distal region of exon 1 (Figure 3a). E2F1 can bind to the RASSF1A promoter at transcriptionally important CpGs Qiu et al. (2004) demonstrated that BLU promoter was regulated by E2F1, and this response was impaired when the promoter was hypermethylated. The present study thus compared the transcriptionally important CpG sites currently identified with the E2F1 binding site previously identified by Qiu et al. (2004). The data revealed that the CpG sites located at -63/-56 bp were indeed E2F1 binding site, thus verifying the finding of transcriptionally important CpG sites for BLU gene. In addition, three putative E2F1-binding sites were revealed at -150/-143, -5/+3 and +13/+23 bp within the RASSF1A CpG island as these were transcriptionally important CpG sites according to the MSO array results (upper panel of Figure 3a). This study therefore performed electrophoretic. mobility. shift. assay. (EMSA),. chromatin. immunoprecipitation (ChIP)-PCR, and luciferase activity assays to determine whether E2F1 can bind to RASSF1A promoter at these putative E2F1 sites. The EMSA assays were conducted using two oligonucleotide probes corresponding to three putative E2F1 binding sites. As shown in Figure 8a, incubation of nuclear extracts with probe 1 or probe 2 produced a retarded band (lane 2 and 6). The intensity increased markedly when E2F1 antibody was used (lane 3 and 7) and the addition of excess unlabeled probe1 or probe 2 completely abolished the binding of E2F1 (lane 4 and 8). This study next performed 45.