Saturday 5 May 2007 POST-GENOMIC TECHNOLOGIES IL7 09.00 – 10.00 INVESTIGATING GENOMES, GENES AND GENOME INSTABILITY Donna G. Albertson. Cancer Research Institute and Comprehensive Cancer Center University of California San Francisco, USA E-mail: [email protected] Genomic DNA copy number aberrations are frequent in solid tumors and genomic profiling of a number of different tumor types by array CGH has revealed variety in the numbers and types of copy number alterations in the tumor genomes. Copy number profiles distinguish tumor subtypes, and specific genomic alterations can be associated with other factors such as hormonal status or TP53 mutation status. Recurrent copy number aberrations, particularly amplifications, which are regions of focal high level copy number change, also focus attention on genes and the pathways in which they participate as candidates with roles in tumor development. Nevertheless, change in DNA copy number is only one way to alter expression of a gene. Other mechanisms for deregulation may be post-transcriptional, post-translational or involve alteration in expression of up stream genes, suggesting that the spectrum of alterations that one sees in fully developed tumors is a composite of selection acting on the variation that is permitted to arise by the particular failures in genome surveillance mechanism(s) present in the tumor. Here we review recent array CGH studies profiling breast and oral squamous cell carcinomas (SCC). We will focus on what we are learning about the underlying genomic instability that gives rise to particular copy number aberrations and discuss how we are taking advantage of the characteristics of oral pre-cancer and cancer genomes to take the next steps to identify genes important for tumor development. FREE PAPER SESSION 5
O30 10.00 – 10.15 MULTIPLEX AMPLICON QUANTIFICATION - MAQ: A NOVEL DIAGNOSTIC TOOL FOR HIGH THROUGHPUT ANALYSIS OF SPECIFIC COPY NUMBER POLYMORPHISMS. Dirk Goossens; Lien Heyrman; Boris Harding; Wim Glassee; Peter De Rijk; Jurgen Del-Favero. Applied Molecular Genomics Group, VIB- Department of Molecular Genetics; University of Antwerp, Belgium E-mail: [email protected] Introduction. Recent studies reporting on genome wide scans for copy number variations (CNVs) based on a whole range of methods lead to the recognition of the importance of CNVs as genetic variation. CNVs detected with these methods, need to be analysed in large sets of individuals for validation and to determine their frequency. For this large scale analysis, new technologies are needed that are simple to use, high-throughput and cheap. Here we present such a method. Methods and Materials. Multiplex Amplicon Quantification (MAQ) consists of the simultaneous PCR amplification of fluorescently labeled target and control amplicons in a one step PCR protocol. The comparison of normalized peak areas between a test individual and reference individuals results in a dosage quotient indicating the copy number of the target amplicon. The PCR primers of the MAQ assays are designed by proprietary software that allows the design of assays with a very high multiplexing degree (> 40). In this way, several genomic regions, genes or even exons of interest can be simultaneously analyzed. Together with the method, we developed a program for straigthforward data analysis, MAQs, which calculates and visualizes the dosage quotients of each amplicon starting from the chromatogram files. Results. We designed, optimized and validated an assay for the genetic diagnosis of CMT1A/HNPP consisting of 34 amplicons. Conclusions. The combination of fast assay development, quick experimental procedure and data analysis, sensitivity and robustness makes the MAQ technology highly suited for the large scale as well as high throughput analysis of any discovered or hypothesized CNV.
Abstracts of the 1st MC-GARD Meeting, 3–5 May 2007: Saturday 5 May
124
O31 10.15 – 10.30 GENOME-WIDE MAP OF MOUSE DNA COPY NUMBER VARIANTS Peggy S. Eis; Timothy Graubert; Patrick Cahan; Deepa Edwin; Rebecca Selzer; Todd Richmond; William Shannon; Xia Li; Howard McLeod; James Cheverud; Timothy Ley. Array CGH Business Unit, NimbleGen Systems, Inc., USA E-mail: [email protected] Introduction. DNA copy number variants (CNVs) are now recognized to play an important role in phenotypic variation in humans [Feuk et al. (2006) Nat. Rev. Gen. 7:85; Freeman, et al. (2006) Genome Res. 16:949] and other organisms such as mouse [Graubert et al. (2007) PLoS Genetics Jan 5, 3:e3] and the worm C. elegans [Mayden et al. (2007) Genome Res. Epub Jan 31]. Since the mouse is an important model organism for human disease, it is important to characterize both single nucleotide polymorphisms (SNPs) and CNVs. Methods and Materials. Array-based comparative genomic hybridization (array CGH) was used to map CNVs genome-wide in 21 mouse phenome database (MPD) strains. The microarray was designed using the mouse reference genome sequence for strain C57BL/6J (MM6, NCBI Build 34). A tiling-path microarray containing 385,000 long oligonucleotide probes was constructed by NimbleGen Systems (http://www.nimblegen.com). The whole-genome mouse CGH array, with a median probe spacing of 5.2 Kb, was used to map CNVs in 20 MPD strains (test samples) against reference sample C57BL/6J. Results. CNVs were detected in all strains tested and on all 19 chromosomes (X and Y were excluded from the analysis). The CNVs ranged in size from 21Kb to 2Mb in size. Of the 80 high confidence CNVs that were detected, 62 (77.5%) contain or overlap at least one gene Conclusions. Mouse CNVs significantly contribute to phenotypic variation and further characterization of this class of genome variation will increase their utility as models for human disease. While this report is currently the highest resolution map of CNVs in the 21 MPD strains studied, the detection of variants as small 21Kb (the effective resolution of the microarray used in the study) is evidence that even higher resolution experiments will be required for a comprehensive CNV map. 10.30 – 11.00 TEA AND COFFEE
EPIGENOMICS FREE PAPER SESSION 6 O32 11.00 – 11.20 GENOME WIDE PROMOTER METHYLATION ANALYSIS IN NEUROBLASTOMA WITH PERSPECTIVES FOR INTEGRATED MOLECULAR PROFILING Jasmien Hoebeeck; Maté Ongenaert; Evi Michels; Katleen De Preter; Joëlle Vermeulen; Nurten Yigit; Anne De Paepe; Wim Van Criekinge; Frank Speleman; Jo Vandesompele. Ghent University Hospital, Ghent, Belgium E-mail: {Jasmien.Hoebeeck, Franki.Speleman}@UGent. be Hypermethylation of normally unmethylated CpG islands located in gene promoter regions can be an alternative for the more classical ways of inactivation of a tumor suppressor gene (TSG), such as mutation and deletion. Methylation-specific PCR of the promoter region of 12 candidate TSGs in neuroblastoma revealed high frequencies (>30%) of methylation of six genes in neuroblastoma cell lines. In primary neuroblastoma tumors, the frequencies of methylation for these genes were PRDM2, 0%; CDH1, 7%; ZMYND10, 17%; CASP8, 54%; RASSF1A, 68% and CD44, 69%. These data, together with re-expression experiments by treatment of neuroblastoma cells with the demethylating agent 5-aza-2'-deoxycytidine (DAC), alone or in combination with histone deacetylase inhibitor, demonstrated the involvement of methylation in neuroblastoma pathogenesis. To obtain more global insights in the genes that are silenced by promoter hypermethylation in neuroblastoma, microarray expression profiling was performed to determine genes that were up-regulated after treatment with DAC using Affymetrix oligonucleotide chips. Comparison of the list of re-activated genes to a list of genes known to be methylated in other tumor types lead to the identification of 12 potentially methylated TSGs located in frequently deleted regions in neuroblastoma, including FABP3 (1p32-33), IGSF4 (11q23.2) and ESR1 (6q25.1). Additionally, a selection of 100 differentially expressed genes was made based on promoter similarity with known methylation marker genes. This gene list contains promising candidate genes that merit further investigation for their involvement in neuroblastoma.
Abstracts of the 1st MC-GARD Meeting, 3–5 May 2007: Saturday 5 May
125
Using a high-throughput nanobiotechnological method adapted for methylation specific PCR, these genes are currently tested on a large panel of well-documented primary neuroblastoma tumors belonging to the different neuroblastoma genetic subtypes. In this way, we aim to establish a methylation signature for improved diagnostic and prognostic classification of neuroblastoma tumors. Integration of genome wide methylation data with other molecular profiling platforms (mRNA, miRNA, BAC arrays) will be discussed. O33 11.20 – 11.35 GENETIC AND EPIGENTIC PROFILING OF OVARIAN SEROUS BORDERLINE TUMORS AND CARCINOMAS Ingrid T. G. W. Bijsmans; Debbie Robbrecht; Manon van Engeland; Nathalie Sieben. Cancer Biology Lab, Research Institute Growth and Development (GROW), Department of Pathology, Maastricht University, Maastricht, The Netherlands E-mail: [email protected] Serous ovarian tumors can be divided into borderline tumors (SBTs, low-malignant potential, 10 year survival > 90%) and carcinomas (SCA, average 10 year survival of 30%). There is ongoing debate whether SBTs are precursors of SCAs or represent separate entities with different biology. So far, only BRAF and KRAS mutation analysis and expression profiling have been used to address this issue. In this study, we performed mutation analysis for BRAF, KRAS and TP53, as well as methylation profiling of 7 tumor suppressor genes in a panel of 15 SBTs and 25 SCAs. BRAF mutations were detected in 4/15 (27%) of SBTs and absent in SCAs (p=0.02). 5/15 SBTs (33%) and 5/25 (20%) SCAs showed mutations in KRAS (not significant). Mutations in TP53 were not detected in SBTs and observed in 7/25 SCAs (28%) (p=0.03). Promoter hypermethylation of the genes p14ARF, p16INK4A, BRCA1 and hMLH1 was detected more frequently in SBTs (7/15 (47%), 5/15 (33%), 8/15 (53%), and 4/14 (29%) respectively) compared to SCAs (5/23,(22%), 4/24 (17%), 5/25 (20%), and 4/25 (16%) respectively). BRCA1 methylation was significantly higher in SBTs compared to SCAs (p=0.03). The genes RASSF1A, APC and SFRP5 showed a lower methylation frequency in SBTs (6/15 (40%), 2/15 (13%; p=0.05) and 0/15 (0%;p=0.05) respectively) compared to SCAs (17/24 (71%), APC 11/25 (44%) and 6/25 (24%) respectively). These data show significant genetic (BRAF and TP53) as well as
epigenetic (BRCA1, APC and SFRP5) differences between SBTs and SCAs. O34 11.35 – 11.50 EPIGENETIC STATUS OF THE G1/S CHECKPOINT GENES IN MISCARRIAGES WITH CHROMOSOMAL MOSAICISM Anna A. Kashevarova; Ekaterina N. Tolmacheva; Igor N. Lebedev. Institute of Medical Genetics, Russian Academy of Medical Sciences, Russia E-mail: [email protected] Introduction. Structural integrity and functioning of genome is under epigenetic regulation. Fail of this control may lead to suppression of genetic regulatory systems and genomic instability. A high frequency of chromosomal abnormalities is observed during early stages of embryo development. Disturbance of epigenetic regulation may occur in early embryogenesis when intensive process of epigenetic reprogramming takes place. So, the origin of chromosomal instability may be partially explained by epigenetic inactivation of the cell-cycle checkpoints. The aim of our study was to investigate the epigenetic status of P14ARF and RB1 genes involved in P53 and RB1 pathways of G1/S transition in miscarriages with chromosomal mosaicism. Methods and Materials. Cytotrophoblast and extraembryonic mesoderm of 73 aneuploid miscarriages were analyzed by interphase FISH. Three groups of abortions with different level of mosaicism were distinguished. The methylation status of P14ARF and RB1 was investigated by MSP in both tissues. Twenty induced abortions were studied as a control group. Results. About half of the miscarriages had the level of aneuploid cells more than 70%. It seems, that origin of mosaicism in the rest groups of embryos with low and middle level of aneuploidy is related with mitotic nondisjunction in somatic cells rather than trisomy zygote rescue mechanism. For the first time hypermethylation of P14ARF and RB1 was revealed in mosaic miscarriages. The frequency of epimutations in cytotrophoblast and extraembryonic mesoderm was 6% and 5.3% (P14ARF) and 11.9% and 27.8% (RB1), respectively. No hypermethylation of both genes was observed in the control group. Significantly, all except one miscarriage with epimutations belong to the groups with low and middle level of mosaicism. Conclusions. Our data provide evidence for the association between epigenetic inactivation of the G1/S
Abstracts of the 1st MC-GARD Meeting, 3–5 May 2007: Saturday 5 May
126
checkpoint genes and induction of chromosomal nondisjunction in somatic cells during early embryogenesis. This study was supported by RFBR (¹ 05-04-48129). O35 11.50 – 12.05 LUMA (LUminometric Methylation Assay) - A HIGH THROUGHPUT METHOD TO THE ANALYSIS OF GENOMIC DNA METHYLATION Mohsen Karimi Arzenani; Sofia Johansson. Center for Molecular Medicine, Karolinska Institute, Sweden E-mail: [email protected] Introduction. Changes in genomic DNA methylation are important events in normal and pathological cellular processes, contributing both to normal development and differentiation as well as to cancer and other diseases. Therefore, there is a need for high through put and quantitative methods to analyse global DNA methylation both to evaluate how DNA methylation is involved in cancer development as well as prognosis of patients. We have developed a new method, so-called LUMA, to measure genome wide DNA methylation in different biological samples. Methods and Materials. The method is based on combined DNA cleavage by methylation-sensitive restriction enzymes and polymerase extension assay by Pyrosequencing™. HpaII is used as methylation sensitive enzyme and MspI is used as methylation insensitive enzyme. DNA methylation level is defined as the ratio of HpaII/MspI. It is performed in 96 well plates and is easy to scale up. Results. The method is quantitative, highly reproducible and easy to scale up. Since no primary modification of genomic DNA, such as bisulfite treatment, is needed, the total assay time is only six hours. In addition, the assay requires only 200-500 ng of genomic DNA, and incorporates an internal control to eliminate the problem of varying amounts of starting DNA. The accuracy and linearity of LUMA was verified by in vitro methylated lambda DNA. In addition, DNA methylation levels were assessed by LUMA in DNA methyltransferase knock-out cell lines and after treatment with the DNA methyltransferase inhibitor (5-AzaCytidine). Conclusions. The LUMA assay may provide a useful method to analyze genome-wide DNA methylation for a variety of physiological and pathological conditions
including etiologic, diagnostic and prognostic aspects of cancer. O36 12.05 – 12.20 MICROARRAY-BASED SCREEN FOR THE IDENTIFICATION OF DIFFERENTIAL FETO-MATERNAL DNA METHYLATION MARKERS AND DEVELOPMENT FOR NIPD OF CHROMOSOME DISORDERS Philippos C. Patsallis; Eliza Papageorgiou; Heike Fiegler; Nigel P. Carter. The Wellcome Trust Sanger Institute, Cambridge, UK E-mail: [email protected] Introduction. The discovery of free fetal DNA in maternal plasma has given the potential for future development of noninvasive prenatal diagnosis (NIPD). We have been using DNA microarray technology to screen for sequences that are differentially methylated between maternal blood and fetal DNA (placenta) to find candidate regions for NIPD assays. Methods and Materials. Methylation DNA Immunoprecipitation (MeDIP) was applied to DNA samples derived from female whole blood and placenta (MeDIP permits highly efficient enrichment of methylated DNA). Hybridisation of the MeDIP enriched fractions to a whole genome tilling path array (mainly constructed by BAC clones) identified large regions of methylation differences between the two tissues by with relatively poor resolution. We then hybridised the same material to higher resolution oligo arrays specific for the chromosomes that have been associated with common euploidies (chr13, 18, 21, X, Y). The probes used in the oligo arrays are 50-60bp long with a median probe spacing of 225bp for chr13, 170bp for chr18, 70bp for chr21, 340bp for chrX and 20bp for chrY. Results. The results from the high resolution oligo arrays have shown small regions of several consecutive oligo array probes to be differentially methylated between whole blood and placenta in all five chromosomes. A detailed analysis of the results obtained from chr18 and chr21 has revealed 42 regions on chr18 and 34 regions on chr21 that are differentially methylated between whole blood and placenta. The Maspin promoter region (located on chr18) was found to be hypermethylated in whole blood compared to placenta as was expected based on the study of Chim et.al. (2004). The methylation status of the maspin promoter region and the methylation status of two additional regions on chr21 where
Abstracts of the 1st MC-GARD Meeting, 3–5 May 2007: Saturday 5 May
127
confirmed by real-time quantitative PCR. Conclusions. Although the results from the high resolution oligo arrays and the real time quantitative PCR look promissing, additional experiments are being performed in order to investigate the degree of MeDiP variability as well as the degree of methylation variability between individual samples. O37 12.20 – 12.35 STROMAL EXPRESSION OF HIF-2A CORRELATES WITH BNIP3 PROMOTER HYPERMETHYLATION AND OVERALL PATIENT SURVIVAL IN COLORECTAL CARCINOMAS Arjen Cleven; Angela Spiertz; Bradley Wouters; Manon van Engeland; Adriaan de Bruïne. University Hospital Maastricht, The Netherlands E-mail: [email protected] Background. Expression of hypoxia-inducible factors (HIF's) in tumor-associated stromal cells is associated with poor outcome in colorectal cancer (CRC). HIF's regulate genes involved in cell survival, and apoptosis. BCL-2/adenovirus E1B-19kDa-interacting protein 3 (BNIP3) is a cell death inducing protein, in most cancers it is upregulated via HIF's in response to hypoxia. BNIP3 plays a functional role in hypoxic survival of tumor cells in CRC. We hypothesized that epigenetic silencing of BNIP3 by promoter methylation could play an important role in the adaptive response of a tumor to hypoxia. Therefore we studied the role of protein expression of hypoxia-inducible factor (HIF)-1a, HIF-2a in relation with BNIP3 promoter hypermethylation in patients with CRC's. Methods. Expression of HIF-1a, and HIF-2a, was analysed by immunohistochemistry in 168 colorectal adenocarcinomas. In the same study population promoter hypermethylation of BNIP3 was determined by using methylation specific PCR (MSP). The expression of HIF-1a, HIF-2a, and BNIP3 promoter hypermethylation was correlated with clinicopathological variables and overall patient survival. Results. Expression of these hypoxia markers was detected in the epithelial compartment of the tumor cells as well as in tumor-associated stromal cells. Only stromal expression of HIF-2a showed a correlation with BNIP3 hypermethylation (p=0.038). Overall we found 59% (n=99) positive and 41% (n=69) negative for BNIP3 promoter hypermethylation. In those cases with BNIP3 hypermethylation there was an overall survival difference of 34% between HIF-2a positive and negative cases, with
HIF-2a positive cases having the worst overall survival. This previous trend in survival did not exist in BNIP3 hypermethylation negative cases (p=0.004). Conclusion. These observations show that the impact of stromal HIF-2a expression on patient survival in CRC is associated with BNIP3 promoter hypermethylation. This supports our hypothesis that epigenetic silencing of the hypoxia induced proapoptotic gene BNIP3 is associated with a survival advantage of stromal cells under hypoxic conditions.
