UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Identification of novel genetic alterations in oral squamous cell carcinoma Baldwin, Corisande Saskia Melody 2005

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-ubc_2005-0366.pdf [ 22.25MB ]
Metadata
JSON: 831-1.0099799.json
JSON-LD: 831-1.0099799-ld.json
RDF/XML (Pretty): 831-1.0099799-rdf.xml
RDF/JSON: 831-1.0099799-rdf.json
Turtle: 831-1.0099799-turtle.txt
N-Triples: 831-1.0099799-rdf-ntriples.txt
Original Record: 831-1.0099799-source.json
Full Text
831-1.0099799-fulltext.txt
Citation
831-1.0099799.ris

Full Text

IDENTIFICATION OF NOVEL GENETIC ALTERATIONS IN ORAL SQUAMOUS CELL CARCINOMA by CORISANDE SASKIA MELODY BALDWIN B.Sc. University of British Columbia, 2002 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES Pathology and Laboratory Medicine THE UNIVERSITY OF BRITISH COLUMBIA June 2005 © Corisande Saskia Melody Baldwin, 2005 ABSTRACT Oral squamous cell carcinoma (OSCC) is the most common head and neck neoplasm, affecting approximately 400,000 people annually worldwide. Most cases are not diagnosed until the advanced stages of the disease resulting in a five-year survival rate of 50%. Application of high resolution genomic analysis techniques for the detection of novel molecular markers and targets will greatly benefit the prevention and management of this disease. Array comparative genomic hybridization (aCGH) enables the detection of segmental gains and losses of DNA. We constructed a 3p-arm specific array comprised of 535 near-overlapping BAC clones for the identification of minimal regions of gain and loss on this chromosome arm. Application of this array to 19 OSCC specimens enabled the identification and characterization of five minimal regions of alteration including 4 regions of loss, and, interestingly, 1 region of gain. 3p loss is a common event in OSCC; however, segmental gains on this arm have not been previously described. Further construction of a whole genome sub-megabase resolution tiling set (SMRT) array comprised of 32,433 over-lapping BAC clones spanning the entire human genome made possible the analysis of the genome of an additional set of 20 OSCC specimens at unprecedented resolution. Comparison of these OSCC genomes allowed the identification of well-known alterations as well as novel minimal regions including gains at 3q23, 5pl5.2, 7pl2.3-13, 7q21.2, and 7q35, and losses at 2pl5, 4q34.3, and 16q23.2. Most of these novel alterations are sub-megabase in size, suggesting that they may have been missed by conventional, lower ii resolution techniques. A selection of the novel gains was confirmed through reverse transcriptase PCR expression analysis of genes within those regions such as TRIO at 5pl5.2, TEM6 at 7pl2.3-13, and CDK6 at 7q21.2, all of which showed elevated expression in tumours compared to normal oral epithelial cells. This study represents the first application of tiling resolution aCGH technology for the detailed analysis of clinical specimens and demonstrates the invaluable power of tiling resolution aCGH technology for the analysis of OSCC genomes. Here, we demonstrate that novel sub-megabase minimal regions of alteration are recurrent events in OSCC genomes. The identification of genes previously implicated in cancer or with functional roles in the cell-cycle and/or signal transduction within these minimal sub-megabase alterations suggests the importance of these alterations in OSCC. Further assessment of these genes and their respective pathways, both at the expression and functional level, may lead to the development of novel drug therapies targeting these gene products. iii TABLE OF CONTENTS ABSTRACT...... ii TABLE OF CONTENTS iv LIST OF TABLES vii LIST OF FIGURES viii ABBREVIATIONS x ACKNOWLEDGEMENTS xi CHAPTER 1: INTRODUCTION 1 1.1 Oral Cancer 1 1.2 Histological Progression 3 1.3 Genetic Progression of Oral Cancer 5 1.4 Utility of Molecular Diagnosis in Disease Management 8 1.5 Conventional methods of detecting genetic alterations in oral cancer 9 1.6 Array CGH Technology 14 1.7 Whole Genome Tiling Resolution BAC Array CGH 17 1.8 Hypotheses & Objectives 19 CHAPTER 2: MATERIALS AND METHODS 21 2.1 Specimen Accrual, Microdissection & DNA Extraction 21 2.2 3p Array CGH 22 2.2.1 3p Array Construction 22 2.2.2 3p Array Probe Labelling 25 2.2.3 3p Array Hybridization 26 2.2.4 3p Array Imaging and Normalization 26 2.3 Whole Genome SMRT Array CGH 27 2.3.1 Whole Genome SMRT Array Construction 27 2.3.2 Whole Genome SMRT Array Probe Labelling 29 2.3.3 Whole Genome SMRT Array Hybridization 30 2.3.4 Whole Genome SMR T A rray Imaging and Normalization 31 2.4 Identification of alterations and alignment of BAC array profiles 31 2.4.1 Identification of alterations using the 3p-array 31 2.4.2 Identification of alterations using the SMRT array 32 2.5 Expression Analysis Confirmation 33 2.5.1 Frozen Tissue Accrual, Microdissection & RNA Extraction 33 2.5.2 Reverse Transcriptase PCR 33 CHAPTER 3: RESULTS & DISCUSSION.. 38 3.1 Summary of 3p Array Results 38 3.1.1 Validation of the Regional Array 38 3.1.2 Detection of Genetic Alteration on Chromosome 3p 40 3.1.3 Fine-mapping Genetic A Iterations in 3p 41 3.2 Summary of Whole Genome SMRT Array Results 46 3.2.1 Validation of the Whole Genome SMRT Array 47 iv 3.2.2 Detection of Genetic Alterations Using the SMRT Array 50 3.2.3 Identification of "Hot Spots " Using Frequency Plot Analysis 53 3.2.4 Array CGH representation of aneuploidy 59 3.2.5 Characterization of Known Minimal Regions of Alterations 60 3.2.6 Detection of putative polymorphisms 68 3.2.7 Identification of Novel Regions ofAlteration 70 3.2.8 "Masking" of single copy number changes due to tissue heterogeneity 91 3.2.9 Comparison of tumour and dysplasia data 92 3.3 Expression Analysis Confirmation of Candidate Genes 93 3.4 Significance 101 3.5 Future Directions 102 CHAPTER 4: CONCLUSIONS 106 REFERENCES 108 APPENDIX 1: Clinical Information of Specimens Analyzed Using the 3p-Array 129 APPENDIX 2: 3p Array Clone Set 130 APPENDIX 3: Clinical Information of Specimens Analyzed Using the Whole Genome SMRT Array 142 APPENDIX 4: Supplemental Karyogram of Tumour 24T 143 APPENDIX 5: Supplemental Karyogram of Tumour 117T 144 APPENDIX 6: Supplemental Karyogram of Tumour 123T 145 APPENDIX 7: Supplemental Karyogram of Matched Normal 123C 146 APPENDIX 8: Supplemental Karyogram of Tumour 125T 147 APPENDIX 9: Supplemental Karyogram of Tumour 161T 148 APPENDIX 10: Supplemental Karyogram of Matched Normal 161C 149 APPENDIX 11: Supplemental Karyogram of Tumour 486T 150 APPENDIX 12: Supplemental Karyogram of Matched Normal 486C 151 APPENDIX 13: Supplemental Karyogram of Tumour 528T 152 APPENDIX 14: Supplemental Karyogram of Tumour 542T... 153 APPENDIX 15: Supplemental Karyogram of Tumour 628T 154 APPENDIX 16: Supplemental Karyogram of Tumour 669T 155 APPENDIX 17: Supplemental Karyogram of Tumour 792T. 156 A P P E N D I X 18: Supplemental Karyogram of Tumour 793T 157 A P P E N D I X 19: Supplemental Karyogram of Tumour 794T . 158 A P P E N D I X 20: Supplemental Karyogram of Tumour 800T 159 A P P E N D I X 21: Supplemental Karyogram of Tumour 801T 160 A P P E N D I X 22: Supplemental Karyogram of Tumour 805T .....161 A P P E N D I X 23: Supplemental Karyogram of Tumour 809T 162 A P P E N D I X 24: Supplemental Karyogram of Tumour 814T 163 A P P E N D I X 25: Supplemental Karyogram of Tumour 819T 164 A P P E N D I X 26: Supplemental Karyogram of Tumour 819C 165 A P P E N D I X 27: Supplemental Karyogram of Tumour 836T 166 vi LIST OF TABLES Table 1: Five minimal regions of alteration identified using the 3p-arm specific array. .44 Table 2: Known minimal regions of alteration identified in 20 OSCC using whole genome array CGH 60 Table 3: Summary of novel minimal regions of alteration detected in 20 OSCC using whole genome array C G H 71 v i i LIST OF FIGURES Figure 1: Lesions of the Oral Cavity 2 Figure 2: Histological progression of oral squamous cell carcinoma (OSCC) 4 Figure 3: Genetic progression model presented by Califano et al. 7 Figure 4: Conventional cytogenetic techniques used to study oral tumour genomes 11 Figure 5: Loss of heterozygosity (LOH) analysis using micro satellite markers 13 Figure 6: Principle of array comparative genomic hybridization 15 Figure 7: Complete coverage whole genome array comparative genomic hybridization. 18 Figure 8: Detection of single copy changes on chromosomes X and Y with the regional array 39 Figure 9: Representative examples of three 3p-array profiles 42 Figure 10: Alignment of 19 oral squamous cell carcinoma (OSCC) array comparative genomic hybridization (aCGH) profiles for chromosome arm 3p 43 Figure 11A,B: Use of SeeGH software to analyze SMRT array data 48 Figure 11C,D: Use of SeeGH software to analyze SMRT array data cont 489 Figure 12: Validation of SMRT array by multiple alignment and comparison of H526 profiles in SeeGH 51 Figure 13: Comparison of 3p profiles generated with 3p-arm specific array and the SMRT array 52 Figure 14: Representative whole genome comparative genomic hybridization (CGH) profile of an OSCC tumour (125T) 54 Figure 15: Frequency plot of all copy number gains and losses present in the 20 oral samples analysed in this study 55 Figure 16: Alignment and comparison of the "raw" or unmanipulated data from 528T with the data manipulated in aCGH-smooth 57 Figure 17: Identification and characterization Of a known gain at 7pl 1.2. 61 Figure 18: Use of the UCSC Human Genome Browser to identify genes within regions. 63 Figure 19: Identification and characterization of a known gain at 1 lql3.3 64 Figure 20: Multiple alignment 5 cases showing a minimal loss at 8p23.2 containing CSMD1...... 66 Figure 21: Detection of polymorphisms using the SMRT array... 69 Figure 22: Alignment of novel alteration at 3q23 among tumours and a matched normal. 73 Figure 23: Alignment of novel alteration at 5pl5.2 among tumours and a matched normal 75 Figure 24: Alignment of novel alteration at 7pl2.3-13 among tumours and a matched normal 77 Figure 25: Alignment of novel alteration at 7q21.2 among tumours and a matched normal 79 Figure 26: Alignment of novel alteration at 7q35 among tumours and a matched normal. 81 viii Figure 27: Alignment of novel high copy number at 1 lq22.2-22.3 containing matrix metalloproteinase cluster 83 Figure 28: Alignment of novel recurring minimal region loss at 2pl5 among 3 tumours. 87 Figure 29: Alignment of novel recurring minimal region loss at 4q34.3 among 3 tumours 89 Figure 30: Alignment of novel recurring minimal region loss at 16q23.2 among 3 tumours 90 Figure 31: Comparison of TRIO expression between oral tumours and normal oral specimens 95 Figure 32: Comparison of TENS 1 expression between oral tumours and normal oral specimens 98 Figure 33: Comparison of CDK6 expression between oral tumours and normal oral specimens 99 Figure 34: Assessment of mystery band present in TENS1 and CDK6 PCR product... 100 ix ABBREVIATIONS aCGH: array comparative genomic hybridization AFP: amplified fragment pool BAC: bacterial artificial chromosome BLAST: basic local alignment search tool cDNA: complementary DNA CGH: comparative genomic hybridization CIS: carcinoma in situ CTD: CaltechDl/D2 FISH: fluorescent in situ hybridization FPC: fingerprinted contigs LM-PCR: linker-mediated polymerase chain reaction LOH: loss of heterozygosity Mbp: megabase pairs mCGH: metaphase comparative genomic hybridization MMP: matrix metalloprotemase MRA: minimal region of alteration NCBI: National Center for Biotechnology Information OPL: oral premalignant lesion OSCC: oral squamous cell carcinoma PAC: PI-derived artificial clone PCR: polymerase chain reaction RPCI: Roswe 11 Park Cancer Institute rtPCR: reverse transcriptase polymerase chain reaction SKY: spectral karyotyping SMRT: sub-megabase resolution tiling set SNP: single nucleotide polymorphism UCSC: University of California Santa Cruz x ACKNOWLEDGEMENTS I would like to extend my gratitude to my supervisor Dr. Wan Lam for his dedication to his students and for his support, guidance and example in helping each of us to achieve our goals. In addition I would to thank the following people for their contributions to this thesis: The array C G H group (Adrian Ishkanian, Chad Malloff, Spencer Watson, Ron deLeeuw, and Miwa Suzuki); Bryan Chi and Philip Wong for their contributions to the development of array C G H technology in our lab; fellow graduate student and mentor Cathie Garnis for her continued support and guidance; and fellow graduate students Bradley Coe, Jonathon Davies, Ashleen Shadeo, William Lockwood, Timon Buys, Laura-Jane Henderson, and Eric Lee for their assistance and valuable discussions. M y gratitude also goes to Hisae Nakamura for microdissection and Michelle Morris for management of specimen material. M y appreciation also extends to my supervisory committee, Dr. Miriam Rosin, Dr. Lewei Zhang, and Dr. Hayden Pritchard for guidance and helpful suggestions. x i C H A P T E R 1: I N T R O D U C T I O N 1.1 Oral Cancer Oral cancer is the most common head and neck neoplasm, affecting ~400,000 people worldwide each year (http://www3.who.int/whosis/discussion_papers/pdf/ paperl3.pdf). Oral squamous cell carcinoma (OSCC) represents over 90% of cases (Bockmuhl, et al. 1998). OSCC is distinctive from other head and neck cancers based on its clinical appearance and outcome (Mao, et al. 2004). The majority of lesions occur on the floor of the mouth or on the tongue (Vokes, et al. 1993). Oral pre-malignant lesions (OPL) often appear as white or red plaques on the oral mucosa which are referred to as a leukoplakia and erythroplakia, respectively (Figure IA). While leukoplakias are visible due to the thickening of the epithelium, erythroplakia develop as a result of atrophy of the epithelium (personal communication, Dr. Lewei Zhang). However, only 36% of these premalignant lesions actually progress to full invasive OSCC (Figure IB) (Papadimitrakopoulou, et al. 1997). In the past, pathologists have relied on histological analysis of biopsies to determine the risk of progression and the course of treatment. Biopsies exhibiting the abnormal cellular morphology and tissue architecture, referred to as regions of "dysplasia," were considered at high risk for progression (Tabor, et al. 2003). However, only visible lesions such leukoplakias and erythroplakias are actually biopsied, while dysplasias invisible to the eye are neglected and the process of taking a biopsy, alone, 1 Figure 1: Lesions of the Oral Cavity. (A) Oral leukoplakia, an O P L characterized by a white plaque, shown here on the lower gum surface. Approximately 1/3 o f oral leukoplakias w i l l progress to full invasive squamous cell carcinoma. (B) Invasive oral squamous cell carcinoma, shown here on the roof of the mouth. (Provided by Dr. Lewei Zhang) 2 often induces histological changes in the lesions resembling those typically seen in dysplasia, making histological risk assessment difficult (Rosin, et al. 2002). In addition, surgical resection of these premalignancies has not been proven to adequately prevent progression to full invasive carcinoma (Zhang, et al. 2001) and, thus, lesions are left untreated as physicians "watch and wait." Additionally, 15% of leukoplakias bearing no dysplastic features, progress nonetheless (Silverman, et al. 1984). As a result, this disease is usually diagnosed at advanced stages, resulting in a poor prognosis (5-year survival rate -50%) (Vokes, et al. 1993). While OSCC is associated with tobacco and alcohol consumption, (Mao 1997) the biological mechanisms underlying the disease are poorly understood (Mao, et al. 2004). 1.2 Histological Progression It is generally accepted that oral cancer progresses through a series of well-defined histological changes, each of which is associated with a clinical stage (Figure 2). Specifically, progression occurs from normal epithelium, to precursor lesions (such as benign hyperplasia, leukoplakias, and erythroplakia), varying degrees of dysplasia, carcinoma in situ (CIS), and, finally, invasive carcinoma (Pindborg 1997). Benign hyperplastic lesions are characterized by over-proliferation of cells exhibiting normal cell morphology. Dysplastic lesions are graded by pathologists as mild, moderate, or severe. Histological characteristics that are used to diagnose dysplasias include 1) loss of basal cell polarity, 2) multiple layers of basal cells, 3) increased nuclear to cytoplasmic ratio, 4) nuclear hyperchromatism, 5) enlarged nucleoli, 6) increase in the number of mitotic cells, 3 Normal Epithelium Dysplasia Carcinoma In Situ Invasive (CIS) Carcinoma Figure 2: Histological progression of ora l squamous cell carcinoma (OSCC). Normal epithelium are driven first into various stages of dysplasia (mild, moderate, severe), then into carcinoma in situ (CIS), and, finally into invasive carcinoma. This histopathological progression is paralleled by a gradual accumulation o f genetic alterations. Following progression to invasive carcinoma, tumour cells are free to migrate to distant sites, a process called "metastasis." (Provided by Dr. Mir iam Rosin) 4 7) presence of abnormal mitotic form, 8) cellular and nuclear pleomorphism, 9) irregular epithelial stratification, 10) loss of intercellular adherence, and 11) keratinization of individual or groups of cells (Pindborg 1997; Warnakulasuriya 2001). Diagnosis of dysplasia is variable and subjective (Bosman 2001; Tabor, et al. 2003; Warnakulasuriya 2001) and depends both on the region of the biopsy that was examined and on the criteria used by the pathologist (Warnakulasuriya 2001). Diagnosis of lesion severity is dependent on the proportion of epithelium that exhibits morphological changes. Since the thickness of the epithelium can vary, this contributes to the subjectivity of histopathological classification (Warnakulasuriya 2001). While mild dysplasias exhibit observable changes confined to the lower third layers, moderate dysplasias show morphological changes in cells present in the lower third extending into the middle layers, and severe dysplasias show changes in the lower and middle layers extending into some but not all of the upper layers of epithelium. When these morphological changes are present in all three layers, the lesion is diagnosed as carcinoma in situ, or CIS (Warnakulasuriya 2001). Finally, when the malignant cells penetrate through the basement membrane into the tissue below the lesion is considered invasive carcinoma (Warnakulasuriya 2001). 1.3 Genetic Progression of Oral Cancer Progression from the pre-malignant to the malignant stage is paralleled by the accumulation of genetic alterations (Califano, et al. 1996; Fearon and Vogelstein 1990; Rosin, et al. 2000; Sciubba 2001; Zhang and Rosin 2001). As alterations are accrued, the genomic stability declines, and this results in further genetic alterations. The challenge, 5 therefore, is to determine which alterations are driving the disease and the order in which they occur in the progressing malignant cell. Califano et al (1996) developed the current genetic model of oral cancer progression (Figure 3). The model is based on a statistical analysis of the loss of heterozygosity (LOH) data from ten critical loci across 87 oral lesions of varying disease stage, ranging from premalignant lesions to full invasive carcinomas. The loci were picked based on previous LOH data that showed > 40% minimal regions of loss. The regions included 3p21 (FHIT locus), 9p21 (pi 6), 4q26-28, 6p, 8, 1 lql3 (cyclin DI locus), 13q21, 14q31-32.1, and 17pl3 (believed to be the second p53 locus). Based on their statistical analysis, they deduced that loss of chromosome 9p is the earliest alteration, driving the normal epithelium into both benign hyperplasia and OPL. This is followed by loss of chromosomes 3p and 17p as the cells progress to dysplasia. Subsequent loss of 1 lq, 13q, and 14q drives dysplasias to the CIS stage and, finally, loss of 6p, 8, and 4q results in the progression from CIS to invasive OSCC. LOH at microsatellite markers on 3p and 9p has been shown to correlate with increased risk (Mao, et al. 1996b; Rosin, et al. 2000; Rosin, et al. 2002). The theory of "field cancerization" was first developed in 1953 by Slaughter et al to explain occurrence of second primary tumours following tumour resection (Slaughter, et al. 1953). The theory postulates that "fields" of tissue chronically exposed to harsh carcinogens incur genetic damage throughout and, therefore, all cells throughout the "field" have the potential to progress to cancer if the necessary changes are induced (Partridge, et al. 2000; Slaughter, et al. 1953). Conversely, it has been theorized that 6 Norma - 11q, 13q, 14q lyperplasia Carcinoma In Situ Invasive Carcinoma 3p, 17p Figure 3: Genetic progression model presented by Califano et al. The model was proposed based on a statistical analysis o f L O H data o f 10 micro satellite markers known to be altered in oral cancer from 87 lesions o f various stages of disease progression ranging from premalignant lesions to full invasive squamous cell carcinoma (Califano et al. 1996). 7 development of a second tumour following resection is caused by residual tumour cells and that both malignancies are clonally related. This has been referred to as "minimal residual disease" (Tabor, et al. 2003). Indeed, Califano et al (1996) show that separate lesions show the same patterns of LOH with identical boundaries, supporting the theory of clonal recurrence. However, recent studies using LOH analysis have shown that both theories are correct (Jiang, et al. 2001). Jang et al (2001) performed LOH analysis of 100 multiple cancerous and precancerous lesions from 26 patients. Assessment of seven well-known markers and analysis for patterns of microsatellite alterations revealed that, while only 14% of the non-invasive lesions from individual patients showed LOH patterns indicative of a clonal relationship between the lesions, 43% of the invasive lesions from individual patients showed LOH patterns indicative of a clonal relationship (Jiang, et al. 2001). While the invasive lesions that show LOH patterns suggesting a clonal relationship would support the "residual disease" theory, the remaining non-invasive cases showing no common microsatellite patterns of alteration support the theory of "field cancerization." 1.4 Utility of Molecular Diagnosis in Disease Management Until recently, prognosis and treatment decisions for oral lesions were dependent on the disease staging as assessed by a pathologist (Tabor, et al. 2003). Generally, prognosis worsened with increasing disease stage. The criterion used in histopathological diagnosis is based on the morphological features of the cells (as described above). While this has been somewhat useful, it has also been shown to be a subjective approach (Warnakulasuriya 2001). Discrepancy between histopathological stage diagnoses for a 8 given lesion among pathologists has been reported (Abbey, et al. 1995; Warnakulasuriya 2001). The process of obtaining a biopsy alone, can induce changes in normal epithelium resembling dysplasia (Rosin, et al. 2002). Indeed, the correlation between histopathological stage and disease outcome is not 100% (Abbey, et al. 1995; Karabulut, et al. 1995; Warnakulasuriya 2001). Compatible with this, most premalignant lesions showing dysplasia will not progress to fully invasive OSCC (Bouquot 1997). This demonstrates the need for an objective approach to disease stage diagnosis. Use of a molecular screening approach to detect alterations necessary for progression would provide an objective and comprehensive approach to stage diagnosis, as opposed to the more subjective histopathological methodologies (Rosin, et al. 2000). LOH analysis has been highly successful in identifying genetic regions associated with risk (Mao, et al. 1996a; Partridge, et al. 1999; Rosin, et al. 2000). In particular, LOH at chromosomes 3pl4 and 9p21 has been shown to be indicative of progression risk (Rosin, et al. 2000; Rosin, et al. 2002; Zhang and Rosin 2001). However, while this offers useful prognostic screening, it does not address the need to identify and characterize genetic alterations driving disease behaviour. The identification of novel molecular markers critical to oral cancer progression will greatly benefit the prevention and management of this disease (Rosin, et al. 2000). 1.5 Conventional methods of detecting genetic alterations in oral cam Conventional tools used to identify critical genetic alterations include cytogenetic analysis (such as G-banding, spectral karyograms (SKY), fluorescence in situ 9 hybridization (FISH), and conventional metaphase comparative genomic hybridization (mCGH)) (Figure 4) and loss of heterozygosity (LOH) analysis (Figure 5). While these techniques have been highly successful in identifying key regions, they have limitations. Cytogenetic techniques such as G-banding and spectral karyotype (SKY) analysis were some of the first techniques used to study tumour genomes. These techniques have been useful in identifying gains, losses, and other large chromosomal rearrangements. However, because these techniques involve the direct examination of tumour metaphase chromosomes under the light microscope, their resolution is limited (Garnis, et al. 2004a). FISH has enabled the fine mapping of breakpoints involved in large chromosomal aberrations (Garnis, et al. 2004a). Similarly, conventional metaphase comparative genomic hybridization (mCGH), which was developed by Kallioniemi et al (Kallioniemi, et al. 1992), has also been useful in identifying segmental gains and losses in tumour genomes including oral tumours (Albertson and Pinkel 2003). The procedure involves differentially labelling tumour and normal diploid DNA with cyanine dyes, co-hybridizing them onto a slide of normal metaphase chromosomes, and observing the ratios of the dyes at each loci along the chromosomes under a light microscope (Figure 4B). While this technique has been successful in identifying key regions, its resolution is limited to 20 Mbp (Pinkel, et al. 1998). 10 t i II 2 f 1 11 F 4 t 6 • 7 II If lie: | 9 10 M I • 1 2 ttt ~4 ft* 14 15 IS ft* fft IS 17 18 9 8 ' : : 13 * 20 • 21 1 22 t | - M [ J | I • ! Y 'Ti.* * Ml Figure 4: Conventional cytogenetic techniques used to study oral tumour genomes. ( A ) Spectral Karyotype ( S K Y ) (http://www.polytec-pi .fr/spectral-imaging2509/inner/EwingSKYg-Band.html) that employs probes specific for each chromosome with separate dye tags and detects translocations, truncations, and duplications o f chromosomes. (B) Metaphase (http://www3.mdanderson.org/depts/ pathology/fish/images/alkmeta.jpg) and (C) interphase FISH (http://www3.mdanderson.Org/depts./pathology/fish/fish/cyclin.html) utilize a single probe specific to one unique region tagged with a fluorescent dye to detect deletions and amplifications of genomic regions. (D) Metaphase comparative genomic hybridization (mCGH) (http://amba.charite.de/cgh /img/01/01e.gif) utilizes a spread o f metaphase cells on a slide onto which differentially labelled normal and tumour D N A are co-hybridized and viewed under the light microscope for regions of copy number gain and loss. 1 1 LOH analysis using microsatellite markers has been widely used to study oral tumour genomes (Rosin, et al. 2000; Rosin, et al. 2002; Zhang and Rosin 2001; Arai, et al. 2002; Beder, et al. 2003; El-Naggar, et al. 1998; El-Naggar, et al. 2001; Epstein, et al. 2003; Grati, et al. 2000; Gupta, et al. 1999; Hibi, et al. 1992; Imai, et al. 2001; Ishwad, et al. 1995; Ishwad, et al. 1996; Ishwad, et al. 1999; Kayahara, et al. 2001; Latif, et al. 1992; Maestro, et al. 1993; Ng, et al. 2000; Ogawara, et al. 1998; Ono, et al. 1999; Partridge, et al. 1996; Partridge, et al. 1999; Rowley, et al. 1996; Sunwoo, et al. 1999; Waber, et al. 1996; Wu, et al. 1997; Wu, et al. 1994; Yamamoto, et al. 2003). The principle of this technique is outlined in Figure 5. LOH analysis involves the PCR amplification of known microsatellite markers, or simple sequence repeat polymorphisms, using primers specific to the sequences flanking them (Garnis, et al. 2004a). In an individual heterozygous for a given marker, the product will appear as two bands of different sizes on an agarose gel. A comparison of the bands produced from patient matched normal and tumour DNA can determine whether there is an alteration at a given loci. While this technique has been highly effective for mapping regions of alteration, the resolution is limited by the availability of microsatellite markers in a given region and the status between two markers is therefore inferred (Garnis, et al. 2004a). Similarly, the number of markers that can be assessed is also limited by the quantity of DNA that is available (Garnis, et al. 2004a). Moreover, while it is able to detect alterations, it is unable to distinguish between an increase or decrease in copy number. 12 Figure 5: Loss of heterozygosity (LOH) analysis using microsatellite markers. L O H analysis utilizes primers specific to sequences flanking microsatellite markers, or regions of repeat sequences, to assess change within that region. A P C R reaction is generated using both matched normal and tumour D N A as the template and the product run on an agarose gel and the bands are compared. If the patient is heterozygous at a given allele, A , two bands wi l l appear on the gel using the normal template. If an alteration is present in the tumour, one band may disappear or the ratio in the intensities of a band may be different in the tumour as compared to the normal, and the tumour is said to show L O H at that marker. However, i f the patient is homozygous at a given allele, B , than a single band wi l l appear on the gel for both the normal and the tumour D N A template, and that given marker is said to be "non-informative." 13 1.6 Array CGH Technology In recent years, conventional CGH has been improved upon and developed into array CGH (aCGH). In contrast to conventional CGH, in which metaphase chromosomes are spread onto a glass slide, segments of normal DNA are spotted onto a slide (Solinas-Toldo, et al. 1997). Test DNA and normal diploid reference DNA are differentially labelled with fluorescent dyes and competitively hybridized to the array. Regions of gain and loss are detected by measuring the signal intensities for each dye and calculating the ratio of the dye signal intensities for a given clone on the array (Pinkel, et al. 1998; Solinas-Toldo, et al. 1997). The principle of aCGH is outlined in Figure 6. DNA segments used to construct CGH arrays include cDNAs (Pollack, et al. 1999), large-insert clones (such as bacterial artificial chromosomes, BACs) (Fiegler, et al. 2003; Pinkel, et al. 1998), cosmids (Buckley, et al. 2002), PI phage artificial chromosomes (PAC) (Fiegler, et al. 2003; Kallioniemi, et al. 1992), and single nucleotide polymorphism (SNP) oligonucleotides (Dumur, et al. 2003; Lindblad-Toh, et al. 2000). In previous studies, these arrays have given dense over-lapping coverage of a region known to be altered or to harbour candidate tumour suppressor genes or oncogenes. One example of this was the 8q21-24 regional BAC array, which was used to identify a novel amplification adjacent to the MYC oncogene in oral and lung cancer (Garnis, et al. 2004c). Another example was the chromosome 20 cosmid/PAC/BAC array (Pinkel, et al. 1998). However, these early tiling path regional arrays only span chromosomal segments previously identified to be commonly altered using conventional 14 BAC/PAC/Cosmid /cDNA/SNP DNA Tumour DNA • • • • • • • • • • • • Normal Reference DNA • DNA Microarray Scanned Image Figure 6: Principle of array comparative genomic hybridization. Briefly, segments of D N A are spotted onto a glass slide to generate a D N A microarray, tumour D N A and normal reference D N A are differentially labelled with cyanine dyes and competitively hybridized to the array, which is scanned for the signal intensity of both cyanine dyes. 15 lower resolution techniques (Davies, et al. 2005). Use of a whole genome tiling path array will allow the assessment of the status of DNA throughout the entire genome and will, therefore, enable fine-mapping of previously identified regions of alteration as well as serving to discover novel narrow regions of alteration not identified with previous, less sensitive techniques (Davies, et al. 2005). The first genome-wide arrays used contained segments of cDNA representing genes present on each of the 46 chromosomes (Pollack, et al. 1999). However, the significant difference between the size of the cDNA clones and the genomic DNA (due to the presence of introns in the genomic DNA) resulted in poor hybridization efficiency and, consequently, low signals and signal-to-noise ratios. The effect was that only very large copy number differences were detectable (Davies, et al. 2005). Moreover, since each cDNA represented a known gene, only those areas known to contain a gene were assessed, creating an experimental bias. Large-insert clone arrays, such as BAC arrays, provide much higher signal-to-noise ratios and can therefore detect single copy number changes with much greater sensitivity (Davies, et al. 2005). Until recently, these arrays were marker-based with the highest density areas containing clones at ~ 1.4 Mbp intervals (Fiegler, et al. 2003; Snijders, et al. 2001). Single-nucleotide polymorphism (SNP) arrays have also been useful in the genome-wide analysis of tumour genomes (Fiegler, et al. 2003; Greshock, et al. 2004; Davies, et al. 2005) and, in particular, oral genomes (Tong, et al. 2004a; Zhou, et al. 2004a; Zhou, et al. 2004b). Previous arrays have consisted of-10,000 25-mer and -85,000 70-mer oligonucleotides (Bignell, et al. 2004; Lucito, et al. 2003; Zhao, et al. 2004) representing regions across the entire genome. 16 While these arrays provide the LOH status at each SNP locus, their ability to detect copy number changes is limited to high copy number differences (Zabarovsky, et al. 2002). In addition, they are compromised by problems with cross-hybridization of oligo probes and the availability of SNP markers (Bignell, et al. 2004; Davies, et al. 2005). While PCR amplification of sample DNA provides complexity reduction of sample tumour DNA in order to reduce the compounding cross-hybridization issue, it also introduces additional noise and variability in hybridization (Davies, et al. 2005). Increased characterization of SNPs will address the need for improved resolution of this technique (Davies, et al. 2005; Matsuzaki, et al. 2004). 1.7 Whole Genome Tiling Resolution BAG Array CGH The recent development of the whole genome sub-megabase tiling array (SMRT array), which contains 32,433 overlapping BAC clones spotted in triplicate onto two glass slides, has increased the resolution of BAC array CGH to ~40-80 kb, allowing detailed genome-wide analysis of tumour genomes (Ishkanian, et al. 2004; Davies, et al. 2005). Figure 7A shows an example of a SMRT array CGH profile for the lung cancer cell line H526 demonstrating the identification of detailed alterations and their breakpoints. This technology has enabled the identification of novel gains and losses not detectable with previous technology in cancer cell line DNA (De Leeuw, et al. 2004). Figure 7B demonstrates the improvement in resolution of the SMRT array as compared with conventional techniques such as LOH and interval-marker based genome-wide array CGH. Application of this technology to archival tumour DNA will be highly useful in 17 Iii III I! !li IK in lit i 10 11 12 13 14 15 I g S Ii 16 Ii f IB \\% i i i»iti i i i 17 18 19 20 21 22 Q Loss of heterozy-gosity (LOH) using Microsatellite Markers Limited resolution Marker-based — _ genome array CGH .. -1 .4 Mbp Resolution Sub megabase Resolution Tiling-Path ~Z_ ""Z_ ""Z_ _ ~Z_ ~I_ _ _ ~ (SMRT) Array CGH — — — — — — — Tiling Resolution Figure 7: Complete coverage whole genome array comparative genomic hybridization. ( A ) Sub-megabase resolution tiling-path (SMRT) array provides comprehensive profiling of the entire human genome. Shown here is the whole genome profile for the H526 lung cancer cell line (Ishkanian, et al. 2004). (B) Resolution comparison of common marker based genome analysis techniques. The resolution of L O H is limited to the availability o f microsatellite markers, while until recently genome-wide marker based arrays provided only ~1.4 Mbp resolution. The S M R T array provides overlapping coverage resulting in "tiling resolution7' (Davies, et al. 2005). 18 the identification of novel genetic changes and the fine mapping of known genetic changes in a variety of cancers. 1.8 Hypotheses & Objectives The hypothesis for this thesis project is three-fold: 1) Archival OSCC specimens will support chromosomal and whole genome array CGH profiling. 2) Chromosomal alterations critical to OSCC progression will be present in multiple patient samples. 3) Candidate tumour suppressor genes and/or oncogenes reside within the minimal regions. Three objectives were designed in order to test this hypothesis: The first objective of this thesis was to demonstrate the utility of array CGH in the analysis of archival OSCC tumour genomes. This was done by hybridizing 19 OSCC specimens using a 3p-arm specific array, comprised of 535 over-lapping BAC clones spanning the 3p arm, and aligning the resulting profiles to identify minimal regions of alteration. 19 The second objective was to identify novel recurring alterations critical to oral carcinogenesis and to define these regions with sub-megabase resolution. This was done by comparing the whole genome array CGH profiles of 20 OSCC in order to identify regions. The third objective was to relate these regions to the map of the human genome to identify genes within recurring regions and to verify these regions by expression analysis of the genes within them. This was achieved by reverse transcriptase PCR of candidate genes on OSCC RNA specimens. 20 CHAPTER 2: MATERIALS AND METHODS 2.1 Specimen Accrual, Microdissection & DNA Extraction Samples were selected based on differing criteria for chromosome 3p-array analysis and whole genome array analysis. Samples used for 3p-array analysis were chosen based on LOH data for 3p. Samples used for whole genome-array analysis were chosen based on the availability of DNA. Formalin-fixed paraffin embedded OSCC specimens were obtained from the British Columbia Oral Biopsy Service and diagnoses confirmed by an oral pathologist (Dr. Lewei Zhang). Tumour cells were microdissected from the normal tissue under the supervision of Dr. Zhang. All microdissected specimens were stored in cryovials in nitrogen tank until extraction. Samples were then thawed and pulse-centrifuged to remove droplets in the cap. The cells were then resuspended, transferred to a 1.5ml Eppendorf tube, and the cryovial was once again rinsed with a Tris-EDTA solution (0.5 M Tris, 0.02 M EDTA, 0.01 M NaCl, pH 8.9) to remove residual cells which were then transferred to the same 1.5 ml Eppendorf tube (final volume was maintained below 1.5 ml). The tubes were then centrifuged at 8,000 rpm at 4°C for 10 minutes. Excess supernatent was removed, confirmed to contain no residual cells (using a light microscope), and discarded, leaving the final volume to be 270 pi in each tube. These samples were then mixed by pipetting with 30 pi SDS-PK (10% sodium dodecyl sulphate, 5 mg/ml proteinase K) and incubated at 48°C for 72 hours. Proteinase K (20 mg/ml in Tris-EDTA solution) was replenished 21 every 2 hours. Tumour DNA was extracted using a standard phenol-chloroform protocol and precipitated with ethanol. The resulting DNA was then fluorometrically quantified using Picogreen (Molecular Probes, Eugene, OR). DNA was extracted and quantified as previously described (Zhang, et al. 1997). The clinical information and demographics for each case is listed in Appendix 1. 2.2 3p Array CGH 2.2.1 3p Array Construction Over 600 overlapping human BAC clones spanning the 3p arm from 3pl2.3-26.3 were selected from the Roswell Park Cancer Institute (RPCI)-l 1 library (Osoegawa, et al. 2001) based on their relative contig position using Fingerprinting Contig (FPC) software and the fingerprint data of -400,000 human BAC clones. The relative positions of the clones were confirmed by comparison with sequence-based BAC contiguous assemblies such as the Ensembl Genome Browser (Clamp, et al. 2003), the NCBI Map Viewer (Wheeler, et al. 2003), and the UCSC Genome Browser (Karolchik, et al. 2003; Kent, et al. 2002). Bacterial clones were obtained from glycerol stocks and streaked onto LB agar plates containing 12.5 pg/ml choramphenicol and incubated overnight. For each clone, a single colony was inoculated in 1.2. ml LB broth containing 12.5 ug/ml choramphenicol in a well of a 2 ml 96-well plate. The filled 2 ml 96-well plate was incubated at 37°C for 24 hours with shaking. Extraction of BAC DNA was performed by the BC Cancer 22 Agency Genome Sciences Centre. BAC DNA was extracted from the resulting pellet of cells using a maxi-prep kit (Clonetech), precipitated with isopropanol and resuspended in 30 pi TE (10 mM Tris-HCl pH8, O.lmM EDTA). The identity of each BAC clone was confirmed via a Hindlll digest fingerprint. A 2 pi aliquot of the extracted DNA was mixed with 1 ul Hindlll restriction enzyme (10 units), 1. pi 10X RE buffer, and 6 pi sterile, distilled water and was incubated at 37°C overnight (Ishkanian, et al. 2004; Watson, et al. 2004b). The digested product was then run on a 1% agarose gel, stained with SYBR green and visualized with the Storm phosophoimager. The Hindlll fingerprint was then compared to the in silico fingerprint with the FPC software. Clones that failed to generate informative or matching hindlll fingerprints were eliminated from the clone set to under 600 clones. Hindlll fingerprint confirmed clones were then amplified by linker-mediated PCR (LM-PCR) to generate sufficient DNA for spotting. To prepare the BAC DNA for L M -PCR, fifty nanograms of BAC DNA from each clone was transferred to a 96 well plate and digested for 8 hours with 5 U Msel in a 40 pi reaction, followed by inactivation at 65°C for 10 minutes. This enabled both the removal and digestion of the genomic DNA insert from the BAC and the exposure of overhangs necessary for ligation with the primer linkers. The digested DNA was then ligated with primer linkers. Msel long (5'-AGTGGGATTCCGCATGCTAGT-3') and Msel short primers (5' -TAACTAGCATCG-3') (Alpha DNA, Quebec) were pre-annealed together for 5 minutes at room temperature prior to addition to the ligation reaction. Ten percent of the digested product was then 23 transferred to a new 96 well plate and mixed with 0.2 uM of both Msel long and short primers, 80 U of T4 DNA ligase, and NEB ligase buffer (New England Biolabs) in a 40 ul volume and ligated at 16°C overnight (12-16 hours). This enabled the ligation of specific primer linker sites to the genomic DNA BAC inserts for PCR amplification. Two rounds of PCR were employed. For the first round of PCR, a 2.5 pi aliquot of the linker-ligated DNA product was brought up in 50 pi reaction consisting of 8 mM MgCh, 1 mM dNTP's (Promega), 0.4 uM Msel long primer and 5 U of Taq polymerase (Promega,storage buffer B) in Promega PCR buffer. The following PCR program was used: Step.l: 95° C, 3 minutes Step 2 95° C, 1 minute Step 3 55° C, 1 minute Step 4 72° C, 3 minutes Step 5 Return to step 2, 30 times Step 6 72° C, 10 minutes For the second round of PCR, 0.25 pi of the first round PCR product PCR amplified under the same conditions for 35 cycles, rather than 30 cycles. This product, the amplified fragment pool (AFP), was then precipitated with 5 pi (1/10 volume) 3 M sodium acetate and 150 pi (2.5 X volume) ethanol at -20°C for 1 hour. The final concentration of the ethanol precipitated DNA was quantified using a ND-1000 spectrophotometer (Nanodrop, Delaware). Generally, the concentration concentrations varied between 40-50 pg per sample. Clones that failed to amplify were also eliminated from the array reducing the clone set to 539 BAC clones. 24 The amplified DNA was then dissolved in 50 pi of 0.5% Micro spotting solution (Telechem) to a final concentration of 1 pg/ul, heat denatured and re-arrayed for robotic printing. Amplified BAC DNA was spotted in triplicate onto SuperAmine substrate slides (Telechem/ArraylT, Sunnyvale, CA) with Stealth Micro Spotting Pins (SMP2.5) using a VersArray Chip Writer Pro system (Bio-Rad, Mississauga, Ontario, Canada). Each "spot" was -100 um in diameter with 180 um spacing between spots. The positive charge of the amine slides enabled the binding of the negatively charged DNA clones to the slide. Baking at 80°C for 1 hour and UV cross linking at 2600 mJ enabled the cross-linking of the DNA to the amine slides. Normalization was achieved by spotting linker-mediated PCR amplified normal male human DNA 48 times. The slides were then washed to remove unbound DNA. The array was finally reduced to 535 clones following the elimination of duplicate redundant clones. The clone list is presented in Appendix 2. 2.2.2 3p Array Probe Labelling Tumour and normal diploid male reference genomic DNA (lOOng each) were random prime labelled with cyanine-5 and cyanine-3 dCTP respectively using a Bioprime Labelling Kit (Invitrogen, Carlsbad, CA). In separate reactions, the DNA was mixed with 10 ul 2.5 X random primers buffer and brought up to a final volume of 20 ul with sterile distilled water. The DNA was then incubated at 100°C for 10 minutes, placed on ice and combined with 40U Klenow, 2.5 pi 10 X dNTPs (2mM dATP, dGTP, dTTP, ImM dCTP) and 2nmoles of cyanine-5 (tumour DNA) or cyanine-3 (normal male reference) labelled dCTP. The probe mixture was then incubated at 37°C for 18 hours, 25 combined and purified together to remove unincorporated nucleotides using ProbeQuant Sephadex G-50 Column (Amersham, Baie d'Urf, PQ). The purified probe was then combined with lOOpg human Cot-1 DNA (Invitrogen, Burlington, ON), precipitated with ethanol and redissolved in 28 ul DIG Easy hybridization buffer (Roche, Laval PQ), 3.5 pi (20 ug/ul) sheared herring sperm DNA (Sigma-Alderich) and 3.5 pi (100 ug/ul) yeast tRNA (Calbiochem). This probe mixture was then denatured at 85° C for 10 minutes and the probe was blocked at 45° C for 1 hour. 2.2.3 3p Array Hybridization 3p-arrays were pre-hybridized by applying 28 u.1 DIG Easy hybridization buffer (Roche, Laval PQ), 4 u.1 (20 u.g/u.1) sheared herring sperm DNA (Sigma-Aldrich, Oakville, ON, Canada), and 4 p.1 10% BSA to the arrays with a 24 x 40 mm cover slip in an ArraylT Array Hybridization Cassette containing lOul water and incubating at 45°C for 1 hour. Arrays were then rinsed with sterile distilled water and placed in 100% isopropanol followed by air-drying. The denatured and pre-blocked probe was then applied to the slide with a 24 x 40 mm cover slip and placed in a hybridization chamber as described above at 45°C for 40 hours. 2.2.4 3p Array Imaging and Normalization Hybridized 3p arrays were scanned for the cyanine-3/cyanine-5 signal intensities using a charge coupled device (CCD) based imaging system and analyzed using Softworx array analysis software (Arrayworx, API, Issaquah, WA). Intensities were 26 normalized using a scale factor based on the signal intensities of the 48 human genomic DNA control spots on the array. The normalized log2 cyanine-5/cyanine-3 signal ratio for each BAC clone was plotted in a graph versus relative tiling path position. A log2 signal ratio of 0 corresponds to equivalent copy number between tumour and normal reference DNA. 2.3 Whole Genome SMRT Array CGH 2.3.1 Whole Genome SMRT Array Construction Clone selection for the whole genome SMRT array was performed by the British Columbia Cancer Agency Genome Sciences Centre (BCCA GSC) (Krzywinski, et al. 2004) from RPCI-11, RPCI-13 and Caltech D1/D2 (CTD1/2) clone libraries using the human BAC fingerprint based physical map developed by the Washington University Genome Sequencing Center (McPherson, et al. 2001). In addition, clones were cultured, the BAC DNA was extracted, and the identity of each clone was confirmed by comparison of Hindlll fingerprints as described with the FPC BAC fingerprint database (http://bacpac.chori.org/genomicRearrays.php) at the BCCA GSC. The BAC DNA for all 32,433 clones was provided to our lab in 96-well plates. Construction of the whole genome SMRT array has been previously described (Ishkanian, et al. 2004; Watson, et al. 2004b). BAC DNA was amplified through linker mediated PCR (LM-PCR) as described above, to generate AFP with sufficient DNA concentration for spotting. However, the primer linkers, Msel long and Msel short, used for the whole genome SMRT array, were linked to an amino acid group. This was done 27 because while the 3p-array was spotted onto SuperAmine slides, the whole genome array was spotted onto aldehyde slides (Erie Scientific). While the amine slides bound the spotted DNA through an electrostatic interaction followed by cross-linking, the aldehyde groups on the Erie Scientific slides formed a covalent bond with the amino acid group on the AFP DNA. Following linker mediated amplification the identities of 3 clones per plate 96-well plate were confirmed by sequencing (Watson, et al. 2004b). Each 10 ul sequencing reaction consisted of 2 ul of the AFP, 4 pi Big Dye (Perkin-Elmer), 0.32 pmol T7 primer (5'-TAATACGACTCACTATAGG-3') or SP6 primer (5'-ATTTAGGTGACACTATAG-3') (Alpha DNA, Quebec) and was amplified using the following program: Stepl: 95° C, 1 minutes Step 2 95° C, 15 seconds Step 3 50° C, 5 seconds Step 4 72° C, 4 minutes Step 5 Return to step 2, 85 times Step 6 72° C, 10 minutes The product was then purified either by ethanol precipitation with 1 ul 3 M sodium acetate and 25 ul ethanol, or a PCR Min-elute kit (Qiagen). The purified product was then resolved using an ABI Model 377 or ABI Model 3700 sequencer (Applied Biosystems) and analysed using NCBI BLAST to query the non-redundant and high throughput genomic sequences database of GeneBank v.2.2.5 (Watson, et al. 2004b). 28 While confirmation of all clones in the array would have been desirable, cost limitations restricted the confirmation to 3 clones per plate. The AFP DNA was then dissolved in 50 pi of 0.5% Micro spotting solution (Telechem, Sunnyvale, CA), denatured by boiling, re-arrayed into 384-well plates for robotic printing and spotted in triplicate onto two aldehyde glass slides as described above, producing an array containing 97,299 elements (Ishkanian, et al. 2004). Normalization was achieved based on median signal intensities. The ability to detect single copy changes was confirmed as part of the quality control of array production (De Leeuw, et al. 2004). This involves hybridization of normal male versus normal female DNA and examination of the X chromosome profiles as previously described by others (De Leeuw, et al. 2004; Ishkanian, et al. 2004). 2.3.2 Whole Genome SMRT Array Probe Labelling Tumour and normal diploid male reference genomic DNA (200 ng each) were random prime labelled with cyanine-3 and cyanine-5 dCTP respectively. In separate reactions, the DNA was mixed with 10 pi 5 X random primer buffer (Promega) containing random octomer primers (Alpha DNA, Quebec) and brought up to a final volume of 33.5 pi with sterile distilled water. The probe was then incubated at 100°C for 10 minutes, placed on ice and combined with 45 U Klenow (Promega), 7.5 pi 10 X dNTPs (2 mM dATP, dGTP, dTTP, 1.2mM dCTP) and 4 nmoles of cyanine-3 (tumour DNA) or cyanine-5 (normal male reference) labelled dCTP. The probe mixture was then incubated at 37°C for 18 hours, combined together, and purified to remove 29 unincorporated nucleotides using ProbeQuant Sephadex G-50 Column (Amersham, Baie d'Urf, PQ). The purified probe was then combined with 400 pg human Cot-1 DNA (Invitrogen, Burlington, ON), precipitated with ethanol and redissolved in 80 pi DIG Easy hybridization buffer (Roche, Laval PQ), 10 pi (20 ug/ui) sheared herring sperm DNA (Sigma-Aldrich, Oakville, ON, Canada) and 10 pi (10 ug/ul) yeast tRNA (Calbiochem, La Jo 11a, CA). This probe mixture was then denatured at 85° C for 10 minutes and the probe blocked at 45° C for 1 hour. 2.3.3 Whole Genome SMRT Array Hybridization Since these arrays were printed onto aldehyde slides, as apposed to the amine slides on which the regional array was spotted, the pre-hybridization steps were much different. Slides were immersed in a blocking solution comprised of lg sodium borohydride, 400 ml 1 X PBS, 100 ml ethanol, for 5 minutes to block the remaining exposed aldehyde groups from binding additional DNA, followed by 3 rinses in sterile, distilled water for 2 minutes each. The slides were then immersed in boiling sterile, distilled water for 2 minutes to denature the DNA and then air dried. The slides were then pre-warmed to 45° C prior to application of the probe. Denatured and pre-blocked probe was then split equally and applied to the two pre-warmed halves of the array with a 24 x 40 mm cover slip in an ArraylT Array Hybridization Cassette (Telechem, Sunnyvale, CA) containing 10 pi water and incubated at 45°C for 40 hours. The hybridized arrays were then washed five times for 5 minutes each at 55° C with 0.1 X SSC/0.1% SDS, followed by five rinses with 0.1 X SSC. 30 2.3.4 Whole Genome SMRT Array Imaging and Normalization Images of the hybridized arrays were captured through Cyanine 3 and Cyanine 5 channels using a charge-coupled device (CCD) camera system (Applied Precision, Issaquah, WA). Images were then analyzed using Softworx Tracker Spot Analysis software (Applied Precision). SeeGH custom software was used to visualize all data as log2 ratio plots (Chi, et al. 2004). This software is available publicly online (http://www. flintbox. com/technology, asp? tech=FB312FB). Clones with standard deviations above 0.075 or with signal-to-noise ratios less than 10 were filtered from the raw data. Clones were normalized based on the mean signal intensity of all spots. 2.4 Identification of alterations and alignment of BAC array profiles 2.4.1 Identification of alterations using the 3p-array A combination of strategies was used to identify copy number gains and losses in all 19 tumours including threshold cut-off analysis and 3p profile alignment. Since spots exhibiting a signal ratio outside of 3 standard deviations from 0 were eliminated, a +/- 0.2 log2 ratio threshold was established for defining copy number alterations in each experiment. Regions of DNA achieving this value were required to involve more than one consecutive clone in order to qualify as a copy number alteration. Alignment of all 3p profiles enabled the identification of minimal regions of alteration recurring in multiple samples. 31 2.4.2 Identification of alterations using the SMRT array A combination of strategies was used to analyse the changes in all 20 tumours, including assessment with aCGH-Smooth with frequency plot analysis, individual karyogram visual analysis, and SeeGH karyotype multiple alignment. Analytical software called aCGH-Smooth was used to detect genetic alterations at a gross level. Briefly, clones with standard deviations above 0.075 or with signal-to-noise ratios less than 10 were filtered from the raw data. Genomic imbalances and their associated breakpoints were identified using genetic local search algorithms within the aCGH-Smooth software package (Jong, et al. 2004) and confirmed by visual inspection of the normalized data. aCGH-Smooth calculates the probability that the signal ratio for each BAC clone resides within that of a set of previous clones using maximum likelihood estimation (Jong, et al. 2004). The "smoothed" data was then aligned in a frequency plot to look for "hot spots" of alteration, or regions frequently altered, in the oral tumour genome. Individual SeeGH karyograms were also visually analyzed for copy number alterations. While threshold analysis was initially employed with the regional array, subsequent experience demonstrated to us that this approach was flawed. That is, each tumour sample is independent and therefore has a different variability often influenced by differing levels of normal DNA contamination as well as differing DNA quality. Use of a uniform threshold, therefore, is not applicable. Indeed, using the SMRT array well known alterations such as 3p and 8p loss, though visible, was often not detected if 32 threshold analysis was applied. Finally, all 20 SteeG//karyograms were aligned to identify minimal regions of gain or loss recurring in multiple tumour samples. 2.5 Expression Analysis Confirmation 2.5.1 Frozen Tissue Accrual, Microdissection & RNA Extraction Frozen tissue was obtained from the BC Oral Biopsy Service, sectioned, diagnosis confirmed, and microdissected by oral pathologist, Dr. Lewei Zhang. Microdissected tissue was suspended in 1 ml RNAlater (Ambion, Austin, TX) and stored at -20° C until RNA extraction. Before extraction, tissue was resuspended in 1 ml 1 X PBS. Total messenger RNA was extracted using a Dynabeads® mRNA DIRECT™ Kit (Dynal, Oslo, Norway). Briefly, tissue was spun down at 14,000g for 15 minutes; supernatant was removed, re-suspended in 1000 ul lysis/binding buffer (Dynal, Oslo, Norway) and homogenized. The suspension was then spun down at 14,000g for 15 minutes and the supernatant was isolated. 250 | i l suspended beads was purified and separated from supernatant using a magnetic tube rack. This supernatant was discarded and supernatant from the tissue lysis was than added to the beads and mixed. The Dynabeads were then washed twice with buffer A and once with buffer B (Dynal, Oslo, Norway). mRNA was eluted from the Dynabeads with 20 pi 10 mM Tris-HCl. Total mRNA yields were assessed using a Nanodrop Spectrometer (Nanodrop Technologies, Wilmington, DE). 2.5.2 Reverse Transcriptase PCR Aliquots of 40 ng of RNA were converted to cDNA in 20 pi reaction volumes using the Superscript II Kit (Invitrogen, Burlington, ON). Briefly, 40 ng RNA and 0.5 ug 33 polydT primers (5'-TTTTTTTTTTTTTTTTTT-3') was suspended in 12 pi sterile distilled water and denatured at 70° C for 10 minutes, followed by addition of 4 pi of 5 X First Strand Buffer (250 mM Tris-Cl, pH 8.3, 375 mM KC1, 15 mM MgCl 2), 2 pi 0.1 M DTT, 1 pi 10 mM dNTPs and incubation at 42° C for 2 minutes. 200 U Superscript II Reverse Transcriptase was then added and the reaction was incubated at 42° C for 50 minutes and stopped by incubation at 70° C for 15 minutes. cDNA yield was confirmed by PCR amplification using primers specific to p-actin. Two sets of primers were used - a positive control set, in which both primers anneal to exon sequences (5'-GATGTGGATCAGCAAGCA-3 75'-GAAAGGGTGTAACGCAACT-3'), and a negative control set, in which one primer anneals to an introns sequence (5'-AGACGATGCAGATCC-375'-GCATGGCAGCAGCAC-3'). cDNA (1 pi) was amplified using a Promega Kit in a 20 pi volume containing 14.2 pi sterile distilled water, 2 pi 10 X PCR buffer, 1.6 pi 25 mM MgCi2, 0.5 pi 10 mM dNTPs, 0.5 pi primers (forward and reverse combined), and 0.2 pi Promega Taq DNA polymerase (5 U/pl). The following PCR program was used: Stepl: 95° C, 1 minute Step 2 95° C, 30 seconds Step 3 55° C, 30 seconds (positive control set)/62° C, 30 seconds (negative control set) Step 4 72° C, 30 seconds Step 5 Return to step 2 35 times 34 Step 6 72° C, 10 minutes 5ul of PCR product was than run on 0.7% agarose gel to assess band intensity. To assess candidate gene (TRIO, TENS1, CDK6) and control gene (B2-microglobulin, /32M) expression 1 pi and 0.33 pi of cDNA, respectively, was PCR amplified in a 20 pi reaction as above using following primers specific to the gene of interest: (forward): 5' - AAGGAACC AAGCGAGC AGT-3' TRIO (reverse): 5' -GGTGTGCTATCCTGCAGTTGT-3' TENS1 (forward): 5' -TCTTCTTCCGGAGGCATTAC-3' TENS1 (reverse): 5' - AAC AGGTGGC AC ACATTATCC-3' CDK6 (forward): 5' -GAAGACTGGCCTAGAGATGTTG-3' CDK6 (reverse): 5' -CCTGAAGTATGGGTGAGAC A-3' B2M (forward): 5' -GATGAGTATGCCTGCCGTGT-3' B2M (reverse): 5' -GC AAGCAAGCAG AATTTGGA-3' Primers were designed to span an intron as a means to assess for genomic DNA contamination. Positive control templates comprised of cDNA from Human MTC™ cDNA Panels I and II (Clonetech, BD Biosciences, Palo Alto, CA) (TRIO: colon, heart, thymus; TENSl: placenta, pancreas, colon; CDK6: testis, thymus, heart; f!2M: heart, placenta, ovary). The following PCR programs were used for each PCR reaction: 35 Step 1 95° C, 1 minute Step 2 95° C, 30 seconds Step 3 62° C (TRIO), 60° C (TENSI), 64° C (CDK6), 56° C (B2M), 30 seconds Step 4 72° C, 30 seconds Step 5 Return to step 2 35 times Step 6 72° C, 10 minutes Five micro litres of B2M PCR product for each sample was loaded onto a 10% polyacrylamide gel (37.5 acrylamide:l bis-acrylamide) and run for 2 hours in 1 X TBE at 400V. An additional 5ul of cDNA from either TRIO, TENSI or CDK6 PCR reactions was loaded into the corresponding sample well and run for an additional hour at 400V. The gel was stained in 1 X SYBR-Green (Roche, Laval, Quebec, Canada)/1 X TBE for 30 minutes and imaged using a Storm Phosphoimager. ImageQuant analysis software (Molecular Dynamics, Piscataway, NJ, USA) was used to measure band and background intensities. The averaged background intensity value was subtracted from each sample band intensity value. For each gene, the ratio of the candidate gene band intensity over the P2M band intensity was calculated and averaged across the tumours and normal samples, and plotted into a bar graph. In addition, for each gene the band intensities of the candidate gene in the tumour samples and normal samples were compared using the Mann-Whitney U Test. The null hypothesis, that there is no difference in the expression level of (candidate gene) between the normal and tumour samples, was tested against the alterative hypothesis, that there is 36 a difference in the expression level of (candidate gene) between the normal and tumour samples. The Mann-Whitney U Test is a non-parametric statistical ranking test employed when the assumptions of normality and/or a sufficient sample size cannot be met. In this case, the candidate gene band intensities for all the tumours and normal samples were ranked. Assuming ni is the number of normal samples (for example, 6), n 2 is the number of tumour samples (for example, 4), R\ is the sum of the ranks assigned to normal samples, and R2 is the sum of the ranks assigned to the tumour samples then, U = ni* n 2 + ni(ni + 1) - R i and U'= n 2* ni + n2(n2 + 1)-R 2 and P0.05(20,6,4) - 22 If U or U ' > po.o5(20,6,4) than the null hypothesis is rejected and the alternative hypothesis is supported. 37 C H A P T E R 3: R E S U L T S & D I S C U S S I O N 3.1 Summary of 3p Array Results The 3p-array employed in this thesis project was constructed as part of the regional array - a pilot project for the whole genome Sub-Megabase Resolution Tiling-set (SMRT) array. The regional array consists of 2770 BAC clones providing overlapping coverage of 26 regions. The regions were selected based on previous studies in print. Included among these regions on the regional array was the 3p-arm specific array which consisted of 535 near-overlapping BAC clones spanning from 3pl2.3-3p26.3. Use of the regional array enabled the optimization of array comparative genomic hybridization (CGH) technology for the detection of single copy number alterations in archival genomic DNA. 3.1.1 Validation of the Regional Array To test the ability of the regional array to detect single copy number alterations in archival DNA, 41 BAC clones representing different autosomes as well as sex chromosomes were included in the regional array and their hybridization signals compared. Normal non-archival male diploid DNA and archival formalin-fixed female DNA were differentially labelled and co-hybridized to the array and the log2 ratios of the dye signal intensities of these 41 BAC clones were calculated and plotted to visualize the differences in the signal intensities of the autosome clones and X and Y clones (Figure 8). As shown, BAC clones representing autosomal regions exhibited log2 ratios at or around 38 BAC Chromosome ' Y Clone t • » , » * :L_* * t > , Autosomal Clones ( Chromosome X Clone Figure 8: Detection of single copy changes on chromosomes X and Y with the regional array. Forty-one BAC clones representing different autosomes and sex chromosomes were included in the regional array for comparison following co-hybridization with normal non-archival male diploid reference and archival formalin-fixed female DNA. When the log2 ratios of the signal intensities are plotted as shown here, single copy number differences are easily visible on the X and Y clones. Clones lying around zero represent autosomal DNA, thus showing no change in the normal samples. The single clone lying above zero represents Y chromosome DNA, thus showing an increased copy number ratio, and the two clones lying below zero represent X chromosome DNA, thus showing a low signal ratio. This demonstrates the sensitivity the regional array (and thus the 3p-arm specific array) to the detection of single copy number changes. 39 0, while those representing chromosome Y and X, exhibited log2 ratios near +1.7 and -0.7, respectively. This demonstrated the sensitivity of the regional array, and thus the 3p array, to detect single copy alterations in archival DNA. 3.1.2 Detection of Genetic Alteration on Chromosome 3p Loss of heterozygosity (LOH) on chromosome 3p is well known and extensively studied in oral squamous cell carcinoma (OSCC) and has been shown to be associated with increased risk of progression in oral premalignant lesions (OPL) (Rosin, et al. 2000; Rosin, et al. 2002). Several studies have sought to fine-map the minimal regions of alteration in oral cancer using LOH (Arai, et al. 2002; Buchhagen, et al. 1996; Hibi, et al. 1992; Killary, et al. 1992; Maestro, et al. 1993; Partridge, et al. 1996; Partridge, et al. 1994; Rowley, et al. 1996; Roz, et al. 1996; Uzawa, et al. 2001; Waber, et al. 1996). However, LOH is unable to distinguish between copy number gains and losses and the resolution is limited by the availability of microsatellite markers. Moreover, following the sequencing of the human genome many of the microsatellite markers have been re-mapped making the data difficult to interpret (Garnis, et al. 2003). Application of array comparative genomic hybridization (aCGH) for the detailed analysis of chromosome 3p in OSCC addressed the need to fine-map and characterize the minimal regions of alteration while, in addition, providing a platform on which to test the utility of array CGH in studying OSCC genomes. Nineteen formalin-fixed OSCC DNA samples were analyzed against non-archival normal male diploid DNA on the 3p-array. The demographics of these cases are shown 40 in Appendix 1. The log2 ratio for each clone was plotted according to the relative position to obtain a "3p profile." In order to identify regions of alteration, individual tumour profiles were first visually analyzed for regions of copy number gain and loss using a +/- 0.2 log2 ratio threshold. By relating the NCBI accession number for bordering clones to the fingerprint map of the human genome (FPC) (Soderlund, et al. 1997), the estimated base pair positions of the clones were determined and the breakpoints identified. Two cases bared no copy number changes on 3p and three cases exhibited a whole arm deletion. The remaining 14 tumours showed multiple segmental copy number changes. Figure 9 shows three examples of 3p profiles showing no change (A), whole arm loss (B), and segmental copy number changes (C). These profiles demonstrate the ability of aCGH technology to detect copy number alterations. Detection of loss on 3p in 17/19 cases (85%) correlates well to that which has previously been reported using LOH and conventional CGH (Buchhagen, et al. 1996; Maestro, et al. 1993; Rosin, et al. 2000; Rosin, et al. 2002; Rowley, et al. 1996; Waber, et al. 1996). 3.1.3 Fine-mapping Genetic Alterations in 3p In order to identify recurring minimal regions of alteration (MRA) in these samples and to define the minimal boundaries of these changes, all nineteen profiles were aligned (Figure 10). Among the 14 samples showing segmental copy number alterations on 3p, 5 distinct recurring MRAs were defined - 4 regions of loss (regions, 1,3,4,5) and 1 region of gain (region 2) (Figure 10). Only alterations present in >6 of the 14 cases showing segmental copy number changes were considered recurrent. These regions are 41 o Figure 9: Representative examples of three 3p-array profiles. The log2 ratios o f the signal intensities of each clone in the 3p-arm specific array were plotted according to relative clone position. Regions o f gain and loss were then identified visually as well as using a +/- 0.2 log2 ratio threshold. Clones lying along zero show no change, while clones lying above and below zero show gain and loss, respectively. Green shading represents areas of copy number decrease, while red shading represents areas of copy number increase. Tumour 539T demonstrates no changes on 3p, and thus a normal profile (A). Tumour 43T shows copy number decrease of the entire 3p arm (B), and tumour 528T shows segmental regions o f both copy number increase and decrease (C). 42 Figure 10: Alignment of 19 oral squamous cell carcinoma (OSCC) array comparative genomic hybridization (aCGH) profiles for chromosome arm 3p. Five minimal regions of alteration ( M R A ) were identified and are marked by grey arrows including 4 regions o f loss (regions 1,3-5) and the first ever reported region of gain on 3p (region 2). Red bars denote regions of copy number increase, while green bars represent regions copy number decrease. Tumour specimens are listed on the left. Microsatellite markers commonly used in the analysis of oral tumour genomes were re-mapped and are listed in relative position along the top. Markers in blue have been reported in previous studies to have L O H frequencies o f 60% (Beder, et al. 2003; Kayahara, et al. 2001; Rowley, et al. 1996; Roz, et al. 1996). The 5 M R A and the genes within them are summarized in Table 1. 4 3 summarized in Table 1. Region 1 at 3pl2.2 (RP11-603J22)-3p21.1 (RP11-122D19), which was lost in 7/14 cases showing segmental copy number changes, was the largest region at ~27 Mbp in size. This region contained 750 known genes. Interestingly, this region encompasses the fragile histidine triad gene, or FHIT, shown to be altered and aberrantly expressed in oral lesions (Lee, et al. 2001; Paradiso, et al. 2004). FHIT, which lies at 3pl4.2, is a large gene comprised of 10 exons that produces a 1.1 kb transcript (Chang, et al. 2002). The resulting protein is a diadenosine triphosphate hydrolase (Barnes, et al. 1996), which has been demonstrated to function in the induction of apoptosis (Barnes, et al. 1996; Fong, et al. 1997; Ohta, et al. 1996; Pace, et al. 1998; Sard, et al. 1999; Siprashvili, et al. 1997). The FHIT gene is prone to interruption due to the presence of the FRA3B common fragile site (Fong, et al. 1997). However, the vast size of region 1 suggests the presence of other tumour suppressor genes within it. Analysis of a larger cohort of samples or the analysis ofearly lesions, such as dysplasias and CIS samples, may assist in narrowing down the region. Table 1: Five minimal regions of alteration identified using the 3p-arm specific array. Region Location Gain/Loss Centromeric B A C Telomeric B A C Size (Mbp) Known Genes 3pl2.2-3p21.2 3p21.3 3p22 3p24.1 3p26.1 loss gain loss loss loss RP11-603J22 (AC107030) RP11-447D11 (AC 104447) RP11-56P22 (AC093557) RP11-35C18 (AC018359) RP1-7715 (AC011327) RP11-122D19 (AC018354) RP 11-509121 (AC 104304) RP11-598J13 (AC073353) RP11-539L2 (AC092503) RP11-7M24 (AC012136) 27 750 (FHIT) 0.7 S(TSP50, TDGF-1) 0.7 no genes 0.8 1 (RBMS3) \(GRM7) 44 The remaining 3 regions of loss (regions 3-5) were much smaller than that of region 1 at less than 2 Mbp. Region 3 at 3p22 (RP11-56P22 to RP11-598J13) was -0.7 Mbp and, interestingly, contains no known genes. However, while this region contains no known genes it may harbour an unknown gene or an important regulatory region. Further studies on this region may shed more light on its potential role in oral tumourigenesis. Region 4 at 3p24.1 (RP11-35C18 to RP11-539102) was ~2 Mbp and contains 1 known gene, the RNA binding motif, single stranded interacting protein 3 (RBMS3) gene. Although RBMS3 is a member of the MYC gene single strand-binding protein family which are believed to be involved in transcription regulation, RBMS3 has been shown to be involved in RNA metabolism (Penkov, et al. 2000). Further functional studies identifying the role, if any, and mechanism of RBMS3 in oral tumourigenesis may clarify the function of this protein. Region 5 at 3p26.1 (RP11-77105 to RP11-7M24) is ~0.8 Mbp and also contains only 1 known gene, a metabotropic glutamate receptor, GRM7. Metabotropic glutamate receptors belong to the G protein-coupled receptor superfamily. The exact function of GRM7 is unknown, however, it has been shown to be involved in the inhibition of cyclic AMP signalling (Schulz, et al. 2002). Although loss of GRM7 has not been reported in oral cancer, methylation of the GRM7 promoter has been reported in chronic lymphocytic leukaemia (Rush, et al. 2004), supporting a putative role of the gene in oral tumourigenesis. Interestingly, there was a recurring minimal region of copy number gain (region 2) at 3p21.3 (RP11-447D11 to RP11-509121) that was -0.7 Mbp in size containing 8 known genes, 2 of which are putative oncogenes: Testes-Specific Protease 50 (TSP50), 45 and Teratocarcinoma-Derived Growth Factor 1 (TDGF-1). TSP50 has been shown to have protease-like activity with expression restricted to the testes (Yuan, et al. 1999). In breast cancer a hypo methylated fragment containing TSP50 was isolated, while in testes the same gene was also found to be demethylated but not methylated in other non-malignant tissue, suggesting that hypomethylation of the promoter results in TSP50 expression (Yuan, et al. 1999). The implication of this gene in spermatogenesis and breast cancer suggests a possible role for this gene in the cell-cycle, supporting its possible role in oral tumourigenesis. TDGF-1 is a glycoprotein sharing structural similarities with epidermal growth factor (EGF) and has been shown to be over-expressed in a variety of cancers including embryonic tumours (Baldassarre, et al. 1997) and colorectal cancer (Baldassarre, et al. 2001). The structural similarity shared between TDGF-1 and EGF suggests the role of this protein in cellular proliferation and tumourigenesis. As mentioned, the large size of region 1 on 3p made it difficult to identify candidate genes despite the presence of the FHIT region within it. 3.2 Summary of Whole Genome SMRT Array Results The successful identification and characterization of recurring MRA on chromosome 3p using the 3p-arm specific array validated the use of aCGH for the analysis of archival OSCC specimens. The whole genome sub-megabase resolution tiling set (SMRT) array contains 32,433 BAC clones spotted in triplicate across two glass slides providing comprehensive overlapping coverage of the entire human genome. Following sample hybridization, the log2 ratio of the clone signals were plotted along their corresponding chromosomes according to their exact base pair positions using SeeGH 46 software (Chi, et al. 2004). The resulting plot appeared as a karyogram. A combination of approaches was used to analyze the data generated from the SMRT array. SeeGH software (Chi, et al. 2004) is invaluable to the analysis of data generated from pathological samples because of the many tools it offers. The entire data set can be viewed at the whole genome level for the analysis of gross genetic alterations (Figure 11 A). However, the data can also be further resolved by "zooming" into individual chromosomes at the whole chromosome view (Figure 1 IB) and at the regional view (Figure 1 IC). Finally, and perhaps the most powerful tool utilized in SeeGH, is the alignment of specific chromosome and chromosomal region data of multiple specimens or experiments to visually analyze for recurring MRA (Figure 1 ID). This enabled the identification of common regions of gain and loss and the minimal breakpoints involved. 3.2.1 Validation of the Whole Genome SMRT Array In order to test the sensitivity of the whole genome SMRT array to detecting single copy changes non-archival normal female diploid DNA was hybridized against non-archival normal male diploid DNA. Closer analysis of the X and Y profiles using the "zoom-in" tool in SeeGH as described above shows that single copy number differences are easily detectable on sex chromosomes, demonstrating the ability of the array to detect single copy number changes. Finally, to test the reproducibility of the array, the lung cancer cell line H526 was hybridized against normal male diploid DNA in numerous independent experiments and the data plotted into karyograms as described above with SeeGH software (shown in Figure 7A above). Alignment of the 47 ' Fttqu-rc* Wot' HenDMtoFtaquwKyl i M f * Pttimwr **+*wt% U n l n n C4C«Oar«t |5~" j j T 9 o . T r . M J l n . f ~ » j t f * C t o - * t * » > fso SHAChl»|o~" » f t Otf^JrT" [ f l * * T ] 13 M f 9 J - v § I • t 5 V I I i I f Mil i t i t i t i t s * fttfoCtirikfe j SP G m * Into i Cfcr-jN«n* M l .1 ** r Accmo.Hvj-1 B 1 I dtMhDni gmta * w i jo OS - HI "i UnqutQ 1 I jP<-*0rt [MPS EL » 3 ' -\ 1 otarrn - th SNR IN* t u n B j *<« M b 1 SNR Chi SNR C M •Ma »«i S t r d w d O i x ^ w firj [io » » J *IHJ -.V. 1 SWO.1 RtBOVAdh ^ j - AJ Bj f B l < \ SUB CM ihoHRaholrat. r- g$ li S? 10 r V5 — 1 r zo mm 3B - y. ! ShgwtrwthoHLrttt — M H It* M B J 1 1 CWStMCr.FltQ | Figure 11: Use of SeeGH software to analyze SMRT array data. At the whole genome view (A) each chromosome is aligned with a chromosome profile on the left in which the log2 ratio of the clones are plotted. Within the array profile, the central blue line represents a log2 ratio of 0. Clones lying to the right and left of 0 show copy number increase and decrease, respectively. At the high resolution chromosome view (B), the clarity o f copy number changes is improved. The red bar on the right indicates a log2 ratio of+1 while the green bar on the left indicates a log2 ratio o f -1. 48 Figure 11 (cont.): Use of SeeGH software to analyze SMRT array data. The program also enables the visual analysis o f whole genome profiles at the regional view (C). Band names are listed between the chromosome drawing and the profiles. Genes are listed to the left of the chromosome drawings. Clicking on a given gene with the mouse enables the determination of the gene name. Multiple alignment o f regions in SeeGH (D) enables the identification o f recurring M R A and the fine-mapping of their minimal boundaries. 40 chromosomal regions as described above allows the careful comparison of the data (Figure 12). While signal ratios commonly varied between slides (due to numerous variables such as slide chemistry, dye labelling efficiency, and washing variability), the alterations were, nonetheless, easily visually identifiable and the consistency of these alterations was 100%. To test the robustness of the whole genome array, the profiles were compared for a tumour analyzed using both the 3p-arm specific array, which consists of 535 BAC clones spanning 3pl2.3-3p26.3, and the SMRT array, which contains a different tiling set of 957 BAC clones spanning 3pl 1.1-the 3p26.3. The 3p profiles generated using each array were visually compared. Although the two arrays have only 104 clones in common, as shown in Figure 13, the profiles are virtually identical. While the exact start and end base pair positions of the whole genome array clones are known and thus plotted accordingly, the base pair positions of the all the 3p specific array clones are not known and are thus plotted in relative order. Exact alignment of the profiles, therefore, is imperfect. Nevertheless, the reproducibility of the data is striking supporting both the robustness of the array CGH technology employed for this thesis and the regions identified. 3.2.2 Detection of Genetic Alterations Using the SMRT Array In order to identify and characterize alterations in OSCC, twenty formalin-fixed OSCC samples were hybridized against non-archival normal male diploid DNA on the SMRT array. Due to a lack of sufficient DNA material only three tumour specimens 50 Figure 12: Validation of SMRT array by multiple alignment and comparison of H526 profiles in SeeGH. Alignment of the data from a region on 3q generated in three independent experiments clearly demonstrates the reproducibility o f S M R T array. The two red bars to the right of 0 in each profile represent a log2 ratio of +0.5 and +1, respectively, while the two green bars to the left of 0 in each profile represent a log2 ratio o f -0.5 and -1, respectively. 51 Tel . o Q. a re OQ Cent. 3p A rm of SMRT Array Tel . l l P s J o 2 Cent. 1 I ft 3p Array Figure 13: Comparison of 3p profiles generated with 3p-arm specific array and the SMRT array. Exact alignment o f profdes was not possible because, unlike the S M R T array clones, the exact base-pair positions o f the 3p-arm specific array clones are unknown and are therefore plotted in relative clone order as apposed to exact base pair position. Nevertheless, the profiles appear to be virtually identical. This demonstrates the robustness and validity o f the whole genome S M R T array. Red shaded areas represent common regions o f copy number increase, while green shaded areas represent common regions of copy number decrease. 52 analyzed with the 3p-array were also analyzed using the SMRT array. The demographics of the cases analyzed with the SMRT array are listed in Appendix 3. For each sample, the log2 ratio for each clone was plotted along the chromosomes according to their exact base pair positions using SeeGH software. The SeeGH karyograms for each of the 20 tumours and the matched normal samples are listed in Appendices 4-27. An example of one such profile, 125T, is shown in Figure 14. At the gross whole genome view, large alterations such as whole arm gains and losses are easily identifiable and the karyogram exhibits numerous large alterations, many of which have been previously reported. A case in point, 125T shows whole arm loss of chromosome 3p, in addition to previously reported losses of chromosomes 4, 5q, 8p, 9, and regions of loss on 1 lp and 1 lq, 17p, 18 and 21. Tumour 125T also shows previously reported gains. Specifically, gain of the well known 8q arm (MYC), gain on 1 lq at the cyclin DI locus, and gains of 5p (hTERT) and 7p (EGFR). However, even at this view we see a novel copy number amplification at 1 lq22.2-22.3 containing a matrix metalloproteinase cluster. This novel region is discussed further in section 3.2.5.1. 3.2.3 Identification of "Hot Spots" Using Frequency Plot Analysis To facilitate the visualization of "hot spots" - regions of frequent gain or loss - a frequency plot of the genetic alterations at each loci among the 20 OSCC samples was constructed (Figure 15). Generation of this frequency plot required the use of aCGH-Smooth software (Jong, et al. 2004) to identify genomic imbalances and their associated breakpoints and SeeGH to plot the data. aCGH-Smooth employs an algorithm that calculates a maximum likelihood for each clone by calculating the probability that a 53 6 8 9 16 10 17 11 18 0QND1 gain 114133 gdn 11qZ2J-22J 12 19 20 13 14 15 i I il I § 21 22 Figure 14: Representative whole genome comparative genomic hybridization (CGH) profde of an OSCC tumour (125T). At the whole genome view gross alterations such as whole arm gains and losses are easily visible as well as high copy number amplifications. This karyogram exhibits many known alterations such as a whole arm loss of 3p, 4, 5q, 8p, and 9. Also visible are gains on 7p (EGFR), 8q arm (MYC), and 1 l q near the cyclin DI locus. Interestingly, there is also a high copy number amplification on 1 l q discussed later. For each chromosome, the central blue line represents a log2 ratio of 0. The red bar to the right of 0 in each profile represents a log2 ratio of+1, while the green bar to the left o f 0 in each profile represents a log+2 ratio of -1. 54 SjjW" ate ' «^ — • 11' • • I I ' - ( III I CM 0 ) r l l i i l M U l T I r - dDMDfo) cBanx £ '^ nm^^^n jimii^ww ^ P I ^ p i ^ M p - ^pmt-Figure 15: Frequency plot of all copy number gains and losses present in the 20 oral samples analysed in this study. Copy number increases and decreases were grossly identified using CGH-Smooth software (explanation in Figure 17). The frequency o f gain and loss at each clone was then tabulated and plotted along the chromosomes in SeeGH. Areas o f red represent copy number increase while areas o f green represent copy number decrease. The blue bars to the left and right of the chromosome indicate a frequency o f 100% (+1) gain and 100% (-1) loss, respectively. 55 given clone is of the same copy number status as a set of previous clones. This enables the identification of putative breakpoints that are then shifted randomly upstream and downstream to calculate an overall fitness. Thus, a group of clones considered to have the same copy number status will be assigned a relative value. When this data is again plotted using SeeGH the data appears to be "smooth" in comparison to the "raw" data (Figure 16). As shown in Figure 15, the frequency of alterations is non-random. Some chromosomes are mostly gained and some are mostly lost. Regions that show an equal frequency of gain and loss are likely alterations resulting from background genomic instability. A variety of "hot spots," however, are evident. Many of these alterations are known. For example, chromosome 3p appears to be the most frequently lost region followed by loss of 4, 5q, 8p, 11, and 18q. Interestingly, loss of 9p, which is thought to be the earliest event in oral tumourigenesis, was evident in only half of the cases. The most frequently gained regions appear to be 8q followed by 3q, and smaller regions on 9q and 1 lq (CCND1). Most noteworthy, at this resolution the frequency plot identifies common alterations that may be missed by conventional CGH. One such example is the telomeric end of 5p containing hTERT which is gained in -50% of cases. Conventionally, loss of chromosome 9p is believed to be the first region lost in oral cancer progression (Figure 3), which has previously been attributed to loss of pi 6, an inhibitor of the cell cycle (Califano, et al. 1996; Wu, et al. 1999). Loss of 3p and 17p is then incurred, and subsequently, loss of 1 lq, 13q, and 14q. The final progression of CIS 56 Multiple Alignment (or Chromosome 3 Ratio Controls | BP and GenejLI r Show Standard Dev Fiag al clones with standard deviation- greater than Hide al clones win standard deviations greater rhart 0 075 Hide clones mth SNR lets than SNR CM SNRCh2 |10 |10 15 Enlaige moused ovet regnris Ratio W i d h Show Ratio Lines: |™ 0,5 Iv 1.0 r 1.5 r to Posrtrve Rano l i a t ! • dRed T | Negative Ratio Unw "3 r ShowThresWdUies Retiesh linage || J 903 t I - 3 1 I 523T_sltdesmerged 1 9 • U K 528T smoottiforalianment Figure 16: Alignment and comparison of the "raw" or non-manipulated data from 528T with the data manipulated in aCGH-smooth. aCGH-Smooth grossly identifies breakpoints and copy number gains and losses using a genetic local search algorithm. The algorithm calculates a maximum likelihood for each clone by determining the probability that a given clone is of the same copy number status as a set of previous clones. This enables the identification of putative breakpoints, which are then shifted randomly upstream and downstream to calculate an overall fitness of the breakpoint. Clones considered to have the same copy number status are assigned a relative value. As described by the name of the software, the manipulated data points appear "smoothened." 57 to invasive carcinoma is thought to be paralleled by the final loss of 6p, 8, and 4q. However, according to the frequency plot, 9p loss is less frequent than 3p perhaps suggesting that the opposite is true. In addition, 17p shows a much lower frequency of loss than 3p. Interestingly, loss of 1 lq was more frequent in these samples than loss of 17p. Finally, loss of 8p and 4q appears far more frequent than loss of 13q, 14q, 17p or 9p, which are thought to come before this. However, one must consider that array C G H detects copy number gain and loss - only one mechanism of gene over-expression and tumour-suppressor inactivation. In addition to copy number loss, genes may be interrupted or silenced via other mechanisms both genetic, such as a single base pair mutation or a frame-shift mutation, and epigenetic, such as methylation - all of which are detectible, but not necessarily distinguishable using L O H , the technique upon which the current progression model was based. Additionally, according to Figure 16, gain of 14q is more frequent than loss. Once again, since L O H is unable to distinguish between copy number gain and loss, this may not be contradictory to the accepted progression model. Interestingly, the frequency plot also shows a greater frequency of whole arm loss than whole arm gain overall. While most frequent losses appear to affect entire chromosome arms, frequently gained regions tend to involve minimal regions within an arm (Figure 15). For example, chromosome arms 3p, 4q, 5q, 8p, and 11 show what appears to be frequent whole arm loss. However, while 5p and 8q show what might appear to be whole arm gain, there are regions of much higher frequency creating "peaks." Similarly, 3q, 7q, 9q, 1 lq, 12p, 13q, and 14q show smaller "peaks" of gain 58 within arms showing no or little whole arm loss. Further analysis of these "peaks" may reveal the minimal region of gain. 3.2.4 Array CGH representation of aneuploidy One possible explanation for the over-representation of whole arm loss compared to whole arm gain in the frequency plot is that the high frequency of whole arm loss is indicative of aneuploidy, or, having more or less than the normal diploid number of chromosomes, characteristic of most solid tumour cells. Aneuploidy has been theorized to result from a tetraploid intermediate early in tumourigenesis. Indeed, near tetraploidy karyotypes is common in epithelial tumours (Mitelman, et al. 2003). This results in unstable subsequent mitosis leading to a loss of chromosomal material, and finally, aneuploidy (Reith and Sudbo 2002). This may perpetuate in further genomic instability producing sub-populations of malignant cells within a single tumour. Although by definition aneuploidy can be characterized by an excess of chromosomes, this may not be represented in aCGH profiles. Rather, since equal quantities of tumour and normal DNA are compared, not equal quantities of cells compared, the aCGH karyogram may show an overall loss for a given chromosome. Hence, unlike minimal copy number gains encompassing putative oncogenes that are selectively amplified (abnormal mitosis leads to loss of chromosomes, not gain of specific chromosome regions), minimal regions of copy number loss are not necessarily selected for because loss of an entire chromosome or chromosome arm is sufficient for the biological effect. Thus, in the frequency plot we see a high frequency of whole arm or whole chromosome loss compared to gain. 59 3.2.5 Characterization of Known Minimal Regions of Alterations A closer look at the frequency plot in SeeGH allowed the assessment of known regions such as 7pl 1.2 (EGFR) (Figure 17A) and 1 lql3.3 (CCND1) (Figure 19A). However, while the frequency plot is useful in identifying "hot spots," one must employ alignment of these chromosomal regions in SeeGH from all 20 OSCC profiled. This approach enables the identification of common regions as well as the definition of the minimal breakpoints with sub-megabase resolution. Several recurring minimal regions of alteration previously associated with OSCC were characterized and their breakpoints defined. These regions and the genes within them are listed in Table 2. Table 2: Known minimal regions of alteration identified in 20 OSCC using whole genome array CGH. Size Region Gain/Loss Frequency Start bp End bp (Mbp) Genes 7p11.2 8p23.2 Gain Loss 8/20 (40%) 13/20 (65%) 54532642 (RP11-16407) 3016057 (CTD-2042C19) 55539345 (RP11-535N12) 3207129 (RP11-234C5) 1.06 0.19 5 (EGFR) 1 (CSMD1) 8p23.2 11q13.3 Loss Gain 13/20 (65%) 9/20 (45%) 3868230 (RP11-567H20) 68951045 (RP11-243A15) 4094513 (RP11-315I17) 69742338 (CTD-2192B11) 0.23 0.79 1 (CSMD1) 8 (CCND1) The small peak in the frequency plot at 7pl 1.2 (Figure 17A) suggested the presence of a recurrent region of gain at the EGFR locus. Further alignment of the region in all 20 OSCC allowed the identification of a 1.06 Mbp minimal region of copy number gain present in 8/20 cases. Of these 8 cases, 3 were high copy number amplifications. 60 -1 -0 5 +0 5+1 -1 -0 5 +0 5+1 t 1' • ' 1 1 1 1 .'l 1 • H i i t 1 i i'. i 1 -0 5 +0 5+1 1 -0.5 +0.5+1 -1 -0 5 +0 5* I • 1 1 1 • K • 11 )' ' 1 a: ' CD UJ 161T 809T 486T 793T 528T Figure 17: Identification and characterization of a known gain at 7 pi 1.2. This region harbours epidermal growth factor receptor (EGFR) which is well known to be over-expressed in a variety of cancers including O S C C . In this study we define the minimal regions of gain at this locus. (A) This frequent copy number gain is evident in the frequency plot (Figure 16). (B) Alignment of this region in 5 tumours using SeeGH. Shaded red regions represent copy number increase and shaded green regions represent copy number decrease. 61 Figure 17B illustrates the alignment of 5 cases (161T, 809T, 486T, 793T, 528T) and the minimal region identified. The minimal region runs from bp 54532642 (RP11-16407) to bp 55539345 (RP11-535N12). Based on the base pair positions of the bordering clones in SeeGH the genes encompassed by this region were identified using the April 2003 version of the UCSC human genome browser (Figure 18). This region contains 5 known genes SEC61G, EGFR, LANCL2, DKFZP564K0822 and FKBP9. EGFR is well known to be associated with a variety of cancers (Clarke, et al. 2003; Dobashi, et al. 2004; Lager, et al. 1994; Tobita, et al. 2003) including head and neck (Rikimaru, et al. 1992; Ulanovski, et al. 2004)and oral squamous cell carcinoma (Chen, et al. 2004; Garnis, et al. 2004b; Kiyota, et al. 2002). Conventional metaphase CGH analyses have indicated high copy number gains on the 7p arm (Jin C 2000), while expression studies have reported EGFR to be over-expressed in oral tumours (Ford and Grandis 2003; Freier, et al. 2003; Grandis and Tweardy 1993a; Grandis and Tweardy 1993b; Santini, et al. 1991). The peak in the frequency plot at 1 lql3.3, the cyclin DI locus, demonstrates a high frequency of gain (Figure 19A). SeeGH alignment of all 20 OSCC at this region identified the 0.79 Mbp minimal region of gain and its boundaries. This gain was present in 9/20 cases with high copy number amplifications in 7/20 cases. The minimal region borders from bp 68951045 (RP11-243A15) to bp 69742338 (CTD-2192B11) and encompasses 8 known genes including IGHMBP2, MGC21621, TPCN2, TPC2, MYEOV, CCDNE and ORAOV. Alignment of 5 cases (125T, 669T, 486T, 542T, 800T) and the minimal region is shown in Figure 19B. Cyclin DI is well known to be associated with a 62 U l (T> 3 5 ' S ^ . oo -oo ca rt 5 n 2 a. 3 oo P . - : ' era _, 3 ° TO <£. OO ? 5 ' a- 3 rt (y, 3 ^ 3- 3-a a> if o C7Q O O 0> 3 oo P o •a g S « rt. „ cro era P z OB r t O re n n = = 3 a n -e = re CO e re - . C rt a rs 3 re or F 3 Human chr7:54,532,642 55,539.345 UCSC Genome Browser v102 Microsoft Internet Explorer . - X File Et* View Favorites Tools Help Address i £ ) http:^ntjtrra.ucsc.edu/ct^fain/hgrracks Gene Sorter Convert UCSC Genome Browser on Human April 2003 Assembly move l < < <! 0 0 0 0 [ E L I zoom «i r^]f^lWl b o s e 1 zoom out [T^[lx~[fiOx] position chr7 54.532.642-55.539.345 ] [ jump |[ dear | ^  1,006,704 bp. I configure Bus* Poafrion SEC61G | co rn M P I 55 • » * • * • « ) B B l M M B l S S 2 M * « » I CrtroMSOM* Bonds L o c a l i z e d by FISH N i p p i n g C l o n e s 7 p n . a 5B3*M4«I 554t*tJ l . l l l l . 1 I I l i ++4-*i Click on a feature for details Click on base posthon to zoom m around cursor. Click on left mini-buttons for track-specific options 0 2 0 i delault tracks [ [ hide all 11 configure""] [ relresh | Use drop down controls below and press refresh to alter tracks displayed Tracks with lots of items will automatically be displayed in more compact modes. Mappiii g and Sequencing Tracks Base Position Chromosome Band full v | U | | STS Markers hide r FISH Clones hide * ?.ecomb Rate hide <» i Internet g 5 pa B' 1 F i g u r e 19: I den t i f i c a t i on a n d c h a r a c t e r i z a t i o n o f a k n o w n g a i n at l lql3.3. This region harbours cyclin DI (CCND1) which is well known to be over-expressed in a variety of cancers including O S C C . In this study we define the minimal regions of gain at this locus. (A) This frequent copy number gain is evident in the frequency plot (Figure 16). (B) Alignment of this region in 5 tumours using SeeGH. Shaded red regions represent copy number increase and shaded green regions represent copy number decrease. 64 variety of cancers (Hedberg, et al. 2003; Janssen, et al. 2000; Specht, et al. 2004; Wong, et al. 2001) and has been previously demonstrated to be over-expressed in OSCC (Carlos de Vicente, et al. 2002; Miyamoto, et al. 2003; Nishimoto, et al. 2004; Papadimitrakopoulou, et al. 2001; Shintani, et al. 2001). Although not highlighted in the frequency plot, multiple alignment of chromosome 8p data showed 2 minimal regions of copy number loss at 8p23.2 (Figure 20). Both minimal regions of loss were present in 13/20 cases. The first region runs from bp 30146057 (M2042C19) to bp 3207129 (RP11-234C05) while the second region runs from bp 3868230 (RP11-567H20) to bp 4094513 (RP11-315117). Referring to the April 2003 version of the UCSC physical map of the human genome revealed that a single large gene, CUB Sushi Domain 1 (CSMD1), spans both regions and is, therefore, interrupted by the copy number loss. Interestingly, 3/20 cases show a narrow copy number decrease that appears within an apparent whole arm loss, suggesting that loss of CSMD1 may represent a homozygous loss in these cases. CSMD1 encodes a 11.5 kb transcript that translates into a transmembrane protein (Sun, et al. 2001). CSMD1 has been previously identified as lost in head and neck squamous cell carcinoma, often by homozygous deletion (Toomes, et al. 2003). Indeed, loss on 8p23 has been previously reported based on LOH (El-Naggar, et al. 1998; Ono, et al. 1999; Sun, et al. 2001; Wu, et al. 1997) and FISH (Ishwad, et al. 1999). Interestingly, this region of loss is not evident in the frequency plot in Figure 16. This may be explained by the fact that aCGH-Smooth often misses alterations that, although visually detectable, do not pass the statistical test applied by the program's algorithm. The reduced ratios may be explained by the 65 -1 -0.5 +0.5 -1 -0.5 +0.5 +1 669T > • 1 1 M • • i i f \ • i i . ' 125T -1 -0.5 +0.5 +1 4-528T •1 -0.5 +0.5 +1 117T -1 -0.5 +0.5 +1 I. 814T Figure 20: Multiple alignment 5 cases showing a recurring loss at 8p23.2 containing CSMD1. Multiple alignment of the chromosome arm 8p data in SeeGH revealed 2 minimal regions o f copy number loss present in 13/20 cases, both of which interrupt the known CUB sushi domain-l (CSMD1) gene. CSMD1 has been previously shown to be lost homozygously in oral cancer (Scholnick and Richter 2003; Toomes, et al. 2003). This region of loss is not evident in the frequency plot (Figure 15) due to the insensitivity o f the software to detecting additional small regions of double copy allele loss within a single copy whole arm loss. Shaded green regions represent copy number decrease, with darker shaded green areas representing an addition level o f decrease. 6 6 presence of normal tissue contamination or the presence of subpopulations of tumours cells not exhibiting this change, in contrast to cell lines which consist of pure populations of genetically identical cells. It is for this reason, that all SeeGH profiles were analyzed visually rather than relying solely on this software. Interestingly, although a high frequency of 3p loss was evident in the frequency plot, further analysis of the 3p arm by multiple alignment of the 3p profiles in SeeGH did not reveal any minimal regions, let alone the same minimal regions detected using the 3p-arm specific array. The vast majority of cases simply showed whole arm or near whole arm loss. Only one case, 528T, showed segmental copy number changes. Incidentally, this case was the only case analyzed with both the 3p-arm specific array and the whole genome array. While initially this may appear to discredit the minimal regions identified with the 3p array, one must also take into account the sample selection process. The samples analysed using the 3p array were selected based on LOH results showing minimal regions of loss on 3p. That is, at least one marker on the 3p arm had to show retention of heterozygosity. In contrast, the samples analyzed using the whole genome SMRT array where chosen based on the amount of DNA available. Since the minimal segmental copy number changes on 3p for 528T appeared identical using both the 3p-arm specific array and the whole genome SMRT array, we can be confident that this observation can be attributed to the small sample size of 20, rather than an artefact of either the 3p array or the whole genome array. Analysis of a larger sample size would certainly reveal the minimal regions identified on chromosome 3p using the 3p array. 67 Pooling the 3p-array and SMRT array data, we observed a high frequency of 3p-arm loss at 92%. 3.2.6 Detection of putative polymorphisms While we were able to detect known copy number alterations relative to the reference DNA using the whole genome SMRT array, some of the differences may be attributed to polymorphisms in the human population. This was assessed by visual comparison of the highly frequent regions of gain or loss that appeared characteristically similar in both the magnitude and size of the change with a matched normal sample. One such example is the loss shown in Figure 21. This loss was present in 15 of 20 cases with each case showing a similar size of the region. Note that the size of the alteration is identical in both tumours. Comparison of two cases showing this loss with their matched normal karyograms shows that this loss is present in the matched normal DNA as well. This suggests that the region is polymorphic. The SMRT arrays sensitivity to detecting polymorphisms is exciting because it will aid in the identification and mapping of polymorphic regions - an interesting and exciting future venture in its own right. However, in the meantime, because the array detects polymorphisms and because these polymorphic regions have not yet been mapped and logged, it is necessary to compare our minimal regions of alteration with matched normal DNA in order to assess whether the novel regions we identify are real. Whenever possible, frequent minimal regions of gain or loss present in a given sample were compared with the corresponding matched normal. However, unfortunately, not all cases used in this analysis were accompanied by matched normal DNA and, therefore, this was not always possible - particularly for the 68 Figure 21: Detection of polymorphisms using the SMRT array. Alignment o f a region on chromosome band 9ql3 showing a minimal region o f copy number decrease in both tumour and matched normal D N A . The identification of this minimal region of apparent loss in both the tumour and the normal D N A suggests that the region represents a polymorphism present in the pooled normal reference D N A , but absent in this particular patient. Since polymorphisms are copy number alterations present in at least 1% of the population, only further assessment of this region in a larger random cohort of individuals wi l l determine whether this region is, in fact, a true polymorphism. Whenever possible suspected minimal regions of alteration present in the tumours were compared with the matched normal D N A to avoid false positives. 69 novel regions of loss identified (see section 3.2.7.2). However, the minimal regions presented here frequently comprise larger regions of change which altered from case to case. In contrast, the size of polymorphisms does not tend to vary from case to case. Hence we are confident that the losses described are not polymorphisms and are, in fact, somatic alterations. 3.2.7 Identification of Novel Regions ofAlteration While characterization of known regions detected in OSCC using the SMRT array is exciting, the most noteworthy achievements were the detection and characterization of novel regions of alteration. Many of these novel regions are sub-megabase in size, often containing a single gene, and thus, may have been missed by previous studies employing conventional, lower resolution techniques such as metaphase CGH and LOH analysis. Both novel gains and losses were detected, fine-mapped, and the candidate genes identified using the whole genome SMRT array (Table 3). 3.2.7.1 Novel Copy Number Gains Interestingly, five novel sub-megabase minimal recurring regions of copy number increase were identified, their boundaries defined, and the candidate genes identified. While some of these regions are exhibited in the frequency plot (Figure 15), some were not. Each of these regions harbour either putative oncogenes or genes already associated with cancer. Most were highly recurrent occurring in a minimum of 6 (30%) cases (the amplification at 1 lq22.2-22.3 being the exception). Each of these regions were also analyzed in matched normal specimens to rule out the possibility that they may be 70 Table 3: Summary of novel minimal regions of alteration detected in 20 OSCC using whole genome array CGH. Region Gain/Loss Frequency Start bp End bp Size (Mbp) Genes 2pl5 3q23 4q34.3 5pl5.2 7p 12.3-13 7q21.2 7q35 llq22.2-22.3 16q23.2 Loss Gain Loss Gain Gain Gain Gain Gain Loss 15/20 (75%) 7/20 (35%) 14/20 (70%) 9/20 (45%) 12/20 (60%) 6/20 (30%) 8/20 (40%) 3/20 (15%) 7/20 (35%) 61407007 (RP11-803E10) 142353819 (RP11-507019) 180186394 (RP11696K24) 14059583 (RP11-744A15) 46947366 (RP11-338N5) 91736598 (RP11-82E23) 143199102 (RP11-466J6) 102097765 (CTD-2003023) 80907943 (RP11-806J24) 61606973 (RP11-803E10) 0.2 4 (USP34) 142536099 (RP11-507019) 0.44 4 (RASA2) 180630413 (RP11-631C12) 0.44 no genes 14641689 (RP11-611H04) 0.58 3 (TRIO) 47204328 (RP11-338N5) 92058838 (RP11-514K1) 143523590 (RP11-714J20) 0.25 0.32 0.32 3 (TEM6, TENS!) 4 (PEX1, CDK6) 4 (ARHGEF5, TPK1) 103342179 (RP11-57016) 1.24 16(MMP cluster) 81058473 (RP11-806J24) .0.15 5 (PKD1L2, BCDOl) 71 polymorphisms. Most noteworthy, this is the first implication of these genes in oral tumorigenesis. The frequency plot in Figure 15 in section 3.2.3 illustrates a high frequency of copy number gain on 3q. Multiple alignment of chromosome 3q in all 20 OSCC revealed a minimal region of gain at 3q23. This minimal region is 0.44 Mbp in size running from bp 142353819 - bp 142536099 (RPT1-507019). The copy number gain is present in 7/20 cases. Figure 22 shows the alignment of 3 cases showing the gain (16IT, 628T, 486T) and the corresponding unchanged region in one matched normal (486C). This region contains 4 genes, including 3 hypothetical genes (AK092355, BC008901, AK055693) and 1 known gene, RASp21 Protein Activator 2 (RASA2), which is interrupted at the telomeric end of the region. RASA2 is a GTPase activating protein (GAP1M) which functions to inactivate RAS by catalyzing the hydrolysis of GTP to GDP (Aspenstrom 2004). Interruption of this gene may result in over-activity of RAS activating downstream pathways. Gains on 3q have been previously reported in a variety of cancers including cervical and vulvar (Huang, et al. 2005), liver (Sy, et al. 2004), neuroendocrine (Welborn, et al. 2004), leukaemia (Casas, et al. 2004), renal (Hoglund, et al. 2004), gastric, lung (Hoglund, et al. 2004; Ashman, et al. 2002; Chujo, et al. 2002), head and neck (Gotte, et al. 2005; Oga, et al. 2001; Squire, et al. 2002), and, specifically, oral cancer (Hannen, et al. 2004; Hermsen, et al. 2001; Hermsen, et al. 1997; Huang, et al. 2002; Jin C 2000; Komiyama, et al. 1997; Okafuji, et al. 1999; Okafuji, et al. 2000; Sreekantaiah, et al. 1994; Weber RG 1998; Wolff, et al. 1998). However this is the first report of a fine-mapped minimal region of gain on 3q in oral cancer. Although the 72 •1 OS +0.5 +1 S i fM r 161T 4 1 1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 1° 628T2 I ,i .:,< 486T 1 -0.5 +0.5 +! 1 r l 486C Figure 22: Alignment of novel alteration at 3q23 among tumours and a matched normal. Shown here is the alignment of 3 tumours (161T, 628T, 486T) and one matched normal (486C) in SeeGH. For each tumour profile, the central purple line represents a log2 ratio o f 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right, and the two green lines to left, represent log2 ratios of +0.5 and +1, and -0.5 and -1, respectively. Red shaded regions mark areas o f copy number increase. The approximate position o f RASA 2 is drawn in purple along the third tumour profile before the matched normal. As shown here the minimal region of gain is marked by 486T. 73 amplification interrupts a known putative tumour suppressor gene, the recurrence of this sub-megabase gain encompassing only hypothetical genes in multiple tumours suggest the presence of novel oncogene on 3q23. Chromosome 5p also exhibited a high frequency copy number gain based on the frequency plot (Figure 15). Subsequent alignment of the 5p arm revealed a 0.58 Mbp minimal region of copy number gain at 5pl5.2, which was present in 9/20 cases. The region runs from bp 14059583 (RP11-744A15) - bp 14641689 (RP11-611H04). Figure 23 shows the alignment of 3 cases showing the gain (123T, 125T, 486T) and the corresponding unchanged region in one matched normal (486C). Interestingly, the matched normal does show a small peak, though of much smaller magnitude than the tumour. This can be attributed to the absence of this sequence in the normal reference DNA, due to polymorphism. This region contains 3 genes including 2 hypothetical genes (AL390214, FIJI 1127) and I known gene, Triple Functional Domain (TRIO). TRIO contains three functional domains - a serme/threonine kinase domain and two guanine nucleotide exchange factor (GEF) domains specific for Racl and RhoA families of Ras (Debant, et al. 1996). The serine/threonine kinase domain suggests kinase activity while the plekstrin homology domains adjacent to the GEF domains, similar to that of other signal transduction molecules, may play a role in protein-protein interactions and membrane sequestration (Debant, et al. 1996). Amplifications on 5p have been reported in a variety of cancers including gastric (Takada, et al. 2005), neuroendocrine (Welborn, et al. 2004), cervical and vulvar (Huang, et al. 2005), rectal adenocarcinoma (Verbeek, et al. 2004), and head and neck cancer (Ashman, et al. 2003). Indeed, a micro-74 -1 -0.5 +0.5 +1 •1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 ll, 125T 161T k 486T 486C Figure 23: normal. Alignment of novel alteration at 5pl5.2 among tumours and a matched Shown here is the alignment of 3 tumours (125T, 161T, 486T) and one matched normal (486C) in SeeGH. For each tumour profile, the central purple line represents a l o g 2 ratio o f 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right and the two green lines to left represent l o g 2 ratios o f +0.5 and +1, and -0.5 and -1, respectively. Red shaded regions mark areas o f copy number increase. The approximate position o f TRIO, is drawn in purple along the third tumour profile before the matched normal. As shown here the minimal region of gain is marked by 486T and 161T. 75 amplification encompassing TRIO detected using array CGH has been reported in small cell lung cancer cell lines (Coe, et al. 2005). Although gains at 5p have been detect in oral cancer (Hannen, et al. 2004; Huang, et al. 2002; Komiyama, et al. 1997; Oga, et al. 2001; Okafuji, et al. 1999; Okafuji, et al. 2000; Sreekantaiah, et al. 1994) (Hermsen, et al. 2001; Hermsen, et al. 1997), this is the first report of a fine-mapped minimal region of gain at 5pl5.2. According to the frequency plot (Figure 15), the 7p chromosome arm does not appear to show frequent gain. However, a novel 0.25 Mbp gain at 7pl2.3-13 was, nonetheless, evident in 12/20 cases using multiple alignment of the karyograms in SeeGH. This region runs from bp 46947366 to bp 4724328 (RPT1-338N05) and encompasses 3 genes including 1 hypothetical gene (AK092864) and 2 known genes, tumour endothelial marker 6 (TEM6) and its transcript variant tensin-like SH2 domain-containing 1 (TENSI). Figure 24 shows the alignment of 3 cases showing the gain (528T, 809T, 123T) and the corresponding unchanged region in one matched normal (123C). Along with TEMs 1-9, TEM6 expression has been shown to be elevated in colorectal tumour endothelial cells (St. Croix B et al 2000) using reverse transcriptase PCR analysis. However, RNA In Situ Hybridization (RISH) analysis failed to show elevated expression of TEM6. Although Carson-Walter et al (2001) showed that TEM6 contains no transmembrane domains, they were unable to locate the signal peptide and subsequently were not confident they had cloned the entire gene (Carson-Walter, et al. 2001). Although this gene would appear only to be expressed and deposited on the cellular membrane of endothelial cells, it may also be expressed and secreted by 76 Figure 24: Alignment of novel alteration at 7pl2.3-13 among tumours and a matched normal. Shown here is the alignment of 3 tumours (528T, 809T, 123T) and one matched normal (123C) in SeeGH. For each tumour profile the central purple line represents a l o g 2 ratio of 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right and the two green lines to left represent l o g 2 ratios of +0.5 and +1, and -0.5 and -1, respectively. Red shaded regions mark areas of copy number increase. The approximate position of TEM6, is drawn in purple along the third tumour profile before the matched normal. As shown here the minimal region of gain is marked by 809T. 77 epithelial cells, and in the case of cancer, malignant epithelium. The protein may function by interacting with proteins on the cell surface of endothelium during angiogenesis. This may explain the expression on the surface of endothelium in colorectal cancer and the lack of a transmembrane domain. Similar to the known region of gain on 8p23.2, which did not appear evident in the frequency plot, the aCGH-Smooth software may not be sensitive enough to the log2 ratio difference in archival tumour samples. Certainly, gains on 7p have been reported in a variety of cancers including gastric (Takada, et al. 2005), breast (Nessling, et al. 2005) (Aulmann, et al. 2005), cervical (Lyng, et al. 2004), renal (Kallio, et al. 2004), and oral cancer (Chen, et al. 2004; Hermsen, et al. 2001; Hermsen, et al. 1997; Huang, et al. 2002; Komiyama, et al. 1997). This is the first report of a minimally defined region of gain in oral cancer. The frequency plot in Figure 15 shows a significant peak in the middle of the 7q arm. Subsequent alignment of SeeGH karyograms enabled the identification of a novel 0.32 Mbp minimal gain at 7q21.2 from bp 91736598 (RP11-82E23) to bp 92058838 (RP11-514K1) which was present in 6/20 cases. This region contains 4 genes including 2 hypothetical genes (DKFZP5640523, MGC40405) and 2 known genes, Peroxisome Biogenesis Factor 1 (PEX1) and Cyclin Dependent Kinase 6 (CDK6). Figure 25 shows the alignment of 3 cases showing the gain (528T, 542T, 486T) and the corresponding unchanged region in one matched normal (486C). CDK6 has been shown to be over-expressed in a number of squamous cell carcinoma cell lines including OSCC cell lines 78 -1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 ;• • J 528T 7C 542T 486T 486C Figure 25: Alignment of novel alteration at 7q21.2 among tumours and a matched normal. Shown here is the alignment of 3 tumours (528T, 542T. 486T) and one matched normal (486C) in SeeGH. For each tumour profile the central purple line represents a l o g 2 ratio of 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right and the two green lines to left represent l o g 2 ratios of +0.5 and +1, and -0.5 and -1, respectively. Red shaded regions mark areas of copy number increase. The approximate positions o f CDK6 and PEX1, are drawn in purple along the third tumour profile before the matched normal. As shown here the minimal region of gain is marked by 542T. 79 and in endocervical adenocarcinomas (Timmermann, et al. 1997; Timmermann, et al. 1998). Inhibition of CDK6 has been shown to result in tumour regression in colon cancer (Fry, et al. 2004). Like CDK4, CDK6 binds to Cyclin D l , which was also shown to be amplified at the gene level in this study and well known to be over-expressed in oral and other cancers. The resulting cyclin D1/CDK6 complex phosphorylates retinoblastoma protein (RB) resulting in progression through the cell cycle. Increased copy numbers of the CDK6 may be one mechanism driving over-proliferation of malignant oral cells. PEX1 is an ATPase involved in peroxisomal matrix protein import. Interruption of the gene is associated with a number of peroxisomal biogenesis disorders including neonatal adrenoleukodystrophy (NALD), immune restoration disease (IRD), and Zellweger syndrome. Peroxisomes contain oxidative enzymes responsible for the elimination of toxic substances such as H2O2. Unlike CDK6, PEX1 has never been implicated in any cancer types nor does it have any obvious roles in cellular proliferation and tumourigenesis. Although the telomeric end of chromosome arm 7q shows no significant peaks of copy number increase, analysis using SeeGH multiple alignment did, nevertheless, reveal a region of recurring copy number gain at 7q35 from bp 143199102 (RP11-466J6) to bp 143523590 (RP11-714J20) which was present in 8/20 cases. This region contains 4 genes including 2 hypothetical genes (AF327904, AF338232) and 2 known genes, Rho Guanine Nucleotide Exchange Factor 5 (ARHGEF5) and Thiamine Pyrophosphokinase 1 (TPK1). Figure 26 shows the alignment of 3 cases showing the gain (125T, 123T, 486T) and the corresponding unchanged region in one matched normal (486C). ARHGEF5 is a 80 -1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 1 i 3 -1 -0.5 +0.S +1 125T 123T 486T 486C Figure 26: Alignment of novel alteration at 7q35 among tumours and a matched normal. Shown here is the alignment of 3 tumours (125T, 123T, 486T) and one matched normal (486C) in SeeGH. For each tumour profile the central purple line represents a l o g 2 ratio of 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right and the two green lines to left represent logi ratios of +0.5 and +1, and -0.5 and -1, respectively. Red shaded regions mark areas of copy number increase. Green shaded regions mark areas of copy number decrease. The approximate positions of ARHGEF5 and TPKl are drawn in purple along the third tumour profile before the matched normal. As shown here the minimal region of gain is marked by 123T. 81 member of the Dbl guanine nucleotide exchange factor (GEF) family specific for Rho GTPases, which are well known to be involved in tumorigenesis (Debily, et al. 2004). Interestingly, the ARHGEF5 gene is 0.5 Mbp centromeric of the fragile site FRA7I, a region prone to chromosomal breakage. Rho-GEFs activate Rho-GTPases by stimulating the exchange of GDP for GTP. Aberrant expression of ARHGEF5, may result in constitutive activation of this signalling pathway. While over-expression of Rho GTPases has been reported in carcinogenesis, this is the first report of an amplification of a GEF in oral cancer. TPK1 regulates thiamine (vitamin BI) metabolism by phosphorylation (Nosaka, et al. 2001). Although TPK1 has kinase activity, neither the amplification of the gene nor the over-expression has been reported in any cancers. Gains on 7q have been reported in a variety of cancers including medulloblastoma (Tong, et al. 2004b), cancers of the peripheral nervous system (Bridge, et al. 2004), uterine (Micci, et al. 2004), head and neck cancer (Huang, et al. 2002; Ng, et al. 2000), and oral cancer (Hermsen, et al. 2001; Jin and Mertens 1993; Komiyama, et al. 1997; Steinhart, et al. 2001; Wang, et al. 1998). Although LOH has been specifically reported at 7q21.2 in thyroid cancer (Trovato, et al. 2004), this is the first report of a minimal region of gain at 7q21.2 in oral cancer. An interesting novel gain on the 1 lq arm was also unapparent in the frequency plot in Figure 15. Indeed it was only present in 3/20 cases, however, 2 of these cases showed a very narrow and high copy number amplification (Figure 27). This 1.24 Mbp gain was detected at 1 lq22.2-l lq22.3 from bp 102097765 (M2003O23) to bp 103342179 82 1 -0.5 +0.5 +1 i i 1 628T •0.5 +0 I -1 -0.5 +0.5 +1 125T T T r 805T BIRC2 B1RC3 PORIMIN MMP7 MMP20 MMP27 MMP8 MMP10 MMP1 MMP3 MMP12 MMP13 Figure 27: Alignment of novel high copy number at llq22.2-22.3 containing a matrix metalloproteinase gene cluster. Shown here is the alignment of 3 tumours (628T, 125T, 805T) in SeeGH. For each tumour profile the central purple line represents a log2 ratio of 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right and the two green lines to left represent log2 ratios of +0.5 and +1, and -0.5 and - I, respectively. Red shaded regions mark areas of copy number increase. Green shaded regions mark areas of copy number decrease. In this case, the minimal region is marked at the centromeric end by 125T and at the telomeric end by 628T. The minimal region is indicated by two dotted blue lines. The genes are listed in their relative order to the right. Interestingly, although this region is present in only 3 tumours, 2 represent high copy number amplification (125T, 805T) and one represents a low copy number increase (628T). 83 (RP 11-57016) and encompasses 21 genes - 3 hypothetical (AB082528, AK095579, AK021818) and 16 known. Included in the known genes is a cluster of 9 matrix metalloproteinases (MMP1, 3, 7, 8, 10, 12, 13, 20, 27). Matrix metalloproteinases (MMP) are secreted zinc-dependent endopeptidases that function in degrading components of the extracellular matrix and basement membrane (Werner, et al. 2002). While MMP are important in many normal physiological functions such as embryogenesis, differentiation, angiogenesis, and tissue remodelling (Birkedal-Hansen, et al. 1993; Foda and Zucker 2001; John and Tuszynski 2001; Sternlicht and Werb 2001), they are also well known to be involved in tumour invasion and metastasis. Over-expression of MMP has been reported in head and neck cancer (Charous, et al. 1997; Franchi, et al. 2002), and more specifically, in oral cancer (Sutinen, et al. 1998) (Lin, et al. 2004a; Lin, et al. 2004b). Alterations on 1 lq have been reported in a variety of cancers including gastric (Takada, et al. 2005), breast (Nessling, et al. 2005), cervical (Zhang, et al. 2005), laryngeal (Rogowski M 2004), endometrial (Moinfar F 2004), ovarian (Lambros, et al. 2005), and oral cancer (Grati, et al. 2000; Komiyama, et al. 1997; Wolff, et al. 1998; Hermsen, et al. 2001; Lin, et al. 2002; Weber RG 1998). Most of these alterations have been attributed to cyclin DI at 1 lql3.3. However, gains at 1 lq22.2 containing this MMP gene cluster have been reported in ovarian (Lambros, et al. 2005) and peripheral nerve sheath cancer (Bridge, et al. 2004). This evidence suggests that additional regions of gain on 1 lq, specifically the gain at 1 lq22.2-22.3, are critical to oral tumorigenesis. 84 In addition to the gains at 1 lql3.3 and 1 lq22.2-22.3, the chromosome also shows frequent regional loss, on both 1 lp and 1 lq, which is also evident in the frequency plot (Figure 15). Multiple alignment in SeeGH revealed mostly whole arm or large regions of loss, and therefore, provided little information with which to identify candidate tumour suppressor genes. Analysis of whole genome SMRT array data generated from early lesions may shed light on this chromosome. 3.2.7.2 Novel Copy Number Losses In addition to novel recurrent copy number gains, novel recurrent copy number losses were also apparent by alignment of karyograms in SeeGH. Unlike the samples exhibiting the aforementioned novel gains, most of the samples (the exception is 486T) possessing novel minimal regions of loss were not accompanied by matched normal DNA. Hence these regions were not compared in matched normal DNA. However, unlike polymorphic regions, which do not vary significantly in the magnitude and size of the copy number difference (compared to the normal male diploid DNA), the minimal regions presented here frequently vary in the magnitude of the copy number difference and comprise larger regions of change, which vary from case to case. This provides some validation of these regions. The minimal region of loss 2pl5 was present in 15 of 20 samples. Figure 28 illustrates the SeeGH alignment of 3 cases showing this deletion (528T, 486T, 794T). This recurrent deletion was 0.20 Mbp in size, running from bp 61407007 to bp 61606973 85 (RP11-803E10). The region encompassed 4 genes - 3 hypothetical (BC022783, AB018272, AL831918) and one known, ubiquitin specific protease 3 (USP34). Ubiquitin specific proteases (USP), also known as deubiquitinating enzymes (DUB), function in the removal of ubiquitin groups from protein substrates (Borodovsky, et al. 2005). To date, the vast majority of USP have not been characterized and their physiological significance remains unclear (Borodovsky, et al. 2005). This is the first implication of USP involvement in tumorigenesis. Loss of heterozygosity on 2p has been reported in a variety of cancers including prostate (Ueda, et al. 2005), colorectal (Tada, et al. 2004), cervical (Miyai, et al. 2004) and salivary gland tumours (Johns, et al. 1996). However, CGH studies on oral tumours have never identified a minimal region of loss or gain. Indeed, this is the first application of complete coverage array CGH for the detailed analysis of oral tumours. Since the region is sub-megabase, previous studies may have simply missed this alteration due to insufficient resolution. This is the first report of a loss on 2p in oral cancer. LOH on 4q has frequently been reported in OSCC (Garnis, et al. 2004b; Grati, et al. 2000; Huang, et al. 2002; Lin, et al. 2002; Ng, et al. 2000; Okafuji, et al. 1999; Rosin, et al. 2000; Weber RG 1998; Zhang and Rosin 2001). Indeed, the frequency plot in Figure 15 exhibits a high frequency of loss on the 4q arm. SeeGH karyograms alignment revealed a minimal region of loss at 4q34.3, which was 0.44 Mbp in size from bp 86 1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 I 528T 486T 794T Figure 28: Alignment of a novel recurring minimal region of loss at 2pl5 among 3 tumours. Shown here is the alignment of 3 tumours (528T, 486T, 794T) in SeeGH. For each tumour profile the central purple line represents a l o g 2 ratio of 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right and the two green lines to left represent l o g 2 ratios of +0.5 and +1, and -0.5 and -1, respectively. Green shaded regions mark areas of copy number decrease. The approximate position of USP34 is drawn in purple along the third tumour profile before the matched normal. As shown here the minimal region o f gain is marked by 486T. 87 180186394 (RP11-696K24) to bp 180630413 (RP11-631C12). This novel loss was present in 14/20 tumours; however, interestingly, there are no genes within this region. Three of these cases are illustrated in Figure 29 (528T, 125T, 809T). Loss of heterozygosity at 4q34 has been reported in a variety of cancers, including colorectal (Douglas, et al. 2004), small cell lung (Cho, et al. 2002), liver (Bluteau, et al. 2002), and cervical (Sherwood, et al. 2000). However, this is the first report of a minimal region of loss at 4q34 in oral cancer. While the absence of genes in this region may appear to halt the analysis of this region, further research should be done on this region as any recurring MRA harbouring no genes may harbour novel undiscovered genes or critical regulatory regions. The final novel region identified in these tumours was a recurrent minimal deletion at 16q23.2 from bp 80907943 to bp 81058473 (RP11-806J24). Once again, this minimal loss was not represented in the frequency plot (Figure 15). This loss occurred in 7/20 cases and encompasses 5 genes - 3 hypothetical (BC014120, BC014157, AK098052) and 2 known, polycystic kidney disease 1-like 2 (PKD1L2) and beta-carotene 15, 15'-dioxygenase 1 (BCDOl). Three cases exhibiting this deletion are aligned in Figure 30 (794T, 800T, 836T). PKD1L2 has been shown to be a novel G-protein binding protein expressed in heart, mammary gland and brain (Yuasa, et al. 2004). Deletion of PKD1L2 results in polycystic kidney disease (Li, et al. 2003), which is believed to result from increased cellular proliferation stimulated by cAMP. PDK1L2 has been shown to bind G-proteins that stimulated cAMP (Yuasa, et al. 2004). This is the first report of a PDK1L2 deletion in tumorigenesis. 88 -1 -0.5 +0.5 +1 \ 1 • l l j ' ir. 1 -0.5 +0.5 +1 J -1 -0.5 +0.5 +1 528T 125T 809T Figure 29: Alignment of a novel recurring minimal region of loss at 4q34.3 among 3 tumours. Shown here is the alignment of 3 tumours (528T, 125T, 809T) in SeeGH. For each tumour profile the central purple line represents a l o g 2 ratio o f 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right and the two green lines to left represent l o g 2 ratios o f +0.5 and +1, and -0.5 and -1 , respectively. Green shaded regions mark areas of copy number decrease. A s shown here, the minimal region o f gain is marked by 125T. Interestingly, there are no known genes within this region but the deletion may interrupt an unknown gene or a regulatory region. 89 -1 -0.5 +0.5 +1 -1 -0.5 +0.5 +1 ' 1 i '94T t 1 II 1 ^ l l 1 1 1 L 11 1 -0.5 +0.5 +1 I 800T 836T Figure 30: Alignment of a novel recurring minimal region of loss at 16q23.2 among 3 tumours. Shown here is the alignment of 3 tumours (794T, 800T, 836T) in SeeGH. For each tumour profile the central purple line represents a logi ratio o f 0. Clones lying along this central line show no change. Clones lying to the right of the central line show copy number increase while clones lying to the left of the central line show copy number decrease. The two red lines to the right and the two green lines to left represent l o g 2 ratios o f +0.5 and +1, and -0.5 and -1, respectively. Green shaded regions mark areas o f copy number decrease. The approximate positions of PKD1L2 and GAN are drawn in purple along the third tumour profile before the matched normal. As shown here the minimal region o f gain is marked by 794T. 90 Similarly, this is also the first report of a deletion of BCDOl; however, there is also no literature reporting information on this gene. Deletions on 16q have also been reported in a variety of cancers, including gall bladder (Kuroki, et al. 2005), breast (Simpson, et al. 2005), brain (Zakrzewska, et al. 2004), endocervical (Hirai, et al. 2004), prostate (Pettus, et al. 2004), and salivary gland (Choi, et al. 2002; Johns, et al. 1996). Losses specific to 16q23.2 have also been reported in prostate (Fromont, et al. 2003; Latil, et al. 1997; McCarthy, et al. 2004; Watson, et al. 2004a), liver (Park, et al. 2004) and gastric cancer (Aqeilan, et al. 2004). However, this is the first report of a minimal loss at 16q23 in oral cancer. Interestingly, 16q23-24 has been reported to be gained in oral cancer (Wolff, et al. 1998) using conventional CGH. 3.2.8 "Masking" of single copy number changes due to tissue heterogeneity. As previously discussed, the frequency plot (Figure 15) shows a higher frequency of whole arm or chromosome loss than whole arm or chromosome gain, and frequent regions of gain appear to be small or sub-arm in size. The results of the SeeGH karyogram alignment analysis correlates with this observation with the identification of more minimal regions of gain than loss (5 vs 3). This is because regions of gain were usually whole arm and many megabases in size. This may be explained by tumour cell aneuploidy, as described in section 3.2.4. However, this may also be explained by the "masking" of alterations in these late stage lesions, caused (i) genomic instability, (ii) tissue heterogeneity due to the presence of subpopulations of tumour cells, and (iii) non-cancerous cell contamination such as by epithelial and connective tissue cells, circulatory cells, and immune cells. In late stage tumours, the tumour cells microdissected do not 91 represent a pure population of genetically identical cells. Rather, there are subpopulations of tumours cells, which, although clonally related, exhibit many different alterations which may look histologically similar. In addition, the technique of microdissection does not completely eliminate contamination of surrounding normal cells, such as normal epithelium and connective tissue as well as lymphocytes from the capillary networks within the tumours produced by angiogenesis. The presence of these normal cells within the microdissected and DNA extracted tissue may mask small single copy number changes. While gains often represent many more copy fold increases, copy number losses can only represent at most a 2-fold difference. Hence, minimal regions of loss may be more readily masked than minimal regions of gain. Studies within the laboratory have assessed the effect of varying levels of contamination in the visual detection of copy number alterations (Garnis 2005b). They demonstrate that increasing degrees of contamination lead to an apparent decrease in the log2 ratio difference. They state that alterations as small as ~1 Mb can tolerate 75% contamination. However, one must keep in mind that these studies were conducted on genetically pure cell lines. At present, the inability to measure the level of contamination from both infiltrating normal cells and tumour cell subpopulations makes it difficult to control the degree of the "masking" effect. The first step might be to apply more precise microdissection techniques, such as laser.microdissection, to eliminate contamination of normal cells. 3.2.9 Comparison of tumour and dysplasia data In addition to the analysis of oral tumour specimens, concurrent studies on oral dysplasia and CIS specimens have also been conducted by fellow graduate student Cathie 92 Garnis. While this data is in the early stages of analysis, multiple alignment of the data from the oral tumours with that of the dysplasias and CIS at the novel regions of alteration outlined in this thesis reveals some similarities. The two minimal regions of gain at 5pl5.2, containing TRIO, and at 7q35, containing ARHGEF5 and TPK1, were also present in a selection of the oral dysplasias and CIS (Garnis 2005a). While this suggests the importance of both regions in tumourigenesis, the absence of the remaining novel alterations in the dysplasia and CIS samples does not undercut the validity of those regions. The process of tumourigensis is well known to be characterized by the gradual accumulation of genetic alterations. The amplification at 1 lq22.2-22.3 containing the matrix metalloproteinases is an example of an alteration that would be accrued at the later stages of disease because it selects for metastasis. Hence, the alterations at 3q23, 7pl2.3-13, 7q21.2, 2pl5 and 16q23.2 may simply be acquired at a later stage. 3.3 Expression Analysis Confirmation of Candidate Genes In order to assess the effects on gene expression associated with the copy number alterations identified in this thesis project, reverse transcriptase rtPCR was performed on a panel of oral tumours and normal oral tissue. Five frozen sections and 7 normal sections were microdissected and the mRNA was isolated. Since single copy number changes are difficult to assess using rtPCR and losses represent low copy number changes, the analysis was restricted to genes located within novel regions of copy number gain. Primers were designed specific to RASA2 (3q23) TRIO (5pl5.2), TENSI (7pl3 -12.3), CDK6 (7q21.2) and ARHGEF5 (7q35) transcripts; however, only primer sets specific to TRIO, TENSI, and CDK6 were successfully optimized and utilized. Four of 93 the 5 tumours and seven of the 8 normal samples analyzed gave informative results for each primer set. TRIO expression in oral tumours was significantly higher than in normal oral samples (Figure 31). When the PCR products for TRIO were run and imaged on an acrylamide gel, lanes loaded with tumour product clearly exhibited bands of the expected band size for TRIO, while lanes loaded with normal product showed no visible bands (Figure 31 A). Further analysis of the bands using ImageQuant enabled the quantification of band intensities, which were then averaged across tumour samples and normal samples separately. The bar graph shown in Figure 3 IB demonstrates the significant difference in the average TRIO band intensity between the tumours and the normal samples. Moreover, application of the Mann-Whitney U Test, a non-parametric rank test applied to data sets when the sample size is small and the assumption of normality cannot be met (Zar 1999), suggested that the expression level in tumours and normal samples were statistically different. The U and U ' were calculated to be 24 and 0 respectively. Since U (24) > p.o.05(2), 4,6 (22), the null hypothesis, that the average level of TRIO expression in tumours and normals is not different, is rejected. Thus, the alterative hypothesis, that the average level of TRIO expression in tumours and normals is different, is accepted. Thus, an increase in the level TRIO expression in tumour samples as compared to normal samples suggest confirmation of the amplified minimal region on 5pl 1.2 containing the TRIO gene. 94 300bp 200bp ^ , lOObpli^ (4) (6) Figure 31: Comparison of T R I O expression between oral tumours and normal oral specimens. P C R products for both TRIO and ^-microglobulin (B2M) using both tumour and normal D N A templates were run on a 12% acrylamide gel (A). As shown, TRIO bands appeared in the lanes loaded with the tumour product, but not in lanes loaded with the normal product. While the B2M bands in the normal lanes also appear to be dimmer for some samples, analysis o f the bands using ImageQuant software followed by a statistical analysis revealed that there is a significant difference in the TRIO expression between tumours and normal samples (B). 95 The results for both TENSI and CDK6 also suggest that their expression is higher in tumours than in normal oral samples. However, interpretation of the results was slightly more complicated due to the presence of an additional band on the acrylamide gel for both genes (Figure 32A and 33A). Since, the sizes of the unidentified bands differ (Figure 34) for each gene and, since the same /32-microglobulin control PCR reaction was loaded onto all 3 gels, the extra bands cannot be attributed to the ^ -microglobulin PCR product. However, as a precaution, the TENSI and CDK6 PCR product was run on an acrylamide gel in the absence of p2-microglobulin. As expected, the additional bands were still present in the lanes (Figure 34). While it is possible that both bands represent a splice variant, this, too, is unlikely, due to the small size of the band. Moreover, while the primers designed for TENSI span a short exon, the primers designed for CDK6 do not. Finally, even if the intervening exon in the TENSI transcript were spliced, the resulting band would still be greater than 100 base pairs (127 vs 153 base pairs). In contrast, it also appears unlikely that the band represents primer dimmers, since the bands are not present in the normal lanes. It is also possible, however, that these extra bands may represent gel artefacts resulting from different conformations of the PCR product. Further optimization of these primers may eliminate these unidentified bands. In order to account for the difference these extra bands may make, ImageQuant analysis of gels for TENSI and CDK6 was done twice, separately, first using only the expected band and then using both bands. Similarly, the average expression for tumours and normal samples were graphed for TENSI and CDK6 (Figures 33B,C and Figure 34B,C). The bar graphs for both genes demonstrate that the differences in the expression levels between the tumours and the normals are significantly different and that the standard error of the 96 means for each do not overlap. Application of the non-parametric Mann-Whitney U Test further suggests a significant difference in the expression of both TENSI and CDK6 between tumours and normal samples. As in the TRIO test, the value for U was calculated to be 24 and, since U (24) > po.05(2), 4,6 (22), the null hypothesis, that the average level of TENSI (or CDK6) expression in tumours and normals is not different, is rejected. The alterative hypothesis, therefore, is accepted, and the average level of TENSI (CDK6) expression in tumours and normals is significantly different. This supports a validation of the regions of gain at 7pl3-12.3 containing TENSI and at 7q21.2 containing CDK6. Although these expression studies are promising for the validation of the recurring MRA detected with the whole genome SMRT array, they are limited by the small sample size of tumours screened. Microdissection of the frozen tumours is a very time-consuming process that it was not possible to achieve within the time allotted for this thesis. However, microdissection of further frozen tumours sections for RNA extraction are currently underway and, thus, the analysis of a larger cohort of samples will provide more concrete data. Similarly, not all the genes within these regions, nor all the regions, were assessed for expression in this thesis. Further troubleshooting focused on designing specific primers, and the optimization of primers specific to these genes in the near future will facilitate the expression analysis of all the known candidate genes identified in this thesis. 97 Figure 32: Comparison of TENSI expression between oral tumours and normal oral specimens. P C R products for both TENSI and ^-microglobulin (/32M) using both tumour and normal D N A templates were run on a 12% acrylamide gel (A). As shown, TENSI bands appeared in the lanes loaded with the tumour product, but not in lanes loaded with the normal product. Interestingly, there also appears to be a secondary mystery band <100bp in size. This product was also run independent of /32M to determine that the band was not attributed to this control gene (Figure 35). This band may represent a splice variant or a conformational artefact. To control for the impact of this and the dimmer J32M bands in some of the normal lanes, analysis o f the bands using ImageQuant software followed by a statistical analysis was also done using only the expected band size (B) and using both the expected and mystery bands (C). As shown, in both instances, there is a significant difference in the TENSI expression between tumours and normal samples. 08 -r Tumours CDK6 unidentified dW UJ Normals "N (-) (•) mm 300bp , 200bp l O O b p I B 2 - M i c r o g l o b u l i n 300bp 2 0 0 b P i Tumours (4) Normals (6) Tumours (4) Normals (6) Figure 33: Comparison of CDK6 expression between oral tumours and normal oral specimens. P C R products for both CDK6 and ^-microglobulin (f32M) using both tumour and normal D N A templates were run on a 12% acrylamide gel (A). As shown, CDK6 bands appeared in the lanes loaded with the tumour product, but not in lanes loaded with the normal product. Interestingly, there also appears to be a secondary mystery band <100bp in size. This product was also run independent of ft2M\o determine that the band was not attributed to this control gene (Figure 35). This band may represent a splice variant or a conformational artefact. To control for the impact of this and the dimmer R2M bands in some of the normal lanes, analysis o f the bands using ImageQuant software followed by a statistical analysis was also done using only the expected band size (B) and using both the expected and mystery bands (C). As shown in both instances there is a significant difference in the CDK6 expression between tumours and normal samples. 99 1 0 0 b p ^ unidentified unidentified Figure 34: Assessment of mystery band present in TENSI and CDK6 PCR product. The same tumour P C R product for both TENSI and CDK6 run on the acyrlamide gels shown in figures 33 and 34, were run a 12% acrylamide gel independent of B2-microglobulin (B2M) to demonstrate that the unexpected bands present on the previous gels was due to the TENSI and CDK6 P C R product, and not that of B2M. Indeed, these mystery bands also appear here. While these bands may represent an alternative splice variant, they are more likely to represent a conformational form o f the expected band. While the TENSI primers do span a small exon, the band size difference would be 127 vs 153 bp, whereas the mystery band appears <100 bp. Similarly, the primers for CDK6 do not span an exon. 1 0 0 3.4 Significance The identification of the novel regions of gain and loss described here and their validation through rtPCR analysis of the genes within these regions demonstrates the utility of whole genome SMRT array analysis in the elucidation of the genetic mechanisms driving OSCC progression. As demonstrated in this thesis, the sub-megabase resolution enables the fine-mapping and characterization of known regions as well as novel regions. Analysis of a much larger cohort of tumour samples will be useful for correlating alterations with disease behaviour and outcome, thus sub-typing the disease. Ultimately, this will lead to better management of the disease, through genetic screening of premalignant lesions, which will aid in treatment decision-making in terms of both surgical and drug therapy. Further, aCGH will also be invaluable in the identification of gene targets for novel drug therapy development. Using new regions and new genes, better and more specific drugs can be developed for different disease sub-types. While the ability of aCGH to detect genetic alterations on a genome-wide basis with high resolution in a single experiment is a great improvement on previous technologies and will greatly benefit our understanding of oral cancer, one must also be aware that this technique will not replace existing techniques used in the analysis of oral genomes, but, rather, will be most useful when integrated with these conventional approaches. Array CGH is highly sensitive to detecting copy number differences; however, it is not able to detect other forms of LOH such as single base pair mutations, reciprocal translocations that interrupt critical genes, or epigenetic alterations such as 101 methylation, which silence or activate gene promoters. In contrast, these alterations are detectable, although not distinguishable, with micro satellite analysis. Similarly, SNP oligonucleotides arrays, although lower in resolution than the whole genome SMRT array, are able to detect single base pair alterations. It is the consolidation of, rather than the competition between, these techniques that will unravel the biological mechanisms driving the behaviour of this disease. 3.5 Future Directions This thesis provides a proof-of-principle analysis of oral tumour genomes using whole genome SMRT array technology. However, while a series of novel minimal regions of alteration were identified and characterized, analysis of a much larger cohort of tumour genomic DNA samples, as well as the matched RNA and normal DNA, will be required for the detailed sub-classification and characterization of this disease. While a selection of dysplasias and CIS samples are in the early stages of SeeGH karyogram analysis, the sample size is similar to that of this study. Thus, a detailed analysis of the DNA, RNA, and matched normal DNA from a much larger cohort of each of mild dysplasia, moderate dysplasia, severe dysplasia and CIS samples, followed by a subsequent comparison with the aforementioned larger cohort of invasive tumour data, will provide a more comprehensive picture of genome copy number alterations driving various stages of oral cancer progression. Analysis of epigenetic alterations will also enable a better understanding of disease progression. Examples of such epigenetic changes include DNA methylation and 102 histone methylation. Both DNA and histone methylation have been shown to be involved in tumourigenesis (Feinberg and Tycko 2004). While hypomethylation of both DNA and histones can lead to activation of oncogenes, hypermethylation can lead to the silencing of tumour suppressor genes. In DNA, methylation is thought to occur at CpG rich regions within the promoter. Interestingly, in this study, 9p, which is commonly believed to be the first loss in oral cancer progression, was less frequently lost than 3p, which is thought to be lost subsequent to the loss of 9p. This raises the possibility that methylation within the 9p arm may be producing the LOH identified in this region and, therefore, may be silencing expression of the pi 6 gene in this region. Certainly, pi 6 has been shown to be methylated in cell lines and primary tumours (Hibi, et al. 2001; Jones and Baylin 2002; Puri, et al. 2005; Raschke, et al. 2005; Sanchez-Cespedes, et al. 2000; Shapiro, et al. 1995). The development of a novel application of the whole genome SMRT array for the analysis of DNA methylation is currently underway. The approach involves isolating DNA bound to methyl transferase proteins in both normal and tumour DNA, purifying the DNA, which is then labelled and the data analyzed in SeeGH. Regions shown to be hypermethylated, therefore, appear as gains in the SeeGH karyograms, while hypomethylated regions appear to be lost. This approach is in the early stages of development, but represents a cutting edge technique useful for the detailed analysis of methylation patterns in tumour DNA. The integration of approaches such as this for epigenetic analysis of a larger cohort OSCC specimen from various stages will be invaluable to understanding the biological mechanisms underlying oral cancer. 103 While detailed analysis of copy number alterations, mutations, and epigenetic changes will be highly useful in understanding the biology of the disease, knowledge of the role of aneuploidy in disease progression will be of equal worth in elucidating the underlying biological mechanisms. Aneuploidy has been shown to be highly common in epithelial tumours (Mitelman, et al. 2003); however, it is not understood whether aneuploidy results from the genomic instability or if genomic instability is caused by aneuploidy (Storchova and Pellman 2004). At present cytogenetic analysis of tumour genomes is the only method of assessing patterns of aneuploidy in epithelial tumours. Previous studies have reported that each tumour has a different pattern of aneuploidy, making identification of recurrent patterns difficult (Storchova and Pellman 2004). Development of a high throughput technique to assess aneuploidy in a large cohort of specimens from various stages of disease progression may reveal some key patterns critical to understanding the disease. Unfortunately, no clinical outcome or history was recorded for the tumours profiled in this study. Although this sample size is not sufficiently large for a comprehensive analysis of the genetic mechanisms driving the disease, this data does provide some insight into the association of novel alterations with the invasive stages. Future analysis of larger cohorts of samples that have recorded clinical history will be crucial to sub-typing the disease, based both on genetic alterations and outcome. This would aid in better screening programs and improved disease management. 104 Following the identification of the novel alterations identified in this study, the next step will be to better characterize expression patterns of the candidate genes lying within these regions. While we assessed gene expression in 4 tumours within 3 regions, this sample size is clearly inadequate, and, furthermore, not all of the genes within these regions, nor all of the regions, were verified. Further expression analysis of all candidate genes within a much larger cohort of samples, perhaps using a tissue microarray, would provide a much more reliable validation of these regions. Furthermore, functional analysis of the genes will aid in the unravelling of the molecular mechanisms underlying the involvement of these genes in tumourigenesis. Such approaches might include the use of cell models and RNA interference, in which the expression of putative tumour suppressor gene is inhibited, and transgenic mouse models, in which a candidate tumour suppressor gene is "knocked-out" or a candidate oncogene is "knocked-in." Only when the functional role of these genes in tumourigenesis is understood will the development of novel drug therapies and, subsequently, improved treatment, begin. 105 C H A P T E R 4: C O N C L U S I O N S The data from this thesis can be summarized in the following key points: 1) High resolution aCGH using both a 3p-arm specific BAC array and whole genome sub-megabase resolution tiling-set (SMRT) array, detects DNA copy number gains and losses in archival formalin-fix oral squamous cell carcinoma (OSCC) DNA. 2) High resolution aCGH detects novel, recurrent sub-megabase genetic alterations not previously identified by conventional approaches. 3) Recurrent, LOH observed on the short arm of chromosome 3 in OSCC can be attributed to 5 recurring minimal regions of copy number alteration. 4) Recurrent sub-megabase alterations, not detected by previously existing technologies, are present throughout the genomes of OSCC. Some examples include the 0.58 Mbp alteration at 5pl5.2 encompassing the gene TRIO, the 0.25 Mbp gain at 7pl2.3-13 comprising TEM6, and the 1.24 Mbp amplification at llq22.2-22.3 encompassing a cluster of matrix metalloproteinase genes. 5) The importance of the sub-megabase alterations in the behaviour of this disease is suggested by the presence of genes within these regions known to be involved in cellular proliferation and apoptosis or previously implicated in cancer. 106 6) A selection of regions, including the copy number gains at 5pl5.2, 7pl2.3-13, and 7q21.2 were confirmed through rtPCR expression analysis of the genes lying within them (TRIO, TEM6, and CDK6, respectively) within a small sample of 4 tumours. 7) In addition to the regions already analyzed using expression analysis, the remaining regions will also be confirmed through rtPCR expression analysis on a larger cohort of OSCC. 8) Investigation of a much larger cohort of OSCC, accompanied by clinical history data using the whole genome SMRT array, may make possible the sub-typing of the disease. 9) Further functional studies of the candidate genes will elucidate the biological mechanisms underlying this disease and will aid in the development of novel drug therapy. 107 REFERENCES Abbey L M , Kaugars GE, Gunsolley JC, Burns JC, Page DG, Svirsky JA, Eisenberg E, Krutchkoff DJ, Cushing M . 1995. Intraexaminer and interexaminer reliability in the diagnosis of oral epithelial dysplasia. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 80(2): 188-91. Albertson DG, Pinkel D. 2003. Genomic microarrays in human genetic disease and cancer. Hum Mol Genet 12 Spec No 2:R145-52. Aqeilan Rl, Kuroki T, Pekarsky Y, Albagha O, Trapasso F, Baffa R, Huebner K, Edmonds P, Croce CM. 2004. Loss of WWOX expression in gastric carcinoma. Clin Cancer Res 10(9):3053-8. Arai K, Shibahara T, Yamamoto N, Noma H. 2002. The presence of candidate tumor suppressor gene loci at chromosome 3p for oral squamous cell carcinomas. Oral Oncology. 38(8):763-71. Ashman JN, Brigham J, Co wen ME, Bahia H, Greenman J, Lind M , Cawkwell L. 2002. Chromosomal alterations in small cell lung cancer revealed by multicolour fluorescence in situ hybridization. Int J Cancer 102(3):230-6. Ashman JN, Patmore HS, Condon LT, Cawkwell L, Stafford ND, Greenman J. 2003. Prognostic value of genomic alterations in head and neck squamous cell carcinoma detected by comparative genomic hybridisation. Br J Cancer 89(5):864-9. Aspenstrom P. 2004. Integration of signalling pathways regulated by small GTPases and calcium. Biochim Biophys Acta 1742(l-3):51-8. Aulmann S, Schnabel PA, Helmchen B, Dienemann H, Drings P, Otto HF, Sinn HP. 2005. Immunohistochemical and cytogenetic characterization of acantholytic squamous cell carcinoma of the breast. Virchows Arch. Baldassarre G, Romano A, Armenante F, Rambaldi M, Paoletti I, Sandomenico C, Pepe S, Staibano S, Salvatore G, De Rosa G and others. 1997. Expression of teratocarcinoma-derived growth factor-1 (TDGF-1) in testis germ cell tumors and its effects on growth and differentiation of embryonal carcinoma cell line NTERA2/D1. Oncogene 15(8):927-36. Baldassarre G, Tucci M, Lembo G, Pacifico FM, Dono R, Lago CT, Barra A, Bianco C, Viglietto G, Salomon D and others. 2001. A truncated form of teratocarcino ma-derived growth factor-1 (cripto-1) mRNA expressed in human colon carcinoma cell lines and tumors. Tumour Biol 22(5):286-93. 108 Barnes LD, Garrison PN, Siprashvili Z, Guranowski A, Robinson AK, Ingram SW, Croce CM, Ohta M, Huebner K. 1996. Fhit, a putative tumor suppressor in humans, is a dinucleoside 5',5"'-P1,P3-triphosphate hydrolase. Biochemistry 35(36): 11529-35. Beder LB, Gunduz M, Ouchida M , Fukushima K, Gunduz E, Ito S, Sakai A, Nagai N , Nishizaki K, Shimizu K. 2003. Genome-wide analyses on loss of heterozygosity in head and neck squamous cell carcinomas. Laboratory Investigation. 83(1):99-105. Bignell GR, Huang J, Greshock J, Watt S, Butler A, West S, Grigorova M , Jones KW, Wei W, Stratton MR and others. 2004. High-resolution analysis of DNA copy number using oligonucleotide microarrays. Genome Res 14(2):287-95. Birkedal-Hansen H, Moore WG, Bodden MK, Windsor LJ, Birkedal-Hansen B, DeCarlo A, Engler JA. 1993. Matrix metalloproteinases: a review. Crit Rev Oral Biol Med 4(2):197-250. Bluteau O, Beaudoin JC, Pasturaud P, Belghiti J, Franco D, Bioulac-Sage P, Laurent-Puig P, Zucman-Rossi J. 2002. Specific association between alcohol intake, high grade of differentiation and 4q34-q35 deletions in hepatocellular carcinomas identified by high resolution allelotyping. Oncogene 21(8):1225-32. Bockmuhl U, Wolf G, Schmidt S, Schwendel A, Jahnke V, Dietel M, Petersen I. 1998. Genomic alterations associated with malignancy in head and neck cancer. Head & Neck. 20(2): 145-51. Borodovsky A, Ovaa H, Meester WJ, Venanzi ES, Bogyo MS, Hekking BG, Ploegh HL, Kessler BM, Overkleeft HS. 2005. Small-molecule inhibitors and probes for ubiquitin- and ubiquitin-like-specific proteases. Chembiochem 6(2):287-91. Bosman FT. 2001. Dysplasia classification: pathology in disgrace? J Pathol 194(2): 143-4. Bouquot J. 1997. The Pathology and Progression of Oral Premalignancy. Epithelial Dysplasia Symposium, 5th International Congress on Oral Cancer. London: Royal College of Physicians. Bridge RS, Jr., Bridge JA, Neff JR, Naumann S, Althof P, Bruch LA. 2004. Recurrent chromosomal imbalances and structurally abnormal breakpoints within complex karyotypes of malignant peripheral nerve sheath tumour and malignant triton tumour: a cytogenetic and molecular cytogenetic study. J Clin Pathol 57(11): 1172-8. Buchhagen DL, Worsham MJ, Dyke DL, Carey TE. 1996. Two regions of homozygosity on chromosome 3p in squamous cell carcinoma of the head and neck: comparison with cytogenetic analysis. Head & Neck. 18(6):529-37. 109 Buckley PG, Mantripragada KK, Benetkiewicz M , Tapia-Paez I, Diaz De Stahl T, Rosenquist M, Ali H, Jarbo C, De Bustos C, Hirvela C and others. 2002. A full-coverage, high-resolution human chromosome 22 genomic microarray for clinical and research applications. Human Molecular Genetics. ll(25):3221-9. Califano J, van der Riet P, Westra W, Nawroz H, Clayman G, Piantadosi S, Corio R, Lee D, Greenberg B, Koch W and others. 1996. Genetic progression model for head and neck cancer: implications for field cancerization. Cancer Research. 56(ll):2488-92. Carlos de Vicente J, Herrero-Zapatero A, Fresno MF, Lopez-Arranz JS. 2002. Expression of cyclin D l and Ki-67 in squamous cell carcinoma of the oral cavity: clinicopathological and prognostic significance. Oral Oncol 38(3):301-8. Carson-Walter EB, Watkins DN, Nanda A, Vogelstein B, Kinzler KW, St Croix B. 2001. Cell surface tumor endothelial markers are conserved in mice and humans. Cancer Res61(18):6649-55. Casas S, Aventin A, Fuentes F, Vallespi T, Granada I, Carrio A, Angel Martinez-Climent J, Sole F, Teixido M , Bernues M and others. 2004. Genetic diagnosis by comparative genomic hybridization in adult de novo acute myelocytic leukemia. Cancer Genet Cytogenet 153(1): 16-25. Chang KW, Kao SY, Tzeng RJ, Liu CJ, Cheng AJ, Yang SC, Wong YK, Lin SC. 2002. Multiple molecular alterations of FHIT in betel-associated oral carcinoma. Journal ofPathology. 196(3):300-6. Charous SJ, Stricklin GP, Nanney LB, Netterville JL, Burkey BB. 1997. Expression of matrix metalloproteinases and tissue inhibitor of metalloproteinases in head and neck squamous cell carcinoma. Ann Otol Rhinol Laryngol 106(4):271-8. Chen YJ, Lin SC, Kao T, Chang CS, Hong PS, Shieh TM, Chang KW. 2004. Genome-wide profiling of oral squamous cell carcinoma. J Pathol 204(3):326-32. Chi B, DeLeeuw RJ, Coe BP, MacAulay C, Lam WL. 2004. SeeGH~a software tool for visualization of whole genome array comparative genomic hybridization data. BMC Bioinformatics 5(1): 13. Cho ES, Chang J, Chung KY, Shin DH, Kim YS, Kim SK. 2002. Identification of tumor suppressor loci on the long arm of chromosome 4 in primary small cell lung cancers. Yonsei Med J 43(2): 145-51. Choi HR, Batsakis JG, Callender DL, Prieto VG, Luna MA, El-Naggar AK. 2002. Molecular analysis of chromosome 16q regions in dermal analogue tumors of salivary glands: a genetic link to dermal cylindroma? Am J Surg Pathol 26(6):778-83. 110 Chujo M, Noguchi T, Miura T, Arinaga M, Uchida Y, Tagawa Y. 2002. Comparative genomic hybridization analysis detected frequent overrepresentation of chromosome 3q in squamous cell carcinoma of the lung. Lung Cancer 38(l):23-9. Clamp M, Andrews D, Barker D, Bevan P, Cameron G, Chen Y, Clark L, Cox T, Cuff J, Curwen V and others. 2003. Ensembl 2002: accommodating comparative genomics. Nucleic Acids Res 31(l):38-42. Clarke LE, Leitzel K, Smith J, Ali SM, Lipton A. 2003. Epidermal growth factor receptor mRNA in peripheral blood of patients with pancreatic, lung, and colon carcinomas detected by RT-PCR. Int J Oncol 22(2):425-30. Coe BP, Henderson LJ, Garnis C, Tsao MS, Gazdar AF, Minna J, Lam S, Macaulay C, Lam WL. 2005. High-resolution chromosome arm 5p array CGH analysis of small cell lung carcinoma cell lines. Genes Chromosomes Cancer 42(3):308-13. Davies JJ, Wilson IM, Lam WL. 2005. Array CGH technologies and their applications to cancer genomes. Chromosome Research 13:1-11. De Leeuw RJ, Davies JJ, Rosenwald A, Bebb G, Gascoyne RD, Dyer MJ, Staudt L M , Martinez-Climent JA, Lam WL. 2004. Comprehensive whole genome array CGH profiling of mantle cell lymphoma model genomes. Hum Mol Genet 13(17): 1827-37. Debant A, Serra-Pages C, Seipel K, O'Brien S, Tang M, Park SH, Streuli M . 1996. The multidomain protein Trio binds the LAR transmembrane tyrosine phosphatase, contains a protein kinase domain, and has separate rac-specific and rho-specific guanine nucleotide exchange factor domains. Proc Natl Acad Sci U S A 93(1.1):5466-71. Debily MA, Camarca A, Ciullo M , Mayer C, El Marhomy S, Ba I, Jalil A, Anzisi A, Guardiola J, Piatier-Tonneau D. 2004. Expression and molecular characterization of alternative transcripts of the ARHGEF5/TIM oncogene specific for human breast cancer. Hum Mol Genet 13(3):323-34. Dobashi Y, Takei N , Suzuki S, Yoneyama H, Hanawa M, Ooi A. 2004. Aberration of epidermal growth factor receptor expression in bone and soft-tissue tumors: protein overexpression, gene amplification and activation of downstream molecules. Mod Pathol 17(12): 1497-505. Douglas EJ, Fiegler H, Rowan A, Halford S, Bicknell DC, Bodmer W, Tomlinson IP, Carter NP. 2004. Array comparative genomic hybridization analysis of colorectal cancer cell lines and primary carcinomas. Cancer Res 64(14):4817-25. i l l Dumur CI, Dechsukhum C, Ware JL, Cofield SS, Best A M , Wilkinson DS, Garrett CT, Ferreira-Gonzalez A. 2003. Genome-wide detection of LOH in prostate cancer using human SNP microarray technology. Genomics 81(3):260-9. El-Naggar AK, Coombes M M , Batsakis JG, Hong WK, Goepfert H, Kagan J. 1998. Localization of chromosome 8p regions involved in early tumorigenesis of oral and laryngeal squamous carcinoma. Oncogene 16(23):2983-7. El-Naggar AK, Mao L, Staerkel G, Coombes M M , Tucker SL, Luna MA, dayman GL, Lippman S, Goepfert H. 2001. Genetic heterogeneity in saliva from patients with oral squamous carcinomas: implications in molecular diagnosis and screening. J Mol Diagn 3(4): 164-70. Epstein JB, Zhang L, Poh C, Nakamura H, Berean K, Rosin M. 2003. Increased allelic loss in toluidine blue-positive oral premalignant lesions. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 95(l):45-50. Fearon ER, Vogelstein B. 1990. A genetic model for colorectal tumorigenesis. Cell 61(5):759-67. Feinberg AP, Tycko B. 2004. The history of cancer epigenetics. Nat Rev Cancer 4(2):143-53. Fiegler H, Carr P, Douglas EJ, Burford DC, Hunt S, Scott CE, Smith J, Vetrie D, Gorman P, Tomlinson IP and others. 2003. DNA microarrays for comparative genomic hybridization based on DOP-PCR amplification of BAC and PAC clones. Genes Chromosomes Cancer 36(4):361-74. Foda HD, Zucker S. 2001. Matrix metalloproteinases in cancer invasion, metastasis and angiogenesis. Drug Discov Today 6(9):478-482. Fong K M , Biesterveld EJ, Virmani A, Wistuba I, Sekido Y, Bader SA, Ahmadian M , Ong ST, Rassool FV, Zimmerman PV and others. 1997. FHIT and FRA3B 3pl4.2 allele loss are common in lung cancer and preneoplastic bronchial lesions and are associated with cancer-related FHIT cDNA splicing aberrations. Cancer Res 57(ll):2256-67. Ford AC, Grandis JR. 2003. Targeting epidermal growth factor receptor in head and neck cancer. Head Neck 25(l):67-73. Franchi A, Santucci M , Masini E, Sardi I, Paglierani M, Gallo O. 2002. Expression of matrix metalloproteinase 1, matrix metalloproteinase 2, and matrix metalloproteinase 9 in carcinoma of the head and neck. Cancer 95(9): 1902-10. Freier K, Bosch FX, Flechtenmacher C, Devens F, Benner A, Lichter P, Joos S, Hofele C. 2003. Distinct site-specific oncoprotein overexpression in head and neck 112 squamous cell carcinoma: a tissue microarray analysis. Anticancer Res 23(5A):3971-7. Fromont G, Joulin V, Chantrel-Groussard K, Vallancien G, Guillonneau B, Validire P, Latil A, Cussenot O. 2003. Allelic losses in localized prostate cancer: association with prognostic factors. J Urol 170(4 Pt 1): 1394-7. Fry DW, Harvey PJ, Keller PR, Elliott WL, Meade M, Trachet E, Albassam M , Zheng X, Leopold WR, Pryer NK and others. 2004. Specific inhibition of cyclin-dependent kinase 4/6 by PD 0332991 and associated antitumor activity in human tumor xenografts. Mol Cancer Ther 3(11):1427-38. Garnis C. 2005a. Identification of Novel Genetic Alterations in the Progression of Lung and Oral Cancer. Ph.D. Dissertation. Garnis C, Baldwin C, Zhang L, Rosin MP, Lam WL. 2003. Use of complete coverage array comparative genomic hybridization to define copy number alterations on chromosome 3p in oral squamous cell carcinomas. Cancer Res 63(24):8582-5. Garnis C, Buys TP, Lam WL. 2004a. Genetic alteration and gene expression modulation during cancer progression. Mol Cancer 3(1):9. Garnis C, Campbell J, Zhang L, Rosin MP, Lam WL. 2004b. OCGR array: an oral cancer genomic regional array for comparative genomic hybridization analysis. Oral Oncol 40(5):511-9. Garnis C, Coe BP, Ishkanian A, Zhang L, Rosin MP, Lam WL. 2004c. Novel regions of amplification on 8q distinct from the MYC locus and frequently altered in oral dysplasia and cancer. Genes Chromosomes Cancer 39(l):93-8. Garnis C, Coe, B.P., Lam, S.L., MacAulay, C , Lam, W.L. 2005b. High resolution array CGH increases heterogeneity tolerance in the analysis of clinical samples. Genomics In Press. Gotte K, Tremmel SC, Popp S, Weber S, Hormann K, Bartram CR, Jauch A. 2005. Intratumoral genomic heterogeneity in advanced head and neck cancer detected by comparative genomic hybridization. Adv Otorhinolaryngol 62:38-48. Grandis JR, Tweardy DJ. 1993a. Elevated levels of transforming growth factor alpha and epidermal growth factor receptor messenger RNA are early markers of carcinogenesis in head and neck cancer. Cancer Res 53(15):3579-84. Grandis JR, Tweardy DJ. 1993b. TGF-alpha and EGFR in head and neck cancer. J Cell Biochem Suppl 17F:188-91. 113 Grati FR, Sirchia SM, Garagiola I, Sironi E, Galioto S, Rossella F, Serafini P, Dulcetti F, Bozzetti A, Brusati R and others. 2000. Losses of heterozygosity in oral and oropharyngeal epithelial carcinomas. Cancer Genetics & Cytogenetics. 118(1):57-61. Greshock J, Naylor TL, Margolin A, Diskin S, Cleaver SH, Futreal PA, deJong PJ, Zhao S, Liebman M , Weber BL. 2004. 1-Mb resolution array-based comparative genomic hybridization using a BAC clone set optimized for cancer gene analysis. Genome Res 14(1): 179-87. Gupta VK, Schmidt AP, Pashia ME, Sunwoo JB, Scholnick SB. 1999. Multiple regions of deletion on chromosome arm 13q in head-and-neck squamous-cell carcinoma. Int J Cancer 84(5):453-7. Hannen EJ, Macville MV, Wienk SM, Slootweg PJ, Manni JJ, Hanselaar AG, de Wilde PC. 2004. Different chromosomal imbalances in metastasized and nonmetastasized tongue carcinomas identified by comparative genomic hybridization. Oral Oncol 40(4):364-71. Hedberg Y, Ljungberg B, Roos G, Landberg G. 2003. Expression of cyclin D l , D3, E, and p27 in human renal cell carcinoma analysed by tissue microarray. Br J Cancer 88(9): 1417-23. Hermsen M, Guervos MA, Meijer G, Baak J, van Diest P, Marcos CA, Sampedro A. 2001. New chromosomal regions with high-level amplifications in squamous cell carcinomas of the larynx and pharynx, identified by comparative genomic hybridization. Journal of Pathology. 194(2): 177-82. Hermsen MA, Joenje H, Arwert F, Braakhuis BJ, Baak JP, Westerveld A, Slater R. 1997. Assessment of chromosomal gains and losses in oral squamous cell carcinoma by comparative genomic hybridisation. Oral Oncol 33(6):414-8. Hibi K, Taguchi M , Nakayama H, Takase T, Kasai Y, Ito K, Akiyama S, Nakao A. 2001. Molecular detection of pi 6 promoter methylation in the serum of patients with esophageal squamous cell carcinoma. Clin Cancer Res 7(10):3135-8. Hibi K, Takahashi T, Yamakawa K, Ueda R, Sekido Y, Ariyoshi Y, Suyama M, Takagi H, Nakamura Y. 1992. Three distinct regions involved in 3p deletion in human lung cancer. Oncogene. 7(3):445-9. Hirai Y, Utsugi K, Takeshima N, Kawamata Y, Furuta R, Kitagawa T, Kawaguchi T, Hasumi K, Noda ST. 2004. Putative gene loci associated with carcinogenesis and metastasis of endocervical adenocarcinomas of uterus determined by conventional and array-based CGH. Am J Obstet Gynecol 191(4): 1173-82. 114 Hoglund M, Gisselsson D, Soller M , Hansen GB, Elfving P, Mitelman F. 2004. Dissecting karyotypic patterns in renal cell carcinoma: an analysis of the accumulated cytogenetic data. Cancer Genet Cytogenet 153(l):l-9. Huang FY, Kwok YK, Lau ET, Tang MH, Ng TY, Ngan HY. 2005. Genetic abnormalities and HPV status in cervical and vulvar squamous cell carcinomas. Cancer Genet Cytogenet 157(l):42-8. Huang Q, Yu GP, McCormick SA, Mo J, Datta B, Mahimkar M , Lazarus P, Schaffer AA, Desper R, Schantz SP. 2002. Genetic differences detected by comparative genomic hybridization in head and neck squamous cell carcinomas from different tumor sites: construction of oncogenetic trees for tumor progression. Genes Chromosomes Cancer 34(2):224-33. Imai FL, Uzawa K, Miyakawa A, Shiiba M, Tanzawa H. 2001. A detailed deletion map of chromosome 20 in human oral squamous cell carcinoma. Int J Mol Med 7(l):43-7. Ishkanian AS, Malloff CA, Watson SK, DeLeeuw RJ, Chi B, Coe BP, Snijders A, Albertson DG, Pinkel D, Marra MA and others. 2004. A tiling resolution DNA microarray with complete coverage of the human genome. Nat Genet 36(3):299-303. Ishwad CS, Ferrell RE, Rossie K M , Appel BN, Johnson JT, Myers EN, Law JC, Srivastava S, Gollin SM. 1995. Microsatellite instability in oral cancer. Int J Cancer 64(5):332-5. Ishwad CS, Ferrell RE, Rossie KN, Appel BN, Johnson JT, Myers EN, Law JC, Srivastava S, Gollin SM. 1996. Loss of heterozygosity of the short arm of chromosomes 3 and 9 in oral cancer. Int J Cancer 69(1): 1-4. Ishwad CS, Shuster M, Bockmuhl U, Thakker N , Shah P, Toomes C, Dixon M, Ferrell RE, Gollin SM. 1999. Frequent allelic loss and homozygous deletion in chromosome band 8p23 in oral cancer. Int J Cancer 80(1):25-31. Janssen JW, Vaandrager JW, Heuser T, Jauch A, Kluin PM, Geelen E, Bergsagel PL, Kuehl WM, Drexler HG, Otsuki T and others. 2000. Concurrent activation of a novel putative transforming gene, myeov, and cyclin DI in a subset of multiple myeloma cell lines witht(ll;14)(ql3;q32). Blood 95(8):2691-8. Jiang WW, Fujii H, Shirai T, Mega H, Takagi M. 2001. Accumulative increase of loss of heterozygosity from leukoplakia to foci of early cancerization in leukoplakia of the oral cavity. Cancer 92(9):2349-56. 115 Jin C JY, Wennerberg J, Dictor M, Mertens F. 2000. Nonrandom Pattern of Cytogenetic Abnormalities in Squamous Cell Carcinoma of the Larynx. Genes, Chromosomes & Cancer. 28:66-76. Jin Y, Mertens F. 1993. Chromosome abnormalities in oral squamous cell carcinomas. Eur J Cancer B Oral Oncol 29B(4):257-63. John A, Tuszynski G. 2001. The role of matrix metalloproteinases in tumor angiogenesis and tumor metastasis. Pathol Oncol Res 7(1): 14-23. Johns M M , 3rd, Westra WH, Califano JA, Eisele D, Koch WM, Sidransky D. 1996. Allelotype of salivary gland tumors. Cancer Res 56(5): 1151-4. Jones PA, Baylin SB. 2002. The fundamental role of epigenetic events in cancer. Nat Rev Genet 3(6):415-28. Jong K, Marchiori E, Meijer G, Van Der Vaart A, Ylstra B. 2004. Breakpoint identification and smoothing of array comparative genomic hybridization data. Bioinformatics. Kallio JP, Mahlamaki EH, Helin H, Karhu R, Kellokumpu-Lehtinen P, Tammela TL. 2004. Chromosomal gains and losses detected by comparative genomic hybridization and proliferation activity in renal cell carcinoma. Scand J Urol Nephrol 38(3):225-30. Kallioniemi A, Kallioniemi OP, Sudar D, Rutovitz D, Gray JW, Waldman F, Pinkel D. 1992. Comparative genomic hybridization for molecular cytogenetic analysis of solid tumors. Science 258(5083):818-21. Karabulut A, Reibel J, Therkildsen MH, Praetorius F, Nielsen HW, Dabelsteen E. 1995. Observer variability in the histologic assessment of oral premalignant lesions. J Oral Pathol Med 24(5): 198-200. Karolchik D, Baertsch R, Diekhans M , Furey TS, Hinrichs A, Lu YT, Roskin K M , Schwartz M, Sugnet CW, Thomas DJ and others. 2003. The UCSC Genome Browser Database. Nucleic Acids Res 31(l):51-4. Kayahara H, Yamagata H, Tanioka H, Miki T, Hamakawa H. 2001. Frequent loss of heterozygosity at 3p25-p26 is associated with invasive oral squamous cell carcinoma. Journal of Human Genetics. 46(6):335-41. Kent WJ, Sugnet CW, Furey TS, Roskin K M , Pringle TH, Zahler A M , Haussler D. 2002. The human genome browser at UCSC. Genome Res 12(6):996-1006. 116 Killary A M , Wolf ME, Giambernardi TA, Naylor SL. 1992. Definition of a tumor suppressor locus within human chromosome 3p21-p22. Proc Natl Acad Sci U S A 89(22):10877-81. Kiyota A, Shintani S, Mihara M , Nakahara Y, Ueyama Y, Matsumura T, Tachikawa T, Wong DT. 2002. Anti-epidermal growth factor receptor monoclonal antibody 225 upregulates p27(KIPl) and pl5(INK4B) and induces G l arrest in oral squamous carcinoma cell lines. Oncology 63(l):92-8. Komiyama T, Matsumura K, Tsuchida N. 1997. Comparison of DNA copy numbers in original oral squamous cell carcinomas and corresponding cell lines by comparative genomic hybridization. Jpn J Cancer Res 88(5):476-83. Krzywinski M , Bosdet I, Smailus D, Chiu R, Mathewson C, Wye N, Barber S, Brown-John M , Chan S, Chand S and others. 2004. A set of BAC clones spanning the human genome. Nucleic Acids Res 32(12):3651-60. Kuroki T, Tajima Y, Matsuo K, Kanematsu T. 2005. Genetic alterations in gallbladder carcinoma. Surg Today 35(2):101-5. Lager DJ, Slagel DD, Palechek PL. 1994. The expression of epidermal growth factor receptor and transforming growth factor alpha in renal cell carcinoma. Mod Pathol 7(5) :544-8. Lambros MB, Fiegler H, Jones A, Gorman P, Roylance RR, Carter NP, Tomlinson IP. 2005. Analysis of ovarian cancer cell lines using array-based comparative genomic hybridization. J Pathol 205(1 ):29-40. Latif F, Fivash M, Glenn G, Tory K, Orcutt ML, Hampsch K, Delisio J, Lerman M, Cowan J, Beckett M and others. 1992. Chromosome 3p deletions in head and neck carcinomas: statistical ascertainment of allelic loss. Cancer Research. 52(6):1451-6. Latil A, Cussenot O, Fournier G, Driouch K, Lidereau R. 1997. Loss of heterozygosity at chromosome 16q in prostate adenocarcinoma: identification of three independent regions. Cancer Res 57(6): 1058-62. Lee Jl, Sqria JC, Hassan K, Liu D, Tang X, El-Naggar A, Hong WK, Mao L. 2001. Loss of Fhit expression is a predictor of poor outcome in tongue cancer. Cancer Research. 61 (3):837-41. Li A, Tian X, Sung SW, Somlo S. 2003. Identification of two novel polycystic kidney disease-1-like genes in human and mouse genomes. Genomics 81 (6):596-608. 117 Lin SC, Chen YJ, Kao SY, Hsu MT, Lin CH, Yang SC, Liu TY, Chang KW. 2002. Chromosomal changes in betel-associated oral squamous cell carcinomas and their relationship to clinical parameters. Oral Oncol 38(3):266-73. Lin SC, Chung MY, Huang JW, Shieh TM, Liu CJ, Chang KW. 2004a. Correlation between functional genotypes in the matrix metalloproteinases-1 promoter and risk of oral squamous cell carcinomas. J Oral Pathol Med 33(6):323-6. Lin SC, Lo SS, Liu CJ, Chung MY, Huang JW, Chang KW. 2004b. Functional genotype in matrix metalloproteinases-2 promoter is a risk factor for oral carcinogenesis. J Oral Pathol Med 33(7):405-9. Lindblad-Toh K, Tanenbaum DM, Daly MJ, Winchester E, Lui WO, Villapakkam A, Stanton SE, Larsson C, Hudson TJ, Johnson BE and others. 2000. Loss-of-heterozygosity analysis of small-cell lung carcinomas using single-nucleotide polymorphism arrays. Nat Biotechnol 18(9):1001-5. Lucito R, Healy J, Alexander J, Reiner A, Esposito D, Chi M, Rodgers L, Brady A, Sebat J, Troge J and others. 2003. Representational oligonucleotide microarray analysis: a high-resolution method to detect genome copy number variation. Genome Res 13(10):2291-305. Lyng H, Beigi M , Svendsrud DH, Brustugun OT, Stokke T, Kristensen GB, Sundfor K, Skjonsberg A, De Angelis PM. 2004. Intratumor chromosomal heterogeneity in advanced carcinomas of the uterine cervix. Int J Cancer 111(3):358-66. Maestro R, Gasparotto D, Vukosavljevic T, Barzan L, Sulfaro S, Boiocchi M . 1993. Three discrete regions of deletion at 3p in head and neck cancers. Cancer Research. 53(23):5775-9. Mao L. 1997. Leukoplakia: molecular understanding of pre-malignant lesions and implications for clinical management. Mol Med Today 3(10):442-8. Mao L, Fan YH, Lotan R, Hong WK. 1996a. Frequent abnormalities of FHIT, a candidate tumor suppressor gene, in head and neck cancer cell lines. Cancer Res 56(22):5128-31. Mao L, Hong WK, Papadimitrakopoulou VA. 2004. Focus on head and neck cancer. Cancer Cell 5(4):311-6. Mao L, Lee JS, Fan YH, Ro JY, Batsakis JG, Lippman S, Hittelman W, Hong WK. 1996b. Frequent microsatellite alterations at chromosomes 9p21 and 3pl4 in oral premalignant lesions and their value in cancer risk assessment. Nat Med 2(6):682-5. 118 Matsuzaki H, Loi H, Dong S, Tsai Y Y , Fang J, Law J, Di X, Liu WM, Yang G, Liu G and others. 2004. Parallel genotyping of over 10,000 SNPs using a one-primer assay on a high-density oligonucleotide array. Genome Res 14(3):414-25. McCarthy RP, Zhang S, Bostwick DG, Qian J, Eble JN, Wang M , Lin H, Cheng L. 2004. Molecular genetic evidence for different clonal origins of epithelial and stromal components of phyllodes tumor of the prostate. Am J Pathol 165(4): 1395-400. McPherson JD, Marra M , Hillier L, Waterston RH, Chinwalla A, Wallis J, Sekhon M , Wylie K, Mardis ER, Wilson RK and others. 2001. A physical map of the human genome. Nature 409(6822):934-41. Micci F, Teixeira MR, Haugom L, Kristensen G, Abeler V M , Heim S. 2004. Genomic aberrations in carcinomas of the uterine corpus. Genes Chromosomes Cancer 40(3):229-46. Mitelman F, Johansson B, Mertens F. 2003. Mitelman Database of Chromosome Aberrations in Cancer. Miyai K, Furugen Y, Matsumoto T, Iwabuchi K, Hirose S, Kinoshita K, Fujii H. 2004. Loss of heterozygosity analysis in uterine cervical adenocarcinoma. Gynecol Oncol 94(1): 115-20. Miyamoto R, Uzawa N, Nagaoka S, Hirata Y, Amagasa T. 2003. Prognostic significance of cyclin DI amplification and overexpression in oral squamous cell carcinomas. Oral Oncol 39(6):610-8. Nessling M , Richter K, Schwaenen C, Roerig P, Wrobel G, Wessendorf S, Fritz B, Bentz M , Sinn HP, Radlwimmer B and others. 2005. Candidate genes in breast cancer revealed by microarray-based comparative genomic hybridization of archived tissue. Cancer Res 65(2):439-47. Ng IO, Xiao L, Lam K Y , Yuen PW, Ng M . 2000. Microsatellite alterations in squamous cell carcinoma of the head and neck - clustering of loss of heterozygosity in a distinct subset. Oral Oncol 36(5):484-90. Nishimoto IN, Pinheiro NA, Rogatto SR, Carvalho A L , Simpson AJ, Caballero OL, Kowalski LP. 2004. Cyclin DI gene polymorphism as a risk factor for squamous cell carcinoma of the upper aerodigestive system in non-alcoholics. Oral Oncol 40(6):604-10. Nosaka K, Onozuka M , Kakazu N, Hibi S, Nishimura H, Nishino H, Abe T. 2001. Isolation and characterization of a human thiamine pyrophosphokinase cDNA. Biochim Biophys Acta 1517(2):293-7. 119 Oga A, Kong G, Tae K, Lee Y, Sasaki K. 2001. Comparative genomic hybridization analysis reveals 3q gain resulting in genetic alteration in 3q in advanced oral squamous cell carcinoma. Cancer Genet Cytogenet 127(l):24-9. Ogawara K, Miyakawa A, Shiba M, Uzawa K, Watanabe T, Wang XL, Sato T, Kubosawa H, Kondo Y, Tanzawa H. 1998. Allelic loss of chromosome 13ql4.3 in human oral cancer: correlation with lymph node metastasis. Int J Cancer 79(4):312-7. Ohta M, Inoue H, Cotticelli MG, Kastury K, Baffa R, Palazzo J, Siprashvili Z, Mori M, McCue P, Druck T and others. 1996. The FHIT gene, spanning the chromosome 3pl4.2 fragile site and renal carcinoma-associated t(3;8) breakpoint, is abnormal in digestive tract cancers. Cell 84(4):587-97. Okafuji M , Ita M, Hayatsu Y, Shinozaki F, Oga A, Sasaki K. 1999. Identification of genetic aberrations in cell lines from oral squamous cell carcinomas by comparative genomic hybridization. J Oral Pathol Med 28(6):241-5. Okafuji M , Ita M , Oga A, Hayatsu Y, Matsuo A, Shinzato Y, Shinozaki F, Sasaki K. 2000. The relationship of genetic aberrations detected by comparative genomic hybridization to DNA ploidy and tumor size in human oral squamous cell carcinomas. J Oral Pathol Med 29(5):226-31. Ono K, Miyakawa A, Fukuda M , Shiiba M , Uzawa K, Watanabe T, Miya T, Yokoe H, Imai Y, Tanzawa H. 1999. Allelic loss on the short arm of chromosome 8 in oral squamous cell carcinoma. Oncol Rep 6(4):785-9. Osoegawa K, Mammoser AG, Wu C, Frengen E, Zeng C, Catanese JJ, de Jong PJ. 2001. A bacterial artificial chromosome library for sequencing the complete human genome. Genome Res ll(3):483-96. Pace HC, Garrison PN, Robinson AK, Barnes LD, Draganescu A, Rosier A, Blackburn GM, Siprashvili Z, Croce CM, Huebner K and others. 1998. Genetic, biochemical, and crystallographic characterization of Fhit-substrate complexes as the active signaling form of Fhit. Proc Natl Acad Sci U S A 95(10):5484-9. Papadimitrakopoulou VA, Hong WK, Lee JS, Martin JW, Lee JJ, Batsakis JG, Lippman SM. 1997. Low-dose isotretinoin versus beta-carotene to prevent oral carcinogenesis: long-term follow-up. J Natl Cancer Inst 89(3):257-8. Papadimitrakopoulou VA, Izzo J, Mao L, Keck J, Hamilton D, Shin DM, El-Naggar A, den Hollander P, Liu D, Hittelman WN and others. 2001. Cyclin D l and pi6 alterations in advanced premalignant lesions of the upper aerodigestive tract: role in response to chemoprevention and cancer development. Clin Cancer Res 7(10):3127-34. 120 Paradiso A, Ranieri G, Stea B, Zito A, Zehbe I, Tommasino M , Grammatica L, De Lena M . 2004. Altered pl6INK4a andFhit expression in carcinogenesis and progression of human oral cancer. Int J Oncol 24(2):249-55. Park SW, Ludes-Meyers J, Zimonjic DB, Durkin M E , Popescu NC, Aldaz C M . 2004. Frequent downregulation and loss of WWOX gene expression in human hepatocellular carcinoma. Br J Cancer 91(4):753-9. Partridge M , Emilion G, Langdon JD. 1996. L O H at 3p correlates with a poor survival in oral squamous cell carcinoma. Br J Cancer 73(3):366-71. Partridge M , Emilion G, Pateromichelakis S, Phillips E , Langdon J. 1999. Location of candidate tumour suppressor gene loci at chromosomes 3p, 8p and 9p for oral squamous cell carcinomas. Int J Cancer 83(3):318-25. Partridge M , Kiguwa S, Langdon JD. 1994. Frequent deletion of chromosome 3p in oral squamous cell carcinoma. Eur J Cancer B Oral Oncol 30B(4):248-51. Partridge M , Li SR, Pateromichelakis S, Francis R, Phillips E , Huang X H , Tesfa-Selase F, Langdon JD. 2000. Detection of minimal residual cancer to investigate why oral tumors recur despite seemingly adequate treatment. Clin Cancer Res 6(7):2718-25. Penkov D, Ni R, Else C, Pinol-Roma S, Ramirez F, Tanaka S. 2000. Cloning of a human gene closely related to the genes coding for the c-myc single-strand binding proteins. Gene 243(1-2):27-36. Pettus JA, Cowley BC, Maxwell T, Milash B, Stephenson RA, Rohr LR, Hoff C, Brothman AR. 2004. Multiple abnormalities detected by dye reversal genomic microarrays in prostate cancer: a much greater sensitivity than conventional cytogenetics. Cancer Genet Cytogenet 154(2): 110-8. Pindborg JJ. 1997. Histological Typing of Cancer and Precancer of the Oral Mucosa. International Histological Classification of Tumours:21-31. Pinkel D, Segraves R, Sudar D, Clark S, Poole I, Kowbel D, Collins C, Kuo WL, Chen C, Zhai Y and others. 1998. High resolution analysis of DNA copy number variation using comparative genomic hybridization to microarrays. Nat Genet 20(2):207-11. Pollack JR, Perou C M , Alizadeh A A , Eisen MB, Pergamenschikov A, Williams CF, Jeffrey SS, Botstein D, Brown PO. 1999. Genome-wide analysis of DNA copy-number changes using cDNA microarrays. Nat Genet 23(l):41-6. Puri SK, Si L, Fan CY, Hanna E. 2005. Aberrant promoter hypermethylation of multiple genes in head and neck squamous cell carcinoma. Am J Otolaryngol 26(1): 12-7. 121 Raschke S, Balz V, Efferth T, Schulz WA, Florl AR. 2005. Homozygous deletions of CDKN2A caused by alternative mechanisms in various human cancer cell lines. Genes Chromosomes Cancer 42(l):58-67. Reith A, Sudbo J. 2002. Impact of genomic instability in risk assessment and chemoprevention of oral premalignancies. Int J Cancer 101(3):205-9. Rikimaru K, Tadokoro K, Yamamoto T, Enomoto S, Tsuchida N. 1992. Gene amplification and overexpression of epidermal growth factor receptor in squamous cell carcinoma of the head and neck. Head Neck 14(1):8-13. Rosin MP, Cheng X, Poh C, Lam WL, Huang Y, Lovas J, Berean K, Epstein JB, Priddy R, Le ND and others. 2000. Use of allelic loss to predict malignant risk for low-grade oral epithelial dysplasia. Clin Cancer Res 6(2):357-62. Rosin MP, Lam WL, Poh C, Le ND, Li RJ, Zeng T, Priddy R, Zhang L. 2002. 3pl4 and 9p21 loss is a simple tool for predicting second oral malignancy at previously treated oral cancer sites. Cancer Res 62(22):6447-50. Rowley H, Jones A, Spandidos D, Field J. 1996. Definition of a tumor suppressor gene locus on the short arm of chromosome 3 in squamous cell carcinoma of the head and neck by means of microsatellite markers. Arch Otolaryngol Head Neck Surg 122(5):497-501. Roz L, Wu CL, Porter S, Scully C, Speight P, Read A, Sloan P, Thakker N. 1996. Allelic imbalance on chromosome 3p in oral dysplastic lesions: an early event in oral carcinogenesis. Cancer Res 56(6): 1228-31. Rush LJ, Raval A, Funchain P, Johnson AJ, Smith L, Lucas DM, Bembea M, Liu TH, Heerema NA, Rassenti L and others. 2004. Epigenetic profiling in chronic lymphocytic leukemia reveals novel methylation targets. Cancer Res 64(7):2424-33. Sanchez-Cespedes M , Esteller M, Wu L, Nawroz-Danish H, Yoo GH, Koch WM, Jen J, Herman JG, Sidransky D. 2000. Gene promoter hypermethylation in tumors and serum of head and neck cancer patients. Cancer Res 60(4):892-5. Santini J, Formento JL, Francoual M, Milano G, Schneider M, Dassonville O, Demard F. 1991. Characterization, quantification, and potential clinical value of the epidermal growth factor receptor in head and neck squamous cell carcinomas. Head Neck 13(2): 132-9. Sard L, Accornero P, Tornielli S, Delia D, Bunone G, Campiglio M, Colombo MP, Gramegna M, Croce CM, Pierotti MA and others. 1999. The tumor-suppressor gene FHIT is involved in the regulation of apoptosis and in cell cycle control. Proc Natl Acad Sci U S A 96(15):8489-92. 122 Scholnick SB, Richter TM. 2003. The role of CSMD1 in head and neck carcinogenesis. Genes Chromosomes Cancer 38(3):281-3. Schulz HL, Stohr H, Weber BH. 2002. Characterization of three novel isoforms of the metabotrobic glutamate receptor 7 (GRM7). Neurosci Lett 326(l):37-40. Sciubba JJ. 2001. Oral cancer. The importance of early diagnosis and treatment. Am J Clin Dermatol 2(4):239-51. Shapiro GI, Park JE, Edwards CD, Mao L, Merlo A, Sidransky D, Ewen ME, Rollins BJ. 1995. Multiple mechanisms of pl6INK4A inactivation in non-small cell lung cancer cell lines. Cancer Res 55(24):6200-9. Sherwood JB, Shivapurkar N , Lin WM, Ashfaq R, Miller DS, Gazdar AF, Muller CY. 2000. Chromosome 4 deletions are frequent in invasive cervical cancer and differ between histologic variants. Gynecol Oncol 79(l):90-6. Shintani S, Mihara M, Ueyama Y, Matsumura T, Wong DT. 2001. Cyclin DI overexpression associates with radiosensitivity in oral squamous cell carcinoma. Int J Cancer 96(3): 159-65. Silverman S, Jr., Gorsky M, Lozada F. 1984. Oral leukoplakia and malignant transformation. A follow-up study of 257 patients. Cancer 53(3):563-8. Simpson PT, Reis-Filho JS, Gale T, Lakhani SR. 2005. Molecular evolution of breast cancer. J Pathol 205(2):248-54. Siprashvili Z, Sozzi G, Barnes LD, McCue P, Robinson AK, Eryomin V, Sard L, Tagliabue E, Greco A, Fusetti L and others. 1997. Replacement of Fhit in cancer cells suppresses tumorigenicity. Proc Natl Acad Sci U S A 94(25): 13771-6. Slaughter DP, Southwick HW, Smejkal W. 1953. Field cancerization in oral stratified squamous epithelium; clinical implications of multicentric origin. Cancer 6(5):963-8. Snijders A M , Nowak N, Segraves R, Blackwood S, Brown N, Conroy J, Hamilton G, Hindle AK, Huey B, Kimura K and others. 2001. Assembly of microarrays for genome-wide measurement of DNA copy number. Nat Genet 29(3):263-4. Soderlund C, Longden I, Mott R. 1997. FPC: a system for building contigs from restriction fingerprinted clones. Comput Appl Biosci 13(5):523-35. Solinas-Toldo S, Lampel S, Stilgenbauer S, Nickolenko J, Benner A, Dohner H, Cremer T, Lichter P. 1997. Matrix-based comparative genomic hybridization: biochips to screen for genomic imbalances. Genes Chromosomes Cancer 20(4):399-407. 123 Specht K, Haralambieva E, Bink K, Kremer M, Mandl-Weber S, Koch I, Tomer R, Hofler H, Schuuring E, Kluin PM and others. 2004. Different mechanisms of cyclin D l overexpression in multiple myeloma revealed by fluorescence in situ hybridization and quantitative analysis of mRNA levels. Blood 104(4): 1120-6. Squire JA, Bayani J, Luk C, Unwin L, Tokunaga J, MacMillan C, Irish J, Brown D, Gullane P, Kamel-Reid S. 2002. Molecular cytogenetic analysis of head and neck squamous cell carcinoma: By comparative genomic hybridization, spectral karyotyping, and expression array analysis. Head Neck 24(9):874-87. Sreekantaiah C, Rao PH, Xu L, Sacks PG, Schantz SP, Chaganti RS. 1994. Consistent chromosomal losses in head and neck squamous cell carcinoma cell lines. Genes Chromosomes Cancer ll(l):29-39. Steinhart H, Bohlender J, Iro H, Jung V, Constantinidis J, Gebhart E, Pahl S, Urbschat S. 2001. DNA amplification on chromosome 7q in squamous cell carcinoma of the tongue. Int J Oncol 19(4):851-5. Sternlicht MD, Werb Z. 2001. How matrix metalloproteinases regulate cell behavior. Annu Rev Cell Dev Biol 17:463-516. Storchova Z, Pellman D. 2004. From polyploidy to aneuploidy, genome instability and cancer. Nat Rev Mol Cell Biol 5(l):45-54. Sun PC, Uppaluri R, Schmidt AP, Pashia ME, Quant EC, Sunwoo JB, Gollin SM, Scholnick SB. 2001. Transcript map of the 8p23 putative tumor suppressor region. Genomics 75(1-3): 17-25. Sunwoo JB, Sun PC, Gupta VK, Schmidt AP, El-Mofty S, Scholnick SB. 1999. Localization of a putative tumor suppressor gene in the sub-telomeric region of chromosome 8p. Oncogene 18(16):2651-5. Sutinen M, Kainulainen T, Hurskainen T, Vesterlund E, Alexander JP, Overall CM, Sorsa T, Salo T. 1998. Expression of matrix metalloproteinases (MMP-1 and -2) and their inhibitors (TIMP-1, -2 and -3) in oral lichen planus, dysplasia, squamous cell carcinoma and lymph node metastasis. Br J Cancer 77(12):2239-45. Sy SM, Wong N, Lai PB, To KF, Johnson PJ. 2004. Regional over-representations on chromosomes lq, 3q and 7q in the progression of hepatitis B virus-related hepatocellular carcinoma. Mod Pathol. Tabor MP, Braakhuis BJ, van der Wal JE, van Diest PJ, Leemans CR, Brakenhoff RH, Kummer JA. 2003. Comparative molecular and histological grading of epithelial dysplasia of the oral cavity and the oropharynx. J Pathol 199(3):354-60. 124 Tada T, Watanabe T, Kanazawa T, Kazama S, Koketsu S, Nagawa H. 2004. Genetic characterization of colorectal cancers in young patients based on chromosomal loss and microsatellite instability. Scand J Gastroenterol 39(11):1134-40. Takada H, Imoto I, Tsuda H, Sonoda I, Ichikura T, Mochizuki H, Okanoue T, Inazawa J. 2005. Screening of DNA copy-number aberrations in gastric cancer cell lines by array-based comparative genomic hybridization. Cancer Sci 96(2): 100-10. Timmermann S, Hinds PW, Munger K. 1997. Elevated activity of cyclin-dependent kinase 6 in human squamous cell carcinoma lines. Cell Growth Differ 8(4):361-70. Timmermann S, Hinds PW, Munger K. 1998. Re-expression of endogenous pl6ink4a in oral squamous cell carcinoma lines by 5-aza-2'-deoxycytidine treatment induces a senescence-like state. Oncogene 17(26):3445-53. Tobita K, Kijima H, Dowaki S, Kashiwagi H, Ohtani Y, Oida Y, Yamazaki H, Nakamura M, Ueyama Y, Tanaka M and others. 2003. Epidermal growth factor receptor expression in human pancreatic cancer: Significance for liver metastasis. Int J Mol Med ll(3):305-9. Tong BC, Dhir K, Ha PK, Westra WH, Alter BP, Sidransky D, Koch WM, Califano JA. 2004a. Use of single nucleotide polymorphism arrays to identify a novel region of loss on chromosome 6q in squamous cell carcinomas of the oral cavity. Head Neck 26(4):345-52. Tong CY, Hui AB, Yin XL, Pang JC, Zhu XL, Poon WS, Ng HK. 2004b. Detection of oncogene amplifications in medulloblastomas by comparative genomic hybridization and array-based comparative genomic hybridization. J Neurosurg Spine 100(2):187-93. Toomes C, Jackson A, Maguire K, Wood J, Gollin S, Ishwad C, Paterson I, Prime S, Parkinson K, Bell S and others. 2003. The presence of multiple regions of homozygous deletion at the CSMD1 locus in oral squamous cell carcinoma question the role of CSMD1 in head and neck carcinogenesis. Genes Chromosomes Cancer 37(2):132-40. Trovato M , Ulivieri A, Dominici R, Ruggeri RM, Vitarelli E, Benvenga S, Barresi G, Trimarchi F, Brunetti E, Vecchione A and others. 2004. Clinico-pathological significance of cell-type-specific loss of heterozygosity on chromosome 7q21: analysis of 318 microdissected thyroid lesions. Endocr Relat Cancer 11(2):365-76. Ueda T, Komiya A, Suzuki H, Shimbo M , Sakamoto S, Imamoto T, Akakura K, Shiraishi T, Ichikawa T. 2005. Loss of heterozygosity on chromosome 2 in Japanese patients with prostate cancer. Prostate. 125 Ulanovski D, Stern Y, Roizman P, Shpitzer T, Popovtzer A, Feinmesser R. 2004. Expression of EGFR and Cerb-B2 as prognostic factors in cancer of the tongue. Oral Oncol 40(5):532-7. Uzawa N, Akanuma D, Negishi A, Iwaki H, Uzawa Y, Amagasa T, Yoshida MA. 2001. Homozygous deletions on the short arm of chromosome 3 in human oral squamous cell carcinomas. Oral Oncol 37(4):351-6. Verbeek W, Schulten HJ, Sperling M, Tiesmeier J, Stoop H, Dinjens W, Looijenga L, Wormann B, Fuzesi L, Donhuijsen K. 2004. Rectal adenocarcinoma with choriocarcinomatous differentiation: clinical and genetic aspects. Hum Pathol 35(ll):1427-30. Vokes EE, Weichselbaum RR, Lippman SM, Hong WK. 1993. Head and neck cancer. N Engl J Med 328(3): 184-94. Waber PG, Lee NK, Nisen PD. 1996. Frequent allelic loss at chromosome arm 3p is distinct from genetic alterations of the Von-Hippel Lindau tumor suppressor gene in head and neck cancer. Oncogene 12(2):365-9. Wang XL, Uzawa K, Miyakawa A, Shiiba M , Watanabe T, Sato T, Miya T, Yokoe H, Tanzawa H. 1998. Localization of a tumour-suppressor gene associated with human oral cancer on 7q31.1. Int J Cancer 75(5):671-4. Warnakulasuriya S. 2001. Histological grading of oral epithelial dysplasia: revisited. J Pathol 194(3) :294-7. Watson JE, Doggett NA, Albertson DG, Andaya A, Chinnaiyan A, van Dekken H, Ginzinger D, Haqq C, James K, Kamkar S and others. 2004a. Integration of high-resolution array comparative genomic hybridization analysis of chromosome 16q with expression array data refines common regions of loss at 16q23-qter and identifies underlying candidate tumor suppressor genes in prostate cancer. Oncogene 23(19):3487-94. Watson SK, deLeeuw RJ, Ishkanian AS, Malloff CA, Lam WL. 2004b. Methods for high throughput validation of amplified fragment pools of BAC DNA for constructing high resolution CGH arrays. BMC Genomics 5(1):6. Weber RG SM, Born IA, Joos S, Ludwig Cobbers JMJ, Hofele C, Reifenberger G, Zoller JE, Lichter P. 1998. Recurrent Chromosomal Imbalances Detected in Biopsy Material from Oral Premalignant and Malignant Lesions by Combined Tissue Microdissection, Universal DNA Amplification, and Comparative Genomic Hybridization. American Journal of Pathology 153(l):295-303. Welborn J, Jenks H, Taplett J, Walling P. 2004. High-grade neuroendocrine carcinomas display unique cytogenetic aberrations. Cancer Genet Cytogenet 155(1):33-41. 126 Werner JA, Rathcke 10, Mandic R. 2002. The role of matrix metalloproteinases in squamous cell carcinomas of the head and neck. Clin Exp Metastasis 19(4):275-82. Wheeler DL, Church DM, Federhen S, Lash AE, Madden TL, Pontius JU, Schuler GD, Schriml L M , Sequeira E, Tatusova TA and others. 2003. Database resources of the National Center for Biotechnology. Nucleic Acids Res 31(l):28-33. Wolff E, Girod S, Liehr T, Vorderwulbecke U, Ries J, Steininger H, Gebhart E. 1998. Oral squamous cell carcinomas are characterized by a rather uniform pattern of genomic imbalances detected by comparative genomic hybridisation. Oral Oncol 34(3): 186-90. Wong SC, Chan JK, Lee KC, Hsiao WL. 2001. Differential expression of pl6/p21/p27 and cyclin D1/D3, and their relationships to cell proliferation, apoptosis, and tumour progression in invasive ductal carcinoma of the breast. J Pathol 194(1):35-42. Wu CL, Roz L, McKown S, Sloan P, Read AP, Holland S, Porter S, Scully C, Paterson I, Tavassoli M and others. 1999. DNA studies underestimate the major role of CDKN2A inactivation in oral and oropharyngeal squamous cell carcinomas. Genes Chromosomes Cancer 25(l):16-25. Wu CL, Roz L, Sloan P, Read AP, Holland S, Porter S, Scully C, Speight PM, Thakker N . 1997. Deletion mapping defines three discrete areas of allelic imbalance on chromosome arm 8p in oral and oropharyngeal squamous cell carcinomas. Genes Chromosomes Cancer 20(4):347-53. Wu CL, Sloan P, Read AP, Harris R, Thakker N. 1994. Deletion mapping on the short arm of chromosome 3 in squamous cell carcinoma of the oral cavity. Cancer Res 54(24):6484-8. Yamamoto N, Mizoe J, Numasawa H, Tsujii H, Shibahara T, Noma H. 2003. Allelic loss on chromosomes 2q, 3p and 21q: possibly a poor prognostic factor in oral squamous cell carcinoma. Oral Oncol 39(8):796-805. Yuan L, Shan J, De Risi D, Broome J, Lovecchio J, Gal D, Vinciguerra V, Xu HP. 1999. Isolation of a novel gene, TSP50, by a hypomethylated DNA fragment in human breast cancer. Cancer Res 59(13):3215-21. Yuasa T, Takakura A, Denker BM, Venugopal B, Zhou J. 2004. Polycystin-lL2 is a novel G-protein-binding protein. Genomics 84(1): 126-38. Zabarovsky ER, Lerman MI, Minna JD. 2002. Tumor suppressor genes on chromosome 3p involved in the pathogenesis of lung and other cancers. Oncogene 21(45):6915-35. 127 Zakrzewska M, Rieske P, Debiec-Rychter M, Zakrzewski K, Polis L, Fiks T, Liberski PP. 2004. Molecular abnormalities in pediatric embryonal brain tumors—analysis of loss of heterozygosity on chromosomes 1, 5, 9, 10, 11, 16, 17 and 22. Clin Neuropathol23(5):209-17. Zar J. 1999. Biostatistical Analysis 4ed.:146-155. Zhang L, Michelsen C, Cheng X, Zeng T, Priddy R, Rosin MP. 1997. Molecular analysis of oral lichen planus. A premalignant lesion? Am J Pathol 151(2):323-7. Zhang L, Poh CF, Lam WL, Epstein JB, Cheng X, Zhang X, Priddy R, Lovas J, Le ND, Rosin MP. 2001. Impact of localized treatment in reducing risk of progression of low-grade oral dysplasia: molecular evidence of incomplete resection. Oral Oncol 37(6):505-12. Zhang L, Rosin MP. 2001. Loss of heterozygosity: a potential tool in management of oral premalignant lesions? J Oral Pathol Med 30(9):513-20. Zhang Z, Gerhard DS, Nguyen L, L i J, Traugott A, Huettner PC, Rader JS. 2005. Fine mapping and evaluation of candidate genes for cervical cancer on 1 lq23. Genes Chromosomes Cancer 43(1):95-103. Zhao X, L i C, Paez JG, Chin K, Janne PA, Chen TH, Girard L, Minna J, Christiani D, Leo C and others. 2004. An integrated view of copy number and allelic alterations in the cancer genome using single nucleotide polymorphism arrays. Cancer Res 64(9):3060-71. Zhou X, Li C, Mok SC, Chen Z, Wong DT. 2004a. Whole genome loss of heterozygosity profiling on oral squamous cell carcinoma by high-density single nucleotide polymorphic allele (SNP) array. Cancer Genet Cytogenet 151(l):82-4. Zhou X, Mok SC, Chen Z, L i Y, Wong DT. 2004b. Concurrent analysis of loss of heterozygosity (LOH) and copy number abnormality (CNA) for oral premalignancy progression using the Affymetrix 10K SNP mapping array. Hum Genet 115(4):327-30. 128 APPENDIX 1: Clinical Information of Specimens Analyzed Using the 3p-Array Age Histological T N M Sample (yrs) Gender Smoker Grade stage 114T 75 F S 2 1 166T 82 F NS 2 2 199T 89 F unknown 2 2 211T 72 M S 1 3 24T 63 F NS 1 1 2T 58 F S 2 2 43T 67 M S 3 1 451T 67 M S 1 3 453T 48 F NS 1 1 469T 75 M S 1 N/A 478T 62 F unknown NS (betel 1 4A 528T 74 M nut) 3 4A 539T 64 F NS 1 4A 566T 78 F unknown N/A N/A 569T 76 F NS 1 3 . 573T 69 M S 2 3 574T 78 F unknown 1 N/A 587T 68 M S 1 0 620T 55 M S 2 1 628T 41 M NS 2 4A 129 APPENDIX 2: 3p Array Clone Set 96 Plate/Well new order July 2003 Band July 2003 ChrStart Fingerprint (RPCI-11) 01-G11 3p-001 N0570H17 01-G12 3p-002 12.1 86136427 N0789F05 01-H01 3p-003 12.1 85926605 N0447J13 01-H02 3p-004 N0259G08 01-H03 3p-005 N0259G08 01-H05 3p-006 N0639P09 01-H06 3p-007 12.1 85112855 N0494I08 01-H07 3p-008 12.1 84928673 N0754J20 01-H08 3p-009 12.1 84921259 N0482K02 01-H09 3p-010 12.1 84729361 N0735B13 01-H10 3p-011 N0276B15 01-H11 3p-012 12.1 84257655 N0731C08 02-A01 3p-013 12.1 84009520 N0472I03 02-A02 3p-014 12.1 83903781 N0412N09 02-A03 3p-015 N0339M21 02-A04 3p-016 12.1 83697556 N0382L10 02-A05 3p-017 N0739C09 02-A06 3p-018 12.2 83138074 N0175J03 02-A07 3p-019 12.2 83101730 N0189B07 02-A08 3p-020 12.2 82895811 N0544A22 02-A09 3p-021 N0657G02 02-A10 3p-022 12.2 82565909 N0001E05 02-A11 3p-023 N0735G06 02-A12 3p-024 12.2 82252566 N0026G10 02-B01 3p-025 N0444J02 02-B03 3p-026 12.2 81662672 N0603J22 02-B04 3p-027 12.2 81576272 N0142L01 02-B05 3p-028 12.2 81307397 N0291P10 02-B06 3p-029 12.2 81341669 N0059E22 02-B07 3p-030 12.2 81087668 N0520D19 02-B08 3p-031 12.2 80946720 N0554B18 02-B10 3p-032 N0755H19 01-B03 3p-033 N0036C14 01-B02 3p-034 12.2 80025983 N0450I19 02-C06 3p-035 12.2 79854980 N0297L21 01-B01 3p-036 12.2 79854980 N0297L21 01-A12 3p-037 12.2 79802538 N0042O01 01-A10 3p-038 12.3 79253880 N0406C14 01-A09 3p-039 12.3 79015796 N0539J20 01-A08 3p-040 12.3 78874632 N0268F24 01-A07 3p-041 12.3 78828508 N0635O13 01-A06 3p-042 N0594C08 01-A05 3p-043 N0347G18 130 01-A04 3p-044 12.3 78217744 N0016M12 01-A03 3p-045 12.3 78124386 N0769N07 01-A02 3p-046 N0765A07 01-A01 3p-047 N0198O02 01-G06 3p-048 N0342C02 01-G05 3p-049 12.3 77488753 N0220O14 01-G04 3p-050 N0653K14 01-G03 3p-051 N0021G05 01-G02 3p-052 N0797J06 01-G01 3p-053 N0003N12 01-F12 3p-054 12.3 76521106 N0536K04 01-F11 3p-055 12.3 76394793 N0264B09 01-F10 3p-056 12.3 76087194 N0484E07 01-F09 3p-057 12.3 75907362 N0089H10 01-F08 3p-058 12.3 75922522 N0058L10 01-F07 3p-059 N0441E02 01-F06 3p-060 12.3 75536844 N0413E06 01-F04 3p-061 N0642N14 01-F03 3p-062 N0718F10 01-F01 3p-063 N0746O08 01-E10 3p-064 12.3 74328750 N0282J12 01-E09 3p-065 13 73914427 N0508O18 01-E08 3p-066 13 73693108 N0020B07 01-E07 3p-067 N0404L21 01-E05 3p-068 13 73189804 N0673E20 01-E04 3p-069 13 73134934 N0610N12 01-E03 3p-070 13 73134934 N0610N12 01-E02 3p-071 N0349E16 01-E01 3p-072 13 72506678 N0328N12 01-D09 3p-073 13 71711525 N0321A23 01-D07 3p-074 13 70958723 N0398F09 . 01-D06 3p-075 13 70854033 N0079P21 01-D05 3p-076 N0056K23 01-D04 3p-077 N0736H02 01-D03 3p-078 13 70483112 N0118O11 01-D02 3p-079 N0767D18 01-D01 3p-080 13 69933080 N0215K24 01-C12 3p-081 N0170F17 01-C11 3p-082 14.1 69552202 N0444P10 01-C10 3p-083 14.1 69417535 N0062G11 01-C09 3p-084 N0152N21 01-C08 3p-085 N0076J08 01-C06 3p-086 14.1 68721081 N0199N21 01-C05 3p-087 14.1 68535723 N0436J20 01-C04 3p-088 N0259O22 01-C03 3p-089 14.1 68323117 N0253K11 01-C01 3p-090 N0384L17 131 0 1 - B 1 2 3p-091 N 0 3 6 4 O 0 9 01-B11 3p-092 14.1 6 7 5 0 3 5 9 2 N0115124 0 1 - B 1 0 3p -093 N 0 5 3 5 G 1 5 0 1 - B 0 7 3p-094 14.1 6 6 7 1 0 7 1 3 N 0 0 2 4 O 1 7 0 1 - B 0 6 3p -095 N 0 8 0 5 G 1 6 0 1 - B 0 5 3p-096 N 0 7 5 1 K 0 5 0 1 - B 0 4 3p-097 14.1 6 6 2 2 6 5 9 3 N 0 5 3 5 L 0 3 02 -C01 3p-098 14.1 6 5 8 0 9 6 6 2 N 0 3 2 0 P 1 1 0 2 - C 0 3 3p -099 14.1 6 5 4 0 8 3 8 3 N 0 7 5 7 A 0 9 0 2 - C 0 4 3p-100 N 0 7 2 5 H 1 3 0 2 - C 0 5 3p-101 14.1 6 5 0 0 7 5 2 0 N 0 5 4 3 A 1 8 0 2 - C 0 8 3p -102 N 0 3 6 7 P 0 8 0 2 - C 0 9 3p -103 14.1 6 4 3 3 4 7 2 9 N 0 0 1 4 D 2 2 0 2 - C 1 0 3p-104 N 0 7 9 5 J 1 9 02 -C11 3p -105 N 0 5 0 7 B 2 1 0 2 - C 1 2 3p-106 N 0 2 8 9 G 1 0 02-D01 3p-107 N 0 4 7 4 A 1 9 0 2 - D 0 2 3p-108 14.2 6 3 6 6 6 1 4 2 N 0 0 5 0 F 2 4 0 2 - D 0 3 3p -109 N 0 5 8 4 G 0 1 0 2 - D 0 4 3p-110 14.2 6 3 4 2 8 5 7 4 N 0 6 7 2 O 0 6 0 2 - D 0 5 3p-111 14.2 6 3 2 6 1 7 5 7 N 0 5 8 5 O 2 1 0 2 - D 0 6 3p -112 14.2 63131951 N 0 6 0 1 G 1 8 0 2 - D 0 7 3p -113 14.2 6 2 9 5 8 9 9 4 N 0 1 2 9 K 2 0 0 2 - D 0 8 3p-114 14.2 6 2 7 9 7 0 3 7 N 0 0 2 4 H 0 1 0 2 - D 1 0 3p -115 14.2 6 2 4 5 5 5 4 3 N 0 1 1 4 P 1 5 02-D11 3p-116 14.2 6 2 1 3 1 3 5 4 N 0 2 3 1 J 1 7 0 2 - D 1 2 3p-117 N 0 0 0 6 O 2 3 02-E01 3p-118 14.2 6 1 9 9 9 0 4 4 N 0 1 5 4 D 0 3 0 2 - E 0 2 3p -119 14.2 6 1 8 6 3 6 6 5 N 0 0 5 8 L 0 9 0 2 - E 0 3 3p-120 14.2 6 1 6 5 8 0 7 4 N 0 7 2 0 N 2 4 0 2 - E 0 4 3p-121 14.2 6 1 5 5 5 8 4 5 N 0 6 1 1 H 1 0 0 2 - E 0 5 3p-122 N 0 6 4 0 B 0 5 0 2 - E 0 6 3p -123 N 0 4 9 5 E 2 3 0 2 - E 0 7 3p-124 14.2 6 1 0 7 2 5 4 6 N 0 1 3 7 N 2 2 0 2 - E 0 9 3p -125 N0468L11 0 2 - E 1 0 3p-126 14.2 60489911 N 0 6 4 1 C 1 7 02-E11 3p-127 N 0 0 2 9 O 2 0 0 2 - E 1 2 3p-128 N 0 1 8 2 G 2 4 0 2 - F 0 2 3p -129 14.2 5 9 8 3 4 4 6 7 N 0 7 6 9 M 2 3 05-A01 3p -130 14.2 5 9 8 3 4 4 6 7 N 0 7 6 9 M 2 3 0 2 - F 0 3 3p-131 N 0 1 7 0 K 1 9 0 4 - E 0 6 3p-132 N 0 1 7 0 K 1 9 0 2 - F 0 5 3p -133 N 0 7 1 9 N 2 2 0 4 - E 0 4 3p-134 N 0 7 1 9 N 2 2 0 2 - F 0 6 3p -135 N0115I11 0 4 - E 0 3 3p-136 N0115I11 0 2 - F 0 7 3p -137 N 0 5 6 7 G 0 2 132 02-F08 3p-138 14.2 59006095 N0734E15 04-E02 3p-139 14.2 59006095 N0734E15 02-F09 3p-140 14.2 58887870 N0057M07 03-B11 3p-141 14.2 58887870 N0057M07 02-F10 3p-142 N0385P13 02-F11 3p-143 14.2 58677137 N0391P04 04-A06 3p-144 14.2 58677137 N0391P04 02-F12 3p-145 N0092A17 04-A04 3p-146 N0092A17 02-G01 3p-147 N0458F14 02-G02 3p-148 N0813L15 02-G03 3p-149 N0316L22 02-G04 3p-150 14.3 57701716 N0332H21 04-A01 3p-151 14.3 57701716 N0332H21 02-G05 3p-152 14.3 57584082 N0680P23 05-E06 3p-153 14.3 57584082 N0680P23 02-G06 3p-154 N0688P04 02-G07 3p-155 N0700D11 02-G08 3p-156 N0767N01 05-E03 3p-157 N0767N01 02-G09 3p-158 14.3 56780966 N0754F19 02-G11 3p-159 14.3 56387454 N0750K11 05-E01 3p-160 14.3 56387454 N0750K11 02-G12 3p-161 14.3 56192945 N0189A01 02-H01 3p-162 N0669E05 02-H02 3p-163 14.3 55983012 N0532I06 02-H03 3p-164 14.3 55807008 N0379L20 05-A05 3p-165 14.3 55807008 N0379L20 04-B05 3p-166 N0372C07 04-F01 3p-167 N0533N15 04-B06 3p-168 14.3 55590746 N0029M01 04-B04 3p-169 14.3 55508998 N0567P18 04-B02 3p-170 N0277M17 04-B01 3p-171 N0064C01 05-E11 3p-172 14.3 54919741 N0096N05 05-E10 3p-173 14.3 54801802 N0015O05 05-E09 3p-174 14.3 54594717 N0554O01 05-E08 3p-175 N0594F16 05-A12 3p-176 N0632O05 05-A11 3p-177 N0122D19 05-A10 3p-178 N0753M04 05-A08 3p-179 N0536P10 05-A07 3p-180 N0417A02 04-E12 3p-181 N0134J14 04-E11 3p-182 N0266E15 04-E10 3p-183 21.1 53238351 N0782N04 04-E09 3p-184 21.1 53320404 N0072H11 133 04-E08 3p-185 N0557M05 04-E07 3p-186 N0178O02 04-A12 3p-187 N0134N13 04-A10 3p-188 N0017O10 04-A09 3p-189 N0447A21 04-A08 3p-190 N0474I07 04-A07 3p-191 21.2 52016695 N0330H06 04-F05 3p-192 N0316H08 04-F04 3p-193 21.2 51775960 N0314A05 04-F03 3p-194 21.2 51616480 N0413B21 04-B10 3p-195 N0665B04 04-B09 3p-196 N0162M15 04-B08 3p-197 21.31 51374604 N0089F17 05-F05 3p-198 N0808I21 05-F04 3p-199 N0095P12 05-F03 3p-200 N0106F04 05-F01 3p-201 N0211E01 05-B06 3p-202 N0646D13 05-B05 3p-203 21.31 50517664 N0804H08 05-B01 3p-204 21.31 49955231 N0493K19 04-F06 3p-205 N0460G14 04-B11 3p-206 N0140N21 04-B12 3p-207 N0171C03 04-F07 3p-208 21.31 48997394 N0804H15 04-F09 3p-209 N0316M24 04-F10 3p-210 N0150N22 06-C10 3p-211 21.31 47155027 N0447D11 06-C09 3p-212 N0677J16 06-C08 3p-213 N0704H13 06-C07 3p-214 N0380M12 06-C06 3p-215 21.31 46843451 N0425J09 06-C05 3p-216 21.31 46755830 N0427P05 06-C04 3p-217 N0538M22 06-C03 3p-218 21.31 46469740 N0509I21 06-C02 3p-219 N0676N17 06-C01 3p-220 N0029H14 06-B12 3p-221 N0458C12 06-B11 3p-222 21.31 46077054 N0793E15 06-B10 3p-223 N0808A02 06-B09 3p-224 N0680N04 06-B08 3p-225 N0307M13 06-B07 3p-226 N0437F20 06-B05 3p-227 21.31 45651411 N0697K23 06-B04 3p-228 N0123N09 06-B03 3p-229 21.31 45458031 N0107O13 06-B02 3p-230 N0047O06 06-B01 3p-231 N0140N09 134 06-A12 3p-232 N0622P17 06-A11 3p-233 N0487I21 06-A10 3p-234 21.31 45013948 N0578F05 06-A09 3p-235 21.31 44908243 N0681O04 06-A08 3p-236 N0674D23 06-A07 3p-237 21.31 44723759 N0272D20 03-A06 3p-238 N0348P10 06-A06 3p-239 N0106E23 06-A05 3p-240 N0753N24 06-A04 3p-241 N0333M05 06-A02 3p-242 N0448P22 06-A01 3p-243 21.33 44162861 N0424N24 05-H12 3p-244 N0445K05 05-H11 3p-245 N0627O21 05-H09 3p-246 N0221D23 05-H08 3p-247 N0723E20 05-H07 3p-248 22.1 43543285 N0606H24 03-A05 3p-249 22.1 43452027 N0024L15 05-D12 3p-250 N0794P05 05-D11 3p-251 N0316I23 05-D10 3p-252 N0117A18 05-D09 3p-253 22.1 43278456 N0188P20 05-D07 3p-254 22.1 43120920 N0625B23 04-H12 3p-255 N0646G12 04-H11 3p-256 22.1 42963135 N0136C24 04-H10 3p-257 N0736F16 04-H08 3p-258 N0312C22 04-H07 3p-259 N0437M11 04-D12 3p-260 22.1 42435367 N0119L02 04-D10 3p-261 N0814E10 04-D09 3p-262 N0698G20 04-D08 3p-263 22.1 41982701 N0193I22 04-D07 3p-264 N0626A01 05-H06 3p-265 22.1 41728499 N0523D22 05-H05 3p-266 22.1 41577833 N0044E08 05-H04 3p-267 22.1 41430754 N0756A10 05-H03 3p-268 22.1 41299743 N0107K23 05-H02 3p-269 22.1 41157668 N0527M19 05-H01 3p-270 N0413K12 05-D06 3p-271 22.1 40789228 N0520A21 05-D05 3p-272 22.1 40666112 N0761N21 05-D03 3p-273 N0809O09 05-D02 3p-274 22.1 40292536 N0391M01 05-D01 3p-275 N0670I16 04-H06 3p-276 22.1 40038176 N0613N24 04-H05 3p-277 N0737L23 04-H04 3p-278 22.2 39837141 N0268J19 135 04-H03 3p-279 N0013C23 04-H02 3p-280 22.2 39652636 N0066G08 04-H01 3p-281 N0365B07 04-D06 3p-282 22.2 39423436 N0667E17 04-D05 3p-283 N0801C10 04-D04 3p-284 N0423B17 04-D01 3p-285 N0623P21 05-G12 3p-286 N0583E05 05-G11 3p-287 N0452P03 05-G10 3p-288 N0134J21 05-G09 3p-289 N0140P20 05-G07 3p-290 N0284F22 05-G08 3p-291 N0284F22 05-C12 3p-292 22.3 38323155 N0090M23 05-C11 3p-293 N0775O14 03-A03 3p-294 N0775O14 05-C09 3p-295 N0091E04 05-C08 3p-296 N0667H18 05-C07 3p-297 N0721J12 04-G12 3p-298 N0802A21 04-G11 3p-299 N0305C24 04-G10 3p-300 22.3 37532207 N0128H05 04-G09 3p-301 N0798L14 04-G08 3p-302 N0285J16 04-G07 3p-303 N0170E06 03-B04 3p-304 N0170E06 04-C12 3p-305 22.3 36852458 N0129K12 04-C10 3p-306 22.3 36638024 N0640L09 04-C09 3p-307 N0709D17 04-C08 3p-308 N0631H24 03-A02 3p-309 N0172A18 05-G06 3p-310 22.3 36095745 N0437B05 05-G03 3p-311 N0774G17 05-G02 3p-312 22.3 35575764 N0380G10 05-C06 3p-313 N0346E13 05-G01 3p-314 N0346E13 05-C04 3p-315 22.3 35375018 N0329N16 05-C03 3p-316 22.3 35258930 N0061J21 05-C02 3p-317 22.3 35228139 N0092I24 05-C01 3p-318 N0518K24 04-G06 3p-319 N0553F16 04-G05 3p-320 23 34977960 N0056P22 04-G02 3p-321 N0316L20 04-G03 3p-322 N0316L20 04-G01 3p-323 N0059J13 04-C06 3p-324 N0408A16 04-C05 3p-325 N0318H08 136 04-C04 3p-326 N0435M24 04-C03 3p-327 N0640M05 04-B07 3p-328 N0151D03 04-C02 3p-329 N0151D03 04-C01 3p-330 N0609H24 05-F12 3p-331 N0521E09 05-F10 3p-332 N0764N08 05-F11 3p-333 N0764N08 05-F08 3p-334 N0150H17 05-F07 3p-335 N0795G20 05-B12 3p-336 N0514O06 05-B11 3p-337 N0033O01 05-B10 3p-338 N0692F18 05-B09 3p-339 N0294O18 05-B08 3p-340 N0421B02 05-B07 3p-341 N0705C20 04-F12 3p-342 N0459C18 06-D04 3p-343 23 32450495 N0524O15 06-D02 3p-344 N0187I16 06-D01 3p-345 24.1 32003419 N0345J06 06-C12 3p-346 N0230N21 06-C11 3p-347 24.1 31631584 N0048E16 06-H12 3p-348 N0671P21 06-H10 3p-349 N0760G13 06-H11 3p-350 N0760G13 06-H09 3p-351 N0167C09 06-H07 3p-352 N0381I06 06-H08 3p-353 N0381I06 06-H06 3p-354 N0727D09 06-H05 3p-355 N0018A10 03-A08 3p-356 23 34492266 N0035C18 03-B08 3p-357 23 34492266 N0035C18 06-H03 3p-358 N0272K09 06-H04 3p-359 N0272K09 06-H02 3p-360 N0717D24 06-H01 3p-361 24.1 29648600 N0009J18 06-G12 3p-362 N0562G23 06-G11 3p-363 N0411F11 06-G10 3p-364 N0077M24 06-G08 3p-365 24.1 28336037 N0301D08 06-G07 3p-366 24.1 28140460 N0539L02 06-G06 3p-367 24.1 27769954 N0057O23 06-G05 3p-368 24.1 27554125 N0222K16 06-G03 3p-369 N0586F17 06-G02 3p-370 N0630B16 06-G01 3p-371 24.1 27349524 N0002G22 06-F12 3p-372 N0342A21 137 06-F11 3p-373 24.1 27178794 N0030A19 06-F10 3p-374 24.1 27003011 N0353I06 06-F09 3p-375 24.2 26717957 N0226E22 04-F11 3p-376 24.2 26520653 N0592A05 06-F08 3p-377 24.2 26520653 N0592A05 06-F07 3p-378 24.2 26359878 N0268I20 06-F06 3p-379 24.2 25893195 N0001C05 06-F05 3p-380 24.2 25680615 N0653E17 06-F04 3p-381 24.2 25469449 N0109D05 06-F03 3p-382 24.2 25253624 N0733H11 06-F02 3p-383 24.2 25029855 N0537O08 06-E12 3p-384 N0689A12 06-E11 3p-385 N0745L10 06-E10 3p-386 N0528H05 06-E09 3p-387 N0763K04 06-E08 3p-388 24.2 23991786 N0118N13 06-E06 3p-389 N0778D16 06-E05 3p-390 N0143O01 06-E04 3p-391 24.3 23207814 N0228H05 06-E03 3p-392 N0703F09 06-E02 3p-393 24.3 22680924 N0053O02 06-E01 3p-394 N0501K22 06-D11 3p-395 N0208G16 06-D10 3p-396 24.3 22209263 N0320M06 06-D09 3p-397 N0644G08 06-D08 3p-398 24.3 21893180 N0587M06 06-D07 3p-399 N0203N02 06-D05 3p-400 N0451A11 07-B04 3p-401 24.3 20979276 N0466A13 07-B03 3p-402 24.3 20827766 N0527L09 07-B02 3p-403 24.3 20635975 N0027J05 07-B01 3p-404 24.3 20489150 N0632N21 07-A12 3p-405 N0669C19 07-A11 3p-406 24.3 20148430 N0197J13 07-A10 3p-407 24.3 20091575 N0809O24 07-A08 3p-408 N0705F24 07-A07 3p-409 N0641E06 07-A06 3p-410 24.3 19229735 N0507A10 07-A05 3p-411 N0566C02 07-A04 3p-412 N0362M23 07-A03 3p-413 N0408A18 07-A02 3p-414 N0566M05 07-A01 3p-415 24.3 18424766 N0054L06 07-B05 3p-416 24.3 18082473 N0554B20 07-B06 3p-417 24.3 17809289 N0008D23 07-B07 3p-418 N0802N17 07-D03 3p-419 24.3 17299996 N0083E07 138 07-D02 3p-420 N0080D24 07-C12 3p-421 24.3 16764377 N0654A18 07-C10 3p-422 25.1 16323060 N0125A16 07-C09 3p-423 25.1 16140472 N0608O08 07-C08 3p-424 N0657N03 07-C06 3p-425 N0317C02 07-C05 3p-426 N0056F22 02-H10 3p-427 25.1 15128467 N0072K12 07-C04 3p-428 25.1 15147208 N0421B21 07-C03 3p-429 25.1 15064381 N0060M01 07-C02 3p-430 N0768A23 07-C01 3p-431 25.1 14651548 N0708D01 07-B11 3p-432 25.1 14236271 N0536I06 07-B10 3p-433 N0320K24 07-E10 3p-434 N0395P16 07-E09 3p-435 N0509B22 02-H09 3p-436 N0488M06 07-E08 3p-437 N0488M06 07-E07 3p-438 N0220D14 07-E06 3p-439 N0011L22 07-E05 3p-440 N0606C06 07-E04 3p-441 25.2 12741771 N0333A02 07-E03 3p-442 25.2 12590778 N0163D23 07-E02 3p-443 25.2 12323574 N0025C10 07-D12 3p-444 25.2 11962174 N0398J15 07-D11 3p-445 25.2 11825108 N0625C09 07-D10 3p-446 25.3 11524143 N0169K17 07-D09 3p-447 25.3 11245310 N0105H19 07-D08 3p-448 N0429F16 07-D07 3p-449 N0680B09 02-H08 3p-450 N0115G03 07-D06 3p-451 N0651I15 07-D05 3p-452 N0094A14 07-D04 3p-453 N0431M22 07-E12 3p-454 N0412A07 07-F01 3p-455 25.3 9667146 N0266J06 07-F02 3p-456 N0810I07 08-C07 3p-457 N0796O13 08-C06 3p-458 25.3 9053942 N0334L22 03-B09 3p-459 25.3 9053942 N0334L22 08-C05 3p-460 25.3 8917082 N0019E08 08-C04 3p-461 N0212O10 08-C03 3p-462 25.3 8376051 N0502K05 08-C02 3p-463 N0427K04 02-H06 3p-464 N0542K24 08-C01 3p-465 N0542K24 08-B12 3p-466 26.1 7745575 N0794G03 139 08-B09 3p-467 N0537O07 08-B08 3p-468 N0812I02 08-B07 3p-469 N0534J21 08-B06 3p-470 N0077I05 03-A10 3p-471 N0077I05 08-B05 3p-472 26.1 6616664 N0529B17 08-B04 3p-473 N0255K15 08-B03 3p-474 26.1 6140905 N0007M24 08-B02 3p-475 26.1 6016133 N0474B13 08-B01 3p-476 26.1 5856605 N0096F18 02-H05 3p-477 26.1 5450383 N0033E18 08-A11 3p-478 26.1 5346972 N0155E14 08-A10 3p-479 N0586C02 08-A09 3p-480 26.1 4916593 N0622A14 08-A08 3p-481 N0608N16 08-A07 3p-482 N0643I08 08-A06 3p-483 26.1 4490451 N0238A09 08-A05 3p-484 N0778H24 08-A04 3p-485 N0800I01 08-A03 3p-486 26.1 4158925 N0453A03 08-A01 3p-487 N0057I04 07-H11 3p-488 N0691I12 07-H09 3p-489 26.2 3818356 N0327M20 07-H10 3p-490 26.2 3818356 N0327M20 07-H08 3p-491 26.2 3754638 N0453F03 07-H07 3p-492 N0367B01 07-H06 3p-493 26.2 3569680 N0090K01 07-H05 3p-494 N0007F04 07-H04 3p-495 N0512O18 07-H03 3p-496 26.2 3274526 N0319I18 07-H01 3p-497 N0772P01 07-G12 3p-498 N0228N13 07-G11 3p-499 26.3 2690182 N0785A07 02-H04 3p-500 26.3 2443360 N0063O01 07-G09 3p-501 N0556O10 07-G08 3p-502 N0567B16 07-G07 3p-503 N0499E07 07-G06 3p-504 N0789M13 07-G05 3p-505 26.3 1962634 N0096G04 07-G04 3p-506 N0201B02 06-G04 3p-507 N0295A13 07-G03 3p-508 26.3 1560711 N0551L04 07-G01 3p-509 N0640N12 07-F12 3p-510 26.3 1424088 N0076O22 07-F11 3p-511 26.3 1309777 N0263P03 07-F10 3p-512 N0432F09 07-F09 3p-513 N0655O17 140 07-F08 3p-514 26.3 1064435 N0392M07 07-F07 3p-515 N0517G20 07-F06 3p-516 N0329A20 07-F05 3p-517 N0645L14 07-F04 3p-518 26.3 768337 N0086C13 07-F03 3p-519 N0440C20 08-D02 3p-520 N0151A04 08-D01 3p-521 26.3 658223 N0775C23 08-C12 3p-522 N0609G18 08-C11 3p-523 N0306H05 03-A09 3p-524 N0806C02 * Clone list reduced to 524 from 535 since publication. 141 A P P E N D I X 3: Clinical Information of Specimens Analyzed Using the Whole Genome S M R T Array Age Tobacco Sample (yrs) Gender usage 117T 34 M NS 123T 63 M S 125T 85 F NS 161T 46 M unknown 24T 63 F NS 486T 67 M S NS (Betel 528T 74 M Nuts) 542T 47 M S 628T 41 M NS 669T 55 M S 792T 42 M unknown 793T 64 M S 794T 59 M NS 800T 78 F unknown 801T 27 M NS 805T 44 M unknown 809T 80 M S 814T 68 F NS 819T 61 M S 836T 40 M S 142 GET -"r' I llll I IIIH in B i M M i K M B U l H niiiLiiiiiiiiiiii.iJiJii iifiiiiimui in i B l S M III llll s H -r rs •-3 S S 3 H 3 -3 c-"3 *— 3 £ "E. c 3 • S i I tan II iii!wiit i nii iiwi iiinB ii :Hrriiiii > i liuuiriitlsiiitS'-li lllllllllll 2 i nil ii mi IIII^ IIIIII i n i in ii III inn » ii II i i i mini 5 IIIIIIB? mini iiimiiD * iiui.ii i i i i i i i i i i iBi i l l i i iiiinnn i l i u m mini N iiiiniiiniiinii ? in mm £ itiuikffittiihw i CM 3 frmiithiriiiini i ffiitiiir i. i riffifrrfiim t ff iniit i nii:»iiiiaiifiiii»liliifi» IIIIIIHIIIIIII l l l l r-M i i i S H i i i s H - i i M S i i i i i i i i m III llll III III f m i n s APPENDIX 5: Supplemental Karyogram of Tumour 117T i B 3 16 ft i = i 17 -• 8 13 18 2 • B 2: 9 14 19 i 10 15 20 21 22 144 iimiiiiiiiiismi iiii si iliSsii! iiii m n n I I I I I I * ii WIIII i IHII i i!tui;1iii.)ii»i)irijhii iiimiiiiii £ illi&BI r v H — i-s e S 3 H **. o £ s-B u E a a 3 y; s a. a, St in i i tn ii t iiumiiu i iiHiiiuinHtSii I I I I I I i i M i i m r a i , * i i i i n i i i i i i n i i i i i i i i u „ iiiiimniiniiii en B t S i S M H M M ii III mini ? D M mn nm inn a iiiiiiiiiiiiiii ? ii IIIIII ? miiiiiiiniiiiii x IIIIIIIIIII IIIIIIIIIIIIIII <M M i n i ) im I I I I I I i h> i n i i n i i i i i i i £ i n IIIIII £ I III a -- r • M U B i i H I l l iiMHSffliHiRifihi oiiiiiniiu^iifoiBiMMhl liBiiBlilSaWlilllilMlil IfifiiiiH JihISrifl iiiiiiiiiHiiimiiiiiiii I H ^ H I ni •••mi «o niiiiiiiiwii - in i i mi ? IIIIII N in APPENDIX 7: Supplemental Karyogram of Matched Normal 123C I m m 2 m 1 3 5? 8 w I 9 5 is' I 10 16 13 18 M . at St » • 14 19 20 21 22 1 i 146 APPENDIX 8: Supplemental Karyogram of Tumour 125T 1 I i IB • H 2 8 f l Bi a 9 10 12 15 17 13 i m 18 I 19 20 21 •5 22 147 APPENDIX 9: Supplemental Karyogram of Tumour 161T I h i fori • • i i I 8 w m 9 i 10 16 1 12 17 13 18 I ft 14 19 = 5=5 15 20 i 21 = 22 148 ui i i i i i i l i i i i i i liitfiiisssiiiiiiii liiHBBillwHtftfli uiiiiilmiiiiilbiimiH in o <N buiiiiiuiiuiiiiiiluiiiusii^ itii]) luai iiimi i Hii i i t i l i^ i^Jimiinif isi i i i i i wiuliTOiiiSiMisJiiisHSi utetiit&mi vnnunuaamttma I iiimi i CO II lllllllllllll CO iniiirBi.asuiii ...i.:;::i::i,i5iijj:j ititliitliitiini liiimBiirjjiUi i : r , « i i ! r n t } ! n ™ II ii m i n t oUttal Siiitriimittii > Hiitiiti '. '. iiwSraiiiii i ii i niii! iii linini : . ; „ . : ; . : : ; , iH'.itmtfil: liiitttJ .... 'ai;; iis-.j J -is an IflMiU IfttMH IIIIMIIIIIIII l l l l IIHI III • • • l l l l & l l l l l l l l l l l WII £ l l l l l l l l l l l ? m i n CM APPENDIX 11: Supplemental Karyogram of Tumour 486T •4 -= I 11 = = 16 > If 1 4 9 • I i 1 8 I 3 i 9 • 14 m i it l=. r: 10 21 22 150 APPENDIX 12: Supplemental Karyogram of Matched Normal 486C s n I f i I §1 Si r - : r*Ti B if n r. I? Si I 8 9 13 18 I 14 • 10 19 15 20 Siit sjj | , i ESS 1. a £ 21 22 I ft I 5J if 151 l l l l l l l l l l l in 1MB i ii iiii£ii*aa!i)ii m i m i i m n i ? H x ir-S-3 3 E 3 S3 st 3 c 3 E "S. c 3 iiiiHi i HW iltiiniiiit i nhism;;;:-;':!: m i mi iiin i »iiiii3aii i iLiiiiiSiiiii; iSi iiirii) MiHBS&i: iisisantUSttfi is . . . . I I I I • ! I l l III 11 BiSHiBiuBi lilllllUMWIIllilWllllg SiiiiSBiiEfii!!! siBMiSiwmifi 3 z w C u Siiliiitiiimiirii i miiisi; i : iwifitSiiitB i ii tiiiii i iiiif iiiwiiiiiiisiiiiii i :iiii;i!i:HSii < IIIIIBIIIHIIIMIIIIIIIII III IMIIIIII •llll (o iiiiii^iiiitiiiiliiiiiiiiii >St"JiSii8;iS o CM iiWiiMiffliBlimimi I * I CN <-r. 'ii iiii'.li.iii; i u l i i i i ) i i i i i S l "iigiKiliSi"imBSn• Itrirn» lllllll £ 1 IIIIIIIIHIIIII x ;;;!.,;:;sss.:V» Bilijlf tjglSsBS N m • m m I*" CM CM liSiliittnllBtllHife i ia i iSt i i i i & i i S K i i m m ii £ mi • II in ? lllllll CM APPENDIX 14: Supplemental Karyogram of Tumour 542T 21 22 X Y 153 n i i i p III! II11 Ml I Hill 111 Bill II1111 tut 1 tu;iH[ii.u[jii^. > -v --— •Jl ^ IIIIII III ^ i IIIIII i mn HI i 11 IIIIII i n i m n in N m i n i i i i i i i i m HID IIIIII mm M S V M M I oo i i i i i i u w III IIIIII I llli II = •a c 3 = IS c JS -5 = 3 c c -s o 00 H . it ro o jrjnMwiffli o II tOMPVtlHI SOTS' "8 APPENDIX 16: Supplemental Karyogram of Tumour 669T 16 17 18 19 20 155 2 IIIIII •VHP t in m i n i ( M l I',!! IIIBIIIIill«IIIIBHIilllll ui yiltM I! vm intuit iii: ' <d&* 111 ti11M9Ptl> 1 tHHSHK ' nu|«a **suM{ngi I I ; K IIIDIII pnpp to -1 IIIIII III I IIIIII " II I II III IIIIII w l l l l l l l l l l l l II 0 I I III Mil' •* Ill III II I ffllJIjJ > -s p i -= s s 3 rt a c — 3 —; = 3 c c -s -4 sC H £11111111 01 IIIIII 111 HU l l l l l l APPENDIX 18: Supplemental Karyogram of Tumour 793T 21 22 X Y 157 i u i i i iU i iu i i d i i i f i i nfi-i i i i i i i i H i l i i i Mi&iStiSiiiiiittiiSii .ti...;-ifi»>iin>i»iH» nm • • i m i i i i i » iiHiaimniii ? iiiniiimi £ iiiimi s H -r o> r-L 3 3 E 3 H E M E 3 rs "3 *» 8 E u "E c 3 i i i i f i i i i n n l i i u u u m i i i r i i iKn i i t i i i i i t ani iiiiiiSiii Uiiuii jHuEsini i&Si n g i I llll III III III 1^1111111 » •S.\ / '"-^i- v:1;a.--t,-.vrjH-' v „ - . ; , -l in nit, i . ; '.'.Eiiriiiiifiiiii a aiiiuiiuuiuSiwiiStBiiiiii • Will .. Mr. i • i u U i i 8 8 f t i i i i B i i i S i M i A M l l l M I um mi i oo II iiimiiiiii ? II II nil ? iniiiE iimin II x S I S n n i B i n B i laintBinWHi i i W H u B S BHUnlinmHi i teB>Sf»ni8 tti^iitlii&ibiHiiliiMii! BJHJiilwijiSHJS i IIII nm ii D HID 11 III II II N III min i III min ii (s. m i in min ^ iininii ^ BlBHfl i nm s <r, 5 z CL, CN < -Si i i i r t t toi i i i i I •BBfirft I R i i * H i H i i i i f t i » i » i ai I a i i 'u i iii ; ii i i i i i f ! i i i i s i i i i i im i i M i i i i i i u u i f / i i i i l i i i t i i i i i i i m i i a i t t i i i i i i i m i l l imiHIIBIIIIHIIIIIIII I I H r I I B S I I I I •ll l l ta I I11 lllllllllll £ Ml 111 III! £ inn N APPENDIX 20: Supplemental Karyogram of Tumour 800T 21 22 X Y 159 APPENDIX 21: Supplemental Karyogram of Tumour 801T iuiiiiiliiuii nffiiSi siirii ismsiii > iii l i imin H IT, O ac u = e E = H •— c E ss -sr o C ss e E a 3 y: C S C M X 5 z CM < Units i I M I IIHWM i iiriiiiinitiifBii : ; i i i :riii i . i ! t i S i t i i t H i -ii m\i r l iRMii i i i i ir . ; II III I IIIIII II i(iiuiim> KS»MiiiiiT, &iSi«S»« »» i* i> > ? I I I B I I I H I I I I x CO IIIIII ii iiimnis IIIIII n iini MIIIIIIII a HSMsi^iiuiiiuiiliiiiS i n 11111111111 cs ifi&iiBuBiuiii Uiihiiiuiiiiiiiii tiiiiitiiiMsm »K»BiB:i!Si»?n »i»*i>roi*»»m N inmnii iHI IIIIII i i is. IflBliiltfMi rTtritiihiiiiiirik i mihiii i 1 iiid IIIIIIIHIIIMIIIII m»iii3Ji & i ii > i l l iS i SiLiiifihSi i tiBiMi I i*MHB»illiafc niiiiiiiimJiiifiiiBliB I IIIIII IIII £ n tnrnmmi «n CN CM S B i t t i r a CM III IIIIII CM APPENDIX 23: Supplemental Karyogram of Tumour 809T 21 22 X Y 162 sljijtiiii . ir.>>r,.,:n;i;» Jiiiiiiiiiiijiijiii (SiLiiiiiiiS m i l i n m i n » iiminiiiiHiii ? IIIIEIIIIIIII in o N H T —< la = O S = S3 si a C BS a S "ft c 9 v: mini i inn iiiiiniiiii i iiriiir.r.ii'.riii: niii i iMStl i i i rttBHr)ann^ lrnaHHkf IIIIIIIIIIIIIIIIMIIIII1I ro i&iinitiu: sijiiiSiiiinsiB iii iii ii ;i»BvTOiii;tiii|vii I MiMtMBrMnl M U g i KffinttiiinnliHni Jl II (0 »«,i.;ii;iiiir.) iiiliiSti 00 iimniiiimii iiiuiiimiiiiiwiii iiifiihUiiuniiiiiHtiiiiiiliiniin •ii»*i*iffiiiiiiii"i rittmrnrniiitl illiiraiiiiMiiii^ litiiliHi^ iitt ffiiiiiH?fii:i£':H:j? l l l l l l l l l l l U l l i n i l l l l l l l l N III l l l l l i n l l l l l l l ! - i i - r i i i i i i i i i i i i i * i i i ianm CN r o -r o z Cm • •ly'iVi—V. ,Mtl,i. .ii.n-jt, in' • ,vW ,v<frft,|r,'itj»,"1%» iraniiiitiiiiiiiBi i MRUIt i Hit«SitHiiiiHiSiiii ajiiiijSiifclV.iiiig i;iiSiii:hfiii H l t K l l O n W i l l t l f M i i M l l i Hlliihiii < m i n i i i i B i i i i m i n i i i i m i ? mi i min «? uiiiiiiiifi APPENDIX 25: Supplemental Karyogram of Tumour 819T 21 22 X Y 164 na mi IIIIII atiMttiiiintfaitti ii't'lti HHfini l f i J M f l / 3 l<4l iMHilUlMlial I I I I I I IMI ? II in ° u — U = o E = E -s u E C a 3 iiiiii: i IIIIII tSiuimhi I iiHiiitiiimmsi auiitiuiiiHUttiiau E L ifiiiSHm t wuEmtiSiifiSi VtU l l i l t t S i i i l S iSiuiBSS tii itui kjiiisutti tSfitfiUii itiisisiiiiiwtautisiii SBiiiiiiSjIt tiil Stiii i iSiiiiiiii»ittit A I I M l l l l l l l l l l l l ? • l r m i n «> iiiiiiiamissuii: ...i.:iiis;..iEi»iii liiiir.liiririi MuifiiSiiTiiiint iiFHrStiiicTiiiHi SW&t^tiiHtlWiSiii BHUKwiiSiiBS i i m i n i N IIIIIIBIIII m i n i N i n IIIIII m n i ? HI IIIIII t in E z QN SL. < SiiiMiihiiiii i inmiii. -, MKJBHBiBiHilu inUia«a»giiasUiio i M H i i HBittlllMtiilBIBIiih iiaiiitii i&iiiiriii IIIIII IIIIIIIIRBIIIIIIII I I I _ I H M I I B i l l l l • • • • • » • • « I I T - „ „ . » , . , u ? mi IIIIII •BiMBl iiiuiiiHiuii iitiiii;ii;r.r.r.v, iiii mn Minium m l l l l ? H»».1i ; i ( i i iHi! i iJ i i i i in lIUHHrfifl IIII Sil l g H •30 U s e £ = H •— o E S3 s-S£ O c-Ti = E C 3 n E z a. 3 . .t:l;:i i i ; i ; : i v . v i } ; » m i n i JfiwfiiiSu I tt&T&ISifiifflSi HifluMMimilttfi „ imiimiiiiiiii o> II iMiim • • j "liliilii III I llll 1 llll 11 lllllll lllllll „ I llll HM ll l l I „ II lllllllllll II ? t i || Uli • llinillll l i lDI x »•Wv-';";;y::f! .i'.'i;r. ;^ ;y*,y~^ i?*-: i>itu<»iiuiii::u i.n;::iiiiiii«!iiiii lauiMitnm mmtmwmtn inn BH [ I l f S nun IIII i min II i i imi N in mil; i HI IIIIII i s IHIIIIIIIIIIII 2 CM CM rnriiiifliiiniint i miti < fll IH UIIHHIIII1HHIIII11 • 11 l l l l „ ID IIIIII CM 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0099799/manifest

Comment

Related Items