Byrði betri berr-at maðr brautu at en sé mannvit mikit; auði betra þykkir þat í ókunnum stað; slíkt er válaðs vera.

Hávamál - Gestaþáttur

Bättre börda man bär ej på vägen än mycket mannavett. På främmande ort går det framför guld. Slikt är den torftiges tillflykt.

Havamål (Den Höges sång)

Better gear than good sense A traveller cannot carry, Better than riches for a wretched man, Far from his own home.

Havamal - Guest's Chapter

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Halldórsdóttir, A.M., Sander, B., Göransson, H., Isaksson, A., Kimby, E., Mansouri, M., Rosenquist, R., Ehrencrona, H. (2011) High-resolution genomic screening in mantle cell lym- phoma - specific changes correlate with genomic complexity, the proliferation signature and survival. , and , 50(2):113-21

II Halldórsdóttir, A.M., Lundin, A., Murray, F., Mansouri, L., Knuutila, S., Sundström, C., Laurell, A., Ehrencrona, H., Sander, B., Rosenquist, R. Impact of TP53 and 17p in mantle cell . , advance online publication, 1 July 2011; doi:10.1038 /leu.2011.162.

III Halldórsdóttir, A.M.*, Kanduri, M.*, Marincevic, M., Mansouri, L., Isaksson, A., Göransson-Kultima, H., Axelsson, T., Agarwal, P., Jernberg-Wiklund, H., Stamatopoulos, K., Sander, B., Ehrencrona, H., Rosenquist R. Mantle Cell Lym- phoma is Characterized by a Strikingly Homogenous Methyla- tion Profile: A Comparative Analysis with Chronic Lymphocyt- ic Leukemia. Submitted.

IV Marincevic, M.*, Kanduri, M.*, Halldórsdóttir. A.M.*, Mansouri, L., Göransson-Kultima, H., Isaksson, A., Juliusson, G., Ehrencrona, H., Stamatopoulos, K., Rosenquist, R. Distinct global DNA methylation profiles among chronic lymphocytic leukemia subsets with stereotyped B cell receptors. Manuscript.

* These authors contributed equally as first authors.

Reprints were made with permission from the respective publishers.

Supervisors and Examination Board

Supervisors: Richard Rosenquist, Professor Dept. of Immunology, Genetics and Pathology Uppsala University Sweden

Hans Ehrencrona, Associate Professor Dept. of Clinical Genetics Lund University Sweden

Faculty opponent: Reiner Siebert, Professor Institute of Human Genetics Christian-Albrechts University Kiel Germany

Review board: Teresita Díaz de Ståhl, Associate Professor Dept. of -Pathology Karolinska Institute Stockholm Sweden

Bengt Smedmyr, Associate Professor Dept. of Medical Sciences, Uppsala University Sweden

Hans Hagberg, Associate Professor Dept. of Radiology, Oncology and Radiation Science Uppsala University Sweden

Contents

INTRODUCTION ...... 13 B-cell maturation and immunoglobulin rearrangement ...... 13 Normal B-cell development ...... 13 Immunoglobulin gene rearrangements ...... 14 B-cell antigenic response in secondary lymphoid tissues ...... 16 B-cell malignancies ...... 16 Classification ...... 17 Cellular origin ...... 17 Pathogenesis ...... 18 Mantle cell lymphoma ...... 19 Epidemiology and clinical features ...... 19 Pathology ...... 20 Pathogenesis ...... 20 Prognostic parameters ...... 24 Therapy ...... 25 Chronic lymphocytic leukemia ...... 26 Epidemiology and clinical features ...... 26 Pathology ...... 26 Pathogenesis ...... 27 Prognostic parameters ...... 29 Therapy ...... 30 Expression and genomic microarray studies in MCL ...... 31 microarrays ...... 31 Gene expression profiling of MCL and the proliferation signature .... 31 Array platforms for analysis of genomic aberrations ...... 32 Copy number aberrations in MCL ...... 33 The tumor suppressor ...... 34 Structure and function ...... 34 TP53 aberrations in cancer ...... 36 Epigenetics and gene silencing by DNA methylation ...... 36 Introduction to epigenetics ...... 36 DNA methylation ...... 37 Aberrant methylation patterns in malignancies ...... 38 Laboratory methods for analyzing DNA methylation ...... 39 Analysis of gene promoter methylation in MCL and CLL ...... 40 AIMS ...... 42

MATERIAL AND METHODS...... 43 Patient Material...... 43 SNP array...... 44 GEP array and proliferation signature ...... 44 Detection of TP53 and deletions...... 45 Methylation array and methylation analysis...... 46 RQ-PCR analysis...... 46 Re-induction of gene expression...... 47 Pyrosequencing...... 47 Statistical analysis...... 47 RESULTS & DISCUSSION...... 49 Paper I: Screening for genomic aberrations in MCL...... 49 Paper II: Impact of TP53 mutation and 17p deletion in MCL...... 51 Paper III: Genome-wide study of methylation profiles in MCL and CLL 53 Paper IV: Genome-wide study of methylation profiles in CLL subsets..56 CONCLUDING REMARKS...... 59 ACKNOWLEDGEMENTS...... 61 REFERENCES ...... 63

Abbreviations

ASCT Autologous stem cell transplantation AID Activation-induced cytidine deaminase BCR B-cell BL C Constant CGH Comparative genomic hybridization cMCL Classical MCL cDNA Complementary DNA CDK Cyclin-dependent kinase CDR3 Complementarity-determining region 3 CHOP , adriamycin, vincristine and CLL Chronic lymphocytic leukemia CNA Copy number aberration CNN-LOH Copy number neutral loss of heterozygosity CNV Copy number variant Ct Cycle threshold D Diversity DLBCL Diffuse large B-cell lymphoma DNMTs DNA methyltransferases DSB Double strand break EBV Epstein-Barr virus FCM , cyclophosphamide, FISH Fluorescent in situ hybridization FL GC GEP Gene expression profiling GOTM tree machine GWAS Genome wide association studies HDAC Histone deacetylases HIV Human immunodeficiency virus HSC Hematopoietic stem cell IARC International Agency for Research on Cancer Ig Immunoglobulin IGHV Immunoglobulin heavy variable ILSG International lymphoma study group

iMCL Indolent MCL IPA Ingenuity pathway analysis J Joining LLMPP Leukemia and Lymphoma Molecular Profiling Project MALT Mucosa-associated lymphoid tissue MAR Minimal altered region MBL Monoclonal B-cell MCL Mantle cell lymphoma MECPs Methylated CpG-binding MI Methylation index MIPI Mantle-cell lymphoma international prognostic in- dex mRNA messenger RNA NHL non- OS Overall survival PCR Polymerase chain reaction PcG Polycomb repressor group RAG Recombination activating genes RSS Recombination signal sequences RQ-PCR Real-time quantitative PCR SHM Somatic hypermutation SLL Small lymphocytic lymphoma SNP Single nucleotide polymorphism T Transitional TdT Deoxynucleotidyl transferase TSG Tumor suppressor gene V Variable WHO World Health Organization

INTRODUCTION

B- and T-lymphocytes are central components of the adaptive immune sys- tem which is antigen-specific and requires the recognition of specific “non- self” antigens during a process called antigen presentation. The evolvement of this system in early vertebrates allowed for a stronger immune response as well as immunological memory. In contrast, the innate immune system de- fends the host from infection by other organisms in a non-specific manner and does not involve memory. The adaptive immune system is highly ad- justable because of critical processes that take place during B-cell matura- tion, namely immunoglobulin (IG) gene recombination and somatic hyper- mutation (SHM). As a result of these mechanisms a limited number of genes can generate a vast number of B-cell receptors (BCR). However, these same processes may also increase the risk of genetic damage, such as mutations and translocations, and thereby predispose the B-cell to malignant transfor- mation. This thesis will focus on B-cell malignancies and therefore a basic introduction to the maturation of B-cells and rearrangements of IG genes is first provided.

B-cell maturation and immunoglobulin gene rearrangement Normal B-cell development B-cells, like all hematopoietic cells, are derived from pluripotent hematopoi- etic stem cells [1]. Specification toward the B-cell lineage is dependent on expression of specific transcription factors that include early B-cell factor (EBF) and 3 (TCF3, also known as E2A) [2, 3]. These DNA-binding proteins regulate the expression of a variety of B-lineage tar- get genes and induce expression of additional transcription factors that play a role in B-cell development. Absolute commitment to the B-cell lineage is however dependent on subsequent expression of the PAX5 transcription factor [4]. Thereafter, the multi-step process of B-cell development starts, first in primary lymphoid tissue (fetal liver and fetal/adult bone marrow) whereas subsequent functional maturation takes place in secondary lym- phoid tissue (human lymph nodes and spleen) as displayed in Figure 1 [5]. IG heavy chain (IGH) gene rearrangement initiates in the earliest B-lineage-

13 committed progenitor which is termed the pro-B cell. If the rearrangement is successful, heavy chains of the µ class are expressed in the cytoplasm, at which stage the cells are defined as pre-B cells. Finally, once light chain gene rearrangements have occurred and light chains are expressed, pre-B cells develop into immature B-cells, expressing IgM on their surface. This generation of immature B-cells occurs in the bone marrow and is antigen- independent (primary B-cell production). The next stage starts with the mi- gration of the newly produced immature B-lymphocytes into the spleen where they differentiate into mature, naïve B-cells in a process referred to as secondary B–cell development (antigen dependent). During this selection process the B-cell goes through two transitional (T) stages; first as T1 cells where those with BCR specificities for blood-born self antigens are deleted by negative selection, and secondly as T2 cells whose maturation is depend- ent on positive selection by antigen. Therefore, only 1-3% of splenic transi- tional cells develop into mature naïve B-cells [6, 7]. Finally, naïve B-cells recirculate through the bloodstream and enter peripheral lymph nodes and spleen where they form primary follicles, before undergoing further differen- tiation into effector cells in response to antigenic challenge [8].

Immunoglobulin gene rearrangements Human antibody molecules (and BCRs) contain heavy and light chains with both constant (C) and variable (V) regions that are encoded by genes on three loci. The IGH locus on 14 contains genes encoding the heavy chain, whereas the IG kappa (IGK) locus on chromosome 2 and IG lambda (IGL) locus on both contain genes encoding the light chain. The IGH locus consists of distinct V, diversity (D), joining (J), and C genes. Because the genes are separated from one another, their juxta- position must occur with deletion of the intervening sequence. The initial event during IGH gene rearrangement juxtaposes an IGHD gene to an IGHJ gene, which is then followed by rearrangement of an IGHV gene to the IGHD-IGHJ complex [8]. The IGHC gene remains separated from the rear- ranged IGHV-D-J complex by an intron which is deleted later on during RNA processing after transcription [8, 9]. This gene recombination process is dependent on enzymatic machinery that deletes intronic sequences and joins coding segments of DNA. The enzymes that mediate these functions act through recognition of unique recombination signal sequences (RSS) [10]. During IG gene recombination the RSS form loops of DNA, which in turn bring the coding exons adjacent to one another, followed by deletion and degradation of the noncoding loops. Two highly conserved proteins, RAG1 and RAG2, cleave at RSS on one DNA strand [10, 11] leading to the creation of a hairpin loop by the other strand which is subsequently opened by the Artemis nuclease [12]. A number of other enzymes are involved in the cleavage and ligation process. One is the nuclear enzyme deoxynucleo-

14 tidyl transferase (TdT) which is recruited to add nucleotides randomly to the DNA ends prior to ligation, thus providing junctional diversity [13]. Another cause of BCR diversity is the variability in the exact position of cleavage of the hairpin loop by the Artemis nuclease. Both mechanisms contribute to a large variability in the DNA sequence even when the same D, J and V seg- ments are joined.

Figure 1. A simplified model of B cell development.

Following translation, the µ heavy chain is expressed on the surface of pre-B cells where it associates with a surrogate light chain (Vpre-B and 5 pro- teins) forming the pre-B cell receptor (pre-BCR) [5, 14]. One role of the surrogate light chains may be to select those heavy chains that will ultimate- ly be capable of pairing with conventional light chains. Another function of pre-BCR signaling is to mediate allelic exclusion, where one successfully rearranged IGH allele inhibits rearrangements at the other IGH allele [15]. At some point pre-BCR expressing cells cease to express surrogate light chains, enter a resting phase and initiate conventional light chain gene rear- rangement. As mentioned, human Ig light chains are encoded by the IGK

15 (60%) or IGL (40%) genes. The light chain loci include V, J and C regions but no D region in contrast to the IGH locus. Rearrangements initiate at the IGK locus with joining of an IGKV gene to an IGKJ gene, where intron splicing after transcription results in formation of a mature IGKV-J-C tran- script. If rearrangements at the first IGK locus are unsuccessful, attempts are made to rearrange the second IGK allele. If this fails, the IGL locus is uti- lized [16]. Following successful recombination at the IGK or IGL locus, the light chain is expressed and associates with the heavy chain to form the Ig molecule. Finally, the multimeric BCR complex is formed by the assembly of a surface Ig homodimer and a non-covalently-bound heterodimer Ig /Ig (CD79A/CD79B) [5]. In order for the organism to mount an effective humoral immune re- sponse, an array of BCRs with unique antigen-binding specificities, together referred to as the Ig repertoire, must be generated. Several mechanisms have evolved to ensure that this occurs; 1) enormous combinatorial diversity cre- ated by the IG V, D and J genes, 2) addition of nucleotides to gene junctions by the TdT nuclear enzyme [13], 3) junctional diversity due to the imprecise formation of DNA joints during recombination, and 4) SHM of V region genes in secondary lymphoid tissues [17], which is described further in the following section.

B-cell antigenic response in secondary lymphoid tissues As described above, naïve B-cells in lymph node primary follicles are de- rived from T2 B-cells that have migrated from the spleen. Following their binding of a T-cell dependent antigen, mature naïve B-cells in primary folli- cles become activated and undergo clonal expansion. The histologic appear- ance of the follicle changes as these events evolve. The nonresponsive B- cells form an outer mantle zone surrounding the proliferating, antigen- responsive B-cells in a central germinal center (GC) [18]. Two regions can be distinguished within the GC of the secondary follicle. At one pole, the cycling B-cell blasts are referred to as centroblasts and form the dark zone. The other pole, the light zone, consists of nonproliferating cells referred to as centrocytes. Some centrocytes will go on to become plasma cells that secrete high-affinity Ig, whereas others will become memory B-cells [8, 19]. The formation of high-affinity Ig occurs as a result of a process termed affinity maturation, which involves Ig class switching and selection of B-cell clones with high antigen-binding potential. This increased affinity is due to changes in the genes that encode the antigen-binding domain of the Ig molecule in a process termed somatic hypermutation (SHM). Here, single nucleotide muta- tions, or more rarely insertions/deletions, are introduced into the genes en- coding the V region of the Ig molecule. The enzyme activation-induced cyti- dine deaminase (AID) is expressed in GC B-cells and is indispensable for both SHM and class switch recombination [20, 21]. Although the process of

16 affinity maturation is highly efficient, one potential unintended consequence is the development of B-cell lymphoma.

B-cell malignancies Classification of B-cell malignancies According to the World Health Organization’s (WHO) classification lym- phomas fall into three major categories; mature B-cell neoplasms, mature T- cell and NK-cell neoplasms and Hodgkin lymphoma [22]. However, mature B-cell neoplasms comprise over 90% of lymphoid neoplasms worldwide [23]. This large category includes such as diffuse large B-cell (DLBCL), follicular (FL), marginal zone (MZL), Burkitt (BL) and mantle cell lymphoma (MCL). Also included among mature B-cell malignancies are such as and chronic lymphocytic leukemia (CLL). MCL and CLL will be the focus of this thesis. Mature B-cell neo- plasms are more common in developed countries, but individual neoplasms vary in their relative frequency in different parts of the world. As an exam- ple, FL is more common in developed countries, particularly the United States and Western Europe, but is uncommon in South America, Eastern Europe, Africa and Asia. Conversely, BL is endemic in equatorial Africa but rare in the United States and Western Europe [22].

Cellular origin of B-cell malignancies The cellular origin of various B-cell malignances is still an area of active investigation. However, available evidence suggests that each disease repre- sents a clonal proliferation of B-cells at a distinct stage of differentiation, ranging from naïve B-cells to mature plasma cells. Therefore, B-cell malig- nancies can to some extent be classified according to the corresponding normal B-cell stage and/or histological location, such as lymph node mantle zone, GC, marginal zone [22]. Important clues to the maturation stage of B- cells can be derived from mutation analysis of IGHV genes [24]. As SHM takes place in GCs of lymph nodes, unmutated IGHV genes may correspond to a pre-GC origin, ongoing mutations to a GC origin and stable mutations to a post-GC origin. Supporting this concept, the GC-derived lymphomas FL and GC B-cell DLBCL frequently have active ongoing SHM, while the ab- sence of ongoing SHM in reflects its post–GC origin [25, 26]. In recent years, however, the concept of cancer stem cells has received increased attention, as it was observed that some malignant cells and stem cells share self-renewal and differentiation capacities [27]. Recently, a small subset of cancer stem cells has been identified among the complete tumor

17 cell population in various malignancies, both solid tumors and leukemias [28-31]. Although the existence of a similar lymphoma stem cell is still con- troversial, it has been proposed that certain lymphomas, such as MCL and FL, originate from committed lymphoid progenitor/precursor cells [32]. Supporting this, a recent study identified a small fraction of CD19 negative MCL cells that possessed self-renewal and tumorigenic activities and were thus proposed to represent MCL-initiating cells [33]. Furthermore, HSCs purified from patients with CLL but not normal HSCs were able to develop monoclonal B cells with a CLL-like phenotype after xenogeneic transplanta- tion into immunodeficient mice [34]. Therefore, HSCs may be involved in leukemogenesis even in mature lymphoid tumors.

Pathogenesis of B-cell malignancies The pathogenesis of B-cell lymphomas is a highly complex process involv- ing genetic alterations in the tumor clone itself as well as biologic alterations in the host. Four main mechanisms of lymphomagenesis are recognized. These include 1) accumulation of genetic alterations in the tumor, 2) infec- tion of the tumor clone by oncogenic viruses/bacteria, 3) stimulation and selection of tumor precursor cells by an antigen, and 4) immunodeficiency of the host [8]. The major known risk factor appears to be an abnormality of the immune system, either immunodeficiency or autoimmune disease. Although evidence of immune system abnormalities are lacking in most patients with mature B-cell neoplasms, immunodeficient patients have a markedly in- creased incidence of B-cell neoplasia, particularly DLBCL and BL [35-37]. Furthermore, patients with autoimmunity are at an increased risk of lym- phomas, particularly DLBCL [38, 39]. Infectious agents have been linked to the development of several types of B-cell lymphomas. The clearest example is BL where Epstein-Barr virus (EBV) is present in nearly 100% of endemic cases and in 40% of sporadic and HIV-associated cases [40, 41]. Other vi- ruses implicated include human herpesvirus-8 (HHV8) in primary effusion lymphoma and in lymphoplasmacytic lymphoma (Walden- ström’s macroglobulinemia) [42, 43]. Bacteria, or immune responses to bac- terial antigens, have been implicated in the pathogenesis of mucosa- associated lymphoid tissue (MALT) lymphoma. Gastric MALT lymphoma in patients infected with depends on the presence of T- cells activated by H. pylori antigens for proliferation and treatment of H. pylori infection causes regression of the lymphoma in many patients [44, 45]. Genetic susceptibility does play a role in B-cell lymphoma development, although the extent of the contribution is presently unknown [46]. Confirmed single nucleotide polymorphism (SNP) associations include TNF-308G->A and IL-10 -3575T->A polymorphisms, both associated with increased NHL risk, in particular DLBCL [47]. Genome wide association studies (GWAS)

18 have linked variations at 6p21.33 and 6p21.32 with FL risk, suggesting that major histocompatibility complex genetic variation influences FL suscepti- bility [48, 49]. Furthermore, GWAS of CLL has identified a total of ten risk loci [50-52]. In contrast to genetic predispositions, acquired genetic aberrations have for a long time been known to play a major role in lymphoma patholo- gy. Several mature B-cell neoplasms have characteristic genetic abnormali- ties that are important in determining their biologic features and that can be useful in differential diagnosis. These include for instance the t(11;14)(q13;q32) in MCL, t(14;18)(q32;q21) in FL, t(8;14)(q24;q32) in BL, and t(11;18)(q21;q21) in MALT lymphoma [22]. The first three place a cellular proto-oncogene under the control of the IGH promoter on chromo- some 14q32, resulting in constitutive activation of the gene. In MCL and BL the overexpressed genes are associated with proliferation (CCND1 or ), while in FL and MALT lymphoma the translocations result in overexpres- sion of an anti-apoptotic gene (BCL2 or API2) [22]. As previously mentioned mature B-cell neoplasms are a very hetero- geneous group. The next sections will be devoted to further outline the fea- tures of MCL and CLL, two very different although related entities.

Mantle cell lymphoma Epidemiology and clinical features MCL is relatively uncommon, comprising 3-10% of NHL. It has a marked male predominance (male:female ratio = 2-7:1) and mainly affects middle- aged to elderly individuals (median age about 60) [22, 53]. MCL typically presents in advanced stage with , hepatosplenomegaly and bone-marrow involvement, but peripheral blood involvement is also present in about 25% of cases . MCL combines the worst features of indolent and aggressive lymphomas, in being incurable by currently available chemother- apy but clinically aggressive, with a median survival of only 3-5 years [22, 54, 55]. Recently however, an indolent form of MCL (iMCL) was described which differs from the classical form of MCL (cMCL) in having a benign clinical course with survival of more than 7 to 10 years, sometimes not even requiring for long periods. This iMCL subgroup often dis- plays nonnodal leukemic disease with predominantly hypermutated IGHV genes, noncomplex karyotypes and weak SOX11 expression [56-58]. MCL with an indolent clinical evolution may therefore represent a distinctive clin- ical and biological subtype of the disease. It has even been proposed that at least some iMCL cases may represent the MCL counterpart to the CLL-like monoclonal B-cell lymphocytosis (MBL) [58]. Recently, very low levels (10-7) of long-lived monoclonal B-cells with a t(11;14) were detected in up

19 to 7% of healthy individuals, which may thus carry MCL-like MBL clones [59].

Pathology In 1992 American and European International lymphoma study group (ILSG) members agreed on the existence of a tumor derived from mantle B- cells that was consequently termed mantle cell lymphoma (MCL) [60]. Pre- vious classifications and nomenclatures included terms such as “intermedi- ately or poorly differentiated lymphocytic lymphoma, diffuse or nodular” (Rappaport) or “malignant lymphoma, diffuse, small cleaved type” (Work- ing formulation) [22]. As these older descriptions imply, MCL demonstrates architectural destruction by a monomorphic lymphoid proliferation with a vaguely nodular or diffuse growth pattern. The malignant cells are small to medium-sized lymphoid cells with slightly to markedly irregular nuclear contours [22]. In addition to the classical form described above, there are some cytological and/or morphological variants of MCL, termed blastoid, pleomorphic, small cell, and marginal zone like. However, only the blastoid and pleomorphic variants are considered to be of potential clinical signifi- cance, with both being clinically aggressive [53]. The blastoid variant con- sists of slightly larger cells resembling lymphoblasts with dispersed chroma- tin and a high mitotic rate. The pleomorphic form is characterized by hetero- geneous cells, with oval to irregular nuclear contours, frequent mitotic fig- ures and at times prominent nucleoli. Both the blastoid and pleomorphic form occur de novo, rarely representing morphological progression of the disease [22, 53]. The immunophenotype of MCL is characterized by expres- sion of mature B-cell antigens (CD19, CD20, CD22, CD79a and BSAP/PAX5) and co-expression of the T-cell associated antigens CD5 and CD43. The IgM/IgD surface Igs are usually intense and frequently associat- ed with the lambda light chain. On the other hand, the neoplastic cells are usually CD10-, BCL6- and IFR4-negative, with occasional weak expression of CD23 [22, 53]. In contrast to iMCL, a strong nuclear expression of the transcription factor SOX11 is a feature of cMCL, including nega- tive forms [61, 62]. SOX11 is not expressed by other mature B-cell lym- phomas with the exception of some BL cases. However, it is uniformly negative in other small B-cell lymphomas which might be confused with MCL, such as CLL, FL and MZL [61, 62].

Pathogenesis of MCL Genetics The t(11;14)(q13;q32) is regarded as the primary genetic event in the patho- genesis and the hallmark of MCL [63]. By analyzing the breakpoint regions

20 it has been deduced that this translocation occurs at the pre-B-cell stage of differentiation in the bone marrow initiated by the recombination of the IGH locus [64]. The translocation juxtaposes the CCND1 locus at 11q13 with the IGH locus at 14q32, causing overexpression of the cyclin D1 at both mRNA and protein levels as displayed in Figure 2 [65, 66]. The t(11;14) is detected by conventional cytogenetics in up to 65% of MCLs, but using fluo- rescent in situ hybridization (FISH) techniques it can be found in virtually all cases of MCL [67]. As cyclin D1 is not generally expressed by normal B cells or other lymphomas, its positive expression has come to represent a highly specific marker for MCL in clinical practice [65]. It should be men- tioned, however, that the t(11;14)(q13;q32) has been detected in other types of hematological malignancies such as multiple myeloma [22]. Furthermore, gene expression profiling studies have identified a subset of MCL cases that is cyclin D1 negative but cyclin D2 or cyclin D3 positive [68, 69]. Translo- cations and resulting protein overexpression involving these alternative cy- clin isoforms may be detected by FISH and RT-PCR respectively but im- munostaining is not specific [70].

Figure 2. The t(11;14)(q13;q32) is a hallmark of MCL and results in excess cyclin D1 expression.

Importantly, studies with transgenic mice suggest that cyclin D1 deregula- tion, although important for MCL initiation, may not be responsible for the complete cell transformation, and that secondary genetic alterations are re- quired [71, 72]. Comparison with other malignant lymphoid neoplasms has revealed that MCL, in particular the blastoid variant, is among those with the highest level of genomic instability [73]. Recurrent genomic aberrations

21 observed in MCL include deletions at 1p, 8p, 9p, 9q, 11q, 17p and gains at 3q, 8q, 15q, 18q [74]. Although the target genes of most genomic aberrations have not been identified, a small group of candidate genes has been charac- terized, including ATM, BCL2, CDK4, CDKN2A, MYC, and TP53. Notably, most of these target genes are involved in regulation and cellular response to DNA damage, as discussed below [75].

Immunogenetics The IGHV genes are unmutated in the majority of MCL cases, consistent with a pre-GC B cell. However, about 20-30% of MCL cases do carry mu- tated IGHV genes (<98% sequence identity to germline) [53, 76, 77]. The fact that SHMs in IGHV genes are detected in a significant proportion of MCL cases indicates that some tumors originate in cells that have undergone the influence of the mutational machinery of the follicular GC. However, the rate of IGHV mutations in MCL is lower than in other lymphoid neoplasms, such as CLL or follicular lymphoma, and the mutational status does not cor- relate with survival in MCL [78]. As in CLL, a biased use of IGHV3-21, IGHV3-23 and IGHV4-34 has been described in MCL, suggesting that these tumors may originate from specific subsets of B cells [53, 76]. Recently, a large collaborative project reported the results of IGHV mutation analysis of 807 sequences from MCL cases [79]. This report agrees with previous stud- ies in that 23% of sequences were defined as “mutated” according to the classical definition (<98% sequence identity to germline). However, using another approach 57% of sequences were classified as minimally/borderline mutated (97-99.9%) whereas 14% were classified as highly mutated (<97%). Notably, 11% of sequences could be assigned to 38 clusters based on se- quence homology (stereotypy) of the heavy complementarity-determining region 3 (VH CDR3) [79].

Molecular pathways Most molecular alterations that are characteristic of MCL affect three major cellular pathways: cell cycle control, DNA damage response and cell surviv- al/, as displayed in Figure 3 [74]. As discussed above cyclin D1 protein overexpression is considered a hallmark of MCL. Cyclin D1 belongs to the cyclin family of proteins, which control the progression of cells through the cell cycle by activating cyclin-dependent kinase (CDK) enzymes [80]. As the name implies the concentration of cyclins varies in a cyclical fashion, as they are produced or degraded during different stages of the cell cycle. Proteins phosphorylated by cyclin/CDK complexes are responsible for specific events during cycle division, such as microtubule formation and remodeling. Specifically, cyclin D1 plays an important role in the regulation of G1-S transition following mitotic growth factor signaling. The protein forms a complex with CDK4/CDK6 which then phosphorylates the tumor suppressor (pRb), facilitating cell cycle pro-

22 gression. pRb plays a key role in the G1-S transition by sequestering and inactivating transcription factors which are involved in the transactiva- tion of genes required for S phase entry and DNA replication [81]. Secondly, genes implicated in DNA damage response play a large role in MCL lymphomagenesis, especially TP53, CDKN2A and ATM. Inactivation of the TP53 locus and/or p53 overexpression occurs in a subset of cases and is a marker of poor prognosis [82]. The p53 tumor suppressor protein has been called “the master guardian of the cell” and is discussed in more detail later in this thesis [83]. Inactivation of CDKN2A by deletion, mutation, or hypermethylation is detectable in a significant proportion of blastoid variant MCL [84]. The CDKN2A locus encodes two key regulatory elements of the cell cycle and survival pathways, the CDK4 inhibitor INK4a and the p53 regulator ARF. As shown in Figure 3 INK4a deletion leads to increased lev- els of CDK4 which forms a complex with cyclin D1, already present in high levels, and thus promotes the G1 to S phase transition in MCL. Loss of ARF, a negative regulator of MDM2, leads to increased levels of MDM2 which promotes p53 degradation. The ATM (ataxia telangiectasia mutated) gene is especially interesting since copy number loss is detected in up to 60% of cases and mutations in up to 75% of MCL cases [85, 86]. The ATM gene encodes a phosphoprotein kinase that is a master controller of cell cycle checkpoint signaling pathways important for cellular response to DNA dam- age, including double strand breaks (DSBs) [87]. The high frequency of ATM gene aberrations in MCL is interesting as these are not as common in other B-cell malignancies [88]. Notably, the ATM gene is strongly expressed in mantle zone naïve B-cells but not in follicular GC B-cells which undergo physiological DSBs during SHM and class switch [89]. ATM inactivation in pre-malignant mantle zone cells may prevent repair of DSBs and thereby facilitate genomic instability and tumorigenesis. In support of this hypothe- sis, MCL cases that carry ATM mutations have a higher number of chromo- somal aberrations than those with a wild-type gene [90]. The third cellular process frequently targeted in MCL involves cell sur- vival and apoptosis pathways. The most relevant apoptotic pathway affected is the BCL2-system [91]. The anti-apoptotic BCL2 gene is amplified and overexpressed in a subset of MCL cases while the pro-apoptotic BIM (BCL2L11) gene has been reported to be targeted by homozygous deletions [91, 92]. Interestingly, a recent study suggested a novel role for cyclin D1 in deregulating the apoptosis pathway in MCL by sequestering the proapototic BAX protein in the cytoplasm and thus facilitating the antiapoptotic function of BCL2 [93]. Thus, the impact of cyclin D1 overexpression in MCL may extend beyond the deregulation of the cell cycle. Furthermore, the NF B and Akt signaling pathways, which are crucial for promoting cell survival and proliferation, are constitutively activated in MCL [94, 95].

23

Figure 3. Molecular pathways involved in the pathogenesis of MCL. Proteins repre- sented in green boxes have been reported to be overexpressed, while proteins in red boxes are underexpressed in MCL. Overexpression of Cyclin D1, represented in a yellow box, is a hallmark of MCL. Adapted from Obrador-Hevia et al [74].

Prognostic parameters With an increasing number of new therapeutic options on the horizon there is a need for correct stratification of patients according to the biological risk of their disease. Recently, a prognostic index specific for MCL (MIPI, Man- tle-cell lymphoma International Prognostic Index) was designed. The MIPI is based on four independent prognostic factors [age, Eastern Cooperative Oncology Group performance score, and leucocyte count] and has been shown to stratify MCL patients into three distinct prog- nostic groups [57, 96]. Several histopathologic and genetic features have been reported to have prognostic value in MCL, including the mitotic index, Ki-67 detection, gene expression profiling and several genomic aberrations, including losses at 8p, 9p, 9q, 13q14, 17p and gains at 3q, 8q and 12q [85, 97, 98]. The prognostic impact of many of these parameters however, merely reflects the tumor proliferation as they lose their predictive value in multi- variate analysis [53]. The importance of proliferation in MCL is reflected by the prognostic power of the proliferation gene expression signature, where the twenty genes with the highest prognostic impact were found to have bio- logical functions related to cell proliferation such as mitosis, DNA synthe- sis/repair and the cell cycle [68]. More recently, a five-gene model (RAN,

24 MYC, TNFRSF10B, POLE2 and SLC29A2) based on RQ-PCR analysis of gene expression was proposed as a prognostic indicator in MCL [99]. Gene expression analysis of MCL is discussed in more detail below.

Therapy The standard treatment for MCL has for many years been polychemothera- py. The combination of chemotherapy regimens such as CHOP (cyclophos- phamide, adriamycin, vincristine and prednisone), hyperCVAD (a dose in- tense, hyperfractionated cyclophosphamide in a CHOP-like combination plus high dose methotrexate and cytarabine), or FCM (fludarabine, cyclo- phosphamide, mitoxantrone) with (a chimeric monoclonal anti- CD20 antibody) can produce impressive overall response rates of up to 80– 95% and complete remission rates of up to 30–87% in previously untreated patients [100-102]. Furthermore, a recent meta-analysis suggests that the addition of rituximab to chemotherapy might improve overall survival (OS) in MCL patients [103]. The combination of high-dose chemotherapy and/or rituximab treatment with autologous stem cell transplantation (ASCT) has achieved high overall response rates in some studies, but this modality is used preferentially in young patients and early in the course of the disease [55, 104, 105]. Furthermore, molecular relapse of MCL may occur many years after ASCT, although a recent study suggests that PCR-based pre- emptive treatment with rituximab can reinduce molecular remission and prevent clinical relapse [106]. Of note, in patients treated within two ran- domized trials of the European MCL Network molecular remission was pre- dictive of clinical outcome after combined immunochemotherapy followed by ASCT or maintenance treatment [107]. Preliminary experience suggests that allogeneic procedures may represent a promising salvage strategy for patients with relapsed and refractory MCL [108, 109]. Despite the relatively high rate of response achieved by these multimodal approaches, the majority of patients relapse, meaning that MCL patients are not cured by conventional therapy strategies. To this end, novel therapeutic approaches that target crucial biological pathways are being actively investi- gated, including cell cycle inhibitors, proteasome inhibitors, mTOR inhibi- tors and histone deacetylase inhibitors [110-112]. The combination of these new strategies with other therapeutic agents, and the correct stratification of patients by applying prognostic parameters, may change the management and outcome of MCL patients in the future.

25 Chronic lymphocytic leukemia Epidemiology and clinical features CLL as a distinct entity was first identified as early as in 1903 by Turks and a detailed clinical description of the disease was presented in 1924 by Minot and Isaacs [8]. CLL is the most common form of adult leukemia in the Western Hemisphere, accounting for nearly 35% of all leukemias [22, 113]. Notably, CLL is rarely seen in Japan, China and other Asian countries but the reason for this wide disparity in incidence is unknown. CLL is primarily a disease of older adults (median age 70), although the disease is being diag- nosed in increasing frequency among younger age groups, probably due to the increased use of diagnostic blood tests in general [8]. There is a male preponderance; the incidence in men is twice that reported for women. Most patients are asymptomatic at diagnosis but some present with weakness, fatigue, autoimmune hemolytic anemia, infections, hepatosplenomegaly or lymphadenopathy [22]. Although the clinical course is indolent, CLL is not usually considered to be curable with available therapy. The overall median survival after diagnosis is 7 years, but can range from months to decades [114]. In 1-10% of CLL patients a transformation to a large cell lymphoma occurs [8]. This has been termed Richter’s syndrome and is frequently char- acterized by sudden clinical deterioration and a rapid increase in the size of a lymphoid mass.

Pathology CLL is a neoplasm of monomorphic small, round B-cells in the peripheral blood, bone marrow and lymph nodes, admixed with prolymphocytes and paraimmunoblasts. The term small lymphocytic lymphoma (SLL) is restrict- ed to cases with the tissue morphology and immunophenotype of CLL, but which are non-leukemic [22]. Enlarged lymph nodes in patients with CLL show effacement of the architecture, with a pseudofollicular pattern of regu- larly distributed pale areas (“proliferation centers”) containing larger cells in a dark background of small cells. Mitotic activity is usually very low. On microscopic examination of bone marrow and peripheral blood smears, CLL cells can be identified as small, homogeneous lymphocytes with clumped chromatin and scant cytoplasm. Smudge or basket cells are often present. The proportion of prolymphocytes (larger cells with prominent nucleoli) in blood films is usually less than 2%, but increasing numbers correlate with a more aggressive disease course [22]. Bone marrow involvement may be nodular, interstitial, diffuse, or a combination of the three. The tumor cells express weak or dim surface IgM or IgM/IgD and CD20. Expression of CD5 and CD23 are characteristic of CLL, whereas FMC7 and CD10 are usually not expressed on CLL cells [22]. According to diagnostic criteria guidelines

26 that were proposed in 2008 by the International Workshop on CLL, the di- agnosis of CLL requires the presence of at least 5 x 109/L B lymphocytes in peripheral blood and clonality needs to be confirmed by flow cytometry. In addition, expression of immunophenotypic markers such as CD5, CD19, CD20 and CD23 should be evaluated by flow cytometry. Other tests are not needed to establish the diagnosis of CLL but may help in predicting progno- sis or estimate disease burden [115].

Pathogenesis Despite the ready availability of tumor cells in CLL, up to now, very little has been known about the pathophysiology of the disease. The cells them- selves do not grow well in vitro and few reliable CLL cell lines exist. Among factors considered important in the pathogenesis of CLL are immu- nogenetics, genetic aberrations, the balance between proliferation and apop- tosis, the cellular microenvironment and antigen stimulation.

Immunogenetics A paradigm shift in the understanding of CLL biology occurred in 1999 when two groups reported that the mutational status of the IGHV genes pre- dicts survival in CLL [116, 117]. Indeed, there are two distinct types of CLL defined by the IGHV mutational status; 40-50% have unmutated IGHV genes (>98% sequence identity to germline), while 50-60% carry a variable amount of SHM. CLL cases that carry mutated IGHV genes are postulated to derive from B-cells that have been activated by T-cells and differentiated into antigen-experienced memory or marginal zone B-cells. CLL cases that carry unmutated IGHV genes are also thought to be antigen-experienced, but probably as a result of T-cell independent immune responses outside the GC [118]. Considerably shorter survival has been reported for CLL patients with unmutated IGHV genes compared to patients with mutated IGHV genes [116, 117]. Furthermore, the presence or absence of SHM is associated with particular IGHV genes, for example specific alleles of the IGHV1-69 and IGHV4-39 genes have an unmutated profile. However, considerable varia- tion is observed within each mutation category suggesting that further strati- fication is possible. In fact, a recent study reported an association between codon 72 TP53 polymorphism and short survival within the IGHV mutated group [119]. In addition, CLL is uniquely characterized by the existence of distinct subsets of cases with closely homologous (stereotyped) CDR3 se- quences [120]. The CDR3 is the most variable of the three CDRs in the Ig molecule, and includes some of the V, all of the D and J, and some of the C domains. As the likelihood of finding highly similar CDR3 sequences in two or more B-cell clones by chance is extremely small, the presence of stereo- typed CDR3s has been suggested to indicate a role for antigen selection in the pathogenesis of CLL [121]. Recently it has been reported that up to 30%

27 of CLL patients carry stereotyped BCRs [120, 122]. For instance, CLL sub- set #1 cases are IGHV unmutated, express IGHV1/5/7 and IGKV1-39/1D-39 genes and display a poor prognosis compared to non-subset #1 cases which use the same IGHV genes but are non-stereotyped and variably mutated [120, 123]. In contrast CLL subset #4 cases are IGHV mutated, IgG switched, express IGHV4-34/IGKV2-30 genes, have a young age of onset and an indolent disease course compared to non-stereotyped, mutated IGHV4-34 patients [120]. Of note, the IGHV3-21 gene is associated with poor prognosis whether expressed in a mutated or unmutated form [124]. Half of all IGHV3-21 patients belong to subset #2 which is characterized by IGLV3-21 gene usage and a stereotyped BCR with a short VH CDR3 [124- 126]. Compared to non-subset #2 cases (which use IGHV3-21 but do not carry a stereotyped BCR) subset #2 patients have been reported to show a shorter time to progression [120].

Genetics CLL is an acquired disorder but the observation that relatives of patients with CLL have an increased risk of CLL and other B-cell malignancies, im- plies that hereditary factors may play an important role in some patients [127]. As previously mentioned GWAS have recently identified 10 risk al- leles associated with CLL [50-52]. Unlike MCL there is no single secondary genetic lesion, analogous to the t(11;14), that characterizes CLL. However, about 80% of CLL cases display chromosomal aberrations when examined by FISH analysis [128]. The most common genetic aberrations in CLL are 12 (11-16% of cases), 13q14 deletions (up to 50%) and 11q22-23 deletions (12-18%), whereas a deletion on chromosome 17p affecting the TP53 gene, is seen less frequently (7-10%). Genetic lesions that target the TP53 TSG (17p deletion), and the ATM gene (11q23 deletion) both affect TP53 function, as ATM regulates p53. As mentioned above, 13q14 deletions are the most common genetic aberrations in CLL. Two loci in this region, miR-15/16 and DLEU7, have been suggested to play a role as tumor suppres- sor [129, 130]. Notably, both mi-R15 and mi-R16 were deleted or down- regulated in 68% of CLL cases [131]. On the other hand, the pathogenic role of trisomy 12 in CLL remains unresolved, although the CLLU1 gene (CLL upregulated gene 1) gene, the first disease specific gene identified in patients with CLL, has been located to 12q22 [132].

Apoptosis and proliferation Another hallmark of CLL is the accumulation of mature B-cells that have escaped programmed cell death and undergone cell-cycle arrest in the G0/G1 phase [114]. The cells have a low proliferative activity, and data suggests that in-vivo defective apoptosis accounts for accumulation of B-cells. BCL2 and other members of the Bcl-2 family, such as anti-apoptotic proteins BCL- XL, BAG1, and MCL1, are overexpressed whereas pro-apoptotic proteins,

28 such as BAX and BCL-XS are underexpressed [133]. In contrast with in- vivo results, apoptosis occurs after in-vitro culture, indicating that the micro- environment is important for CLL cell survival. Surprisingly, a recent study using an isotopic labeling method to measure CLL kinetics showed that CLL is not a static disease, resulting simply from accumulation of long-lived lymphocytes, but a dynamic entity in which cells proliferate and die [134]. This finding is in conflict with the conventional view that CLL is character- ized almost exclusively by cell accumulation due to a defect in apoptosis. Consequently, an emerging model emphasizes the importance of prolifera- tion centers, a characteristic of CLL histology, where CLL cells are stimulat- ed by endothelial cells and T cells in the microenvironment provoking cell division and reinforcing resistance to apoptosis [114].

Antigen stimulation The role of antigens in CLL pathogenesis has gained increased attention in recent years [121]. Following reports that elevated titers of IGHV4-34 Ig antibodies were linked with chronic infections caused by the herpes viruses EBV and CMV, IGHV4-34 BCRs were found to be overrepresented among CLL cases with detectable EBV and/or CMV DNA [135]. The restricted IG gene repertoire and the presence of stereotyped BCRs in CLL further suggest a role for antigen drive in disease development and progression. Recent stud- ies on antibody reactivity in CLL indicate that monoclonal CLL antibodies resemble natural antibodies recognizing motifs on apoptotic cells and exoge- nous bacteria [136]. It has therefore been postulated that CLL arises from innate immunity like B-cells rather than adaptive immune responses [121, 136].

Prognostic parameters The prognosis and clinical course in CLL is variable. Some patients die within a few years from diagnosis even with modern therapy, while others have long survival and never require treatment. Most patients fall into an intermediate group having an indolent disease for many years but then even- tually develop a more aggressive form. The unpredictable clinical course has led to the development of several staging systems and prognostic markers. The Rai and Binet clinical staging systems have been used for both research protocols and clinical practice for more than 20 years [137, 138]. Both are based on physical examination and standard laboratory tests to evaluate the presence of anemia/thrombocytopenia, lymphadenopathy and hepato- /. Both systems distinguish three major prognostic subgroups with different survival. However, neither Rai nor Binet systems can reliably predict the clinical course of patients in the nonadvanced stages of CLL. Several prognostic parameters have been evaluated to this end, including a) genetic markers such as cytogenetic aberrations, TP53 mutations, and IGHV

29 gene mutation status/BCR stereotype, b) immunophenotypic markers such as CD38 and ZAP70, and c) RNA-based markers such as lipoprotein lipase (LPL), and CLLU1 [139]. As mentioned above, IGHV mutational status cor- relates strongly with survival in CLL where the expression of mutated (<98% identity) IGHV genes predicts a longer time to treatment and a pro- longed median survival compared with cases expressing unmutated genes. Notably this was also evident among patients with early stages of disease [117]. Mutational analysis of IGHV genes has now become the gold standard for prognostication in CLL, at least in clinial trials, especially since the marker does not change during the clinical course. Both CD38 and ZAP70 were initially considered potential surrogate markers for IGHV mutational status, but have since been confirmed to be independent prognostic markers [139]. Two common genetic aberrations in CLL, deletions of 17p and 11q22-23, are associated with unmutated IGHV status and poor prognosis, whereas 13q14 deletions correlate with the presence of IGHV SHM and a benign disease course. Notably, 17p deletions and TP53 mutations are the only predictive markers for CLL, identifying patients with aggressive dis- ease which is often resistant to standard chemotherapy and with a very short survival [140].

Therapy In general, newly diagnosed patients with asymptomatic early-stage CLL are monitored (“watch and wait”) without therapy unless there is evidence of disease progression. This practice is based on randomized clinical trials which failed to show improvement in overall survival after early treatment with chlorambucil in patients with stage A CLL [141, 142]. Indications for initiation of treatment include progressive marrow failure, symptomatic splenomegaly or lymphadenopathy, progressive lymphocytosis or autoim- mune cytopenias [115]. Monotherapy with chlorambucil, an alkylating agent, has served as initial, front-line therapy for CLL for several decades and this is still considered appropriate by many specialists, particularly in frail patients [143]. Standard frontline treatment for CLL is currently fludar- abine and cyclophosphamide (FC). However, recent guidelines recommend adding rituximab to FC in the standard protocol, and to base the choice of therapy on disease stage, patient performance and cytogenetics [143]. In the presence of the p53 mutation/deletion or in refractory disease, more aggres- sive therapy with alemtuzumab and allogeneic stem cell transplantation is warranted in younger patients [144]. Not all patients can tolerate these in- tense regimens, but can instead be treated with chlorambucil, or alemtuzumab monotherapy. With optimal chemoimmunotherapy regi- mens, complete responses with durable remission can be expected in nearly 50% of CLL patients treated [143].

30 Expression and genomic array studies in MCL Gene expression microarrays Gene expression profiling (GEP) has become an important tool for modern molecular biology that allows the expression of thousands of genes to be examined simultaneously. While the contains more than 20,000 protein-coding genes, only a small fraction is expressed as messenger RNA (mRNA) at any given time. GEP is widely used in current cancer re- search to evaluate global gene expression. A variety of microarray platforms has been developed for this purpose but the basic technique is similar: a glass slide or membrane is spotted or "arrayed" with DNA probes, each rep- resenting a single mRNA molecule, or transcript. Expression arrays are gen- erally categorized as cDNA arrays, where the array features are composed of PCR products, or oligonucleotide arrays (oligoarrays) where the probes are made from synthesized oligomers. In the experiment, purified test RNA is fluorescently labeled and hybridized to the slide/membrane, often simultane- ously with reference RNA to facilitate comparison of data across multiple experiments. After repeated washing, the raw data is obtained by laser scan- ning and analyzed by statistical software [145]. In Paper I in this thesis Affymetrix GeneChip® arrays (Affymetrix, Santa Clara, CA, USA), including HG-U95Av2 (10,000 transcripts), HG-U133A (22,000 transcripts), and HG-U133 Plus 2.0 (39,000 transcripts), were used. The Affymetrix arrays are designed such that probes, 20-25 nucleotides long oligonucleotides, are synthesized directly on the array, 16-20 unique probes for each transcript. RNA is extracted from the cells to be analyzed, and sub- jected to reverse transcriptase to form cDNA. The cDNA is then transcribed and fluorescently labeled creating complementary RNA (cRNA) which is hybridized to the array.

Gene expression profiling of MCL and the proliferation signature In 2003, the Lymphoma/Leukemia Molecular Profiling Project (LLMPP) group reported that MCL has a unique gene expression signature that differs from the signatures of CLL/SLL and DLBCL [68]. Notably, this expression signature was identified both in the classic cyclin D1-positive MCL cases and a small subset of cyclin D1-negative MCL. Besides giving clues to the pathogenesis of the tumor, the GEP revealed that the expression of only twenty genes could identify subsets of patients whose median survival dif- fered by more than 5 years. Interestingly, these twenty genes have primarily functions related to proliferation, such as mitosis (e.g. CENPF, ASPM), DNA synthesis/repair (e.g. POLE2, TOP2A) and the cell cycle (e.g. CDC20, CDK1, CDKN3). Other GEP studies have compared MCL tumors with nor-

31 mal B-cells [146-150]. However, comparative analysis of differentially ex- pressed genes among independent studies, shows a relatively limited degree of overlap [74]. Differences in methodology likely explain the observed differences among these studies. A recent review identified only five genes, all upregulated in MCL, that were reported in at least three papers; CCND1 (cyclin D1), BCL2, CNN3 (Calponin 3), ENPP2 (Autotaxin) and HOMER3 [74]. Overexpression of the anti-apoptotic protein BCL2 has been reported in various types of NHL as well as MCL. The other three genes had not been previously reported to be altered in MCL, but are involved in other tumors and/or cell growth.

Array platforms for analysis of genomic aberrations In the early nineties metaphase comparative genomic hybridization (CGH) was developed as a molecular cytogenetic technique that compensated for difficulties in conventional cytogenetics and FISH [151]. Briefly, the ge- nomic DNA of test and reference samples is isolated and labeled in red and green fluorescent dyes, respectively. Each labeled DNA is subject to com- petitive hybridization to metaphase chromosomes from normal cells and the ratio of red and green fluorescent signals is measured along each chromo- some. Although metaphase CGH had a significant impact in the field of can- cer cytogenetics, the resolution for detection of CNAs was limited to 5-10 Mb. To circumvent this limitation array-based CGH was developed. In array- CGH cloned or synthesized DNA fragments, which are spotted onto a micro- scopic glass-slide using robotic devices, replace chromosomal targets. The ratio between fluorescence levels of control and test DNA for each clone can be plotted using data analysis software to visualize dosage variations. The array resolution is limited by the size of the DNA targets and the density of coverage over the genome, but can be as low as 5-10 kilobases. Array-CGH has enabled the characterization of DNA copy number "signatures" or pro- files for various malignancies, including MCL [152]. SNP array is yet another type of DNA microarray which is used to de- tect polymorphisms within a population. A SNP, a variation at a single DNA nucleotide, is the most frequent type of variation in the genome - more than 10 million SNPs have been identified [153]. The basic principles of SNP array are the same as for the CGH microarray, including the convergence of DNA hybridization, fluorescence microscopy, and solid surface DNA cap- ture. The difference lies in the application of allele-specific oligonucleotide probes in the SNP array, which makes it possible to specifically detect the two different alleles of each SNP. An important application of SNP arrays is in the area of GWAS, seeking association between phenotype and genotype. However, in cancer genomics SNP arrays have also been successfully used to detect CNAs. Dosage variations can be determined from SNP array da- tasets by comparing the hybridization signal intensities at each SNP locus

32 with corresponding signal intensities from normal genomes. An additional benefit of SNP arrays is the ability to detect copy number neutral loss of heterozygosity (CNN-LOH), also known as , by combin- ing genotyping and copy number data [154]. In our study of CNAs in MCL, we employed the Affymetrix GeneChip1 Mapping 250K NspI array. Affymetrix SNP arrays involve oli- gonucleotide probes spotted on gene chips and were adapted from the micro- array hybridization chip technology utilized for gene expression studies. The arrays contain up to 40 separate 25-mer oligonucleotide probes for each SNP locus, representing both mismatch and perfect match probes. Genomic DNA is prepared by restriction enzyme digestion followed by adapter ligation and single-primer PCR. After DNA labeling and hybridization, fluorescence intensities are measured for each allele probe of each SNP (Figure 4). A large body of work exists for copy number analysis using Affymetrix SNP arrays, and several algorithms have been developed, including dChipSNP, Copy Number Analysis Tool (CNAT) and Copy Number Analyzer for GeneChip (CNAG) [154].

Copy number aberrations in MCL The various technologies described above have all been utilized to analyze genomic aberrations in MCL, including metaphase CGH arrays [98, 155- 157], bacterial artificial chromosome CGH arrays [92, 158-161], and SNP- arrays [149, 162-164]. These studies have revealed that MCL is character- ized by a high load of CNAs compared to many other lymphomas [73]. As mentioned above, the most common recurrent aberrations include losses at 1p, 8p, 9p, 9q, 11q, 13q, 17p and gains at 3q, 8q, 15q, 18q [74, 85]. It seems plausible that these chromosomal regions contain crucial genes implicated in pathogenesis. Several candidate genes have been identified in these regions as previously mentioned, such as ATM (11q), TP53 (17p), MYC (8q) and BCL2 (18q), and notably most of them are involved in cell cycle regulation and DNA damage response pathways [74, 85]. However, for many common- ly affected regions, such as 1p, 3q, 8p, 9q and 13q, the search for candidate gene(s) is still ongoing. The important role of the ATM gene in the patho- genesis of MCL was discussed above. Recurrent high-level gains (amplifica- tions) have been observed at a few loci in MCL, including 3q, 8q (MYC), 10p (BMI1), 12q (CDK4) and 18q (BCL2). The CDKN2A locus at 9p21 is targeted by recurrent homozygous deletions in 20-30% of highly prolifera- tive variants of MCL, but in less than 5% of typical cases [165, 166]. Ampli- fication of the CDK4 locus and overexpression of CDK4 is also associated with highly proliferative MCL, further highlighting the importance of the G1 to S phase transition in MCL.

33

Figure 4. The principle behind the Affymetrix GeneChip1 Mapping 250K NspI array (adapted from Affymetrix GeneChip® Human Mapping 500K Datasheet).

The tumor suppressor p53 As mentioned above p53 is considered the master guardian of the cellular genome and a pivotal tumor suppressor. It is among the most extensively studied of any human genes and proteins with more than 58.000 PubMed- listed publications so far. Below follows a brief summary on the structure and function of p53 and its role in cancer.

Structure and function The transcription factor p53 is encoded by the TP53 gene on the short arm of (17p13.1) [167]. The TP53 gene is 20 kb long and has 11 exons but translation starts in exon 2 to yield a 2.2 kb mRNA molecule

34 [168]. The human p53 protein can be divided into five domains, each with a specific function: 1. A transactivation domain that mediates the interaction of p53 with a multitude of interacting proteins, such as coactivator mol- ecules and negative regulators. 2. A proline rich domain whose role is unclear, but may be im- portant for apoptotic activity. 3. A DNA binding domain, the core domain, where roughly 90% of TP53 mutations in cancer are located. 4. An oligomerization domain that mediates first dimerization and then tetramerization of p53 molecules. 5. A C-terminal regulatory domain [169]. As indicated above, p53 normally exists in the cellular nucleus as a homo- tetramer, or a dimer of dimers, which is essential for its activity [169]. Under normal circumstances p53 has a very short half-life in the cell as it is rapidly ubiquinated, transported from the nucleus and degraded [83]. This negative regulation of p53 is largely mediated by the Mdm2 protein whose expression is induced by p53 itself, forming a negative feedback loop [170]. Therefore, p53 function in non-stressed cells is held at a low basal state. A variety of cellular physiologic stresses can lead to the activation of p53, in- cluding anoxia, genomic damage and disturbance in the growth-regulatory circuit [171]. The activation of p53 involves both its rapid accumulation in the nucleus and a conformational change that facilitates its transcriptional regulator function. The most important event in the activation of p53 is phosphorylation of its N-terminal transactivation domain, which can be me- diated by a variety of kinases, such as kinases involved in the genome integ- rity checkpoint (ATR, ATM, CHK1/2) [83, 172]. After activation by some kind of a stress signal, p53 can promote cell cy- cle arrest, apoptosis and cellular senescence (Figure 3). p53 functions mainly as a sequence-specific transcription factor that binds directly to specific DNA sequences and transactivates nearby genes [173, 174]. Many of the p53 target genes that have currently been identified encode proteins that are inti- mately involved in apoptosis (e.g. BAX) or cell cycle progression (e.g. CDKN1A/p21) [175, 176]. However, the p53 response is complex and is increasingly recognized to be cell type restricted or context restricted. In addition to its transcription factor activity, p53 has recently been found to possess diverse biochemical activities. As an example, the interaction of p53 with members of the Bcl2 family in the cytoplasm leads to the release of cytochrome c and apoptosis [177]. The consequences of p53 activation are thought to be proportionate to the activating stress signal. Severe stress in- duces apoptosis and senescence, whereas milder stress leads to a transient growth arrest while the cell attempts to repair the damage caused by it [171].

35 TP53 aberrations in cancer As previously stated somatic TP53 mutations are frequent in human especially in advanced stage or in subtypes with aggressive clinical behavior [178]. The prevalence is lower in hematological malignancies (<20%) than in solid tumors (>50%) [179]. Nevertheless, TP53 mutations have been associated with poor prognosis in a number of lymphoid malignancies, such as DLBCL and CLL [179]. Although not extensively studied in MCL, TP53 mutations were detected in 20% of samples in one cohort and were associated with poor survival [180]. The two main mechanisms of TP53 inactivation are mutations, usually single-base substitutions, and loss of alleles [178]. In contrast to other tumor suppressor genes (TSGs) most mutations targeting TP53 are missense, rather than nonsense or frameshift, and thus cause single amino-acid substitutions rather than truncating or destroying the protein [178]. Almost 90% of muta- tions are located between codons 125 and 300, corresponding roughly to the DNA binding domain. The single most common mutation type is C:G>T:A substitution at CpG sites. CpG dinucleotides have a high mutation rate be- cause the cytosine at CpG sites may be methylated at position 5’. This desta- bilizes the nucleotide which undergoes spontaneous deamination into thy- mine much more rapidly than unmethylated cytosine [178]. Almost a third of all missense TP53 mutations arise in six “hotspot” codons, namely 175, 245, 248, 249, 273 and 282, [181] and in five of these residues CpG dinucleotides are mutated [178]. These hotspot mutations can be further classified as “con- tact” (248 and 273) or “structural” (175, 245, 249, and 282), depending on whether they make contacts with DNA or support the structure of the DNA- binding surface [178]. The structural and functional impact of TP53 muta- tions can thus vary. Some hotspot mutations have been reported to demon- strate allele-specific “gain-of-function” effects which confer new properties, such as oncogene activity, to the mutant p53 protein. Alternatively, the bind- ing of a mutant p53 to wild-type p53 may compromise the function of the entire tetramer. Such dominant–negative effects may explain why, unlike other TSGs, TP53 is so frequently targeted by missense mutations [178].

Epigenetics and gene silencing by DNA methylation Introduction to epigenetics It has been long established that human gene expression is tissue specific. Since the genetic material is virtually the same in all nucleated cells in an organism, differential gene expression cannot be explained by genetics. In contrast, epigenetic mechanisms represent reversible modifications that af- fect gene expression without altering the DNA sequence itself. In recent times, several epigenetic mechanisms have been identified, including DNA

36 methylation, histone modification and Polycomb-trithorax gene regulation. Two groups of regulators, the Polycomb repressors and trithorax activators, maintain the correct expression of certain key developmental regulators (in- cluding the homeotic genes) by changing chromatin structure. DNA methyl- ation, discussed in more detail below, is now recognized as an important epigenetic mechanism which interacts with histone modification.

DNA methylation DNA methylation is a powerful mechanism for silencing gene expression and maintaining stability of the genome [182]. In mammals DNA methyla- tion usually takes place at the 5 position of the cytosine base within CpG dinucleotides, which are found primarily in regions of large repetitive se- quences or in CpG islands associated with 60% of gene promoters [183]. While the majority of the 30,000 CpG islands in the human genome remains unmethylated, repetitive sequences are heavily methylated [183, 184]. Dramatic changes in DNA methylation occur during human development. During gametogenesis in gonads the demethylated genome of primordial germ cells undergoes rapid do novo methylation so that sperm and egg cells have substantially methylated genomes. After fertilization, a wave of de- methylation occurs at the preimplantation stage, to be succeeded shortly after by large-scale de novo methylation beginning at the pre-gastrulation stage [185]. Thus decisions concerning the transcriptional stage of many genes are made early in embryogenesis. The methylation pattern is “heritable”, as it is maintained through many rounds of DNA replication. Maintenance of meth- ylation occurs following DNA replication because almost invariably both strands of the parent DNA are methylated [183]. Both hemi-methylated daughter strands are subsequently methylated by a class of enzymes termed DNA methyltransferases (DNMTs) [83, 185]. The means by which CpG methylation affects gene transcription are not entirely resolved. CpG methylation in the vicinity of a gene promoter causes transcriptional repression of the associated gene, whereas removal of methyl groups often causes de-repression of transcription. There is some evidence, however, that DNA methylation primarily targets genes that are already si- lent rather than active promoters, and thus may be a secondary event [184]. At any rate there seems to be a complex interaction between DNA methyla- tion, histone acetylation and histone methylation, the three covalent modifi- cations most strongly associated with a repressed chromatin state [186]. Alt- hough some studies suggest that histone modifications precede DNA meth- ylation, there is also evidence that DNA methylation influences the histone modification pattern [184, 187]. But what are the potential mechanisms of DNA methylation-mediated transcriptional repression? DNA methylation may repress transcription both directly, by inhibiting the binding of specific transcription factors, and indi-

37 rectly, by recruiting methylated CpG-binding proteins (MECPs) [184]. MECPs act as transcriptional repressors by recruiting a corepressor complex consisting of the transcription factor repressor Sin3A and histone deacetylas- es (HDAC) [185]. MECPs can also bind histone methyltransferase so that after the deacetylation of H3K9 (the lysine at position 9 in the histone 3 pro- tein) it becomes methylated. The methylated H3K9 is a target for hetero- chromatinizing proteins such as HP1 which cause the chromatin to become condensed and transcriptionally inactive [185]. There are a growing number of human diseases associated with failure of the proper establishment and/or maintenance of epigenetic information, in- cluding DNA methylation. Examples of diseases characterized by aberrant DNA methylation include malignancies and imprinting disorders [182]. The next section discusses genomic methylation in malignant disorders.

Aberrant methylation patterns in malignancies As mentioned above DNA methylation in normal cells occurs mainly in repetitive genomic regions, such as satellite DNA and parasitic elements (transposable elements and endogenous retroviruses). In contrast CpG is- lands are generally unmethylated, although exceptions are increasingly being identified [182]. Aberrant DNA methylation in tumors, on the other hand, typically manifests as global hypomethylation and local hypermethylation of CpG islands [188, 189]. Much of this “global hypomethylation” can be at- tributed to the loss of methyl groups attached to the DNA of highly repeated sequences in the cell genome [182]. This loss is correlated with chromoso- mal instability, but it remains unclear whether it actually causes this instabil- ity. As CpG islands, the targets of hypermethylation in cancer, are usually affiliated with gene promoters, this means that the de novo DNMTs are actu- ally inappropriately silencing certain genes. In recent years it has become apparent that promoter methylation is as important in shutting down TSGs as are the various mechanisms of somatic mutations. More than half of the TSGs that are involved in familial cancer syndromes because of germline mutation have been found to be silenced by promoter methylation in sporad- ic cancers. Examples of genes that have been reported to be hypermethylated in human tumor cell genomes include the cell-cycle regulator RB1 (sporadic retinoblastoma), the CDK4/6 inhibitor CDKN2A (diverse tumors) and the MDM2 inhibitor ARF (lymphoma) [83]. All three genes have potential roles in the pathogenesis of MCL as described earlier in this thesis. As tumor for- mation proceeds, the silencing of genes through promoter methylation can also involve the “caretaker” genes that are responsible for maintaining the integrity of the DNA sequences in the genome.. As implied above, the fre- quency of methylation of a specific gene varies dramatically from one type of tumor to the next. Consequently, the overall methylation profile may be specific for a particular malignancy. In support of this, studies of hematolog-

38 ical malignancies have revealed that the methylation pattern is quite hetero- geneous among this group of neoplasms [190]. Notably, a recent study of hematological malignancies found that the local DNA hypermethylation in mature B-cell lymphomas is accompanied by gene-specific hypomethylation in contrast to the global hypomethylation described previously [190]. It has been postulated that aberrant hypomethylation can lead to increased tran- scription of genes with oncogenic potential. Furthermore, it has been report- ed that CpG sites located in CpG island shores (within 2 kb of islands), ra- ther than in the CpG islands themselves, are more differently methylated in cancer and during cellular differentiation [191, 192].

Laboratory methods for analyzing DNA methylation A variety of techniques and tools has been developed to detect and analyze DNA methylation patterns [193, 194]. A number of studies have used meth- ylation sensitive restriction enzymes that digest DNA based on methylation status [195, 196], while immunoprecipitation methods use antibodies that bind methylcytosine or methyl-CpG binding proteins [197, 198]. Currently, bisulfite reaction based methods are considered the gold standard to analyze the methylation state of individual cytosines [199, 200]. Bisulfite treatment of DNA leaves 5-methylcytosine residues unmodified while unmethylated cytosine residues are converted to uracil and read as thymine. Bisulfite se- quencing yields precise information but has limited coverage, although re- cently genome-wide bisulfite sequencing techniques have been developed [201]. In contrast to bisulfite sequencing, pyrosequencing gives a quantita- tive estimate of DNA methylation levels at specific CpG sites and has been reported to correlate well with array platforms [202]. Microarray-based methods allow for genome-wide analysis of DNA methylation. In papers III and IV we used the genome-wide Illumina Infinium HumanMethylation27 BeadChip array (Illumina, San Diego, USA) which measures methylation levels at 27,578 CpG dinucleotides, covering 14,495 genes. Prior to applica- tion to the array each DNA sample is bisulfite converted, whole-genome amplified and enzymatically fragmented. Illumina’s BeadArray Technology is based on 3-micron silica beads in microwells, where each bead is covered with multiple (>105) copies of a specific oligonucleotide probe. In the meth- ylation assay two bead types target each CpG locus. One bead type is cov- ered with allele-specific probes specific to the methylated (C) state whereas the other bead has probes specific to the unmethylated (T) state. Annealing of allele-specific primers is followed by single-base extension using DNP- and Biotin-labeled ddNTPs (figure 5). After extension, the array is fluores- cently stained with antibodies against the ddNTPs, scanned and the intensi- ties of the unmethylated and methylated bead types measured. An earlier version of the Illumina array, the GoldenGate Methylation Cancer Panel I, targets 1,505 CpG sites and has been successfully used for methylation pro-

39 filing in hematological malignancies, including mature B-cell lymphomas [190, 203, 204]. Thus, microarray platforms are currently widely applied to study genomic methylation in malignancies, although sequencing-based profiling methods will probably gain ground in the future [205].

Figure 5. The principle behind the Illumina Infinium HumanMethylation27 Bead- Chip array. The DNA specimen to be analyzed is bisulfite converted, amplified and fragmented (left panel). The Illumina array uses two different bead types per CpG site; the M type is covered with oligonucleotide probes that match the methylated site, while the U type has probes that match the unmethylated site. When the probe matches the bisulfite treated DNA sequence annealing is followed by single base- extension and detection (right panel). Adapted from Weisenberger et al, Compre- hensive DNA Methylation Analysis on the Illumina® Infinium® Assay Platform (Ap- plication Note: Illumina® Epigenetic Analysis).

Analysis of gene promoter methylation in MCL and CLL The pattern and impact of promoter methylation in lymphomas, including MCL and CLL, has been the subject of a few studies. One study of 367 he- matological neoplasms revealed a highly heterogeneous picture of DNA methylation among mature B-cell malignancies. Tumors of germinal center origin (i.e. DLBCL, FL, BL) had the most prominent promoter hypermethyl- ation, whereas MCL had intermediate and CLL low degrees of de novo DNA methylation [190]. Notably, polycomb repressor group (PcG) targets in embryonic stem cells were significantly enriched among hypermethylated

40 genes in several entities, among them MCL [190]. Caldwell et al have com- pared DNA methylation between three small B-cell lymphoma subtypes (i.e. FL, CLL/SLL and MCL) using oligonucleotide microarrays and/or sequenc- ing. The first microarray study targeted only 156 loci but detected preferen- tial methylation of multiple gene promoters in FL compared to MCL or CLL. However, no statistically significant difference in methylation between MCL and CLL was detected [206]. A later study using a microarray contain- ing 8544 CpG island clones revealed a subtype-specific methylation pattern. In hierarchical clustering all the MCL cases clustered together whereas the methylation pattern of CLL was more heterogeneous, with one subgroup clustering close to MCL and a second subset clustering with FL [207]. Again, more CpG loci were hypermethylated in FL compared to MCL and a subset of CLL cases. A study applying a novel massively parallel sequenc- ing-by-synthesis method (454 sequencing) targeting 25 genes in ALL, FL, CLL and MCL reported both an increase in methylation and a progressive spreading of methylation from the periphery toward the center of CpG is- lands in ALL and FL compared to CLL and MCL [208]. DNA methylation patterns have also been studied individually for each disease. A recent study applied the HELP (HpaII tiny fragment En- richment by Ligation mediated PCR) assay to compare genome-wide meth- ylation profiles of MCL specimens to normal naïve B-cells [209]. Four genes (CDKN2B, MLF-1, PCDH8, HOXD8) were confirmed to be hyper- methylated in MCL and four genes (CD37, HDAC1, NOTCH1, CDK5) were hypomethylated [209]. Another study focused on genes whose expression was induced in MCL cell lines after treatment with DNMTs and/or HDAC inhibitors. Subsequently promoter methylation levels were measured in pri- mary MCL samples using the MassArray assay [210]. In this way a set of genes (SOX9, HOXA9, AHR, NR2F2, ROBO1) was identified where the methylation degree correlated with higher proliferation, increased number of chromosomal abnormalities, and shorter survival. Using the Infinium HumanMethylation27 BeadChip array our group was the first to demonstrate significant differences in gene methylation pat- terns between IGHV mutated and unmutated CLL cases [211]. In IGHV un- mutated CLL several tumor suppressor genes were methylated, whereas genes enhancing cell proliferation and tumor progression were unmethylat- ed. Previous reports in CLL had shown a strong correlation between promot- er methylation and transcriptional silencing for specific genes (e.g. ZAP70, TWIST2, DAPK1, HOXA4) [212-215].

41 AIMS

The general aim of the studies in this thesis was to investigate genetic aber- rations in MCL, and genomic methylation profiles in MCL and CLL. More specifically, the aims were as follows:

I. To apply high-resolution SNP arrays to screen for novel aberrations in MCL and to correlate individual CNAs with genetic complexity, the proliferation signature and survival.

II. To investigate the spectrum and clinical impact of TP53 mutations and 17p deletions in MCL.

III. To apply DNA methylation microarrays to study the genomic meth- ylation profiles in MCL and CLL and to identify genes that differed in methylation status between the two diseases.

IV. To apply DNA methylation microarrays to compare the genomic methylation profiles in selected CLL subsets with stereotyped BCRs.

42 MATERIALS AND METHODS

Patient Material Paper I includes specimens from 31 newly diagnosed MCL patients treated at Karolinska University Hospital, Huddinge, Sweden. Most tissue speci- mens were collected from lymph nodes (n=26), while a few were from spleen (n=3) or tonsils (n=2). The median tumor cell content was 82% as determined by flow cytometry. All cases met standard diagnostic criteria for MCL [22], including over-expression of the cyclin D1 protein and/or pres- ence of the t(11;14) [22]. Survival data were available for all 31 patients, and the median OS was 51 months. Paper II includes a large cohort of 119 MCL cases that were collected from university hospitals in Sweden (n=111) and Finland (n=8). Tumor samples were collected between 1983 and 2004 (median 1999), mainly from lymph nodes, but also from spleen, tonsils, bone marrow or other tissues. All cases fulfilled the WHO diagnostic criteria for MCL [22]. Survival data were available for all 119 patients, and the median OS was 36 months. Paper III includes a subset (n=20) of the primary MCL tumors ana- lyzed in paper I. Here, the 10 cases with the lowest and highest proliferation gene expression signature, respectively, were selected from the original co- hort of MCL cases (n=31). In paper III, the 20 MCL specimens were com- pared with 30 CLL tumor samples that were collected from collaborating institutes in Greece (n=18) and Sweden (n=12). All CLL samples were col- lected from peripheral blood and had a tumor percentage of 70 percent. The diagnosis of CLL was based on recently revised diagnostic criteria [115]. The selected CLL cases all carried stereotyped BCRs, half (n=15) were IGHV mutated with subset #4 stereotyped BCRs and the other half were IGHV unmutated with subset #1 stereotyped BCRs [120]. Paper IV analyzed the same 30 CLL subset #1 and #4 CLL cases as Paper III, but additionally included 9 CLL cases that belong to subset #2. In Papers III and IV CLL and/or MCL cell lines were used for valida- tion experiments. The HG3 CLL cell line is EBV transformed and derived from an IGHV unmutated CLL patient [136]. The MCL cell lines JVM-2 and Jeko-1 were purchased from Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ), Braunschweig, Germany, (http://www.dsmz.de/).

43 SNP array In paper I, the Affymetrix GeneChip1 Mapping 250K NspI array was ap- plied to 31 MCL samples according to standard protocols (Gene Chip Map- ping 500K Assay Manual, Affymetrix, Santa Clara, CA, USA) as described in the paper. A reference set consisting of 70 normal human specimens ana- lyzed at the Uppsala Array Platform was included. The Nexus Copy Number 5.0 software was employed for copy number analysis. The Nexus SNP Rank Segmentation algorithm was used to define segments of copy number change, using P = 10-6 and a minimum of 20 consecutive SNPs as segmenta- tion settings [216]. To define losses and gains log2ratio thresholds of 0.15 for segments <5 Mbp, and 0.10 for segments >5 Mbp were applied, while high-level gains (amplifications) were defined as regions with log2ratios 0.5. These cutoffs and settings were based on previous experience, includ- ing studies on CNAs in CLL tumor specimens [217, 218]. However, as the signal to noise ratio was generally lower in MCL than CLL (probably due to a higher proportion of non-tumor cells) somewhat lower log2 thresholds were applied for MCL as compared to CLL ( 0.20). To reduce false-positive find- ings, small segments (<200 Kbp) and those located in telomeric/centromeric regions were excluded from further analysis, as were. CNAs overlapping more than 50% with known CNVs assembled in the ”Database of Genomic Variants” (http://projects.tcag.ca/variation/). A recently developed algorithm to identify for tumor specific CNN-LOH regions, in which the LOH is not caused by altered copy number, was applied [219].

GEP array and proliferation signature In paper I the global gene expression profiles of the 31 MCL specimens were examined using microarray technology. Total RNA was prepared using TRI- zol method and the cRNA synthesis for microarray experiments was per- formed according to standard Affymetrix protocols. The gene expression profiling was carried out using three different generations of the Affymetrix expression microarrays: HG-U133 Plus 2.0 arrays (17 cases), HG-U133A arrays (10 cases) and HG-U95Av2 arrays (4 cases). The use of several dif- ferent generations of arrays resulted from the fact that the samples were not analyzed in one batch. Instead each sample was analyzed immediately on arrival to the facility using the most recent version of the microarray. The raw gene expression data was analyzed using the R statistical computing language (http://www.r-project.org) and Bioconductor project packages (www.bioconductor.org). To calculate the proliferation gene expression sig- nature for each case the expression levels of 20 genes were used as previous- ly described [68]. First, the expression levels for each gene were log- normalized and mean centered across all samples. Next, the proliferation

44 signature average for a given sample was computed as the average of these 20 mean centered gene expression values. The resulting proliferation signa- ture is a continuous numerical value which has been reported to inde- pendently predict survival among MCL patients [68].

Detection of TP53 mutations and deletions The gold standard for detection of TP53 mutations in patient specimens is direct (Sanger) sequencing. Alternative methods for mutation screening in- clude single stranded conformation polymorphism, denaturing gradient gel electrophoresis and denaturing high performance liquid chromatography [220]. More recently developed techniques include mutation arrays and fluo- rescent melting curve analysis [221, 222]. The AmpliChip p53 Research Test is a microarray based resequencing assay that queries exons 2-11. Comparison of the AmpliChip p53 with direct sequencing in 98 CLL cases revealed increased detection of mutations using the microarray although it failed to detect microdeletions [221]. Similarly, high-resolution melting curve analysis was able to confirm all TP53 mutations detected by direct sequencing in ovarian and breast tumor samples [222]. In the future, these novel techniques may enable faster and more efficient screening of TP53 mutations in patient samples. In Paper II we chose direct sequencing as this method had been well established in our laboratory. Genomic DNA from 119 MCL cases was PCR amplified using three primer sets, targeting TP53 exons 4, 5/6 and 7/8 re- spectively, prior to product purification and sequencing as previously de- scribed [223]. The decision to sequence only exons 4 to 8 can be justified as these exons encode for the core DNA-binding domain of p53, where roughly 90% of reported cancer-related TP53 mutations are located. Identified TP53 mutations were analyzed using the p53 Knowledgebase (http://p53.bii.a- star.edu.sg/index.php), the International Agency for Research on Cancer (IARC) database (http://www-p53.iarc.fr/) and the UMD p53 mutation data- base (http://p53.free.fr/index.html). For detection of 17p copy number loss, one of the array techniques described previously can be used, such as CGH or SNP microarrays. An alternative approach, used for selected cases in Paper II, is to apply quantita- tive real-time PCR, where the copy number status for 17p in the tumor spec- imen is compared to a reference chromosomal region (Z) and a reference DNA specimen according to the Delta Delta Ct (cycle threshold) formula:

45 Methylation array and analysis of methylation data In papers III and IV, we applied the Illumina Infinium HumanMethylation27 BeadChip array (Illumina, San Diego, USA) to measure methylation levels at 27,578 CpG dinucleotides (14,495 genes) in 20 MCL and 30 CLL bisul- fite converted specimens according to the standard Illumina protocol [211]. Notably, over 70 percent of the CpG sites targeted by the HumanMethyla- tion27 BeadChip array are located within CpG islands. This may be relevant and limit the scope of our studies, as it has been reported that methylation of CpG sites located in CpG island shores just outside the CpG islands may have a stronger association with carcinogenesis and tissue differentiation than methylation of the islands themselves [191, 192]. The methylation status of each CpG site is represented by the meth- ylation index (MI, or Beta value). The MI equals C/(C+T), where C repre- sents the fluorescence signals from the methylated alleles and T signals from the unmethylated alleles. To identify genes that differed significantly in methylation status between subgroups we applied an empirical Bayes mod- erated t test using a p-value level <0.05 as a cutoff. A difference in mean MI 0.30 was added as an additional filter to focus on genes with the largest differences in mean methylation levels between groups. Principal component analysis (PCA) was carried out using the “princomp” function in the Matlab computing software and generation of heatmaps and dendrograms was done using Genesis TreeView version 1.2.7 free software (www.genome.tugraz.at) [224]. The gene ontology tree machine (GOTM) online tool was used for gene ontology analysis of differentially methylated genes and the Ingenuity Pathway Analysis (IPA) program was used to inves- tigate biological functions of the genes.

RQ-PCR analysis Quantitative genomic PCR was used in paper I to confirm the presence of a novel loss at 20q. The same technique was used in Paper II to detect dele- tions at 17p. Gene expression was measured using RQ-PCR in Papers III and IV. RQ-PCR primers were designed using the Primer3 software (Broad insti- tute, Boston, USA). RQ-PCR analysis was performed using Maxima® SYBR Green/ROX qPCR master mix and the Stratagene Mx 3005p detec- tion system. Differences in expression were calculated using the Max Pro QPCR software with ACTB (actin, beta) as a reference gene and the Ct method.

46 Re-induction of gene expression Gene promoter methylation may cause or contribute to the silencing of gene expression. The impact of gene methylation on expression was therefore investigated for selected genes. To this end MCL cell lines (JVM-2 and Jeko-1), primary CLL cells or CLL cell lines (HG3) were treated with the DNMT inhibitor 5-aza-2´-deoxycytidine (DAC) and the HDAC inhibitor trichostatin A (TSA). Previous studies have indicated a possible additive or synergistic effect when TSA is used in combination with DAC treatment in transcriptional reactivation experiments [225]. The cells were cultured in RPMI supplemented with glutamine, fetal bovine serum and penicil- lin/streptomycin until confluent. After counting and splitting, the cells were cultured for 72 hours in RPMI media supplemented with one of the follow- ing two treatments: (i) medium containing DAC in increasing concentrations or (ii) medium containing DAC for 72 hours followed by addition of TSA in the last 12 hours. Negative control cells were included.

Pyrosequencing Pyrosequencing was performed for selected genes in MCL and CLL pa- tient samples to confirm that they were differentially methylated. In addition, pyrosequencing was applied to measure the reduction in gene methylation after treatment. Prior to sequencing the DNA samples were bisulfite convert- ed and PCR amplified. Three pyrosequencing primers were designed for each gene using the PyroMarkTM software. A forward and reverse primer pair (one biotin labelled in the 5’ end) was designed for PCR amplification while the third primer was made for sequencing the amplified region. The amplicon was designed to contain the specific CpG site targeted by the Illu- mina methylation array. Following PCR amplification the product was im- mobilized to Streptavidin Sepharose, followed by incubation with annealing buffer containing the sequencing primer. The analysis was carried out using the PyroMarkTM Q24 pyrosequencer instrument and CpG site methylation analysis was performed with the PyroMark Q24 software.

Statistical analysis ANOVA, Fisher’s exact, Mann-Whitney U and t-test were applied to deter- mine statistical differences between studied groups. OS was estimated using the Kaplan-Meier method and comparison of survival between subgroups was made using the Log-Rank test. For estimating the correlation between survival and the proliferation signature, a Cox proportional hazards model was fit and the likelihood ratio P value was reported. All calculations were

47 performed using Statistica version 8.0/9.0/10.0 (StatSoft, Scandinavia AB, Uppsala, Sweden).

48 RESULTS & DISCUSSION

Paper I: Screening for genomic aberrations in MCL MCL tumors typically carry many genomic aberrations, such as gains and deletions, which have been studied extensively using various techniques [73, 92, 98, 149, 155-164]. However, few studies have analyzed the correlation of specific CNAs with prognostic parameters such as the proliferation gene expression signature or the overall load of CNAs. Therefore, in the present study, we systematically explored the relationship between the most com- mon individual CNAs and the proliferation signature on one hand and over- all genomic complexity on the other hand. In addition, we performed a care- ful literature search to identify potential novel CNAs in MCL in our dataset. Thirty-one primary MCL tumor specimens from untreated patients were included in the study. All cases carried at least one genomic aberration, with a median number of 14 CNAs (8 losses and 6 gains) per case. The pattern of CNAs observed was in close agreement with earlier studies of MCL, where the most commonly detected aberrations were losses at 1p, 6q, 8p, 9p, 9q, 11q, 13q, 17p and gains at 3q, 8q, 15q, 18q. A total of 192 recurrent (present in 2 cases) minimal altered regions (MARs) were defined by overlapping genomic CNA segments.We were able to confirm a number of MARs that had just recently been reported [226]. Furthermore, one interesting novel loss at 20q13.2 was identified and subsequently confirmed by RQ-PCR in all five (16%) cases. Although the 20q loss varied in size, the MAR only con- tained a single gene, ZFP64, which encodes for a poorly characterized protein. The ZFP64 gene has recently been reported to by hypermeth- ylated in gastric and colorectal cancers, which suggests a tumor suppressor function [227, 228]. A relationship between high tumor proliferation and short survival in MCL is well established [85, 229-231]. A high proliferation gene expression signature predicted inferior survival in our cohort (P < 0.0001), confirming the prognostic power of this 20-gene signature. To investigate the associa- tion of CNAs with the proliferation signature, cases were divided into high and low proliferation signature groups (with scores above or below the me- dian, respectively). High proliferation signature was associated with an in- creased number of CNAs larger than 15 Mbp (Figure 6), but there was no statistical correlation with smaller CNAs. Therefore, clinically aggressive MCL cases seem to be characterized by many large genomic aberrations. In

49 addition to the correlation with CNA size and number, the proliferation sig- nature was also associated with specific CNAs, including gains at 7p21.3- p22.1 and losses at 9q13-q22.31. Indeed, both 7p gains and 9q losses have previously been associated with clinically aggressive variants of MCL [98, 157, 160] and recently another study identified a relationship between the same CNAs and a high proliferation signature [226].

Figure 6. A high proliferation signature was associated with many large CNAs.

The next step was to explore whether individual MARs correlated with increased genomic complexity (here defined as a higher number of CNAs), and this was found to be true for losses at 1p, 8p, 13q and 17p. The strong relationship of 8p losses to genomic complexity was present regardless of CNA size (using cutoffs of <1 Mbp, 1-5 Mbp and >5 Mbp). Losses at 8p have previously been associated with inferior survival in MCL supporting the view that certain aberrations may be related to tumor proliferation, ac- cumulation of DNA damage and clinical behavior [98, 159]. Cases with losses at 13q, in particular large (>5 Mbp) losses, were also characterized by a high degree of genetic complexity. Likewise, 17p losses were associated with increased number of CNAs >5 Mbp long. The same observation has been made in CLL where patients with 17p deletions targeting TP53 carried a higher frequency of large genomic aberrations than other clinical sub- groups studied [218]. As most of the 31 MCL cases studied were characterized by a large num- ber of CNAs, unsupervised hierarchical clustering was performed based on the CNA regions to see if certain aberrations tended to co-occur. More than half of cases clustered together in the largest of four subgroups, which was not characterized by any specific pattern of CNAs. In contrast, the second cluster had a distinct pattern characterized by the presence of 11q loss but complete absence of 13q loss, larger 8q gains, 3q gains and 7p gains. Inter-

50 estingly, 3q and 7p gains often co-existed as five of six cases with 7p gain also had 3q gains. This implies that some aberrations tend to occur together in MCL as a few studies have previously suggested [97, 232]. However, larger studies are needed to further clarify the relationship between different CNAs in MCL. Finally, losses at two chromosomal regions, 13q and 19p, were associated with improved survival. In previous studies of MCL 13q14 deletions have been associated with short survival [97, 159]. In our cohort, however, a - tionship was identified between the size of the 13q deletion and prognosis, as patients with relatively small, single losses targeting 13q14 had short surviv- al (< 1 year) while patients with large and/or complex losses at 13q had much better survival (mean 7 years). These results suggest that it might be of clinical importance to distinguish between different types of 13q deletions in MCL. However, this finding should be confirmed in a larger cohort. To summarize, this high-resolution genomic analysis of 31 primary MCL cases identified a novel loss at 20q targeting a single gene, ZFP64. Unsuper- vised hierarchical clustering identified subgroups that differed in CNA pat- terns. Known genomic aberrations were linked to high proliferation scores (7p gains and 9q losses), increased genomic complexity (losses at 8p, 13q and 17p) or improved survival (losses at 13q and 19p). Further studies are required to identify the target genes at each of these genomic loci and to investigate their contribution to MCL pathogenesis.

Paper II: Impact of TP53 mutations and 17p deletions in MCL TP53 mutations have been associated with poor prognosis in both DLBCL and CLL [179]. Less has been reported on TP53 mutations in MCL although a study from the Leukemia and Lymphoma Molecular Profiling Project (LLMPP) group identified TP53 mutations in 20% of MCL samples and an association with inferior survival [180]. The prognostic relevance of 17p deletions targeting the TP53 locus in MCL is even less clear although several studies have reported an association with poor prognosis [155, 158, 160, 233]. The aim of this study was to further investigate the prognostic impact of TP53 mutations and 17p deletion in a large cohort of MCL patients (n=119). In addition, the spectrum and characteristics of TP53 mutations in MCL were analyzed. Direct sequencing of TP53 in 119 MCL cases identified 18 cases that car- ried a total of 19 TP53 mutations, including one case with a silent mutation. Therefore, the prevalence of non-silent TP53 mutations was 14% in the en- tire cohort, which is slightly lower than reported by the LLMPP group (18- 20%) [180, 226]. One potential explanation, in addition to differences in the

51 patient material, relates to the different sensitivities of the techniques applied in the two studies, as the LLMPP study used denaturing high performance liquid chromatography for mutation screening and subsequently direct se- quencing for confirmation [180]. MCL cases with blastoid vs. classic histol- ogy had comparable mutation rates in our study (13 vs. 18%, p = 0.6), alt- hough blastoid cases had significantly shorter survival (median 24 vs 39 months, P <0.01). We were interested in investigating the spectrum and characteristics of TP53 mutations in MCL. To this end we combined our dataset of TP53 mu- tations (n=18) with previously reported TP53 mutations in MCL (n=55) that had been accumulated in the IARC database. No significant differences be- tween the two cohorts with regards to mutation characteristics were identi- fied. The mutation pattern was also similar to what has been reported for TP53 mutations in cancer in general; missense mutations were most frequent (77%), the majority of mutations were transitions (62%), predominantly GC>AT transitions (55%) at CpG sites (31%). Therefore, no TP53 mutations or patterns specific to MCL were identified. In contrast, specific TP53 muta- tion patterns have been described in a few non-haematological malignancies, including lung and liver carcinomas, and are associated with exposure to certain carcinogens (tobacco and aflatoxin B1 respectively) [180]. Deletion at 17p was identified in 32% of cases and was therefore consid- erably more common than TP53 mutation. Of the 38 cases with 17p loss 24% also carried TP53 mutations, whereas 53% of cases with a non-silent TP53 mutation had a concomitant 17p deletion. These rates are almost iden- tical to those recently reported for the LLMPP MCL cohort [226]. This can be contrasted with CLL and myeloid malignancies (myelodysplastic syn- dromes and acute myeloid leukemia) where more than 70 percent of cases with a 17p loss also carry a TP53 mutation [223, 234, 235]. One of the most important results of this study was the striking impact that TP53 mutation had on survival, as the median overall survival was only 13 months for TP53 mutated cases compared to 43 months for cases without a mutation (p=0.006). This was in agreement with previous studies [180, 226]. In contrast, the presence or absence of 17p deletion did not influence survival in our cohort (median overall survival, 39 vs. 35 months respective- ly, p = 0.7). Previous studies have yielded mixed results with regards to the prognostic impact of 17p deletions in MCL, with a few studies indicating a relationship with poor prognosis [155, 158, 160], while others have not re- ported a correlation [98, 157, 159, 236]. Ours is the largest study to date including 119 MCL cases. Nevertheless, the different impact of TP53 mutation and 17p deletion on survival in MCL is interesting. These results are also in sharp contrast with the scenario in CLL, where any type of mono- or biallelic TP53 disruption (del(17p) only, mutation only, or both mutation and del(17p)) predicts poor clinical outcome [234, 237]. Possible explanations for the stronger prognos-

52 tic impact of TP53 mutations relative to 17p deletions in MCL include the dominant negative and/or gain-of-function effects that have been reported for certain TP53 mutations. The lack of prognostic impact of 17p deletions in MCL compared to CLL may be related to the high level of genomic insta- bility in the former. Only a few of the various CNA commonly identified in MCL have been associated with clinical outcome. Most CNA in MCL may in fact be secondary to general genomic instability while CNA in CLL are typically very few in number, and more likely to result from selective pres- sure. To summarize, we have confirmed that TP53 mutation is a marker of poor prognosis in MCL and studied the spectrum of TP53 mutations in this dis- ease. In contrast, the presence of 17p deletion was not associated with clini- cal outcome. However, the retrospective nature of this analysis requires that the implications of TP53 mutations and/or 17p deletions in MCL be further investigated in future studies.

Paper III: Genome-wide study of methylation profiles in MCL and CLL Global genomic methylation patterns in MCL have not been extensively studied. As a result it is not known whether DNA methylation patterns differ between MCL cases with good vs. poor prognostic markers. In CLL, howev- er, our group has recently demonstrated significant differences in DNA methylation pattern between IGHV mutated and unmutated subgroups [211]. Furthermore, comparing methylation profiles between MCL and CLL may be of interest since a number of similarities exist between the two diseases. For example the surface expression of CD5 is a defining immunophenotypic feature of both MCL and CLL. CD5 may play a role in ontogeny, immune regulation and maintenance of tolerance [238]. There are also similarities with regards to immunogenetics; as both can be divided into groups with or without IGHV SHM and in both diseases a subset of cases carries stereo- typed BCRs. Notably, the proportion of cases with mutated IGHV genes and/or stereotyped BCRs is lower in MCL. Another common feature of MCL and CLL is the frequent deletion at 13q14 [239]. DNA methylation profiling was performed for primary MCL (n=20) and CLL (n=30) specimens. Only two normal controls (peripheral blood mono- nuclear cells (PBMCs) and CD19+ B-cells from peripheral blood) were in- cluded in the analysis as the main focus was on comparing prognostic sub- groups within MCL and MCL vs. CLL. Furthermore, it is presently unclear from which exact B-cell compartment CLL and MCL develop. In order to compare methylation profiles between good vs. poor prognostic groups, MCL cases with the lowest and highest proliferation signature (n=10 for

53 each group) were selected to represent both ends of the clinical spectrum for MCL. For comparison, CLL cases including 15 IGHV mutated (stereotyped BCR subset #4) and 15 IGHV unmutated (stereotyped BCR subset #1) spec- imens were included. The two subsets represent the extremes in clinical out- come within CLL, with subset #4 having the best prognosis and subset #1 the worst [120]. DNA methylation was quantified for >27,000 CpGs using the Illumina Infinium BeadChip array. We found that the methylation profile of MCL was remarkably homoge- nous without formation of clusters in contrast to the more heterogeneous CLL collection. When stringent statistical criteria were applied to search for differentially methylated genes between cases with low versus high prolifer- ation signature only six genes were identified. This apparent homogeneity of MCL is in agreement with previous studies that have failed to define discrete prognostic/clinical subgroups based on parameters such as IGHV mutation status, genomic aberrations or gene expression profiling [53, 78, 150]. In contrast, DNA methylation profiling clearly separated MCL and CLL patient samples into discrete groups using unsupervised hierarchical clustering, in- dicating that the distribution of promoter DNA methylation differs consider- ably between these genomes. This is in agreement with a previous report comparing methylation profiles in small B-cell lymphoma subtypes using CpG island microarrays, where all the MCL cases clustered together unlike the CLL collection which separated into two distinct subgroups [207]. How- ever, their CpG island microarray had limited genomic resolution and the analysis was dependent on restriction enzyme recognition sequences. Fur- thermore, the functions of the differentially methylated genes were not ex- plored systematically. In order to further characterize the differences between MCL and CLL we applied stringent statistical testing that identified a total of 176 genes as dif- ferentially methylated between MCL and all CLL samples (subsets #1 and #4 combined). The GOTM online tool was used to search for enriched bio- logical processes and molecular functions among genes that were either methylated in CLL (n=90) or methylated in MCL (n=86). Among genes methylated in MCL there was a significant enrichment for transcription fac- tors/regulators involved in developmental biological processes. Specifically, nine of the transcription regulators belonged to the group of genes (DLX2, HLXB9, HOXA2, HOXA9, HOXA13, LHX1, PAX7, PITX3 and POU4F1). Furthermore, when analyzed on a global scale, the mean methyla- tion level of all homeobox genes on the array was higher for MCL than CLL (MI = 0.29 vs. 0.23, p = 0.002). Of note, these nine homeobox genes were all unmethylated in the normal PBMC and CD19+ B-cell controls included in the array. Despite the small number of normal control samples, these prelim- inary data indicate that these homeobox genes are not normally methylated in B-lymphocytes. Therefore, homeobox gene methylation in MCL may represent an aberrant evolution.

54 Homeobox genes represent a highly conserved family of transcription fac- tors that have crucial roles in development. Although methylation of homeo- box genes has previously been reported in various malignancies, including FL, this is the first study to describe homeobox gene methylation in MCL [225, 240-242]. The expression of homeobox genes during development is under the control of the polycomb repressor group (PcG) of genes [243]. Interestingly, several studies have reported that genes hypermethylated in cancers are enriched for PcG target genes [244-246]. This is true for many hematological malignancies as well, including MCL, but not CLL [190]. Therefore, the lack of homeobox gene methylation in CLL and the presence of methylation in MCL may reflect the differential methylation of PcG target genes in general in these two diseases. Although the mean methylation of PcG target genes was significantly higher in MCL than CLL in our study using a Student’s t-test, the difference was not significant using the nonpar- ametric Mann-Whitney U test (data not shown). When gene expression of four homeobox genes (HLXB9, HOXA13, PAX7 and LHX1) was measured with RQ-PCR, no expression was detected for CLL or MCL. Furthermore, expression could be re-induced (HOXA13 and HLXB9) in MCL cell lines but not consistently in CLL cells after treatment with methylation inhibitors. Therefore, homeobox gene expression in CLL may be silenced by mechanisms alternative to DNA methylation such as histone modifications. Notably, a recent study that identified prominent hy- permethylation of homeobox genes in FL compared to normal B-cells re- ported undetectable transcription of many of the hypermethylated develop- mental genes in both normal B-cells and FL in gene expression array analy- sis [225]. Therefore, methylation of homeobox genes in FL and MCL may reflect preferential DNA methylation of previously silenced genes. Further- more, as hinted above, repressive PcG marks may predispose DNA for aber- rant methylation in some tumors. We also used IPA to characterize the biological functions of differentially methylated genes. Among genes methylated in CLL (n=90) most (n=19) belonged to the molecular and cellular function category of cell death, whereas this category was not enriched in MCL. Admittedly, the apoptosis genes methylated in CLL comprised a heterogeneous group which is ex- pected as the assignment of genes to functional groups can be arbitrary. Therefore, although most are proapoptotic (e.g. CYFIP2, DYRK2, PRDM2) others have antiapoptotic (e.g. PAK6) functions. A few pro-apoptotic genes, such as CD80, were methylated in MCL when compared to CLL subset #4 [247]. Pyrosequencing of patient samples confirmed the differential methyla- tion of the CD80 gene, and subsequently CD80 expression as measured by RQ-PCR was significantly higher in CLL subset #4. Finally, several TSGs and oncogenes were identified among the differen- tially methylated genes. Among identified TSGs were SH3GL2 (methylated in MCL), SEPT9 and PRDM2 (both methylated in CLL). In support of our

55 findings both SEPT9 and PRDM2 were hypomethylated in MCL compared to normal B-cells in a recent study [190]. We were able to re-induce the ex- pression of the SH3GL2 gene after treatment of MCL cells with DNMT/HDAC inhibitors, suggesting a role for methylation in silencing of this gene. The proto-oncogene MERTK and the proliferation-associated gene CAMP were unmethylated in MCL. Both have been reported to be overex- pressed in MCL and various cancers [248-251] and we were able to confirm the high expression of CAMP in MCL with RQ-PCR. In conclusion, several observations were made in this genomic methyla- tion profiling study of MCL and CLL. The methylation pattern of MCL ap- peared homogenous and no correlation with the proliferation signature was detected. Specific functional classes of genes were preferentially methylated in either MCL or CLL, most notably homeobox genes which had higher degree of methylation in MCL compared to CLL. However, further studies are needed to determine the causes and potential significance of homeobox gene methylation in MCL. Finally, an extended comparison with other B- cell malignancies and a larger number of normal B-cell controls would allow us to draw more definite conclusions.

Paper IV: Genome-wide study of methylation profiles in stereotyped CLL subsets Recently our group compared the genomic methylation profiles of three CLL prognostic groups, IGHV mutated, unmutated and IGHV3-21 CLL, and iden- tified significant differences in promoter methylation pattern [211]. Accumu- lating data suggests, however, that classification of CLL according to stereo- typed BCRs can further subdivide CLL into prognostic subgroups independ- ent of IGHV mutation status [120, 123, 126]. Some stereotyped CLL subsets have been reported to display specific gene expression profiles [252, 253], but DNA methylation patterns have not been previously compared between stereotyped CLL subsets. The aim of this study was to analyze the global genomic methylation patterns for a unique collection of 39 CLL cases that belong to three different subsets, i.e. the poor-prognostic subsets #1 (unmu- tated IGHV1/5/7, IGKV1-39/1D-39) and #2 (IGHV3-21 CLL) and the favor- able prognostic subset #4 (mutated IGHV4-34, IGKV2-30). Results of genome-wide profiling using Illumina 27K methylation arrays revealed that each CLL subset displayed a unique methylation profile as many genes were differentially methylated between each two subsets. Im- portantly, only a partial overlap was observed between genes identified in our current subset study and our previous IGHV mutation status study [211]. Therefore, investigation of DNA methylation patterns in stereotyped subsets may give additional information to studying IGHV prognostic groups. As in

56 Paper III only two normal controls (PBMCs and CD19+ B-cells) were in- cluded in the analysis, since the primary aim was to compare the profiles of CLL subgroups. Using the GOTM tool significant enrichment of immune response genes, such as CD80, CD86, CD43, and IL10, was identified when comparing sub- set #1 to #4 or subset #1 to #2. Most were methylated in subset #1 but not in subsets #2 and #4. To investigate this in more detail, a separate GOTM anal- ysis was performed for genes methylated only in subset #1 or subset #4. In short, targets methylated in subset #1 were enriched for the same immune response genes (e.g. CD80, CD86 and IL10) as well as NF- B-inducing kinase activity (IKBKE, PPP4C, RIPK3), while the only enriched categories among targets methylated in subset #4 were cell adhesion (e.g. ITGAM, PECAM1, LAMA1). Again, only a limited overlap was observed with our previous methylation array study, as enriched categories among genes meth- ylated in different IGHV mutation groups were only partially the same as for corresponding stereotyped subsets. Among these overlapping genes were the adaptive immune response genes (CD80, CD86 and IL10) which were en- riched among methylated targets in both subset #1 and IGHV unmutated CLL. The differential methylation of two genes, CD80 and CD86, observed in the array data was validated by pyrosequencing of subset #1 and #4 patient samples. Subsequently the gene expression of CD80 and CD86 as measured by RQ-PCR was found to be higher in subset #4 than subset #1 cases. Final- ly, both CD80 and CD86 were re-expressed after treatment with DNMT/HDAC inhibitors in subset #1 CLL cells. These preliminary findings need to be verified as only two subset #1 cells were included in the re- expression experiment. In support of our results, CD86 was recently reported to be highly expressed and upregulated in subset #4 CLL cells after stimula- tion via toll-like receptors as compared to subset #1 [254, 255]. CD80 and CD86 are co-stimulatory molecules involved in B-cell signaling and are only expressed on activated B cells [256, 257]. In agreement with this both CD80 and CD86 were methylated in our normal controls, suggesting that the lack of methylation in CLL subset #4 may be aberrant. These data are prelimi- nary, however, due to the low number of normal controls included. Both CD80 and CD86 have been reported to be up-regulated on CLL B cells after activation via CD40 or pre-activated T-cells in vitro [258, 259]. In addition several genes with known functions in apoptosis were differ- entially methylated between CLL subsets. PRF1 was methylated in subset #4, which is in agreement with our previous work where PRF1 was less methylated and more highly expressed in IGHV unmutated cases compared to mutated cases [211]. The PRF1 gene encodes the perforin 1 protein and is not normally expressed in B cells but in cytotoxic T cells and natural killer cells [260]. In the current study, the expression of PRF1 was successfully re-

57 induced in subset #4 cells after treatment with DNMT/HDAC inhibitors, suggesting that promoter methylation regulates PRF1 expression. In summary, each of three stereotyped CLL subsets had a characteristic DNA methylation profile. Only a partial overlap was observed with our pre- vious methylation array study of IGHV mutated/unmutated/IGHV3-21 CLL, both with regards to differentially methylated genes and enriched gene on- tology categories. One of our main observations was the differential DNA methylation and mRNA expression of the CD80 and CD86 genes between subset #1 and subset #4 samples. Future work is needed to confirm the dif- ferential CD80/CD86 protein expression between CLL subsets and subse- quently to determine their impact on in vitro activation of T lymphocytes by CLL cells.

58 CONCLUDING REMARKS

In 1992, American and European ILSG members agreed on the definition of a new type of lymphoma derived from mantle B-cells that was consequently termed mantle cell lymphoma (MCL) [60]. In less than 20 years, significant advances have been made in the field of MCL research, both with regards to refining the diagnostic criteria and understanding the pathobiology of MCL. It has been firmly established that the hallmark of MCL is over-expression of the cyclin D1 protein, usually as a consequence of the chromosomal trans- location t(11;14)(q13;q32) [261, 262]. However, it is also well documented that this chromosomal event alone is not sufficient to result in lym- phomagenesis and that secondary genomic alterations are required for ma- lignant transformation [71, 72]. Indeed, MCL is characterized by a high load of CNAs compared to many other lymphoma subtypes [73]. Furthermore, aberrant DNA methylation may contribute to lymphomagenesis in MCL as has been suggested for other lymphoma types, such as DLBCL [203]. Much remains to be learned regarding the influence of individual genomic aberrations on the pathogenesis and clinical behavior of MCL. Despite using a relatively small dataset of MCL cases, in paper I, we were able to show that a high proliferation gene expression signature was associated with spe- cific CNAs and a higher load of aberrations of large size. Furthermore, other CNAs, including some that have previously been associated with aggressive variants of MCL, were associated with high genomic complexity (a high number of CNAs). Interestingly, despite the relatively large number of pub- lications on genomic aberrations in MCL, we were able to identify a small novel loss which targets a single gene, ZFP64, with a potential role in tumor- igenesis. The TP53 gene at 17p13 has been referred to as a “master guardian” of the cell, protecting its genome from damage by facilitating DNA repair or else apoptosis. No other gene is more frequently altered in human malignan- cies. TP53 mutations have previously been associated with poor prognosis in MCL, but the prognostic impact of 17p deletions has been somewhat contro- versial. In paper II, which is the largest study to date, we confirmed the asso- ciation of TP53 mutations with poor survival in MCL. Notably, 17p dele- tions were quite common but did not have any impact on survival. There- fore, a mutation of the TP53 gene seems to be more harmful than loss of one copy of the gene. Most mutations of the TP53 gene were missense mutations which have the potential to confer new properties to the mutated protein,

59 such as gain-of-function or dominant negative effects. On the other hand 17p deletions may, at least in some cases, merely be consequences of the general genetic instability characteristic of MCL. In recent years, increased focus has been directed towards epigenetic reg- ulation of gene expression in malignancies. Aberrant DNA methylation is an epigenetic phenomenon widely reported in tumors including B-cell malig- nancies, where the methylation pattern has been reported to be heterogene- ous [190]. In paper III, we decided to study the correlation of genomic me- thylation to the proliferation signature in MCL and to compare the methyla- tion profile of MCL and CLL. In contrast to our genomic aberration study (paper I), we did not find any association between DNA methylation and the proliferation signature. However, we were able to identify specific functional classes of genes that were preferentially methylated in either MCL or CLL. These findings may reflect distinct epigenetic silencing mechanisms in- volved in the pathogenesis of these B-cell malignancies. In paper IV, we further extended our studies of global DNA methylation to compare the profiles of three different CLL subsets expressing “stereo- typed” BCRs. Patients that belong to subsets #1 and #2 typically have a poor clinical outcome, while patients in subset #4 have a more indolent course. In short, each subset had a unique DNA methylation profile, and with regards to differentially methylated genes there was only a limited overlap with the previous study of methylation in IGHV prognostic groups. Gene ontology analysis revealed that genes involved in various aspects of the immune re- sponse were frequently methylated in subset #1 but not in subsets #2 or #4. Among these immune response genes were the co-stimulatory molecules CD80 and CD86 that are involved in B-cell signaling and activation through the BCR. Although the causes for subset-specific methylation of immune response genes are not known, potential explanations include differences in cellular origin, antigen exposure and micro-environmental interactions.

60 ACKNOWLEDGEMENTS

This work was carried out at the Department of Immunology, Genetics and Pathology at Uppsala University. Financial support was provided by the Nordic Cancer Union, the Swedish Cancer Society, the Swedish Medical Research Council and Lion’s Research Foundation in Uppsala.

First of all I would like to express my deepest gratitude to my supervisor Professor Richard Rosenquist for giving me the chance to pursue a doctoral degree while working as a physician at the Department of Clinical Immunology and Transfusion Medicine. Your expertise, encouragement and support have inspired me to do my best - you are a true role model!

I would also like to sincerely thank my co-supervisor Dr. Hans Ehrencrona. You stuck with me despite all the changes in your life, including the birth of your two babies, moving to Lund, starting a new job, and not to mention those microorganisms! Your help, high standards and patience have been invaluable.

Warm greetings to my co-authors: Special thanks to Birgitta Sander for providing us with the MCL clinical samples and cell lines, for collecting the clinical data and for good collaboration on the papers! Many thanks to Hanna Göransson Kultima for performing bioinformatic analysis on microarray data, and for always being so helpful. To Andreas Lundin for your productive and hard work on TP53 sequencing – good luck with your medical career! To Eva Kimby, Sakari Knuutila, Anna Laurell, Christer Sundström and Kostas Stamatopoulos for providing patient samples and clinical data. To Prasoon Agarwal, Tomas Axelsson, Anders Isaksson, Gun- nar Juliusson and Helena Jernberg-Wiklund for good collaboration.

To past and present members of the Molecular Hematology group; Arifin, Fiona, Ingrid, Ja, Larry, Lesley, Maria, Marie, Mattias, Meena, Millaray, Nikki and Rebeqa! I was very fortunate to become a part of such a good team, I have learned much from you and enjoyed your company. Thanks to each and everyone, especially for being so understanding and helpful with the lab work, and for tolerating my divided lab/hospital existence .

61 To the administrative staff at Rudbeck that makes sure everything runs smoothly; Birgitta Gustafsson, Britt-Marie Carlberg, Christina Andersson, Elisabeth Sandberg, Gunilla Åberg, Pirkko Boox and Ulla Steimer. To the various facilities that provided technical assistance and good service; the Rudbeck IT department, the Uppsala Genome Center, the Uppsala Array Platform and the SNP&SEQ Technology Platform.

To Dr. Richard Burack, my mentor at Washington University, St. Louis, Missouri for sparking my interest in lymphoma biology. Your enthusiasm, energy and knowledge laid the foundation for my future studies.

To the colleagues at Clinical Immunology and Transfusion Medicine (KITM): Thank you for giving me the chance to practice transfusion medicine at Akademiska while pursuing my doctoral degree. Special thanks to the nurses at the apheresis unit.

To my husband, Harri. You are the rock in my life, best friend and of course the father of our wonderful children, Halldór Alexander, Jökull Ari and Hugrún Eva. We have been through so much together, moving from Iceland to St. Louis, USA, later from USA to Sweden, and finally from Sweden back to Iceland. I cannot imagine there is a better husband around!

Special thanks to the staff at Uppsala International School (Kvarngärdesskolan) and Bulans preschool for taking such good care of our children. I could not pursue my work and studies if I didn’t know that the children were in good hands.

And last but not least, big thanks and warm greetings to my family and friends in Iceland. It feels good to be back!

62 REFERENCES

1. Ikuta, K., et al., Lymphocyte development from stem cells. Annu Rev Immunol, 1992. 10: p. 759-83. 2. Reya, T. and R. Grosschedl, Transcriptional regulation of B-cell differentiation. Curr Opin Immunol, 1998. 10(2): p. 158-65. 3. Henderson, A. and K. Calame, Transcriptional regulation during B cell development. Annu Rev Immunol, 1998. 16: p. 163-200. 4. Nutt, S.L., et al., Commitment to the B-lymphoid lineage depends on the transcription factor Pax5. Nature, 1999. 401(6753): p. 556-62. 5. LeBien, T.W. and T.F. Tedder, B lymphocytes: how they develop and function. Blood, 2008. 112(5): p. 1570-80. 6. Su, T.T. and D.J. Rawlings, Transitional B lymphocyte subsets operate as distinct checkpoints in murine splenic B cell development. J Immunol, 2002. 168(5): p. 2101-10. 7. Petro, J.B., et al., Transitional type 1 and 2 B lymphocyte subsets are differentially responsive to antigen receptor signaling. J Biol Chem, 2002. 277(50): p. 48009-19. 8. Hoffman, R., et al., eds. Hematology. Basic Principles and Practice. Fourth ed. 2005, Elsevier, Churchill Livingstone. 9. Schatz, D.G., M.A. Oettinger, and M.S. Schlissel, V(D)J recombination: molecular biology and regulation. Annu Rev Immunol, 1992. 10: p. 359-83. 10. Gellert, M., Recent advances in understanding V(D)J recombination. Adv Immunol, 1997. 64: p. 39-64. 11. Oettinger, M.A., et al., RAG-1 and RAG-2, adjacent genes that synergistically activate V(D)J recombination. Science, 1990. 248(4962): p. 1517-23. 12. Ma, Y., et al., Hairpin opening and overhang processing by an Artemis/DNA- dependent protein kinase complex in nonhomologous end joining and V(D)J recombination. Cell, 2002. 108(6): p. 781-94. 13. Komori, T., et al., Lack of N regions in antigen receptor variable region genes of TdT-deficient lymphocytes. Science, 1993. 261(5125): p. 1171-5. 14. Melchers, F., et al., The surrogate light chain in B-cell development. Immunol Today, 1993. 14(2): p. 60-8. 15. Kitamura, D. and K. Rajewsky, Targeted disruption of mu chain membrane exon causes loss of heavy-chain allelic exclusion. Nature, 1992. 356(6365): p. 154-6. 16. Ritchie, K.A., R.L. Brinster, and U. Storb, Allelic exclusion and control of endogenous immunoglobulin gene rearrangement in kappa transgenic mice. Nature, 1984. 312(5994): p. 517-20. 17. Wagner, S.D. and M.S. Neuberger, Somatic hypermutation of immunoglobulin genes. Annu Rev Immunol, 1996. 14: p. 441-57.

63 18. Liu, Y.J., et al., Germinal centres in T-cell-dependent antibody responses. Immunol Today, 1992. 13(1): p. 17-21. 19. Kosco, M.H. and D. Gray, Signals involved in germinal center reactions. Immunol Rev, 1992. 126: p. 63-76. 20. Yoshikawa, K., et al., AID enzyme-induced hypermutation in an actively transcribed gene in fibroblasts. Science, 2002. 296(5575): p. 2033-6. 21. Martin, A., et al., Activation-induced cytidine deaminase turns on somatic hypermutation in hybridomas. Nature, 2002. 415(6873): p. 802-6. 22. Swerdlow, S.H., et al., eds. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Fourth Edition ed. IARC WHO Classification of Tumours. 2008, World Health Organization. 23. Armitage, J.O. and D.D. Weisenburger, New approach to classifying non- Hodgkin's lymphomas: clinical features of the major histologic subtypes. Non- Hodgkin's Lymphoma Classification Project. J Clin Oncol, 1998. 16(8): p. 2780-95. 24. Stevenson, F., et al., Insight into the origin and clonal history of B-cell tumors as revealed by analysis of immunoglobulin variable region genes. Immunol Rev, 1998. 162: p. 247-59. 25. Ottensmeier, C.H., et al., Analysis of VH genes in follicular and diffuse lymphoma shows ongoing somatic mutation and multiple isotype transcripts in early disease with changes during disease progression. Blood, 1998. 91(11): p. 4292-9. 26. Kosmas, C., et al., Origin and diversification of the clonogenic cell in multiple myeloma: lessons from the immunoglobulin repertoire. Leukemia, 2000. 14(10): p. 1718-26. 27. Reya, T., et al., Stem cells, cancer, and cancer stem cells. Nature, 2001. 414(6859): p. 105-11. 28. Singh, S.K., et al., Identification of human brain tumour initiating cells. Nature, 2004. 432(7015): p. 396-401. 29. Bonnet, D. and J.E. Dick, Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med, 1997. 3(7): p. 730-7. 30. Ricci-Vitiani, L., et al., Identification and expansion of human colon-cancer- initiating cells. Nature, 2007. 445(7123): p. 111-5. 31. Lapidot, T., et al., A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature, 1994. 367(6464): p. 645-8. 32. Martinez-Climent, J.A., et al., Lymphoma stem cells: enough evidence to support their existence? Haematologica, 2010. 95(2): p. 293-302. 33. Chen, Z., et al., Prospective isolation of clonogenic mantle cell lymphoma- initiating cells. Stem Cell Res, 2010. 5(3): p. 212-25. 34. Kikushige, Y., et al., Self-renewing hematopoietic stem cell is the primary target in pathogenesis of human chronic lymphocytic leukemia. Cancer Cell, 2011. 20(2): p. 246-59. 35. Beral, V., et al., AIDS-associated non-Hodgkin lymphoma. Lancet, 1991. 337(8745): p. 805-9.

64 36. Canioni, D., et al., Lymphoproliferative disorders in children with primary immunodeficiencies: immunological status may be more predictive of the outcome than other criteria. Histopathology, 2001. 38(2): p. 146-59. 37. Nalesnik, M.A., et al., The pathology of posttransplant lymphoproliferative disorders occurring in the setting of cyclosporine A-prednisone immunosuppression. Am J Pathol, 1988. 133(1): p. 173-92. 38. Smedby, K.E., E. Baecklund, and J. Askling, Malignant lymphomas in autoimmunity and inflammation: a review of risks, risk factors, and lymphoma characteristics. Cancer Epidemiol Biomarkers Prev, 2006. 15(11): p. 2069-77. 39. Baecklund, E., et al., Association of chronic inflammation, not its treatment, with increased lymphoma risk in rheumatoid arthritis. Arthritis Rheum, 2006. 54(3): p. 692-701. 40. Prevot, S., et al., Analysis of African Burkitt's and high-grade B cell non- Burkitt's lymphoma for Epstein-Barr virus genomes using in situ hybridization. Br J Haematol, 1992. 80(1): p. 27-32. 41. Hamilton-Dutoit, S.J., et al., In situ demonstration of Epstein-Barr virus small RNAs (EBER 1) in acquired immunodeficiency syndrome-related lymphomas: correlation with tumor morphology and primary site. Blood, 1993. 82(2): p. 619-24. 42. Cesarman, E., et al., Kaposi's sarcoma-associated herpesvirus-like DNA sequences in AIDS-related body-cavity-based lymphomas. N Engl J Med, 1995. 332(18): p. 1186-91. 43. Pozzato, G., et al., Low-grade malignant lymphoma, hepatitis C virus infection, and mixed cryoglobulinemia. Blood, 1994. 84(9): p. 3047-53. 44. Hussell, T., et al., The response of cells from low-grade B-cell gastric lymphomas of mucosa-associated lymphoid tissue to Helicobacter pylori. Lancet, 1993. 342(8871): p. 571-4. 45. Wotherspoon, A.C., et al., Regression of primary low-grade B-cell of mucosa-associated lymphoid tissue type after eradication of Helicobacter pylori. Lancet, 1993. 342(8871): p. 575-7. 46. Wang, S.S. and A. Nieters, Unraveling the interactions between environmental factors and genetic polymorphisms in non-Hodgkin lymphoma risk. Expert Rev Anticancer Ther, 2010. 10(3): p. 403-13. 47. Rothman, N., et al., Genetic variation in TNF and IL10 and risk of non- Hodgkin lymphoma: a report from the InterLymph Consortium. Lancet Oncol, 2006. 7(1): p. 27-38. 48. Conde, L., et al., Genome-wide association study of follicular lymphoma identifies a risk locus at 6p21.32. Nat Genet, 2010. 42(8): p. 661-4. 49. Skibola, C.F., et al., Genetic variants at 6p21.33 are associated with susceptibility to follicular lymphoma. Nat Genet, 2009. 41(8): p. 873-5. 50. Di Bernardo, M.C., et al., A genome-wide association study identifies six susceptibility loci for chronic lymphocytic leukemia. Nat Genet, 2008. 40(10): p. 1204-10. 51. Crowther-Swanepoel, D., et al., Verification that common variation at 2q37.1, 6p25.3, 11q24.1, 15q23, and 19q13.32 influences chronic lymphocytic leukaemia risk. Br J Haematol, 2010. 150(4): p. 473-9.

65 52. Crowther-Swanepoel, D., et al., Common variants at 2q37.3, 8q24.21, 15q21.3 and 16q24.1 influence chronic lymphocytic leukemia risk. Nat Genet, 2010. 42(2): p. 132-6. 53. Jares, P. and E. Campo, Advances in the understanding of mantle cell lymphoma. Br J Haematol, 2008. 142(2): p. 149-65. 54. Herrmann, A., et al., Improvement of overall survival in advanced stage mantle cell lymphoma. J Clin Oncol, 2009. 27(4): p. 511-8. 55. Geisler, C.H., et al., Long-term progression-free survival of mantle cell lymphoma after intensive front-line immunochemotherapy with in vivo-purged stem cell rescue: a nonrandomized phase 2 multicenter study by the Nordic Lymphoma Group. Blood, 2008. 112(7): p. 2687-93. 56. Fernandez, V., et al., Genomic and gene expression profiling defines indolent forms of mantle cell lymphoma. Cancer Res, 2010. 70(4): p. 1408-18. 57. Geisler, C.H., et al., The Mantle Cell Lymphoma International Prognostic Index (MIPI) is superior to the International Prognostic Index (IPI) in predicting survival following intensive first-line immunochemotherapy and autologous stem cell transplantation (ASCT). Blood, 2010. 115(8): p. 1530-3. 58. Ondrejka, S.L., et al., Indolent mantle cell leukemia: clinicopathologic variant characterized by isolated lymphocytosis, interstitial bone marrow involvement, kappa light chain restriction, and good prognosis. Haematologica, 2011. 59. Lecluse, Y., et al., t(11;14)-positive clones can persist over a long period of time in the peripheral blood of healthy individuals. Leukemia, 2009. 23(6): p. 1190-3. 60. Banks, P.M., et al., Mantle cell lymphoma. A proposal for unification of morphologic, immunologic, and molecular data. Am J Surg Pathol, 1992. 16(7): p. 637-40. 61. Ek, S., et al., Nuclear expression of the non B-cell lineage Sox11 transcription factor identifies mantle cell lymphoma. Blood, 2008. 111(2): p. 800-5. 62. Mozos, A., et al., SOX11 expression is highly specific for mantle cell lymphoma and identifies the cyclin D1-negative subtype. Haematologica, 2009. 94(11): p. 1555-62. 63. Campo, E., M. Raffeld, and E.S. Jaffe, Mantle-cell lymphoma. Semin Hematol, 1999. 36(2): p. 115-27. 64. Welzel, N., et al., Templated nucleotide addition and immunoglobulin JH-gene utilization in t(11;14) junctions: implications for the mechanism of translocation and the origin of mantle cell lymphoma. Cancer Res, 2001. 61(4): p. 1629-36. 65. Bosch, F., et al., PRAD-1/cyclin D1 gene overexpression in chronic lymphoproliferative disorders: a highly specific marker of mantle cell lymphoma. Blood, 1994. 84(8): p. 2726-32. 66. de Boer, C.J., et al., Cyclin D1 protein analysis in the diagnosis of mantle cell lymphoma. Blood, 1995. 86(7): p. 2715-23. 67. Vaandrager, J.W., et al., Direct visualization of dispersed 11q13 chromosomal translocations in mantle cell lymphoma by multicolor DNA fiber fluorescence in situ hybridization. Blood, 1996. 88(4): p. 1177-82.

66 68. Rosenwald, A., et al., The proliferation gene expression signature is a quantitative integrator of oncogenic events that predicts survival in mantle cell lymphoma. Cancer Cell, 2003. 3(2): p. 185-97. 69. Fu, K., et al., Cyclin D1-negative mantle cell lymphoma: a clinicopathologic study based on gene expression profiling. Blood, 2005. 106(13): p. 4315-21. 70. Quintanilla-Martinez, L., et al., Differential diagnosis of cyclin D2+ mantle cell lymphoma based on fluorescence in situ hybridization and quantitative real-time-PCR. Haematologica, 2009. 94(11): p. 1595-8. 71. Gladden, A.B., et al., Expression of constitutively nuclear cyclin D1 in murine lymphocytes induces B-cell lymphoma. Oncogene, 2006. 25(7): p. 998-1007. 72. Lovec, H., et al., Cyclin D1/bcl-1 cooperates with myc genes in the generation of B-cell lymphoma in transgenic mice. EMBO J, 1994. 13(15): p. 3487-95. 73. Salaverria, I., et al., Mantle cell lymphoma: from pathology and molecular pathogenesis to new therapeutic perspectives. Haematologica, 2006. 91(1): p. 11-6. 74. Obrador-Hevia, A., et al., Molecular biology of mantle cell lymphoma: from profiling studies to new therapeutic strategies. Blood Rev, 2009. 23(5): p. 205- 16. 75. Fernandez, V., et al., Pathogenesis of mantle-cell lymphoma: all oncogenic roads lead to dysregulation of cell cycle and DNA damage response pathways. J Clin Oncol, 2005. 23(26): p. 6364-9. 76. Thorselius, M., et al., Somatic hypermutation and V(H) gene usage in mantle cell lymphoma. Eur J Haematol, 2002. 68(4): p. 217-24. 77. Walsh, S.H., et al., Mutated VH genes and preferential VH3-21 use define new subsets of mantle cell lymphoma. Blood, 2003. 101(10): p. 4047-54. 78. Kienle, D., et al., VH mutation status and VDJ rearrangement structure in mantle cell lymphoma: correlation with genomic aberrations, clinical characteristics, and outcome. Blood, 2003. 102(8): p. 3003-9. 79. Hadzidimitriou, A., et al., Is there a role for antigen selection in mantle cell lymphoma? Immunogenetic support from a series of 807 cases. Blood, 2011. Jul 26. [Epub ahead of print]. 80. Galderisi, U., F.P. Jori, and A. Giordano, Cell cycle regulation and neural differentiation. Oncogene, 2003. 22(33): p. 5208-19. 81. Harbour, J.W. and D.C. Dean, The Rb/E2F pathway: expanding roles and emerging paradigms. Genes Dev, 2000. 14(19): p. 2393-409. 82. Louie, D.C., et al., p53 overexpression as a marker of poor prognosis in mantle cell lymphomas with t(11;14)(q13;q32). Blood, 1995. 86(8): p. 2892-9. 83. Weinberg, R.A., The biology of cancer. 1st ed. 2007, New York: Garland Science, Taylor & Francis Group. 84. Pinyol, M., et al., p16(INK4a) gene inactivation by deletions, mutations, and hypermethylation is associated with transformed and aggressive variants of non-Hodgkin's lymphomas. Blood, 1998. 91(8): p. 2977-84. 85. Jares, P., D. Colomer, and E. Campo, Genetic and molecular pathogenesis of mantle cell lymphoma: perspectives for new targeted therapeutics. Nat Rev Cancer, 2007. 7(10): p. 750-62. 86. Schaffner, C., et al., Mantle cell lymphoma is characterized by inactivation of the ATM gene. Proc Natl Acad Sci U S A, 2000. 97(6): p. 2773-8.

67 87. Shiloh, Y., ATM and related protein kinases: safeguarding genome integrity. Nat Rev Cancer, 2003. 3(3): p. 155-68. 88. Fang, N.Y., et al., Oligonucleotide microarrays demonstrate the highest frequency of ATM mutations in the mantle cell subtype of lymphoma. Proc Natl Acad Sci U S A, 2003. 100(9): p. 5372-7. 89. Starczynski, J., et al., Variations in ATM protein expression during normal lymphoid differentiation and among B-cell-derived neoplasias. Am J Pathol, 2003. 163(2): p. 423-32. 90. Camacho, E., et al., ATM gene inactivation in mantle cell lymphoma mainly occurs by truncating mutations and missense mutations involving the phosphatidylinositol-3 kinase domain and is associated with increasing numbers of chromosomal imbalances. Blood, 2002. 99(1): p. 238-44. 91. Rummel, M.J., et al., Altered apoptosis pathways in mantle cell lymphoma. Leuk Lymphoma, 2004. 45(1): p. 49-54. 92. Tagawa, H., et al., Genome-wide array-based CGH for mantle cell lymphoma: identification of homozygous deletions of the proapoptotic gene BIM. Oncogene, 2005. 24(8): p. 1348-58. 93. Beltran, E., et al., A cyclin-D1 interaction with BAX underlies its oncogenic role and potential as a therapeutic target in mantle cell lymphoma. Proc Natl Acad Sci U S A, 2011. 94. Pham, L.V., et al., Inhibition of constitutive NF-kappa B activation in mantle cell lymphoma B cells leads to induction of cell cycle arrest and apoptosis. J Immunol, 2003. 171(1): p. 88-95. 95. Rudelius, M., et al., Constitutive activation of Akt contributes to the pathogenesis and survival of mantle cell lymphoma. Blood, 2006. 108(5): p. 1668-76. 96. Hoster, E., et al., A new prognostic index (MIPI) for patients with advanced- stage mantle cell lymphoma. Blood, 2008. 111(2): p. 558-65. 97. Sander, S., et al., Genomic aberrations in mantle cell lymphoma detected by interphase fluorescence in situ hybridization. Incidence and clinicopathological correlations. Haematologica, 2008. 93(5): p. 680-7. 98. Salaverria, I., et al., Specific secondary genetic alterations in mantle cell lymphoma provide prognostic information independent of the gene expression- based proliferation signature. J Clin Oncol, 2007. 25(10): p. 1216-22. 99. Hartmann, E., et al., Five-gene model to predict survival in mantle-cell lymphoma using frozen or formalin-fixed, paraffin-embedded tissue. J Clin Oncol, 2008. 26(30): p. 4966-72. 100. Forstpointner, R., et al., The addition of rituximab to a combination of fludarabine, cyclophosphamide, mitoxantrone (FCM) significantly increases the response rate and prolongs survival as compared with FCM alone in patients with relapsed and refractory follicular and mantle cell lymphomas: results of a prospective randomized study of the German Low-Grade Lymphoma Study Group. Blood, 2004. 104(10): p. 3064-71. 101. Lenz, G., et al., Immunochemotherapy with rituximab and cyclophosphamide, doxorubicin, vincristine, and prednisone significantly improves response and time to treatment failure, but not long-term outcome in patients with previously untreated mantle cell lymphoma: results of a prospective randomized trial of

68 the German Low Grade Lymphoma Study Group (GLSG). J Clin Oncol, 2005. 23(9): p. 1984-92. 102. Romaguera, J.E., et al., High rate of durable remissions after treatment of newly diagnosed aggressive mantle-cell lymphoma with rituximab plus hyper- CVAD alternating with rituximab plus high-dose methotrexate and cytarabine. J Clin Oncol, 2005. 23(28): p. 7013-23. 103. Schulz, H., et al., Immunochemotherapy with rituximab and overall survival in patients with indolent or mantle cell lymphoma: a systematic review and meta- analysis. J Natl Cancer Inst, 2007. 99(9): p. 706-14. 104. Gianni, A.M., et al., Long-term remission in mantle cell lymphoma following high-dose sequential chemotherapy and in vivo rituximab-purged stem cell autografting (R-HDS regimen). Blood, 2003. 102(2): p. 749-55. 105. Dreyling, M., et al., Early consolidation by myeloablative radiochemotherapy followed by autologous stem cell transplantation in first remission significantly prolongs progression-free survival in mantle-cell lymphoma: results of a prospective randomized trial of the European MCL Network. Blood, 2005. 105(7): p. 2677-84. 106. Andersen, N.S., et al., Pre-emptive treatment with rituximab of molecular relapse after autologous stem cell transplantation in mantle cell lymphoma. J Clin Oncol, 2009. 27(26): p. 4365-70. 107. Pott, C., et al., Molecular remission is an independent predictor of clinical outcome in patients with mantle cell lymphoma after combined immunochemotherapy: a European MCL intergroup study. Blood, 2010. 115(16): p. 3215-23. 108. Maris, M.B., et al., Allogeneic hematopoietic cell transplantation after fludarabine and 2 Gy for relapsed and refractory mantle cell lymphoma. Blood, 2004. 104(12): p. 3535-42. 109. Khouri, I.F., et al., Nonablative allogeneic stem-cell transplantation for advanced/recurrent mantle-cell lymphoma. J Clin Oncol, 2003. 21(23): p. 4407-12. 110. Kouroukis, C.T., et al., Flavopiridol in untreated or relapsed mantle-cell lymphoma: results of a phase II study of the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol, 2003. 21(9): p. 1740-5. 111. Goy, A., et al., Phase II study of proteasome inhibitor in relapsed or refractory B-cell non-Hodgkin's lymphoma. J Clin Oncol, 2005. 23(4): p. 667-75. 112. O'Connor, O.A., et al., Phase II clinical experience with the novel proteasome inhibitor bortezomib in patients with indolent non-Hodgkin's lymphoma and mantle cell lymphoma. J Clin Oncol, 2005. 23(4): p. 676-84. 113. Jemal, A., et al., Cancer statistics, 2010. CA Cancer J Clin, 2010. 60(5): p. 277-300. 114. Dighiero, G. and T.J. Hamblin, Chronic lymphocytic leukaemia. Lancet, 2008. 371(9617): p. 1017-29. 115. Hallek, M., et al., Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood, 2008. 111(12): p. 5446-56.

69 116. Hamblin, T.J., et al., Unmutated Ig V(H) genes are associated with a more aggressive form of chronic lymphocytic leukemia. Blood, 1999. 94(6): p. 1848- 54. 117. Damle, R.N., et al., Ig V gene mutation status and CD38 expression as novel prognostic indicators in chronic lymphocytic leukemia. Blood, 1999. 94(6): p. 1840-7. 118. Klein, U. and R. Dalla-Favera, New Insights into the Pathogenesis of Chronic Lymphocytic Leukemia. Semin Cancer Biol, 2010 Oct 25. [Epub ahead of print]. 119. Majid, A., et al., TP53 codon 72 polymorphism in patients with chronic lymphocytic leukaemia: identification of a subgroup with mutated IGHV genes and poor clinical outcome. Br J Haematol, 2011. 153(4): p. 533-5. 120. Stamatopoulos, K., et al., Over 20% of patients with chronic lymphocytic leukemia carry stereotyped receptors: Pathogenetic implications and clinical correlations. Blood, 2007. 109(1): p. 259-70. 121. Rosen, A., et al., Antigens in chronic lymphocytic leukemia-Implications for cell origin and leukemogenesis. Semin Cancer Biol, 2010. 20(6): p. 400-9. 122. Murray, F., et al., Stereotyped patterns of somatic hypermutation in subsets of patients with chronic lymphocytic leukemia: implications for the role of antigen selection in leukemogenesis. Blood, 2008. 111(3): p. 1524-33. 123. Bomben, R., et al., Molecular and clinical features of chronic lymphocytic leukaemia with stereotyped B cell receptors: results from an Italian multicentre study. Br J Haematol, 2009. 144(4): p. 492-506. 124. Thorselius, M., et al., Strikingly homologous immunoglobulin gene rearrangements and poor outcome in VH3-21-using chronic lymphocytic leukemia patients independent of geographic origin and mutational status. Blood, 2006. 107(7): p. 2889-94. 125. Tobin, G., et al., Chronic lymphocytic leukemias utilizing the VH3-21 gene display highly restricted Vlambda2-14 gene use and homologous CDR3s: implicating recognition of a common antigen epitope. Blood, 2003. 101(12): p. 4952-7. 126. Ghia, P., et al., Geographic patterns and pathogenetic implications of IGHV gene usage in chronic lymphocytic leukemia: the lesson of the IGHV3-21 gene. Blood, 2005. 105(4): p. 1678-85. 127. Cuttner, J., Increased incidence of hematologic malignancies in first-degree relatives of patients with chronic lymphocytic leukemia. Cancer Invest, 1992. 10(2): p. 103-9. 128. Dohner, H., et al., Chromosome aberrations in B-cell chronic lymphocytic leukemia: reassessment based on molecular cytogenetic analysis. J Mol Med, 1999. 77(2): p. 266-81. 129. Palamarchuk, A., et al., 13q14 deletions in CLL involve cooperating tumor suppressors. Blood, 2010. 115(19): p. 3916-22. 130. Cloonan, N., et al., The miR-17-5p microRNA is a key regulator of the G1/S phase cell cycle transition. Genome Biol, 2008. 9(8): p. R127. 131. Calin, G.A., et al., Frequent deletions and down-regulation of micro- RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proc Natl Acad Sci U S A, 2002. 99(24): p. 15524-9.

70 132. Buhl, A.M., et al., Identification of a gene on chromosome 12q22 uniquely overexpressed in chronic lymphocytic leukemia. Blood, 2006. 107(7): p. 2904- 11. 133. Caligaris-Cappio, F. and T.J. Hamblin, B-cell chronic lymphocytic leukemia: a bird of a different feather. J Clin Oncol, 1999. 17(1): p. 399-408. 134. Messmer, B.T., et al., In vivo measurements document the dynamic cellular kinetics of chronic lymphocytic leukemia B cells. J Clin Invest, 2005. 115(3): p. 755-64. 135. Kostareli, E., et al., Molecular evidence for EBV and CMV persistence in a subset of patients with chronic lymphocytic leukemia expressing stereotyped IGHV4-34 B-cell receptors. Leukemia, 2009. 23(5): p. 919-24. 136. Lanemo Myhrinder, A., et al., A new perspective: molecular motifs on oxidized LDL, apoptotic cells, and bacteria are targets for chronic lymphocytic leukemia antibodies. Blood, 2008. 111(7): p. 3838-48. 137. Rai, K.R., et al., Clinical staging of chronic lymphocytic leukemia. Blood, 1975. 46(2): p. 219-34. 138. Binet, J.L., et al., A new prognostic classification of chronic lymphocytic leukemia derived from a multivariate survival analysis. Cancer, 1981. 48(1): p. 198-206. 139. Stamatopoulos, K., P. Ghia, and R. Rosenquist, eds. New Biological Prognostic Markers in Chronic Lymphocytic Leukemia. 2010, Wolters Kluwer/Lippincott, Williams & Wilkins: Milan. 140. Zenz, T., et al., TP53 Mutation and Survival in Chronic Lymphocytic Leukemia. J Clin Oncol, 2010. 28(29): p. 4473-9. 141. Dighiero, G., et al., Chlorambucil in indolent chronic lymphocytic leukemia. French Cooperative Group on Chronic Lymphocytic Leukemia. N Engl J Med, 1998. 338(21): p. 1506-14. 142. Chemotherapeutic options in chronic lymphocytic leukemia: a meta-analysis of the randomized trials. CLL Trialists' Collaborative Group. J Natl Cancer Inst, 1999. 91(10): p. 861-8. 143. Foon, K.A. and M.J. Hallek, Changing paradigms in the treatment of chronic lymphocytic leukemia. Leukemia, 2010. 24(3): p. 500-11. 144. Dreger, P., et al., Indications for allogeneic stem cell transplantation in chronic lymphocytic leukemia: the EBMT transplant consensus. Leukemia, 2007. 21(1): p. 12-7. 145. Schulze, A. and J. Downward, Navigating gene expression using microarrays- -a technology review. Nat Cell Biol, 2001. 3(8): p. E190-5. 146. Rizzatti, E.G., et al., Gene expression profiling of mantle cell lymphoma cells reveals aberrant expression of genes from the PI3K-AKT, WNT and TGFbeta signalling pathways. Br J Haematol, 2005. 130(4): p. 516-26. 147. Ek, S., et al., From gene expression analysis to tissue microarrays: a rational approach to identify therapeutic and diagnostic targets in lymphoid malignancies. Mol Cell Proteomics, 2006. 5(6): p. 1072-81. 148. Ortega-Paino, E., et al., Functionally associated targets in mantle cell lymphoma as defined by DNA microarrays and RNA interference. Blood, 2008. 111(3): p. 1617-24.

71 149. Rinaldi, A., et al., Genomic and expression profiling identifies the B-cell associated tyrosine kinase Syk as a possible therapeutic target in mantle cell lymphoma. Br J Haematol, 2006. 132(3): p. 303-16. 150. Martinez, N., et al., The molecular signature of mantle cell lymphoma reveals multiple signals favoring cell survival. Cancer Res, 2003. 63(23): p. 8226-32. 151. Kallioniemi, A., et al., Comparative genomic hybridization for molecular cytogenetic analysis of solid tumors. Science, 1992. 258(5083): p. 818-21. 152. Bejjani, B.A. and L.G. Shaffer, Application of array-based comparative genomic hybridization to clinical diagnostics. J Mol Diagn, 2006. 8(5): p. 528- 33. 153. Hinds, D.A., et al., Whole-genome patterns of common DNA variation in three human populations. Science, 2005. 307(5712): p. 1072-9. 154. Dutt, A. and R. Beroukhim, Single nucleotide polymorphism array analysis of cancer. Curr Opin Oncol, 2007. 19(1): p. 43-9. 155. Allen, J.E., et al., Identification of novel regions of amplification and deletion within mantle cell lymphoma DNA by comparative genomic hybridization. Br J Haematol, 2002. 116(2): p. 291-8. 156. Bentz, M., et al., t(11;14)-positive mantle cell lymphomas exhibit complex karyotypes and share similarities with B-cell chronic lymphocytic leukemia. Genes Chromosomes Cancer, 2000. 27(3): p. 285-94. 157. Bea, S., et al., Increased number of chromosomal imbalances and high-level DNA amplifications in mantle cell lymphoma are associated with blastoid variants. Blood, 1999. 93(12): p. 4365-74. 158. Flordal Thelander, E., et al., Detailed assessment of copy number alterations revealing homozygous deletions in 1p and 13q in mantle cell lymphoma. Leuk Res, 2007. 31(9): p. 1219-30. 159. Kohlhammer, H., et al., Genomic DNA-chip hybridization in t(11;14)-positive mantle cell lymphomas shows a high frequency of aberrations and allows a refined characterization of consensus regions. Blood, 2004. 104(3): p. 795- 801. 160. Rubio-Moscardo, F., et al., Mantle-cell lymphoma genotypes identified with CGH to BAC microarrays define a leukemic subgroup of disease and predict patient outcome. Blood, 2005. 105(11): p. 4445-54. 161. Schraders, M., et al., Novel chromosomal imbalances in mantle cell lymphoma detected by genome-wide array-based comparative genomic hybridization. Blood, 2005. 105(4): p. 1686-93. 162. Bea, S., et al., Uniparental disomies, homozygous deletions, amplifications, and target genes in mantle cell lymphoma revealed by integrative high- resolution whole-genome profiling. Blood, 2009. 113(13): p. 3059-69. 163. Kawamata, N., et al., Identified hidden genomic changes in mantle cell lymphoma using high-resolution single nucleotide polymorphism genomic array. Exp Hematol, 2009. 37(8): p. 937-46. 164. Vater, I., et al., GeneChip analyses point to novel pathogenetic mechanisms in mantle cell lymphoma. Br J Haematol, 2009. 144(3): p. 317-31. 165. Pinyol, M., et al., Deletions and loss of expression of p16INK4a and p21Waf1 genes are associated with aggressive variants of mantle cell lymphomas. Blood, 1997. 89(1): p. 272-80.

72 166. Dreyling, M.H., et al., Alterations of the cyclin D1/p16-pRB pathway in mantle cell lymphoma. Cancer Res, 1997. 57(20): p. 4608-14. 167. Isobe, M., et al., Localization of gene for human p53 tumour antigen to band 17p13. Nature, 1986. 320(6057): p. 84-5. 168. May, P. and E. May, Twenty years of p53 research: structural and functional aspects of the p53 protein. Oncogene, 1999. 18(53): p. 7621-36. 169. Joerger, A.C. and A.R. Fersht, Structural biology of the tumor suppressor p53. Annu Rev Biochem, 2008. 77: p. 557-82. 170. Wu, X., et al., The p53-mdm-2 autoregulatory feedback loop. Genes Dev, 1993. 7(7A): p. 1126-32. 171. Levine, A.J. and M. Oren, The first 30 years of p53: growing ever more complex. Nat Rev Cancer, 2009. 9(10): p. 749-58. 172. Toledo, F. and G.M. Wahl, Regulating the p53 pathway: in vitro hypotheses, in vivo veritas. Nat Rev Cancer, 2006. 6(12): p. 909-23. 173. Farmer, G., et al., Wild-type p53 activates transcription in vitro. Nature, 1992. 358(6381): p. 83-6. 174. Pietenpol, J.A., et al., Sequence-specific transcriptional activation is essential for growth suppression by p53. Proc Natl Acad Sci U S A, 1994. 91(6): p. 1998-2002. 175. Miyashita, T. and J.C. Reed, Tumor suppressor p53 is a direct transcriptional activator of the human bax gene. Cell, 1995. 80(2): p. 293-9. 176. el-Deiry, W.S., et al., WAF1, a potential mediator of p53 tumor suppression. Cell, 1993. 75(4): p. 817-25. 177. Vaseva, A.V. and U.M. Moll, The mitochondrial p53 pathway. Biochim Biophys Acta, 2009. 1787(5): p. 414-20. 178. Olivier, M., M. Hollstein, and P. Hainaut, TP53 mutations in human cancers: origins, consequences, and clinical use. Cold Spring Harb Perspect Biol, 2010. 2(1): p. a001008. 179. Cheung, K.J., D.E. Horsman, and R.D. Gascoyne, The significance of TP53 in lymphoid malignancies: mutation prevalence, regulation, prognostic impact and potential as a therapeutic target. Br J Haematol, 2009. 146(3): p. 257-69. 180. Greiner, T.C., et al., Mutation and genomic deletion status of ataxia telangiectasia mutated (ATM) and p53 confer specific gene expression profiles in mantle cell lymphoma. Proc Natl Acad Sci U S A, 2006. 103(7): p. 2352-7. 181. Brosh, R. and V. Rotter, When mutants gain new powers: news from the mutant p53 field. Nat Rev Cancer, 2009. 9(10): p. 701-13. 182. Robertson, K.D., DNA methylation and human disease. Nat Rev Genet, 2005. 6(8): p. 597-610. 183. Rodriguez-Paredes, M. and M. Esteller, Cancer epigenetics reaches mainstream oncology. Nat Med, 2011. 17(3): p. 330-9. 184. Bird, A., DNA methylation patterns and epigenetic memory. Genes Dev, 2002. 16(1): p. 6-21. 185. Strachan, T. and A.P. Read, Human molecular genetics 3. 3rd ed. 2004, London and New York: Garland Science, Taylor & Francis Group. 186. Fuks, F., DNA methylation and histone modifications: teaming up to silence genes. Curr Opin Genet Dev, 2005. 15(5): p. 490-5.

73 187. Bachman, K.E., et al., Histone modifications and silencing prior to DNA methylation of a tumor suppressor gene. Cancer Cell, 2003. 3(1): p. 89-95. 188. Esteller, M., Epigenetics in cancer. N Engl J Med, 2008. 358(11): p. 1148-59. 189. Jones, P.A. and S.B. Baylin, The epigenomics of cancer. Cell, 2007. 128(4): p. 683-92. 190. Martin-Subero, J.I., et al., A comprehensive microarray-based DNA methylation study of 367 hematological neoplasms. PLoS One, 2009. 4(9): p. e6986. 191. Doi, A., et al., Differential methylation of tissue- and cancer-specific CpG island shores distinguishes human induced pluripotent stem cells, embryonic stem cells and fibroblasts. Nat Genet, 2009. 41(12): p. 1350-3. 192. Irizarry, R.A., et al., The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nat Genet, 2009. 41(2): p. 178-86. 193. Rozhon, W., et al., Rapid quantification of global DNA methylation by isocratic cation exchange high-performance liquid chromatography. Anal Biochem, 2008. 375(2): p. 354-60. 194. Ammerpohl, O., et al., Hunting for the 5th base: Techniques for analyzing DNA methylation. Biochim Biophys Acta, 2009. 1790(9): p. 847-62. 195. Lippman, Z., et al., Profiling DNA methylation patterns using genomic tiling microarrays. Nat Methods, 2005. 2(3): p. 219-24. 196. Irizarry, R.A., et al., Comprehensive high-throughput arrays for relative methylation (CHARM). Genome Res, 2008. 18(5): p. 780-90. 197. Weber, M., et al., Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normal and transformed human cells. Nat Genet, 2005. 37(8): p. 853-62. 198. Ballestar, E., et al., Methyl-CpG binding proteins identify novel sites of epigenetic inactivation in human cancer. EMBO J, 2003. 22(23): p. 6335-45. 199. Huang, Y.W., T.H. Huang, and L.S. Wang, Profiling DNA methylomes from microarray to genome-scale sequencing. Technol Cancer Res Treat, 2010. 9(2): p. 139-47. 200. Fernandez, A.F., et al., A DNA methylation fingerprint of 1628 human samples. Genome Res, 2011. 201. Li, Y., et al., The DNA methylome of human peripheral blood mononuclear cells. PLoS Biol, 2010. 8(11): p. e1000533. 202. Rajendram, R., et al., Assessment of methylation level prediction accuracy in Methyl-DNA immunoprecipitation and sodium bisulfite based microarray platforms. Epigenetics, 2011. 6(4). 203. Martin-Subero, J.I., et al., New insights into the biology and origin of mature aggressive B-cell lymphomas by combined epigenomic, genomic, and transcriptional profiling. Blood, 2009. 113(11): p. 2488-97. 204. Richter, J., et al., Array-based DNA methylation profiling of primary lymphomas of the central nervous system. BMC Cancer, 2009. 9: p. 455. 205. Harris, R.A., et al., Comparison of sequencing-based methods to profile DNA methylation and identification of monoallelic epigenetic modifications. Nat Biotechnol, 2010. 28(10): p. 1097-105.

74 206. Guo, J., et al., Differential DNA methylation of gene promoters in small B-cell lymphomas. Am J Clin Pathol, 2005. 124(3): p. 430-9. 207. Rahmatpanah, F.B., et al., Differential DNA methylation patterns of small B- cell lymphoma subclasses with different clinical behavior. Leukemia, 2006. 20(10): p. 1855-62. 208. Taylor, K.H., et al., Ultradeep bisulfite sequencing analysis of DNA methylation patterns in multiple gene promoters by 454 sequencing. Cancer Res, 2007. 67(18): p. 8511-8. 209. Leshchenko, V.V., et al., Genome wide DNA methylation analysis reveals novel targets for drug development in mantle cell lymphoma. Blood, 2010. 210. Enjuanes, A., et al., Identification of methylated genes associated with aggressive clinicopathological features in mantle cell lymphoma. PLoS One, 2011. 6(5): p. e19736. 211. Kanduri, M., et al., Differential genome-wide array-based methylation profiles in prognostic subsets of chronic lymphocytic leukemia. Blood, 2010. 115(2): p. 296-305. 212. Corcoran, M., et al., ZAP-70 methylation status is associated with ZAP-70 expression status in chronic lymphocytic leukemia. Haematologica, 2005. 90(8): p. 1078-88. 213. Raval, A., et al., TWIST2 demonstrates differential methylation in immunoglobulin variable heavy chain mutated and unmutated chronic lymphocytic leukemia. J Clin Oncol, 2005. 23(17): p. 3877-85. 214. Strathdee, G., et al., Promoter hypermethylation silences expression of the HoxA4 gene and correlates with IgVh mutational status in CLL. Leukemia, 2006. 20(7): p. 1326-9. 215. Raval, A., et al., Downregulation of death-associated protein kinase 1 (DAPK1) in chronic lymphocytic leukemia. Cell, 2007. 129(5): p. 879-90. 216. Olshen, A.B., et al., Circular binary segmentation for the analysis of array- based DNA copy number data. Biostatistics, 2004. 5(4): p. 557-72. 217. Gunnarsson, R., et al., Screening for copy-number alterations and loss of heterozygosity in chronic lymphocytic leukemia--a comparative study of four differently designed, high resolution microarray platforms. Genes Chromosomes Cancer, 2008. 47(8): p. 697-711. 218. Gunnarsson, R., et al., Large but not small copy-number alterations correlate to high-risk genomic aberrations and survival in chronic lymphocytic leukemia: a high-resolution genomic screening of newly diagnosed patients. Leukemia, 2010. 24(1): p. 211-5. 219. Goransson, H., et al., Quantification of normal cell fraction and copy number neutral LOH in clinical lung cancer samples using SNP array data. PLoS One, 2009. 4(6): p. e6057. 220. Greiner, T.C., Enhanced detection of TP53 mutations using a GC-clamp in denaturing high performance liquid chromatography. Diagn Mol Pathol, 2007. 16(1): p. 32-7. 221. Chiaretti, S., et al., Evaluation of TP53 mutations with the AmpliChip p53 research test in chronic lymphocytic leukemia: Correlation with clinical outcome and gene expression profiling. Genes Chromosomes Cancer, 2011.

75 222. Krypuy, M., et al., High resolution melting for mutation scanning of TP53 exons 5-8. BMC Cancer, 2007. 7: p. 168. 223. Zainuddin, N., et al., TP53 Mutations are infrequent in newly diagnosed chronic lymphocytic leukemia. Leuk Res, 2011. 35(2): p. 272-4. 224. Sturn, A., J. Quackenbush, and Z. Trajanoski, Genesis: cluster analysis of microarray data. Bioinformatics, 2002. 18(1): p. 207-8. 225. Bennett, L.B., et al., DNA hypermethylation accompanied by transcriptional repression in follicular lymphoma. Genes Chromosomes Cancer, 2009. 48(9): p. 828-41. 226. Hartmann, E.M., et al., Pathway discovery in mantle cell lymphoma by integrated analysis of high-resolution gene expression and copy number profiling. Blood, 2010. 116(6): p. 953-61. 227. Sugai, T., et al., Molecular analysis of gastric differentiated-type intramucosal and submucosal cancers. Int J Cancer, 2010. Epub Feb 22. 228. Habano, W., et al., Novel approach for detecting global epigenetic alterations associated with tumor cell . Int J Cancer, 2007. 121(7): p. 1487-93. 229. Argatoff, L.H., et al., Mantle cell lymphoma: a clinicopathologic study of 80 cases. Blood, 1997. 89(6): p. 2067-78. 230. Bosch, F., et al., Mantle cell lymphoma: presenting features, response to therapy, and prognostic factors. Cancer, 1998. 82(3): p. 567-75. 231. Raty, R., et al., Ki-67 expression level, histological subtype, and the International Prognostic Index as outcome predictors in mantle cell lymphoma. Eur J Haematol, 2002. 69(1): p. 11-20. 232. Hutter, G., et al., Allelic genotyping reveals a hierarchy of genomic alterations in mantle cell lymphoma associated to cell proliferation. Ann Hematol, 2009. 88(9): p. 821-8. 233. Espinet, B., et al., Incidence and prognostic impact of secondary cytogenetic aberrations in a series of 145 patients with mantle cell lymphoma. Genes Chromosomes Cancer, 2010. 49(5): p. 439-51. 234. Zenz, T., et al., Monoallelic TP53 inactivation is associated with poor prognosis in chronic lymphocytic leukemia: results from a detailed genetic characterization with long-term follow-up. Blood, 2008. 112(8): p. 3322-9. 235. Jasek, M., et al., TP53 mutations in myeloid malignancies are either homozygous or hemizygous due to copy number-neutral loss of heterozygosity or deletion of 17p. Leukemia, 2010. 24(1): p. 216-9. 236. Halldorsdottir, A.M., et al., High-resolution genomic screening in mantle cell lymphoma--specific changes correlate with genomic complexity, the proliferation signature and survival. Genes Chromosomes Cancer, 2010. 50(2): p. 113-21. 237. Rossi, D., et al., The prognostic value of TP53 mutations in chronic lymphocytic leukemia is independent of Del17p13: implications for overall survival and chemorefractoriness. Clin Cancer Res, 2009. 15(3): p. 995-1004. 238. Hardy, R.R. and K. Hayakawa, B cell development pathways. Annu Rev Immunol, 2001. 19: p. 595-621. 239. Stilgenbauer, S., et al., Expressed sequences as candidates for a novel tumor suppressor gene at band 13q14 in B-cell chronic lymphocytic leukemia and mantle cell lymphoma. Oncogene, 1998. 16(14): p. 1891-7.

76 240. Lind, G.E., et al., Novel epigenetically deregulated genes in include homeobox genes and SCGB3A1 (HIN-1). J Pathol, 2006. 210(4): p. 441-9. 241. Rauch, T., et al., Homeobox gene methylation in lung cancer studied by genome-wide analysis with a microarray-based methylated CpG island recovery assay. Proc Natl Acad Sci U S A, 2007. 104(13): p. 5527-32. 242. Rodriguez, B.A., et al., Epigenetic repression of the estrogen-regulated Homeobox B13 gene in breast cancer. Carcinogenesis, 2008. 29(7): p. 1459- 65. 243. Ringrose, L. and R. Paro, Epigenetic regulation of cellular memory by the Polycomb and Trithorax group proteins. Annu Rev Genet, 2004. 38: p. 413- 43. 244. Ohm, J.E., et al., A stem cell-like chromatin pattern may predispose tumor suppressor genes to DNA hypermethylation and heritable silencing. Nat Genet, 2007. 39(2): p. 237-42. 245. Schlesinger, Y., et al., Polycomb-mediated methylation on Lys27 of histone H3 pre-marks genes for de novo methylation in cancer. Nat Genet, 2007. 39(2): p. 232-6. 246. Widschwendter, M., et al., Epigenetic stem cell signature in cancer. Nat Genet, 2007. 39(2): p. 157-8. 247. Suvas, S., et al., Distinct role of CD80 and CD86 in the regulation of the activation of B cell and B cell lymphoma. J Biol Chem, 2002. 277(10): p. 7766-75. 248. Islam, T.C., et al., High level of cannabinoid receptor 1, absence of regulator of G protein signalling 13 and differential expression of Cyclin D1 in mantle cell lymphoma. Leukemia, 2003. 17(9): p. 1880-90. 249. von Haussen, J., et al., The host defence peptide LL-37/hCAP-18 is a growth factor for lung cancer cells. Lung Cancer, 2008. 59(1): p. 12-23. 250. Heilborn, J.D., et al., Antimicrobial protein hCAP18/LL-37 is highly expressed in breast cancer and is a putative growth factor for epithelial cells. Int J Cancer, 2005. 114(5): p. 713-9. 251. Coffelt, S.B., et al., Ovarian cancers overexpress the antimicrobial protein hCAP-18 and its derivative LL-37 increases ovarian cancer cell proliferation and invasion. Int J Cancer, 2008. 122(5): p. 1030-9. 252. Marincevic, M., et al., Distinct gene expression profiles in subsets of chronic lymphocytic leukemia expressing stereotyped IGHV4-34 B-cell receptors. Haematologica. 95(12): p. 2072-9. 253. Falt, S., et al., Distinctive gene expression pattern in VH3-21 utilizing B-cell chronic lymphocytic leukemia. Blood, 2005. 106(2): p. 681-9. 254. Arvaniti, E., et al., Toll-Like Receptor Signaling Pathway In Chronic Lymphocytic Leukemia: Distinct Gene Expression Profiles of Potential Pathogenetic Significance In Specific Subsets of Patients, in ASH. 2010, Blood: Orlando, Florida, USA. 255. Ntoufa, S., et al., Differential Functional Outcomes After Stimulation Via Innate Immunity Receptors In Chronic Lymphocytic Leukemia Subtypes Defined by the Molecular Features of the Immunoglobulin Receptor, in ASH. 2010, Blood: Orlando, Florida, USA.

77 256. Chaperot, L., et al., Functional expression of CD80 and CD86 allows immunogenicity of malignant B cells from non-Hodgkin's lymphomas. Exp Hematol, 1999. 27(3): p. 479-88. 257. Vasu, C., et al., CD80 and CD86 C domains play an important role in receptor binding and co-stimulatory properties. Int Immunol, 2003. 15(2): p. 167-75. 258. Van den Hove, L.E., et al., CD40 triggering of chronic lymphocytic leukemia B cells results in efficient alloantigen presentation and cytotoxic T lymphocyte induction by up-regulation of CD80 and CD86 costimulatory molecules. Leukemia, 1997. 11(4): p. 572-80. 259. Romano, C., et al., Effects of preactivated autologous T lymphocytes on CD80, CD86 and CD95 expression by chronic lymphocytic leukemia B cells. Leuk Lymphoma, 2003. 44(11): p. 1963-71. 260. Trapani, J.A. and I. Voskoboinik, The complex issue of regulating perforin expression. Trends in Immunology, 2007. 28(6): p. 243-245. 261. Raffeld, M. and E.S. Jaffe, bcl-1, t(11;14), and mantle cell-derived lymphomas. Blood, 1991. 78(2): p. 259-63. 262. Espinet, B., et al., Translocation (11;14)(q13;q32) and preferential involvement of chromosomes 1, 2, 9, 13, and 17 in mantle cell lymphoma. Cancer Genet Cytogenet, 1999. 111(1): p. 92-8.

78