Molecular Pathogenesis of Diffuse Large B-cell Lymphoma

Introduction

Diffuse Large B-Cell Lymphoma (DLBCL) is the most common type of Non-Hodgkin-Lymphoma (NHL), accounting for about 40% of NHL cases ¹. DLBCL is a heterogeneous disease with a highly variable clinical course, currently treated with combinations of immuno- and chemotherapy. Based on gene expression profile analysis, this single diagnostic category can be classified into distinct phenotypic subtypes, differing in molecular and clinical features and reflecting the origin from specific stages of B cell differentiation during the germinal center reaction ². During the past decade, multiple recurrent genetic alterations associated with DLBCL have been identified. This review will provide a brief summary of the germinal center reaction as a basis to understand the biological heterogeneity of DLBCL, and then focus on individual genetic lesions contributing to the pathogenesis of this disease.

Most DLBCLs derive from Germinal Center B-cells

The germinal center (GC) is the site where B-cells undergo distinct genetic processes to generate high-affinity antibodies (Fig 1). GCs are formed by proliferating B-cells in secondary lymphoid tissues upon T-cell dependent antigen stimulation. Within the dark zone of the GC, which consists of highly proliferating centroblasts (CBs), the cells undergo somatic hypermutation (SHM) of the variable region of the immunoglobulin gene (IgV) ^3-4. SHM produces primarily single nucleotide substitutions, but also deletions and duplications in the IgV heavy and light chain genes, resulting in the production of antibodies with high affinity for the antigen ^3-5. SHM can also target a number of non-immunoglobulin genes in normal B-cells, for example the 5′ untranslated region of B-cell lymphoma 6 (BCL-6) ^6-8. SHM occurs via DNA strand breaks and requires activation-induced cytidine deaminase (AID), which initiates the process by converting deoxycytidines to uracils, which are then further processed by DNA repair enzymes, leading to the creation of abasic sites and error-prone repair ^9-11.

The initiation and maintenance of the GC is dependant on BCL-6, a transcriptional repressor belonging to the BTB/POZ/ZincFinger family of transcription factors. BCL-6 is essential in the GC reaction, as evidenced by the observation that mice lacking BCL-6 cannot form GCs nor can produce high affinity antibodies ^12-13. BCL-6 is highly expressed in CBs, where it directly binds to and represses more than 1200 genes, as recently identified through integrated biochemical, functional and bioinformatics approach ¹⁴. BCL-6 target genes are involved in a variety of signaling pathways that are important for the GC reaction, including: i) DNA damage response, ii) apoptosis iii) plasma cell differentiation, iv) B-cell receptor (BCR) signaling, v) CD40 signaling, vi) TNFβ signaling, vii) Interferon (INF) signaling, viii) Toll-like receptor (TLR) signaling and ix) WNT signaling as well as x) T-cell mediated activation ^14-22. Taken together, these data indicate that BCL-6 is essential for the rapid proliferation of CBs, while allowing GC B-cells to undergo DNA modifications without inducing an unwanted DNA-damage response. Furthermore, BCL-6 inhibits the expression of several transcription factors that are essential for plasma cell differentiation ^{14,17-18,23-24}.

In the light zone of the germinal center, CBs differentiate into centrocytes (CCs), which are re-challenged by the antigen in order to allow the selection for B-cells that produce high-affinity antibodies, while cells with a low-affinity Ig-receptor are eliminated by apoptosis ²⁵. Furthermore, CCs undergo class-switch recombination (CSR), an intrachromosomal DNA recombination event that confers distinct effector functions to the antibodies by changing their immunoglobulin class from IgD and IgM to IgG, IgA or IgE ²⁶. CSR occurs via non-homologous end-joining and requires AID ^27-28. Another critical process that is initiated in the light zone of the GCs is the differentiation of B-cells with high-affinity Ig-receptor into effector plasma cells or memory B-cells. The down-regulation of BCL-6 is essential to allow terminal B-cell differentiation, and is accomplished in these cells through at least two distinct mechanisms, i.e. activation of CD40 and stimulation of the BCR. CD40 activation via CD40 ligand, expressed on CD4+ T cells, leads to NF-κB-mediated activation of interferon regulatory factor 4 (IRF4) and subsequent transcriptional silencing of BCL-6 ^29-30. The stimulation of the BCR promotes mitogen-activated protein kinases (MAPKs) mediated phosphorylation of BCL-6, followed by its ubiquitination and subsequent proteasomal degradation ^5,25,31. Down-regulation of BCL-6, in turn, restores DNA-damage responses, arrests proliferation, and allows for the expression of positive-regulatory-domain-containing 1 (PRDM1/ BLIMP1) a transcription factor required for plasma cell differentiation ^18,23.

All B-cell NHLs –with the exception of mantle-cell and lymphoblastic lymphoma – derive from either GC-cells or B-cells that have passed through the GC, as indicated by the fact that these lymphomas carry hypermutated IgV genes ³². In addition, two main mechanisms of genetic lesion in B-NHL, i.e. chromosomal translocations and aberrant somatic hypermutation (ASHM), occur as byproducts of AID-dependent DNA remodeling events that take place in the GC. The requirement for AID in GC-derived lymphomagenesis was recently confirmed by the analyses of transgenic mice where it was shown that GC-derived lymphomas do not develop in animals lacking AID ³³.

Diffuse large B-cell lymphoma subtypes derive from distinct B-cell differentiation stages

DLBCL is remarkably diverse in both clinical presentation and outcome, likely reflecting its pathogenetic and biologic heterogeneity. Over the past decade, the usage of genome-wide expression profiling has not only allowed a better understanding of the molecular mechanisms underlying the development of this disease, but has also revealed a number of features associated with an unfavorable clinical outcome ^2,34-38. According to similarities to the putative cell of origin, DLBCL can be divided into at least three different groups: i) germinal center B-cell like (GCB) DLBCL, which derives from centroblasts, ii) activated B-cell like (ABC) DLBCL, which resembles features of plasmablastic B-cells committed to terminal B cell differentiation, and iii) primary mediastinal large B-cell lymphoma (PMBCL), presumably arising from thymic B-cells ^36,39-40. However, 15-30% of DLBCL cannot be classified into any of the above subgroups ^2,41. The cell-of-origin based classification has prognostic value, since ABC-DLBCL have a worse overall survival as compared to GCB-DLBCL, and respond less effectively to current therapeutic regimes, with cure rates of around 40% ^35,42. Immunohistochemical markers have been shown to be able to discriminate the individual subgroups, and several of them –CD10, BCL-6, MUM1, BCL-2 and CYCLIN D2 – have been shown to be predictive of survival ^40,43. The combination of CD10, MUM1 and BCL-6 can divide DLBCL in GCB-DLBCL and non-GCB-DLBCL with about 80% concordance with the GEP ⁴⁰. A combination of 5 makers –GCTE1, CD10, BCL-6, MUM1 and FOXP1– can achieve about 90% concordance with the GEP ⁴⁴. In addition to the difference in cell of origin, these subgroups are associated with diverse genetic alterations (see below), suggesting that they depend on distinct oncogenic programs.

A separate classification scheme using gene-set enrichment analyses identified three phenotypic subsets characterized by the expression of genes involved in oxidative phosphorylation (OxPhos), B-cell receptor signaling (BCR) and host inflammatory response (HR) ³⁷. Tumors in the latter subset show increased expression of macrophage/dendritic cell markers, T/natural killer cell receptor and activation pathway components, as well as complement cascade members, and inflammatory mediators, suggesting an increased inflammatory response ³⁷. In HR tumors an increased number of infiltrating T cells and dendritic cells was found. Despite the increased immune response, these tumors do not have a favorable clinical outcome ³⁷.

Mechanisms of genetic alteration in DLBCL

Genetic alterations found in NHLs and in DLBCL in particular, include chromosomal translocations, mutations caused by ASHM, sporadic somatic mutations and copy number alterations, denoted by deletions and amplifications.

Chromosomal translocations in NHLs represent reciprocal and balanced recombination events frequently but not exclusively involving the immunoglobulin locus, with the breakpoint either located in the switch region or in the target region of SHM ^45-46. With very few exceptions, NHL-associated translocations do not lead to gene fusions, but cause dysregulated expression of the target gene. Given its critical function in both CSR and SHM, AID has been suggested to contribute to B-cell lymphomagenesis by facilitating the occurrence of chromosomal translocations and ASHM. Indeed, a direct role for AID in the development of GC-derived lymphomas has been demonstrated in mice ^33,47.

The term ASHM defines a mechanism of genetic lesion resulting from the aberrant activity of the physiologic SHM machinery, as strongly suggested by the features of the observed mutations – specifically, the pattern of nucleotide exchanges, the requirement of transcription, and the distribution within 2kb from the transcription initiation site ⁴⁸. By introducing mutations in the 5′ regulatory region of multiple genes, including coding sequences, ASHM is believed to play a major role in lymphomagenesis by causing deregulated expression of the target genes (often represented by proto-oncogenes) or by altering their protein function ^48,49,50. ASHM is primarily found in DLBCL, with more than 50% of cases affected ^48,51. Among the target genes identified so far are the well known proto-oncogenes MYC and PIM1 ⁴⁸.

In addition to chromosomal translocations and ASHM, altered gene expression in DLBCL can be caused by copy number alterations or somatic point mutations, analogous to non-lymphoid tumors. With the advent of novel techniques such as genome wide SNP array analysis and next-generation whole exome/genome sequencing analysis, novel lesions are likely to be identified. Recent examples include mutations and deletions of CREBBP and EP300, which are result in defective histone and non-histone protein acetylation in a significant fraction of DLBCL and FL (see below) ⁵².

Genetic Lesions associated with GCB-DLBCL

The gene expression pattern of GCB-DLBCL is similar to that of normal GC-B-cells ². In addition, these tumors retain typical features of normal GC-B-cells such as CSR and SHM ⁵³. GCB-DLBCLs are characterized by several, different genetic alterations (Tab. 1).

In approximately 35% of cases, the translocation t(14;18) leads to the ectopic expression of the B-cell lymphoma 2 (BCL-2) oncogene, a key anti-apoptotic molecule also expressed in follicular lymphoma (FL) and chronic lymphocytic leukemia ⁵⁴. BCL-2 translocations deregulate BCL-2 by juxtaposing potent regulatory elements from the Ig locus in close proximity to the BCL-2 locus, as well as by disrupting negative suppression by BCL-6 ^16,20. Additionally, about 40% of DLBCL cases without a t(14;18) translocation co-express BCL-2 and BCL-6 as the results of several mechanisms including i) deregulation of Miz1, the coactiator molecule by which BCL-6 binds to the BCL-2 promoter; i) ASHM of BCL-2 promoter sequences, iii) mutations in the BCL-2 coding sequence ¹⁶. Increased levels of BCL-2 have been associated with an inferior outcome in DLBCL ^55-56.

In 15% of DLBCL, the MYC gene, encoding for a transcription factor associated with Burkitt lymphoma, is deregulated due to chromosomal translocations –most commonly t(8;14) ⁵⁵– which bring the MYC coding sequence under the control of the immunoglobulin promoter ⁵⁷. Amplifications of the MIGH1 region containing the microRNA (miR) 17-92 cluster on chromosome 13q are found in around 12% of GCB-DLBCL ⁵⁸. This microRNA polycistron acts as a potential oncogene and accelerates MYC-induced lymphomagenesis in mice ^59-60. Furthermore, the cluster enhances oncogenesis by increasing proliferation and survival via inhibition of the tumor suppressor PTEN and the pro-apoptotic protein BIM ⁶⁰. Interestingly, deletions of PTEN on chromosome 10q, which are found in around 11% of cases and lead to the activation of AKT, are more frequent in GCB-DLBCL than ABC-DLBCL, and are mutually exclusive to amplifications involving the miR-17-92-cluster ^58,61. Activation of phosphatidylinositol 3 kinase (PI3K) activates AKT and other pathways that inhibit apoptosis, promote cellular growth, cell motility and angiogenesis. Activating Mutations in PI3K are found in about 8% of DLBCL and are mutually exclusive with loss of PTEN ⁶².

A number of chromatin modifying genes have been reported to be mutated in DLBCL with a prevalence in GCB-DLBCL. Recurrent somatic mutations affecting a single residue in the polycomb-group oncogene EZH2 have been found in 21.7% of GCB-DLBCL ⁶³. EZH2 is part of the Polycomb Repressive Complex and encodes for a histone methyltransferase trimethylating Lys27 of histone H3. The identified mutations result in the replacement of a single tyrosine within the gene Set domain and have been recently shown to enhance the ability of EZH2 to trimethylate histone H3, in part by increasing its affinity for the substrate ^63-65. The histone methyltransferase MLL2 is somatically mutated in around 30% of DLBCL. The majority of the mutations introduce premature stop codons and frame-shift insertions and deletions that most likely inactivate MLL2 ^66-67 (and unpublished data). Myocyte enhancer factor 2B (MEF2B) is a member of the MADS/MEF2 family of DNA binding proteins which cooperates with histone modifying enzymes to regulate expression of genes. MEF2B is mutated in about 9% of DLBCL ^66-67(and unpublished data). A recent study has identified monoallelic deletions and mutations inactivating CREBBP and EP300 in nearly 39% of GCB-DLBCL and less frequently in ABC-DLBCL (17% of samples) ⁵². CREBBP and EP300 are acetyltransferases that act as transcriptional co-activators in multiple signaling pathways. As a consequence of the mutations, CREBBP/EP300 lose their ability to acetylate BCL-6 and p53, a post-translational modification that inactivates BCL-6 by disrupting the recruitment of histone deacetylases (HDACs) and thus hindering its capacity to repress transcription, while representing an essential requirement for p53 activation ^68-69. Thus, CREBBP and EP300 mutations may contribute to lymphomagenesis by favoring the decreased activity of the tumor suppressor and constitutive activation of the oncogene ⁵². Interestingly, these mutations are also found in about 40% of follicular lymphoma, suggesting that DLBCL and FL share common pathogenetic events ⁵².

Mutations of p53 have been associated with transformation from follicular lymphoma, and should thus be more frequent in GCB-DLBCL, although conflicting results have been reported in other series. Nevertheless, p53 mutations were able to stratify GCB-DLBCL patients in subgroups with different survival ^70-72.

Mutations of the BCL-6 5′ regulatory region are found in about 75% of all DLBCL, and reflect the activity of the physiologic SHM mechanism operating in GC B-cells ^7-8. However, a subset of these mutations affecting the gene 5′ untranslated exon 1 were exclusively found in DLBCL – particularly, in GCB-DLBCL – where they impair a negative autoregulatory loop through which BCL-6 controls its own expression ^49,73.

Evidence of ASHM is found in GCB-DLBCL as well as in ABC-DLBCL, overall accounting for more than 50% of DLBCL patients ^50-51. However, different mutation frequencies have been observed at certain target genes in the two DLBCL phenotypic subtypes. MYC and BCL-2 are preferentially targeted by ASHM in GCB-DLBCL ¹⁶.

Genetic Lesions associated with ABC-DLBCL

The ABC subtype of DLBCL has a gene expression pattern that is similar to normal B-cells activated by BCR cross-linking in vitro and to a subset of GC B cells committed to plasma cell differentiation ². A prominent feature of the ABC-DLBCL gene expression signature is the enrichment in NF-κB target genes, suggesting that constitutive activation of NF-κB plays an important role in this disease subtype ^74-75. Indeed, ABC-DLBCL cell lines are specifically dependent on NF-κB activity, as interference with NF-κB signaling kills ABC- but not GCB-DLBCL, and the NF-kB transcription complex is found in the nuclei of the tumor cells in a large fraction of cases ^74,76. Constitutive NF-κB activation can be due to several distinct genetic alterations, which affect both positive and negative regulators of the pathway predominantly in ABC-DLBCL (Fig. 2 and Tab. 1) ^75,77-79. One of the most commonly involved genes is TNFAIP3, encoding for the negative NF-κB regulator A20, with about 30% of cases displaying biallelic inactivation by mutations and/or deletions ⁷⁷. A20 is a dual-function ubiquitin-modifying enzyme involved in termination of NF-κB responses ⁸⁰. Consistently, enforced expression of A20 in DLBCL cell lines carrying biallelic TNFAIP3 inactivation induced cell growth arrest and apoptosis by blocking NF-κB signaling, as demonstrated by the cytoplasmic relocation of p50 ^77,81. Genetic alterations of A20 are also present in other lymphomas characterized by constitutive NF-κB activation (e.g. marginal zone lymphoma, Hodgkin lymphoma), while they are rarely found in GCB-DLBCL ^77,81-83.

In normal B-cells, BCR induced activation of NF-κB requires CARD11, a scaffold protein that coordinates the activation of IκB kinase β ⁸⁴. Mutations of CARD11 are found in around 10% of ABC-DLBCL as well as in a smaller subset of GCB-DLBCL, and typically affect amino acids within or adjacent to the coiled-coil domain ^77,85. Introduction of CARD11 mutants into lymphoma cell lines leads to constitutive NF-κB activation, suggesting that lymphoma cells with CARD11 mutations are engaging the NF-κB pathway in the absence of BCR signaling ⁸⁵. CD79A and CD79B are proximal BCR subunits and were found to be mutated in around 20% of ABC-DLBCL ⁷⁴. The mutations target the immunoreceptor tyrosine-based activation motif (ITAM) of CD79A and CD79B, most frequently at a conserved tyrosine residue, thus increasing surface BCR expression and attenuating a feedback inhibitor of BCR signaling ⁷⁴. Overall, CD79A and CD79B mutations are thought to induce chronic active BCR signaling, with consequent activation of NF-κB, PI3K and MAP-kinase pathways.

During the normal immune response, NF-κB is also activated after stimulation of TLRs and receptors for interleukin (IL) 1 and 8 ^86-87. MYD88 functions as a signaling adaptor protein, coordinating the assembly of a complex that activates NF-κB following TLRs and IL-1 and IL8 receptors stimulation ^86-87. MYD88 mutations are found in about 30% of ABC-DLBCL, harboring the same amino acid substitution (L265P) in the Toll/IL-1 receptor domain ⁷⁸. The L265P mutant protein may promote cell survival by activating NF-κB-signalling, JAK kinase activation of STAT3, and secretion of IL-6, IL-10 and INFβ ⁷⁸.

Other mutations affecting NF-κB modulators in DLBCLs include TRAF2 (3%), TRAF5 (5%), MAP3K7 (5%) and RANK (8%) ⁷⁷. A TRAF2 mutant isolated from an ABC-DLBCL was able to activate NF-κB, while the functional consequences of the remaining mutations have not been studied in detail ⁷⁷.

Consistent with these findings, treatment with the proteasome inhibitor Bortezomib, which blocks the degradation of IκBα, leads to significantly higher response rates in ABC-DLBCL patients ⁸⁸.

Normal GC-B-cells require the down regulation of BCL-6 – mediated by NF-kB – and expression of IRF4, BLIMP1 and XBP1 for plasma cell differentiation ^17-18,24. In ABC-DLBCL, several lesions involving this pathway have been found. Chromosomal translocations of BCL-6, located on chromosome 3q27, are detected in about 35% of cases, with a two-fold higher incidence in ABC-DLBCL as compared to GCB-DLBCL ^73,89-91. The translocations are balanced, reciprocal and involve various alternative partners ^12,91-94. Most of the translocations result in a fusion transcript, with the promoter region and the first non-coding exon of BCL-6 being replaced by the translocation partner ^95-96. The most common translocations involve the immunoglobulin heavy-chain promoter, resulting in constitutive expression of BCL-6 ^12,91. Deregulated expression of BCL-6 is thought to play a critical role in blocking differentiation, decreasing the p53-mediated apoptotic response to DNA damage as well as providing a proliferative advantage ^14-15,17-20. This is supported by the fact that mice with deregulated BCL-6 expression develop DLBCL ⁹⁷.

The PRDM1 gene – located on chromosome 6q21 – encodes for BLMP1, a zinc finger transcriptional repressor. BLIMP1 is essential for the differentiation of B-cells into plasma cells ^98-100, which it promotes in part by repressing genes involved in BCR signaling and proliferation ^101-102. Inactivating mutations and deletions of PRDM1 are found in up to 30% of ABC-DLBCL ^103-106. Additionally, BLIMP1 can be inactivated by transcriptional repression through constitutively active BCL-6, as it is the case in patients carrying BCL-6 translocations. Consistent with this model, chromosomal transloactions of BCL-6 and genetic alterations affecting BLIMP1 are mutually exclusive ^103-105. Notably, the remaining over 30% of ABC-DLBCL lack BLIMP1 protein despite the expression of IRF4 and the absence of genetic alterations in BLIMP1 or BCL-6 ¹⁰⁵. Since IRF4 is invariably co-expressed with BLIMP1 in plasma cells and differentiating GC-B-cells, these observations suggest that alternative mechanisms exist to inactivate BLIMP1 in ABC-DLBCL. The role of BLIMP1 as a tumor suppressor has been recently demonstrated in a mouse model where conditional deletion of the gene leads to the development of lymphoproliferative disorders recapitulating features of ABC-DLBCL ^105-106.

Amplifications of the telomeric segment of chromosome 19q are found in about 25% of ABC-DLBCL ⁵⁸. SPIB – an ETS family transcription factor – is supposed to be one functionally important gene on chromosome 19q, a model supported by the fact that down regulation of SPIB was toxic to ABC-DLBCL cell lines ⁵⁸. In addition, a translocation between SPIB and the Ig heavy chain locus was found in one ABC-DLBCL cell line ⁴⁶.

Homozygous or heterozygous deletions of the INK4a/ARF locus are observed in about 30% of ABC-DLBCL ⁵⁸. P16^INK4a and p14^ARF regulate the pRB and the p53 tumor suppressor pathways, and inactivation of the p53 pathway via INK4a/ARF is found to inhibit apoptosis in aggressive lymphoid malignancies ^107-108.

Genetic Lesions associated with PMBCL

Primary mediastinal large B-cell lymphoma (PMBCL) accounts for less than 3% of NHLs and is a tumor of young adults with a median age of about 35 years ^109-110, and females being affected more frequently than men ¹¹¹. The only site of lymphoma involvement is the anterior mediastinum, but the rapidly growing bulky tumor can extend locally into other thoracic structures ¹¹². This distinctive location, together with its gene expression profile, suggest that this DLBCL subtype arises from thymic B-cells in the mediastinum ^111,113-114. PMBCL are characterized by several genetic alterations (Tab. 1). A common lesion in PMBCL is represented by gains of band 9p24, found in about 70% of the cases ^115-117. The amplified region contains several genes of possible pathogenetic significance, including the gene encoding for the tyrosine kinase JAK2 and the PDL1 and PDL2 loci, which are regulators of T-cell responses. These three genes were all expressed at high levels in PMBCL as compared to other DLBCL subtypes ³⁸. Recently PDL1 and PDL2 have been identified as key targets of these amplifications using high-resolution copy number data ¹¹⁸. Overexpression of JAK2 was proposed to be responsible for the constitutive activation of the transcription factor STAT6, due to its ability to activate IL-3 and IL-4, which are both involved in the regulation of STAT6 ^36,119. Moreover, SOCS1, a negative regulator of JAK2, is commonly affected by genetic lesions in PMBCL ^120-121.

PMBCL also shares with ABC-DLBCL the evidence of NF-κB pathway constitutive actiation, with PMBCL expressing increased levels of NF-κB targets that promote cell survival and inhibit apoptosis ¹²². Gains of chromosome 2p14-16 encompassing the NF-κB family member REL, are found in about 70% PMBCL and include ^115,117. Also affected by these gains is BCL11A, a transcriptional repressor that may provide a survival advantage ¹¹⁷.

Interestingly PMBCL and classic Hodgkin’s lymphoma have a notable genomic similarity, including gains on 2p and 9p ¹²³. Furthermore, remarkable similarities in the gene expression profiles of these two diseases have been found ^36,38. However, the gene expression signature of PMBCL retains the expression of genes typical of mature B-cells, while these transcripts are shut down in Hodgkin’s lymphoma.

Concluding remarks

Within recent years our understanding of the pathogenesis of DLBCL has significantly improved. Genome-wide expression profiling has illustrated the heterogeneity of DLBCL and allowed a better understanding of the molecular mechanisms underlying the disease. The known genetic lesions do not account for all DLBCL cases and therefore only represent a subset of those required for tumor initiation. New techniques like whole genome/exome sequencing and genome-wide SNP analyses have already led to the identification of novel genetically altered genes in DLBCL and further improved our understanding of the disease. Identification of tumors that are dependent on particular signaling pathways is essential to improve disease stratification and will provide more effective and less toxic targeted therapy approaches.