The Genetic Landscape of Diffuse Large B-Cell Lymphoma

Laura Pasqualuccia and Riccardo Dalla-Faverab

Diffuse large B-cell lymphoma (DLBCL), the most common lymphoid malignancy in the western world,

0037-1963/& 2015 Elshttp://dx.do

Conflicts ointerest oinapprop

SupportedFrom the InCancer C

aAssociateCancerColumbi

bProfessorUniversit

Address coCancer GNew Yor

Seminars

is an aggressive disease that remains incurable in approximately 30% of patients. Over the past decade,the rapid expansion of sequencing technologies allowing the genome-wide assessment of genomic andtranscriptional changes has revolutionized our understanding of the genetic basis of DLBCL by providinga comprehensive and unbiased view of the genes/pathways that are disrupted by genetic alterations in thisdisease, and may contribute to tumor initiation and expansion. These studies uncovered the existence ofseveral previously unappreciated alterations in key cellular pathways that may also influence treatmentoutcome. Indeed, a number of newly identified genetic lesions are currently being explored as markers forimproved diagnosis and risk stratification, or are entering clinical trials as promising therapeutic targets.This review focuses on recent advances in the genomic characterization of DLBCL and discusses howinformation gained from these efforts has provided new insights into its biology, uncovering potentialtargets of prognostic and therapeutic relevance.Semin Hematol 52:67–76. C 2015 Elsevier Inc. All rights reserved.

INTRODUCTION

Diffuse large B-cell lymphoma (DLBCL) is the mostprevalent B-cell non-Hodgkin lymphoma (B-NHL) in theadult, comprising 30% to 40% of all new diagnoses andincluding cases that arise de novo and cases that result fromthe histologic transformation of various, less aggressiveB-NHL types (i.e., follicular lymphoma [FL] and chroniclymphocytic leukemia).1 Although curable in a substantialproportion of patients by contemporary R-CHOP chemo-immunotherapy, as many as 40% of patients do notachieve durable remissions and will succumb to theirdisease. It has become clear that one of the reasons forsuch lack of success is the remarkable heterogeneity of thismalignancy, which encompasses multiple distinct sub-groups reflecting the origin from B cells at variousdevelopmental stages or the coordinated expression ofcomprehensive consensus clusters. These molecular sub-groups differ not only in the expression of specific gene

$ - see front matterevier Inc. All rights reserved.i.org/10.1053/j.seminhematol.2015.01.005

f interest: The authors declare that they have no conflicts ofr competing financial or personal relationships that couldriately influence the content of this article.in part by grant no. R01CA172492 (to L.P.).stitute for Cancer Genetics, Herbert Irving Comprehensiveenter, Columbia University, New York, NY.

Professor of Pathology and Cell Biology, Institute forGenetics, Department of Pathology and Cell Biology,a University, New York, NY.and Director, Institute for Cancer Genetics, Columbiay, New York, NY.

rrespondence to Laura Pasqualucci, MD. Institute forenetics, Columbia University, 1130 St. Nicholas Ave,k, NY 10032. E-mail: lp171@columbia.edu

in Hematology, Vol 52, No 2, April 2015, pp 67–76

signatures, but also in the oncogenic pathways that drivetumor development, often predicting discrete overallsurvival rates. Thus, a more precise definition of thegenetic changes that are associated with DLBCL isfundamental to improve our understanding of the disease,identify new therapeutic targets, and develop stratifiedapproaches to treatment.

Here we review current knowledge about the molecularpathogenesis of DLBCL, with emphasis on major bio-logical programs/pathways that are dysregulated by geneticlesions in the two main subtypes of the disease, as revealedby recent genomic profiling efforts.

CELLULAR ORIGIN OF DLBCL

The Germinal Center Reaction

Analogous to most B-NHL, DLBCL arises from theclonal expansion of B cells in the germinal center (GC), aspecialized microenvironment that forms in secondarylymphoid organs upon encounter of a naïve B cell withits cognate antigen, in the context of T-cell–dependent co-stimulation.2 GCs are highly dynamic structures wheremature B cells undergo rapid proliferation (o12 hoursdoubling time) and iterative rounds of somatic hyper-mutation (SHM), affinity maturation and clonal selection,as well as class switch recombination (CSR), with the aimof favoring the emergence of cells that produce antibodieswith increased affinity for the antigen and capable ofdistinct effector functions.3

These processes are compartmentalized within twoanatomically distinct areas where B cells recirculate bidir-ectionally: the dark zone (DZ), populated by rapidlydividing centroblasts, and the light zone (LZ), which is

67

NaiveB cell

Plasma cell

Memory B cell

PlasmablastThymicB cell

Ag

Dark zone Light zone

T cell

T cellFDC

epigenetic modification immune escape

proliferation apoptosis

BCL6 deregulation terminal differentiation

NF- B/BCR signaling JAK/STAT signaling

DNA damage response cell cycle other

MLL2/MLL3 MCREBBP/EP300 M/D

B2M/CD58 M/D

BCL6 Tx

MEF2B MFOXO1 M

TP53 M

32-38

3221-29

20-40

108

20

%TNFAIP3 M/D

MYD88 M

CD79A/B MCARD11 M

BCL2 AmpPRDM1 M/D

CDKN2A/B D

3030

209

24-3025

30

% PDL1/2 Amp/Tx

JAK2 Amp

CIITA TxSTAT6 M

SOCS1 M

TNFAIP3 M

TP53 MPTPN1 M

49

30

3836

45

36

20

%

20

BCL2 Tx/M

MYC Tx

EZH2 MGNA13 M

BCL6 BSE1 M

PTEN DmiR17-92 G

34

10

2225

15

6-116-12

%

Figure 1. GC and DLBCL pathogenesis. Schematics of the GC reaction and its relationship with major molecular subtypesof DLBCL. The most common shared and subtype-specific genetic alterations are shown, with color codes indicating theinvolved biological pathway. Blue, loss of function; red, gain of function; FDC, follicular dendritic cell.

L. Pasqualucci and R. Dalla-Favera68

composed of smaller non-dividing lymphocytes admixedwith a reticulum of follicular dendritic cells (Figure 1). DZand LZ B cells are characterized by unique biologicalprograms that are executed by a network of transcriptionfactors required for orderly GC development and whosederegulated expression is implicated in lymphomagenesis.The initiation of the GC reaction (i.e., the formation ofthe DZ) is orchestrated by a transitory peak in theexpression of NF-κB, IRF4, and MYC by a few GCfounder cells, followed by their downregulation in theoverall DZ population.3,4 In particular, MYC transcrip-tion is directly silenced by the GC master regulatorBCL6,5 a potent transcriptional repressor that, in theB-cell lineage, is expressed specifically during the GCreaction. BCL6 enables the DZ phenotype by modulatingthe activity of a broad set of genes involved in multiplesignaling pathways (see Supplementary data I), and isthought to sustain the proliferative status of GC cells whileallowing the execution of DNA remodeling eventsrequired for SHM and CSR, without eliciting DNA damageresponses. Additionally, BCL6 prevents the premature acti-vation and differentiation of GC B cells before the selectionfor the survival of high-affinity clones.6 Additional tran-scription factors that are required for GC formation and arerelevant for lymphomagenesis include TCF3 (E2A), whichenforces tonic BCR signaling in DZ B cells by regulating theexpression of downstream effectors, and EZH2, a histonemethyltransferase that helps establish bivalent chromatindomains at key regulatory loci, transiently suppressingterminal differentiation.7,8

On completion of this proliferative expansion in theDZ, B cells migrate to the LZ, where a variety of signals,including engagement of the BCR by the antigen,activation of the CD40 receptor by CD40 ligand, andstimulation of the BAFF and Toll-like receptors (TLR)activate downstream signaling cascades such as PI3K,MEK, and NF-κB. One consequence of this reactionand particularly of NF-κB activation is the re-expression ofIRF4, which binds to the BCL6 promoter and turns off itstranscription,9 thus relieving the expression of anothermaster regulator of plasma cell differentiation, BLIMP1(Supplementary data). Downregulation of BCL6 will thusrestore the ability of the B cell to become activated anddifferentiate into an antibody-secreting plasma blast.

The identification of the molecular circuits that regu-late the transition between these two phases is central tothe understanding of lymphomagenesis because thesetranscriptional programs can be recognized to some extentin the two major subtypes of systemic DLBCL based oncell of origin (COO) (see below). Moreover, the samecircuits are recurrent targets of genetic lesions in DLBCL,indicating that tumor cells exploit the unique properties ofGC B cells for their own selective advantage.

DLBCL Subtypes Derive from Distinct Phasesof B-Cell Differentiation

The clinical and biological heterogeneity of DLBCLhas been known to pathologists and clinicians for decades;however, it was the introduction of gene expression

The genetic landscape of DLBCL 69

profiling (GEP) technologies that allowed the formalrecognition of multiple distinct subtypes, reflecting eitherthe derivation from discrete B-cell differentiation stages orthe coordinated expression of specific transcriptionalsignatures. According to the current taxonomy forDLBCL, and based on resemblance to the transcriptionalprofiles of their presumed COO, at least three molecularsubgroups have been recognized within this diagnosticentity: GC B-cell–like (GCB) DLBCL, activated B-cell–like (ABC) DLBCL, and primary mediastinal large B-celllymphoma, while an additional 15% to 30% of casesremain unclassified.10–12 Both GCB and ABC-DLBCLappear more related to LZ B cells13; however, GCB-DLBCLs are defined by the elevated expression of BCL6and CD10, the absence of post-GC markers (e.g., IRF4and BLIMP1), and highly mutated immunoglobulin geneswith ongoing SHM.14 Conversely, the GEP of ABC-DLBCL resembles that of BCR-activated B cells orplasmablastic B cells, suggesting that its putative normalcounterpart has received signals to downregulate the GC-specific program and is poised to terminal B-cell differ-entiation. ABC-DLBCLs do not show evidence ofongoing SHM, consistent with their late GC origin.14,15

Primary mediastinal large B-cell lymphoma, a tumor ofthe mediastinum that shares significant similarities withHodgkin lymphoma, is thought to develop from a thymicpost-GC B cell or from a GC B cell migrated to thethymus.11,12 This lymphoma is recognized as a separateclinical-pathological entity from systemic DLBCL in therevised 2008 WHO Classification of Lymphoid Malig-nancies,1 and has been extensively reviewed recently16;thus, it will not be discussed.

Although imperfectly reproduced by immunohisto-chemistry,17 risk stratification based on the COO classi-fication holds prognostic significance because patientsdiagnosed with GCB-DLBCL have superior overall sur-vival compared with those presenting with ABC-DLBCL.18

A distinct classification scheme reflecting the coordi-nated transcription of comprehensive consensus clustersidentified three robust signatures defined by the expressionof genes involved in oxidative phosphorylation, B-cellreceptor/proliferation, and tumor microenvironment/hostinflammatory response.19 Both COO and comprehensiveconsensus cluster classifications are highly reproducible,yet they do not overlap, indicating that they capturedifferent aspects of DLBCL biology, further underscoringthe complexity of this disease.

MECHANISMS OF GENETIC LESION IN DLBCL

SHM and CSR are essential for the execution ofeffective immune responses20,21; yet, because of theirability to introduce DNA breaks, they expose the genomeof GC B cells to a constant risk. Moreover, these reactionstake place in an environment –the GC– wherein B cellsreplicate at remarkably fast rates and DNA damage

checkpoints are silenced as the result of the activity ofBCL6 transcriptional repressor. Accordingly, many of thestructural alterations implicated in DLBCL development(namely, chromosomal translocations and aberrantsomatic hypermutation [ASHM]) derive from errorsoccurring during one of these two reactions.22 Formalproof to this model came from the demonstration thatablation of AID, the enzyme required for both SHM andCSR, in lymphoma-prone mouse models was able toprevent the formation of MYC-IgH rearrangements andthe development of DLBCL.23,24

ASHM is a mechanism of genomic instability thattargets the 50 sequences of actively transcribed genes as theresult of a malfunction in the physiologic SHM process.25

In GC B cells, SHM is restricted to a few genes, includingthe immunoglobulin genes and BCL6,26,27 becausealthough AID can bind to multiple DNA sequences,28

mutations at off-target genes are normally repaired withhigh accuracy,29 preventing widespread mutational activ-ity. On the contrary, nearly half of DLBCL cases displaymultiple somatic mutations in a large number of activelytranscribed genes, including the proto-oncogenes PIM1and MYC.25,30 ASHM-generated lesions are typicallydistributed within promoter-proximal sequences and,depending on the genomic configuration of the targetgene, may affect untranslated as well as coding regions. Assuch, ASHM has the capability to either alter genetranscriptional regulation or modify key structural/func-tional properties.25 While a comprehensive characteriza-tion of the potentially extensive genetic damage caused byASHM is still missing, this mechanism likely contributesto the heterogeneity of DLBCL via the alteration ofdiverse cellular pathways in different cases.

Analogous to other cancers, DLBCL also harborsgenomic deletions, amplifications, and point mutationsthat lead to oncogenic activation or to inactivation oftumor suppressor genes.

THE GENOMIC LANDSCAPE OF DLBCL

In recent years, the development of powerful sequenc-ing technologies has offered an unprecedented opportunityto interrogate the cancer genome in a comprehensive andunbiased manner. The integration of whole-genome,whole-exome, and RNA sequencing approaches has sig-nificantly improved our understanding of the geneticlandscape of DLBCL by defining its degree of complexityand by revealing previously unrecognized genes/pathwaysthat may have contributed to its clonal expansion, includ-ing many that impact on remodeling the epigenome.While, for some such candidates, a detailed functionalcharacterization is still lacking, these studies have providedan incremental gain in our knowledge of the pathogenesisof DLBCL. Most importantly, they identified depend-encies of the tumor cells on specific molecules/circuits, which represent attractive targets for therapeuticintervention.

L. Pasqualucci and R. Dalla-Favera70

Compared with other B-cell malignancies such aschronic lymphocytic leukemia and acute leukemias, thecoding genome of DLBCL is relatively complex, with anaverage of 50 to over 100 lesions per case and greatvariability across different patients.31–34 Nonetheless, thecollection of genetic changes observed in each patientoften converge on common cellular pathways, suggestingcritical roles in DLBCL pathogenesis. It is important tonote that, because whole-exome sequencing approaches donot examine non-coding portions of the genome (i.e.,sterile transcripts, micro RNAs, and 50 sequences, thelatter representing the target domain of ASHM), theoverall mutation load of the DLBCL genome is expectedto be even higher.

DYSREGULATED CELLULAR PATHWAYS

The molecular complexity of DLBCL can be recapitu-lated by genetic lesions that are shared across differentCOO-defined phenotypic subtypes, and alterations thatpreferentially or even exclusively segregate with GCB- andABC-DLBCL, suggesting the addiction to distinct onco-genic pathways (Figure 1). The following sections willfocus on the most relevant genes/functional programs thatare derailed in DLBCL, selected based on historicalrelevance, functional characterization, recurrence in thedisease, and evidence for specific targeting by geneticalterations (e.g., genes encompassed by focal lesions).

Genetic Lesions Common to GCB- andABC-DLBCL

Alterations of histone modification genes

One commonly disrupted program only recently appre-ciated in DLBCL is represented by epigenetic remodeling.

Up to 30% of cases, with some preference for GCB-DLBCL, harbor mutations and/or deletions inactivatingCREBBP and, more rarely, EP300, two ubiquitouslyexpressed acetyltransferases that modify lysine residueson both histone and non-histone nuclear proteins, mod-ulating the activity of a large number of DNA-bindingtranscription factors.35,36 CREBBP mutations includetruncating events that remove the C-terminal HATdomain and amino acid changes that impair its affinityfor Acetyl-CoA, severely reducing its enzymatic activity.Except in few cases, they are observed in heterozygosis andare accompanied by expression of the residual wild-typeallele, suggesting a role as haploinsufficient tumor sup-pressors. Indeed, loss of a single CREBBP (or EP300) alleleis the cause of a rare congenital disorder (Rubinstein-Taybisyndrome) associated with developmental defects andtumor predisposition, providing evidence for a dose-dependent pathogenic effect of these genes.35 Whileadditional efforts will be needed to comprehensively definethe transcriptional network regulated by CREBBP in GCB cells and disrupted in DLBCL patients carryingCREBBP mutations, one mode by which these alterations

contribute to lymphomagenesis is through impaired ace-tylation of its substrates BCL6 and p53, which leads toconstitutive activation of the oncoprotein and decreasedfunction of the tumor suppressor.36,37 The balancedactivity of these two proteins is key for regulating DNAdamage responses during immunoglobulin remodeling inthe GC38; thus, one consequence of BCL6 activity over-riding p53 would be an increased tolerance for genomicinstability in the context of impaired apoptotic responses.The discovery of mutations in CREBBP and EP300 hastherapeutic implications because of the availability ofdrugs that inhibit deacetylation mechanisms and couldprovide therapeutic benefits by re-establishing physiologicacetylation levels in these patients.

At least one third of DLBCLs feature mutations in themixed lineage lymphoma/leukemia 2 (MLL2).33 MLL2encodes for a methyltrasferase that controls epigenetic tran-scriptional regulation by trimethylating the lysine 4 positionof histone 3 (H3K4). While the consequences of MLL2mutations in DLBCL have not yet been elucidated, mostevents are predicted to generate severely truncated proteinslacking the catalytic SET domain, which is required for itsmethyltransferase activity. Thus, MLL2 mutations are likelyto have a broad effect on chromatin regulation, which may inturn contribute to lymphomagenesis by reprogramming theepigenome of the precursor cancer cell. Importantly, inacti-vating mutations of both CREBBP and MLL2 are alsoobserved in cases of FL (40% and �89%, respec-tively),33,34,36 emerging as one of the most common geneticalterations reported in B-NHL to date. Moreover, recentstudies reconstructing the history of clonal evolution duringFL transformation to DLBCL suggest that MLL2 andCREBBP mutations represent early events introduced in acommon mutated ancestral clone before divergent evolutionto FL/transformed FL (tFL).39,40 Alterations in epigeneticmodifiers may thus facilitate the initial stages of trans-formation, by creating a permissive environment for theproliferation and survival of the cancer clone.

Alterations deregulating BCL6

Deregulation of BCL6 activity represents a key mech-anism of transformation in DLBCL, achieved via multipledirect and indirect modalities (Figure 2). Chromosomalrearrangements of the BCL6 locus characterize as many as35% of DLBCL patients, although with two- to three-foldhigher frequencies in ABC-DLBCL.41,42 These balanced,reciprocal recombination events juxtapose the codingdomain of BCL6 downstream to heterologous promotersderived from alternative chromosomal partners, leading toderegulated expression of an intact protein, in part bypreventing its downregulation during post-GC differentia-tion. It remains to be studied whether the consequences ofBCL6 translocations are different in ABC- and GCB-DLBCL because BCL6 controls functionally separatebiological programs depending on the interaction withdistinct co-repressor molecules.43 The BCL6 50 sequences

CBP/EP300*

NF- B

IRF4MEF2B*

FBXO11*BCL6 gene*

BCL6 protein

25%

30% 4%

10%

* mutated in DLBCL

DNA damage response

(p53, ATR)

Plasma cell differentiation

(Blimp1)

Programmed cell death

(Bcl2)

Cell Cycle arrest(p21)

B cell activation(CD80)

Figure 2. Deregulation of BCL6 activity by geneticlesions in DLBCL. Recurrent genetic alterations deregulat-ing BCL6 function in DLBCL, either directly or indirectly.Representative biological programs modulated by BCL6in the GC, and disrupted as a consequence of theselesions are shown at the bottom.

The genetic landscape of DLBCL 71

are also targeted by multiple point mutations in 470%of cases26,27; while these events largely reflect thephysiologic activity of SHM in the GC, a subset ofmutations clustering in the first BCL6 noncoding exonare exquisitely restricted to lymphoma, where theyderegulate BCL6 expression by at least two modalities:1) they disrupt a negative autoregulatory loop by whichthe BCL6 protein controls its own transcription44,45; 2)they prevent IRF4 binding to and transcriptional repres-sion of BCL6 following CD40 signaling.9 Intriguingly,BCL6 deregulating mutations are only found in GCB-DLBCL, suggesting that tumor cells exploit differentmechanisms depending on their functional state. Becausethe full extent of BCL6 mutations with documentedfunctional consequences has not been characterized, theexact percentage of cases carrying BCL6 genetic abnor-malities cannot be determined.

In addition to genetic lesions directly affecting theBCL6 gene, DLBCL have devised a number of ways toderegulate the BCL6 function indirectly. BesidesCREBBP/EP300 loss, which impairs acetylation-mediatedinactivation of BCL6, 10% to 15% of patients harborgain-of-function somatic mutations in the MEF2B tran-scription factor, a protein highly expressed in the GC andinvolved in the transcriptional activation of BCL6.33,46

These lesions promote the activity of MEF2B by at leasttwo mechanisms, entailing enhanced transcriptional acti-vation (mutations in the N-terminal MADS-box andMEF2 domain) or the loss of phosphorylation- andsumoylation-mediated negative regulatory motifs.46 In4% of cases, loss-of-function mutations/deletions ofFBXO11 impair proteasomal-mediated degradation ofthe BCL6 protein, which is controlled by this E3ubiquitin ligase.47 Consistent with a prominent oncogenicrole in lymphomagenesis, deregulated BCL6 expression in

a mouse model mimicking the DLBCL-associated trans-location leads to the development of clonal lymphoproli-ferative disorders recapitulating features of the humandisease.48

Loss of immune surveillance mechanisms

A large proportion of DLBCLs have evolved mecha-nisms to escape both CTL-mediated and NK-cell-mediated immune surveillance. In 29% of cases, thebeta-2-microgloblin (B2M) gene is lost because of struc-turally disruptive mutations and/or deletions, and another30% of cases lack B2M expression in the absence ofgenetic lesions, suggesting the existence of additionalgenetic or epigenetic mechanisms of inactivation.49 B2Mencodes for an invariant subunit of the HLA class Icomplex, which is expressed on the surface of all nucleatedcells and is required for recognition by cytotoxic Tlymphocytes.50 As a result, more than 60% of DLBCLlack surface HLA class I expression, which in turn mayfavor lymphomagenesis by allowing evasion from immunesurveillance.

Other lesions

Mutations and deletions of TP53 remain an importantpathogenic lesions in �20% of all DLBCL,34,51 includingthose derived from FL transformation39; recent studies alsosuggest that TP53 mutations affecting its DNA binding siteare most important from a prognostic standpoint. Also sharedacross both DLBCL subtypes are mutations of the FOXO1transcription factor. These events cluster around a phosphor-ylation site required for AKT-mediated nuclear-cytoplasmictranslocation and inactivation of FOXO1, and were suggestedto enhance its activity by preventing its nuclear exportfollowing PI3K signaling52; however, a systematic examina-tion of the effects of these mutations in the context of B cellsis still lacking, warranting further studies. FOXO1 mutationsare significantly enriched in patients with aggressive disease,suggesting a role for prognostication and risk stratification.52

GCB-DLBCL

Until recently, only a few lesions had been foundpreferentially associated with GCB-DLBCL, includingchromosomal translocations of BCL2 (34% to 45% ofcases) and MYC (10% to 14% of cases), which both leadto ectopic expression of the involved protein, in part byallowing escape from BCL6-mediated transcriptionalrepression5,53; deletions of the tumor suppressor PTEN(6-10% of cases)54; and, less frequently, amplifications ofthe region encompassing the mir-17-92 microRNA clus-ter, also a negative regulator of PTEN.54 This picturechanged significantly after genomic profiling studiesuncovered recurrent mutations in several previously unrec-ognized genes specifically in this subtype. The followingparagraphs will cover two novel mutation targets with

20%

9%

30%

30%

25% 25%

Figure 3. Disrupted pathways in ABC-DLBCL. Key signaling cascades engaged in LZ B cells and disrupted by geneticlesions in ABC-DLBCL. Bolts, gain-of-function mutations; crosses, loss-of-function mutations.

L. Pasqualucci and R. Dalla-Favera72

well-defined functional roles; other lesions that, althoughfrequent, are less characterized will not be discussed.

Mutations of EZH2

Heterozygous somatic mutations of the polycomb-group oncogene EZH2 have been reported in �22% ofGCB-DLBCL patients.55 With few exceptions, EZH2mutations result in the replacement of a single evolu-tionary conserved residue (Tyr641) within the proteinSET domain, leading to enhanced catalytic specificity andincreased levels of H3K27me3.56 In line with this, condi-tional expression of mutant EZH2 alleles in mice pro-motes GC hyperplasia and cooperates with BCL2 ininducing DLBCL.7 Notably, small-molecule EZH2 inhib-itors have just entered clinical trials for the treatment ofpatients with NHL, with promising results.57

Mutations in the Ga13 pathway

Approximately 30% of GCB-DLBCLs are character-ized by structurally damaging mutations in various com-ponents of a G-protein coupled inhibitory circuit thatregulates the growth and local confinement of GC B cells(namely GNA13, S1PR2 and, more rarely, ARHGEF1 andP2RY8) (Figure 1).58 Loss of these genes in the mouse wasassociated with increased GC B cell survival and dissem-ination to the lymph and bone marrow, ultimately leadingto lymphoma development.58

ABC-DLBCL

Of the three main subtypes of DLBCL, the genomiclandscape of ABC-DLBCL is the best characterized, beingassociated with a constellation of genetic abnormalities

that converge on two signaling pathways: activation ofNF-κB and block in terminal B cell differentiation.Additional recurrent lesions include amplifications of theBCL2 locus and deletions or lack of expression of theCDKN2A/2B tumor suppressor genes.54

Genetic lesions leading to constitutiveactivation of the NF-kB transcription factor

A prominent feature of ABC-DLBCL is the constitu-tive activation of the NF-κB signaling pathway, firstevidenced by the enriched expression of NF-κB targetgenes and the requirement of NF-κB for the proliferationand survival of ABC-DLBCL, but not GCB-DLBCL celllines.59 A number of studies have subsequently provideddirect evidence for the presence of genetic alterations inmolecules whose common denominator is the ability toinduce activation of NF-κB.

Mutations activating the BCR signaling pathway. MatureB cells require “tonic” signaling from the BCR to survive;however, ABC-DLBCL cells were found to display achronic, active form of BCR signaling, which requiresCARD11 and is sustained by the presence of geneticalterations in proximal members of the pathway.60

More than 20% of patients harbor somatic mutationsin the Ig superfamily members CD79B and, at lowerfrequencies, CD79A (Figure 3).60 In most cases, themutations replace the first tyrosine residue (Y196) in thecytoplasmic immunoreceptor tyrosine-based activationmotifs (ITAMs). These events are thought to circumventnegative feedback circuits that attenuate BCR signaling,thus maintaining it chronically active. Consistently,knockdown of several BCR proximal and distal subunits

The genetic landscape of DLBCL 73

is specifically toxic to ABC-DLBCL, supporting its directinvolvement in the pathogenesis of this disease andproviding the basis for the development of targetedtherapies directed against this pathway.

In �9% of ABC-DLBCL (and a smaller subset ofGCB-DLBCL), activation of BCR and NF-κB is sustainedby oncogenic mutations of the CARD11 gene. CARD11 isa major component of the “signalosome” complex, thecoordinated recruitment of which is required for propertransduction of BCR signaling (Figure 3).61 These eventscluster in the exons encoding for the protein coiled-coildomain and enhance the ability of CARD11 to trans-activate NF-κB target genes.62

Because activation of the BCR can trigger multipledownstream signaling cascades, besides canonical NF-κB(i.e., PI3K, ERK/MAP kinase, and NF-AT), it is expectedthat additional programs contribute to the neoplastictransformation of ABC-cells, as was also suggested byinitial experiments demonstrating the cooperative toxicityof NF-κB and PI3K inhibitors in ABC-DLBCL celllines.62

Mutations activating the TLR pathway. Oncogenicallyactive MYD88 mutations are found in one third ofABC-DLBCLs, where they target an evolutionarilyinvariant residue within the TIR (Toll/IL1 receptor)domain, leading to a L265P substitution.63 Thismutation induces IRAK4 kinase activity andphosphorylation through the spontaneous assembly ofa protein complex containing IRAK1 and IRAK4,which in turn can activate NF-κB as well as JAK/STAT3 transcriptional responses, also a phenotypic traitof ABC-DLBCL and a requirement for their survival.Other mutations in the MYD88 TIR domain wereobserved in both ABC- and GCB-DLBCL, but theirsignificance remains to be established.

Mutations inactivating negative regulators of NF-

kB. Almost one third of ABC-DLBCLsGCB_DLBCLs harbor biallelic TNFAIP3 truncatingmutations and/or deletions.64,65 TNFAIP3 encodes fora dual function ubiquitin-modification enzyme (A20)involved in the termination of NF-κB responsestriggered by TLR and BCR stimulation. A20 regulatesthese functions through the post translationalmodification of several substrates, where it firstremoves K63-linked regulatory ubiquitines via itsOTU domain, and subsequently conjugates K48-linked ubiquitines via its zinc finger domains, targetingthem for proteasome-mediated degradation. By removingthese domains, A20 mutations are thought to induceinappropriately prolonged NF-κB responses.64,65

Consistently, A20 knock-out mice display aninflammatory phenotype reflecting overactive NF-κB andTLR responses.66

In summary, multiple alterations dysregulate theNF-κB cascade at different levels in ABC-DLBCL. These

vulnerabilities offer a unique opportunity for the develop-ment of tailored therapeutic strategies; indeed, the Brutontyrosine kinase (BTK) inhibitor ibrutinib is rapidlyemerging as a novel paradigm for the treatment of ABC-DLBCL, where it should be effective in those casesharboring mutations in the BCR signaling cascadeupstream of BTK (Figure 3).57

Genetic lesions preventing terminaldifferentiation

The differentiation of GC B cells into plasma cellsrequires PRDM1, a sequence-specific transcriptionalrepressor that is upregulated in a subset of LZ B cellspoised to undergo plasma cell differentiation and in allplasma cells.67 However, �25% of ABC-DLBCLs havelost the PRDM1 gene because of truncating mutations,missense mutations, and/or genomic deletions,68–70 whilean additional sizeable fraction of patients lacks thePRDM1 protein because of transcriptional repression byconstitutively active, translocated BCL6 alleles.68 Rear-rangements of BCL6 and alterations inactivating PRDM1are mutually exclusive, supporting a complementary rolein promoting lymphomagenesis by blocking terminaldifferentiation. In line with this model, conditionaldeletion of PRDM1 in GC B cells in vivo leads tohuman-like ABC-DLBCL.68,71

PERSPECTIVE

Over the past decade, targeted resequencing and genomicprofiling have led to the discovery of recurrent, previouslyunappreciated genetic lesions, revealing the involvement ofbiological programs and signaling pathways that are central toDLBCL pathogenesis and identifying, in some cases, impor-tant new regulators of GC development and thereforehumoral immunity. These pathways represent vulnerabilitiesof the lymphoma cell that could be exploited for improveddiagnosis, prognostication, and therapeutic intervention.Notably, a number of drugs have been either newlydeveloped or “repositioned” to target genetically disruptedprograms in DLBCL (Figure 3). While these drugs areexpected to impact the standard of care for this malignancy,the complexity of the involved pathways and the overallheterogeneity of the disease suggest that precise patientstratification will be necessary to identify sensitive andresistant cases.

AcknowledgmentsThe authors thank all the members of the Dalla-Favera

and Pasqualucci laboratories for their contribution to thegeneration of data reported in this manuscript, andapologize to those whose work could not be described orcited because of space limitations.

L. Pasqualucci and R. Dalla-Favera74

APPENDIX A. SUPPORTING INFORMATION

Supplementary data associated with this article can befound in the online version at http://dx.doi.org/10.1053/j.seminhematol.2015.01.005.

REFERENCES

1. Swerdlow SH, Campo E, Harris NL, et al. WHO classi-fication of tumors of haematopoietic and lymphoid tissues,4th ed. Lyon, France: IARC, 2008.

2. Klein U, Dalla-Favera R. Germinal centres: role in B-cellphysiology and malignancy. Nat Rev Immunol. 2008;8(1):22-33.

3. Victora GD, Nussenzweig MC. Germinal centers. Ann RevImmunol. 2012;30:429-57.

4. Calado DP, Sasaki Y, Godinho SA, et al. The cell-cycleregulator c-Myc is essential for the formation and main-tenance of germinal centers. Nat Immunol. 2012;13(11):1092-100.

5. Dominguez-Sola D, Victora GD, Ying CY, et al. The proto-oncogene MYC is required for selection in the germinalcenter and cyclic reentry. Nat Immunol. 2012;13(11):1083-91.

6. Basso K, Dalla-Favera R. Roles of BCL6 in normal andtransformed germinal center B cells. Immunol Rev. 2012;247(1):172-83.

7. Beguelin W, Popovic R, Teater M, et al. EZH2 is requiredfor germinal center formation and somatic EZH2 mutationspromote lymphoid transformation. Cancer Cell. 2013;23(5):677-92.

8. Caganova M, Carrisi C, Varano G, et al. Germinal centerdysregulation by histone methyltransferase EZH2 promoteslymphomagenesis. J Clin Invest. 2013;123(12):5009-22.

9. Saito M, Gao J, Basso K, et al. A signaling pathwaymediating downregulation of BCL6 in germinal center Bcells is blocked by BCL6 gene alterations in B celllymphoma. Cancer Cell. 2007;12(3):280-92.

10. Alizadeh AA, Eisen MB, Davis RE, et al. Distinct types ofdiffuse large B-cell lymphoma identified by gene expressionprofiling. Nature. 2000;403(6769):503-11.

11. Rosenwald A, Wright G, Leroy K, et al. Molecular diagnosis ofprimary mediastinal B cell lymphoma identifies a clinicallyfavorable subgroup of diffuse large B cell lymphoma related toHodgkin lymphoma. J Exper Med. 2003;198(6):851-62.

12. Savage KJ, Monti S, Kutok JL, et al. The molecularsignature of mediastinal large B-cell lymphoma differs fromthat of other diffuse large B-cell lymphomas and sharesfeatures with classical Hodgkin lymphoma. Blood. 2003;102(12):3871-9.

13. Victora GD, Dominguez-Sola D, Holmes AB, Deroubaix S,Dalla-Favera R, Nussenzweig MC. Identification of humangerminal center light and dark zone cells and their relation-ship to human B-cell lymphomas. Blood. 2012;120(11):2240-8.

14. Lossos IS, Alizadeh AA, Eisen MB, et al. Ongoing immu-noglobulin somatic mutation in germinal center B cell-likebut not in activated B cell-like diffuse large cell lymphomas.Proc Natl Acad Sci U S A. 2000;97(18):10209-13.

15. Lenz G, Nagel I, Siebert R, et al. Aberrant immunoglobulinclass switch recombination and switch translocations in

activated B cell-like diffuse large B cell lymphoma. J ExperMed. 2007;204(3):633-43.

16. Steidl C, Gascoyne RD. The molecular pathogenesis ofprimary mediastinal large B-cell lymphoma. Blood.2011;118(10):2659-69.

17. Hans CP, Weisenburger DD, Greiner TC, et al. Confirma-tion of the molecular classification of diffuse large B-celllymphoma by immunohistochemistry using a tissue micro-array. Blood. 2004;103(1):275-82.

18. Rosenwald A, Wright G, Chan WC, et al. The use ofmolecular profiling to predict survival after chemotherapyfor diffuse large-B-cell lymphoma. N Engl J Med. 2002;346(25):1937-47.

19. Monti S, Savage KJ, Kutok JL, et al. Molecular profiling ofdiffuse large B-cell lymphoma identifies robust subtypesincluding one characterized by host inflammatory response.Blood. 2005;105(5):1851-61.

20. Muramatsu M, Kinoshita K, Fagarasan S, Yamada S,Shinkai Y, Honjo T. Class switch recombination andhypermutation require activation-induced cytidine deami-nase (AID), a potential RNA editing enzyme. Cell.2000;102(5):553-63.

21. Revy P, Muto T, Levy Y, et al. Activation-induced cytidinedeaminase (AID) deficiency causes the autosomal recessiveform of the Hyper-IgM syndrome (HIGM2). Cell.2000;102(5):565-75.

22. Kuppers R, Dalla-Favera R. Mechanisms of chromosomaltranslocations in B cell lymphomas. Oncogene. 2001;20(40):5580-94.

23. Pasqualucci L, Bhagat G, Jankovic M, et al. AID is requiredfor germinal center-derived lymphomagenesis. Nat Genet.2008;40(1):108-12.

24. Ramiro AR, Jankovic M, Eisenreich T, et al. AID is requiredfor c-myc/IgH chromosome translocations in vivo. Cell.2004;118(4):431-8.

25. Pasqualucci L, Neumeister P, Goossens T, et al. Hyper-mutation of multiple proto-oncogenes in B-cell diffuse large-cell lymphomas. Nature. 2001;412(6844):341-6.

26. Pasqualucci L, Migliazza A, Fracchiolla N, et al. BCL-6mutations in normal germinal center B cells: evidence ofsomatic hypermutation acting outside Ig loci. Proc NatlAcad Sci U S A. 1998;95(20):11816-21.

27. Shen HM, Peters A, Baron B, Zhu X, Storb U. Mutation ofBCL-6 gene in normal B cells by the process of somatichypermutation of Ig genes. Science. 1998;280(5370):1750-2.

28. Yamane A, Resch W, Casellas R. Deep-sequencing identi-fication of the genomic targets of the cytidine deaminaseAID and its cofactor RPA in B lymphocytes. Nat Immunol.2011;12(1):62-9.

29. Liu M, Duke JL, Richter DJ, et al. Two levels of protectionfor the B cell genome during somatic hypermutation.Nature. 2008;451(7180):841-5.

30. Khodabakhshi AH, Morin RD, Fejes AP, et al. Recurrenttargets of aberrant somatic hypermutation in lymphoma.Oncotarget. 2012;3(11):1308-19.

31. Lawrence MS, Stojanov P, Polak P, et al. Mutationalheterogeneity in cancer and the search for new cancer-associated genes. Nature. 2013;499(7457):214-8.

32. Lohr JG, Stojanov P, Lawrence MS, et al. Discovery andprioritization of somatic mutations in diffuse large B-celllymphoma (DLBCL) by whole-exome sequencing. ProcNatl Acad Sci U S A. 2012;109(10):3879-84.

The genetic landscape of DLBCL 75

33. Morin RD, Mendez-Lago M, Mungall AJ, et al. Frequentmutation of histone-modifying genes in non-Hodgkinlymphoma. Nature. 2011;476(7360):298-303.

34. Pasqualucci L, Trifonov V, Fabbri G, et al. Analysis of thecoding genome of diffuse large B-cell lymphoma. NatGenet. 2011;43(9):830-7.

35. Goodman RH, Smolik S. CBP/p300 in cell growth, trans-formation, and development. Genes Devel. 2000;14(13):1553-77.

36. Pasqualucci L, Dominguez-Sola D, Chiarenza A, et al.Inactivating mutations of acetyltransferase genes in B-celllymphoma. Nature. 2011;471(7337):189-95.

37. Bereshchenko OR, Gu W, Dalla-Favera R. Acetylationinactivates the transcriptional repressor BCL6. Nat Genet.2002;32(4):606-13.

38. Phan RT, Dalla-Favera R. The BCL6 proto-oncogenesuppresses p53 expression in germinal-centre B cells. Nature.2004;432(7017):635-9.

39. Pasqualucci L, Khiabanian H, Fangazio M, et al. Genetics offollicular lymphoma transformation. Cell Rep. 2014;6(1):130-40.

40. Green MR, Gentles AJ, Nair RV, et al. Hierarchy insomatic mutations arising during genomic evolution andprogression of follicular lymphoma. Blood. 2013;121(9):1604-11.

41. Iqbal J, Greiner TC, Patel K, et al. Distinctive patterns ofBCL6 molecular alterations and their functional consequen-ces in different subgroups of diffuse large B-cell lymphoma.Leukemia. 2007;21(11):2332-43.

42. Ye BH, Lista F, Lo Coco F, et al. Alterations of a zinc finger-encoding gene, BCL-6, in diffuse large-cell lymphoma.Science. 1993;262(5134):747-50.

43. Parekh S, Polo JM, Shaknovich R, et al. BCL6 programslymphoma cells for survival and differentiation throughdistinct biochemical mechanisms. Blood. 2007;110(6):2067-74.

44. Pasqualucci L, Migliazza A, Basso K, Houldsworth J,Chaganti RS, Dalla-Favera R. Mutations of the BCL6proto-oncogene disrupt its negative autoregulation indiffuse large B-cell lymphoma. Blood. 2003;101(8):2914-23.

45. Wang X, Li Z, Naganuma A, Ye BH. Negative autoregu-lation of BCL-6 is bypassed by genetic alterations in diffuselarge B cell lymphomas. Proc Natl Acad Sci U S A. 2002;99(23):15018-23.

46. Ying CY, Dominguez-Sola D, Fabi M, et al. MEF2Bmutations lead to deregulated expression of the oncogeneBCL6 in diffuse large B cell lymphoma. Nat Immunol.2013;14(10):1084-92.

47. Duan S, Cermak L, Pagan JK, et al. FBXO11 targets BCL6for degradation and is inactivated in diffuse large B-celllymphomas. Nature. 2012;481(7379):90-3.

48. Cattoretti G, Pasqualucci L, Ballon G, et al. DeregulatedBCL6 expression recapitulates the pathogenesis of humandiffuse large B cell lymphomas in mice. Cancer Cell. 2005;7(5):445-55.

49. Challa-Malladi M, Lieu YK, Califano O, et al. Combinedgenetic inactivation of beta2-Microglobulin and CD58reveals frequent escape from immune recognition indiffuse large B cell lymphoma. Cancer Cell. 2011;20(6):728-40.

50. Cresswell P, Ackerman AL, Giodini A, Peaper DR, WearschPA. Mechanisms of MHC class I-restricted antigen process-ing and cross-presentation. Immunol Rev. 2005;207:145-57.

51. Monti S, Chapuy B, Takeyama K, et al. Integrative analysisreveals an outcome-associated and targetable pattern of p53and cell cycle deregulation in diffuse large B cell lymphoma.Cancer Cell. 2012;22(3):359-72.

52. Trinh DL, Scott DW, Morin RD, et al. Analysis of FOXO1mutations in diffuse large B-cell lymphoma. Blood. 2013;121(18):3666-74.

53. Saito M, Novak U, Piovan E, et al. BCL6 suppression ofBCL2 via Miz1 and its disruption in diffuse large B celllymphoma. Proc Natl Acad Sci U S A. 2009;106(27):11294-9.

54. Lenz G, Wright GW, Emre NC, et al. Molecular subtypesof diffuse large B-cell lymphoma arise by distinct geneticpathways. Proc Natl Acad Sci U S A. 2008;105(36):13520-5.

55. Morin RD, Johnson NA, Severson TM, et al. Somaticmutations altering EZH2 (Tyr641) in follicular and diffuselarge B-cell lymphomas of germinal-center origin. NatGenet. 2010;42(2):181-5.

56. Yap DB, Chu J, Berg T, et al. Somatic mutations at EZH2Y641 act dominantly through a mechanism of selectivelyaltered PRC2 catalytic activity, to increase H3K27 trime-thylation. Blood. 2011;117(8):2451-9.

57. Roschewski M, Staudt LM, Wilson WH. Diffuse largeB-cell lymphoma-treatment approaches in the molecularera. Nat Rev Clin Oncol. 2014;11(1):12-23.

58. Muppidi JR, Schmitz R, Green JA, et al. Loss of signallingvia Ga13 in germinal centre B-cell-derived lymphoma.Nature. 2014;516(7530):2548-52.

59. Davis RE, Brown KD, Siebenlist U, Staudt LM. Constitu-tive nuclear factor kappaB activity is required for survival ofactivated B cell-like diffuse large B cell lymphoma cells.J Exper Med. 2001;194(12):1861-74.

60. Davis RE, Ngo VN, Lenz G, et al. Chronic active B-cell-receptor signalling in diffuse large B-cell lymphoma. Nature.2010;463(7277):88-92.

61. Thome M. CARMA1. BCL-10 and MALT1 in lymphocytedevelopment and activation. Nat Rev Immunol. 2004;4(5):348-59.

62. Lenz G, Davis RE, Ngo VN, et al. Oncogenic CARD11mutations in human diffuse large B cell lymphoma. Science.2008;319(5870):1676-9.

63. Ngo VN, Young RM, Schmitz R, et al. Oncogenically activeMYD88 mutations in human lymphoma. Nature. 2011;470(7332):115-9.

64. Compagno M, Lim WK, Grunn A, et al. Mutations ofmultiple genes cause deregulation of NF-kappaB indiffuse large B-cell lymphoma. Nature. 2009;459(7247):717-21.

65. Kato M, Sanada M, Kato I, et al. Frequent inactivation ofA20 in B-cell lymphomas. Nature. 2009;459(7247):712-6.

66. Boone DL, Turer EE, Lee EG, et al. The ubiquitin-modifying enzyme A20 is required for termination ofToll-like receptor responses. Nat Immunol. 2004;5(10):1052-60.

67. Shapiro-Shelef M, Lin KI, McHeyzer-Williams LJ, Liao J,McHeyzer-Williams MG, Calame K. Blimp-1 is required for

L. Pasqualucci and R. Dalla-Favera76

the formation of immunoglobulin secreting plasma cells andpre-plasma memory B cells. Immunity. 2003;19(4):607-20.

68. Mandelbaum J, Bhagat G, Tang H, et al. BLIMP1 is a tumorsuppressor gene frequently disrupted in activated B cell-likediffuse large B cell lymphoma. Cancer Cell. 2010;18(6):568-79.

69. Pasqualucci L, Compagno M, Houldsworth J, et al. Inactiva-tion of the PRDM1/BLIMP1 gene in diffuse large B celllymphoma. J Exper Med. 2006;203(2):311-7.

70. Tam W, Gomez M, Chadburn A, Lee JW, Chan WC,Knowles DM. Mutational analysis of PRDM1 indicates atumor-suppressor role in diffuse large B-cell lymphomas.Blood. 2006;107(10):4090-100.

71. Calado DP, Zhang B, Srinivasan L, et al. Constitutive canon-ical NF-kappaB activation cooperates with disruption ofBLIMP1 in the pathogenesis of activated B cell-like diffuselarge cell lymphoma. Cancer Cell. 2010;18(6):580-9.