Molecular pathogenesis of germinal center-derived B cell lymphomas

Summary
B cell lymphomas comprise a heterogeneous group of genetically, biologically, and clinically distinct neoplasms
that, in most cases, originate from the clonal expansion of B cells in the germinal center (GC). In recent years,
the advent of novel genomics technologies has revolutionized our understanding of the molecular
pathogenesis of lymphoid malignancies as a multistep process that requires the progressive accumulation of
multiple genetic and epigenetic alterations. A common theme that emerged from these studies is the ability of
lymphoma cells to co-opt the same biological programs and signal transduction networks that operate during
the normal GC reaction, and misuse them for their own survival advantage. This review summarizes recent
progress in the understanding of the genetic and epigenetic mechanisms that drive the malignant
transformation of GC B cells. These insights provide a conceptual framework for the identification of cellular
pathways that may be explored for precision medicine approaches.

INTRODUC TION
The adaptive immune system has evolved to allow the
effective recognition and response against a virtually
unlimited number of invading pathogens. These
processes require the formation of germinal centers (GC),
a specialized microenvironment that ensures the
generation and selection of cells producing antibodies with
high affinity for the antigen.1–4 The GC reaction is thus
critical for the establishment of humoral immunity.
However, the unique and complex events that are
associated with this reaction pose a significant risk to the
genome of GC B cells, which have to endure high
replication stress while undergoing multiple DNA
breakage and recombination events required by the
antibody diversification processes of somatic
hypermutation (SHM) and class switch recombination
(CSR).5
Additionally, GC B cells must have elevated epigenetic
plasticity in order to rapidly reprogram their transcriptional
networks while recirculating between the GC dark zone
(DZ) and light zone (LZ) compartments to assume distinct
functional states. Therefore, it does not come as a
surprise that the majority of mature B cell lymphomas
derive from the clonal expansion of a GC B cell, as
documented by multiple evidences that include the
presence of somatically mutated immunoglobulin (IG)
genes,6 the strong similarity of their gene expression
profiles with that of B cells in various phases of the GC
reaction,7,8 and the frequent association with genetic
lesions—ie, chromosomal translocations and aberrant
SHM—that result from errors occurring during GC-specific
DNA remodeling events.9,10 More recently, next-
generation sequencing studies and functional genomics
analyses of lymphomas have refined this concept by
uncovering multiple genetic alterations, the consequence
of which is the dysregulation of epigenetic and
transcriptional programs that operate during distinct
phases of the normal GC reaction, while facilitating
evasion from immune recognition mechanisms. Indeed,
the common theme of these studies has been the
realization that malignant cells co-opt the same signaling
pathways and regulatory networks that are normally
utilized by GC B lymphocytes to sustain their own growth
and survival. This review will summarize current
knowledge about the genetics and pathogenesis of the
most common GC-derived B-cell non-Hodgkin
lymphomas, including Burkitt lymphoma (BL), follicular
lymphoma (FL), and diffuse large B cell lymphoma
(DLBCL).
Among the plethora of genes and biological programs
identified to date as targets of genetic alterations in these
diseases, we will focus on those that are most frequently
affected and have been functionally characterized in the
context of the normal biology of the GC. We refer readers
to other reviews for a comprehensive overview on the
interaction between the tumor cells and the
microenvironment, 11 the role of infectious agents,12 and
the clinical implications for mechanism-based therapeutic
approaches.13,14
THE D OUBLE-EDG ED S WORD O F T HE
GC R E AC TION
Germinal centers are highly dynamic structures that form
transiently in secondary lymphoid organs upon encounter
of a B cell with a foreign pathogen to enable the generation
and selection of clones with high affinity antibodies.1,2,4,5
In 2007, an elegant series of in vivo imaging studies
demonstrated that, within these structures, B cells
constantly recirculate between two anatomically separate
compartments15– 18: the DZ, which consists of rapidly
proliferating cells (also known as centroblasts) capable of
dividing every 6-12 hours, and the LZ, which comprises
more quiescent cells termed centrocytes, admixed with
resident accessory cells such as follicular dendritic cells
(FDC), follicular helper T (Tfh) cells, and macrophages
(Figure 1).2,8 These two compartments correspond to
distinct functional states of the B lymphocytes.
Specifically, the DZ is the site where GC B cells undergo
secondary diversification of their antibody genes through
the process of SHM,19 a DNA remodeling event initiated
by the activation induced cytidine deaminase (AID)
enzyme20,21 that introduces single nucleotide
substitutions, as well as small deletions and duplications,
in the variable region of the IG genes (IGV), the final goal
being to increase the affinity of their B cell receptor (BCR)
for the antigen.22,23 DZ B cells then cease proliferation and
move to the LZ, where they compete for signals delivered
by a limited number of resident Tfh cells that trigger
positive selection.8 Based on affinity for the antigen, LZ B
cells will either re-enter the DZ for iterative rounds of cell
division and selection, which lead to affinity maturationat
the population level, exit the GC to become a memory B
cell or a plasma cell, or be eliminated by the default GC
apoptotic program. The LZ is also the site where B cells
undergo CSR, an intrachromosomal deletional
recombination event that requires the activity of AID and
confers distinct effector functions to antibodies with
identical specificities.24,25 Ultimately, cells will terminate
the cyclic DZ-LZ re-entry process and be selected for
survival and differentiation into long-lived memory B cells
and plasma cells.25,26
The two distinct phases of GC development reflect
discrete transient states within the same B cell
developmental stage, which can be recognized to some
extent in the transcriptional signature of the various
lymphoma subtypes, with BL being more similar to a DZ B
cell, and FL and DLBCL being more related to a GC LZ
cell.8
TR ANSCRIPTION F AC TOR N E TWORKS
WIRING T HE G C
As mentioned, a common theme of recent genomic
analyses has been the realization that most genetic
alterations associated with B lymphoid malignancies
converge on transcription factor networks that are
normally used by the GC B cell to sense antigens and
assume distinct functional states during the GC reaction.
This section focuses on selected proteins that, among
those implicated in the control of GC physiology, are most
commonly targeted by genetic lesions in lymphoma (see
ref. 2 for a comprehensive review of the molecular
switches that coordinate the complex dynamics of the GC
reaction).
As a sequence specific DNA-binding transcription factor,
the MYC protein controls a broad range of cellular
programs, including proliferation, cell growth, energy
metabolism, telomerase maintenance, differentiation, and
apoptosis,27 its target gene network being estimated to
include ~15% of all protein-coding genes as well as non-
coding RNAs.28 In addition, MYC was shown to control
DNA replication through mechanisms that are
independent of its transcriptional activity, a property that
may promote genomic instability by inducing replication
stress.29
During the GC reaction, MYC expression is tightly
regulated through a bimodal distribution pattern whereby
the protein is first induced in the GC founding B cells,
shortly upon antigen binding to the BCR, but is then
transcriptionally repressed by BCL6 (and most likely other
factors) in the DZ B cell population, before being re-
induced in a small subset of LZ B cells that are undergoing
affinity-based positive selection (Figure 2).30 Given its
critical role in almost all proliferating cells, the absence of
MYC expression in the GC DZ population has been a
matter of confusion and represents a paradox that is still
not completely understood.31 However, elegant in vivo
studies have shed light into the stage-specific role of MYC
during the GC response. In particular, it has been
demonstrated that upregulation of MYC expression in the
LZ is required for antigen selected B cells to re-enter the
cell cycle and recirculate to the DZ for additional rounds of
proliferation, SHM and selection.30 This process, known
as “cyclic re-entry,”
is critical to sustain affinity maturation and to maintain the
GC reaction.8,30,32 Consistently, conditional deletion of
MYC early upon antigenic stimulation abrogates GC
formation,32 and inhibition of its activity specifically in the
LZ leads to dissolution of established GCs.30 Recent
elegant work suggests that, in LZ B cells, MYC is
induced through BCR and CD40 ligation via NF-κB and
FOXO1 activation respectively
MEF2B and MEF2C
The myocyte-specific enhancer factor 2B (MEF2B) and
MEF2C proteins belong to an ancient family of
transcription factors involved in the regulation of multiple
developmental programs through interaction with specific
transcriptional co-factors. 34 Recent studies have
uncovered critical roles for these proteins in both normal
and malignant GC B cells, where MEF2B and MEF2C
appear to play slightly distinct functions. In particular,
MEF2C, which is expressed at all stages of mature B cell
differentiation, was shown to regulate the proliferation of
GC B cells by functioning as a direct transcriptional
effector of BCR signaling via the p38 MAPK pathway.35
Consistently, loss of MEF2C function caused reduced
immune responses to antigen and defective GC formation
due to impaired B-cell proliferation. 35 In contrast, MEF2B
is exquisitely expressed in the GC, where its transcription
is rapidly induced after T-cell dependent antigen
activation, slightly before the upregulation of BCL6.36,37
This is in line with the ability of MEF2B to directly bind
promoter/enhancer sequences in the BCL6 locus and
transactivate its expression.37,38 In addition, MEF2B is at
the center of a broad transcriptional network that includes,
among other cis-regulatory elements, the full set of GC-
specific super-enhancers, indicating an important role in
instructing the GC reaction, and particularly the DZ.39 In
this view, the modest effects on GC formation observed in
vivo upon single loss of Mef2b may suggest partially
redundant functions of Mef2b and Mef2c39, which can
dimerize with each other. In support of this hypothesis, co-
deletion of these two genes in GC B cells completely
abrogated GC formation
BCL6
Known as the “master regulator” of the GC reaction, BCL6
is a potent transcriptional repressor that, within the mature
B cell lineage, is selectively expressed in GC B cells40,41
and is absolutely required for GC development. The
specific signals that drive upregulation of BCL6 following
antigenic stimulation are in large part unknown; however,
multiple transcription factors have been implicated in this
process, including IRF4,42 IRF8,43 and MEF2B.37 BCL6
orchestrates the GC reaction by negatively modulating the
expression of a diverse set of genes involved in multiple
biological programs, which it recognizes through specific
motifs in their promoter-proximal ordistal sequences,44
recruiting distinct co-repressor complexes.45,46
Functions critical to the GC B cell phenotype and
controlled by BCL6 include T-cell mediated B cell
activation,47,48 BCR and CD40 signaling, 49,50 the induction
of apoptosis (eg, by suppression of BCL2),51
the sensing and response to DNA damage (eg, by
suppression of TP53, ATR, CHEK1),52–54 several cytokine
and chemokine signaling pathways,47,48 Notch2
signaling,55 and plasma cell differentiation (via
suppression of BLIMP1).56 Additionally, BCL6 regulates
gene transcription by repressing specific microRNAs (eg,
miR-155, which has inhibitory roles on AID).57,58 As such,
BCL6 is critical to establish the hyper-proliferative status
of DZ B cells, while allowing them to tolerate the DNA
breaks associated with SHM and CSR without eliciting cell
cycle arrest and apoptotic responses; moreover, BCL6
prevents the transduction of receptor signals that could
lead to premature activation and differentiation prior to the
selection of high affinity B cell clones. Upon completion of
these processes, expression of BCL6 needs to be turned
off in order to license B cell exit from the GC; the
mechanisms responsible for this molecular switch have
been partially elucidated and include at least two signals
that operate at the translational and transcriptional level
respectively: (a) engagement of the BCR by the antigen
acquired by FDCs, and (b) activation of the CD40 receptor
by the CD40 ligand present on Tfh-cells. 49,50,59 By
relieving the expression of its target genes,
downregulation of BCL6 will restore the ability of the cell
to respond to microenvironmental cues that lead to
termination of the GC reaction. The requirement of BCL6
for the GC response was conclusively demonstrated by
the lack of GC formation and affinity maturation in Bcl6-
deficient mouse models,60,61 a phenotype also induced by
deletion of a highly interactive, GC-specific locus control
region upstream of Bcl6.62
FOXO1
FOXO1 is a member of the Fox-O family of Forkhead
transcription factors that plays critical roles at defined
stage transitions during B cell development.63 Within the
GC, FOXO1 is almost ubiquitously expressed in the DZ;
this pattern is in keeping with the low activity in this
compartment of PI3K-AKT signaling, a major negative
regulator of FOXO1 that is only detectable in the GC LZ.64
Indeed, FOXO1 is necessary for the establishment and
maintenance of GC polarity, as well as for the DZ B cell
phenotype.64,65 The role of FOXO1 in sustaining the DZ
program entails its ability to transactivate multiple genes
involved in functions typical of DZ B cells, such as the
chemokine receptor CXCR4, cell proliferation, and
negative modulation of DNA repair, while silencing
signaling pathways characteristic of the GC LZ program
(eg, activation and differentiation), in part in cooperation
with BCL6.64,65 The importance of this dual activity is
demonstrated by the phenotype of GC-specific Foxo1 null
mouse models, which form apparently normal GCs
entirely composed of LZ B cells and devoid of DZ gene
programmes.64,65 Intriguingly, Foxo1-negative GC B cells
displayed normal SHM but defective affinity maturation
and CSR, suggesting that the AID-mediated SHM
machinery can be activated independently of the DZ
program.64,65
E2A (TCF3)
The E2A transcription factor plays an instructive role in
normal B-cell development and GC formation by
regulating the transcription of several B-cell restricted
genes, including those encoding the immunoglobulin
heavy and light chains, the negative BCR regulator SHP1
(a key contributor to the absence of BCR signaling in most
DZ GC B cells),66 and CCND3, which is required for the
proliferative expansion of GC B cells.67,68 Consistent with
its pro-proliferative function, E2A is expressed at higher
levels in DZ B cells8 and its deletion in vivo impaired the
expansion of GCs, although this gene appears to be
dispensable during the initial phases of antigen
activation.69
The NF-κB-IRF4-BLIMP1 axis
Activation of the NF-κB signaling cascade downstream of
signals emanating from the BCR, CD40 receptor, and
TLR,70 represents acritical step in both the initiation and
the termination of the GC reaction (Figure 2).42,71,72 In
particular, biochemical evidence and in vivo mouse
models have allowed to reconstruct a critical circuit that is
triggered by CD40:CD40L interaction in a subset of LZ B
cells and involves the upregulation of IRF4, a known NF-
κB target and a master regulator of terminal B cell
differentiation.71,72 At high concentrations, IRF4 can also
repress BCL6,59 a mechanism required for termination of
the mature GC programme and induction of the other
plasma cell master regulator, BLIMP1.73 BLIMP1, in turn,
maintains the plasma cell programme by suppressing
BCL6 through a negative feedback loop.56 Accordingly, a
small subpopulation of LZ B cells can be detected in the
GC LZ by immunofluorescence or immunohistochemistry
analysis, which displays nuclear translocation of the NFκB
subunit RELA and co-expression of IRF4 and BLIMP1, in
the absence of BCL6.49,74 Consistently, conditional GC-
specific deletion of RelA in mice blocks plasma cell
differentiation, 75 analogous to the phenotype observed in
Irf4 or Prdm1-deficient mice
EPIGENETIC CONTROL OF THE GC
The highly dynamic nature of the GC reaction requires the
rapid and coordinated switching between discrete
transcriptional programs that contribute to cell fate
determination in response to signals delivered by the
microenvironment.76 GC B cells must therefore be able
to quickly remodel their epigenetic landscape. Indeed,
recent studies showed that, within the GC
microenvironment, B cells undergo massive
reorganization of the genomic architecture that encodes
the GC B cell transcriptome.62 This reprogramming
process requires the activity of dedicated
histone/chromatin modifying enzymes that catalyze the
deposition of specific histone marks associated with open
or close chromatin. The identification of recurrent somatic
mutations in the genes encoding for these very same
enzymes was a major, unanticipated finding of recent
whole exome sequencing efforts in GC B lymphoid
malignancies.
EZH2
Enhancer of zeste homologue 2 (EZH2) is a SET-domain
Histone methyltransferase that silences gene transcription
by trimethylating the lysine 27 residue of histone H3
(H3K27me3) and recruiting the Polycomb Repressive
Complex-2 PRC2).77 Within the mature B cell lineage,
EZH2 is expressed specifically in the GC, where it creates
bivalent promoters that control the transcription of genes
involved in the negative regulation of cell cycle (CDKN1A)
and in terminal differentiation (IRF4, PRDM1).78,79 In
particular, recent studies defined a positive feedback loop
by which EZH2 enables the characteristic proliferative
phenotype of GC B cells by suppressing CDKN1A and
releasing the expression of E2F1.80 In vivo, loss of Ezh2
completely abrogated GC formation, establishing this
methyltransferase as a master regulator of the GC.78,79
CREBBP/EP300
CREBBP and EP300 are ubiquitously expressed,
pleiotropic regulators of gene expression that catalyze the
addition of acetyl groups to specific lysine residues in both
histone and non-histone proteins.81 Of particular
relevance to GC B cells, CREBBP has been shown to
oppose the proto-oncogenic activity of BCL6 by two
mechanisms: (a) the direct acetylation of the BCL6
protein, which prevents the recruitment of HDACs and
thus impairs its trans-repressor function82,83; and (b)
acetylation of H3K27 at the promoter/enhancer
sequences of virtually all BCL6 target genes, which
facilitates their transcription84,85 (Figure 3). Moreover,
CREBBP-mediated acetylation is an essential
requirement for the activation of the p53 tumor
suppressor, which is itself a target of BCL6.52,86 A
comprehensive characterization of the transcriptional
network modulated by CREBBP in the GC has been
recently obtained and revealed a preferential enrichment
in GC LZ signature genes,84,85 including components of
the BCR and CD40 signaling cascade, master regulators
of terminal differentiation, the CIITA transactivator,
multiple MHC-class II loci, and additional genes involved
in antigen presentation/processing84,85,87 (Figure 3).
CREBBP therefore appears to be particularly important in
the decision making of B cells that exit the GC.

KMT2D

The KMT2D gene (also known as MLL2 or MLL4) encodes

a member of the SET1 family of histone
methyltransferases that facilitates transcription by
predominantly mono-and di-methylating
the lysine at position 4 of histone H3 (H3K4), two
modifications that mark competent enhancers.88 The
genes bound by KMT2D in GC B cells and negatively
influenced in expression upon its deletion play key
roles in B-cell physiology, including the positive regulation
of apoptosis, CD40 signaling, and the control of cell
migration and proliferation. 89,90 Intriguingly, loss of Kmt2d
in CD19-Cre conditional knock-out
models, that is, before antigen encounter by naive B cells,
leads to a significant expansion of the GC B cell
population. These data suggest a role for this
methyltransferase in the commitment of mature B cells to
enter the GC reaction, as well as at later stages of GC
differentiation, and support current models of FL
pathogenesis inferring that KMT2D mutations, frequently
observed in FL and DLBCL, are early events in the history
of tumor evolution (see below).