R E V I E W S
`
`LYMPHOID MALIGNANCIES:
`THE DARK SIDE OF B-CELL
`DIFFERENTIATION
`
`A. L. Shaffer, Andreas Rosenwald and Louis M. Staudt
`
`When the regulation of B-cell differentiation and activation is disrupted, lymphomas and
`leukaemias can occur. The processes that normally create immunoglobulin diversity might be
`misdirected, resulting in oncogenic chromosomal translocations that block differentiation,
`prevent apoptosis and/or promote proliferation. Prolonged or unregulated antigenic stimulation
`might contribute further to the development and progression of some malignancies. Lymphoid
`malignancies often resemble normal stages of B-cell differentiation, as shown by molecular
`techniques such as gene-expression profiling. The similarities and differences between malignant
`and normal B cells indicate strategies for the treatment of these cancers.
`
`D E C I S I O N M A K I N G I N T H E I M M U N E S Y S T E M
`
`Normal lymphocyte differentiation is, in some sense, a
`disaster waiting to happen. B cells put their genomic
`integrity in danger during the formation and revision
`of their antigen receptors. A second potentially dan-
`gerous event is the response to antigen. When this
`response functions normally, the clonal expansion of
`B cells is regulated tightly by homeostatic controls.
`However, chronic infections can wreak havoc on lym-
`phocyte homeostasis, as can abnormal responses to
`self-antigens, and both of these mechanisms might
`contribute to lymphoid malignancies. Finally, many of
`the oncogenic events that occur in lymphomas and
`leukaemias disrupt the molecular pathways that regu-
`late B-cell differentiation, proliferation and apoptosis.
`A confounding issue in the study of human lym-
`phoid malignancies has been imprecision of diagnosis.
`Recent studies using gene-expression profiling and
`genomic mutational analysis have shown that lym-
`phomas and leukaemias that are difficult to distinguish
`histologically can nevertheless be molecularly distinct
`diseases. Using more-precise disease definitions, malig-
`nancies can be related often to distinct stages of B-cell
`differentiation. In this review, we focus on advances in
`the molecular definition of mature B-cell malignancies
`
`and discuss how the relationship between a lymphoma
`and its normal B-cell counterpart might be exploited
`to understand and treat these cancers. We discuss also
`how oncogenic alterations in these cancers subvert
`homeostatic regulation of lymphocyte responses.
`
`The perils of normal B-cell differentiation
`The first dangerous hurdle in B-cell differentiation is
`rearrangement of the immunoglobulin genes of B-cell
`precursors in the bone marrow to form a B-cell receptor
`(BCR). This molecular process, V(D)J RECOMBINATION,
`involves double-stranded DNA breaks that are initiated
`by recombination-activating genes (RAG1 and RAG2)
`and resolved by the non-homologous end-joining
`repair apparatus1. Occasionally, these breaks are resolved
`aberrantly, leading to chromosomal translocations. In
`lymphomas, chromosomal translocations typically
`replace the normal regulatory sequence of a gene with
`heterologous regulatory elements that drive inappropri-
`ate gene expression near the breakpoints. Clear examples
`of such mistakes in V(D)J recombination are t(14;18) —
`that is, a translocation between chromosomes 14 and 18
`— which involves the BCL2 gene and the immuno-
`globulin heavy-chain (IgH) locus in follicular lym-
`
`V(D)J RECOMBINATION
`The somatic rearrangement of
`variable (V), diversity (D) and
`joining (J) regions of antigen-
`receptor genes, which leads to
`the repertoire diversity of both
`
`Metabolism Branch,
`Center for Cancer Research,
`National Cancer Institute,
`National Institutes of
`Health, Bethesda,
`Maryland 20892, USA.
`Correspondence to L.M.S.
`e-mail: lstaudt@mail.nih.gov
`doi:10.1038/nri953
`
`NATURE REVIEWS |
`
`IMMUNOLOGY
`
`VOLUME 2 | DECEMBER 2002 | 1
`
`IPR2018-00685
`Celgene Ex. 2035, Page 1
`
`

`

`R E V I E W S
`
`T- and B-cell receptors.
`
`GERMINAL CENTRE
`The structure that is formed by
`the clonal expansion of antigen-
`activated B-cell blasts that have
`migrated to the follicles of
`lymph nodes. The B cells in these
`structures proliferate and their
`immunoglobulin genes undergo
`somatic hypermutation, before
`the cells leave as plasma cells or
`
`phoma (BOX 1), and t(11;14), which involves the gene
`encoding cyclin D1 and the IgH locus in mantle-cell
`lymphoma (BOX 1). The structures of the recombination
`breakpoints in these translocations are consistent with
`RAG-mediated cleavage of the IgH locus, guided by
`recombination signal sequences (RSSs). However, the
`recombination sites in the partner genes lack clear RSSs
`and do not have cryptic sequences that could function
`
`as RSSs2, although they sometimes involve DNA
`regions with altered structure3. The IgH breaks of
`t(11;14) in mantle-cell lymphoma seem to occur before
`heavy-chain diversity and joining segment (DH–JH)
`rearrangement, which indicates that this translocation
`occurs early in B-cell differentiation4. As RAGs are not
`expressed after the immature B-cell stage5,6, t(14;18)
`might occur in a pre-GERMINAL CENTRE (GC) B cell as well.
`
`Box 1 | Mature B-cell malignancies
`
`Follicular lymphoma
`An often indolent B-cell lymphoma with a follicular growth pattern. Most are characterized by the overexpression
`of BCL-2, owing to t(14;18). They comprise ~22% of non-Hodgkin lymphomas (NHLs). They cannot be cured by
`conventional chemotherapy and the survival rate is 73% at 10 years.
`
`Mantle-cell lymphoma
`A B-cell lymphoma that localizes to the mantle region of secondary follicles. Mantle-cell lymphoma (MCL) is associated
`with t(11;14), which results in the overexpression of cyclin D1. MCLs comprise 6% of all NHLs, have a male
`predominance and occur at a median age of 60. With current chemotherapy regimens, patients with MCL can achieve
`complete remission, but long-term remission is rare and median survival is 3–5 years.
`
`Burkitt lymphoma
`An aggressive B-cell lymphoma of children and young adults that is associated invariably with translocations of c-MYC.
`The endemic form involves Epstein–Barr virus (EBV) infection of malignant cells, whereas the sporadic form is EBV
`independent. These lymphomas can be cured in more than 80% of cases.
`
`Multiple myeloma
`An incurable malignancy of plasma cells with a median survival of three years. Multiple myeloma constitutes ~10%
`of all haematological malignancies, with a median age at diagnosis of ~65. Neoplastic cells are located in the bone
`marrow, and osteolytic bone lesions are characteristic. Reciprocal chromosomal translocations between one of the
`immunoglobulin loci and various other genes, including those that encode cyclin D1, cyclin D3, c-MAF, MMSET
`(multiple myeloma SET-domain protein) or fibroblast growth factor receptor 3 (FGFR3), are considered to be primary
`oncogenic events.
`
`Diffuse large B-cell lymphoma
`Diffuse large B-cell lymphoma (DLBCL) is the most common type of NHL (30–40% of cases). Up to one third
`of cases have abnormalities of BCL6, and ~20% of cases have translocations of BCL2. DLBCLs are clinically,
`morphologically and molecularly heterogeneous. 40% of patients with DLBCL can be cured by conventional
`chemotherapy.
`
`Hodgkin lymphoma
`This type of lymphoma accounts for ~10% of all lymphoid malignancies, and it usually arises in the lymph nodes of
`young adults. It can be subdivided into a classical subtype and a less common nodular lymphocyte predominant
`subtype. Cure rates of more than 80% can be achieved with present therapies.
`
`Lymphoplasmacytic lymphoma
`This is a rare form of NHL that comprises ~1.5% of nodal lymphomas. It is usually indolent and frequently involves
`bone marrow, lymph nodes and spleen. Most patients have monoclonal immunoglobulin M in their serum, and the
`tumour cells have a plasmacytic morphology. A subset of lymphoplasmacytic lymphomas is characterized by recurrent
`t(9;14), which involves the PAX5 (paired box gene 5) and immunoglobulin heavy-chain loci.
`
`Marginal-zone lymphoma
`This extranodal lymphoma occurs in organs that normally lack organized lymphoid tissue (such as the stomach,
`salivary glands, lungs and thyroid glands), and it comprises 7–8% of all B-cell lymphomas. In many cases, chronic
`inflammation or an autoimmune process precedes development of the lymphoma. Gastric mucosal-associated
`lymphoid tissue (MALT) lymphoma, the most common type, is associated with Helicobacter pylori infection, and 70%
`of patients at early stages have complete remission after eradication of this bacterium. At later stages, the acquisition
`of genetic abnormalities might lead to H. pylori-independent growth of the tumour cells or to transformation to an
`aggressive DLBCL.
`
`Chronic lymphocytic leukaemia
`The most common type of leukaemia, chronic lymphocytic leukaemia (CLL), is often an indolent disease with
`a median age of onset of 65. CLL is molecularly and clinically related to a nodal lymphoma known as small
`lymphocytic lymphoma. Current therapy can reduce symptoms, but it is not curative and does not prolong
`survival.
`
`2 | DECEMBER 2002 | VOLUME 2
`
`www.nature.com/reviews/immunol
`
`IPR2018-00685
`Celgene Ex. 2035, Page 2
`
`

`

`memory B cells.
`
`FOLLICULAR DENDRITIC CELLS
`(FDCs). Cells with a dendritic
`morphology that are present in
`lymph nodes, where they present
`intact antigens held in immune
`complexes to B cells.
`
`PLASMABLAST
`A dividing B cell that is
`committed to plasma-cell
`differentiation.
`
`CLASS-SWITCH
`RECOMBINATION
`DNA rearrangement of the VDJ
`region from immunoglobulin M
`to any of the IgG, IgA or IgE
`constant genes at the heavy-
`chain locus. Recombination
`occurs in repetitive sequences of
`DNA that are located upstream
`of each constant gene.
`
`SOMATIC HYPERMUTATION
`The substitution of
`‘untemplated’ nucleotides or
`small deletions targeted to a
`rearranged VDJ or VJ segment,
`which occurs only in B cells.
`The mutations are found
`between the promoter and
`enhancer of the rearranged gene
`(including non-coding regions),
`but they are found at the highest
`frequency in ‘hotspots’
`(RGYW) that are located in the
`complementarity-determining
`regions.
`
`This possibility is interesting because follicular lym-
`phomas seem to be arrested at the GC stage of differen-
`tiation (see later), which indicates that a naive B cell
`that has acquired a BCL2 translocation can nevertheless
`participate in an antigen-driven GC response.
`After antigen encounter, naive B cells follow one of
`three pathways: they can enter the GC microenviron-
`ment, where they interact with T cells, FOLLICULAR DENDRITIC
`CELLS and antigen7; they can differentiate into short-
`lived PLASMABLASTS outside of the GC8; or they can enter
`an unresponsive state known as anergy (FIG. 1a). In the
`GC, two molecular processes remodel DNA —
`immunoglobulin CLASS-SWITCH RECOMBINATION (CSR) and
`immunoglobulin SOMATIC HYPERMUTATION (SHM).
`Both CSR and SHM generate DNA breaks9–11 and
`are, therefore, dangerous mechanisms that might pre-
`dispose to chromosomal translocations. The DNA
`breaks that are induced by CSR and SHM coincide
`with the sites of chromosomal translocations that
`involve the IgH locus in certain lymphoid malignan-
`cies. SHM is probably involved in t(8;14) in endemic
`Burkitt lymphomas (BOX 1) because the c-MYC gene is
`often joined to the IgH locus in a rearranged and
`somatically mutated IgH variable (V) region12–14. SHM
`can also target non-immunoglobulin loci, such as
`BCL6 (REFS 15–17), and the involvement of these genes
`in translocations is probably a byproduct of this
`process. CSR is the culprit in many of the transloca-
`tions that occur in multiple myeloma (BOX 1) and spo-
`radic Burkitt lymphoma, because the translocation
`breakpoints occur in IgH switch regions18,19.
`
`Cell of origin
`Historically, the relationship between normal B-cell
`subpopulations and types of lymphoma has been
`assessed by a combination of microscopic appearance
`and immunophenotype. By these criteria, most
`mature B-cell malignancies seem to be ‘trapped’ at par-
`ticular stages of normal B-cell development. Follicular
`lymphomas, for example, have growth patterns that
`resemble those of normal GC B cells, and they are
`infiltrated with follicular dendritic cells and T cells.
`The tumour cells also express the membrane metallo-
`endopeptidase CD10, which is a hallmark of human
`GC B cells, leaving little doubt that follicular lym-
`phoma is a disease of GC B cells. However, in some
`lymphomas, the tumour cells show a spectrum of
`morphological differentiation, ranging from GC-like
`cells to plasmacytic cells, which indicates that the block
`in differentiation is not complete.
`When we speak of cell of origin we are, by necessity,
`referring to the relationship between the phenotype of
`the tumour on clinical presentation and a normal
`stage of B-cell differentiation. We cannot observe
`human lymphoid tumours during their natural evolu-
`tion from a normal B cell. Therefore, as mentioned
`earlier, oncogenic translocations might occur at an
`early stage of B-cell differentiation, after which the
`transformed B cell might differentiate further and
`arrest at a later stage of differentiation. The important
`point is that the phenotype of the tumour at clinical
`
`R E V I E W S
`
`presentation will influence its clinical behaviour and
`responsiveness to therapy.
`Several possible mechanisms could account for the
`apparent developmental arrest in many lymphoid
`malignancies. First, oncogenic alterations could inter-
`fere with regulatory networks that control lymphocyte
`differentiation. As discussed later, translocation of BCL6
`might cause lymphomas, in part, by blocking plasma-
`cytic differentiation. Second, the malignant lymphocyte
`might lose responsiveness to external cues, such as anti-
`gen or other immune cells that regulate normal differ-
`entiation. Third, it is conceivable that an oncogenic
`event might activate pathways that mimic a particular
`stage of normal differentiation. This possibility seems
`less likely in some lymphoid malignancies, as described
`later, that share extensive gene-expression profiles and
`biological functions with particular stages of B-cell
`differentiation.
`The analysis of somatic mutations in the rearranged
`immunoglobulin loci of lymphoid malignancies shows
`that there are clear differences between the diagnostic
`categories (TABLE 1). Most types of non-Hodgkin lym-
`phoma have highly mutated immunoglobulin genes
`that bear the hallmarks of SHM. A prominent exception
`to this rule might be mantle-cell lymphoma, which
`indicates that this lymphoma might be pre-GC in ori-
`gin. Of course, the mere presence of immunoglobulin
`mutations in a lymphoid malignancy only indicates that
`the cell that gave rise to the tumour had passed through
`a stage of B-cell differentiation during which SHM
`occurs. In some lymphomas, however, individual
`tumour cells in the malignant clone have distinct
`immunoglobulin sequences, which indicates that the
`tumour is frozen at a stage of differentiation at which
`SHM is ongoing (TABLE 1).
`The presence of immunoglobulin mutations in lym-
`phoid malignancies is usually taken as evidence that the
`cell of origin of the tumour passed through the GC
`microenvironment. Although most SHM takes place in
`GCs20, recent work indicates that it can occur also out-
`side of classical GC structures. Signalling through CD40
`is required to initiate and maintain the GC reaction21,22,
`and studies of hyper-IgM patients with genetic deficien-
`cies in CD40 ligand (CD40L) have shown that some
`SHM can take place in the absence of CD40 signalling23.
`In particular, a CD27+IgM+IgD+ subpopulation of
`somatically mutated memory B cells is retained in the
`peripheral blood of these patients, whereas other mem-
`ory B-cell subpopulations are absent. These studies
`indicate that SHM can occur outside of the GC, and
`they are reminiscent of earlier work in lymphotoxin-α-
`deficient mice, which lack GCs but can initiate SHM
`after several immunizations24. A direct observation of
`SHM outside of GCs was reported recently using a trans-
`genic mouse engineered to synthesize anti-IgG antibod-
`ies (rheumatoid factors)25. Clonal expansion of anti-IgG-
`specific B cells was observed in the T-zone–red-pulp
`border of the spleen, and the B cells in these proliferative
`foci were shown to have ongoing SHM at a rate similar
`to that seen in GC B cells. Given the possibility of extra-
`GC SHM, the presence of immunoglobulin mutations
`
`NATURE REVIEWS |
`
`IMMUNOLOGY
`
`VOLUME 2 | DECEMBER 2002 | 3
`
`IPR2018-00685
`Celgene Ex. 2035, Page 3
`
`

`

`R E V I E W S
`
`in a lymphoid malignancy cannot be taken as definitive
`evidence for a GC or post-GC cell of origin.
`Recently, the relationship of B-cell malignancies to
`normal stages of B-cell differentiation and activation has
`been clarified using genomic-scale gene-expression pro-
`filing. A unique gene-expression signature distinguishes
`GC B cells from other stages of B-cell differentiation,
`including resting naive and memory blood B cells and
`mitogenically activated blood B cells26,27 (FIG. 1b). The GC
`B-cell signature contains several hundred genes, includ-
`ing well-known GC markers (such as the genes encoding
`
`CD10, CD77 synthase and BCL-6) and many new genes
`of unknown function that were identified by high-
`throughput sequencing of complementary DNA libraries
`from normal GC B cells28. Expression of the GC B-cell
`signature genes is maintained in some lymphoma cell
`lines26, which indicates that this signature is a stable
`change in gene expression and does not require the cell-
`ular interactions that are present in the GC microenvir-
`onment to be maintained. So, the GC B cell is at a discrete
`stage of B-cell differentiation and is not just a specialized
`type of activated lymphocyte.
`
`Memory B cell
`
`Plasma cell
`
`CD10
`
`JAW1
`
`BCL-6
`
`CD77
`synthase
`
`FOXP1
`
`CD44
`
`Cyclin D2
`
`a
`
`Naive B cell
`
`Anergic
`B cell
`
`Antigen
`
`Activated
`B cell
`
`GC reaction
`
`Extra-GC
`B cells
`
`b
`
`Normal B cells
`
`B-cell lymphomas
`
`GC B cell
`
`Activated
`B cell
`
`Follicular
`lymphoma
`
`Burkitt
`lymphoma
`
`GC B-cell-like
`DLBCL
`
`Activated
`B-cell-like
`DLBCL
`
`signature
`GC B-cell
`
`signature
`B-cell
`Activated
`
`IRF4
`Figure 1 | Mature B-cell lymphomas: cell of origin. a | When naive B cells encounter antigen they become activated and
`face three cell fates: clonal expansion and selection in a germinal centre (GC), clonal expansion and differentiation at extra-GC
`sites, or anergy. Eventually, B cells either die or differentiate to memory B cells or antibody-secreting plasma cells. b | Gene-
`expression profiling shows a relationship between stages of B-cell differentiation and several types of mature B-cell lymphoma.
`Each column represents the results of gene-expression profiling from a single messenger RNA sample of normal or malignant
`B cells. Each row represents the expression of a single gene. Genes were chosen on the basis of their ability to distinguish
`between diffuse large B-cell lymphomas (DLBCLs) of GC and non-GC phenotype. Samples are compared with a common
`reference RNA pool, and relative gene expression is shown using a colour scale in which shades of red indicate genes that are
`expressed at a higher than median level, shades of green indicate genes that are expressed at a lower than median level, and
`black indicates genes that are expressed at the median level across all samples. A 16-fold range of gene expression is shown.
`Germinal-centre B-cell signature genes — for example, those that encode CD10, JAW1, BCL-6 and CD77 synthase — relate
`normal GC B cells to some lymphomas (follicular lymphomas, Burkitt lymphomas and the GC B-cell-like DLBCL subgroup).
`Genes that are expressed at a higher level in mitogenically activated peripheral-blood B cells than in GC B cells — for example,
`those that encode FOXP1 (forkhead box P1), CD44, cyclin D2 and IRF4 (interferon-regulatory factor 4) — uniquely identify the
`activated B-cell-like DLBCL subgroup. Potential cell types of origin for these lymphomas are indicated in pink (activated B cell)
`and blue (GC B cell) in part a.
`
`4 | DECEMBER 2002 | VOLUME 2
`
`www.nature.com/reviews/immunol
`
`IPR2018-00685
`Celgene Ex. 2035, Page 4
`
`

`

`Table 1 | Characteristics of mature B-cell malignancies
`Malignancy
`SHM
`Ongoing SHM
`
`Mantle-cell lymphoma
`
`Chronic lymphocytic
`leukaemia (CLL)‡
`Burkitt lymphoma
`Follicular lymphoma
`Marginal-zone lymphoma
`— nodal, extranodal (MALT)
`and splenic
`GC B-cell-like DLBCL
`Activated B-cell-like DLBCL
`
`No (except
`for a small
`percentage131)
`Yes and no
`
`Yes
`Yes
`Yes (except for
`some splenic
`variants132)
`Yes
`Yes
`
`No
`
`No
`
`No
`Yes
`Yes (prevalent
`in MALT
`lymphomas133)
`Yes
`No
`
`GC B-cell
`expression
`profile27,71,99,130
`No
`
`No
`
`Yes
`Yes
`No§
`
`Yes
`No
`
`R E V I E W S
`
`Putative cell of origin*
`
`Pre-GC B cell
`
`Antigen-experienced
`B cell (pre- or post-GC)
`GC B cell
`GC B cell
`GC B cell or post-GC
`B cell
`
`GC B cell
`GC B-cell subset or
`extra-GC mutated B cell23,25
`Post-GC B cell
`
`Post-GC B cell
`GC or post-GC B cell
`
`GC B cell
`
`Yes
`
`Yes
`Yes
`
`Yes
`
`Yes
`
`No
`No
`
`Yes
`
`No¶
`
`No
`No
`
`Yes
`
`Lymphoplasmacytic
`lymphoma (LPL)
`Multiple myeloma
`Hodgkin lymphoma
`(classical type)
`Hodgkin lymphoma
`(nodular lymphocyte
`pre-dominant type)
`*Based on the presence or absence of somatic hypermutation (SHM) and the gene-expression profile. ‡CLL consists of two clinically
`distinct subtypes, one with SHM and one without134. In some CLLs, subclones can accumulate additional mutations through SHM or
`another mutational process. §The gene-expression profile of marginal-zone lymphomas (MZLs) has yet to be determined, but MZL B cells
`lack germinal-centre (GC) markers (such as CD10 and BCL-6) and express marginal-zone markers (such as CD21 and CD35)86. ¶The
`gene-expression profile has yet to be determined, but LPL has characteristics that relate it to post-GC plasma cells (such as cytoplasmic
`immunoglobulin)97. DLBCL, diffuse large B-cell lymphoma; MALT, mucosal-associated lymphoid tissue.
`
`Several types of B-cell lymphoma express GC B-cell
`signature genes, including follicular lymphomas, Burkitt
`lymphomas and a subgroup of diffuse large B-cell lym-
`phomas (DLBCLs)26,29 (BOX 1 and FIG. 1). This finding
`establishes that these malignancies are derived from a
`GC B cell per se and not from a post-GC somatically
`mutated B cell. Although these malignancies retain
`expression of most of the GC B-cell signature genes, the
`lymphoma from an individual patient might have lost
`expression of any one GC B-cell marker. Furthermore,
`expression of a single gene is usually insufficient to estab-
`lish the relationship between a malignancy and its nor-
`mal counterpart, because many of the gene-expression
`differences between stages of differentiation are quanti-
`tative, not qualitative, in nature. Therefore, a ‘diagnosis’
`of a GC B-cell origin must be based on the expression of
`several GC B-cell signature genes to be accurate. By con-
`trast, other types of lymphoid malignancy fail to express
`these GC B-cell genes, and they have their own gene-
`expression signatures that relate them to other stages of
`B-cell differentiation (TABLE 1).
`About half of all DLBCLs fall into a gene-expression
`subgroup known as GC B-cell-like DLBCLs (GCB
`DLBCLs), which have a gene-expression profile that
`closely resembles that of normal GC B cells26,29 (FIG. 1
`and BOX 2). Furthermore, these lymphomas have highly
`mutated immunoglobulin genes and SHM is ongoing
`in malignant clones30. Gene-expression profiling indi-
`cates also that most GCB DLBCLs have undergone
`immunoglobulin class switching29 (A.R. and L.M.S.,
`
`unpublished observations). Together, these observations
`point to a GC B cell as the cell of origin for GCB
`DLBCLs, and they show that these tumours are trapped
`at this stage of differentiation.
`Another subgroup of DLBCLs, representing ~30%
`of cases, are known as activated B-cell-like DLBCLs
`(ABC DLBCLs), because these lymphomas resemble
`mitogenically activated peripheral B cells, and not GC
`B cells, in their gene-expression profile26,29 (FIG. 1). An
`important feature of ABC DLBCLs is the high level of
`expression of nuclear factor-κB (NF-κB) target genes,
`including those that encode BCL-2, interferon regula-
`tory factor 4 (IRF4), CD44, FLIP (FLICE-like inhibitory
`protein) and cyclin D2 (see below)31. These lymphomas
`have a high level of immunoglobulin somatic muta-
`tions, but they do not have ongoing SHM30. Nearly all
`ABC DLBCLs express a high level of IgM29 (A.R. and
`L.M.S., unpublished observations), which indicates that
`they have not undergone immunoglobulin class-switch
`recombination, a finding that is unexplained so far.
`The cell of origin for ABC DLBCLs is less clear than
`for GCB DLBCLs, although the absence of the GC
`B-cell gene-expression signature and the lack of ongoing
`SHM do not indicate a GC B-cell origin. ABC DLBCLs
`resemble pre-plasma cells in terms of gene expression in
`that they have higher levels of expression of immuno-
`globulin, X-box binding protein 1 (XBP1), IRF4 and
`other plasma-cell genes than GCB DLBCLs, and a lower
`level of expression of BCL-6 (REF. 29; A.R. and L.M.S.,
`unpublished observations). GCs contain a subpopulation
`
`NATURE REVIEWS |
`
`IMMUNOLOGY
`
`VOLUME 2 | DECEMBER 2002 | 5
`
`IPR2018-00685
`Celgene Ex. 2035, Page 5
`
`

`

`R E V I E W S
`
`of BCL-6−BLIMP1+IRF4+ cells32,33, which are a possible
`normal counterpart of ABC DLBCLs. Alternatively, as
`plasma-cell differentiation and SHM can occur outside of
`GCs23,25, it is possible that ABC DLBCLs are derived from
`B cells that have never entered a GC.
`
`Oncogenic lesions in lymphomas
`The various chromosomal translocations, amplifica-
`tions, mutations and deletions that occur in B-cell lym-
`phomas disrupt normal B-cell homeostasis in three
`ways: by driving the cells through the cell cycle, by pre-
`venting the normal induction of cell death and by
`blocking terminal differentiation. Analysis of normal
`GC B-cell gene expression and function indicates how
`these oncogenic events might perturb mature B-cell dif-
`ferentiation to give a selective advantage to the trans-
`formed cells (FIG. 2).
`
`Enhancing cell growth and proliferation. GC B cells
`are some of the most rapidly proliferating cells in the
`body, doubling in number every seven hours7, and gene-
`expression profiling has shown that they have a corre-
`spondingly high level of expression of cell-cycle progres-
`sion genes that function in the G2/M phase of the cell
`cycle — for example, the genes that encode CDC2 (cell-
`division cycle 2), PLK (polo-like kinase) and BUB1 (bud-
`ding uninhibited by benzimidazoles 1 homologue)27,29.
`Interestingly, however, genes that control cell growth
`(increase in cell size) are expressed at low levels in GC
`B cells. These genes encode components of the protein-
`translation machinery (for example, ribosomal proteins
`and translation-initiation factors) and intermediary
`
`metabolism (for example, glycolytic enzymes), and they
`are turned on as lymphocytes ‘blast’ in response to mito-
`genic stimuli. Many of these genes are targets of the
`transcription factor c-MYC34. Although GC B cells
`express c-MYC35,36, they express lower levels of c-MYC
`messenger RNA than other dividing cells27,29. Given the
`drive for positive and negative selection in the GC, it
`might be appropriate that cells tip the balance in favour
`of proliferation rather than cell growth to expand the
`pool of selectable B cells as rapidly as possible.
`The expression of c-MYC is altered by transloca-
`tions, mutations and/or overexpression in many GC
`B-cell-derived lymphomas. All Burkitt lymphomas have
`translocations of c-MYC to immunoglobulin loci37, and
`these translocations occur also in some DLBCLs38.
`Burkitt lymphomas and DLBCLs can also accumulate
`somatic mutations of c-MYC that might alter its func-
`tion as a transcription factor17,39. These oncogenic
`lesions of c-MYC could increase cell growth and pro-
`mote tumour-cell proliferation34.
`
`Blocking apoptosis. GC B cells seem to be poised for
`apoptosis unless they are rescued by positive selection7.
`Many anti-apoptotic proteins, such as BCL-2, A1 and
`BCL-XL, are expressed at low levels in most GC
`B cells27,29,40–42. The NF-κB signalling pathway transcrip-
`tionally activates several of these genes and delivers a
`potent anti-apoptotic stimulus to cells43. NF-κB tran-
`scription factors are kept in an inactive state in the cyto-
`plasm by interactions with inhibitor of NF-κB proteins
`(IκBs). Signalling through various cell-surface receptors
`activates IκB kinase (IKK), which phosphorylates IκB,
`
`Box 2 | Diffuse large B-cell lymphoma: many diseases in one diagnostic category
`
`On the basis of morphological and clinical criteria, the diagnostic framework that is used at present places diffuse large
`B-cell lymphomas (DLBCLs) in a single category and, consequently, all patients receive the same therapy97. However,
`patients with DLBCL are markedly heterogeneous in their response to multi-agent chemotherapy in that ~40% can be
`cured, whereas the remainder succumb to the disease124.
`Recent gene-expression profiling of DLBCL has shown that this single diagnostic category includes more than one
`molecularly and clinically distinct disease. DLBCL consists of at least three gene-expression subgroups, known as
`germinal-centre B-cell-like (GCB), activated B-cell-like (ABC) and type 3 (REFS 26,29). The gene-expression subgroups
`differ by the expression of more than 1,000 genes, which makes them as distinct as acute lymphoblastic and acute
`myelogenous leukaemias. As discussed in detail in this review, these subgroups seem to be derived from different stages
`of normal B-cell differentiation.
`The DLBCL gene-expression subgroups have distinct mechanisms of malignant transformation, which shows that
`they are pathogenetically distinct diseases. The t(14;18), which involves BCL2, is seen exclusively in GCB DLBCLs and is
`present in ~20% of these cases29,125. Similarly, amplification of the c-REL locus on chromosome 2p occurs only in GCB
`DLBCLs29. By contrast, activation of the anti-apoptotic nuclear factor-κB (NF-κB) pathway is a feature of ABC DLBCL,
`but not GCB DLBCL31.
`The molecular distinctions between subgroups of DLBCL are important because the subgroups differ in their ability
`to be cured by the multi-agent chemotherapy that is used at present26,29. Patients with GCB DLBCL have the most
`favourable cure rate (60% five-year survival), whereas patients with ABC and type 3 DLBCL have five-year survival
`rates of only 36% and 39%, respectively29. These clinical differences might be owing, in part, to the ability of the NF-κB
`pathway to block many forms of cell death, including that induced by chemotherapy126.
`Further gene-expression differences between DLBCLs can affect the success of chemotherapy29,127. The expression of
`genes that are associated with proliferation predicts poor outcome29. By contrast, expression of MHC class II genes by
`lymphoma cells predicts a favourable outcome, which indicates that DLBCLs might downregulate expression of these
`genes to evade an immune response29. Some patients mount a reactive lymph-node response to DLBCL cells that
`involves macrophages, natural killer cells and stromal cells, and this innate immune response is associated with survival
`after chemotherapy29.
`
`6 | DECEMBER 2002 | VOLUME 2
`
`www.nature.com/reviews/immunol
`
`IPR2018-00685
`Celgene Ex. 2035, Page 6
`
`

`

`GC B cell
`
`Antigen
`
`FDC
`
`Plasma cell
`
`BCL-6+
`IRF4–
`BLIMP1–
`
`T
`
`B
`
`Apoptosis
`
`Plasmablast/plasma cell
`
`B
`
`B
`
`PB
`
`B
`
`B
`
`B
`
`BCL-6–
`IRF4+
`BLIMP1+
`
`B
`
`B
`
`B
`
`Oncogenic
`mechanism
`
`BCL-6
`PAX5
`
`BCL-2
`NF-κB
`
`c-MYC
`
`Figure 2 | Mechanisms of malignant transformation of germinal-centre B cells. The
`germinal centre (GC) is a microenvironment in which B cells undergo rapid clonal expansion in the
`presence of T cells, follicular dendritic cells (FDCs) and antigen. In a process of positive selection,
`GC B cells mutate their immunoglobulin genes, and those that acquire mutations that maintain or
`improve the affinity of the B-cell receptor (BCR) for antigen are rescued from programmed cell
`death and can differentiate further. Most GC B cells express BCL-6 but lack expression of
`interferon-regulatory factor 4 (IRF4) and B-lymphocyte-induced maturation protein 1 (BLIMP1).
`Some GC B cells have a plasmablastic (PB) phenotype, have turned off expression of BCL-6 and
`have turned on expression of IRF4 and BLIMP1. These cells are likely to become plasma cells
`(PCs). Several oncogenic mechanisms can subvert normal GC B-cell homeostasis in lymphomas.
`c-MYC, which is not expressed highly by normal GC B cells, is often expressed as a result of
`translocation by Burkitt lymphoma cells, which promotes clonal expansion. Apoptosis is blocked
`by oncogenic activation of the nuclear factor-κB (NF-κB) pathway, which occurs in diffuse large
`B-cell lymphomas (DLBCLs), marginal-zone lymphomas, Hodgkin lymphomas and Epstein–Barr
`virus-related lymphomas. Apoptosis is also abrogated by translocation, amp