Mechanisms of chromosomal translocations in B cell ... - Nature

0 downloads 0 Views 672KB Size Report
Mechanisms of chromosomal translocations in B cell lymphomas. Ralf KuČppers*,1,2 and Riccardo Dalla-Favera1. 1Institute of Cancer Genetics, Columbia ...
ã

Oncogene (2001) 20, 5580 ± 5594 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

Mechanisms of chromosomal translocations in B cell lymphomas Ralf KuÈppers*,1,2 and Riccardo Dalla-Favera1 1

Institute of Cancer Genetics, Columbia University, New York, NY 10032, USA; 2Institute for Genetics and Department of Internal Medicine I, University of Cologne, Cologne, Germany

Reciprocal chromosomal translocations involving the immunoglobulin (Ig) loci are a hallmark of most mature B cell lymphomas and usually result in dysregulated expression of oncogenes brought under the control of the Ig enhancers. Although the precise mechanisms involved in the development of these translocations remains essentially unknown, a clear relationship has been established with the mechanisms that lead to Ig gene remodeling, including V(D)J recombination, isotype switching and somatic hypermutation. The common denominator of these three processes in the formation of Ig-associated translocations is probably represented by the fact that each of these processes intrinsically generates double-strand DNA breaks. Since isotype switching and somatic hypermutation occur in germinal center (GC) B cells, the origin of a large number of B cell lymphomas from GC B cells is likely closely related to aberrant hypermutation and isotype switching activity in these B cells. Oncogene (2001) 20, 5580 ± 5594. Keywords: chromosomal translocation; class switch recombination; germinal center; lymphomagenesis; somatic hypermutation; V(D)J recombination Introduction Distinct types of cancers often show characteristic genetic changes involved in tumorigenesis. A hallmark of mature B cell lymphomas are balanced chromosomal translocations, mostly involving the immunoglobulin (Ig) genes and a variety of partner genes (DallaFavera and Gaidano, 2001). Some of these are associated with speci®c subtypes of mature B cell lymphoma. Prominent examples include the bcl-1/Ig translocation in mantle zone lymphoma, the bcl-2/Ig translocation in follicular lymphoma and the c-myc/Ig translocation in Burkitt's lymphoma. In most instances, the translocated partner gene becomes transcriptionally deregulated as a consequence of its transposition into the Ig locus. The consequences of such dysregulated expression of proto-oncogenes can often be inferred from the normal functions of their protein products. For example, bcl-1 is a cyclin

*Correspondence: R KuÈppers, University of Cologne, Department of Internal Medicine I, LFI E4 R706, Joseph-Stelzmannstr. 9, D-50931 Cologne, Germany; E-mail: [email protected]

involved in control of the cell cycle, bcl-2 functions by inhibiting apoptosis, and c-myc is a major control factor for cellular growth (Dalla-Favera and Gaidano, 2001). As will be discussed below, the frequent involvement of Ig loci in chromosomal translocations re¯ects the remodeling of Ig genes that occurs during B cell development as a result of V gene segment assembly, somatic hypermutation, and class switch recombination. This review will focus on the involvement of these processes in chromosomal translocations associated with tumors derived from mature B cells, including B cell non Hodgkin lymphomas and multiple myeloma. For a discussion of chromosomal translocation in immature B cell tumors, which often lack an association with Ig loci, the reader is refered to recent reviews (Harrison, 2000; Rowley, 1999).

Remodeling Ig genes: V(D)J recombination, somatic hypermutation and class switch recombination V(D)J recombination During B cell development in the bone marrow, B cell precursors form intact exons for the variable region of antibody molecules by recombining individual Ig gene segments. For the variable region exon of the Ig heavy chain, V, D and J gene segments are joined, whereas the V region exons of k and l light chains are each composed of V and J segments (Rajewsky, 1996; Tonegawa, 1983). These gene rearrangements occur in an ordered fashion by a process called V(D)J recombination (Figure 1). First, a DH segment is rearranged to a JH segment. This is followed by a VH to DHJH joining, which can be either in-frame (i.e. in the correct reading frame for encoding antibody sequences) or out-of-frame. In the latter case, a second attempt can be made on the other heavy chain allele, whereas in the ®rst case, the pre B cell can proceed with development by undergoing Ig light chain gene rearrangements. V(D)J recombination is initatiated when two lymphocyte-speci®c endonucleases, the recombination activating gene (RAG) 1 and 2 proteins cut the rearranging gene segments at speci®c recombination signal sequences (RSS) that ¯ank the coding sequences of the V, D and J segments (reviewed in Fugmann et al., 2000). Each RSS is composed of a conserved heptamer sequence and a conserved nonamer separated

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

5581

Figure 1 Molecular processes modifying immunoglobulin genes. V(D)J recombination is a process by which gene segments coding for the variable region of antibody molecules are assembled. Three gene segments (V, D and J) are joined to create a heavy chain V region gene, the k and l light chain genes (not shown) are each composed of V and J segments. Heavy chain assembly occurs in two steps. In the ®rst rearrangment, a DH gene segment is rearranged to a JH segment, in the second rearrangement, a VH gene is rearrranged to a DHJH joint. In the human IgH locus on chromosome 14, there are ca. 50 functional VH, 27 DH and six JH gene segments. In class switch recombination, the expressed heavy chain constant region (CH) gene(s) (usually the originally expressed cm and cd genes) is (are) replaced by a downstream CH gene. The recombination process involves deletion of the DNA between repetitive DNA regions (switch regions, sm, sg and sa) upstream of the recombining CH genes. The speci®city of the antibody remains unaltered, but the e€ector functions of the antigen receptor are changed. Somatic hypermutation introduces point mutations and also some deletions and duplications at a very high rate speci®cally into the V region genes and ¯anking sequences (mutations are indicated by vertical lines)

by a 12 or 23 bp long spacer. RAG1 and 2 cleave precisely between the RSS and the coding sequences of the two rearranging gene segments in a concerted fashion, yielding two blunt RSS signal ends and two covalently sealed hairpin coding ends (Figure 2a). The signal ends are precisely joined, releasing the intervening DNA between the rearranging gene segments as a DNA circle. The hairpin structures of the coding ends are opened and the two ends are joined. During this reaction, nucleotides may be removed from the coding ends by exonucleolytic digestion, and non-germline nucleotides (N sequences) may be added by the lymphocyte-speci®c terminal nucleotidyltransferase (TdT), further increasing the diversity of V region genes (Figure 2a). This phase of the reaction is catalyzed, in addition to the RAG proteins, by enzymes that are also involved in non-homologous double-strand DNA break repair, including DNA ± PK (which is composed of the DNA-PK catalytic subunit and the Ku70 and Ku80 proteins), DNA ligase IV and XRCC4 (Fugmann et al., 2000). Somatic hypermutation When mature B cells are activated through binding of cognate antigen with their antigen receptors and interaction with T helper cells, they migrate into B cell follicles of secondary lymphoid organs (like lymph nodes and spleen) and establish histological structures called germinal centers (GC) (MacLennan, 1994). In the GC, the B cells vigorously proliferate and activate a somatic hypermutation mechanism which speci®cally introduces mutations into the V regions of Ig genes (Berek et al., 1991; Jacob et al., 1991; KuÈppers et al., 1993) (Figure 1). Through the acquisition of somatic mutations, antibody variants are generated, and cells expressing a B cell receptor (BCR) with increased

anity to the respective antigen are positively selected. GC B cells undergo several rounds of proliferation, mutation and selection before they ®nally di€erentiate into memory B cells or plasma cells. The process of somatic hypermutation introduces mostly nucleotide substitutions into rearranged V genes; deletions and duplications account for about 5% of the mutation events (Goossens et al., 1998; Neuberger and Milstein, 1995). Mutations are introduced at a very high rate (ca. 1073 to 1074/bp/ generation) in a region about 1 ± 2 kb downstream of the transcriptional promoter of both productively as well as non-productively rearranged V genes. It appears that hypermutation is dependent on the presence of the Ig enhancers and on transcription of the Ig genes, and that in particular transcription initiation is involved (Betz et al., 1994; Fukita et al., 1998; Peters and Storb, 1996). The mechanism of somatic hypermutation is still unknown. However, it has recently been demonstrated that hypermutation is associated with double-strand DNA breaks (Bross et al., 2000; Papavasiliou and Schatz, 2000). The most favored model suggests that an error-prone DNA polymerase is involved in the introduction of mutations in proximity of the DNA breaks. Class switch recombination A fraction of GC B cells undergoes class switch recombination and thereby changes the isotype of the expressed BCR, resulting in altered e€ector functions of the antibody (Figure 1) (Liu et al., 1996; Maizels, 1999). Through class switching, the Cm and Cd genes that are originally expressed by a naive B cell are subsequently replaced by the Cg, Ca or Ce genes. In the human, there are nine functional IgH constant region genes (1 Cm, 1 Cd, 4 Cg, 2 Ca and 1 Ce), located Oncogene

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

5582

Figure 2 V(D)J recombination and class switching (a) In V(D)J recombination, the RAG1 and RAG2 enzymes cleave the DNA at conserved recombination signal sequences (RSS) ¯anking the rearranging gene segments (here shown for a DH and JH segment). At the coding joints, hairpin structures are generated, whereas the signal joints are blunt end ligated. Upon opening of the hairpins, nucleotides may be removed from the DNA ends, or non-germline encoded nucleotides may be added before the gene segments are joined. (b) In class switch recombination, DNA is cleaved within repetitive switch regions upstream of the two recombining CH genes. The 5' part of the upstream switch region is joined to the 3' part of the downstream switch region, thereby replacing the originally expressed CH gene by a downstream CH gene. The excised DNA is circularised to a switch circle and later lost from the cell

downstream of the V, D and J gene cluster. Upstream of each CH gene, there are repetitive DNA sequences, the switch regions (Figure 2b). During class switch recombination, DNA strand breaks are introduced into Oncogene

both the switch m (sm) region and into a switch region associated with one of the downstream CH genes. The DNA fragment between the switch regions is removed from the chromosome and the switch regions are

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

joined, such that a new CH gene is placed downstream of the V region exon (Figure 2b). Although B cells carry only one functional VH gene rearrangement, class switching often takes place on both IgH alleles (Irsch et al., 1994). The putative `class switch recombinase' is still unknown, but the DNA ± PK complex, which is involved in V(D)J recombination and non-homologous double-strand DNA break repair (see above) is also needed for class switch recombination (Maizels, 1999). Chromosomal translocations in B cells of healthy mice and humans B cells that carry chromosomal translocations involving the Ig loci can be found at a low frequency in normal mice and humans. For example, B cells harbouring a bcl-2/IgH translocation have been detected at a frequency of about 1 in 104 to 106 in healthy individuals (Limpens et al., 1991; Summers et al., 2001), and c-myc/IgH translocations occur at a frequency of ca. 1 in 106 cells in normal mice and have also been detected in 2% of healthy humans (Muller et al., 1995; Roschke et al., 1997). These ®ndings are consistent with a multistep transformation process during B cell lymphomagenesis, in which chromosomal translocations involving Ig loci ± while playing a decisive role in B cell lymphomagenesis and the type of lymphoma that develops from a transformed B cell ± alone are not sucient to render a B cell malignant. In line with this view, transgenic mice overexpressing bcl-2 or c-myc in B cells develop B cell lymphomas that are monoclonal, implying that only rare transgene expressing cells will give rise to a malignant lymphoma, most likely after acquisition of additional transforming events (Adams et al., 1985; McDonnell and Korsmeyer, 1991). Predominant targeting of translocations to non-productively rearranged Ig alleles Most B cell lymphomas express a BCR, and the Ig chromosomal translocations of these tumors are predominantely located on the non-functional Ig alleles. As there is no structural reason for why a translocation event should target an Ig allele with a functional Ig gene rearrangement less frequently than an allele with a non-productive rearrangement (discussed below), this indicates that there is strong selection for B lymphoma cells to produce a functional BCR, as is the case for normal B cells (Lam et al., 1997). Thus, it appears that even B cells that acquire a chromosomal translocation still depend on BCR expression for survival. Although there are some examples of B cell non Hodgkin lymphomas where translocation or mutation events resulted in loss of Ig expression (de Jong et al., 1989), these cases may have acquired some additional transforming events which overcame the dependence on a functional BCR. A notable exception to this rule is classical Hodgkin's

lymphoma, where the tumor cells appear to derive from a population of preapoptotic GC B cells that often acquired destructive mutations preventing expression of a BCR (Kanzler et al., 1996; KuÈppers and Rajewsky, 1998).

5583

Chromosomal translocations involving V(D)J recombination Several recurrent chromosomal translocations involving Ig loci in B cell lymphomas show structural features which suggest involvement of V(D)J recombination in their generation. Paradigms of these translocations are the t(11;14) bcl-1/IgH translocation in mantle zone lymphoma and the t(14;18) bcl-2/IgH translocation in follicular lymphoma (Tsujimoto et al., 1985, 1988). Other examples include the t(1;14) involving the bcl-9 gene in an acute lymphoblastic lymphoma and the t(1;14) involving the bcl-10 gene in mucosa-associated lymphatic tissue (MALT)-type lymphoma (Willis et al., 1998, 1999; Zhang et al., 1999). While the present discussion is focussed on B cell lymphomas, it should be mentioned that V(D)J recombinase-associated translocations are not restricted to B cell lymphomas but also occur in T cell malignancies, involving the T cell receptor loci (Tycko and Sklar, 1990). Indeed, since T cells use the same enzymatic machinery for the assembly of T cell receptor genes as B cells use to assemble Ig genes, most of the following discussion on the proposed mechanisms of V(D)J recombination-associated translocations in B cell malignancies also applies to chromosomal translocations in T cell tumors. Translocation breakpoints resembling products of V(D)J recombination The proposed involvement of V(D)J recombination in bcl-2/IgH-type translocation events is mainly based on the observation that the translocation breakpoints are close to the RSS site of a JH gene segment or the 5' RSS of a DHJH joint, implying that the translocation occurred as the cell attempted a DH to JH or VH to DHJH rearrangement (JaÈger et al., 2000; Tsujimoto et al., 1985, 1988). This idea is further supported by the fact that the DH or JH gene segments at the breakpoints often show trimming of the ends and that nongermline encoded nucleotides resembling N sequences are added at the breakpoints, two features which are also typical for V(D)J joining (see above) (Bakhshi et al., 1987; Cotter et al., 1990; JaÈger et al., 2000; Tsujimoto et al., 1985, 1988). Moreover, analysis of the reciprocal translocation breakpoints of bcl-1/ and bcl-2/IgH translocations revealed that these breaks involve DH segments (or in rare cases VH segments) (Bakhshi et al., 1987; Cotter et al., 1990; JaÈger et al., 2000; Tsujimoto et al., 1988; Welzel et al., 2001). Since the germline fragment on chromosome 14 that lies between the recombining DH and JH gene segments is deleted from the chromosome, the translocation most Oncogene

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

5584

likely happened when there were DNA breaks at both a DH and a JH gene (Figure 3a). The fact that the DH genes in DH/bcl-1 and DH/bcl-2 joints show germline sequences at their 5' ends (and not VHDH joints) indeed argues against these translocations being mediated through a DNA strand break in an already rearranged VHDHJH joint. A comparison of a large number of t(11;14) bcl-1/ IgH and t(14;18) bcl-2/IgH breakpoints revealed that in the t(11;14) of mantle zone lymphomas, the usage of JH genes is comparable to normal B cells, whereas in follicular lymphomas, the most downstream JH genes are preferentially involved, as are the most 5' DH genes (JaÈger et al., 2000; Welzel et al., 2001). This led the authors to speculate that the bcl-2/IgH translocations may occur during a particular developmental stage as the cells are undergoing secondary DH ± JH recombinations. However, it is still unclear why secondary DH ± JH recombinations should be preferentially involved in the generation of bcl-2/IgH translocations. Taken together, it is likely that translocations of the bcl-2/IgH type into JH loci arise as a mistake of V(D)J recombination and therefore involve RAG-mediated DNA strand breaks within the Ig loci. As opposed to

most other types of chromosomal translocations, the translocation event likely involves three DNA breaks (at DH and JH on one chromosome and near the oncogene locus on the other chromosome). Due to the loss of the DNA fragment between the DH and JH gene segments involved in the translocation, the bcl-1/ and bcl-2/IgH translocations are, formally, not completely balanced. Models for chromosomal breakage at the oncogene loci in V(D)J-recombinase-associated translocations While the involvement of aberrant V(D)J recombination in the generation of the DNA breaks of the Ig partners in translocations involving bcl-1, bcl-2 or the other genes mentioned above appears clear-cut, the origin the DNA breaks in the partner chromosomes is less clear. Since convincing RSS motifs are seldom found near the breakpoints in the oncogene loci, it is unlikely that RSS-guided (i.e. V(D)J recombinationlike) DNA cleavage is involved. Moreover, a considerable fraction of breakpoint junctions involving bcl-1 and bcl-2 show duplication of nucleotides at the junction. This is suggestive of staggered DNA cleavage

Figure 3 Ig-gene remodeling processes involved in lymphomagenesis. Coding sequences of Ig genes are indicated in green. RSS sites are shown in (a) and (b) as black squares, and RSS-like sequences in non-Ig genes in (c) by grey quares. Non-Ig loci are shown in brown. In (g) bcl-6 is shown as an example of one of six known non-Ig genes (bcl-6, Fas, Pax-5, Rho/TTF, c-myc and Pim-1) that undergo somatic hypermutation. While bcl-6 and Fas accumulate somatic mutations in normal GC B cells, hypermutation of the four latter genes is restricted to di€use large cell lymphomas. In (h), bcl-6 is shown as an example of one of ®ve known mutating non-Ig genes (see Figure 6) that undergo translocations with breakpoints located in the mutation domain Oncogene

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

(JaÈger et al., 2000; Welzel et al., 2001), a feature that is incompatible with RAG-mediated DNA cleavage. Several models have been proposed to explain the chromosomal breaks in translocations of the bcl-2/IgH type. One model is based on the fact that chi-like sequences are found within the two predominant breakpoint regions in the bcl-2 locus, the major breakpoint region (mbr) and the minor cluster region (mcr), as well within the major translocation cluster (mtc) of the bcl-1 gene (Wyatt et al., 1992). Chi sequences were originally described as hot spots of recombination in E. coli. In support of a potential role for Chi-like sequences in the bcl-1 and bcl-2 translocations, an endogenous endonuclease activity detected in B lineage cells speci®cally cleaves within mbr (Jaeger et al., 1993). Furthermore, oligonucleotides containing the Chi-like sequence bind to a 45 kD protein (bp45) predominantly expressed in immature B cells (Jaeger et al., 1993, 1994). Based on these observations it was speculated that bp45 may bind to the Chi-like sequences and function as an accessory molecule for the endogenous nuclease(s) (Jaeger et al., 1993, 1994). Since Chi-like sequences are also present near DH and JH regions, 45 bp may also have a role in bringing together the bcl-2 and IgH loci, thereby facilitating their recombination (Jaeger et al., 1994). In two recent elegant studies it was shown that the RAG proteins, besides their function in V(D)J recombination, can also act as transposases (Agrawal et al., 1998; Hiom et al., 1998). These are enzymes that catalyze the excision of a DNA fragment ¯anked by speci®c recognition sequences and subsequent integration of this fragment into another DNA molecule. The integration is mediated through a transesteri®cation reaction and results in a duplication of several basepairs at the integration site of the recipient DNA (Agrawal et al., 1998; Hiom et al., 1998). In the transposition reaction catalyzed by RAG1 and 2, DNA fragments with RSS sites at their ends are integrated into another DNA molecule. Based on these features of the RAG proteins, Hiom et al. (1998) speculated that this activity might represent a source of oncogene translocations. They suggested two models for the generation of chromosomal translocations involving transposase activity of the V(D)J recombination machinery. The ®rst model proposes that a one-ended RAG-mediated cleavage at a DH or JH RSS creates a blunt-ended signal end which is then linked to DNA in another chromosome in a transpositional strand transfer (Hiom et al., 1998). This branched structure can be further processed to generate a hairpin end and an interchromosomal junction containing the RSS. The reciprocal translocation is completed when the two hairpin structures ± one at the coding end of the DH or JH gene segment and one in the other chromosome ± are joined. The second model proposes a two-ended insertion event (Hiom et al., 1998). Two blunt-ended RSS ends are involved in a coupled transposition reaction on the two opposite DNA strands of the target DNA on another chromosome (Figure 4). If the intermediate structure is not resolved by a conventional

transposition event (which would result in insertion of the RSS-¯anked DNA into the recipient DNA and not a translocation), but by RAG-catalyzed hairpin formation, the RSS-containing DNA fragment would be excised again, leaving behind a chromosomal break with two hairpin ends (Roth and Craig, 1998). Finally, a chromosomal translocation is generated when the hairpin ends at the IgH locus are joined crosswise to the hairpin ends generated in the other chromosome (Figure 4). The two distinct models of chromosomal translocations via transposase activity of the RAG proteins make clear predictions for the structures of the chromosomal breakpoints. While the same type of joints is generated at the DHJH or JH breaks on chromosome 14, the reciprocal joints are di€erent. In the one-end transposition model, the 5' RSS site of the

5585

Figure 4 A model for RAG-mediated translocation via doubleended transposition. An RSS-¯anked DNA fragment generated in an attempted DH to JH joining (a) can be inserted into another DNA molecule through RAG-catalyzed transesteri®cation reactions of both RSS sites (b). If the intermediate structure is not resolved in a transposition-like reaction, but by RAG-mediated cleavage (c), the RSS-¯anked DNA fragment is cut out again with short single stranded overhangs, leaving behind two hairpin ends (d). If these hairpin ends are crosswise ligated to the DH and JH coding hairpin ends (d), a chromosomal translocation is generated (e). The structure of t(11;14) (bcl-1/IgH) and t(14;18) (bcl-2/IgH) breakpoints in mantle zone lymphoma and follicular lymphoma, respectively, are largely compatible with this model. Modi®ed from Roth and Craig (1998) Oncogene

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

5586

recombining DHJH or JH gene segment is found at the reciprocal joint, whereas in the two-end model, a DH (or VH) gene segment is present at that joint, with deletion of a large DNA fragment from chromosome 14 (Figure 3a and 4). In a collection of more than 40 reciprocal breakpoints of bcl-2/IgH translocations and in a small number of reciprocal bcl-1/IgH breakpoints, only a single example shows an RSS site on the reciprocal chromosome, whereas all other breakpoints show a DH (or rarely a VH) gene segment, consistent with the second model (Bakhshi et al., 1987; Cotter et al., 1990; JaÈger et al., 2000; Meijerink et al., 1995; Tsujimoto et al., 1988). Thus, if the breaks in the bcl-2 or bcl-1 loci are indeed generated by a transposase activity of the RAG proteins, it occurs through a twoended reaction in the vast majority of cases. In summary, there are currently two models to explain DNA cleavage at the oncogene loci involved in V(D)J recombination-associated translocations into Ig loci, a model proposing two-ended transposase activity of the RAG enzymes, and a model suggesting involvement of another endonuclease cutting at chilike sequences in the breakpoint regions. Since the structures of the chromosomal breakpoints in bcl-1 and bcl-2 translocations are compatible with both scenarios, additional studies are needed to determine whether one of the proposed pathways indeed mediates the translocations. Detection of T nucleotides at bcl-1/ and bcl-2/IgH breakpoints The models discussed above o€er straighforward pathways for the generation of translocations of the bcl-2/IgH type. However, two recent studies suggest that the molecular events associated with these translocations are more complex. JaÈger et al. (2000) showed that a considerable fraction of bcl-1/ and bcl-2/ IgH breakpoints contain templated nucleotide insertions, designated T nucleotides (Welzel et al., 2001). These T nucleotides consist of short copies of DH, JH or mbr sequences surrounding the breakpoints, and occur as direct as well as inverted copies. The copies are often not directly adjacent to the donor sequences and may harbor mutations, suggesting that they are generated by an unusual process of error-prone shortpatch repair. It will be interesting to know how these T nucleotides are generated and whether they are also present in other types of chromosomal translocations. RAG-mediated translocation events in the GC? Several recent studies suggested that V(D)J recombination may not be restricted to immature B cells in the bone marrow but may also take place in GC B cells, allowing these cells to revise their antigen receptor by performing novel rearrangements of their heavy or light chain genes (Han et al., 1997; Me€re et al., 1998; Papavasiliou et al., 1997). Based on these reports, it has been speculated that (a fraction of) RAG-mediated translocations in GC B cell-derived lymphomas ± like

Oncogene

the bcl-2/IgH translocation in follicular lymphoma ± occur in the GC microenvironment (JaÈger et al., 2000; KuÈppers et al., 1999b; Willis and Dyer, 2000). At present, there is only indirect evidence supporting formation of the t(14;18) in GC B cells. JaÈger et al. (2000) noted that some bcl-2/DH/JH joints in follicular lymphoma carry point mutations in the DH genes. Since DHJH joints can undergo low levels of somatic hypermutation, and since there are usually no mutations at the bcl-2/JH breakpoints (presumably because they are too far downstream of the bcl-2 promoter), it was speculated that the cells had acquired somatic mutations in their DHJH joints in the GC prior to the translocation event. The idea that bcl-2 translocations may occasionally take place in mature B cells in the GC is also supported by an analysis of rare Ig-de®cient follicular lymphomas, many of which carry only one rearranged IgH allele, the one with a bcl-2/IgH translocation (de Jong et al., 1989). Follicular lymphoma is a malignancy of GC B cells, which presumably express a functional BCR when they are antigen-activated and driven into a GC reaction. Thus, the bcl-2/IgH translocations of Ig-de®cient follicular lymphomas may have occurred after the cell became a GC B cell, perhaps by joining of bcl-2 to a downstream JH element on the rearranged IgH allele, with inactivation or deletion of the originally expressed VHDHJH joint. Arguing against a role for RAG expression in GC B cells in the generation of chromosomal translocations, however, are recent experiments in the mouse indicating that the RAG proteins are not reexpressed in the GC and that the rearranging B cells in murine spleen are immature B cells that have immigrated from the bone marrow and yet still express RAG proteins (Gartner et al., 2000; Yu et al., 1999). Indeed, Ignegative (and hence presumably immature) B cells expressing RAG mRNA may also be found in human peripheral blood (Me€re et al., 2000). Moreover, many RAG-expressing cells are present outside GCs in human tonsils (Girschick et al., 2001; Me€re et al., 1998). Hence, V(D)J recombination-associated and thus RAG-dependent translocations may occur in secondary lymphoid organs although outside the GC. Other V(D)J recombinase-mediated recombination or transposition events While the translocations discussed above involve RAG-mediated DNA cleavage at DHJH joints or JH genes and result in juxtaposition of oncogenes to Ig loci, there is also indication that the V(D)J recombination machinery may erroneously target genes without an involvement of Ig or TCR loci. Illegitimate V(D)J recombinase activity has been implicated in large deletions a€ecting the hprt locus in T cells (Fuscoe et al., 1991), the transcription factor TAL1 (Aplan et al., 1990; Brown et al., 1990) and the MTS1 gene (Cayuela et al., 1997). While deletions in the hprt gene have little consequences and are found at low frequency in normal T cells, the deletions in the TAL1

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

and MTS1 loci are likely involved in T cell leukemogenesis. Tal1 is a protooncogene, and deletions disrupting 5' regulatory regions of the gene have been detected in ca. 20% of T cell acute lymphoblastic leukemias (T-ALL) (Aplan et al., 1990; Brown et al., 1990). MTS1 is a tumor suppressor gene, which is inactivated by deletion or disruption in most human T-ALL (Cayuela et al., 1997). In each of the three genes, heptamer-like sequences are located near the breakpoints of the deletions, and the breakpoints show `nibbling' of nucleotides as well as addition of non-templated nucleotides, features which support the idea that the deletions are mediated by V(D)J recombinase activity. While each of these examples has been identi®ed in T lineage cells, there is a priori no reason why similar deletion events should not occur also in B cells. A di€erent type of illegitimate V(D)J recombinase activity was recently described in follicular lymphoma. In two cases of the disease, the bcl-2 gene was inserted into the IgH locus, between a DH and JH gene segment (Vaandrager et al., 2000). At the deletion breakpoints 5' and 3' of the bcl-2 gene on chromosome 18, cryptic RSS sites were identi®ed. Each of the three breakpoint junctions (bcl-2 5'/bcl-2 3', bcl-2/DH and bcl-2/JH), showed N sequence insertions (Vaandrager et al., 2000). These features strongly suggest that the transposition of the bcl-2 gene from chromosome 18 into the IgH locus on chromosome 14 was RAG-mediated and happened when the cells were undergoing a DHJH recombination. Therefore, aberrant RAG-mediated recombination is likely not only involved in the generation of reciprocal chromosomal translocations, but also in deletions and transposition events involving non-antigen receptor genes (Figure 3b, c). Chromosomal translocations involving class switch recombination Chromosomal breakpoints located in IgH switch regions have been detected in a large number of di€erent translocations, including c-myc [t(8;14)] in sporadic Burkitt's lymphoma (Dalla-Favera et al., 1983; Gelmann et al., 1983; Showe et al., 1985; Taub et al., 1982), bcl-3 [t(14;19)] in B-cell chronic lymphocytic leukemia (McKeithan et al., 1997; Ohno et al., 1993), bcl-6 [t(3;14)] in di€use large cell lymphoma (Baron et al., 1993; Ye et al., 1993), Pax5 [t(9;14)] in lymphoplasmacytoid lymphoma (Iida et al., 1996), lyt-10 [t(10;14)] in di€use large cell lymphoma (Neri et al., 1991), MUC1 [t(1;14)] in an extranodal lymphoma (Gilles et al., 2000), and ®broblast growth factor receptor 3 [t(4;14)], c-maf [t(14;16)] and MUM1/IRF4 [t(6;14)] in multiple myeloma (Chesi et al., 1997, 1998a; Iida et al., 1997). Because of the location of the breakpoints in switch regions, these translocations likely happen as a mistake of class switch recombination, when DNA strand breaks are introduced into the switch regions

of the recombining CH genes (Figure 3d). Usually, the oncogenes involved in the switch recombinationassociated translocations are translocated to the der14, juxtaposing them to the IgH 3' enhancer, which results in their dysregulated expression. However, since translocation events involving switch regions result in translocation of the JH-Cm intron enhancer to the translocation partner, this could cause dysregulation of genes also on the other derivative chromosome. Indeed, two examples of such a combined dysregulation have been reported recently (Chesi et al., 1998b; Janssen et al., 2000). Since normal class switch recombination involves DNA cleavage at two switch regions, it is remarkable that a large number of class switch-associated translocations involve only one switch region. This holds true for all cases involving sm on the der14 chromosome; as this is the most 5' switch region, it follows that the reciprocal joints also involve sm (Chesi et al., 1997; Gelmann et al., 1983; Gilles et al., 2000; Iida et al., 1997; McKeithan et al., 1997; Ohno et al., 1993; Showe et al., 1985). In addition, the same phenomenon is seen in many translocations involving the sg or sa regions. This suggests that a structure in the recombination complex may favor interaction of only one of the switch regions in the complex with the other recombining chromosome during the translocation event. Alternatively, some normal human B cells may undergo a recombination event which results in deletion of parts of the sm region (Vaandrager et al., 1998; Zhang et al., 1995), as a means of `isotype stabilization' to prevent class switching in IgMexpressing B cells with a shortened sm region. As such, some of the translocation events involving sm regions on both derivative chromosomes may occur during an attempted sm shortening and not during a conventional class switch recombination event. Class switch recombination is involved not only in reciprocal translocation events, but may also lead to oncogene dysregulation by insertion events (Figure 3e). An example for this type of aberration is a case of multiple myeloma in which a DNA sequence deleted in a switch recombination was inserted on chromosome 11 near the cyclin D1 oncogene (Gabrea et al., 1999). Since this DNA fragment harbours the IgH a1 enhancer, the insertion results in overexpression of cyclin D1. Since such insertions are invisible by standard karyotype analysis, they may be more frequent in various B cell tumors than currently known. Since the detailed structure of the reciprocal translocation breakpoints has been determined only for a few chromosomal translocations into Ig switch regions, and since the molecular mechanism of class switch recombination has not yet been identi®ed, the events causing DNA breakage in the partner chromosomes of such translocations are largely unknown. However, as discussed below, in some cases, somatic hypermutation may be involved in causing DNA breaks in the oncogenes translocated to Ig switch regions.

5587

Oncogene

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

5588

The role of somatic hypermutation in chromosomal translocations and lymphomagenesis Chromosomal translocations into the Ig loci: involvement of somatic hypermutation The process of somatic hypermutation not only generates nucleotide exchanges but also a signi®cant frequency of deletions and duplications (see above). As the generation of deletions/duplications is intimately associated with DNA cleavage, it was suggested that somatic hypermutation is accompanied by DNA strand breaks (Goossens et al., 1998). This model was indeed veri®ed by two recent studies demonstrating DNA double strand breaks in the V region genes of mutating B cells (Bross et al., 2000; Papavasiliou and Schatz, 2000). Since such breaks are potentially recombinogenic, somatic hypermutation may be involved directly in the generation of chromosomal translocations in B cell lymphomas (Figure 3f). An inspection of published translocation breakpoints indeed revealed several examples that suggest their generation as byproducts of somatic hypermutation (Figure 5) (Goossens et al., 1998; KuÈppers et al., 1999a). In Burkitt's lymphoma, the breakpoint of c-myc/IgH or /IgL translocations often lies within rearranged V genes or the J introns downstream of the rearranged V genes, i.e. throughout the region that is targeted by the hypermutation machinery (Cario et al., 2000; Denny et al., 1985; Haluska et al., 1986; Hartl and Lipp, 1987; Henglein et al., 1989; Kato et al., 1991; Klobeck et al., 1987; Neri et al., 1988). Moreover, in each of these cases, somatic mutations were identi®ed within the rearranged V genes, indicating that the loci were indeed targeted by somatic hypermutation. Importantly, RSS sites were not found near the breakpoints in these cases, arguing against involvement of the RAG enzymes. Further strong support for the notion that chromosome translocations can be generated during the course of the GC reaction is based on sequence analysis of the reciprocal translocation breakpoints which uncovered somatic mutations in the 5' sequences of the V region genes (Cario et al., 2000; Henglein et al., 1989; Kato et al., 1991; Neri et al., 1988) (Figure 5). Since this part of the Ig locus is separated by the translocation event from the Ig enhancers, which are mandatory for hypermutation (Betz et al., 1994; Goyenechea and Milstein, 1996), the translocations are likely to have occurred after hypermutation of the rearranged V gene, i.e. after onset of the GC reaction. In addition to c-myc/Ig translocations in Burkitt's lymphoma, some translocations of the bcl-6 gene in di€use large cell lymphomas and the IgG Fc receptor IIb in a follicular lymphoma also show translocation breakpoints in somatically mutated V region genes (Akasaka et al., 2000; Callanan et al., 2000), consistent with an involvement of somatic hypermutation in the formation of these translocations (Figure 5). Somatic hypermutation of non-Ig genes in GC B cells It was originally assumed that somatic hypermutation is a process that acts only on rearranged Ig V genes.

Oncogene

Figure 5 Translocation breakpoints suggesting an involvement of somatic hypermutation. Several examples of translocation breakpoint within or adjacent to rearranged V genes are shown. In all examples, somatic mutations were described in the rearranged Ig genes. The breakpoints are not associated with RSS sites and are distributed throughout the region that is targeted by somatic hypermutation. A question mark indicates that the respective region has not been sequenced. Arrows indicate the position of the translocation breakpoints. Stars indicate presence of somatic mutations. The location of the enhancers is exemplarily shown only in the ®rst example for the k locus. In cell line B593, the breakpoint is in the intron between the leader and V segment exons. In cell line Ag876, a somatically mutated, non-productive VH2 gene rearrangement (VH2 ± 5/DH6 ± 13/JH4) is present on the 8q- chromosome (unpublished observation). References for the lines and cases: KOBK101 (Kato et al., 1991), JI (Klobeck et al., 1987), Ly91 (Henglein et al., 1989), Ag876 (Neri et al., 1988), 514, 881 (Akasaka et al., 2000), B593 (Callanan et al., 2000). sBL, sporadic Burkitt's lymphoma; eBL, endemic Burkitt's lymphoma; DLCL, di€use large cell lymphoma, FL, follicular lymphoma

However, the detection of frequent mutations in the bcl-6 gene on translocated as well as normal alleles in di€use large cell lymphomas (Migliazza et al., 1995) prompted an analysis of this gene in normal B cells.

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

These studies revealed that the gene is often mutated in normal human GC and memory, but not naive, B cells (Pasqualucci et al., 1998; Peng et al., 1999; Shen et al., 1998). About one third of normal human GC and memory B cells carry bcl-6 mutations, with a mutation frequency ca. 50 times lower than the corresponding mutation load in rearranged VH genes. Besides the speci®c association of these mutations with GC-derived B cells, the pattern of the mutations also supports their generation through the process of somatic hypermutation (Pasqualucci et al., 1998; Shen et al., 1998). Recently, the CD95 (Fas) gene has been identi®ed as a second example of a gene that acquires somatic mutations in the normal GC reaction (MuÈschen et al., 2000b). As in the case of bcl-6, most CD95 mutations cluster downstream of the promoter. However, some mutations were also observed in exon 9, which encodes the death domain. Since mutations in exon 9 can confer resistance to CD95-induced apoptosis, exon 9 mutations detected in some human B cell lymphomas (Gronbaek et al., 1998; Landowski et al., 1997; MuÈschen et al., 2000a; Takakuwa et al., 2001) might allow the tumor cells to escape CD95mediated elimination by tumor-speci®c e€ector T cells. Thus, the hypermutation process may also promote lymphomagenesis by targeting regulatory and coding sequences of the bcl-6 proto-oncogene and the CD95 tumor suppressor gene, resulting in either dysregulated expression (bcl-6) or loss of function (CD95).

associated with hypermutation in these genes may target them for translocation events, identifying hypermutation-associated DNA breaks in non-Ig genes as a further potential source for chromosomal breaks that may result in chromosomal translocations (Figure 3h). Thus, in bcl-6/Ig translocations involving rearranged V genes, the DNA breaks in both the Ig and the bcl-6 loci may be hypermutation-associated. Furthermore, bcl-6 translocations into Ig switch regions may be caused by a combination of hypermutation-associated DNA breaks in bcl-6 and class switching-associated breaks in the Ig switch regions. Moreover, two of the lymphoma-speci®c hypermutating loci (Rho/TTF and pim-1) loci are among the many non-Ig partners of bcl-6 translocations in di€use large cell lymphomas (Akasaka et al., 2000; Yoshida et al., 1999). Thus, somatic hypermutation may also be responsible for the origin of bcl-6 translocations that do not involve Ig loci.

5589

Translocations with an unclear association to Ig remodeling processes and complex translocations While most translocation breakpoints in the Ig loci can be attributed to either V(D)J recombination, class

Aberrant somatic hypermutation of multiple genes in diffuse large cell lymphoma In a search for other genes that are targeted by somatic hypermutation, we have recently identi®ed four genes (pax-5, c-myc, Rho/TTF and pim-1) which are frequently mutated in di€use large cell lymphomas (Pasqualucci et al., 2001). The clustering of the mutations ca 1 ± 2 kb downstream of the promoters and their pattern indicates that the mutations result from somatic hypermutation. Surprisingly, these genes are neither mutated in a number of other GC-derived B cell lymphomas nor in normal GC B cells. Thus, there appears to be a speci®c malfunction of the somatic hypermutation machinery in di€use large cell lymphoma cells or their precursors. Given the relatively high frequency of mutated genes among those analysed (4/17), it is likely that the number of loci targeted in di€use large cell lymphomas is considerably higher. Therefore, aberrant hypermutation of coding or regulatory sequences of many genes may be the basis for di€use large cell lymphoma pathogenesis and represent a novel type of genomic instability involved in tumor development. Intriguingly, each of the ®ve proto-oncogenes showing aberrant targeting of somatic hypermutation (pax-5, c-myc, Rho/TTF, pim-1 and bcl-6) are also involved in chromosomal translocations in di€use large cell lymphomas, with the breakpoint regions of these genes largely overlaping with their mutated regions (Figure 6). This suggests that the DNA breaks

Figure 6 Hypermutable regions are susceptible to chromosomal translocations. Shown are Ig genes and bcl-6 that undergo physiological somatic hypermutation, and four genes that are targets for aberrant somatic hypermutation in di€use large cell lymphoma. The line below each gene indicates the region targeted by somatic hypermutation. The arrows indicate positions of chromosomal breakpoints. For Ig, bcl-6 and c-myc, where many breakpoints have been described, only representative examples are shown. The exact position of the breakpoint involving the Rho/ TTF gene is unknown. Noncoding exons are shown by open boxes, coding exons by black boxes. For bcl-6, not all exons are shown. The colocalisation of the mutation and breakpoint regions suggests that the hypermutation-associated DNA strand breaks might increase the risk for chromosomal translocations in the same regions. Modi®ed from Pasqualucci et al. (2001) Oncogene

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

5590

switching or somatic hypermutation, some show features which do not allow a clear assignment to either of these three processes. Examples of this include translocations of either c-myc, bcl-1 or bcl-2 into the IgL loci, where the translocation breakpoints lie 2 ± 5 kb upstream of unrearranged J gene segments in the k or l locus (Adachi and Tsujimoto, 1989; Cario et al., 2000; Hollis et al., 1984; Komatsu et al., 1994). In an example involving the IgH locus, the IRTA-1 gene is translocated into the intron between the CH3 and transmembrane exons of the Ca1 gene (Hatzivassiliou et al., 2001). In none of these cases have sequences with homology to the RSS heptamer or the switch regions been identi®ed near the breakpoints, and, of course, unrearranged Ig loci or CH genes do not undergo somatic hypermutation. Perhaps the open structure of the Ig loci in B lineage cells results in a somewhat increased risk for DNA damage that may in rare instances give rise to translocation events independent of the B cell-speci®c recombination or hypermutation machinery. In addition to the simple reciprocal translocations discussed above, the Ig loci can also be involved in more complex translocation events. Several examples of these have been identi®ed in multiple myeloma, including an insertion of part of the IgH locus at a t(3;8) breakpoint involving c-myc (Shou et al., 2000). Similar insertion events of IgH sequences have also been observed in di€use large cell lymphomas with t(1;3) and t(3;6) involving the bcl-6 gene (Chaganti et al., 1998). These cases underscore the genetic instability of the Ig loci in B cells and the variety of aberrant rearrangements, in which the Ig loci can be involved. Heavy chain predominance of translocations Most translocations involving Ig loci in B cell lymphomas are targeted to the heavy chain locus. There are several aspects of B cell biology that likely account for this phenomenon. Obviously, class switch-associated translocation can only occur on heavy chain loci. Given the frequent involvement of switch regions in various types of translocations, this is likely one of the main factors to explain the heavy chain predominance in translocation targeting. The heavy chain locus may also be at a higher risk to acquire somatic hypermutationassociated translocations, because the average mutation load in VH region genes is nearly twice as high as the mutation load in light chain genes (Klein et al., 1998b), suggesting a higher activity of the hypermutation machinery at the IgH locus. A low frequency of translocations at the Igk locus is perhaps also due to the frequent inactivation of non-productive VkJk joints by rearrangements of the kappa deleting element (Siminovitch et al., 1985). These recombinations, which silence the respective k loci by deletion of the k enhancers, inactivate both k alleles in most l-expressing B cells, as well as the non-expressed k allele in a fraction of k-expressing B cells (Klein et al., 1998a). Besides these factors relating to features of the Ig loci, a predominance

Oncogene

of IgH-associated translocations involving V(D)J recombinase may also be caused by di€erent accessibility of the partner loci at di€erent stages of B cell development. For example, the bcl-2 gene is expressed at a higher level in pro B cells (which undergo heavy chain gene rearrangements) than in pre B cells (which undergo light chain gene rearrangements) (Li et al., 1993). If transcription of the bcl-2 gene renders the bcl-2 locus more accessible for translocation events, this could explain the preferential targeting of bcl-2 translocations to the heavy chain locus. Occurrence of chromosomal translocations as byproducts or speci®c defects of the Ig-remodeling processes? The description of chromosomal translocations as `mistakes' of Ig-remodeling processes leaves it open as to whether these events occur as rare byproducts of the normal processes, or represent speci®c defects in the targeting of V(D)J recombination, class switching or somatic hypermutation. The ®nding that bcl-2/IgH translocations can be detected in a large fraction of normal individuals (see above) argues against an involvement of somatic or inherited genetic defects in the mistargeting of V(D)J recombination. In line with the view that V(D)J recombination may have some low level intrinsic in®delity in targeting is the observation that antigen-receptor trans-rearrangements (analysed for T cell receptor b and g cross-rearrangements) can be regularly detected in healthy humans (Kirsch and Lista, 1997; Tycko et al., 1989). Likewise, the ®nding that about 30% of human GC and memory B cells carry mutations of the bcl-6 gene (Pasqualucci et al., 1998; Shen et al., 1998), argues against a role for a somatic or germline defect in targeting of the bcl-6 gene for chromosomal translocation by aberrant somatic hypermutation (Figure 7).

Figure 7 Targets of somatic hypermutation. Mutations in bcl-6 and Fas (CD95) are regarded as byproducts of physiological somatic hypermutation as they occur in 10 ± 30% of normal GC B cells. In di€use large cell lymphomas, four genes have been identi®ed that undergo aberrant somatic hypermutation. Since these genes were identi®ed among 17 genes tested, the true number of genes targeted by aberrant hypermutation is likely higher. Whereas mutations at the 5' end of the Fas gene ful®l several criteria of somatic hypermutation, the association between somatic hypermutation and the mutations in the death domain of the Fas gene is less certain. DLCL: di€use large cell lymphoma

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

Table 1 Genetic aberrations in B cell tumors generated as mistakes of B cell-speci®c Ig-modifying processes Enzyme machinery

Type of molecular process

Examples

References

V(D)J recombination

(i) translocation of oncogene to Ig (D)J locus;

± bcl-1/lgH in MZL ± bcl-2/lgH in FL ± a(TAL 1 and MTS1 in T cell leukemias) ± bcl-2 in FL ± bcl-3/lgH in B-CLL ± bcl-6/lgH in DLCL ± c-myc/lgH in BL ± FGFR/lgH in MM ± insertion into cylin D1 in MM ± c-myc/lgH in BL ± bcl-6/lgL in DLCL ± bcl-6, CD95/Fas in various NHL ± Pax-5, c-myc, Pim-1, Rho/TTF in DLCL ± bcl-6, Pax-5, Pim-1, Rho/TTF

(Bakhshi et al., 1987; Cotter et al., 1990; JaÈger et al., 2000; Tsujimoto et al., 1988) (Aplan et al., 1990; Brown et al., 1990; Cayuela et al., 1997) (Vaandrager et al., 2000) (Chesi et al., 1997; Gelmann et al., 1983; Ohno et al., 1993; Showe et al., 1985; Baron et al., 1993; Ye et al., 1993)

Ig class switching

Somatic hypermutation

(ii) deletions in oncogenes or tumor suppressor genes; (iii) transposition of genes into the Ig locus (i) translocation of oncogene to IgH switch regions; (ii) insertion of excised switch sequences into other loci (i) translocation of oncogene into rearranged V genes; (ii) somatic mutations in non-lg genes;

(iii) translocation of mutating oncogene

5591

(Gabrea et al., 1999) see Figure 4 (Migliazza et al., 1995; MuÈschen et al., 2000b; Pasqualucci et al., 1998; Shen et al., 1998; Pasqualucci et al., 2001) Pasqualucci et al., 2001

a

So far no example in B cell lymphomas identi®ed. Abbreviations: MZL: mantle zone lymphoma, FL: follicular lymphoma, B-CLL: B cell chronic lymphocytic leukemia, MM: multiple myeloma, BL: Burkitt's lymphoma, DLCL: di€use large cell lymphoma, NHL: Non Hodgkin lymphoma

The situation may be di€erent in case of genes that are speci®cally mutated in di€use large cell lymphomas (e.g., pim-1, Rho/TTF, c-myc and Pax-5). Since these genes are not mutated at signi®cant frequency in normal B cells and other types of lymphomas, their occurrence may be due to a genetic alteration in di€use large cell lymphomas or their precursors (Figure 7). This would be reminescent of somatically acquired defects in DNA mismatch repair genes which result in accumulation of mutations in many genes and an increased risk of colorectal carcinogenesis (Kinzler and Vogelstein, 1996). The role of the GC in B cell lymphomagenesis The frequent involvement of somatic hypermutation and class switch recombination in the generation of chromosomal translocations points to the GC as the microenvironment in which many B cell lymphomas develop (KuÈppers et al., 1999b). But how good is the evidence that these molecular processes, and hence their aberrant activities, are restricted to the GC? It is known that class switch recombination can also take place in T cell-independent immune responses that lack GCs and are not associated with somatic hypermutation (Garcia de Vinuesa et al., 1999; Maizels and Bothwell, 1985). Moreover, indication that somatic hypermutation may also occur outside the GC is provided by the detection of somatically mutated B cells in a gene knock-out mouse model that lacks GCs and in patients with germline mutations of the CD40ligand gene and absence of GCs in lymph nodes (Matsumoto et al., 1996; Weller et al., 2001). However, the consistent presence of a high load of somatic mutations in the rearranged V genes of lymphomas

that frequently have switch-associated translocations (like Burkitt's lymphoma, di€use large cell lymphoma and multiple myeloma) (Klein et al., 1998a) argues against a GC-independent development of these lymphomas in the course of T-independent immune responses. Regarding the GC speci®city of somatic hypermutation, the presence of somatic Ig gene mutations in seemingly GC-de®cient animals and humans may re¯ect the presence of GC (-like) structures in organs other than the lymph nodes and spleen. Intriguingly, both the CD19 and the CD40 knock-out mice, which were originally reported as lacking GCs, show GC (-like) structures in the Peyer's patches of the intestine, and CD40-ligand de®cient mice develop GCs under particular activation conditions (Chirmule et al., 2000; Gardby and Lycke, 2000; G Cattoretti, personal communication). Hence, it is still uncertain whether somatic hypermutation can occur outside the GC microenvironment. Finally, the derivation of many B cell lymphomas from GC B cells is also supported by the morphology and immunophenotype of the malignant cells and the histological appearance of the tumor. For example, Burkitt's lymphoma cells express the marker CD77, which is speci®c of GC centroblasts. Indeed, the presence of structures that resemble GCs in follicular lymphoma and the morphological resemblance of the tumor cells to centroblasts and centrocytes argues that this tumor type represents a GC B cell lymphoma par excellance (Fyfe et al., 1987; Isaacson, 1996). Although it cannot be excluded that some rare chromosomal translocations involving class switching or somatic hypermutation may take place outside the GC microenvironment, there is little doubt that these translocations usually occur in the GC. Hence, besides the vigorous proliferation of GC B cells, which may by Oncogene

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

5592

itself increase the risk for the acquisition of genetic damage, the remodeling of Ig genes in the GC through somatic hypermutation and class switching likely plays a decisive role in the development of many B cell lymphomas in this microenvironment (KuÈppers et al., 1999b). In many instances, aberrant activity of these two processes results in the generation of balanced chromosomal translocations. Sometimes, these two processes may even cooperate. For example, as discussed above, translocations of bcl-6 into IgH switch regions may involve hypermutation-associated DNA breaks in the bcl-6 gene and class switch recombination-associated DNA cleavage in the switch regions. Besides the generation of translocations, hypermuta-

tion and class switching may also contribute to lymphomagenesis in the GC by causing other types of mutations (Table 1, Figure 3). A particularly intriguing example of this is the aberrant targeting of non-Ig genes by the somatic hypermutation process, which might be a key event in the development of di€use large cell lymphoma. Acknowledgments R KuÈppers is supported by the Deutsche Forschungsgemeinschaft through SFB502 and the Heisenberg programme. We thank Richard Baer for critically reading the manuscript.

References Adachi M and Tsujimoto Y. (1989). Oncogene, 4, 1073 ± 1075. Adams JM, Harris AW, Pinkert CA, Corcoran LM, Alexander WS, Cory S, Palmiter RD and Brinster RL. (1985). Nature, 318, 533 ± 538. Agrawal A, Eastman QM and Schatz DG. (1998). Nature, 394, 744 ± 751. Akasaka H, Akasaka T, Kurata M, Ueda C, Shimizu A, Uchiyama T and Ohno H. (2000). Cancer Res., 60, 2335 ± 2341. Aplan PD, Lombardi DP, Ginsberg AM, Cossman J, Bertness VL and Kirsch IR. (1990). Science, 250, 1426 ± 1429. Bakhshi A, Wright JJ, Graninger W, Seto M, Owens J, Cossman J, Jensen JP, Goldman P and Korsmeyer SJ. (1987). Proc. Natl. Acad. Sci. USA, 84, 2396 ± 2400. Baron BW, Nucifora G, McCabe N, Espinosa R, Le Beau MM and McKeithan TW. (1993). Proc. Natl. Acad. Sci. USA, 90, 5262 ± 5266. Berek C, Berger A and Apel M. (1991). Cell., 67, 1121 ± 1129. Betz AG, Milstein C, Gonzalez-Fernandez A, Pannell R, Larson T and Neuberger MS. (1994). Cell., 77, 239 ± 248. Bross L, Fukita Y, McBlane F, Demolliere C, Rajewsky K and Jacobs H. (2000). Immunity, 13, 589 ± 597. Brown L, Cheng JT, Chen Q, Siciliano MJ, Crist W, Buchanan G and Baer R. (1990). EMBO J., 9, 3343 ± 3351. Callanan MB, Le Baccon P, Mossuz P, Duley S, Bastard C, Hamoudi R, Dyer MJ, Klobeck G, Rimokh R, Sotto JJ and Leroux D. (2000). Proc. Natl. Acad. Sci. USA, 97, 309 ± 314. Cario G, Stadt UZ, Reiter A, Welte K and Sykora KW. (2000). Br. J. Haematol., 110, 537 ± 546. Cayuela JM, Gardie B and Sigaux F. (1997). Blood, 90, 3720 ± 3726. Chaganti SR, Rao PH, Chen W, Dyomin V, Jhanwar SC, Parsa NZ, Dalla-Favera R and Chaganti RS. (1998). Genes Chrom. Cancer, 23, 328 ± 336. Chesi M, Bergsagel PL, Shonukan OO, Martelli ML, Brents LA, Chen T, Schrock E, Ried T and Kuehl WM. (1998a). Blood, 91, 4457 ± 4463. Chesi M, Nardini E, Brents LA, Schrock E, Ried T, Kuehl WM and Bergsagel PL. (1997). Nat. Genet., 16, 260 ± 264. Chesi M, Nardini E, Lim RS, Smith KD, Kuehl WM and Bergsagel PL. (1998b). Blood, 92, 3025 ± 3034. Chirmule N, Tazelaar J and Wilson JM. (2000). J. Immunol., 164, 248 ± 255. Cotter F, Price C, Zucca E and Young BD. (1990). Blood, 76, 131 ± 135. Oncogene

Dalla-Favera R and Gaidano G. (2001). Cancer. DeVita VT, Hellman S and Rosenberg SA. (eds). Lippincott Williams and Wilkins: Philadelphia, USA, pp. 2215 ± 2235. Dalla-Favera R, Martinotti S, Gallo RC, Erikson J and Croce CM. (1983). Science, 219, 963 ± 967. de Jong D, Voetdijk BM, Van Ommen GJ, Kluin-Nelemans JC, Beverstock GC and Kluin PM. (1989). J. Exp. Med., 169, 613 ± 624. Denny CT, Hollis GF, Magrath IT and Kirsch IR. (1985). Mol. Cell. Biol., 5, 3199 ± 3207. Fugmann SD, Lee AI, Shockett PE, Villey IJ and Schatz DG. (2000). Ann. Rev. Immunol., 18, 495 ± 527. Fukita Y, Jacobs H and Rajewsky K. (1998). Immunity, 9, 105 ± 114. Fuscoe JC, Zimmerman LJ, Lippert MJ, Nicklas JA, O'Neill JP and Albertini RJ. (1991). Cancer Res., 51, 6001 ± 6005. Fyfe G, Cebra-Thomas JA, Mustain E, Davie JM, Alley CD and Nahm MH. (1987). J. Immunol., 139, 2187 ± 2194. Gabrea A, Bergsagel PL, Chesi M, Shou Y and Kuehl WM. (1999). Mol. Cell., 3, 119 ± 123. Garcia de Vinuesa C, O'Leary P, Sze DM, Toellner KM and MacLennan IC. (1999). Eur. J. Immunol., 29, 1314 ± 1323. Gardby E and Lycke NY. (2000). Eur. J. Immunol., 30, 1861 ± 1871. Gartner F, Alt FW, Monroe RJ and Seidl KJ. (2000). J. Exp. Med., 192, 1745 ± 1754. Gelmann EP, Psallidopoulos MC, Papas TS and DallaFavera R. (1983). Nature, 306, 799 ± 803. Gilles F, Goy A, Remache Y, Shue P and Zelenetz AD. (2000). Blood, 95, 2930 ± 2936. Girschick HJ, Grammer AC, Nanki T, Mayo M and Lipsky PE. (2001). J. Immunol., 166, 377 ± 386. Goossens T, Klein U and KuÈppers R. (1998). Proc. Natl. Acad. Sci. USA, 95, 2463 ± 2468. Goyenechea B and Milstein C. (1996). Proc. Natl. Acad. Sci. USA, 93, 13979 ± 13984. Gronbaek K, Straten PT, Ralfkiaer E, Ahrenkiel V, Andersen MK, Hansen NE, Zeuthen J, Hou-Jensen K and Guldberg P. (1998). Blood, 92, 3018 ± 3024. Haluska FG, Finver S, Tsujimoto Y and Croce CM. (1986). Nature, 324, 158 ± 161. Han S, Dillon SR, Zheng B, Shimoda M, Schlissel MS and Kelsoe G. (1997). Science, 278, 301 ± 305. Harrison CJ. (2000). Baillieres Best Pract. Res. Clin. Haematol., 13, 427 ± 439. Hartl P and Lipp M. (1987). Mol. Cell. Biol., 7, 2037 ± 2045.

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

Hatzivassiliou G, Miller I, Takizawa J, Palanisamy N, Pulivarthi R, Iida S, Tagawa S, Russo J, Neri A, Cattoretti G, Clynes R, Mendelsohn C, Chaganti RSK and DallaFavera R. (2001). Immunity, 14, 277 ± 289. Henglein B, Synovzik H, Groitl P, Bornkamm GW, Hartl P and Lipp M. (1989). Mol. Cell. Biol., 9, 2105 ± 2113. Hiom K, Melek M and Gellert M. (1998). Cell, 94, 463 ± 470. Hollis GF, Mitchell KF, Battey J, Potter H, Taub R, Lenoir GM and Leder P. (1984). Nature, 307, 752 ± 755. Iida S, Rao PH, Butler M, Corradini P, Boccadoro M, Klein B, Chaganti RS and Dalla-Favera R. (1997). Nat. Genet., 17, 226 ± 230. Iida S, Rao PH, Nallasivam P, Hibshoosh H, Butler M, Louie DC, Dyomin V, Ohno H, Chaganti RS and DallaFavera R. (1996). Blood, 88, 4110 ± 4117. Irsch J, Irlenbusch S, Radl J, Burrows PD, Cooper MD and Radbruch AH. (1994). Proc. Natl. Acad. Sci. USA, 91, 1323 ± 1327. Isaacson PG. (1996). Histopathology, 28, 487 ± 495. Jacob J, Kelsoe G, Rajewsky K and Weiss U. (1991). Nature, 354, 389 ± 392. Jaeger U, Karth GD, Knapp S, Friedl J, Laczika K and Kusec R. (1994). Leuk. Lymphoma, 14, 197 ± 202. Jaeger U, Purtscher B, Karth GD, Knapp S, Mannhalter C and Lechner K. (1993). Blood, 81, 1833 ± 1840. JaÈger U, Bocskor S, Le T, Mitterbauer G, Bolz I, Chott A, Kneba M, Mannhalter C and Nadel B. (2000). Blood, 95, 3520 ± 3529. Janssen JW, Vaandrager JW, Heuser T, Jauch A, Kluin PM, Geelen E, Bergsagel PL, Kuehl WM, Drexler HG, Otsuki T, Bartram CR and Schuuring E. (2000). Blood, 95, 2691 ± 2698. Kanzler H, KuÈppers R, Hansmann ML and Rajewsky K. (1996). J. Exp. Med., 184, 1495 ± 1505. Kato S, Tachibana K, Takayama N, Kataoka H, Yoshida MC and Takano T. (1991). Gene, 97, 239 ± 244. Kinzler KW and Vogelstein B. (1996). Cell, 87, 159 ± 170. Kirsch IR and Lista F. (1997). Semin. Immunol., 9, 207 ± 215. Klein U, Goossens T, Fischer M, Kanzler H, Braeuninger A, Rajewsky K and KuÈppers R. (1998a). Immunol. Rev., 162, 261 ± 280. Klein U, Rajewsky K and KuÈppers R. (1998b). J. Exp. Med., 188, 1679 ± 1689. Klobeck HG, Combriato G and Zachau HG. (1987). Nucleic Acids Res., 15, 4877 ± 4888. Komatsu H, Iida S, Yamamoto K, Mikuni C, Nitta M, Takahashi T, Ueda R and Seto M. (1994). Blood, 84, 1226 ± 1231. KuÈppers R, Goossens T and Klein U. (1999a). Curr. Top. Microbiol. Immunol., 246, 193 ± 198. KuÈppers R, Klein U, Hansmann ML and Rajewsky K. (1999b). N. Engl. J. Med., 341, 1520 ± 1529. KuÈppers R and Rajewsky K. (1998). Ann. Rev. Immunol., 16, 471 ± 493. KuÈppers R, Zhao M, Hansmann ML and Rajewsky K. (1993). EMBO J., 12, 4955 ± 4967. Lam KP, KuÈhn R and Rajewsky K. (1997). Cell, 90, 1073 ± 1083. Landowski TH, Qu N, Buyuksal I, Painter JS and Dalton WS. (1997). Blood, 90, 4266 ± 4270. Li YS, Hayakawa K and Hardy RR. (1993). J. Exp. Med., 178, 951 ± 960. Limpens J, de Jong D, van Krieken JH, Price CG, Young BD, van Ommen GJ and Kluin PM. (1991). Oncogene, 6, 2271 ± 2276.

Liu YJ, Malisan F, de Bouteiller O, Guret C, Lebecque S, Banchereau J, Mills FC, Max EE and Martinez-Valdez H. (1996). Immunity, 4, 241 ± 250. MacLennan IC. (1994). Ann. Rev. Immunol., 12, 117 ± 139. Maizels N. (1999). Am. J. Hum. Genet., 64, 1270 ± 1275. Maizels N and Bothwell A. (1985). Cell, 43, 715 ± 720. Matsumoto M, Mariathasan S, Nahm MH, Baranyay F, Peschon JJ and Chaplin DD. (1996). Science, 271, 1289 ± 1291. McDonnell TJ and Korsmeyer SJ. (1991). Nature, 349, 254 ± 256. McKeithan TW, Takimoto GS, Ohno H, Bjorling VS, Morgan R, Hecht BK, Dube I, Sandberg AA and Rowley JD. (1997). Genes Chrom. Cancer, 20, 64 ± 72. Me€re E, Davis E, Schi€ C, Cunningham-Rundles C, Ivashkiv LB, Staudt LM, Young JW and Nussenzweig MC. (2000). Nature Immunol., 1, 207 ± 213. Me€re E, Papavasiliou F, Cohen P, de Bouteiller O, Bell D, Karasuyama H, Schi€ C, Banchereau J, Liu YJ and Nussenzweig MC. (1998). J. Exp. Med., 188, 765 ± 772. Meijerink JP, Raemaekers JM and Mensink EJ. (1995). Br. J. Haematol., 91, 630 ± 639. Migliazza A, Martinotti S, Chen W, Fusco C, Ye BH, Knowles DM, Ot K, Chaganti RS and Dalla-Favera R. (1995). Proc. Natl. Acad. Sci. USA, 92, 12520 ± 12524. Muller JR, Janz S, Goedert JJ, Potter M and Rabkin CS. (1995). Proc. Natl. Acad. Sci. USA, 92, 6577 ± 6581. MuÈschen M, Re D, BraÈuninger A, Wolf J, Hansmann ML, Diehl V, KuÈppers R and Rajewsky K. (2000a). Cancer Res., 60, 5640 ± 5643. MuÈschen M, Re D, Jungnickel B, Diehl V, Rajewsky K and KuÈppers R. (2000b). J. Exp. Med., 192, 1833 ± 1840. Neri A, Barriga F, Knowles DM, Magrath IT and DallaFavera R. (1988). Proc. Natl. Acad. Sci. USA, 85, 2748 ± 2752. Neri A, Chang CC, Lombardi L, Salina M, Corradini P, Maiolo AT, Chaganti RS and Dalla-Favera R. (1991). Cell, 67, 1075 ± 1087. Neuberger MS and Milstein C. (1995). Curr. Opin. Immunol., 7, 248 ± 254. Ohno H, Doi S, Yabumoto K, Fukuhara S and McKeithan TW. (1993). Leukemia, 7, 2057 ± 2063. Papavasiliou F, Casellas R, Suh H, Qin XF, Besmer E, Pelanda R, Nemazee D, Rajewsky K and Nussenzweig MC. (1997). Science, 278, 298 ± 301. Papavasiliou FN and Schatz DG. (2000). Nature, 408, 216 ± 221. Pasqualucci L, Migliazza A, Fracchiolla N, William C, Neri A, Baldini L, Chaganti RSK, Klein U, KuÈppers R, Rajewsky K and Dalla-Favera R. (1998). Proc. Natl. Acad. Sci. USA, 95, 11816 ± 11821. Pasqualucci L, Neumeister P, Gossers T, Nanjangud G, Chaganti RSK, KuÈppers R and Dalla-Favera R. (2001). Nature, in press. Peng HZ, Du MQ, Koulis A, Aiello A, Dogan A, Pan LX and Isaacson PG. (1999). Blood, 93, 2167 ± 2172. Peters A and Storb U. (1996). Immunity, 4, 57 ± 65. Rajewsky K. (1996). Nature, 381, 751 ± 758. Roschke V, Kopantzev E, Dertzbaugh M and Rudiko€ S. (1997). Oncogene, 14, 3011 ± 3016. Roth DB and Craig NL. (1998). Cell, 94, 411 ± 414. Rowley JD. (1999). Semin. Hematol., 36, 59 ± 72. Shen HM, Peters A, Baron B, Zhu X and Storb U. (1998). Science, 280, 1750 ± 1752.

5593

Oncogene

Mechanisms of chromosomal translocations in B cell lymphomas R KuÈppers and R Dalla-Favera

5594

Oncogene

Shou Y, Martelli ML, Gabrea A, Qi Y, Brents LA, Roschke A, Dewald G, Kirsch IR, Bergsagel PL and Kuehl WM. (2000). Proc. Natl. Acad. Sci. USA, 97, 228 ± 233. Showe LC, Ballantine M, Nishikura K, Erikson J, Kaji H and Croce CM. (1985). Mol. Cell. Biol., 5, 501 ± 509. Siminovitch KA, Bakhshi A, Goldman P and Korsmeyer SJ. (1985). Nature, 316, 260 ± 262. Summers KE, Go€ LK, Wilson AG, Gupta RK, Lister TA and Fitzgibbon J. (2001). J. Clin. Oncol., 19, 420 ± 424. Takakuwa T, Dong Z, Takayama H, Matsuzuka F, Nagata S and Aozasa K. (2001). Cancer Res., 61, 1382 ± 1385. Taub R, Kirsch I, Morton C, Lenoir G, Swan D, Tronick S, Aaronson S and Leder P. (1982). Proc. Natl. Acad. Sci. USA, 79, 7837 ± 7841. Tonegawa S. (1983). Nature, 302, 575 ± 581. Tsujimoto Y, Gorham J, Cossman J, Ja€e E and Croce CM. (1985). Science, 229, 1390 ± 1393. Tsujimoto Y, Louie E, Bashir MM and Croce CM. (1988). Oncogene, 2, 347 ± 351. Tycko B, Palmer JD and Sklar J. (1989). Science, 245, 1242 ± 1246. Tycko B and Sklar J. (1990). Cancer Cells, 2, 1 ± 8. Vaandrager JW, Schuuring E, Kluin-Nelemans HC, Dyer MJS, Raap AK and Kluin PM. (1998). Blood, 92, 2871 ± 2878. Vaandrager JW, Schuuring E, Philippo K and Kluin PM. (2000). Blood, 96, 1947 ± 1952. Weller S, Faili A, Garcia C, Braun MC, Le Deist FF, de Saint Basile GG, Hermine O, Fischer A, Reynaud CA and Weill JC. (2001). Proc. Natl. Acad. Sci. USA, 98, 1166 ± 1170.

Welzel N, Le T, Marculescu R, Mitterbauer G, Chott A, Pott C, Kneba M, Du M-Q, Kusec R, Drach J, Raderer M, Mannhalter C, Lechner K, Nadel B and Jaeger U. (2001). Cancer Res., 61, 1629 ± 1636. Willis TG and Dyer MJ. (2000). Blood, 96, 808 ± 822. Willis TG, Jadayel DM, Du MQ, Peng H, Perry AR, AbdulRauf M, Price H, Karran L, Majekodunmi O, Crook T, Hamoudi R, Isaacson PG and Dyer MJS. (1999). Cell, 96, 35 ± 45. Willis TG, Zalcberg IR, Coignet LJ, Wlodarska I, Stul M, Jadayel DM, Bastard C, Treleaven JG, Catovsky D, Silva ML and Dyer MJ. (1998). Blood, 91, 1873 ± 1881. Wyatt RT, Rudders RA, Zelenetz A, Delellis RA and Krontiris TG. (1992). J. Exp. Med., 175, 1575 ± 1588. Ye BH, Rao PH, Chaganti RS and Dalla-Favera R. (1993). Cancer Res., 53, 2732 ± 2735. Yoshida S, Kaneita Y, Aoki Y, Seto M, Mori S and Moriyama M. (1999). Oncogene, 18, 7994 ± 7999. Yu W, Nagaoka H, Jankovic M, Misulovin Z, Suh H, Rolink A, Melchers F, Me€re E and Nussenzweig MC. (1999). Nature, 400, 682 ± 687. Zhang K, Cheah HK and Saxon A. (1995). J. Immunol., 154, 2237 ± 2247. Zhang Q, Siebert R, Yan M, Hinzmann B, Cui X, Xue L, Rakestraw KM, Naeve CW, Beckmann G, Weisenburger DD, Sanger WG, Nowotny H, Vesely M, Callet-Bauchu E, Salles G, Dixit VM, Rosenthal A, Schlegelberger B and Morris SW. (1999). Nat. Genet., 22, 63 ± 68.