Chromosome translocations between oncogenes and the region spanning the immunoglobulin (Ig) heavy chain (IgH) variable (V), diversity (D), and joining (J) gene segments (Ig V-JH region) are found in several mature B cell lymphomas in humans and mice. The breakpoints are frequently adjacent to the recombination signal sequences targeted by recombination activating genes 1 and 2 during antigen receptor assembly in pre–B cells, suggesting that these translocations might be the result of aberrant V(D)J recombination. However, in mature B cells undergoing activation-induced cytidine deaminase (AID)-dependent somatic hypermutation (SHM), duplications or deletions that would necessitate a double-strand break make up 6% of all the Ig V-JH region–associated somatic mutations. Furthermore, DNA breaks can be detected at this locus in B cells undergoing SHM. To determine whether SHM might induce c-myc to Ig V-JH translocations, we searched for such events in both interleukin (IL) 6 transgenic (IL-6 tg) and AID−/− IL-6 tg mice. Here, we report that AID is required for c-myc to Ig V-JH translocations induced by IL-6.
Chromosome translocations are products of unresolved double-strand DNA breaks (1–4) and therefore occur frequently at Ig genes because these loci undergo programmed DNA damage during antigen receptor gene diversification. In developing T and B cells, RAG1 and RAG2 proteins produce double-strand breaks at recombination signal sequences (RSS) found adjacent to variable, diversity, and joining (V(D)J) gene segments (5, 6). These RSS are found in close proximity or within translocation breakpoints in numerous B cell lymphomas (for review see references 7 and 8).
In mature B cells, expression of activation-induced cytidine deaminase (AID) (9–11) leads to deamination of cytidine residues in Ig V-JH and Ig switch regions resulting in U:G mismatches that are processed to produce somatic hypermutations (SHMs) and initiate class switch recombination (for review see references 12–14). Double-strand breaks are obligate intermediates in the class switch reaction, and translocations involving switch regions are frequently found in sporadic Burkitt's lymphoma, diffuse large B cell lymphoma, and multiple myeloma, suggesting that AID is responsible for the lesions that lead to such translocations (15–22). Consistent with this idea, AID-induced breaks in the switch region activate the cellular DNA damage response machinery (23), and AID is essential for c-myc translocations to the Ig switch region in IL-6 transgenic (tg) (24, 25). In addition, AID appears to target several oncogenes that are frequently mutated and often translocated to antibody genes in mature B cell malignancies (26–30). In agreement with these observations, deregulated expression of AID is associated with malignancy (31–34).
Double-strand breaks are not obligate intermediates in SHM of the Ig V-JH (35). Nevertheless, duplications or deletions that would necessitate a double-strand break make up 6% of all the Ig V-JH region–associated somatic mutations, and DNA breaks can be detected in this region in B cells undergoing mutation (36–42). In addition, endemic Burkitt's lymphoma, multiple myeloma, follicular lymphoma, and diffuse large B cell lymphoma contain mutated V genes as well as translocations to the Ig V-JH or Ig V-JL regions (for review see references 7, 8, and 43). This suggests that translocations in these malignancies may have occurred in mature B cells that express AID but not RAG1/2 (44), and that some Ig V-JH region–associated translocations are byproducts of lesions induced by AID during hypermutation. To ascertain the extent to which AID contributes to genomic instability at the IgH locus, we tested if AID is required for translocations between c-myc and the Ig V-JH region in IL-6 tg mice.
IL-6 tg mice develop hyperplastic lymph nodes that contain large numbers of class-switched plasmacytes, a portion of which express GL7 and CD138 (24, 45, 46). Plasmacytosis is believed to develop in these mice because IL-6 attenuates apoptosis and promotes proliferation and differentiation of late-stage B cells, allowing for the accumulation of translocations between IgH and c-myc (45). Although a majority of the c-myc translocation breakpoints are at the Ig switch region, a small fraction occur at the V-JH region (46) and therefore resemble the translocations found in endemic Burkitt's lymphoma (for review see references 7 and 8). To determine whether AID is required for translocations between the Ig V-JH region and c-myc, we generated AID-deficient IL-6 tg mice (AID−/−IL-6 tg) by breeding (24). AID−/−IL-6 tg mice developed lymph node hyperplasia and plasmacytosis with a slightly delayed onset compared with IL-6 tg mice, and there was no detectable class switching in the AID−/−IL-6 tg mice (Fig. 1, A and B) (24).
To document translocations between the Ig V-JH region and c-myc, we developed PCR assays for these events and examined cells from hyperplastic lymph nodes from IL-6 tg and AID−/− IL-6 tg mice (Fig. 2 A). Southern blotting and DNA sequencing were used to verify candidate translocations (Figs. 2 and Figs.3). Assaying four different lymph node pools per mouse for derivative 12 and derivative 15 translocations (Fig. 2 A), we identified 37 unique translocations in 14 IL-6 tg mice, but none in 12 AID−/−IL-6 tg mice (P = 0.0025) (Fig. 2 C). Derivative 12 and 15 translocations were similar in both the number of translocations identified (21 and 16, respectively) as well as in their breakpoint distribution along the chromosome (Fig. 3), providing strong evidence that variable region translocations are reciprocal. As expected, c-myc to Ig switch region translocations were far more frequent (Fig. 2 B) (24, 46). We conclude that translocations between c-myc and the Ig V-JH region are AID dependent in IL-6 tg mice.
To gain further insight into the etiology of the identified translocations, we analyzed and mapped the breakpoints. Sequence analysis revealed that the majority of the breakpoints were in or around JH segments, as commonly seen for oncogenic translocations in B cell non-Hodgkin's (Fig. 3) (for review see references 7 and 8). Junction sequences resembled those previously characterized for c-myc to Ig switch region translocations from IL-6 tg mice (24) in that they involved either blunt ends or 1–3 nucleotide stretches of micro-homology or nucleotide insertions (Tables I and II), implying that nonhomologous end joining resolves these breaks. Finally, the translocation breakpoints in c-myc were all in the first intron, which differs from those to the Ig switch region in IL-6 tg mice, where a majority of the breakpoints are in the first exon of c-myc (Fig. 3) (24).
Although there was no correlation between the position of the breakpoints and the RGYW motifs, which are the preferred targets of AID, the translocated Ig V-JH region was somatically mutated at a frequency of ∼0.6 × 10−3 mutations per basepair (Fig. 4). This rate of mutation is similar to that reported for the Ig V-JH region in B cells undergoing hypermutation (47). Derivative 12 and 15 IgH sequences had a similar frequency of hypermutation, suggesting that SHM must have occurred before or during translocation. Interestingly, the overall position of the mutations mirrored the positions of the translocation breakpoints (compare Figs. 3 and Figs.4), supporting the idea that regions prone to SHM are susceptible to translocations. Excluding nucleotide insertions at the breakpoint, the rate of mutation within 20 bp of either side of the translocation breakpoints was almost 10-fold higher (5.4 × 10−3) than that seen overall for IgH in our translocations. This implies that c-myc/IgH translocations are resolved through error-prone repair, or that translocation breakpoints occur at or immediately adjacent to hypermutated sequences. Discounting mutations within 20 bp of the breakpoint, we identified three mutations in c-myc, resulting in a mutation rate that is above background (∼0.2 × 10−3) but approximately threefold lower than the rate observed for IgH (Fig. 4). These results suggest that c-myc translocations to the Ig V-JH region in IL-6 tg mice occur during or after Ig V-JH region SHM (28, 47, 48).
Molecular characterization of translocations found in lymphomas showed that in addition to Ig switch regions, many breakpoints are near RSS, suggesting a role for aberrant V(D)J recombination in their etiology. This idea is supported by experiments in mice deficient in DNA damage response and apoptosis pathways that develop spontaneous lymphomas harboring oncogenic translocations involving the TCR or IgH locus (1, 2, 49–53). For example, ataxia telangectasia mutated mice (ATM−/−) develop thymic lymphomas with TCR gene translocations (54–58). These translocations can be ascribed to V(D)J recombination because they fail to occur in the absence of RAG expression (51, 53). Similarly, mice deficient in p53 and nonhomologous end joining factors or histone H2AX develop RAG-dependent translocations that lead to pro–B cell lymphomas (1, 2, 49, 50, 52, 59–63).
However, some translocation breakpoints in proximity of RSS are found in mature B cell tumors, such as endemic Burkitt's lymphoma, diffuse large B cell lymphoma, and multiple myeloma (for review see references 7, 8, and 43). These translocations are believed to arise during or after the germinal center reaction because the Ig genes involved in the translocations are usually somatically mutated (7). Nevertheless, they could be mediated by V(D)J recombination if RAG1/2 were reexpressed in the germinal center. An alternative possibility is that the translocations to the Ig V-JH region in mature B cells are byproducts of double-strand DNA breaks created by AID during SHM.
Although double-strand breaks are not obligate intermediates in SHM, a measurable fraction of all Ig hypermutations involve deletions or insertions that require repair of double-strand breaks (36, 37). A region spanning 1–2 kbp downstream of Ig VH genes undergoes somatic mutation at a rate of 10−3 per basepair per generation in germinal center B cells (for review see reference 12) (47). Given the large numbers of B cells in the germinal center and their rapid rates of division, AID-induced double-strand breaks in the Ig V-JH region are not infrequent events (36–39). However, to date there has been no direct evidence for the role of AID-dependent Ig V-JH region breaks in chromosome rearrangement.
AID induces c-myc translocations to the Ig switch region by a mechanism that resembles class switching in that formation of the initial lesion requires cytidine deamination and uracil removal from DNA (25). However, resolution of the lesion proceeds by distinct pathways for switching and translocation (25). Factors that act in cis to promote switch region synapsis, such as 53BP1 and H2AX, have no impact on c-myc to Ig switch translocations despite their effects on genomic stability (25, 64). In contrast, factors that transmit damage signals to the nucleus, such as p53, do not appear to affect switching but are essential in suppressing Ig switch translocations, possibly by promoting the death of cells that overexpress c-myc (65, 66).
We have shown that mature B cells in IL-6 tg mice develop translocations involving c-myc and the Ig V-JH region and that they are AID dependent. These rearrangements closely resemble those found in endemic Burkitt's lymphoma, multiple myeloma, and diffuse large B cell lymphoma in that both of the translocated genes are hypermutated (for review see references 7 and 8). Ig V-JH regions are direct targets of AID, which is likely to produce the DNA double-strand break intermediates in the translocation reaction. Although translocated c-myc was also mutated, germline c-myc is not thought to be a target for AID (28), and c-myc mutation was substantially lower than the Ig partner. We speculate that c-myc mutation may have occurred after the translocation when it was under the control of Ig regulatory elements (28, 48, 67, 68).
The higher level of mutation we found proximal to the translocation breakpoints is analogous to what is observed for switch region junctions (69). However, the mutation breakpoints proximal to the c-myc to Ig V-JH region translocations are not RGYW hotspot biased and differ from breakpoint distal mutations in that they are not enriched for transitions. This suggests that the higher mutation frequency within 20 nts of the junction is the result of error-prone repair by nonhomologous end joining, whereas many of the mutations further away from the breakpoint are AID induced.
We conclude that translocations involving the Ig V-JH region of the Ig locus can be attributed to lesions produced by AID.
Materials And Methods
DNA preparation and PCR.
Hyperplastic lymph nodes from individual male and female IL-6 tg (14) and AID−/−IL-6 tg (12) mice were combined into four pools. Total DNA was prepared from 2 × 107 cells for each pool. 0.5 × 106 cells from each of the four pools for 12 different mice was assayed for derivative 12 translocations by PCR using primer set 1 and 2.5 × 106 cells using primer set 2 (see below). For derivative 15 translocations, 0.5 × 106 cells from each of the four pools from 12 different mice was amplified using set 3 (see below).
For derivative 12 translocations from c-myc to the IgH variable region, we performed nested PCR (Long Expand PCR system; Roche) using the following primers: primer set 1: first round with 5-GCAATGACTGAAGACTCAGTCCCTCTTAAG-3 (IgH) and ACTTAGCCCTGCAGACGCCCAGGAATCGCC (c-myc), followed by nested PCR with TACCATTTGCGGTGCCTGGTTTCGGAGAGG (IgH) and TTGGCTTCAGAGGCTGAGGGAGGCGACTG (c-myc); primer set 2: first round with GTGCCCCACTCCACTCTTTGTCCCTATGC (IgH) and GAAATAAAAGGGGAGGGGGTGTCAAATAATAAGAG (c-myc), followed by nested PCR with ATCATCCAGGGACTCCACCAACACCATCAC (IgH) and CCTCCCTTCTACACTCTAAACCGCGACGCCAC (c-myc). 500 ng DNA (105 cells) was amplified in each 20 μl first-round PCR reaction; 1 μl of the first reaction was template for the nested PCR reaction. PCR conditions for the first round were 94°C for the first 2 min, followed by 10 cycles of 94°C for 30 s, 61°C for 30 s, and 68°C for 7 min, followed by 19 cycles of 94°C for 30 s, 61°C for 30 s, 68°C for 7min plus 20 s/cycle. The conditions for the nested PCR were as follows: 94°C for 15 s, 61°C for 30 s, 68°C for 4 min, followed by 15 cycles of 94°C for 30 s, 61°C for 30 s, 68°C for 4 min plus 20 s/cycle. The combined total number of cycles of amplification was 45.
Conditions for amplification of derivative 15 translocations from c-myc to the IgH variable region were the same as those for derivative 12, with the exception of a 2-min extension time in the first-round PCR and a 1-min extension time in the nested PCR. Amplification of derivative 15 translocations was performed using the following primers: primer set 3: GTTGAGACATGGGTCTGGGTCAGGGAC (IgH) and ATCAGCGGCCGCAACCCTCGCCGCCGC (c-myc), followed by a nested PCR reaction with CTCTGCCTGCTGGTCTGTGGTGACATTAG (IgH) and GAAGGCTGGATTTCCTTTGGGCGTTGG (cmyc).
For amplification of derivative 12 c-myc to switch region translocations, primers described previously (24) were used following the PCR conditions described above for primer set 1.
All animal experiments were performed in accordance with the rules and regulations of The Rockefeller University Institutional Laboratory of Animal Care and Use committee.
Southern blot analysis.
PCR products from primer set 1 were separated on 0.8% agarose gels and denatured for 15 min in 0.4 M NaOH before transfer to nylon membranes and probing with P32 radiolabeled primers. The IgH probe sequence is GGTGGCAGAAGCCACAACCATACATTCCCA, and the c-myc probe sequence is GCGCCTCGGCTCTTAGCAGACTGTAT.
The PCR reactions were separated on 0.8% agarose gels, and the bands were gel extracted using the QIAGEN gel extraction kit according to the manufacturer's instructions and sent for direct sequencing. Translocations were sequenced with the nested PCR primers. If the translocation breakpoint was not identified with the first round of sequencing, we designed new primers for sequencing to walk along the translocation until we reached the breakpoint.
PCR products were sent directly for sequencing without cloning so that we could discount the error rate of the polymerase in our analysis. Mutations that arose during early PCR cycles could be identified by chromatogram analysis by the presence of more than one base signal for the same nucleotide. Such nucleotides were discounted from our analysis. All sequenced translocations were aligned using both SeqMan from DNA Star and the Codon Code Aligner software. Overlapping traces from all the translocation sequences allowed for distinction between real mutations and single nucleotide polymorphisms. In the case of c-myc, we also amplified and sequenced the first intron from the IL-6 tg mouse to verify that the c-myc mutations identified were not single nucleotide polymorphisms.
FACS analysis was conducted as described previously (24). All antibodies were from BD Biosciences.
The authors are grateful to Dr. Eva Besmer, Dr. Nina Papavasiliou, Dr. Andre Nussenzweig, and members of the Nussenzweig laboratory for suggestions.
D.F. Robbiani is a Fellow of the Leukemia and Lymphoma Society, and Y. Dorsett is a CRI predoctoral fellow. This work was supported in part by National Institutes of Health grants to M.C. Nussenzweig. M.C. Nussenzweig is a Howard Hughes Medical Institute Investigator.
The authors have no conflicting financial interests.
Abbreviations used: AID, activation-induced cytidine deaminase; IgH, Ig heavy chain; Ig V-JH, genomic region spanning the heavy chain V(D)J; RSS, recombination signal sequences; SHM, somatic hypermutation; tg, transgenic; V(D)J, variable, diversity, and joining.