Erroneous class switching and false VDJ recombination: Molecular dissection of t(8;14)/MYC‐IGH translocations in Burkitt‐type lymphoblastic leukemia/B‐cell lymphoma

Abstract

The chromosomal translocation t(8;14)(q24;q32) with juxtaposition of MYC to enhancer elements in the immunoglobulin heavy chain (IGH) gene locus is the genetic hallmark of the majority of Burkitt lymphoma and a subset of Diffuse large B‐cell lymphoma patients. Around 3% of adult B‐lineage acute lymphoblastic leukemia (ALL) patients show this aberration. Flow cytometry mostly reveals a “mature B‐ALL” or “Burkitt‐type” ALL immunophenotype. Using long‐distance PCR for t(8;14)/MYC‐IGH fusion, we investigated bone marrow, peripheral blood and a few other samples with suspected Burkitt‐ALL or mature B‐ALL and identified 133 MYC‐IGH‐positive cases. The location of the chromosomal breaks in the IGH joining and the 8 different switch regions was determined using a set of long‐distance PCRs. The chromosomal breakpoints with the adjacent MYC regions on 8q24 were characterized by direct sequencing in 49 cases. The distribution of chromosomal breaks among the IGH joining and switch regions was the following: JH 23.3%, M 21.8%, G1 15.0%, G2 7.5%, G3 3.8%, G4 4.5%, A1 12.8%, A2 3.8%, E 7.5%. Two breakpoint clusters near MYC were delineated. There was no clear correlation between the degree of somatic hypermutation and the chromosomal break locations. Epstein Barr virus was detected in 5 cases (4%). This detailed and extensive molecular analysis illustrates the molecular complexity of the MYC‐IGH translocations and the detected distribution of breakpoints provides additional evidence that this translocation results from failed switch and VDJ recombinations. This study may serve as a model for the analysis of other IGH translocations in B‐cell lymphoma.

Highlights

• Establishes an extensive map of MYC‐IGH translocations with previously unmatched accuracy.

• Shows an unexpectedly high frequency of MYC translocations to the Ce switch region.

• Reveals complex MYC translocations in a switch‐rearranged IGH gene locus.

• Investigates correlations between somatic hypermutation and IGH and MYC break location.

• Provides a map of the most frequently mutated amino acids in the MYC protein.

1. Introduction

The diagnosis of acute lymphoblastic leukemia (ALL) is currently mainly based on immunophenotyping, usually performed by flow cytometry of bone marrow aspirates and/or peripheral (leukemic) blood. Immunophenotyping allows the distinction between T‐ and B‐lineage. Further immunologic subdivisions of B‐lineage ALL lead to the classification in pro B, pre B, common or mature B/Burkitt‐ALL. Around two‐thirds of the latter patients carry a t(8;14)(q24;q32) chromosomal translocation that on the molecular level fuses the MYC gene to the immunoglobulin heavy chain (IGH) locus (Küppers and Dalla‐Favera, 2001; Janz, 2006). This translocation is a characteristic hallmark of the majority of Burkitt lymphomas. The translocation results in a dysregulation of MYC which is thought to be the pivotal event in molecular pathogenesis. Some arguments have been raised on the optimal treatment of these patients. While several therapy study groups treat them basically according to protocols similar to those developed for B‐precursor ALL, the German Multicenter ALL Therapy Study Group (GMALL) has reported significant improvements with treatment regimens consisting of repetitive therapy blocks including high‐dose methotrexate as an essential element (Hoelzer et al., 1996).

In the current work we report results obtained within the framework of the central molecular genetic diagnostics of the GMALL study group. During the 9‐year period between 2002 and 2011 a large number of patient samples with either a Burkitt‐ALL immunophenotype or suspected Burkitt lymphoma were investigated prospectively by long‐distance PCR for the translocation t(8;14). Archived samples from the 1990s were included in this analysis. One‐hundred and thirty‐three t(8;14)/MYC‐IGH‐positive samples were identified. The location of the breakpoint in the IGH joining and the different switch regions was subsequently characterized by a set of different long‐distance PCRs. A subset of these samples was further analyzed by sequencing the break region near MYC, including MYC exon 2. Various clinical and immunological data of the t(8;14)‐positive patients are reported.

2. Methods

2.1. Patient samples

Most patient samples were obtained within the framework of the GMALL B‐ALL/B‐NHL 2002 study (ClinicalTrials.gov Identifier: NCT00199082) and in part (retrospectively) also the preceding GMALL B‐ALL/B‐NHL 1990 study. Both are phase‐IV studies for adults and adolescents with newly diagnosed mature B‐cell acute lymphoblastic leukemia (B‐ALL), Burkitt lymphoma and other high‐grade lymphomas. Both studies have been approved by local ethics committees, and patients had given their written consent for scientific investigations. Eight samples were acquired locally. Five hundred eighty‐eight samples were investigated altogether. Since these samples were acquired from a large number of >100 clinics in Germany over a long time period the collection may be considered relatively unbiased and “quasi‐random”. All our investigations were undertaken either for diagnostic purposes or with residual material obtained through diagnostic procedures. Our study complied with the principles set forth in the Declaration of Helsinki.

2.2. Flow cytometry

Immunophenotyping by flow cytometry was done basically as described previously (Burmeister et al., 2005). Samples were classified as mature B‐ALL if they showed a B‐lineage commitment (CD19⁺, CD79α⁺, cyCD22⁺), positivity for HLA‐DR, CD10 and monotypical (kappa or lambda) surface immunoglobulin (sIg) and lacked antigen expressions typical of other hematopoietic neoplasms, such as CD5, CD34 and TdT. However, atypical immunophenotypes were also included in the analysis, e.g. those lacking sIg or CD10, with polyclonal sIg or positivity for TdT.

2.3. Cell lines

DNA from the following cell lines was used as positive control in the PCRs: Raji (IGHG), Ramos (IGHM), CA46 (IGHA), BLUE‐1 (IGHJ). The cell lines Raji, Ramos and CA‐46 were obtained from the DSMZ Braunschweig/Germany (Acc No 319, 603, 73), while BLUE‐1 was established in our laboratory (DSMZ Acc No 594) (Türkmen et al., 2011).

2.4. Long‐distance PCR (LD PCR) for t(8; 14)/MYC‐IGH

DNA was isolated using the Gentra PureGene method. LD PCR for t(8; 14) was performed as reported previously using the Expand Long Template PCR System kit (Roche, Mannheim, Germany) with a modified PCR cycler program as described by Basso (Basso et al., 1999). The analysis of each sample involved 11 different LD PCRs. One of these implied the amplification of a 9.0‐kb control product from the PLAT gene (primers GGAAGTACAGCTCAGAGTTCTGCAGCACCCCTGC, CAAAGTCATGCGGCCATCGTTCAGACACACC) to safeguard genomic DNA integrity. In 5 PCRs an antisense primer in MYC exon 2 (Myc‐04 ACAGTCCTGGATGATGATGTTTTTGATGAAGGTCT) was combined with an antisense primer in the CH1 exons of IGHA1/2 (Ca, TCGTGTAGTGCTTCACGTGGCATGTCACGGACTTG), the CH1 exons of IGHG1‐4 (Cg, GGTCACCACGCTGCTGAGGGAGTAGAGT), the CH1 exon of IGHM (Cm, at first: TGCTGCTGATGTCAGAGTTGTTCTTGTATTTCCAG, later: GTGTTCGTCTGTGCCCTGCATGACGTCCTTGGAAG), the CH3 exon of IGHE (Ce, ATCGTGGGCGACTTGCGGATGAAC) or in the 6th joining segment IGHJ6 (J6, CTTACCTGAGGAGACGGTGACCGTGGTCCC). Each investigation included 5 positive controls amplified in parallel (DNA from cell lines or, in case of IGHE, a positive patient sample).

2.5. LD PCRs and sequencing for the discrimination of MYC‐IGHG1‐4

Samples showing a PCR product with primer Cg were further investigated by 3 different LD PCRs. In each PCR primer, Cg was replaced by either primer Cg13 GGGTTTTGGGGGGAAGAGGAAGACTGACGGTC, Cg2 GGCACTCGACACAACATTTGCGCT or Cg4 CATGATGGGCATGGGGGACCATAT, showing specificity for IGHG1/IGHG3, IGHG2 or IGHG4, respectively. PCR products obtained with the primer combination Cg13 were sequenced with primer Cg‐seq ATATCCRGGAGGACCCTGCCCCTGACCT located between IGHG1/3 exons CH1 and CH2 to determine whether they were MYC‐IGHG1 or MYC‐IGHG3‐positive. LD PCR conditions were the same as described above.

2.6. LD PCRs and sequencing for the discrimination of MYC‐IGHA1 and MYC‐IGHA2

Samples showing a PCR product with primer Ca were subjected to a second LD PCR with primer Ca12 GCCGGTCAGTGTGCACGTGAGGTTCGC in combination with Myc‐04. The resulting PCR product was sequenced with primer Ca‐seq AGGCCAGGAGGGGCGAGGCGGGGGCA to distinguish MYC‐IGHA1 and MYC‐IGHA2. LD PCR conditions were the same as described above.

2.7. PCR for Epstein–Barr virus

The PCR investigation for Epstein–Barr virus (EBV) was done as reported previously using a nested PCR. EBV‐positive cases were subsequently analyzed by quantitative PCR to quantify the virus load (Burmeister et al., 2005).

2.8. DNA sequencing

DNA sequencing of PCR products was performed at the Max Planck institute's facilities on ABI sequencers using standard methods. All PCR products were sequenced double‐stranded. Sequences were submitted to the EMBL/Genbank nucleotide database (Acc. No. AM287138–AM287158AM287138AM287139AM287140AM287141AM287142AM287143AM287144AM287145AM287146AM287147AM287148AM287149AM287150AM287151AM287152AM287153AM287154AM287155AM287156AM287157AM287158, AM287160–AM287187AM287160AM287161AM287162AM287163AM287164AM287165AM287166AM287167AM287168AM287169AM287170AM287171AM287172AM287173AM287174AM287175AM287176AM287177AM287178AM287179AM287180AM287181AM287182AM287183AM287184AM287185AM287186AM287187).

3. Results

3.1. Mapping of chromosomal breaks in IGH

During the time period between 2002 and 2011, samples that showed an immunophenotype by flow cytometry compatible with Burkitt‐type lymphoblastic leukemia/Burkitt lymphoma (HLA‐DR⁺, CD19⁺, CD79α⁺, cyCD22⁺, CD10⁽⁺⁾, sIg⁽⁺⁾, TdT⁽⁻⁾, CD34^‐) were investigated by long‐distance (LD) PCR for t(8;14)/MYC‐IGH as described previously (Basso et al., 1999; Burmeister et al., 2005). Investigations within the B‐NHL2002 study (since 2002) were done prospectively while samples from the previous B‐NHL90 study (since 1990) were analyzed retrospectively using archived material. Atypical immunophenotypes, e.g. without CD10 or sIg expression or with TdT positivity were also included in this analysis. Bone marrow samples lacking bone marrow involvement by flow cytometry but with other evidence of a high‐grade lymphoma, e.g. by lymph node biopsy, or suspected diagnosis of Burkitt lymphoma were also investigated by LD PCR. Samples lacking sufficient DNA quality as revealed by a control PCR were excluded from further analysis. Cytogenetic data were only available in a minority of cases. One‐hundred and thirty‐three samples showed the presence of a t(8;14)/MYC‐IGH translocation by LD PCR. Eighty‐two samples were taken from bone marrow, 41 from peripheral blood, 4 from ascites, 1 from a lymph node, and 5 were not precisely specified (blood or bone marrow).

Thirty‐one of the positive samples showed a translocation of MYC to the IGHJ (IGHJ1‐IGHJ6) region, 41 to one of the IGHG regions (IGHG1‐IGHG4), 29 to the IGHM region, 22 to one of the IGHA regions (IGHA1 and IGHA2) and 10 to the IGHE region. To further refine this analysis, PCR primers based on individual sequence differences in IGHG1‐IGHG4 and IGHA1‐2 were developed (Figure 1A). While identifying MYC‐IGHG2 and MYC‐IGHG4 by specific LD PCRs was possible, it was impossible to establish reliable PCRs for the distinction of IGHG1 and IGHG3 and for the distinction of IGHA1 and IGHA2. Therefore, a consensus PCR primer in exon CH2 of IGHG1 and IGHG3 was used, and the 3′ end of the resulting PCR product was sequenced to determine whether it was MYC‐IGHG1 or MYC‐IGHG3 (primers Cg13 and Cg‐seq, Figure 1A+B). The same approach with a consensus PCR primer in exon CH2 of IGHA1 and IGHA2 was pursued in the case of IGHA1 and IGHA2 (primers Ca12 and Ca‐seq, Figure 1). In this way, it was reliably and unambiguously possible to identify the approximate point of insertion of MYC into the IGH locus. The final distribution of IGH breakpoints was thus the following: IGHJ 31 (23.3%), IGHM 29 (21.8%), IGHG1 20 (15.0%), IGHG2 10 (7.5%), IGHG3 5 (3.8%), IGHG4 6 (4.5%), IGHA1 17 (12.8%), IGHA2 5 (3.8%), IGHE 10 (7.5%) (Figure 2).

Figure 1.

PCR primer locations in the IGH constant region genes. (A) Exon organization of the constant region genes IGHM, IGHG1‐IGHG4, IGHA1/IGHA2, and IGHE with PCR primer locations and orientations (Lefranc and Lefranc, 2001). Primer binding sites are shown as triangles (◀ antisense, ▶ sense = sequencing primer). All antisense primers were used in combination with primer Myc‐04. Three different IGHG3 genotypes differing in the number of hinge (H) regions are known. Primers Cg, Ca, Cm, Ce were used in the first PCRs. Cases showing a PCR product with primer Cg were further investigated with primers Cg2 and Cg4 and those showing no PCR product with both were subjected to PCR with primer Cg13. Cases showing a PCR product with primer Ca were further investigated with primer Ca12. The PCR products resulting from primer Cg13 or primer Ca12 were sequenced with primers Cg‐seq and Ca‐seq. (B) Examples of PCR products generated with the primer combination Myc‐04/Cg13. Above the lanes the sample numbers (std = size standard).

Figure 2.

Location of breakpoints in the IGH J‐C region and PCR products exemplified. The upper part of the figure shows the gene organization of the IGH J‐C region. Chromosomal breaks occur in the 5′ vicinity of the 6 functional J segments and in the 8 different switch regions. The IGH J‐C region comprises 3 major enhancer elements (Eμ and the two 3′RR elements 3′ of CA1 and CA2). The diagram below shows the distribution of breaks as detected by LD PCR and sequencing. Thirty‐one cases showed a break in the joining region with some of them displaying additional rearrangements of the constant region genes (smaller pie chart).

The genetic distance between the IGHJ6 and IGHM genes comprises around 7 kb (Figure 2). Since every sample was rigorously investigated by 5 LD PCRs for MYC‐IGHJ, ‐IGHM, ‐IGHG, ‐IGHA, ‐IGHE), we usually observed a MYC‐IGHM PCR product that was approximately 7 kb larger than the MYC‐IGHJ product in cases that were MYC‐IGHJ‐positive. In some MYC‐IGHJ‐positive samples a MYC‐IGHG, MYC‐IGHA or MYC‐IGHE PCR product was observed instead of the expected MYC‐IGHM PCR product (Figure 3A). In these cases the IGH locus had apparently undergone isotype recombination IGHM > IGHG or IGHM > IGHA, such that the intervening DNA sequence between the switch regions of IGHM and the respective IGHG or IGHA locus had been deleted (Figure 3B). We determined the involvement of the respective IGHG and IGHA genes by LD PCR and sequencing as described above. Seven out of 31 MYC‐IGHJ‐positive cases had undergone switch recombination. The distribution of involved constant gene loci was the following: MYC‐IGHJ‐IGHM 24 (77%, i.e. no switch recombination), 2 MYC‐IGHJ‐IGHG1 (6%, i.e. switch recombination IGHM > IGHG1), 1 MYC‐IGHJ‐IGHG3 (3%), 3 MYC‐IGHJ‐IGHA1 (10%), 1 MYC‐IGHJ‐IGHE (3%) (Figure 2). It was not possible to decide whether switch recombination had occurred before or after MYC translocation.

Figure 3.

MYC translocation in the IGHJ (joining) region. (A) Agarose gel image of two exemplary cases with MYC translocation to the IGHJ region. In the first case a larger PCR product is also generated with the primer for the IGHM constant region. In the second case, the IGHM PCR product is missing but an IGHG PCR product is generated (IGHG1, as revealed by sequencing). (B) Two hypothetical molecular mechanisms are conceivable for case 2: either a translocation of MYC to a switch‐rearranged IGH locus (above) or a switch recombination of a MYC‐IGHJ‐rearranged IGH locus (below).

3.2. Clinical correlations and EBV involvement

The clinical and immunological characteristics of the MYC‐IGH‐positive patients are summarized in Table 1. Sixty‐seven percent of the patients were male. This male preponderance was present in all genetic subgroups. The median age at diagnosis was 50 years (range 17–82). MYC‐IGHGM‐, MYC‐IGHA1‐, and MYC‐IGHA2‐positive patients appeared younger, while MYC‐IGHG3‐ and MYC‐IGHG4‐positive were older but especially in the latter three, the numbers were too small for statistically significant conclusions. Most evaluable patients showed a kappa light chain restriction with positivity for CD10 and surface immunoglobulin and lack of TdT. Aberrant antigen expression patterns (TdT⁺, CD10^‐, sIg^‐) were observed in less than 10% of cases (Table 1). Two samples from 70‐ and 71‐year‐old males (MYC‐IGHG2, CD10⁺/sIg⁺ and MYC‐IGHA2, CD10^‐/TdT^‐/sIg^‐) showed expression of CD5. A third sample (ascites from a 58‐year‐old male, MYC‐IGHA1, with high EBV load) had a CD10^‐/TdT^‐/sIg⁺ immunophenotype with expression of CD38.

Table 1.

Characteristics of MYC‐IGH‐positive patients.

	All	J	M	G3	G1	A1	G2	G4	E	A2
Number	133	31	29	5	20	17	10	6	10	5
Median age (range) [ys]	50 (17–82)	48 (17–78)	37 (18–76)	72 (50–76)	52.5 (17–75)	43 (19–81)	52 (23–82)	63.5 (50–74)	58.5 (44–79)	28 (26–70)
Sex (m/f)	86/47	19/12	18/11	3/2	16/4	9/8	8/2	4/2	6/4	3/2
Kappa/lambda (N = 101)	63/38	13/9	15/8	2/3	11/7	9/3	4/4	2/3	4/5	3/0
CD10‐ (N = 111)	9	0	4	0	1	2	0	0	0	2
TdT+ (N = 108)	4	0	1	0	0	0	1	1	1	0
sIg‐ (N = 112)	11	4	0	1	0	1	1	2	1	1
EBV‐DNA+	5	2	1	1	0	1	0	0	0	0

All samples were tested for the presence of Epstein–Barr virus (EBV) DNA by nested PCR. PCR‐positive samples were subsequently investigated by a quantitative PCR for EBV to quantify the virus load. Forty (30.1%) samples were EBV‐positive by nested PCR but a higher virus copy number of >1000 genomes per 100 ng genomic DNA (adjusted to the relative blast count) was detected in only 5 cases (3.8% of all, Table 1). All other samples showed virus copy numbers below 20 per 100 ng DNA and were regarded as EBV‐negative.

3.3. Detailed sequence analysis of 49 samples

In 49 randomly selected cases the entire sequence from the 3′ end of MYC exon 2 to the chromosomal break position was determined by sequencing. The sequences were submitted to the EMBL/Genbank database (AM287138–AM287158, AM287160–AM287187). The chromosomal breakpoints were dispersed over an area from some bp 5′ of the second MYC exon to 2 kb 5′ of the first MYC exon (Figure 4A). Although most breaks appeared to be somewhat randomly scattered there was a clustering around the boundaries of MYC exon 1. Two break clusters were delineated, the first about 2100–2650 bp and the second 1250–1850 bp 5′ of MYC exon 2. Thirty‐eight of all breaks (78%) occurred in these 2 regions with some of them only a few bp apart (Figure 4A and supplement 1). Remarkably, somatic hypermutations appeared also to be clustered in these two regions. No association of the break position on chromosome 14 with the break position on chromosome 8 was apparent. Analysis of the sequences discerned samples which had been variably affected by somatic hypermutation in the exonic and intronic regions (supplement 1 and 2). Thirty‐two samples (65%) showed a mutated amino acid sequence of the 252 N‐terminal amino acids (aa) encoded by MYC exon 2 (a sequence alignment can be found in supplement 2) with altogether 86 aa substitutions. The average number of aa mutations per sample was 2.8 (median 2.0), and none of the mutated samples showed a premature stop codon in the MYC sequence. One sample was heavily mutated with 25 aa substitutions (10% of all aa), another sample showed an insertion of 5 aa (PPLPTP) between aa 59 and 60, possibly a hitherto unreported MYC variant. The most frequent mutations were P57T (4 cases), F138S (4), S249N (4), and E39D (3); the most frequently mutated amino acid positions were aa 57 (7 mutations), aa 138 (6), aa 58 (5), and aa 249 (4) (Figure 4B). Twenty mutations occurred in the MYC homology box 1 region (Mb1), 11 in the MYC homology box 2 region (Mb2), and one in the MYC homology box 3a (Mb3a) region (Figure 4B). Sixty‐seven (78%) of all aa mutations were found in the first 149 aa, the “transactivation domain”.

Figure 4.

Breakpoint locations near MYC and amino acid mutations in MYC exon 2. (A) Location of breakpoints near MYC. The chromosomal breakpoint was determined in 49 cases by sequencing. Break locations are represented as bullets (●). MYC has two alternative translation start sites, one non‐canonical CTG start site in exon 1 and one ATG start site in exon 2. The coding sequence starting from the ATG site is shaded in black. Four different promoters (P0, P1, P2, P3) have been described with the great majority of transcripts resulting from the use of P1 or P2 under normal circumstances. (B) Amino acid (aa) mutations in MYC exon 2 (ATG codon = aa 1). Altogether 86 amino acid mutations were found distributed among 32 cases (supplement 2). The 3 detected N11S substitutions were regarded as polymorphisms (rs4645959) and are not included. The numbers above the bars represent amino acid numbers. All amino acids with more than one mutation are numbered. The most frequent mutations were P57T (4 cases), F138S (4), S249N (4) and E39D (3). Below the scale the regions of MYC exon 2 important for transformation (Mb1, Mb2, Mb3a = MYC homology box 1, 2 and 3a).

A number of single nucleotide polymorphisms (SNPs) were detected in the sequenced cases at low frequency (supplement 1). Three cases showed the rs4645959 SNP in MYC exon 2 with N11S aa substitution. Further SNPs leading to amino acid substitutions were not detected. The degree of hypermutation of the samples ranged from 0% (no mutation at all) to a maximum of 3.12% nucleotide mutations. Four of the samples showed no mutations (3 MYC‐IGHJ, 1 MYC‐IGHM), 18 a low mutation level (defined as >0% and ≤0.3%), 20 an intermediate (>0.3% to <0.8%), and 7 a high level of somatic hypermutation (supplement 1). These latter cases comprised one MYC‐IGHM‐ and ‐IGHG3‐, two MYC‐IGHG1‐, and three MYC‐IGHJ‐positive cases. One of the three MYC‐IGHJ‐positive cases showed a switch rearrangement to IGHG1 (see Figure 3). Remarkably, the correlation between the degree of nucleotide hypermutation and the number of amino acid exchanges in MYC exon 2 was relatively weak. Highly mutated cases not necessarily showed a high degree of amino acid substitutions and vice versa. There was also no clear correlation between the degree of hypermutation and the break in the IGH locus, e.g. there were highly mutated and unmutated MYC‐IGHJ‐positive cases. Cases with MYC translocation in a switch region did not show a different degree of hypermutation as compared to those with MYC translocation to the joining region.

4. Discussion

Around 25% of the observed chromosomal breaks in the IGH region occurred in the joining region, i.e. closely 5′ to one of the IGHJ genes. Seventy‐five percent of the breaks were located in one of the 8 switch regions. The relative distribution of the chromosomal breaks in the C_G1–C_G4 and C_A1–C_A2 switch regions remarkably mirrors the relative immunoglobulin serum concentrations of the 4 IgG and 2 IgA subclasses (Berth et al., 1999; Alyanakian et al., 2003). This supports the hypothesis that breaks in these regions are indeed the result of randomly occurring errors in switch recombination. Breaks in the C_M and C_E switch regions were, however, significantly more frequent than expected from their relative serum immunoglobulin concentrations. This may be explained by the fact that IgM and IgE are largely membrane‐bound and their serum concentrations do not truly reflect the extent to which recombinations involving the switch regions of C_M and C_E occur. Some cases with MYC juxtaposition to the joining region showed a switch‐recombined constant region (Figures 2 and 3). This observation can be explained by two different mechanisms: either by a switch recombination in a MYC‐IGHJ‐recombined IGH locus or by a failed secondary recombination of a V_H‐D_H‐J_H‐rearranged region that already underwent switch recombination. This secondary recombination may be a failed V_H‐J_H recombination involving a more 5′ upstream V_H gene. This process is known as “V replacement” and occurs through cryptic recombination signal sequence (RSS) sites near the 3′ end of the V_H‐D_H‐J_H‐rearranged V_H gene (Darlow and Stott, 2005; Zhang, 2007). While the first process would presumably involve activation‐induced cytidine deaminase (AICDA), the enzyme that is responsible for somatic hypermutation and class switch recombination (Pavri and Nussenzweig, 2011; Stavnezer, 2011), the second would likely be mediated by RAG1 and 2 proteins, i.e. occur at an earlier stage of differentiation in a “pre‐germinal center” cell (Darlow and Stott, 2005; Zhang, 2007).

Three major cis elements in the IGH gene locus regulate the machinery of V_HD_HJ_H recombination and isotype switching (Figure 1). The first is the Eμ “intronic enhancer”, which is located between the joining and C_M constant region. Eμ promotes V_HD_HJ_H rearrangement in B‐cell progenitors, easily yields high expression of transgenes in B‐cells but is dispensable for IGH expression in mature B cells (Perlot et al., 2005; Pinaud et al., 2011). The second and third cis regulatory elements are two virtually identical “3′ regulatory regions” (3′RR) or “3′ locus control regions” (3′LCR), which have been mapped 3′ of the C_A1 and C_A2 constant regions (Chen and Birshtein, 1996; Mills et al., 1997). Both 3′RRs harbor 3 enhancers (HS_1,2, HS₃ and HS₄), which are mostly active in mature B‐cells and are important for IGH transcription at the plasma cell stage and for efficient switch recombination (Vincent‐Fabert et al., 2010). A translocation of MYC to IGH thus may lead to a dysregulation of MYC by different enhancer elements. Translocations to the joining region leave all 3 regulatory elements intact while the Eμ enhancer is removed by translocations to the switch regions. Additionally, translocations to the first 3 switch regions (those of C_G3, C_G1, C_A1) leave both 3′RRs intact while the 5′ proximal 3′RR is deleted by translocation to the switch regions of C_G2, C_G4, C_E or C_A1. The role of these enhancer elements upon MYC expression has been investigated in murine models bearing either Eμ‐Myc or 3′RR‐Myc transgenes. Eμ‐Myc transgenic mice mostly developed B‐cell precursor malignancies, while those with a 3′RR‐Myc transgene developed more mature lymphomas immunologically and morphologically resembling Burkitt lymphoma (Truffinet et al., 2007). This suggests that the 3′RRs are more important for the development of murine Myc‐driven Burkitt‐type lymphoma (Gostissa et al., 2009). However, the results from murine models may not be simply transferable to humans.

The detailed sequence analysis of 49 cases showed evidence of somatic hypermutation in most samples, but a few were not mutated. Somatic nucleotide mutations are an indicator of the action of AICDA (Stavnezer, 2011). In addition to single nucleotide substitutions several samples showed deletions more typical of switch recombination junctions (supplement 1) (Pavri and Nussenzweig, 2011). Hypermutated samples could possibly originate from a post‐germinal center cell while unmutated samples could represent an early B‐cell phenotype. However, there was no significant difference in the degree of hypermutation between samples with MYC juxtaposition to the joining region and those with MYC juxtaposition to one of the switch regions. There were also no significant differences in antigen expression between the two groups (Table 1).

The 49 analyzed chromosomal breaks near MYC were primarily clustered in two regions, each comprising around 500–600 bp at the 5′ and 3′ ends of MYC exon 1. A possible explanation would be the existence of secondary structures in these regions, facilitating the formation of a single DNA strand (ssDNA) or RNA‐DNA hybrid during transcription. ssDNA is the substrate of AICDA. Both regions have a high G and C content (>55%) and could form structures like R loops or G loops which are known targets of AICDA in the switch regions (Duquette et al., 2005; Ruiz et al., 2011). Somatic hypermutation appeared to be concentrated in these two regions. The association of break location near various oncogenes and somatic hypermutation has been previously described and found to be suggestive of the involvement of AICDA in these translocations (Küppers and Dalla‐Favera, 2001; Pasqualucci et al., 2001). Mice with an Igk‐Aicda transgene had an increased frequency of Myc‐Igh translocations in activated B cells (Robbiani et al., 2009). Recently, Greisman et al. (2012) proposed a hypothetical model in which chromosomal breaks at the earliest stage of B‐cell development, e. g. in t(11;14)/CCND1‐IGH ‐positive mantle cell lymphoma, occur predominantly near CpG dinucleotides, and to a lesser extent near the (A/T)GC(A/T) consensus motif (Greisman et al., 2012). Chromosomal breaks at a germinal center B‐cell stage, e.g. with t(8;14)/MYC‐IGH, were found predominantly near the (A/T)GC(A/T) consensus motif. Both sequence motifs are targets of AICDA and the authors speculated that the higher expression of AICDA in later stage B cells might be the cause of this different clustering. However, functional test are needed to prove these correlations.

The location of the chromosomal break near MYC may influence the level of MYC transcription, as reported by Wilda et al. (2004), who found that MYC mRNA levels were inversely correlated with the distance between break position and 5′ end of MYC exon 2 (Wilda et al., 2004).

The effect of amino acid mutations on MYC function has been investigated in previous studies. Oster et al. analyzed the N‐terminus of the human MYC protein by using various deletion variants and MYC proteins with specifically introduced mutations between amino acids 45–63 (Mb1 = Myc box 1 region) or 128–143 (Mb2 = Myc box 2 region) (Oster et al., 2003). When expressing these various MYC proteins in HO15.19 cells, they observed a doubling time that was not significantly different as compared to wild‐type MYC. Although 20 and 11 mutations (23% and 13%) detected in our series were located in Mb1 and Mb2 (Figure 4B, supplement 2) we observed none of the mutants investigated by Oster at al. (D48A, K51A, K52A, L55A, I130E, Q131A, C133A, W135A, W135F, W135E, S136A). Therefore only limited conclusions can be drawn from this work. Hemann et al. found that the 2 Myc mutants P57S and T58A were unable to induce the pro‐apoptotic protein Bcl2l11 (Bim) in transgenic mice, thus leading to an evasion of p53 tumor suppressor function (Hemann et al., 2005). The observed distribution of mutations in MYC exon 2 was similar to that reported previously (Smith‐Sørensen et al., 1996) although most previous reports heavily relied on cell lines and investigated quite heterogeneous cohorts including many endemic Burkitt lymphomas. Cell lines tend to acquire additional mutations during in vitro cultivation (Yano et al., 1993; Albert et al., 1994). In contrast, the cohort investigated here included only patients from a limited geographic area and no cell lines.

The translocation t(8;14) with MYC‐IGH juxtaposition is typical of Burkitt lymphoma (BL) and a subset of diffuse large B‐cell lymphomas (DLBCL). Occasionally, it is also found in multiple myeloma. Various attempts have been made to better distinguish BL from DLBCL. Gene expression analyses have defined a BL‐typical gene expression profile, characterized primarily by the overexpression of key MYC target genes (Dave et al., 2006; Hummel et al., 2006). DLBCL cases on the other hand have shown a gene expression profile with pronounced expression of genes involved in the NFKB1 pathway. However, remarkably little attention has been paid to the precise molecular structure of MYC translocations in these gene expression studies. Earlier WHO classifications have suggested that t(8;14)/MYC‐IGH translocations would affect the joining regions in endemic BL and the switch regions in sporadic BL (Diebold et al., 2001) but this assumption has been eliminated in the current classification (Leoncini et al., 2008). Translocations of MYC in the switch regions are supposed to occur at a later stage of B‐cell maturation than those involving the joining region. However, this simple, albeit attractive view is disproven by numerous reports of endemic cases, with chromosomal breaks in the switch regions (e.g. endemic BL cell lines Raji, Ramos) on the one hand and sporadic cases with juxtaposition of MYC to the joining regions on the other hand (Shiramizu et al., 1991; Basso et al., 1999; Burmeister et al., 2005).

Busch et al. (2007) reported the chromosomal breakpoints of 78 pediatric sporadic Burkitt lymphoma patients (Busch et al., 2007). They analyzed in detail the breakpoint locations near MYC and found a similar clustering as reported here but did not precisely specify the break location in IGH. Instead they summarized all four C_G and the two C_A switch regions, because they used one PCR for both and they did not detect any C_E breakpoints, because no PCR for MYC‐IGHE was used.

EBV was detected in significant copy numbers in around 4% of cases in our series. Despite the fact that the association of this virus with BL is known for more than 40 years, it's role is still unclear and has been called “curious” and “enigmatic” (Thorley‐Lawson and Allday, 2008; Grömminger et al., 2012). Bellan et al. have suggested that EBV‐positive and ‐negative cases might arise from different cells of origin (Bellan et al., 2005), the former from memory B‐cells or late germinal center lymphoblasts and the latter from more immature early centroblasts. In our series, the few EBV‐positive cases appeared to be randomly distributed among the genetic subgroups, and there was no molecular hallmark that distinguished EBV‐positive from EBV‐negative cases (Table 1). One EBV‐positive case showed a plasmacytic differentiation and was CD10‐negative, while the others were CD10‐positive, thus resembling a more mature “post germinal” phenotype. Even in EBV‐PCR‐negative cases an involvement of the virus in the pathogenesis cannot be completely ruled out, because it may only be a factor in the induction of neoplastic growth, but may no longer be necessary in its maintenance. Thus, it may successively be lost during growth of the malignant clone.

In summary, the results presented here illustrate the remarkable complexity of the t(8;14)/MYC‐IGH aberration. The present study represents by far the largest investigation on this topic in adult patients and no previous investigation has consistently mapped the chromosomal breakpoints to individual IGH switch regions. Future gene expression studies investigating t(8;14)/MYC‐IGH‐positive lymphomas, i.e. those aimed at distinguishing BL from DLBCL, should take account of this complexity and include molecular details of the t(8;14)‐positive cases. The analysis presented here may also serve as a model for the analysis of chromosomal translocations involving the IGH joining and switch regions in various other B‐cell lymphomas.

Conflict of interest disclosures

All authors declare that they have no potential conflict of interests.

Supporting information

The following are the supplementary data related to this article:

Supplementary data

Supplementary data

Supplementary data 1.

1.1.

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.molonc.2013.04.006.

Burmeister Thomas, Molkentin Mara, Schwartz Stefan, Gökbuget Nicola, Hoelzer Dieter, Thiel Eckhard, Reinhardt Richard, (2013), Erroneous class switching and false VDJ recombination: Molecular dissection of t(8;14)/MYC‐IGH translocations in Burkitt‐type lymphoblastic leukemia/B‐cell lymphoma, Molecular Oncology, 7, doi: 10.1016/j.molonc.2013.04.006.