Class switching and Myc translocation: how does DNA break?

Abstract

Chromosomal translocations involving immunoglobulin switch regions are commonly thought to arise from aberrant AID-induced DNA lesions. New data, however, suggest AID does not initiate such lesions, but acts subsequently in the B cell transformation process.

Early events in B cell malignant transformation often involve chromosomal translocations between the Myc oncogene and immunoglobulin switch (S) regions. These translocations are believed to arise from abberant immunoglobulin class-switch recombination (CSR), involving doublestrand DNA breaks (DSBs) in both Myc and the S regions. The enzyme activationinduced cytidine deaminase (AID) is required for CSR. DSBs in S regions are essential intermediates in CSR and are, together with DSBs in Myc, the initiating lesions leading to such translocations. Thus, one theory poses that AID is responsible for generating the aberrant lesions that initiate chromosomal translocations. However, new data from Schatz and colleagues¹ indicate that AID is not required for the first step of CSR: the generation of S-region DSBs. Unniraman et al.¹ show that this cytidine deaminase makes no measurable contribution to the generation of initial Myc–S-region translocations, thereby indicating that S regions accumulate DSBs in a way that does not require the intervention of AID.

The genes encoding the B cell receptor for antigen and secreted antibody show a high degree of variability, which is generated by two sequential stages of genetic alteration. In the early stage, gene recombination dependent on the recombination activating gene product selects one of each of the variable (V), diversity (D) and joining (J) segments from the respective gene pools and combines them into a single immunoglobulin V(D)J exon. After a B cell encounters its corresponding antigen, immunoglobulin genes undergo two types of DNA modification: CSR and somatic hypermutation (SHM). CSR and SHM somatically diversify the antibody molecule in different ways, although both require AID. CSR substitutes the constant region of the antibody heavy chain (C_H) with a downstream C_H region by introducing DSBs in the S regions of an upstream and a downstream C_H gene and perfecting an intrachromosomal and inter-S-S-region recombination entailing the excision of intervening genomic DNA. SHM inserts single-nucleotide changes (mismatches) in V(D)J genes.

The demonstration that AID can directly deaminate DNA in Escherichia coli² and is upregulated in germinal center B cells, which actively undergo CSR and can incur into neoplastic Myc translocation, has led to the assumption that AID effects DSBs in the S regions. This assumption has been further strengthened by the demonstration that uracil DNA glycosylase (UDG) is also required for CSR². Indeed, a widely accepted model of CSR entails the direct deamination by AID of cytosine bases into uracil residues in the S-region DNA made single-stranded by transcription, which precedes CSR. Subsequently, UDG, which is also required for CSR, would remove the uracil bases to generate abasic sites. Such abasic sites would then be removed by apurinic/apyrimidinic endonuclease 1 (APE1), resulting in a single-strand break (Fig. 1a, major pathway). A similar nearby lesion on the opposite strand would result in a DSB, and DSBs in both an upstream and a downstream S region would lead to excision of the intervening DNA segment. A crucial prediction of this model is that UDG is necessary for the generation of DSBs during CSR. In absence of UDG, DSBs could be generated by intervention of the mismatch repair protein on the emerging mismatched U-G pairs (Fig. 1a, minor pathway). A different but not mutually exclusive possibility is that AID functions as an mRNA-editing enzyme, possibly allowing for the expression of a yet-to-be-identified endonuclease (Fig. 1b). The idea of an mRNA-editing function for AID is supported by the structural similarity of AID to APOBEC-1, an apolipoprotein B–editing enzyme, and the finding that de novo protein synthesis is required for CSR and for the formation of DNA repair γ-H2AX (phosphorylated H2AX, a mammalian core histone H2A variant) foci in the immunoglobulin heavy chain (Igh) locus³.

Figure 1.

Generation of DSBs in S regions. (a) DNA deamination: AID directly deaminates cytidine residues in DNA, converting them to uridine residues. The G:U mismatch can then be processed by either the base-excision repair (BER) pathway (the major pathway) or by the mismatch-repair (MMR) machinery, which includes mutS homolog 1 (Msh1), Msh6, mutL homologue 1 (Mlh1) and postmitotic segregation (Pms), to introduce gaps or nicks on opposite strands of the S-region DNA. The nicks induced by the base-excision repair pathway are thought to be generated by the following process: UNG removes the AID-introduced deoxyuridine, thereby creating an abasic site that is processed by APE1 to yield a DNA nick. (b) RNA deamination and editing: AID deaminates mRNA through recognition by specific cofactors, and mRNA thus edited is translated into a putative endonuclease, which cleaves DNA to generate DSB. (c) The work by Unniraman et al.¹ together with the findings by Begum et al.⁵ cast considerable doubt on the models summarized in a and b and lend strong support to the idea of AID-independent generation of DSB: Blunt DSBs are generated in yet-to-be-determined way independently of AID. In the absence of AID, these DSB ends are ligated intra-S region through nonhomologous end-joining. After stimulation by activated CD4⁺ T cells, AID is upregulated and recruited to DSB, perhaps through a cofactor, to deaminate cytidine near the free DNA ends and generate U:G mismatches. UDG mediated attack on the mispaired U would lead to the generation of an abasic site, which becomes the substrate of APE1 to yield DNA nicks and give rise to resected DNA ends (U:G mispairs could also be processed through the BER or MMR pathways, as shown in a). A second possibility is that AID, together with specific cofactors, deaminates and edits mRNA, specifying an endonuclease that resects DNA ends. Finally, a third possibility is that AID, together with specific cofactors, deaminates and edits mRNA that is translated into a recombinase, which remodels chromatin in a way that facilitates S-region synapsis and inter-S-S-region recombination.

To test the hypothesis that translocations involving the S regions arise from aberrations of the AID-dependent generation of DSBs, Unniraman et al.¹ used AID-deficient mice, in which CSR is absent. Unexpectedly, these mice had abundant translocations of Myc into S regions, indicating that the generation of the DSBs that are responsible for these translocations is independent of AID. As shown by their further experiments, outgrowth of translocation-positive cells was dependent on AID, raising the possibility that AID, which is not required for the generation of the S-region DSBs, is important for tumor progression, perhaps by virtue of its mutagenic properties. These are novel and important findings. Their relevance extends far beyond the involvement of CSR events in Myc–S-region translocation and neoplastic transformation to challenge the widely held assumption that cytidine deamination by AID is a critical step in the generation of the initial DSB that lead to CSR. This conclusion is only apparent in contrast with that recently drawn by Ramiro et al.⁴, who failed to detect similar Myc translocations in mice transgenic for the gene (Il6) encoding interleukin 6 (IL-6) that were mutant for AID. Instead, it provides an explanation for the failure of Ramiro et al. to detect B cells with Myc translocations and their conclusion that “AID is required for c-myc/IgH chromosome translocations in vivo.” Chronic and abundant in vivo production of IL-6, as induced by Unnirmann et al.¹ by multiple injections with pristane, leads, in the absence of AID, to a counterselection of outgrowth of cells with Myc translocation. This stage is comparable to the stage well characterized by Ramiro et al.⁴ in a situation of chronic IL-6 secretion as achieved in Il6-transgenic mice.

The putative dependence on AID for γ-H2AX focus formation in the Igh locus, during CSR⁵, and the apparently defective generation of DSBs in S_μ (ref. 6) of patients with mutations in the gene encoding uracil-DNA glycosylase (UNG), have been used as evidence supporting the contention that AID cleaves double-stranded DNA to generate DSBs. However, H2AX phosphorylation is dispensable for the initial recognition of DNA breaks and CSR still occurs at substantial levels in the absence of H2AX, indicating that AID-dependent γ-H2AX foci do not reflect the presence of DSBs, but instead constitute intermediates in the DNA repair process itself^7,8. Some DSBs are generated during DNA replication and these DSBs must be repaired. The fact that H2AX deficiency is not detrimental to life further indicates that γ-H2AX foci formation does not necessarily reflect the presence of DSBs. This contention is further strengthened by the failure to detect γ-H2AX foci in the Igh locus of AID-deficient (Aicda^−/−) mice despite the demonstrated presence of considerable DSBs in the V_H gene segments of these mice⁹. The reason for the apparent defective generation of DSBs in S_μ in patient P2 (ref. 6) with a UNG mutation and in hyperimmunoglobulin M syndrome remains elusive. Unexpectedly, the UNG mutant reproducing the P2 patient mutation is active in both uracil removal and CSR⁵. Accordingly, experiments in our laboratory clearly show the patient P3 DNA displays significant DSBs in S regions (unpublished data). Finally, the uracil-removing activity of UDG is dispensable for CSR⁵.

The dispensability of AID in the generation of S-region DSBs parallels the dispensability of AID in generation of DSBs that characteristically target the (mutational) RGYW ‘hotspot’ in immunoglobulin V genes^9,10. Such AID-independent DSBs occur not only in immunoglobulin V genes but also in human MYC and the human B cell leukemia gene BCL6, and are blunt-ended¹¹. They, however, are ‘processed’ by AID to yield resected ends in hypermutating B cells in vitro and in vivo¹¹. These resected free DNA ends would be essential for initiation of the process that leads to the insertion of somatic mutations through homologous recombination involving the intervention of error-prone DNA polymerases. A similar process is probably at work in S-region DNA, in which AID-independent DSBs occur at a high frequency. These DSBs are blunt-ended but become resected once AID is induced after B cell CD40 engagement by CD154 on activated CD4⁺ T cells and B cell receptor for antigen cross-linking (Fig. 1c). These data suggest that the initiation step in CSR and SHM is similar and consists of AID-dependent processing of an AID-independent double-stranded DNA cleavage. The cytidine deaminase catalytic core of AID is crucial for CSR and SHM. AID could be specifically recruited to blunt DSB ends and ‘process’ them, together with UDG and APE1, to generate resected DSBs by attacking a cytosine near the DSB free ends. Or AID could function as an RNA-editing deaminase and edit mRNA for an endonuclease that would resect the blunt DSB ends. Whether generated through direct DNA deamination or through mRNA editing, the resected DSB ends could become direct substrates for synapsis and recombination.

The discovery of AID and the subsequent delineation of its enzymatic activity have provided a model by which the initiation of CSR is effected. The work by Unniraman et al.¹ together with the findings by Begum et al.⁵ challenge the prevailing model of CSR and force reconsideration of the possible mechanism that underlies the generation of DSBs. DSBs would be generated as blunt-ended DNAs by a yet-to-be-identified endonuclease. Like in immunoglobulin V genes, DSBs mainly target the RGYW hotspot in S regions, indicating that a putative endonuclease or cleavage factor would have to specifically interact with RGYW, which is highly represented in the S regions, particularly S_μ. However, this motif occurs throughout the genome, including immunoglobulin C_H regions, suggesting that DSB in immunoglobulin S regions and V genes may also be the expression of an ‘intrinsic fragility’ of these DNA sequences¹. The resulting AID-independent blunt-ended DSBs could target AID and DNA repair factors to the RGYW hotspot. The nature of this targeting process and the involved elements need to addressed, as does the nature of the endonuclease or cleavage factor responsible for effecting the initial DSB.

The appearance of DSBs causes retention of AID in the nucleus¹², suggesting DSBs or the related repair patches are substrates for AID binding. An absence of AID would abort inter-S-S-region recombination and result in an intra-S-region repair process, possibly through nonhomologous end-joining. In addition to processing blunt DNA ends, AID could be involved, perhaps together with cofactors, in S-S-region synapsis or could edit the putative recombinase that is involved in joining of DNA ends and effect inter-S-S-region recombination, thereby having substantial involvement in the synaptic or recombinogenic stage of CSR (Fig. 1c, ‘recombinase’). Future experiments should address the function of AID in S-S-region synapsis and other postcleavage recombination events. They should also assess and define the function of AID in mRNA editing, for which isolation of target mRNA is essential.