A new study by Honjo and colleagues, recently published in Science, adds to the controversy about the role of activation-induced cytidine deaminase (AID) in class-switch recombination (CSR) — the process by which antibodies switch their isotype from IgM to IgG, IgA or IgE by recombination at the heavy-chain (IgH) loci. AID is known to be required for the double-stranded DNA breaks that allow recombination, but the way in which it achieves this is hotly debated.

The most widely accepted model (see Further reading) proposes that AID deaminates cytosine residues to uracil residues in immunoglobulin-gene DNA; the uracil residues are then removed by the DNA base-excision repair enzyme uracil-DNA glycosylase (UNG), resulting in strand breaks. However, Honjo and colleagues have proposed an alternative model — on the basis of similarities between AID and the mRNA-editing enzyme APOBEC1 — in which AID deaminates an unknown mRNA such that it encodes a novel endonuclease that cleaves the DNA. By looking at the temporal relationship between DNA breaks and UNG activity, they now put forward further evidence in favour of the RNA-editing model.

In these experiments, phosphorylated histone H2AX (γ-H2AX), which forms next to double-stranded DNA breaks, was used as a marker for cleavage sites involved in CSR. When a mouse lymphoma cell line was stimulated to switch to IgA production, γ-H2AX accumulated at the IgH loci. Expression of the UNG inhibitor Ugi by these cells reduced the level of CSR to IgA but did not affect γ-H2AX accumulation. Because γ-H2AX focus formation at the IgH loci does not depend on UNG, but CSR is reduced in the absence of UNG, this indicates that UNG has a role in CSR downstream of DNA cleavage, which contradicts the DNA-deamination model. In support of this, there was no difference in the IgH mutation rate between cells expressing or not expressing Ugi; because mutation results from DNA cleavage during CSR, it seems likely that UNG is not required for DNA cleavage.

UNG mutants that lack UNG catalytic activity but retain DNA-binding activity were shown to recover wild-type levels of CSR in UNG-deficient B cells. The importance of UNG for CSR in the absence of its catalytic activity indicates that it might have a structural role. Honjo and colleagues elaborate on their RNA-editing model by suggesting that UNG functions as a scaffold to recruit repair proteins to the DNA after DNA cleavage caused by an AID-processed endonuclease. The role of such an endonuclease is also supported by a further paper from Honjo's group showing that de novo protein synthesis is required for DNA cleavage in CSR.