The roots of antibody diversity

Article metrics

When faced with foreign molecules our antibodies mutate, allowing them to bind to the intruders more strongly. In a story full of surprises, it looks as though the mechanism of mutation has finally been revealed.

One of the tactics our immune system uses to fight off intruders is the production of protective antibody proteins, which recognize and neutralize foreign molecules. Antibodies are secreted by the B cells of the immune system, and are unique among our proteins in their unlimited potential for diversity. Half of each antibody molecule is encoded by the so-called variable genes, and when a B cell recognizes and responds to a foreign molecule, these variable genes undergo mutation at a tremendously high frequency.

In a stunning set of papers, culminating in that on page 43 of this issue1, Neuberger and colleagues seem to have discovered the elusive mechanism behind this 'hypermutation' — and the answer is totally unexpected. First they showed that the activation-induced cytosine deaminase (AID) protein, which is needed for hypermutation2, leads to DNA mutations when expressed in bacteria3. More mutations of the cytosine 'letters' in DNA occur in bacteria that lack the enzyme uracil glycosylase, hinting that AID converts cytosines to uracils. (Normally, these erroneous uracils would be removed by the glycosylase before they caused a problem.) These findings are now followed1 by a convincing demonstration that the generation of uracil from cytosine in B-cell DNA sparks hypermutation, which occurs as the cells try to fix the errors.

After vaccination or infection, the body first produces antibodies of relatively low affinity for the foreign molecule, the antigen. As an immune response progresses, these antibodies become hypermutated. The purpose of hypermutation is to create new protein sequences that can bind the antigen more strongly and specifically than their precursors. This allows us to respond quickly and effectively to pathogens that we have encountered previously. People who cannot mutate their antibodies suffer recurring bacterial and viral infections, and do not respond to vaccination.

Hypermutation occurs only in B cells that have been stimulated with antigen. During the development of B cells, before they encounter antigen, genetic shuffling has recombined blocks of DNA into long segments representing the basic antibody-encoding genes. Then, during hypermutation, many single mutations — and, rarely, small insertions and deletions — are introduced into the variable regions of these genes.

A breakthrough in understanding the mechanism behind hypermutation came two years ago, when Muramatsu et al.2 discovered that the AID protein — which is expressed only in hypermutating B cells — is needed for the process. At the time it was thought that AID might convert ('deaminate') cytosine to uracil in the messenger RNA encoding an endonuclease enzyme, thereby enabling an active form of the enzyme to be produced. Endonucleases cut nicks in DNA strands, and it was proposed that nicks generated by this putative endonuclease in the variable regions of antibody genes could be the lesions that trigger hypermutation. This notion was attractive because numerous DNA breaks had been detected around variable genes. But the theory was questioned when similar numbers of breaks were found in mice lacking AID4,5. Researchers now suspect that these nicks are unrelated to hypermutation.

A second possibility, proposed by Neuberger and colleagues, was that AID converts cytosine to uracil in DNA (not RNA), as suggested by the increase in mutations of cytosines in AID-expressing bacteria3. In this model, AID produces a lot of uracils in the variable genes (Fig. 1a). The uracils are mistakes, as the usual bases in DNA are cytosine, guanine, adenine and thymine, and need to be corrected — but that often leads to more errors. One 'correction' mechanism involves the DNA polymerase enzymes that copy DNA, which mistakenly read uracil as thymine during replication, and introduce errors accordingly (Fig. 1b). Other correction mechanisms rely on the uracil first being removed by uracil glycosylase, leaving a gap. Mutations can then occur as the cell fills in the gap by one of three pathways — one DNA-replication and two DNA-repair pathways (Fig. 1c–e).

Figure 1: A model for antibody diversity.

a, The AID protein converts cytosines (C) in the antibody-gene variable region to uracils (U). These can be processed in several ways to generate mutations. b, High-fidelity DNA-replicating enzymes (polymerases) recognize U as thymine (T), and pair it with adenine (A) during replication. c–e, These pathways rely on the initial removal of U by uracil glycosylases. c, Low-fidelity polymerases bypass the resulting gap during replication, filling it more or less at random. One example is shown. d, In this repair pathway, a T is preferentially inserted in the gap, leading to errors after replication. e, In this repair process, the gap is filled in and extended, generating mutations that are mostly located opposite As and Ts. In chicken B cells lacking the protein XRCC2, the replication pathways predominate. All four pathways work in mice and humans. 'Transitions' swap one purine base (G or A) for the other; likewise for pyrimidines (C, T and U). 'Transversions' swap a pyrimidine for a purine, or vice versa. Blue lines, old DNA strands; orange lines, new strands produced after DNA replication.

Di Noia and Neuberger1 now provide support for the conversion of cytosine to uracil in DNA in eukaryotic cells. They took advantage of a chicken B-cell line engineered previously by Neuberger's group. These cells lack XRCC2, a protein involved in DNA recombination and repair, and so tend to deal with uracil in DNA by the replication pathways (Fig. 1b, c), not by DNA repair. Moreover, they mainly generate transversion-type mutations in the variable genes. In transversions, a pyrimidine base (cytosine, thymine or uracil) is swapped for a purine (adenine or guanine). In transitions, one pyrimidine is swapped for another, or one purine for another. The transversions in these B cells generally involve replacements of cytosine or guanine (rather than of adenine or thymine), and this probably occurs by the replication pathway that relies on the initial removal of uracil by the glycosylase (Fig. 1c).

The authors' experiments were simple yet effective. The uracil glycosylase was inhibited in the B cells, forcing the cells to deal with uracils through the other replication pathway. By its nature this pathway generates transitions of an initial cytosine–guanine base pair to a thymine–adenine pair (Fig. 1b). Di Noia and Neuberger reasoned that if the production of uracil in variable-region genes is the trigger for hypermutation, and if cells cannot deal with the uracil by removing it but instead copy over it, then the predominant mutation pattern would shift from transversions of cytosine and guanine to transitions. And this is exactly what they saw.

So what about hypermutation in mammalian B cells? Again, the generation of uracil could be the crucial starting point. In hypermutating mammalian B cells all four nucleotides are mutated with similar frequencies in variable genes, and transitions are favoured. All replication and repair pathways could be used to process AID-generated uracils, causing these patterns of hypermutation. After uracil is removed by either uracil glycosylase or another glycosylase, SMUG1, processing may be shunted into the repair pathways, allowing mutations of adenine and thymine — rather than just cytosine and guanine — to occur (Fig. 1d, e). During error-prone base-excision repair (Fig. 1d), the gap may be incorrectly filled in by DNA polymerase ι (ref. 6), which erroneously inserts thymine opposite guanine7. During short-patch repair synthesis (Fig. 1e), polymerase η could fill in the gap and synthesize several bases by strand displacement opposite adenines and thymines8,9. The mismatch-repair proteins MSH2 and MSH6, and polymerase ζ, may also be involved10,11,12.

The identification of AID-generated uracils as the likely starting point is a major breakthrough in understanding hypermutation. Nonetheless, questions remain. The most important one is how AID targets the rearranged variable genes. Perhaps AID is delivered to these genes by interacting with transcription factors13. Also, why is the usually accurate base-excision repair using polymerase β subverted to mutation-generating repair with polymerase ι in variable genes? What is the exact role of MSH2 and MSH6? AID also initiates two other antibody-alteration processes: 'class switch recombination' in mammals, and variable-gene conversion in chickens. How can the DNA-deaminase model explain these very different events? Finally, before the model is totally convincing it must be shown that AID actually does deaminate cytosine in DNA, and that uracils are present in the variable genes. But these are exciting times for immunology, and it appears that one of the last black boxes has been opened.


  1. 1

    Di Noia, J. & Neuberger, M. S. Nature 419, 43–48 (2002); advance online publication, 31 July 2002 (doi:10.1038/nature00981).

  2. 2

    Muramatsu, M. et al. Cell 102, 553–563 (2000).

  3. 3

    Petersen-Mahrt, S. K., Harris, R. S. & Neuberger, M. S. Nature 418, 99–103 (2002).

  4. 4

    Bross, L., Muramatsu, M., Kinoshita, K., Honjo, T. & Jacobs, H. J. Exp. Med. 195, 1187–1192 (2002).

  5. 5

    Papavasiliou, F. N. & Schatz, D. G. J. Exp. Med. 195, 1193–1198 (2002).

  6. 6

    Bebenek, K. et al. Science 291, 2156–2159 (2001).

  7. 7

    Frank, E. G. et al. EMBO J. 20, 2914–2922 (2001).

  8. 8

    Zeng, X. et al. Nature Immunol. 2, 537–541 (2001).

  9. 9

    Pavlov, Y. I. et al. Proc. Natl Acad. Sci. USA 99, 9954–9959 (2002).

  10. 10

    Phung, Q. H. et al. J. Exp. Med. 187, 1745–1751 (1998).

  11. 11

    Wiesendanger, M., Kneitz, B., Edelmann, W. & Scharff, M. D. J. Exp. Med. 191, 579–584 (2000).

  12. 12

    Diaz, M., Verkoczy, L. K., Flajnik, M. F. & Klinman, N. R. J. Immunol. 167, 327–335 (2001).

  13. 13

    Peters, A. & Storb, U. Immunity 4, 57–65 (1996).

Download references

Author information

Correspondence to Patricia J. Gearhart.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Gearhart, P. The roots of antibody diversity. Nature 419, 29–31 (2002) doi:10.1038/419029a

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.