J. Exp. Med.
Volume 195, Number 9, May 6, 2002 F37–F41
The Rockefeller University Press • 0022-1007/2002/05/F37/5 $5.00
The Function of AID in Somatic Mutation and Class Switch
Recombination: Upstream or Downstream of DNA Breaks
Katrin F. Chua, Frederick W. Alt, and John P. Manis
Howard Hughes Medical Institute and Children’s Hospital, Center for Blood Research and Department of Genetics,
Harvard Medical School, Boston, MA 02115
The immune system has evolved specific mechanisms to
combat a potentially limitless number of foreign pathogens
using a limited arsenal of Ig genes. To diversify the coding
potential of the Ig genes, B cells undergo several processes
of regulated genetic alterations. Early in their development,
B cells in the bone marrow undergo V(D)J recombination
to juxtapose variable region V, D, and J segments in differ-
ent combinations, creating a large repertoire of antibodies
(1). Later in B cell development, usually after antigen-
dependent activation of B cells, the genetic alteration pro-
cesses of somatic mutation (SM), class switch recombina-
tion (CSR), and gene conversion further diversify the
antigen-recognition repertoire as well as the effector func-
tion of encoded antibodies. In SM, which is the dominant
means of secondary alteration of variable region gene se-
quences in humans and mice, mutations are introduced in
the Ig variable region genes at a tremendous rate, which
allows for evolution of high affinity antibodies (2). In
some vertebrates, such as chickens and pigs, diversification
of assembled Ig variable regions occurs by a gene conver-
sion mechanism rather than SM (3). In CSR, to diversify
the effector function of specific antibodies, recombination
occurs within the downstream portion of the IgH locus to
join variable region genes with different constant (C
gion genes (4).
SM introduces mutations, small deletions, and insertions
at a high rate in a
2 kb region downstream of the Ig pro-
moter, altering the specificities of the encoded antibodies
(2). SM usually occurs within the specific microenviron-
ment of germinal centers, which is thought to be critical
for this process. Within germinal centers, antibodies with
high affinity for antigen are then selected, while low-affin-
ity antibodies are weeded out in a process termed affinity
maturation. The SM mutations commonly occur at con-
served sequence motifs (hotspots). The mechanism of SM
has been proposed to involve generation of DNA breaks
followed by a repair process that involves an error-prone
polymerase (5). In gene conversion, the assembled variable
region sequences are altered via homologous recombina-
tion using other unrearranged variable region genes or
pseudogenes as templates.
DNA breaks that occur during SM were first detected
by overexpressing the enzyme terminal deoxynucleotidyl
transferase (TdT), which catalyzes nontemplated addition
of nucleotides to free DNA ends, in a constitutively hyper-
mutating B cell line (6). This study revealed that nucle-
otides were specifically inserted at SM hotspots, suggesting
that these hotspots were sites of DNA breaks. Subse-
quently, three groups investigated the nature of these
breaks (single versus double-strand breaks, blunt versus
staggered or hairpin ends) using ligation-mediated PCR
(LM-PCR) strategies. These studies detected blunt-ended
double-stranded DNA breaks (DSBs) preferentially at hot-
spot sequences in B cells undergoing SM (7, 8). Papavasil-
iou and Schatz further showed that the majority of the
DSBs were detected during the G2 cell cycle phase when
the homologous recombination repair pathway is dominant
(7), while Bross et al. demonstrated the reliance of these
breaks on transcriptional activity (8). In addition, Kong and
Maizels also detected single-stranded DNA breaks at hy-
permutation sites (9). Although breaks in genomic DNA
can arise by a number of mechanisms, including apoptotic
DNA fragmentation and in vitro shearing, these studies
provided considerable evidence that the breaks detected by
the LM-PCR assays were associated with the SM mecha-
nism. Specifically, they showed that the breaks occurred
preferentially at SM hotspots, were dependent on transcrip-
tion, and occurred preferentially in hypermutating B cells.
CSR is another genetic modification employed by B
cells to boost the immune response by changing the con-
stant region of the antibody while retaining the antigen-
specific variable region. This DNA recombination event
occurs between two switch (S) regions, consisting of
stretches of repetitive sequence, located just upstream of
gene (except C
). CSR is a recombination/dele-
tion mechanism that juxtaposes a downstream C
) to the expressed V(D)J segment, allowing switching
from expression of IgM to IgG, IgA, or IgE (4). In vivo,
CSR requires germline transcription of S region sequences,
the generation of DSBs within S regions, and resolution of
these breaks by a process that requires NHEJ factors (4). It
has been suggested that CSR may, like SM, involve DNA
Address correspondence to John P. Manis, The Children’s Hospital, 300
Longwood Ave., Boston, MA 02115. Phone: 617-355-7290; Fax: 617-
738-0163; E-mail: firstname.lastname@example.org
synthesis by an error-prone polymerase, because mutations
have been detected near recombination junctions (10).
Recently, the discovery that the AID gene is required
for SM, gene conversion, and CSR has linked the mecha-
nisms of these three processes (11–14). AID encodes a cyti-
dine deaminase and shares sequence homology to the
RNA editing gene APOBEC-1 (15). Although cytidine
deaminase activity has been demonstrated for AID in vitro,
neither the function nor the substrate of AID is known.
Like APOBEC-1, AID could edit an RNA transcript to
change the function of the encoded protein, for example,
an endonuclease or error-prone polymerase. Another pos-
sibility is that AID acts on relevant RNA or DNA se-
quences, targeting them for recombination or mutation.
To date, there is no evidence for either possibility. The
function of AID has been studied in light of the three gen-
eral requirements (transcription, DNA breaks, and repair)
for CSR and SM. Analysis of germline transcription of Ig
exons after activation of CSR in AID-deficient B cells,
revealed that AID is not required for the transcription step.
Considering DNA repair, AID is also unlikely to function
in NHEJ during CSR, because V(D)J recombination,
which requires NHEJ, appears to be intact in AID-defi-
cient B cells (11). It has therefore been tempting to specu-
late that AID (or a RNA transcript edited by AID) func-
tions in DNA cleavage.
In this issue, two manuscripts describe the use of LM-
PCR strategies to ask whether AID is required for the
DSBs that they previously have found to be associated with
SM, and demonstrate, quite surprisingly, that the answer is
no (16, 17). Based on the current studies, Papavasiliou and
Schatz (16) propose that AID has a post-DNA cleavage
function, because they detect DNA breaks by LM-PCR in
germinal center B cells from AID-deficient mice, as well as
in a B cell line made functionally deficient for AID with a
dominant-negative protein. Bross et al. use a similar LM-
PCR assay to demonstrate DNA breaks in both germline
(unrearranged) and rearranged V gene segments of B cells
induced for SM, although SM is preferentially targeted to
rearranged V genes (17). Moreover, in the absence of AID,
no mutations were found despite an abundance of DNA
breaks. Thus, Bross et al. (17) also find that DSBs are gen-
erated on Ig variable region genes in cells undergoing so-
matic mutation in the absence of AID, and conclude that
these DSB are not sufficient to induce SM.
With respect to the observed DSBs and the role of AID,
Bross et al. provide a more equivocal interpretation than
Papavasiliou and Schatz, which includes three possible sce-
narios: (a) AID is upstream of DSBs in SM, and the DSBs
observed in their study are irrelevant to SM; (b) AID is up-
stream of DSBs in SM, but only a minor proportion of the
DSBs seen in wild-type B cells are relevant to the process
and these are not above the high background levels seen in
AID-deficient B cells; and (c) AID is downstream of the
breaks in SM as suggested by Papavasiliou and Schatz. Bross
et al. feel that possibility (a) is unlikely based on direct and
indirect evidence of others that DSBs are involved in SM
(2), and that possibility (c) is unlikely based on evidence
that AID may function upstream of DSBs in CSR (see be-
low). Thus, they appear to favor the second possibility,
which in actuality is not distinguishable by their studies
from possibility (a), in that most (and potentially all) DSBs
detected by their assays would be irrelevant to SM. If this
possibility were correct, it would raise the further questions
of why these DSBs are detected at such high and specific
levels, in both wild-type and AID-deficient cells, and
whether their previous conclusions that such DSBs are re-
lated to SM may need re-evaluation.
The findings of Papavasiliou and Schatz (16) and Bross et
al. (17) are particularly intriguing in light of recent work
suggesting that AID is required for the DNA breaks associ-
ated with CSR. Some form of DNA breaks are clearly in-
termediates in CSR, as the intervening DNA between re-
combining S regions is deleted (10). Previous studies
provided evidence that DSBs can be detected in S regions
by LM-PCR in cells activated for CSR (18), and there is
evidence that resolution of these breaks occurs by a process
that requires the non homologous end-joining factors (19–
22). To address the potential role of AID in CSR-related
DSBs, Petersen et al. (23) took advantage of the observa-
tion that the histone H2AX becomes phosphorylated
within seconds of DSB-inducing DNA damage, and the
phosphorylated H2AX proteins (
in nuclear foci that likely represent sites of damaged DNA.
The formation of
-H2AX foci is thought to represent a
very early event in the response to introduction of DSBs in
DNA (24). Petersen et al. found that
be detected at Ig C
genes in B cells undergoing CSR, but
not in B cells from AID-deficient mice (23). In the absence
of H2AX, CSR was diminished but still occurred at sub-
stantial levels. Based on these data, Petersen et al. reason-
ably argued that AID likely functions upstream of the DNA
modifications that initiate CSR. Examination of the factors
responsible for H2AX phosphorylation in CSR, as well as
its timing relative to the generation of DNA breaks will
certainly shed light on its role in this process. However,
one must also consider the possibility that the AID-depen-
-H2AX foci do not reflect the initial DNA lesion in
CSR, but rather are intermediates in the repair process it-
self. This could occur if resolution of an initial DNA
break is resolved via a replication-dependent mechanism, as
-H2AX foci have been correlated with stalled replication
forks (25). While Petersen et al. argued that CSR is per-
formed in G1 (23), there is other evidence that CSR is rep-
lication dependent, and that switch region sequences can
form secondary structures that could predispose to replica-
tional stress (10).
In their study, Petersen et al. also reported that mutations
accumulate in the S
region of wild-type cells after stimu-
lation but in the absence of CSR (23). Only background
levels of mutations were detected in similarly stimulated
AID-deficient B cells. Another report from Nagaoka et al.
(26) described similar results and showed that many of these
mutations are reminiscent of those found in SM, occur-
ring at similar hotspot sequence motifs. The Peterson et al.
(23) and Nagaoka et al. (26) studies argued that the S re-
-H2AX) can be detected
-H2AX foci could
Chua et al.
gion mutations are markers for the DNA lesions that ini-
tiate CSR, because they occur on alleles that had not un-
dergone bona fide CSR (e.g., did not delete the very 3
core region). They therefore proposed that the ab-
sence of the mutations in AID-deficient cells is evidence
that AID is required for DNA cleavage in the initiation of
CSR. Notably, however, previous studies showed that
CSR is accompanied by a very high level of mutations and
small deletions in regions flanking CSR junctions (27–29);
this was attributed to an error-prone repair process in the
resolution stage of CSR. In this regard, it is conceivable
that the much lower level of mutations observed by Pe-
tersen et al. and Nagoaka et al. might represent an exten-
sion of the same process. Thus, while the authors clearly
show that such mutations occur in the absence of CSR in-
volving two S regions, they have not ruled out possible
contributions of related intra-S region recombination. In
this regard, large internal S-region deletions, detectable by
Southern blotting, can accompany and precede actual CSR
between different S regions and are thought to occur by
the same mechanism as bona fide CSR (4). Notably, such
deletions occur at frequencies as high as 25–30% in IgM-
producing cells activated for CSR and at much higher lev-
els on unswitched alleles in IgG1- and IgG3-producing
cells (30–32). Related smaller deletions, undetectable by
Southern blotting, also occur. Thus it remains a formal
possibility that the S region mutations observed by Petersen
et al. and Nagoaka et al. could have been due to error-
prone resolution of internal S region recombination events.
Thus, more direct DSB assays will be required to draw de-
finitive conclusions about the role of AID in generating
CSR-related DSBs (6–9, 16–18, 23).
Taken together, these recent reports imply that while
AID is required for DSB formation in CSR, it is dispens-
able for the SM-associated DSBs detected by Papavasiliou
and Schatz, and Bross et al. (16, 17). One explanation for
these seemingly disparate findings is that despite the mech-
anistic similarities between these two processes, AID func-
tions at different steps of the CSR versus SM reactions,
e.g., involved in the generation of DSBs in CSR, but in
the repair of DSBs in SM. This could be the case if AID
edits two different mRNAs, one encoding a factor required
for SM and the other for CSR. The requirement for AID
in generating DNA breaks during Ig variable region gene
conversion has not yet been explored. Could AID function
after DNA cleavage in both CSR and SM? As proposed by
Papavasiliou and Schatz (16), AID could favor the use of an
error-prone DNA repair pathway during SM. In the ab-
sence of this AID-dependent pathway, the generation of
DNA breaks would be resolved by an error-free mecha-
nism, and mutations would not occur. In the case of CSR,
AID could also be involved in a post-cleavage event, if this
event is before formation of
could also involve recruitment of an error-prone repair
pathway, which would introduce mutations around the re-
gions of CSR junctions. In the absence of AID, DNA
breaks would be sealed correctly and preclude mutations
and recombination. Finally, although the Papavasiliou and
-H2AX foci. This function
Schatz (16) and Bross et al. (17) reports provide strong evi-
dence that AID is dispensable for detection of DSBs in Ig
variable regions genes in cells stimulated for SM, it remains
formally possible that AID functions upstream of the breaks
that initiate SM. In this case, one would have to propose
that although most or all DSBs detected in these studies are
very specifically correlated with essentially all known as-
pects of SM, they are nonetheless not directly related to the
SM mechanism (16).
One important issue raised by the possibility that AID is
required for DNA cleavage in SM, CSR, and gene conver-
sion is how AID-initiated breaks in each process result in
different outcomes. It is possible that the breaks are gener-
ated during different phases of the cell cycle, when differ-
ent repair pathways are preferentially employed for DSB
repair. Indeed, CSR has been argued to occur during G1
(23), when NHEJ is preferentially active, whereas SM has
been argued to occur during G2 (7), when homologous re-
combination pathways are activated. It may be that SM
breaks can be repaired by homologous recombination
pathways in G2 because of the presence of a sister chroma-
tid (7), while CSR breaks occur in G1 in the absence of
such a template and therefore are resolved by NHEJ. Addi-
tionally, SM was not impaired in stimulated B cells in the
absence of the NHEJ factor DNA-PKcs, which is required
for efficient CSR (19, 22, 33). More generally, the means
by which DSBs are resolved may be modulated by factors
that preferentially activate one repair pathway over an-
other. Such a model, where tipping the balance of repair
factors alters the outcome of DNA breaks, has been re-
cently proposed in a study of gene conversion in the DT40
chicken B cell line (34). This cell line undergoes gene con-
version at a high frequency, but when made deficient for
the homologous repair factors XRCC2/3, the cell line in-
stead diversified its assembled Ig light chain variable region
gene using a SM-like mechanism. If AID functions in a
post-cleavage event, it could play a role in altering the bal-
ance of factors that are available to repair a break. For ex-
ample, Papavasiliou and Schatz (16) argue that in the ab-
sence of AID, the SM-associated DSBs cannot be repaired
by error-prone pathways, and are therefore repaired by er-
The factors affecting the outcome of Ig locus DNA
breaks may well be generally expressed and found in many
cell types. Evidence for this was recently demonstrated in
SM studies in plasma cell lines that do not express AID and
normally do not undergo SM (35). Expression of AID in
these cells allowed SM to occur, indicating that AID is the
only factor required for SM that is normally missing in
these cells. Remarkably, another recent study showed that
AID is the only B cell specific factor required for CSR, be-
cause expression of AID in a fibroblast cell line (NIH3T3)
was sufficient to cause recombination of a model CSR sub-
strate (36). However, it is clear that within endogenous loci
in normal lymphocytes, the factors and processes associated
with germline transcription are also required (4), providing
two separate levels for control of CSR in activated B cells.
The finding that AID can effect CSR in transcribed S re-
gions in nonlymphoid cells raises the intriguing possibility
that, regardless of whether AID functions upstream or
downstream of DNA breaks, ectopic or dysregulated ex-
pression of AID could predispose to tumorigenesis. Specifi-
cally, ectopically expressed AID could generate DNA
breaks that predispose to translocation, or alternatively,
could inactivate specific repair pathways that suppress trans-
location. Clearly the recent studies examining DNA breaks
in CSR and SM in the absence of functional AID have set
the groundwork for further investigation of the role of this
protein in regulated genetic modifications.
This work is supported by National Institutes of Health grant
(AI31541 to F. Alt), by a Charles Hood Foundation and Lym-
phoma Research Foundation grant (to J. Manis), and by a Jane
Coffin Childs Memorial Fund for Medical Research fellowship (to
Submitted: 10 March 2002
Revised: 19 March 2002
Accepted: 20 March 2002
1. Hesslein, D.G., and D.G. Schatz. 2001. Factors and forces
controlling V(D)J recombination.
2. Jacobs, H., and L. Bross. 2001. Towards an understanding of
Curr. Opin. Immunol.
3. Sonoda, E., C. Morrison, Y.M. Yamashita, M. Takata, and S.
Takeda. 2001. Reverse genetic studies of homologous DNA
recombination using the chicken B-lymphocyte line, DT40.
Philos. Trans. R. Soc. Lond. B Biol. Sci.
4. Manis, J.P., M. Tian, and F.W. Alt. 2002. Mechanism and
control of class-switch recombination.
5. Brenner, S., and C. Milstein. 1966. Origin of antibody varia-
6. Sale, J.E., and M.S. Neuberger. 1998. TdT-accessible breaks
are scattered over the immunoglobulin V domain in a consti-
tutively hypermutating B cell line.
7. Papavasiliou, F.N., and D.G. Schatz. 2000. Cell-cycle-regu-
lated DNA double-stranded breaks in somatic hypermutation
of immunoglobulin genes.
8. Bross, L., Y. Fukita, F. McBlane, C. Demolliere, K. Rajew-
sky, and H. Jacobs. 2000. DNA double-strand breaks in im-
munoglobulin genes undergoing somatic hypermutation.
9. Kong, Q., and N. Maizels. 2001. DNA breaks in hypermu-
tating immunoglobulin genes: evidence for a break-and-
repair pathway of somatic hypermutation.
10. Kinoshita, K., and T. Honjo. 2001. Linking class-switch re-
combination with somatic hypermutation.
11. Muramatsu, M., K. Kinoshita, S. Fagarasan, S. Yamada, Y.
Shinkai, and T. Honjo. 2000. Class switch recombination
and hypermutation require activation-induced cytidine de-
aminase (AID), a potential RNA editing enzyme.
12. Arakawa, H., J. Hauschild, and J.M. Buerstedde. 2002. Re-
quirement of the activation-induced deaminase (AID) gene
for immunoglobulin gene conversion.
Nat. Rev. Mol.
13. Revy, P., T. Muto, Y. Levy, F. Geissmann, A. Plebani, O.
Sanal, N. Catalan, M. Forveille, R. Dufourcq-Labelouse, A.
Gennery, et al. 2000. Activation-induced cytidine deaminase
(AID) deficiency causes the autosomal recessive form of the
Hyper-IgM syndrome (HIGM2).
14. Harris, R.S., J.E. Sale, S.K. Petersen-Mahrt, and M.S. Neu-
berger. 2002. AID is essential for immunoglobulin V gene
conversion in a cultured B cell line.
15. Muramatsu, M., V.S. Sankaranand, S. Anant, M. Sugai, K.
Kinoshita, N.O. Davidson, and T. Honjo. 1999. Specific ex-
pression of activation-induced cytidine deaminase (AID), a
novel member of the RNA-editing deaminase family in ger-
minal center B cells.
J. Biol. Chem.
16. Papavasiliou, F.N., and D.G. Schatz. 2002. The activation-
induced deaminase functions in a postcleavage step of the so-
matic hypermutation process.
17. Bross, L., M. Muramatsu, K. Kinoshita, T. Honjo, and H. Ja-
cobs. 2002. DNA double-strand breaks: prior to but not suf-
ficient in targeting hypermutation.
18. Wuerffel, R.A., J. Du, R.J. Thompson, and A.L. Kenter.
1997. Ig Sgamma3 DNA-specific double strand breaks are
induced in mitogen-activated B cells and are implicated in
19. Rolink, A., F. Melchers, and J. Andersson. 1996. The SCID
but not the RAG-2 gene product is required for S mu-S ep-
silon heavy chain class switching.
20. Casellas, R., A. Nussenzweig, R. Wuerffel, R. Pelanda, A.
Reichlin, H. Suh, X.F. Qin, E. Besmer, A. Kenter, K. Ra-
jewsky, and M.C. Nussenzweig. 1998. Ku80 is required for
immunoglobulin isotype switching.
21. Manis, J.P., Y. Gu, R. Lansford, E. Sonoda, R. Ferrini, L.
Davidson, K. Rajewsky, and F.W. Alt. 1998. Ku70 is re-
quired for late B cell development and immunoglobulin
heavy chain class switching.
J. Exp. Med.
22. Manis, J.P., D. Dudley, L. Kaylor, and F.W. Alt. 2002. IgH
class switch recombination to IgG1 in DNA-PKcs-deficient
. In press.
23. Petersen, S., R. Casellas, B. Reina-San-Martin, H.T. Chen,
M.J. Difilippantonio, P.C. Wilson, L. Hanitsch, A. Celeste,
M. Muramatsu, D.R. Pilch, et al. 2001. AID is required to
initiate Nbs1/gamma-H2AX focus formation and mutations
at sites of class switching.
24. Paull, T.T., E.P. Rogakou, V. Yamazaki, C.U. Kirchgessner,
M. Gellert, and W.M. Bonner. 2000. A critical role for his-
tone H2AX in recruitment of repair factors to nuclear foci af-
ter DNA damage.
25. Ward, I.M., and J. Chen. 2001. Histone H2AX is phos-
phorylated in an ATR-dependent manner in response to rep-
J. Biol. Chem.
26. Nagaoka, H., M. Muramatsu, N. Yamamura, K. Kinoshita,
and T. Honjo. 2002. Activation-induced deaminase (AID)-
directed hypermutation in the immunoglobulin Smicro re-
gion: implication of AID involvement in a common step of
class switch recombination and somatic hypermutation.
27. Dunnick, W., M. Wilson, and J. Stavnezer. 1989. Mutations,
duplication, and deletion of recombined switch regions sug-
gest a role for DNA replication in the immunoglobulin
Mol. Cell. Biol.
28. Dunnick, W., and J. Stavnezer. 1990. Copy choice mecha-
nism of immunoglobulin heavy-chain switch recombination.
J. Exp. Med
J. Exp. Med
Chua et al.
Mol. Cell. Biol.
29. Dunnick, W., G.Z. Hertz, L. Scappino, and C. Gritzmacher.
1993. DNA sequences at immunoglobulin switch region re-
Nucleic Acids Res.
30. Winter, E., U. Krawinkel, and A. Radbruch. 1987. Directed
Ig class switch recombination in activated murine B cells.
31. Gu, H., Y.R. Zou, and K. Rajewsky. 1993. Independent
control of immunoglobulin switch recombination at individ-
ual switch regions evidenced through Cre-loxP-mediated
32. Bottaro, A., F. Young, J. Chen, M. Serwe, F. Sablitzky, and
F.W. Alt. 1998. Deletion of the IgH intronic enhancer and
associated matrix-attachment regions decreases, but does not
abolish, class switching at the mu locus.
33. Bemark, M., J. Sale, H. Kim, C. Berek, R. Cosgrove, and M.
Neuberger. 2000. Somatic Hypermutation in the absence of
DNA-dependent protein kinase catalytic subunit (DNA-
) or recombination-activating gene (RAG) 1 activity.
34. Sale, J.E., D.M. Calandrini, M. Takata, S. Takeda, and M.S.
Neuberger. 2001. Ablation of XRCC2/3 transforms immu-
noglobulin V gene conversion into somatic hypermutation.
35. Martin, A., P.D. Bardwell, C.J. Woo, M. Fan, M.J. Shulman,
and M.D. Scharff. 2002. Activation-induced cytidine deami-
nase turns on somatic hypermutation in hybridomas.
36. Okazaki, I., K. Kinoshita, M. Muramatsu, K. Yoshikawa, and
T. Honjo. 2002. The AID enzyme induces class switch re-
combination in fibroblasts.