MOLECULAR AND CELLULAR BIOLOGY, Jan. 2006, p. 131–139
Copyright © 2006, American Society for Microbiology. All Rights Reserved.
Vol. 26, No. 1
Alternative Pathways for the Repair of RAG-Induced DNA Breaks
David M. Weinstock1and Maria Jasin2*
Department of Medicine1and Molecular Biology Program,2Memorial Sloan-Kettering Cancer Center,
1275 York Avenue, New York, New York 10021
Received 15 May 2005/Returned for modification 13 June 2005/Accepted 13 October 2005
RAG1 and RAG2 cleave DNA to generate blunt signal ends and hairpin coding ends at antigen receptor loci
in lymphoid cells. During V(D)J recombination, repair of these RAG-generated double-strand breaks (DSBs)
by the nonhomologous end-joining (NHEJ) pathway contributes substantially to the antigen receptor diversity
necessary for immune system function, although recent evidence also supports the ability of RAG-generated
breaks to undergo homology-directed repair (HDR). We have determined that RAG-generated chromosomal
breaks can be repaired by pathways other than NHEJ in mouse embryonic stem (ES) cells, although repair by
these pathways occurs at a significantly lower frequency than NHEJ. HDR frequency was estimated to be
>40-fold lower than NHEJ frequency for both coding end and signal end reporters. Repair by single-strand
annealing was estimated to occur at a comparable or lower frequency than HDR. As expected, V(D)J recom-
bination was substantially impaired in cells deficient for the NHEJ components Ku70, XRCC4, and DNA-PKcs.
Concomitant with decreased NHEJ, RAG-induced HDR was increased in each of the mutants, including cells
lacking DNA-PKcs, which has been implicated in hairpin opening. HDR was increased to the largest extent in
Ku70?/?cells, implicating the Ku70/80 DNA end-binding protein in regulating pathway choice. Thus, RAG-
generated DSBs are typically repaired by the NHEJ pathway in ES cells, but in the absence of NHEJ
components, a substantial fraction of breaks can be efficiently channeled into alternative pathways in these
Nonhomologous end-joining (NHEJ) of DNA double-
strand breaks (DSBs) created by the RAG recombinase during
V(D)J recombination generates a substantial portion of the
diversity found in antigen receptors (6, 11). The RAG recom-
binase, composed of the RAG1 and RAG2 proteins, initiates
recombination by introducing nicks at recombination signal
sequences (RSS elements), each composed of conserved hep-
tamer and nonamer sequences separated by a nonconserved
spacer of either 12 bp or 23 bp. Through a transesterification
reaction the nicks become converted to DSBs, resulting in two
hairpin coding ends and two blunt signal ends. The signal ends
undergo precise joining, whereas the hairpin coding ends un-
dergo further processing prior to joining, resulting in a diverse
set of junctions. In addition to the cleavage reaction, the RAG
proteins participate in end processing and joining, possibly by
maintaining the ends in a postcleavage complex, which serves
as a scaffold to facilitate repair (1, 14, 38, 48, 52).
The proteins involved in NHEJ of RAG-induced DSBs dur-
ing V(D)J recombination include the DNA end-binding pro-
tein Ku70/80, the Ser/Thr kinase DNA-PKcs, the XRCC4/
ligase IV complex, and the Artemis protein, which has
endonuclease activity (6, 24). Mutant cell lines for each of
these NHEJ proteins have severely impaired formation of cod-
ing joints during V(D)J recombination, whereas only Ku70/80-
and XRCC4/ligase IV-deficient cells are markedly impaired
for signal joint formation (2, 8, 12, 23, 40, 46). These proteins
are also involved in the repair of DSBs generated by ionizing
radiation and other DNA-damaging agents, since mutant cell
lines are hypersensitive to these agents (15, 40). In addition,
mutant cells can exhibit spontaneous chromosomal rearrange-
ments, in some cases leading to oncogenic translocations in
mice (7). NHEJ is not totally abrogated in cells deficient for
these components, however, since efficient joining of DSBs in
plasmids (see, e.g., reference 50) and a low level of V(D)J
recombination are observed. In addition, sequence analysis of
oncogenic translocation junctions derived from NHEJ mutant
mice indicates repair by nonhomologous processes (4, 53).
Thus, one or more poorly defined, alternative NHEJ pathways
exist in cells, which may be prone to causing rearrangements.
To distinguish NHEJ involving these alternative pathways
from NHEJ involving the components necessary for robust
V(D)J recombination, the latter pathway has often been
termed “classical” NHEJ.
NHEJ involves little or no sequence homology at the DSB.
Mammalian cells also efficiently repair DSBs by homologous
recombination, also termed homology-directed repair (HDR),
as demonstrated by studies using the rare-cutting endonucle-
ase I-SceI (22). Central to HDR is a DNA strand exchange
involving the invasion of single-stranded DNA into a homolo-
gous duplex DNA, a reaction that is promoted by the Rad51
protein (33, 45). HDR is generally considered to be a precise
form of repair, because it can restore the original sequence of
a damaged chromosome by using a homologous sequence,
preferably the sister chromatid (16), as a template for repair
synthesis. This contrasts with NHEJ, in which joining may be
imprecise (15, 24). In addition to HDR, a second pathway
involving sequence homology, termed single-strand annealing
(SSA), can mediate repair if sequence repeats are located near
the DSB (33, 45). Like HDR, SSA involves single strands
formed by resection at a DSB, but, unlike HDR, SSA occurs by
the annealing of the single strands rather than strand invasion.
* Corresponding author. Mailing address: Molecular Biology Pro-
gram, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue,
New York, NY 10021. Phone: (212) 639-7438. Fax: (212) 717-3317.
We investigated repair pathway choice of RAG-generated
chromosomal breaks using a mouse embryonic stem (ES) cell
system and NHEJ-deficient cell lines. Lee and colleagues have
recently demonstrated that the RAG proteins can induce
HDR at RSS elements (17). We now extend these studies and
find that HDR is induced by expression of full-length RAG
proteins using either a coding end or signal end chromosomal
reporter and that RAG-induced HDR requires the Rad51
strand exchange protein. RAG-generated breaks can also be
repaired by SSA. However, HDR and SSA occur at low fre-
quencies compared with NHEJ [i.e., V(D)J recombination]. In
addition, the frequency of HDR is substantially higher in
Ku70?/?, DNA-PKcs?/?, and XRCC4?/?ES cell lines, with the
highest frequency in Ku70?/?cells. These results indicate that
classical NHEJ factors not only mediate repair of RAG-gen-
erated breaks in these cells but also channel repair into V(D)J
recombination, thereby suppressing repair by HDR.
MATERIALS AND METHODS
DNA manipulations and cell transfections. The cDNAs for full-length RAG1,
RAG2, and RAG2-T490A were cloned from pcRAG-1, pcRAG-2 (25), and
pcRAG-2 T490A (21) into the pCAGGS expression vector (29) to create the
pCAGGS-RAG1, pCAGGS-RAG2, and pCAGGS-RAG2-T490A vectors, re-
spectively. The targeting vector hprtDRGFP-NICE was created by cloning a
fragment containing a 12-bp spacer RSS (5?-CACAGTGCTACAGACTGGAA
CAAAAACC-3?), NsiI site, BstBI site, and 23-bp spacer RSS (5?-GGTTTTTG
TACAGCCAGACAGTGGAGTACTACCACTGTG-3?) (13) into the unique
I-SceI site in the hprtDRGFP targeting vector (35), such that the RSS elements
are oriented to produce chromosomal coding ends upon RAG1/RAG2 cleavage.
The DRGFP-CE substrate was created by cloning a 333-bp intronic sequence
from the human ?-globin gene into the NsiI and BstBI sites in DRGFP-NICE to
expand the distance between the RSS elements. DRGFP-SE is identical to
DRGFP-CE, except the RSS elements are oriented to produce chromosomal
signal ends. DRGFP-CE, DRGFP-SE, and DRGFP-NICE were targeted to the
hypoxanthine phosphoribosyltransferase gene (hprt) locus in wild-type and
NHEJ mutant ES cells (9, 10, 12), as previously described (36). Targeting was
confirmed by Southern blotting (36). In addition to targeted clones, Ku70?/?
clones with a single copy of DRGFP-CE randomly integrated into the genome
(hprt?/puroR) were also isolated and were used to analyze individual repair
products after RAG1/RAG2 cleavage.
For HDR assays, 0.8 ?g of empty vector (pCAGGS), the I-SceI expression
vector pCBASce (39), or pCAGGS-RAG1 and pCAGGS-RAG2 was transfected
into ES cells in 24-well plates using Lipofectamine 2000 (Invitrogen, Inc.), ac-
cording to the manufacturer’s instructions. The same quantities of expression
vectors for Ku70 and Rad51-K133R (42) and pCAGGS-RAG2-T490A were used
in cotransfection experiments. Flow-cytometric analysis was performed on a
Becton Dickinson FACScan 96 h after transfection, as described previously (37).
PCR and sequencing. A combined PCR-Southern blotting strategy was used to
examine RAG-induced NHEJ, HDR, and SSA. Genomic DNA was isolated 7
days after transfection and used as the template for PCR with previously de-
scribed primers (31), except that amplification was performed for 22 cycles with
a 1 min 20 s elongation time. The products were separated on a 1.5% agarose gel,
transferred to a nylon membrane, and probed with a32P-labeled fragment from
DR-GFP, which was cut with HindIII/BamHI, as previously described (31).
Quantification was performed using ImageJ software (rsb.info.nih.gov/ij/).
To analyze individual repair products, genomic DNA was digested with NdeI
and MfeI, which cleave within the sequence between the RSS elements, and then
0.4 ?g was used as the template for PCR with primers DRGFP1 and DRGFP2,
as previously described (35). The PCR product (5 ?l) was subjected to TOPO TA
cloning (Invitrogen, Inc.) and transformed into Top10-competent Escherichia
coli. Individual colonies were isolated and analyzed by BcgI digestion and se-
quencing (Bio Resource Center, Cornell University, Ithaca, NY).
HDR induced by RAG recombinase. To determine the fre-
quency of RAG-induced HDR, we modified our green fluo-
rescence protein (GFP)-based reporter, DR-GFP, to contain
RSS elements. DR-GFP contains the cleavage site for the
rare-cutting I-SceI endonuclease within one of two defective
GFP repeats (Fig. 1A) (37). We inserted two RSS elements—
one containing a 12-bp spacer and the other containing a 23-bp
spacer—into the I-SceI site of DR-GFP, with the RSS ele-
ments separated by 333 bp of intronic sequence derived from
the human ?-globin gene. In DRGFP-SE, the RSS elements
are oriented such that RAG-mediated cleavage will produce
two blunt signal ends (SE) and excise a 333-bp fragment with
two hairpin coding ends. In DRGFP-CE, the RSS elements are
in the opposite orientation, such that RAG-mediated cleavage
produces two hairpin coding ends (CE) and excises a 400-bp
fragment with blunt signal ends. After cleavage, DNA ends
repaired by HDR through a noncrossover gene conversion, the
most frequent type of HDR event, will restore a functional
We introduced the HDR reporters into mouse ES cells.
Colonies that were hprt?/puroRwere selected, as each reporter
is flanked by hprt sequences to target the reporters to the hprt
locus (35) and, as well, contains the selectable puromycin re-
sistance gene (puroR) between the GFP repeats (Fig. 1A).
Correct targeting was confirmed by Southern hybridization
(Fig. 1B). With the HDR reporters integrated at the hprt locus
in each of the cell lines, possible position effects on cleavage or
repair that may arise from random integration should be ab-
In the absence of RAG1/RAG2 expression, GFP-positive
(GFP?) cells were rarely observed: in cells containing the
DRGFP-SE and DRGFP-CE reporters, 0.001 and 0.002% of
cells, respectively, were measured by flow cytometry to be
GFP?. However, after transient expression of full-length
RAG1/RAG2 into the appropriate cell lines, GFP?cells were
detected at significantly higher frequencies, indicating HDR
(Fig. 1C). An average of 0.026% of cells obtained from cells
containing the DRGFP-SE reporter were GFP?, and 0.073%
of cells obtained from cells containing the DRGFP-CE re-
porter were GFP?, an ?30-fold-higher level than that in the
absence of RAG1/RAG2 expression (Fig. 1D). For compari-
son, approximately 1.8% of cells obtained after I-SceI expres-
sion from cells containing the DR-GFP reporter were GFP?
(Fig. 1C), similar to our previous results with this reporter (35).
Thus, 70-fold (DRGFP-SE)- and 25-fold (DRGFP-CE)-lower
rates of HDR were obtained with RAG1/RAG2 expression
compared with I-SceI expression using similar reporter sub-
strates. The recovery of GFP?cells was dependent on the
HDR protein Rad51, as expression of a mutant Rad51 protein
(Rad51-K133R), which can interfere with HDR in a dominant
manner (42, 44), reduced the number of GFP?cells to back-
ground levels (0.001%) (Fig. 1C and D).
The RAG-induced GFP?cells formed a distinct population
of green fluorescent cells by flow cytometry and were obtained
at a substantially higher frequency than in the absence of
RAG1/RAG2 expression. However, given their low frequency,
we verified that they arose from HDR by enriching for the
GFP?population by flow sorting (Fig. 1C). Genomic DNA
was isolated from the sorted populations and confirmed to
contain the expected sequence for a functional GFP gene (data
132WEINSTOCK AND JASINMOL. CELL. BIOL.
phenotype associated with defective V(D)J coding end processing in Arte-
mis-deficient mice. Mol. Cell 10:1379–1390.
41. Roth, D. B., J. P. Menetski, P. B. Nakajima, M. J. Bosma, and M. Gellert.
1992. V(D)J recombination: broken DNA molecules with covalently sealed
(hairpin) coding ends in scid mouse thymocytes. Cell 70:983–991.
42. Stark, J. M., P. Hu, A. J. Pierce, M. E. Moynahan, N. Ellis, and M. Jasin.
2002. ATP hydrolysis by mammalian RAD51 has a key role during homol-
ogy-directed DNA repair. J. Biol. Chem. 277:20185–20194.
43. Stark, J. M., and M. Jasin. 2003. Extensive loss of heterozygosity is sup-
pressed during homologous repair of chromosomal breaks. Mol. Cell. Biol.
44. Stark, J. M., A. J. Pierce, J. Oh, A. Pastink, and M. Jasin. 2004. Genetic
steps of mammalian homologous repair with distinct mutagenic conse-
quences. Mol. Cell. Biol. 24:9305–9316.
45. Symington, L. S. 2002. Role of RAD52 epistasis group genes in homologous
recombination and double-strand break repair. Microbiol. Mol. Biol. Rev.
46. Taccioli, G. E., G. Rathbun, E. Oltz, T. Stamato, P. A. Jeggo, and F. W. Alt.
1993. Impairment of V(D)J recombination in double-strand break repair
mutants. Science 260:207–210.
47. Takata, M., M. S. Sasaki, E. Sonoda, C. Morrison, M. Hashimoto, H.
Utsumi, Y. Yamaguchi-Iwai, A. Shinohara, and S. Takeda. 1998. Homolo-
gous recombination and non-homologous end-joining pathways of DNA
double-strand break repair have overlapping roles in the maintenance of
chromosomal integrity in vertebrate cells. EMBO J. 17:5497–5508.
48. Tsai, C. L., A. H. Drejer, and D. G. Schatz. 2002. Evidence of a critical
architectural function for the RAG proteins in end processing, protection,
and joining in V(D)J recombination. Genes Dev. 16:1934–1949.
49. Van Dyck, E., A. Z. Stasiak, A. Stasiak, and S. C. West. 1999. Binding of
double-strand breaks in DNA by human Rad52 protein. Nature 398:728–
50. Verkaik, N. S., R. E. Esveldt-van Lange, D. van Heemst, H. T. Bruggenwirth,
J. H. Hoeijmakers, M. Z. Zdzienicka, and D. C. van Gent. 2002. Different
types of V(D)J recombination and end-joining defects in DNA double-
strand break repair mutant mammalian cells. Eur. J. Immunol. 32:701–709.
51. Yamamoto, A., T. Taki, H. Yagi, T. Habu, K. Yoshida, Y. Yoshimura, K.
Yamamoto, A. Matsushiro, Y. Nishimune, and T. Morita. 1996. Cell cycle-
dependent expression of the mouse Rad51 gene in proliferating cells. Mol.
Gen. Genet. 251:1–12.
52. Yarnell Schultz, H., M. A. Landree, J. X. Qiu, S. B. Kale, and D. B. Roth.
2001. Joining-deficient RAG1 mutants block V(D)J recombination in vivo
and hairpin opening in vitro. Mol. Cell 7:65–75.
53. Zhu, C., K. D. Mills, D. O. Ferguson, C. Lee, J. Manis, J. Fleming, Y. Gao,
C. C. Morton, and F. W. Alt. 2002. Unrepaired DNA breaks in p53-deficient
cells lead to oncogenic gene amplification subsequent to translocations. Cell
VOL. 26, 2006ALTERNATIVE REPAIR PATHWAYS FOR RAG-INDUCED DNA BREAKS139