Trex1 Exonuclease Degrades ssDNA
and Autoimmune Disease
Yun-Gui Yang,1Tomas Lindahl,1and Deborah E. Barnes1,*
1Cancer Research UK London Research Institute, Clare Hall Laboratories, South Mimms, Hertfordshire EN6 3LD, UK
Trex1 is the major 30DNA exonuclease in mam-
gene can cause Aicardi-Goutie `res syndrome,
characterized by perturbed immunity. Similarly,
Trex1?/?mice have an autoinflammatory phe-
notype; however, the mechanism of Trex1-defi-
cient disease is unknown. We report that Trex1,
ordinarily associated with the endoplasmic re-
ticulum (ER), relocalizes to the S phase nucleus
after g irradiation or hydroxyurea treatment.
Notably, Trex1-deficient cells show defective
G1/S transition and chronic ATM-dependent
checkpoint activation, even in the absence of
exogenous stress, correlating with persistent
single-stranded DNA molecules produced in S
phase, which accumulate in theER. Our data in-
polynucleotide species generated from pro-
cessing of aberrant replication intermediates
to attenuate DNA damage checkpoint signaling
and prevent pathological immune activation.
Trex1, the major 30/ 50DNA exonuclease in mammalian
cells, acts preferentially on single-stranded DNA (ssDNA),
to which it binds avidly. Trex1 is not found in yeasts and,
although a ‘‘Trex’’ ortholog is detected in the Xenopus
and Drosophila genomes, appears to have arisen from
a recent gene duplication. Hence, TREX1 and the related
TREX2 open reading frame (sharing 44% amino acid iden-
tity within the exonuclease domain but lacking a unique
C-terminal extension present in TREX1) are only docu-
2001a) is very closely linked (in both mouse and man)
downstream of the gene encoding the ATR-interacting
protein ATRIP (Cortez et al., 2001) and tail to tail with the
gene encoding the p53-inducible proapoptotic protein
a mismatched 30terminal deoxyribonucleotide at a DNA
strand break, and weak homology with known editing
enzymes, suggested that it may serve a DNA-editing role
in DNA replication or gap filling during DNA repair (Ho ¨ss
et al., 1999; Mazur and Perrino, 2001b). We generated
gene-targeted Trex1?/?knockout mice in order to eluci-
date the function of the enzyme in vivo (Morita et al.,
ous mutation frequency or cancer incidence that would
indicate editing of DNA mismatches by Trex1 in vivo.
Rather, Trex1 null mice displayed inflammatory myocardi-
tis of apparently autoimmune etiology, leading to dilated
cardiomyopathy and a dramatically reduced lifespan.
In a subsequent development, TREX1 mutations have
been identified in patients with the inherited disease
Aicardi-Goutie `res syndrome (AGS), leading to loss of
TREX1 exonuclease activity in cell lines from patients
with homozygous mutations at the AGS1 locus (Crow
et al., 2006a). AGS presents as a severe neurological syn-
drome but has phenotypic overlap with the autoimmune
disease systemic lupus erythematosus (SLE; Aicardi and
Goutie `res, 2000). More recently, certain heterozygous
TREX1 mutations have been implicated in SLE, as well
as dominant AGS and other distinct monogenic diseases
(Lee-Kirsch et al., 2007a, 2007b; Rice et al., 2007a; Ri-
chards et al., 2007). The encephalopathy in AGS is, like
the cardiomyopathy of Trex1?/?mice, associated with
an autoinflammatory response in the absence of infection
(Goutie `res et al., 1998; Morita et al., 2004). Elevated levels
of interferon alpha (IFN-a) in AGS and SLE (Alarco ´n-
Riquelme, 2006; Banchereau and Pascual, 2006; Gou-
tie `res et al., 1998) are reminiscent of antiviral-like immune
responses (Stetson and Medzhitov, 2006), and Trex1-
deficient disease can be considered a recessive genetic
mimic of congenital viral infection (Crow et al., 2006a).
As AGS may also be caused by mutations in any of the
three subunits of the RNaseH2 holoenzyme (Crow et al.,
2006b), a common nucleic acid substrate, or activation
of convergent signaling pathways, might underscore this
genetically heterogeneous disease (Rice et al., 2007b).
ase from cell nuclei (Ho ¨ss et al., 1999; Mazur and Perrino,
1999), is predominantly localized within the cytoplasmic
compartment but can be mobilized to the cell nucleus,
Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc. 873
874 Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc.
apparently as part of a caspase-independent cell death
pathway (Chowdhury et al., 2006). The crystal structure
of Trex1 explains the specificity of the enzyme and the
molecular basis for understanding mutations that lead to
disease (Brucet et al., 2007; de Silva et al., 2007). How-
ever, the in vivo substrate of Trex1, and the mechanism
whereby persistence of this nucleic acid substrate in the
absence of Trex1 exonuclease activity provokes patho-
logical activation of the immune system, is unclear.
Here, we have characterized the cellular phenotype of
a Trex1?/?mouse embryo fibroblast (MEF) cell line, dem-
onstrating chronic DNA damage checkpoint activation in
cells lacking Trex1 and extranuclear accumulation of an
endogenous ssDNA substrate, with consequences for
Trex1 Relocalizes to Replication Foci
in the Cell Nucleus after g Irradiation
or Hydroxyurea Treatment
Trex1 has a perinuclear distribution in mammalian cells
and is colocalized with ER-associated proteins (Chowd-
hury et al., 2006). As Trex1 has the properties of a 30
DNA repair exonuclease, we examined whether the
protein might relocate to the nucleus upon treatment of
cells with DNA-damaging agents. Wild-type MEFs were
transfected with an epitope-tagged GFP-Trex1 construct
and treated with ionizing radiation (IR), and the subcellular
localization of GFP-Trex1 was monitored by fluorescence
microscopy (Figure 1A). Nonirradiated cells showed
ER-associated localization of GFP-Trex1, with no fluores-
cence detectable in the nucleus. However, ectopically
expressed Trex1 protein partially relocalized from the ER
to form foci in the nucleus after cellular irradiation with
sublethal doses of IR. Similar results were obtained in re-
sponse to reduction of deoxyribonucleoside triphosphate
pools by hydroxyurea (HU; Figure 1A), indicating that
Trex1 is specifically recruited to stalled replication forks.
Furthermore, in synchronized cell populations, IR- or
HU-induced relocalization of Trex1 to nuclear foci oc-
curred specifically in S phase and was not seen in
G0/G1 cells (Figure 1B). In order to further define the sites
to which Trex1 is recruited in the nucleus, cells expressing
epitope-tagged Trex1 were labeled in S phase with bro-
modeoxyuridine (BrdU) prior to IR or HU treatment. Strik-
ingly, Trex1 within the nucleus was colocalized with BrdU-
labeled DNA (Figure 1C); 70%–80% of BrdU foci were
merged with Trex1 foci in dual-labeled cells after 90 min,
and this did not increase significantly up to 180 min post-
irradiation (data not shown). These results indicate that
Trex1 is specifically recruited to replication foci after g
radiation or HU stress.
Trex1?/?MEFs Show Abnormal Responses
after g Irradiation
We examined the cellular response to IR of Trex1?/?ver-
sus Trex1+/+MEFs. Wild-type Trex1+/+MEFs displayed
dose-dependent cell killing by IR, as shown by fluores-
cence-activated cell sorting (FACS), with the appearance
of a sub-G1 peak and concomitant reduction in the G1
population (Figures 1D and 1E); time-course analysis
showed that the percentage of sub-G1 cells peaked
sharply at 22–24 hr postirradiation (data not shown). In
marked contrast, Trex1?/?MEFs appeared resistant to
cell killing by IR, as there was no dose-dependent appear-
ance of a sub-G1 population (Figures 1D and 1E), even 72
hr postirradiation at 4 Gy (data not shown). The G2/M
checkpoint may fail to operate correctly in Trex1?/?
MEFs after g irradiation, provoking cell-cycle arrest in
subsequent G0/G1 rather than apoptosis. Furthermore,
nonirradiated Trex1?/?MEFs appeared to have a reduced
proportion of cells in S phase compared to the Trex1+/+
control (0 Gy; Figure 1D).
Trex1?/?MEFs Are Defective in G1/S Transition
and IR-Induced G2 Arrest
To evaluate a possible cell-cycle defect in nonirradiated
Trex1?/?cells, we determined the proportion of cells in
each phase of the cycle in asynchronous cultures of
Trex1?/?versus Trex1+/+MEFs. Quantitative FACS analy-
sis revealed a greatly reduced proportion of Trex1?/?cells
in S phase compared with the Trex1+/+wild-type control
and significant accumulation in G0/G1 (Figure 2A). In
the Trex1?/?MEF cell line proliferates more slowly in
exponential culture (Figure 2B), with an increased cell
doubling time (?36 hr) compared to the normal Trex1+/+
control (?20 hr).
In order to follow a defined population throughout the
cell cycle, cells undergoing DNA replication in S phase
were pulse labeled with BrdU, and the progression of
Figure 1. Relocalization of Epitope-Tagged Trex1 and Abnormal Responses of Trex1?/?MEFs after Exogenous Stress
(A)LocalizationofTrex1. Wild-typeMEFs weretransfected withtheepitope-tagged Trex1expressionconstruct, treated withor without IRorHU24hr
posttransfection, and the localization of GFP-Trex1 was monitored (after recovery postirradiation as indicated) by fluorescence microscopy.
(B) Cell-cycle-specific relocalization. Wild-type MEFs were synchronized in G0/G1 or S phase prior to FACS analysis (shown) or treatment as in (A).
(C) Colocalization with BrdU. Transfected cells were labeled with BrdU prior to IR or HU treatment. Cells were processed (after 90 min recovery post-
irradiation) for indirect immunofluorescence analysis by confocal microscopy using anti-BrdU/secondary Cy3-conjugated antibodies (red) and anti-
Flag/secondary FITC-conjugated antibodies (green) todetect epitope-tagged Trex1 (due tothe extinction of GFP fluorescence inethanol-fixed cells).
Representative images are shown; colocalization of Trex1 and BrdU foci was calculated in 45 dual-labeled IR-treated cells. Nuclei were visualized
with DAPI (A–C).
(D and E) Cell killing after g irradiation. DNA content of asynchronous Trex1+/+wild-type and Trex1?/?MEFs was analyzed by FACS (D), and cells with
sub-G1 DNA content were quantified (E), 24 hr postirradiation with the doses indicated. Error bars represent the SEM from two independent
Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc. 875
Figure 2. Defects in Cell-Cycle Progression and DNA Damage Checkpoints in Trex1?/?MEFs
(A) Cell-cycle distribution. Asynchronous cultures were analyzed by FACS (upper panel) and cells in G0/G1, S, and G2/M phases were quantified
(B) Proliferation rate. Cells (5 3 104seeded in duplicate wells) were cultured in 6-well plates and counted after the times indicated.
(C) Incorporation ofBrdU. Flow cytometry showsthegated population (%)of Trex1+/+wild-type andTrex1?/?MEFscostained with7-AAD(xaxis) and
FITC-conjugated anti-BrdU antibodies (y axis) after incubation without (control, upper panels) or with BrdU (+BrdU, lower panels).
(D) Cell-cycle progression. BrdU-positive MEFs (as in [C]), with or without g irradiation, were returned to BrdU-free medium (0 hr) and analyzed by
FACS at the times indicated.
Error bars are as in Figure 1; these are too small to discern in (B).
876 Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc.
BrdU-positive cells was monitored at various time points
by FACS. Labeling of cells was highly efficient with
?35% of wild-type and ?12% of Trex1?/?MEFs incorpo-
rating detectable levels of BrdU (Figure 2C), consistent
with the proportion of Trex1?/?versus Trex1+/+cells in S
phase (Figure 2A). Immediately after the BrdU pulse
(time 0 hr), BrdU-positive cells were broadly distributed
in S phase as well as G2/M in both the Trex1?/?and
Trex1+/+cell lines (Figure 2D). Wild-type Trex1+/+MEFs,
in the absence of g irradiation (Figure 2D, first column),
continued to progress into G2/M during 0–2 hr and by
4 hr had exited G2/M, forming a characteristic G0/G1
peak before starting to recycle again. Trex1?/?MEFs (Fig-
ure 2D, second column) also progressed from S to G2/M
during 0–2 hr, somewhat more rapidly than Trex1+/+
MEFs—perhaps indicating an intra-S checkpoint failure
(Figure 2A)—and, similarly to the wild-type control, had
exited G2/M after 4 hr and entered G0/G1. However,
Trex1?/?cells exhibited a major delay in entering the
next S phase, forming a peak of cells accumulated in
G0/G1 that had not re-entered the cell cycle even after
12 hr. These data show that unstressed Trex1?/?MEFs
are defective in G1/S transition.
The abnormal cellular responses ofTrex1?/?MEFsto IR
indicated that there were defects in the G2/M DNA dam-
age checkpoint (Figure 1D). We further investigated this
in irradiated BrdU-labeled cells (Figure 2D). Wild-type
Trex1+/+MEFs progressed normally from S phase into
G2/Mover2hrbutdid notprogressinto G0/G1,remaining
arrested in G2, even 12 hr postirradiation (Figure 2D, third
column), indicating a functional G2/M DNA damage
checkpoint. Irradiated Trex1?/?MEFs again showed a
less ordered progression into G2/M during 0–2 hr but
then did not progress into G0/G1 and appeared to arrest
in G2 up to 8 hr postirradiation (Figure 2D, fourth column).
However, unlike wild-type MEFs, irradiated Trex1?/?
MEFs started to re-enter G0/G1 at 10 hr postirradiation,
where cells continued to cycle (Figure 2D, fourth column).
These data show that the G2/M DNA damage checkpoint
is compromised and not maintained in Trex1?/?MEFs.
Chronic Activation of ATM-Mediated DNA Damage
Checkpoint Signaling in Trex1?/?MEFs
in the Absence of Exogenous Stress
As Trex1?/?MEFs showed cell-cycle defects and abnor-
mal responses to g irradiation, we examined the expres-
sion and activation of several core protein components
of the DNA damage checkpoint pathways that control
cell-cycle progression in mammalian cells (for recent
review, see Niida and Nakanishi, 2006). These included
the central transducers of DNA damage signaling, the
Chk1 and Chk2 kinases, which are phosphorylated by
the apical ATR and ATM kinases in response to replication
stress and IR. g radiation-induced phosphorylation of
Chk2 was clearly seen in wild-type MEFs (Figure 3A). In
contrast, there was much slower phosphorylation of
Chk2 in Trex1?/?MEFs, and strikingly, the level of Chk2
protein was greatly reduced in the Trex1?/?versus
Trex1+/+MEF cell line, most notably in nonirradiated cells.
Reduced levels of Chk2 protein were confirmed in
Trex1?/?MEFs by using a second independent antibody;
Chk2 mRNA levels were similar in Trex1?/?and control
cells (Figures S1A and S1B in the Supplemental Data
available with this article online). There was also reduced
phosphorylation of Chk2, as well as Chk1, in Trex1?/?
MEFs treated with HU. However, in marked contrast to
Chk2, the level of Chk1 protein was similar in untreated
Trex1?/?versus Trex1+/+MEFs (Figure 3B). Levels of
ATM and ATR were equivalent in Trex1?/?versus
of cellular responses to DNA damage (reviewed in Ahn
et al., 2004), is apparently destabilized in Trex1?/?cells.
A key downstream effector phosphorylated by Chk2 is
p53, which is also directly phosphorylated by ATM, bring-
ing about G1 arrest via transactivation of the cyclin D/E-
dependent kinase inhibitor p21 (el-Deiry et al., 1993;
Harper et al., 1993); further ubiquitylation of p53 may
also contribute to transactivation and cell-cycle arrest
(Le Cam et al., 2006). We examined the phosphorylation
of p53 in irradiated Trex1?/?
(Figure 3A). Wild-type Trex1+/+MEFs showed a normal
response to g radiation; phosphorylated p53 was not
detected in nonirradiated cells but was induced by IR, its
disappearance 2 hr postirradiation coinciding with the ap-
pearance of a higher molecular weight, possibly ubiquity-
lated, species. In marked contrast to wild-type cells, p53
was readily detectable in nonirradiated Trex1?/?cells
and as the high molecular weight species. That p53 is
ubiquitylated in nonirradiated Trex1?/?MEFs was con-
firmed by transfecting cells with a construct expressing
epitope-tagged ubiquitin (Figure S2). Consistent with
p53 activation, p21 was also detected in nonirradiated
Trex1?/?MEFs, whereas it was only induced 2 hr postirra-
diation in IR-treated wild-type cells. Substantial accumu-
lation of p21 in nonirradiated Trex1?/?MEFs was also
verified by indirect immunofluorescence (Figure 3C); no
p21-positive cells were seen in wild-type cultures, but
?38% of Trex1?/?MEFs showed p21 staining. These
data are consistent with defective G1/S transition of
Trex1?/?MEFs (Figure 2).
We examined possible mechanisms for the reduction of
was not apparently phosphorylated in nonirradiated
Trex1?/?cells (Figure 3A), the IR-independent activation
of downstream DNA damage checkpoint proteins indi-
cates chronic activation of Chk2 in Trex1?/?cells that
may target the protein for degradation. Indeed, adding
the MG132 proteasome inhibitor to the growth medium
of Trex1?/?MEFs restored the Chk2 protein level in a con-
centration-dependent manner but had no effect on the
level in Trex1+/+cells (Figure 3D). The half-life of endoge-
nous Chk2 protein was reduced to ?2.5 from ?5.5 hr in
Trex1?/?versus Trex1+/+MEFs (Figure S1C). That the de-
pletion of Chk2 in Trex1?/?MEFs is a direct consequence
of the loss of Trex1 was confirmed by the restoration
of Chk2 protein to wild-type levels in Trex1?/?MEFs
Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc. 877
Figure 3. Chronic Activation of DNA Damage Signaling in Trex1?/?MEFs
(A and B) Expression and modification of checkpoint proteins in Trex1+/+wild-type and Trex1?/?MEFs treated with or without (A) IR or (B) HU. Cell
molecular weight species detected by anti-Ser15-phosphorylated p53 antibodies were verified as ubiquitylated p53 (see text). Actin is a control for
equal gel loading. Irradiated cells were returned to cell culture after an unavoidable lag of ?20 min (0 hr) and analyzed after recovery for the times
(C) Detection of p21 by indirect immunofluorescence. Cells were stained with anti-p21/FITC-conjugated secondary antibodies (red) and analyzed by
fluorescence microscopy (left panels). The percentage of p21-positive cells was calculated from R250 cells analyzed for each cell line (right panel).
Error bars as in Figure 1.
(D) Chk2 protein is targeted for degradation. Trex1?/?and Trex1+/+MEFs were treated with increasing doses (1, 5, and 10 mM) of the proteasome
inhibitor MG132 and Chk2 detected in extracts by immunoblotting in comparison with untreated (?) cells. Trex1?/?MEFs expressing epitope-tagged
Trex1 (+) were similarly analyzed, 24 hr posttransfection with the expression construct; Trex1 was detected with anti-Flag antibodies.
878 Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc.
transfected with a Trex1 expression construct (Figure 3D).
efforts to restore this protein to wild-type levels by ectopic
overexpression (Figure S1D). However, Chk2 protein
levels were restored after treatment of Trex1?/?MEFs
with the ATM-specific inhibitor Ku-55933 (Hickson et al.,
2004), as well as by inhibition of both ATM and ATR with
caffeine (Figure 3E), indicating that DNA damage signaling
by ATM to Chk2 targets the latter for degradation in
Trex1?/?MEFs. The data indicate chronic ATM-mediated
checkpoint activation in Trex1?/?cells in the absence of
exogenous DNA damage.
A Persistent ssDNA Species Arising in S Phase
Accumulates in the ER of Trex1?/?MEFs
Localization of Trex1 with BrdU foci in stressed S phase
cells (Figure 1), defective G1/S transit in unstressed
Trex1?/?MEFs (Figure 2), and the affinity of Trex1 for
ssDNA (Ho ¨ss et al., 1999; Mazur and Perrino, 1999,
2001b; Morita et al., 2004) would indicate that Trex1
acts on ssDNA arising during DNA replication that might
accumulate in Trex1-deficient cells. We utilized an anti-
body specific for ssDNA in immunofluorescence analysis
of Trex1?/?versus Trex1+/+MEFs that had been fixed
without any further processing that might generate inci-
dental ssDNA (Figure 4). Strikingly, Trex1?/?MEFs
showed positive staining for ssDNA under these condi-
tions, and this was entirely extranuclear (Figure 4A). Z
stack scanning confirmed that ssDNA staining was con-
fined to the cytoplasmic compartment (data not shown).
That the antibody specifically detected ssDNA in Trex1?/?
MEFs was verified by pretreating samples with S1 nucle-
ase, which eliminated staining of Trex1?/?cells in a dose-
dependent manner, at S1 concentrations that do not
degrade duplex nucleic acids (Figure 4A). FACS analysis
(measuring fluorescence of the Cy3-conjugated second-
ary antibody) showed that ?42% Trex1?/?MEFs were
positive for ssDNA, as indicated by the shift in fluorescent
density of Cy3-positive versus Cy3-negative cells; this
shift was eliminated by S1 nuclease treatment and not
seen in wild-type cells (Figure 4B).
Colocalization with the ER protein calreticulin showed
that ssDNA in Trex1?/?MEFs was associated with the
ER, as was GFP-Trex1 in transfected wild-type cells (Fig-
ure 4C). To estimate the size of the accumulated ssDNA
species, DNA in cytoplasmic extracts of Trex1?/?and
Trex1+/+MEFs was 32P labeled. Denaturing gel electro-
phoresis confirmed the presence of cytoplasmic DNA in
Trex1?/?MEFs, and a surprisingly discrete band was
resolved, corresponding to an ssDNA polynucleotide spe-
cies of between 60 and 65 residues (Figure 4D); there was
no detectable labeled DNA in cytoplasmic extract from
Trex1+/+MEFs. To investigate the origin of the ER-associ-
ated ssDNA in Trex1?/?MEFs, cells were pulse labeled
with BrdU prior to detection of extranuclear ssDNA as
above. Under these nondenaturing conditions, BrdU-
labeled DNA in the nucleus is undetectable and there
is no BrdU staining of Trex1+/+MEFs. But in labeled
Trex1?/?MEFs, BrdU was colocalized with extranuclear
is seen in Trex1?/?cells processed conventionally for
immunofluorescence (Figure S3). Thus, aberrant extra-
nuclear ssDNA polynucleotides in Trex1?/?MEFs had
apparently originated in S phase.
Cells from Patients with Inactivating TREX1
Mutations at the AGS1 Locus Show the Same
Molecular Defects as Trex1?/?MEFs
We analyzed the cell-cycle distribution and expression of
homozygous for TREX1 mutations that abolished TREX1
30DNA exonuclease activity (Crow et al., 2006a; Figure 5).
TREX1-deficient AGS fibroblasts accumulated in G0/G1
with a reduced S phase population (Figure 5A) and
showed chronic activation of checkpoint proteins, includ-
ing depletion of Chk2, elevation of p21, and increased
phosphorylation/ubiquitylation of p53, in the absence of
were normal (Figure 5B). There were higher background
levels of phosphorylated p53 and p21 in the nonimmortal-
ized human fibroblasts (Figure 5B) and a correspondingly
higher background of p21-positive cells detected by
immunofluorescence, although the increase above back-
ground was ?46% in AGS cells (Figure 5C), similar to
Trex1?/?MEFs. Furthermore, accumulation of extranu-
clear ssDNA was detected in AGS fibroblasts (Figure 5D)
and in a similar proportion of cells (?39%; Figure 5E), as
seen in Trex1?/?MEFs. Thus, primary human fibroblasts
from AGS patients with inactivating TREX1 mutations
recapitulate the cellular phenotype of a spontaneously im-
define suitable conditions for simultaneous staining of
extranuclear ssDNA and p21 in the same cells, but in
both Trex1?/?MEFs and AGS fibroblasts, the proportion
?40% of cells in an asynchronous culture.
Trex1 is a DNA 30exonuclease (Ho ¨ss et al., 1999; Mazur
and Perrino, 1999), although paradoxically, it is not resi-
dent in the nucleus (Chowdhury et al., 2006). Here we
show that (1) Trex1 is mobilized to replication foci in the
nucleus after g irradiation or HU treatment and (2) Trex1
acts on an ssDNA polynucleotide species arising from
the processing of aberrant DNA replication intermediates,
leading to chronic ATM-dependent checkpoint activation
and extranuclear accumulation of ssDNA in Trex1-defi-
cient cells, even in the absence of exogenous stress.
(E) Inhibition of ATM restores Chk2 protein levels. Trex1?/?MEFs were treated with 10 mM of the ATM-specific inhibitor Ku-55933 (ATM-i) or 10 mM
caffeine for the times indicated prior to immunoblot analysis.
Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc. 879
880 Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc.
ssDNA at the Replication Fork and Extranuclear
Accumulation in Trex1-Deficient Cells
Transient exposure of ssDNA in the mammalian cell nu-
cleus occurs during processes such as transcription and
replication, and at any one time, 1%–2% of the genomic
DNA may be in single-stranded form in proliferating cells
(Bjursell et al., 1979). DNA polymerases can pause or stall
at regions of DNA that are damaged, repetitive in
sequence, or heavily transcribed and may also bypass
DNA damage. Elegant studies in yeast have directly dem-
onstrated the presence of ssDNA due to uncoupling of
leading- and lagging-strand synthesis (Sogo et al., 2002)
or gaps left by repriming on the leading strand (Lopes
et al., 2006). If the replication fork encounters an immov-
able block, dissociation of the replisome and fork collapse
can lead to abnormal four-branched recombination inter-
mediates (reversed forks) and DNA breaks that have to be
processed to allow replication to restart and extensive
single-stranded regions may be generated by nucleolytic
processing of nascent strands (Cotta-Ramusino et al.,
2005). The accumulation of ssDNA that becomes coated
with the single-strand binding protein complex RPA is
the key trigger that activates the DNA damage checkpoint
(Zou and Elledge, 2002), initiating a complex network of
responses to stabilize, repair, and/or restart the fork in
order to preserve genome integrity. Although key ele-
ments of checkpoint pathways are conserved from yeast
to mammals (for recent reviews, see Harrison and Haber,
2006; Niida and Nakanishi, 2006), yeast do not appear to
have a robust 30DNA exonuclease. Thus, Trex1, the most
abundant 30DNA exonuclease in mammalian cells, is not
found in yeast.
We have shown that Trex1 is relocalized from the ER to
the nucleus at replication foci after g irradiation or HU,
where it would bind ssDNA, suggesting that resolution of
some forks requires a 30DNA exonuclease activity in
mammalian cells. However, in nonirradiated Trex1-defi-
cient cells, persistent ssDNA, which BrdU-labeling shows
has originated in S phase, accumulates in the ER, indicat-
ing that Trex1 does not act directly on aberrant single-
stranded 30termini at the replication fork. Consistent
withthis,sister chromatidexchanges, indicative ofhomol-
elevated in Trex1?/?MEFs (Morita et al., 2004). Rather,
Trex1 apparently acts on discrete ssDNA molecules that
have already been released from replication sites. Exonu-
clease processing to resect DNA ends during resolution/
repair of recombination intermediates and double-strand
breaks (DSBs) would generate mononucleotides or possi-
bly short oligomers (see Harrison and Haber, 2006). Our
data by contrast demonstrate that Trex1 processes a dis-
Mechanistically, this could be generated by combined
helicase/endonuclease activity; of enzymes present at
the replication fork, a possible candidate would be Dna2.
Processing of the Lagging Strand
During DNA Replication
parently cooperates with the FEN1 enzyme in Okazaki
fragment processing during lagging-strand DNA synthe-
sis. In in vitro assays, the ability of Dna2 to remove occa-
sional long 50ssDNA flaps displaced by DNA polymerase
dthatareRPAcoated orhavesecondarystructure, aswell
as both 50and 30regions of equilibrating flaps, produces
a suitable substrate for FEN1 to then generate a ligatable
specifically require not only a DNA exonuclease with 30/
plete degradation. It is interesting to note that AGS is
caused not only by mutations in the human TREX1 gene
but also in any of the genes encoding the three subunits
of RNaseH2 (Crow et al., 2006b). RNaseH2 is the major
RNaseH activity hydrolyzing the RNA strand of an
RNA:DNA hybrid in yeast and mammalian cells and,
although its cellular function is not clearly defined, might
aid Dna2 and FEN1 in removing Okazaki fragment RNA
primers (Arudchandran et al., 2000; Qiu et al., 1999).
RNaseH2 can also excise single ribonucleotides embed-
ded in double-stranded DNA (Eder et al., 1993; Rydberg
and Game, 2002), as may arise by ligation of incompletely
long flaps released by Dna2 have secondary structure,
there may be a specific requirement for RNaseH2 to
degrade the 50RNA primer reannealed within the folded-
back ssDNA polynucleotide; the 30DNA exonuclease of
Trex1 could act at a reduced rate on double-stranded
Figure 4. Extranuclear Accumulation of an S Phase-Specific ssDNA Polynucleotide in Trex1?/?MEFs
(A) Extra-nuclear accumulation of ssDNA. Methanol-fixed Trex1+/+wild-type and Trex1?/?MEFs were stained directly with anti-ssDNA/Cy3-conju-
gated secondary antibodies (red) or pretreated with S1 nuclease, as indicated. The nucleus is stained with DAPI.
(B) FACS analysis of ssDNA staining. Cell suspensions were stained as in (A), and Cy3 fluorescence was monitored by FACS either with (blue) or
without (red) S1 nuclease pretreatment. Note different scales on the y axes.
(C) Localization of ssDNA to the ER. Fixed Trex1?/?MEFs stained for ssDNA (red) as in (A) were costained with anti-calreticulin/FITC-conjugated
secondary antibodies (green), and wild-type cells expressing GFP-Trex1 were stained with anti-calreticulin/Cy3-conjugated secondary antibodies
and analyzed by confocal microscopy.
(D) Sizing of ssDNA. DNA from equal amounts of cytoplasmic extract, verified by blotting with anti-calreticulin antibodies, was 32P labeled and
analyzed (2 ml and 20 ml) by denaturing 5% PAGE in comparison with labeled synthetic oligonucleotide markers. Where indicated, 20 ml extract
was treated with 200 U/ml S1 nuclease prior to DNA labeling.
(E)Extranuclearaccumulation ofnewlyreplicatedDNA.Cellswerepulselabeled withBrdUand,2hrafterthepulse,costainedforssDNA(red)asin(A)
and with FITC-conjugated anti-BrdU antibodies (green) and analyzed by confocal microscopy.
Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc. 881
Figure 5. Cellular Defects of AGS Fibroblasts
Control and AGS primary human fibroblasts were analyzed in key assays shown for wild-type and Trex1?/?MEFs in preceding figures.
(A) Cell-cycle distribution. See Figure 2A.
(B) Expression and modification of checkpoint proteins. See Figures 3A and 3B.
(C) Detection of p21 by indirect immunofluorescence. See Figure 3C.
(D) Extranuclear accumulation of ssDNA. See Figure 4A.
(E) Flow cytometry of ssDNA staining. See Figure 4B.
882 Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc.
DNA in regions of secondary structure, as well as ssDNA
(Ho ¨ss et al., 1999; Mazur and Perrino, 1999).
It has been proposed that removal of the entire RNA-
DNA primer region synthesized by the pola-primase
complex, which lacks proof-reading activity, may abro-
gate the need to correct any errors inserted by pola (Bae
merase pausing, and a discrete 61 nucleotide replication
intermediate has been described in Drosophila cells that
may correspond to such constraints (Blumenthal and
Clark, 1977). Our data, showing a 60–65 nucleotide
ssDNA species that accumulates in Trex1-deficient cells,
might be consistent with either model. Although irrelevant
to Trex1-deficient disease, we cannot formally exclude
here that Trex1 mobilized to the nucleus after g irradiation
may act on an IR-induced lesion. However, such damage
would be distributed randomly in the genome, not just at
replication foci (Figure 1), and furthermore, Trex1 does
not degrade IR-generated 30termini (Inamdar et al.,
2002).ThatTrex1 issimilarly mobilizedto the nucleusafter
treatment with HU (which stalls replication forks) and only
during S phase clearly points to a replication-associated
problem that is exacerbated in stressed Trex1-deficient
cells. The present study focuses on Trex1-deficient cells.
However, in initial studies using fractionated cell extracts
of synchronized wild-type MEFs, we can detect endoge-
nous Trex1 at low levels in the nucleus of unstressed cells
exclusively during S phase (Figure S4).
Chronic Checkpoint Activation
in Trex1-Deficient Cells
Persistence of the ssDNA polynucleotide substrate of
Trex1 causes chronic activation of checkpoint signaling
in Trex1-deficient cells. Phosphorylation of ATR in re-
sponse to RPA-coated ssDNA is a necessary early event
in the DNA damage response to replication stress (Zou
and Elledge, 2002), although it is becoming increasingly
clear that both ATM and ATR are activated in response
to replication stress as well as IR (reviewed in Cuadrado
et al., 2006; Hurley and Bunz, 2007). We have shown that
there is chronic ATM-dependent signaling via Chk2 and
p53 in the absence of exogenous stress in Trex1-deficient
radation of Chk2. Accumulation of p21 is consistent with
the observed cellular defect in G1/S transit (el-Deiry
et al., 1993; Harper et al., 1993). The detailed mechanism
of Chk2 turnover has not been investigated here but may
involve ubiquitylation (Zhang et al., 2006). Similarly to
Chk2-deficient cells (see Ahn et al., 2004), Trex1?/?
MEFs are resistant to apoptosis after g irradiation, indicat-
ing that this property of Trex1-deficient cells is likely to be
due to depletion of Chk2. Low levels of Chk2, as well as
reduced phosphorylation of Chk1, could contribute to
a failure of G2 arrest in irradiated Trex1?/?MEFs, where
ATM may act upstream of ATR-Chk1 (see Cuadrado
et al., 2006; Hurley and Bunz, 2007). Our data indicate
fork (which may no longer be coated with RPA) cause
chronic ATM signaling, and their degradation by Trex1 is
edented activation of ATM by ssDNA requires further
mechanistic investigation. As movement of Trex1 from
the ER to the nucleus is not affected by caffeine treatment
(Figure S5), this relocalization does not appear to be
mediated by the checkpoint response.
Inthe absence of exogenous stress, proliferating Trex1-
deficient cells eventually adapt and continue to cycle
but—if Trex1 does not act directly in DNA repair—may
not accumulate genomic DNA damage; this is consistent
with there being no increase in spontaneous mutation
frequency measured in Trex1?/?mice (Morita et al.,
2004). Aninability to mount a G2 arrestleaves cellsvulner-
able to induced DNA damage, but in Trex1-deficient cells,
it is the ability to maintain rather than initiate G2 arrest that
is compromised after g irradiation (Figure 2D, right-hand
column). Phosphorylation of histone H2AX by ATM and
ATR is a molecular marker of DSBs and endogenous oxi-
dative damage (Stucki and Jackson, 2006). Notably, there
was reduced gH2AX focus formation in Trex1?/?versus
Trex1+/+MEFs (Figure S6). This might be an indirect effect
of deregulated checkpoints, or chronic signaling may pro-
mote DNA repair in Trex1?/?MEFs. Rather than covalent
DNA damage, it is apparently the end products of the pro-
cessing of such damage that persist in Trex1-deficient
cells. ssDNA polynucleotides emerge outside the nucleus
after a single S phase (Figure 4E) where, in the absence of
ER-associated Trex1, they accumulate in the cytoplasmic
compartment. Here, as DNA that is no longer normally
sequestered, they become immunostimulatory molecules
(Ishii and Akira, 2006).
Relevance of the Trex1-Deficient Cellular Defect
to Autoimmune Disease
Autoimmunity is a feature of Trex1-deficient inflammatory
disease (Crow et al., 2006a; Morita et al., 2004), with inap-
propriate activation of antiviral-like immune responses
leading to elevated levels of IFN-a in the absence of infec-
tion (Goutie `res et al., 1998). The innate immune system
responds to infectious pathogen (nonself) as well as un-
necessary/abnormal host (self) nucleic acids via comple-
mentary receptor systems that culminate in the produc-
tion of type I IFNs or NF-kB-dependent proinflammatory
cytokines (Ishii and Akira, 2006). A member of the family
of Toll-like receptors (TLRs), TLR9, is uniquely located in
the ER of immune cells, responding to endocytosed or
transfected DNA, including ssDNA, that contains unme-
thylated CpG motifs (Barton et al., 2006). Importantly,
from nascent DNA at the replication fork would not have
undergone de novo methylation or chromatin assembly.
TLR9-independent recognition of cytosolic DNA has
also been reported (reviewed in Chi and Flavell, 2007).
That Trex1 functions in cellular defense against DNA
viruses might be suggested by its ER location and the
similarity of Trex1-deficient genetic disease to congenital
viral nfection. Alternatively, sequestering the Trex1 30
Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc. 883
exonuclease in the cytoplasmic compartment may serve
to keep it away from DNA until it is needed, as is the
case with some other potent DNA modifying enzymes
(Brar et al., 2004).
DNA damage signaling can directly engage the immune
system in a noncell type-specific manner. Activation of
ATM by genotoxic agents or stalled DNA replication in-
system that is expressed by natural killer cells and acti-
vated CD8+ T cells (Gasser et al., 2005), the same cell
types that secrete Granzyme A to induce the caspase-
independent proapoptotic pathway mediated by Trex1
(Chowdhury et al., 2006). Chowdhury et al. demonstrated
relocalization of TREX1 to the nucleus after granzyme A
treatment, reduced DNA breaks after granzyme A treat-
ment in TREX1 siRNA-suppressed cells, and that TREX1
siRNA-suppressed cells were resistant to granzyme A-
induced apoptosis; that we see similar results after g irra-
diation or HU treatment of Trex1 null cells might indicate
that this is actually a more nonspecific stress response.
Importantly, we now report a chronic cellular phenotype
even in unstressed cells that directly relates to the Trex1-
deficient disease phenotype. ATM signaling also activates
NF-kB that—like Trex1—is mobilized to the nucleus in re-
activation of this pathway in early tumorigenesis may alert
immune surveillance to suppress malignant progression
(reviewed in Gasser and Raulet, 2006). Trex1-deficient
disease does not appear to be associated with increased
cancer risk, albeit within the reduced lifespan resulting
from pathological autoinflammatory responses (Crow
et al., 2006a; Morita et al., 2004). Our data place Trex1 at
DNA damage checkpoints, and immunity.
Cell Lines and Procedures
Wild-type and Trex1?/?MEF cell lines and primary human fibroblast
control (MRC5) and AGS patient (F39A) cell lines have been described
with a137Cs source; irradiation was at a dose of 4 Gy unless otherwise
specified. Cells were treated with HU at 2 mM for 3 hr. For BrdU label-
ing, cells were pulse labeled with 10 mM BrdU (BD Bioscience) for 2 hr
before returning to BrdU-free medium. Cells were treated with the
proteasome inhibitor MG132 (Alexis Biochemicals) at various concen-
trations for 24 hr. Cells were synchronized in G0/G1, S, or G2/M by
serum starvation (48 hr), double thymidine block (and 3 hr release),
or 0.4 mg/ml nocodazole (16 hr), respectively.
Epitope-Tagged Recombinant Trex1
Total RNA was extracted from wild-type MEFs with Tri-Reagent
(Sigma). The murine Trex1 open reading frame encoding the full-length
314 amino acid protein was amplified by reverse transcription (RT)-
PCR of 1 mg RNA with Superscript II enzyme (Invitrogen) and the
following forward/reverse adaptor primers (complementary sequence
Trex1 was epitope tagged at the amino terminus with a Flag-
(Glycine)5-EGFP-(Glycine)5sequence in plasmid C1B (kindly provided
by Jan-Michael Peters, IMP, Vienna) after restriction digest of both
plasmid and PCR product with BglII and SalI. The construct was veri-
fied by DNA sequencing and transfected into cells by using Lipofect-
amine 2000 (Invitrogen). Transfection efficiency was routinely 75%–
Cells were sorted by FACS, and cell profiles were analyzed by using
Cellquest (BD Bioscience) and ModFIT LT (Verity, Maine) software.
Cellular DNA was labeled with the Cycletest Plus DNA Reagent Kit
(BD Bioscience). BrdU-labeled cells were analyzed with the FITC
BrdU Flow Kit (BD Bioscience) and 7-amino-actinomycin D (7-AAD).
Antibodies used are listed in Table S1.
Cell Extracts and Immunoblotting
To prepare whole-cell extracts, cell pellets were lysed in ten volumes
ice-cold lysis buffer (20 mM Tris-HCl [pH 7.4], 20% glycerol, 0.5 M
NaCl, 1.5 mM MgCl2, 0.1 mM EDTA, 0.1 mM EGTA, 0.5% Nonidet
NP-40, and 0.1% Triton X-100, plus protease inhibitors). Protein
concentration was measured by Bradford assay. Cell lysates were
separated by10% SDS-PAGE,transferred tonitrocellulose, incubated
with antibodies in TBST-5% dried milk, and detected by horseradish
peroxidase-conjugated secondary antibodies using the ECL Western
Blotting Detection Kit (Amersham Biosciences).
Purification and Labeling of Cytoplasmic DNA
To prepare cytoplasmic extracts, cells were incubated in 10 mM
HEPES (pH 7.9), 10 mM KCl, 1.5 mM MgCl2, 0.34 M sucrose, 10%
glycerol, plus protease inhibitors for 5 min on ice with 0.1% Triton
X-100, and nuclei were removed by low-speed centrifugation (1300
3 g, 4 min). Cytoplasmic protein extract (?500 mg/400 ml) was treated
with1mg/mlProteinase K at55?C for 1hr.After phenol/chloroformex-
traction, the supernatant was incubated with 500 mg/ml DNase-free
RNase A (QIAGEN) for 30 min at 37?C, again followed by phenol/chlo-
roform extraction. DNA was purified by using a DyeEx 2.0 Dye-Termi-
nator spin column (QIAGEN) and a 20 ml aliquot 30-end-labeled using
f-32P-ddATP (GE Healthcare) and terminal deoxynucleotidyl transfer-
ase(Promega). Unincorporated ddATP was removed bya second spin
column. The DNA-containing eluate was analyzed by denaturing 5%
PAGE and Typhoon PhosphorImager (Amersham Biosciences).
Fluorescence Microscopy and Cell Imaging
Cells grown on glass coverslips were fixed with 4% paraformaldehyde
in PBS for 15 min and permeabilized with 0.1% Triton X-100 for 10 min
on ice. Incubation with antibodies was in TBST-5% dried milk. Nuclear
tured with 2 N HCl, 0.5% Triton X-100, and neutralized with 0.1 M
Na3B4O7(pH 8.5). For the detection of ssDNA, cells were fixed in
80% methanol in PBS, ±150 mM NaCl, without further treatment prior
to adding antibody. Cells were pretreated for 1 hr at 37?C with
200 mg/ml RNase A (QIAGEN) and with S1 nuclease plus reaction
buffer (Promega) where indicated. Slides were mounted in Vectashield
mounting medium with DAPI (Vector Laboratories, Burlingame, CA)
and visualized under a Zeiss Axioskop fluorescence microscope
equipped with a CCD imaging system (IP Lab Spectrum, NY) or by
confocal microscopy (Zeiss).
Supplemental Data include six figures and one table and can be found
with this article online at http://www.cell.com/cgi/content/full/131/5/
884 Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc.
We thank Vincenzo Costanzo for critical discussions and reading
the manuscript and Graeme Smith (KuDOS Pharmaceuticals/
AstraZeneca) for a gift of Ku-55933 inhibitor. This work was supported
by Cancer Research UK.
Received: April 27, 2007
Revised: July 24, 2007
Accepted: October 8, 2007
Published: November 29, 2007
Ahn, J., Urist, M., and Prives, C. (2004). The Chk2 protein kinase. DNA
Repair (Amst.) 3, 1039–1047.
Aicardi, J., and Goutie `res, F. (2000). Systemic lupus erythematosus or
Aicardi-Goutie `res syndrome? Neuropediatrics 31, 113.
Alarco ´n-Riquelme, M.E. (2006). Nucleic acid by-products and chronic
inflammation. Nat. Genet. 38, 866–867.
Arudchandran, A., Cerritelli, S.M., Narimatsu, S.K., Itaya, M., Shin, D.-
Y., Shimada, Y., and Crouch, R.J. (2000). The absence of ribonuclease
urea, caffeine and ethyl methanesulphonate: implications for roles of
RNases H in DNA replication and repair. Genes Cells 5, 789–802.
Bae, S.-H., Bae, K.-H., Kim, J.-A., and Soo, Y.-S. (2001). RPA governs
endonuclease switching during processing of Okazaki fragments in
eukaryotes. Nature 412, 456–461.
Banchereau, J., and Pascual, V. (2006). Type I interferon in systemic
lupus erythematosus and other autoimmune diseases. Immunity 25,
Barton, G.M., Kagan, J.C., and Medzhitov, R. (2006). Intracellular
localization of the Toll-like receptor 9 prevents recognition of self
DNA but facilitates access to viral DNA. Nat. Immunol. 7, 49–56.
Bjursell, G., Gussander, E., and Lindahl, T. (1979). Long regions of
single-stranded DNA in human cells. Nature 280, 5721–5723.
Blumenthal, A.B., and Clark, E.J. (1977). Discrete sizes of replication
intermediates in Drosophila cells. Cell 12, 183–189.
Bourdon, J.C., Renzing, J., Robertson, P.L., Fernandes, K.N., and
Lane, D.P. (2002). Scotin, a novel p53-inducible proapoptotic protein
located in the ER and the nuclear membrane. J. Cell Biol. 158, 235–
Brar, S.S., Watson, M., and Diaz, M. (2004). Activation-induced
cytosine deaminase (AID) is actively exported out of the nucleus but
retained by the induction of DNA breaks. J. Biol. Chem. 279, 26395–
Brucet, M., Querol-Audi, J., Serra, M., Ramirez-Espain, X., Bertlik, K.,
Ruiz, L., Lloberas, J., Macias, M.J., Fita, I., and Celada, A. (2007).
Structure of the dimeric exonuclease TREX1 in complex with DNA
displays a proline-rich binding site for WW domains. J. Biol. Chem.
Chi, H., and Flavell, R.A. (2007). Sensing the enemy within. Nature 448,
Chowdhury, D., Beresford, P.J., Zhu, P., Zhang, D., Sung, J.S.,
Demple, B., Perrino, F.W., and Lieberman, J. (2006). The exonuclease
TREX1 is in the SET complex and acts in concert with NM23–H1 to
degrade DNA during granzyme A-mediated cell death. Mol. Cell 23,
Cortez, D.S., Guntuku, S., Qin, J., and Elledge, S.J. (2001). ATR and
ATRIP: partners in checkpoint signaling. Science 294, 1713–1716.
Cotta-Ramusino, C., Fachinetti, D., Lucca, C., Doksani, Y., Lopes, M.,
Sogo, J., and Foiani, M. (2005). Exo1 processes stalled replication
forks and counteracts fork reversal in checkpoint-defective cells.
Mol. Cell 17, 153–159.
Crow, Y.J., Hayward, B.E., Parmar, R., Robins, P., Leitch, A., Ali, M.,
Black, D.N., van Bokhoven, H., Brunner, H.G., Hamel, B.C., et al.
(2006a). Mutations in the gene encoding the 30-50DNA exonuclease
TREX1 cause Aicardi-Goutie `res syndrome at the AGS1 locus. Nat.
Genet. 38, 917–920.
Crow, Y.J., Leitch, A., Hayward, B.E., Garner, A., Parmar, R., Griffith,
E., Ali, M., Semple, C., Aicardi, J., Babul-Hirji, R., et al. (2006b).
Mutations in genes encoding ribonuclease H2 subunits cause
Aicardi-Goutie `res syndrome and mimiccongenital viral brain infection.
Nat. Genet. 38, 910–916.
Cuadrado, M., Martinez-Pastor, B., and Fernandez-Capetillo, O.
(2006). ATR activation in response to ionizing radiation; still ATM
territory. Cell Div. 1, 1–4.
de Silva, U., Choudhury, S., Bailey, S.L., Harvey, S., Perrino, F.W., and
Hollis, T. (2007). The crystal structure of TREX1 explains the 30
specificity and reveals a polyproline II helix for protein partnering.
J. Biol. Chem. 282, 10537–10543.
Eder, P.S., Walder, R.Y., and Walder, J.A. (1993). Substrate specificity
of human RNaseH1 and its role in excision repair of ribose residues
misincorporated in DNA. Biochimie 75, 123–126.
el-Deiry, W.S., Tokino, T., Velculescu, V.E., Levy, D.B., Parsons, R.,
Trent, J.M., Lin, D., Mercer, W.E., Kinzler, K.W., and Vogelstein, B.
(1993). Waf1, a potential mediator of p53 tumor suppression. Cell 75,
Gasser, S., and Raulet, D.H. (2006). The DNA damage response,
immunity and cancer. Semin. Cancer Biol. 16, 344–347.
Gasser, S., Orsulic, S., Brown, E.J., and Raulet, D.H. (2005). The DNA
damage pathway regulates innate immune system ligands of the
NKG2D receptor. Nature 436, 12186–12190.
Goutie `res, F., Aicardi, J., Barth, P.G., and Lebon, P. (1998). Aicardi-
Goutie `res syndrome: an update and results of interferon-a studies.
Ann. Neurol. 44, 900–907.
Harper, J.W., Adami, G.R., Wei, N., Keyomarsi, K., and Elledge, S.J.
(1993). The p21 Cdk interacting protein Cip1 is a potent inhibitor of
G1 cyclin-dependent kinases. Cell 75, 805–816.
Harrison, J.C., and Haber, J.E. (2006). Surviving the breakup: the DNA
damage checkpoint. Annu. Rev. Genet. 40, 209–235.
Hickson, I., Zhao, Y., Richardson, C.J., Green, S.J., Martin, N.M.B.,
Orr, A.L., Reaper, P.M., Jackson, S.P., Curtain, N.J., and Smith, G.C.
(2004). Identification and characterization ofa novel and specific inhib-
itor of the ataxia-telangiectasia mutated kinase ATM. Cancer Res. 64,
Ho ¨ss, M., Robins, P., Naven, T.J., Pappin, D.J., Sgouros, J., and
Lindahl, T. (1999). A human DNA editing enzyme homologous to the
Escherichia coli DnaQ/MutD protein. EMBO J. 18, 3868–3875.
Hurley, P.J., and Bunz, F. (2007). ATM and ATR. components of an
integrated circuit. Cell Cycle 6, 414–417.
Inamdar, K.V., Yu, Y., and Povirk, L.F. (2002). Resistance of 30-phos-
phoglycolate DNA ends to digestion by mammalian DNase III. Radiat.
Res. 157, 306–311.
Ishii, K.J., and Akira, S.(2006). Innate recognitionof, and regulation by,
DNA. Trends Immunol. 27, 525–532.
Kim,J.-H., Kim, H.-D., Ryu, G.-H., Kim, D.-H., Hurwitz, J., and Seo, Y.-
S. (2006). Isolation of human Dna2 endonuclease and characterisation
of its enzymatic properties. Nucleic Acids Res. 34, 1854–1864.
Le Cam, L., Linares, L.K., Paul, C., Julien, E., Lacroix, M., Hatchi, E.,
Triboulet, R., Bossis, G., Shmueli, A., Rodriguez, M.S., et al. (2006).
E4F1 is an atypical ubiquitin ligase that modulates p53 effector func-
tions independently of degradation. Cell 127, 775–788.
Lee-Kirsch, M.A., Chowdhury, D., Harvey, S., Gong, M., Senenko, L.,
Engel, K., Pfeiffer, C., Hollis, T., Gahr, M., Perrino, F.W., et al.
(2007a). A mutation in TREX1 that impairs susceptibility to granzyme
Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc. 885
A-mediated cell death underlies familial chilblain lupus. J. Mol. Med.
Lee-Kirsch, M.A., Gong, M., Chowdhury, D., Senenko, L., Engel, K.,
Lee, Y.-A., de Silva, U., Bailey, S., Witte, T., Vyse, T.J., et al. (2007b).
Mutations in the gene encoding the 30-50DNA exonuclease TREX1
are associated with systemic lupus erythematosus. Nat. Genet. 39,
Li, N., Banin, S., Ouyang, H., Li, G.C., Coutois, G., Shiloh, Y., Karin, M.,
and Rotman, G. (2001). ATM is required for IkB kinase (IKK) activation
in response to DNA double strand breaks. J. Biol. Chem. 276, 8898–
Lopes, M., Foiani, M., and Sogo, J.M. (2006). Multiple mechanisms
control chromosome integrity after replication fork uncoupling and
restart at irreparable UV lesions. Mol. Cell 21, 15–27.
Masuda-Sasa, T., Imamura, O., and Campbell, J.L. (2006). Biochemi-
cal analysis of human Dna2. Nucleic Acids Res. 34, 1865–1875.
Mazur, D.J., and Perrino, F.W. (1999). Identification and expression of
the TREX1 and TREX2 cDNA sequences encoding mammalian 30/50
exonucleases. J. Biol. Chem. 274, 19655–19660.
Mazur, D.J.,andPerrino,F.W.(2001a).Structureand expressionofthe
TREX1 and TREX2 30/50exonuclease genes. J. Biol. Chem. 276,
Mazur, D.J., and Perrino, F.W. (2001b). Excision of 30termini by the
TREX1 and TREX2 30/50exonucleases. J. Biol. Chem. 276, 17022–
Morita, M., Stamp, G., Robins, P., Dulic, A., Rosewell, I., Hrivnak, G.,
Daly, G., Lindahl, T., and Barnes, D.E. (2004). Gene-targeted mice
lacking the Trex1 (DNase III) 30/50DNA exonuclease develop inflam-
matory myocarditis. Mol. Cell. Biol. 24, 6719–6727.
Niida, H., and Nakanishi, M. (2006). DNA damage checkpoints in
mammals. Mutagenesis 21, 3–9.
Qiu, J., Qian, Y., Frank, P., Wintersberger, U., and Shen, B. (1999).
Saccharomyces cerevisiae RNaseH(35) functions in RNA primer
eration with Rad27 nuclease. Mol. Cell. Biol. 19, 8361–8371.
Rice, G., Newman, W.G., Dean, J., Patrick, T., Parma, R., Flintoff, K.,
Robins, P., Harvey, S., Hollis, T., O’Hara, A., et al. (2007a). Heterozy-
gous mutations in TREX1 cause familial chilblain lupus and dominant
Aicardi-Goutie `res syndrome. Am. J. Hum. Genet. 80, 811–815.
Rice, G., Patrick, T., Parmar, R., Taylor, C.F., Aeby, A., Aicardi, J.,
Artuch, R., Attard Montalto, S., Bacino, C.A., Barroso, B., et al.
(2007b). Clinical and Molecular Phenotype of Aicardi-Goutieres
Syndrome. Am. J. Hum. Genet. 81, 713–725.
Richards, A., van den Maagdenberg, A.M.J.M., Jen, J.C., Kavanagh,
D., Bertram, P., Spitzer, D., Liszewski, M.K., Barilla-LaBarca, M.-L.,
30-50DNA exonuclease TREX1 cause autosomal dominant retinal vas-
culopathy with cerebral leukodystrophy. Nat. Genet. 39, 1068–1070.
Rossi, M.L., and Bambara, R.A. (2006). Reconstituted Okazaki
fragment processing indicates two pathways of primer removal.
J. Biol. Chem. 281, 26051–26061.
Rumbaugh, J.A., Murante, R.S., Shi, S., and Bambara, R.A. (1997).
Creation and removal of embedded ribonucleotides in chromosomal
DNA during Mammalian Okazaki fragment processing. J. Biol.
Chem. 272, 22591–22599.
Rydberg, B., and Game, J. (2002). Excision of misincorporated ribonu-
cleotides in DNA by RNase H (type 2) and FEN-1 in cell-free extracts.
Proc. Natl. Acad. Sci. USA 99, 16654–16659.
Sogo, J.M., Lopes, M., and Foiani, M. (2002). Fork reversal and ssDNA
accumulation at stalled replication forks owing to checkpoint defects.
Science 297, 599–602.
Stetson, D.B., and Medzhitov, R. (2006). Type I interferons in host
defense. Immunity 25, 373–381.
Stucki,M.,and Jackson, S.P. (2006).gH2AX and MDC1:anchoring the
DNA-damage-response machinery to broken chromosomes. DNA
Repair (Amst.) 5, 534–543.
Zhang, D., Zaugg, K.,Mak, T.W., and Elledge,S.J. (2006). A rolefor the
deubiquitinating enzyme USP28 in Control of the DNA-damage
response. Cell 126, 529–542.
Zou, L., and Elledge, S.J. (2002). Sensing DNA damage through ATRIP
recognition of RPA-ssDNA complexes. Science 300, 1542–1548.
886 Cell 131, 873–886, November 30, 2007 ª2007 Elsevier Inc.