DNA damage sensor MRE11 recognizes cytosolicdouble-stranded DNA and induces type I interferonby regulating STING traffickingTakeshi Kondoa,b, Junya Kobayashic, Tatsuya Saitoha,b, Kenta Maruyamaa,b, Ken J. Ishiid,e, Glen N. Barberf,Kenshi Komatsuc, Shizuo Akiraa,b,1, and Taro Kawaia,b

aLaboratory of Host Defense and dLaboratory of Vaccine Science, World Premier International Research Center Initiative (WPI) Immunology Frontier ResearchCenter, Osaka University, Osaka 565-0871, Japan; cRadiation Biology Center, Kyoto University, Kyoto 606-8501, Japan; bDepartment of Host Defense, ResearchInstitute for Microbial Diseases, Osaka University, Osaka 565-0871, Japan; eLaboratory of Adjuvant Innovation, National Institute of Biomedical Innovation,Ibaraki, Osaka 567-0085, Japan; and fDepartment of Cell Biology and Sylvester Comprehensive Cancer Center, University of Miami Miller School of Medicine,Miami, FL 33136

Contributed by Shizuo Akira, January 8, 2013 (sent for review December 7, 2012)

Double-stranded DNA (dsDNA) derived from pathogen- or host-damaged cells triggers innate immune responses when exposed tocytoplasm. However, the machinery underlying the primary recog-nition of intracellular dsDNA is obscure. Herewe show that the DNAdamage sensor, meiotic recombination 11 homolog A (MRE11),serves as a cytosolic sensor for dsDNA. Cells with a mutation ofMRE11 gene derived from a patient with ataxia-telangiectasia–likedisorder, and cells inwhichMre11was knockeddown, haddefects indsDNA-induced type I IFN production. MRE11 physically interactedwith dsDNA in the cytoplasm and was required for activation ofstimulator of IFN genes (STING) and IRF3. RAD50, a binding proteinto MRE11, was also required for dsDNA responses, whereas NBS1,another binding protein to MRE11, was dispensable. Collectively,our results suggest that theMRE11–RAD50 complex plays importantroles in recognition of dsDNA and initiation of STING-dependentsignaling, in addition to its role in DNA-damage responses.

signal transduction | innate immunity | pattern recognition receptor

Nucleic acids derived from pathogens, or released from hostcells during tissue damage, have the potential to activate

the immune system. Tremendous progress has been made in ourunderstanding of how host cells recognize these nucleic acid mol-ecules and trigger immune responses (1, 2). RNA-sensing mecha-nisms in particular, which aremediated bymembers of the Toll-likereceptor (TLR) family and by retinoic acid-inducible gene I (RIG-I)-like receptors (RLRs), have been assessed. In contrast, despitethe important possibility that DNA derived from either pathogensor damaged host cells could trigger innate immune responses, themechanisms underlying DNA sensing are relatively obscure.TLR9 is amembrane-boundDNA sensor; it senses CpGDNAat

the endosome to trigger immune responses, particularly in plas-macytoid dendritic cells (pDCs). It also contributes to antiviralresponses by producing type I IFN (IFN), and its aberrant recog-nition of host DNA is linked to autoimmune diseases (3–5). How-ever, it appears that there are TLR9-independent pathways thatrecognize double-stranded DNA (dsDNA) in the cytoplasm (6–8).DNA-dependent activator of IFN-regulatory factors (DAI; alsoknown as ZBP1 or DLM-1) has been reported as an intracellulardsDNA sensor (9), but Dai-deficient mice are able to produce typeI IFN after intracellular DNA stimulation (10), suggesting that theroles ofDAI are either redundant or restricted to specific cell types.An endoplasmic reticulum (ER) resident transmembrane pro-

tein, STING (stimulator of IFNgenes, also known asMITA,ERIS,MPYS or TMEM173), was found to play essential roles for cyto-solic dsDNA-mediated production of type I IFN and inflammatorycytokines through activating the TANK-binding kinase 1 (TBK1)-IFN regulatory factor 3 (IRF3) axis (11). Sting-deficient mice havelost their ability to produce type I IFN in response to dsDNA.Although the STING–TBK1–IRF3 axis is crucial for cytosolicDNA-induced IFN production, STING fails to bind and colocalizewith DNA, suggesting existence of a DNA sensor(s) upstream of

STING. Previous studies have also demonstrated that intracellulardelivery of DNA, but not RNA, elicits dynamic translocation ofSTING from the ER to the Golgi apparatus, followed by for-mation of a vesicle-like structure in which TBK1 is recruited (11, 12).These studies showed that STING translocation is essential for ac-tivation of downstream events. However, the regulatory mechanismsfor this trafficking of STING are largely unknown.Absent in melanoma 2 (AIM2) functions as a cytosolic DNA

sensor forming inflammasomes to release IL-1β, but it is not es-sential for type I IFN production (1). Recently, two DNA-bindingproteins IFN, gamma-inducible protein 16 (IFI16) andDEADboxpolypeptide 41 (DDX41) were identified to be involved in theSTING pathway (13, 14). These proteins interact with DNA andSTINGand are critical for dsDNA-induced type I IFN production.However, the contributions of thesemolecules to responses againstdsDNA under physiological conditions, and their roles in differentcell types, are still unknown.Early studies about the link between DNA damage and innate

immune responses indicated that DNA damage events can triggertype I IFN induction (15, 16). However, it was shown that Ku70,DNA-PKcs, p53, or ataxia-telangiectasia mutated kinase (ATM) isdispensable for intracellular dsDNA-induced type I IFN pro-duction (8). Whereas other DNA repair-related genes were dem-onstrated to be involved in viral infection (17), their roles in innateimmune responses to dsDNA have not been well characterized.In the present study, we showed that a DNA damage sensor,

meiotic recombination 11 homolog A (MRE11), serves as a keysensor for exogenous DNA and activates the STING-dependentpathway.

ResultsIntracellular dsDNA Promotes Phosphorylation of ATM. To identifymolecules involved in innate immune responses to cytosolicdsDNA, we stimulated mouse embryonic fibroblast cells (MEFs)with dsDNA or dsRNA, and mRNA extracted from the cells wassubjected to microarray analysis. Because the B-form dsDNA li-gand, polydeoxyadenine-polydeoxythymidine [poly (dA:dT)], cantrigger multiple pathways including RLRs (18, 19), we used a non–AT-rich sequence dsDNA ligand, IFN stimulatoryDNA (ISD) (8).No genes were strongly and specifically induced by liposome-de-livered ISDcomparedwith a dsRNA synthetic analog polyinosinic-polycytidylic acid [poly (I:C)] [which is dependent on RIG-I/melanoma differentiation-associated gene 5 (MDA5) signaling](Fig. S1A). However, we found that Atm mRNA was preferentially,

Author contributions: T. Kondo, S.A., and T. Kawai designed research; T. Kondo, T.S., andK.M. performed research; J.K., K.J.I., G.N.B., and K.K. contributed new reagents/analytictools; and T. Kondo and T. Kawai wrote the paper.

The authors declare no conflict of interest.1To whom correspondence should be addressed. E-mail: sakira@biken.osaka-u.ac.jp.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1222694110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1222694110 PNAS | February 19, 2013 | vol. 110 | no. 8 | 2969–2974

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

albeit only weakly, up-regulated by ISD. We confirmed theincreased Atm mRNA expression after ISD stimulation usingquantitative reverse transcription-PCR (qPCR) assays (Fig. S1B).These findings imply that exogenous dsDNA mimics signalingassociated with DNA double-strand breaks (DSBs). To verify thishypothesis, we stimulated MEFs with ISD or poly (I:C), and eval-uated ATM activation. ISD, but not poly (I:C), promoted ATMphosphorylation, similar to UV treatment (Fig. 1A). Nevertheless,it has previously been reported that ATM is not required for type IIFN production in response to ISD (8). We also found that therewas no obvious change in IFNB1 mRNA induction betweenATM-deficient AT5BIVA human cells and ATM-complementedAT5BIVA cells (Fig. S1C). Collectively, these findings suggest thepossibility that other DNA repair protein(s) may contribute toimmune responses against intracellular dsDNA.

MRE11 Is Involved in Cytosolic DNA Responses. We next investigatedwhether or not upstream molecules involved in ATM phosphory-lation trigger type I IFN responses. TheMRN complex, containingMRE11, RAD50 homolog (RAD50), and nijmegen breakage

syndrome 1 (NBS1), plays crucial roles in the early phase of DSBsand is required for ATM phosphorylation (20, 21). To addresswhether or not theMRNcomplex is involved in exogenous dsDNAresponses, we stimulated granulocyte macrophage colony-stimu-lating factor (GM-CSF)-induced bone marrow-derived dendriticcells (GM-DCs) prepared from C57BL/6J mice with ISD in thepresence or absence ofMirin, a chemical inhibitor ofMRE11 (22).Mirin diminished the expression of Ifnb1mRNA induced by ISD ina dose-dependent manner (Fig. 1B, Left), whereas Ifnb1 mRNAinduction in response to poly (I:C) was not affected (Fig. 1B,Right). To confirm the contribution of MRE11 to dsDNA res-ponses, we used a knockdown strategy using two distinct siRNAstargeting the Mre11 gene. MEFs treated with these siRNAs dis-played a considerable reduction of Mre11 expression (Fig. 1C). Inaccordance with a previous study (23), reduction of Mre11 ex-pression resulted in destabilization of Rad50. Knockdown ofMre11 led to reduced Ifnb1 induction in response to ISD, but notchange of Ifnb1 induction in response to poly (I:C) (Fig. 1D, Left).In addition, the expression of chemokine (C-X-C motif) ligand 10(Cxcl10) and Interleukin 6 (Il6) was suppressed by Mre11 knock-down (Fig. 1D, Right). Predictably, Rad50 knockdown displayedsimilar results (Fig. S1 D and E). Likewise, Mre11 knockdown inGM-DCs by shRNA abrogated (Fig. 1E), albeit not completely,the induction of Ifnb1 in response to ISD (Fig. 1F). In contrast,Mre11 knockdown did not impair the responses to CpGDNA (Fig.1F). Moreover, production of IFNβ by ISD was also abolished byMre11 knockdown, as measured by ELISA (Fig. S1F).We next examined whether MRE11 is involved in pathogen-

induced type I IFN production. We treated GM-DCs with herpessimplex virus (HSV)-1 and Listeria monocytogenes, which are un-derstood to induce type I IFN production by exposing their ge-nomic DNA. However, Mre11 knockdown did not impair type IIFN induction by these pathogens (Fig. 1G). These results furthersupport the assertion that MRE11 is absolutely required for geneinduction by cytoplasmic dsDNA in various cell types.

Cytosolic DNA Responses Are Abrogated in Ataxia-Telangiectasia–Like Disorder Cells. We examined dsDNA-mediated type I IFNresponses in cells derived fromapatient with ataxia-telangiectasia–like disorder (ATLD), in which the MRE11 protein is truncateddue to a genetic mutation, and both MRE11 and RAD50 proteinsare destabilized (Fig. 2A) (20, 23). In ATLD cells, IFNB1 mRNAexpression was not increased after stimulation with variousdsDNA ligands such as ISD, Escherichia coli DNA, and plasmidDNA (Fig. 2B). In contrast, MRE11-complemented cells wereable to induce IFNB1 against dsDNA stimulation (Fig. 2B). Bothcell types have comparable ability to induce IFNB1 after poly (I:C)stimulation. The response to poly (dA:dT) was partially impairedby loss of MRE11 expression, consistent with a previous reportthat STING is partially involved in poly (dA:dT) responses (11).This unresponsiveness in ATLD cells was observed in a variety ofDNA dosage conditions and IFNB1 mRNA expression was notinduced by 24 h after stimulation (Fig. S2A). We also analyzedtype I IFNproduction by ELISA and confirmed that production ofIFNβ after ISD stimulation was dependent on MRE11 (Fig. 2C).Furthermore, we examined pathogen-induced IFNB1 induction inATLD cells. HSV-1–induced IFNB1 expression was comparablebetween ATLD and MRE11-complemented cells (Fig. S2B).These results strongly suggest that MRE11 is crucial for type IIFN induction following cytosolic DNA stimulation but notHSV-1 infection.

MRE11 Functions Upstream of the STING–TBK1–IRF3 Axis. We nextaddressed whether or not MRE11 functions as a cytosolic sensorfor exogenous dsDNA. Type I IFN responses induced by dsDNAare dependent on TBK1 (7), which promotes phosphorylation ofIRF3 in cytoplasm, followed by translocation of IRF3 to the nu-cleus. As shown in Fig. 2D and Fig. S2C, ATLD cells had defec-tive IRF3 phosphorylation and nuclear translocation in responseto ISD. We then generated ATLD cells stably expressing HA-tagged STING (STING-HA) by retrovirus infection to visualize

A

ATM

p-ATM

-Actin

ISD

Untrea

ted

poly

(I:C)

UV

B

E

GM-DC Ifnb1

400

300

200

100

0

600

Mirin:Stimulation:

mR

NA

fold

indu

ctio

n 500 400

300

200

100

0

500ISD poly (I:C)

C

Mre11

-Actin

EGFPmM11

shRNA:

Rad50

Mre11

-Actin

Contro

l

mM11-2

mM11-3

siRNA:

Rad50

mM11

EGFP250

200

150

100

50

0

mR

NA

fold

indu

ctio

n

Unstim IS

D

GM-DC Ifnb1

CpG

250

200

150

100

50

0

mR

NA

fold

indu

ctio

n

Unstim

GM-DC Ifnb1

HSV-1

Lister

ia

mM11

EGFPF

D

G

mR

NA

fold

indu

ctio

n

Ifnb1

siRNA:

poly (I:C)ISD

80

60

40

20

0

120

100

Contro

l

Contro

l

mM11-2

mM11-3

6000

4000

2000

0

10000

8000

Contro

l

Contro

l

mM11-2

mM11-3

*

**

** **

*

**

40

20

10

0

60 Il-6

50

30

ISD

Contro

l

Contro

l

mM11-2

mM11-3

Cxcl10

800

600

400

200

0

1000

ISD

Contro

l

Contro

l

mM11-2

mM11-3

** **

* **

Fig. 1. MRE11 is involved in dsDNA-mediated type I IFN production. (A)Immunoblot analysis of phospho-ATM proteins. HEK293 cells were trans-fected with 1 μg·mL−1 of ISD or poly (I:C) (with Lipofectamine 2000) for 2 h orwere exposed to UV light (10 Jm−2). Data are from one experiment repre-sentative of two. (B) qPCR analysis in GM-DCs. GM-DCs were transfected with1 μg·mL−1 of ISD or poly (I:C) for 6 h with DMSO or Mirin (10 μM or 100 μM). (Cand D) Immunoblot analysis of MRE11 and RAD50 (C) and qPCR analysis ofmouse Ifnb1, Cxcl10, and Il6 (D) in MEFs treated with a control siRNA or siR-NAs targeting Mre11 (mM11-2 or mM11-3) and stimulated with 1 μg·mL−1 ofISD or poly (I:C) for 6 h. (E–G) Immunoblot analysis of MRE11 and RAD50 (E)and qPCR analysis of mouse Ifnb1 in GM-DCs infected with shRNA-codingretrovirus targeting Mre11 or EGFP (used as control) and stimulated with1 μg·mL−1 of ISD or 250 nM CpG for 6 h (F) or infected with HSV-1 ata multiplicity of infection (MOI) of 10 or L. monocytogenes for 12 h (G).Results for mRNA are represented relative to those of control samples. Dataare from three independent experiments in triplicate (mean and SD in B, D, F,and G). *P < 0.05 and **P < 0.005 compared with controls.

2970 | www.pnas.org/cgi/doi/10.1073/pnas.1222694110 Kondo et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

STING localization. STING-HA was localized at the ER andfailed to translocate to the Golgi in ATLD cells during ISD stim-ulation. In contrast, the translocation was rescued by restoration ofMRE11 expression (Fig. 2E). Similar loss of translocation of IRF3and STING was observed in Mre11-knockdown MEFs (Fig. S2 DandE). Taken together, these results show thatMRE11 is involvedinDNA responses upstreamof STING.Notably, althoughHSV-1–orL.monocytogenes-induced type I IFN production was dependenton STING, we were unable to observe STING translocation aftertreatment with these pathogens in MEFs, ATLD, and MRE11-complemented ATLD cells. This was consistent with our datasuggesting that MRE11 and RAD50 are dispensable for type IIFN production by these pathogens.

MRE11 Senses dsDNA in Cytoplasm. To address whether MRE11recognizes exogenous dsDNA in the cytoplasm, we determined thecellular localization of ISD and MRE11. MRE11 was localized inboth nuclei and cytoplasm with dot-like structures, and rhodamine(ROX)-labeled ISD was partially colocalized with MRE11 in thecytoplasm in MEF and HeLa cells (Fig. 3A and Fig. S3A). Incontrast, there was no colocalization between MRE11 and ROX-poly (I:C). Furthermore, Rad50 was also colocalized with MRE11and exogenous dsDNA (Fig. S3B). To confirm these results, wetransiently transfected MEFs with expression plasmids for GFP-fusedMRE11 or RAD50. Fig. 3B shows that ectopically expressedGFP-MRE11 and GFP-RAD50 were both colocalized with plas-mid DNA in the cytoplasm. We further examined this interactionusing a biochemical binding assay. Cell lysates prepared fromHEK293 cells transfected with biotin-labeled ISD were subjectedto a pull-down assay using streptavidin-conjugated beads. Asshown in Fig. 3C, MRE11, but not RAD50, was detected by im-munoblot, indicating an interaction between MRE11 and DNA.This was consistent with previous reports showing that MRE11 isthe major component of the MRN complex for DNA binding(reviewed in ref. 21). In addition, MRE11 senses sugar-phosphatecontacts and has no base interactions, which is consistent withprevious results that intracellular DNA detection requires a nativesugar-phosphate backbone and lacks sequence preference (8) andwith our own results showing that MRE11 is required for various

dsDNA ligand-induced immune responses (Fig. 2B). Thus, thesefindings suggest that MRE11, rather than RAD50, functions as amajor component to recognize ISD.

NBS1 Is Dispensable for Cytosolic DNA Responses. To further un-derstand howMRE11 induces type I IFN responses to dsDNA, weconstructed a series of MRE11 mutants (Fig. 4A). MEFs wereinfected with retroviruses encodingMRE11mutants (Fig. 4B) andsubjected to qPCR analysis after ISD stimulation. MRE11 hasphosphoesterase motifs in the N-terminal region, which serve toprocess the end ofDNA. Thus, we first examinedwhether nucleaseactivity is required for type I IFN induction. As shown in Fig. 4C,Left, retroviral introduction of FLAG-tagged wild-type hMRE11(WT) in MEFs displayed similar type I IFN induction to that ofempty vector-introduced control cells. In contrast, the H129Nmutation, which lost the nuclease activity (24–26), was found tofacilitate type I IFN induction, suggesting that the nuclease activityhas inhibitory effects on downstream signal transduction. MRE11contains two predictedDNA-binding regions in the central (DB-A)and theC-terminal (DB-B) regions (Fig. 4A). To further investigatethe contributions of these DNA-binding regions, we generatedtwo mutants: ΔDB-A and R633Z. The R633Z mutant, which isidentical to amutation found inATLDcells, has lost theC-terminalDNA binding region. As shown in Fig. 4C,Center, stable expressionof these mutants led to lower induction of type I IFN than controlcells. Interestingly, DB-A showed a small contribution to type IIFN induction compared with DB-B. This result may reflect thefunction served by DB-A to exert nuclease activity by cappingcaptured DNA, a function which is dispensable for forming theDSB complex. On the other hand, the requirement of DB-B isconsistent with the defect of IFN induction in ATLD cells.Furthermore, we addressed the effect of other mutations identi-

fied in patients with ATLD. We introduced N117S or W210C mu-tations, both of which had a loss of binding with NBS1 (23, 27, 28).However, there was no significant change in type I IFN induction byexpression of these mutants (Fig. 4C, Right), suggesting that NBS1-binding is not essential for dsDNA-induced type I IFN production.To further investigate the involvement of NBS1, we performed

binding assays andmicroscopy analysis using GM7166 cells derived

A

C

MRE11

β-Actinm

RN

A fo

ld in

duct

ion

IFNB1

Stimulation:

ATLD2SV + MRE11ATLD2SV

MRE11:

ISD104

103

102

101

100

10-1

E. ColiDNA

104

103

102

101

100

10-1

poly (dA:dT)104

103

102

101

100

10-1

poly (I:C)107

105

103

101

10-1

plasmidDNA

104

103

102

101

100

10-1

D

Mer

geG

M13

0S

TIN

G-H

Aunstimulated ISD

ATLD2SV ATLD2SV+MRE11 ATLD2SV ATLD2SV

+MRE11

P-IRF3

β-Actin

Unstim IS

D

poly

(I:C)

ATLD2SVATLD2SV+MRE11

Unstim IS

D

poly

(I:C)

E

B

ATLD2SV

100

80

60

40

20

0

120

Stimulation:

poly (I:C)2500

2000

1500

1000

500

0

ISD

IFN

-β p

g m

l-1

ND

ND

ND

ND

ND

ATLD2SV + MRE11ATLD2SV

**** **

**

Fig. 2. Hypomorphic mutation of MRE11 inATLD cells drastically abolishes type I IFN pro-duction against various types of DNA ligands.(A) Immunoblot analysis of MRE11 in ATLD2SVand MRE11 complemented cells. (B) qPCR anal-ysis of human IFNB1 in ATLD2SV cells trans-fected with 1 μg·mL−1 of the indicated nucleicacid ligands for 8 h. Results for mRNA are rep-resented relative to those of control samples.Data are from three independent experimentsin triplicate (mean and SD). (C) ELISA of humanIFNβ in ATLD2SV cells transfected with 1 μg·mL−1

of ISD or poly (I:C) for 24 h. Data are fromthree independent experiments in triplicate(mean and SEM). (D) Immunoblot analysis ofphospho-IRF3 proteins in ATLD2SV cells stim-ulated with 5 μg·mL−1 of ISD or poly (I:C) for 3 h.Data are from one experiment representative oftwo. (E) Immunofluorescence staining of stablyexpressed STING-HA in ATLD2SV cells stimulatedwith 5 μg·mL−1 of ISD for 1 h. The localization ofSTING and GM130 (Golgi apparatus marker) wasobserved by confocal microscopy. Data are fromone experiment representative of three. **P <0.005 compared with controls.

Kondo et al. PNAS | February 19, 2013 | vol. 110 | no. 8 | 2971

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

from a patient with Nijmegen breakage syndrome in which NBS1expression was particularly low (Fig. 4D) (29). In these cells,MRE11 colocalized and interacted with exogenous DNA (Fig. 4EandFig. S4A). Furthermore, ISD stimulation significantly increasedlevels of type I IFN expression in GM7166 cells (Fig. 4F). Notably,NBS1 complementation resulted in a reduction of IFNB1 expres-sion. Because it had been reported that loss of NBS1 increasescytosolic distribution of MRE11 (30) (Fig. S4B) and transfectedDNA is mainly sensed in the cytosol (31), this result implies thatNBS1 expression decreases cytosolic DNA sensing by MRE11.Taken together, these results suggest that MRE11 is the major

component required for DNA sensing, whereas its nuclease ac-tivities and one of the binding partners, NBS1, are not requiredfor type I IFN production.

5,6-Dimethylxanthenone-4-Acetic Acid Induction of Type I IFNsDepends on STING but Not MRE11. MRE11 plays crucial roles inDNA repair. In addition, previous reports indicated that DNAdamage can induce type I IFN (15). Thus, we assessed whether ornot DNA damage can trigger the MRE11–STING pathway. Wefound that cisplatin and etoposide treatment induced translocationof STING (Fig. S5A). In addition, this translocation was impairedin ATLD2SV cells, but rescued by replacement of MRE11 (Fig.5A). However, this trafficking was observed in less than 0.1% cells.Indeed, these stimuli were not sufficient to induce gene expressionin MEFs although they modestly induced IFNβ gene (Fig. S5B).Notably, Sting was not required for these weak responses. Theseresults suggest that cisplatin and etoposide slightly activate theMRE11–STING pathway, but is insufficient for type I IFNinduction.Previous studies have shown that a vascular-disrupting agent

(VDA) 5,6-dimethylxanthenone-4-acetic acid (DMXAA) has the

ability to induce type I IFN via TBK1 and IRF3, but not TLR- andRLR-signaling pathways (32). Furthermore, DMXAA inducestype I IFN without activation of NFκB or MAPKs. These similarresponses to DNA prompted us to examine whether DMXAAcan activate the MRE11–STING pathway. First, we investigatedthe dependency on STING. Ifnb1 and Cxcl10 induction afterDMXAA treatment was impaired in Sting-deficient MEFs (Fig.5B). Because DMXAA is implicated in induction of cell deaththrough up-regulation of promyelocytic leukemia (Pml) (Fig. 5B,Right) and down-regulation of Bcl-2 in a type I IFN-dependentmanner (32), we tested whether or not STING affects DMXAA-induced cell death. The TUNEL assay shown in Fig. 5C dem-onstrates that Sting-deficient cells are resistant to DMXAA-in-duced apoptosis, suggesting that STING-mediated geneinduction contributes to DMXAA function. Next, we furtherassessed the effect of DMXAAon the STING pathway. DMXAAtreatment triggered STING translocation, indicating thatDMXAA activates the STING pathway. However, the trans-location also occurred in ATLD cells (Fig. 5D) and Mre11 orRad50 knockdown had no measurable effect on gene expressionchanges induced by DMXAA (Fig. 5E). These results indicatethat DMXAA activates STING independently of MRE11.

DiscussionIn the present study, we identified MRE11 as a sensor for exoge-nous dsDNA, which is required for STING trafficking and type IIFN induction. Our data show that MRE11 contributes to recog-nition of a broad spectrum of dsDNA and the contribution ofMRE11 is not restricted to certain cell types. Analysis of a series ofMRE11 mutants indicated that the nuclease activity of MRE11 isnot required for type I IFN induction. Indeed, theH129N nuclease-inactive form ofMRE11 induces higher amounts of IFNβ than wildtype, indicating that nuclease activity may negatively regulate type IIFN induction. It is speculated that the interaction of MRE11 withdsDNA initially elicits the STING-dependent signaling pathway,and subsequent DNA processing via MRE11 nuclease activityterminates downstream signaling. Thus, MRE11 may also have

A

B

Mergeun

stim

ulat

edR

OX

-pol

y (I:

C)

RO

X-IS

DMre11 ROX

Enlarge

Streptavidinpull-down

MRE11

MRE11

Wholecell lysate

β-Actin

RAD50

RAD50

bio-ISDtransfection

CGFP DAPI

GFP

-hM

RE

11G

F P-h

RA

D50

Merge

Fig. 3. MRE11 and RAD50 are recruited to exogenous dsDNA delivered intothe cytosol. (A) Immunofluorescence staining of MRE11 in MEFs transfectedwith 0.5 μg·mL−1 of ROX-ISD or poly (I:C) for 2 h. Data are from one ex-periment representative of three. (B) Localization of ectopically expressedMRE11 and RAD50. MEFs were transiently transfected with expressionplasmids for GFP-fused MRE11 or RAD50. After 48 h, GFP signals colocalizedwith DAPI stained plasmids were observed. Data are from one experimentrepresentative of two. (C) Binding assay of exogenous DNA and MRE11.HEK293 cells were transfected with 5 μg·mL−1 of biotin-labeled ISD for 2 h.Cell lysates were incubated with streptavidin beads for 1 h and then beadswere washed three times with lysis buffer. Purified proteins were analyzedby immunoblot with indicated antibodies. Data are from one experimentrepresentative of two. (Scale bars, 2 μm.)

A

B

Nuclea

se ac

tivity

DNA bind

ing

RAD50 bi

nding

NBS1 bind

ing

Nuclease DNA bindingA B

708407421

643692

1 246

WT

H129N

N117S

ΔDB-A

W210C

R633Z

**

*

EV

H129N

mR

NA

fold

i nd u

ctio

n

4

3

2

1

0

5

WT

4

3

2

1

0

5

WT

N117S

W210C

MEF Ifnb1

ISD

0.8

0.6

0.4

0.2

0

1.2

WT

1.0

ΔDB-AR63

3Z

FLAG(MRE11)β-Actin

ΔDB-AEV H12

9N

N117S

WT W210C

R633Z

D

E

NBS1β-Actin

GM7166 + NBS1GM7166

NBS1:

IFNB1

mR

NA

fold

indu

ctio

n

Stimulation:

Poly (I:C)10000

1000

100

10

1

0.1

ISD1000

100

10

1

0.1

MergeMRE11 ROX-ISD

F

C

**

***

***

*

Fig. 4. NBS1 is dispensable for intracellular DNA-mediated type I IFN in-duction. (A) Schematic representation of the series of human MRE11 mutantsand their proposed properties. (B and C) Immunoblot analysis of MRE11 (B)and qPCR analysis of mouse Ifnb1 (C) in MEFs stimulated with 1 μg·mL−1 ofISD for 6 h. (D) Immunoblot analysis of NBS1 in GM7166 and complementedcells. (E) Immunofluorescence staining of MRE11 in GM7166 cells treated as inFig. 3A and endogenous MRE11 was observed by immunofluorescencestaining. Data are from one experiment representative of two. (F) qPCRanalysis of human IFNB1 in GM7166 cells transfected with 1 μg·mL−1 of ISD orpoly (I:C) for 8 h. Results for mRNA are represented relative to those ofcontrol samples. Data are from three independent experiments in triplicate(mean and SD in C and F). *P < 0.05 and **P < 0.005 compared with controls.

2972 | www.pnas.org/cgi/doi/10.1073/pnas.1222694110 Kondo et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

a function to suppress excessive DNA responses, similar to Trex1(33). Notably, we found that Mirin, which inhibits ATM activationby targeting MRE11, blocked type I IFN induction. Although it isconsidered thatMirin suppressesMRE11nuclease activity, a recentstudy using H129N knock-in mice demonstrated that the nucleaseactivity of MRE11 is not required for ATM activation (26). Thus,the suppressive effect of Mirin on type I IFN induction is not re-stricted to inhibition of MRE11 nuclease activity. The precisemechanisms by which Mirin modifies MRE11 function to suppressdsDNA-induced type I IFNproduction require future investigation.Results from the microscopy experiments and the analysis of

MRE11 mutants indicate that RAD50 is also involved in in-tracellular DNA responses. However, unlike MRE11, RAD50 ismainly present in the nucleus in resting cells and is recruited tothe MRE11–DNA complex in the cytoplasm after DNA stimu-lation (Fig. S3B). RAD50 has a binding protein, RAD50 interactingprotein 1 (RINT1), which is distributed in the ER during interphaseand forms a complex to regulate ER–Golgi traffic (34, 35), suggestingthe possibility that RINT1 may act as a bridge between MRE11/RAD50 and STING by mediating membrane trafficking. Furtherstudies are required to clarify how RAD50 contributes to exogenousDNA responses. Given that Rad50 knockdown only partially

abrogated poly (I:C)-mediated response (Fig. S1E), it may bepossible that Rad50 is involved in RNA-mediated innate immuneresponses, which should be clarified in the future. Another com-ponent of the MRN complex, NBS1, is not essential for cytosolicDNA responses. NBS1 was shown to be important for recruitmentof MRE11 to endogenous DSB foci, and for the interaction be-tween MRE11 and ATM (30, 36), neither of which seem to benecessary for the exogenous DNA-induced STING pathway.Furthermore, loss of NBS1 increases cytosolic MRE11, sug-gesting that MRE11 recognition of exogenous dsDNA in the cy-toplasm is enhanced by a loss of NBS1.Our present results reveal that MRE11 is not necessary for type

I IFN responses against pathogens such as HSV-1 or L. mono-cytogenes, which are considered to activate immune responses viaintracellular DNA sensors. Although these pathogens induce type IIFN production through STING (11), we failed to observe STINGtranslocation after treatment with these pathogens in various con-ditions. Indeed, our results indicate that MRE11 is specifically in-volved in the responses associatedwith STING trafficking (Fig. 2E).These findings propose that there are multiple pathways; dsDNAinduces STING translocation and pathogens induce STING acti-vation at the ER membrane. The latter pathway may also be im-portant in the case of infection with RNA viruses such as vesicularstomatitis virus (VSV), becauseVSV induces type I IFNproductionthrough STING (11) in the absence of STING translocation. Itremains unclear why HSV-1 and L. monocytogenes do not induceSTING trafficking. One possibility is the location at which DNAresponses are triggered. HSV-1 generally replicates in the nucleusand does not appear to expose genomic DNA to the cytoplasm ofhost cells. IFI16 has been shown to localize in the nucleus in manycells and sense Kaposi’s sarcoma-associated herpesvirus in nucleus(37). Another study has also demonstrated that nuclear localizationsequence of IFI16 locates IFI16 in the nucleus, where it recognizesHSV-1 (31). Furthermore, a cytosolic DNA sensor, AIM2, is un-likely required for inflammasome activation by HSV-1 (38). Inaddition, MRE11/RAD50 seems to affect the trafficking systemsbetween the ER and the Golgi apparatus, suggesting that cyto-plasmic DNA recognition by MRE11/RAD50 influences STINGtrafficking, which is linked to activation of downstream signaling.Alternatively, HSV-1 may possess the evasion machinery to avoidSTING trafficking by collapsing theMRNcomplex, becauseHSV-1circumvents host defense machinery and exploits the MRN com-plex for replication (17). On the other hand, L. monocytogenesresides in the cytoplasm and its genomic DNA is considered a li-gand for the STING pathway (8). It is currently unknown whyL. monocytogenes infection does not cause STING translocation.However, recent studies have shown that cyclic di-GMP of bacteriais a major component responsible for type I IFN induction and,notably, STING directly recognizes cyclic di-GMP and activatesdownstream signaling (39, 40), suggesting that L. monocytogenesdirectly activates STINGwithoutMRE11 andRAD50.Meanwhile,we also found that STING is critical for DMXAA-induced immuneresponses without MRE11. This result implies that DMXAA (thestructure of which ismore similar to cyclic di-GMP than to dsDNA)may interact with STING directly. However, we cannot exclude thepossibility that other unknown receptors for DMXAA trigger theSTING pathway.Our observations that MRE11/RAD50 mediates recognition of

dsDNA rather than pathogens suggest that the biological signifi-cance of MRE11-mediated intracellular DNA recognition is torespond to damaged host cells, rather than defense against foreignpathogens. In resting eukaryote cells, DNA is strictly stored inparticular compartments such as the nucleus andmitochondria andthe release of endogenous DNA to the cytoplasm or extracellularmilieu provides signals that alert the host cell to danger. WhereasDNases mediate the clearance of self DNA in the apoptotic state(5), an excessive amount of DNA escaping from DNases is re-sponsible for induction of type I IFN, probably through activationof DNA sensors. In addition to its function within the nucleus toguard the integrity of the genome, MRE11 may have acquiredadditional functions against various stresses from the extracellular

B

ED

C Sting+/+ Sting−/−

DAPITUNEL

Medium

DMXAA

DAPITUNEL

Sting −/−Sting +/+

MEF

mR

NA

fold

i ndu

ctio

n

16

12

8

4

0

Ifnb120

DMXAA:

8000

6000

4000

2000

0

Cxcl1012000

100006

4

2

0

Pml8

mR

NA

fold

indu

ctio

n

Ifnb1

DMXAA:

800

600

400

200

0

1200

ControlsiRNA:

MEF

1000

mM11-3 mR50-1

Aun

stim

ulat

ed

ATLD2SV ATLD2SV+MRE11

STING-HA

DM

XA

AATLD2SV ATLD2SV

+MRE11

unst

imul

ated

cisp

latin

STING-HA

****

**

***

Fig. 5. DMXAA-mediated gene expression requires STING but notMRE11. (A)Immunofluorescence staining of stably expressed STING-HA in ATLD2SV cellsstimulated with 10 μM cisplatin for 24 h. (B) qPCR analysis of mouse Ifnb1,Cxcl10, and Pml inwild-typeand Sting−/−MEFs treatedwith 50 μg·mL−1 DMXAAfor 4 h. (C) TUNEL assay in wild-type and Sting−/−MEFs treated with 50 μg·mL−1

DMXAA for 12 h. Data are from one experiment representative of two. (D)Immunofluorescence staining of stably expressed STING-HA in ATLD2SV cellsstimulated with 100 μg·mL−1 DMXAA for 1 h. Data are from one experimentrepresentative of two. (E) qPCR analysis of mouse Ifnb1 in MEFs treated witha control siRNAor siRNAs (mM11-3 ormR50-1) and stimulatedwith 100 μg·mL−1

DMXAA for 4 h. Results for mRNA are represented relative to those of controlsamples. Data are from three independent experiments in triplicate (mean andSD in B and E). *P < 0.05 and **P < 0.005 compared with controls.

Kondo et al. PNAS | February 19, 2013 | vol. 110 | no. 8 | 2973

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

environment over the course of evolution, in a similar manner toATM, which has been shown to function as a redox sensor (41). Anumber of studies have demonstrated that inappropriate pro-duction of type I IFN is harmful and leads to autoimmune diseases(5). Further studies with regard to MRE11-mediated DNAsensing may be helpful for better understanding the mechanismsof DNA-associated autoimmunity.In conclusion, we have demonstrated a critical function ofMRE11

in intracellular dsDNA responses. Our data provide knowledgeabout the role of theMRN complex in the recognition of dsDNAin the cytoplasm. Although the consequences of activity of theMRN complex in innate immune responses in vivo remain to beelucidated, these findings suggest a rationale for further studies,which will enable us to understand the precise mechanisms ofnucleic acid recognition, and the biological significance of this onhost defense against pathogens, as well as responses to damagedhost or tumor cells.

Materials and MethodsRNA Interference. Double-strandedRNAduplexes correspondingtomouseMre11(MSS206766 and MSS206767) and Rad50 (MSS208386 and MSS276712) mRNAwere purchased from Invitrogen. A nonspecific siRNA (45–2001; Invitrogen) was

used as a negative control. MEFs were transfected with 50 nM siRNA using Lip-ofectamine2000 (Invitrogen)according to themanufacturer’s instructions.At60hafter transfection, the cells were used for further experiments.

Quantitative RT-PCR. Total RNA of cultured cells was isolated using TRIzolreagent (Invitrogen) and reverse transcribed with ReverTra Ace (Toyobo)according to the manufacturer’s instructions. PCR analysis was performedwith power SYBR Green (Applied Biosystems) and the primers described in SIMaterials and Methods.

Statistical Analysis.Differences were analyzed for statistical significance usingStudent’s t test. A P value of less than 0.05 was considered significant.

Additional Information. The details are available in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank M. Nishino and C. Funamoto for technicalassistance, the members of the Laboratory of Host Defense for valuablediscussion, and E. Kamada and M. Kageyama for secretarial assistance. Thisstudy was supported by a KAKENHI Grant-in-Aid for Research Activity (A)(23689030) and supported by the Special Coordination Funds of the JapaneseMinistry of Education, Culture, Sports, Science and Technology; as well as theMinistry of Health, Labour, and Welfare in Japan and the Japan Society for thePromotion of Science through the Funding Program for World-Leading Inno-vative R&D on Science and Technology (FIRST Program).

1. Hornung V, Latz E (2010) Intracellular DNA recognition. Nat Rev Immunol 10(2):123–130.

2. Desmet CJ, Ishii KJ (2012) Nucleic acid sensing at the interface between innate andadaptive immunity in vaccination. Nat Rev Immunol 12(7):479–491.

3. Hemmi H, et al. (2000) A Toll-like receptor recognizes bacterial DNA. Nature 408(6813):740–745.

4. Baccala R, Hoebe K, Kono DH, Beutler B, Theofilopoulos AN (2007) TLR-dependentand TLR-independent pathways of type I interferon induction in systemic autoim-munity. Nat Med 13(5):543–551.

5. Nagata S, Hanayama R, Kawane K (2010) Autoimmunity and the clearance of deadcells. Cell 140(5):619–630.

6. Okabe Y, Kawane K, Akira S, Taniguchi T, Nagata S (2005) Toll-like receptor-independentgene induction program activated by mammalian DNA escaped from apoptotic DNAdegradation. J Exp Med 202(10):1333–1339.

7. Ishii KJ, et al. (2006) A Toll-like receptor-independent antiviral response induced bydouble-stranded B-form DNA. Nat Immunol 7(1):40–48.

8. Stetson DB, Medzhitov R (2006) Recognition of cytosolic DNA activates an IRF3-de-pendent innate immune response. Immunity 24(1):93–103.

9. Takaoka A, et al. (2007) DAI (DLM-1/ZBP1) is a cytosolic DNA sensor and an activatorof innate immune response. Nature 448(7152):501–505.

10. Ishii KJ, et al. (2008) TANK-binding kinase-1 delineates innate and adaptive immuneresponses to DNA vaccines. Nature 451(7179):725–729.

11. Ishikawa H, Ma Z, Barber GN (2009) STING regulates intracellular DNA-mediated, typeI interferon-dependent innate immunity. Nature 461(7265):788–792.

12. Saitoh T, et al. (2009) Atg9a controls dsDNA-driven dynamic translocation of STINGand the innate immune response. Proc Natl Acad Sci USA 106(49):20842–20846.

13. Unterholzner L, et al. (2010) IFI16 is an innate immune sensor for intracellular DNA.Nat Immunol 11(11):997–1004.

14. Zhang Z, et al. (2011) The helicase DDX41 senses intracellular DNA mediated by theadaptor STING in dendritic cells. Nat Immunol 12(10):959–965.

15. Kim T, et al. (1999) Activation of interferon regulatory factor 3 in response to DNA-damaging agents. J Biol Chem 274(43):30686–30689.

16. Karpova AY, Trost M, Murray JM, Cantley LC, Howley PM (2002) Interferon regulatoryfactor-3 is an in vivo target of DNA-PK. Proc Natl Acad Sci USA 99(5):2818–2823.

17. Weitzman MD, Lilley CE, Chaurushiya MS (2010) Genomes in conflict: Maintaininggenome integrity during virus infection. Annu Rev Microbiol 64:61–81.

18. Ablasser A, et al. (2009) RIG-I-dependent sensing of poly(dA:dT) through the in-duction of an RNA polymerase III-transcribed RNA intermediate. Nat Immunol 10(10):1065–1072.

19. Chiu YH, Macmillan JB, Chen ZJ (2009) RNA polymerase III detects cytosolic DNA andinduces type I interferons through the RIG-I pathway. Cell 138(3):576–591.

20. Uziel T, et al. (2003) Requirement of the MRN complex for ATM activation by DNAdamage. EMBO J 22(20):5612–5621.

21. Stracker TH, Petrini JH (2011) The MRE11 complex: Starting from the ends. Nat RevMol Cell Biol 12(2):90–103.

22. Dupré A, et al. (2008) A forward chemical genetic screen reveals an inhibitor of theMre11-Rad50-Nbs1 complex. Nat Chem Biol 4(2):119–125.

23. Stewart GS, et al. (1999) The DNA double-strand break repair gene hMRE11 is mu-tated in individuals with an ataxia-telangiectasia-like disorder. Cell 99(6):577–587.

24. Moreau S, Ferguson JR, Symington LS (1999) The nuclease activity of Mre11 is re-quired for meiosis but not for mating type switching, end joining, or telomeremaintenance. Mol Cell Biol 19(1):556–566.

25. Arthur LM, et al. (2004) Structural and functional analysis of Mre11-3. Nucleic AcidsRes 32(6):1886–1893.

26. Buis J, et al. (2008) Mre11 nuclease activity has essential roles in DNA repair andgenomic stability distinct from ATM activation. Cell 135(1):85–96.

27. Lee JH, et al. (2003) Regulation of Mre11/Rad50 by Nbs1: Effects on nucleotide-dependent DNA binding and association with ataxia-telangiectasia-like disordermutant complexes. J Biol Chem 278(46):45171–45181.

28. Fernet M, et al. (2005) Identification and functional consequences of a novel MRE11mutation affecting 10 Saudi Arabian patients with the ataxia telangiectasia-likedisorder. Hum Mol Genet 14(2):307–318.

29. Ito A, et al. (1999) Expression of full-length NBS1 protein restores normal radiationresponses in cells from Nijmegen breakage syndrome patients. Biochem Biophys ResCommun 265(3):716–721.

30. Kobayashi J, et al. (2002) NBS1 localizes to gamma-H2AX foci through interactionwith the FHA/BRCT domain. Curr Biol 12(21):1846–1851.

31. Li T, Diner BA, Chen J, Cristea IM (2012) Acetylation modulates cellular distributionand DNA sensing ability of interferon-inducible protein IFI16. Proc Natl Acad Sci USA109(26):10558–10563.

32. Roberts ZJ, et al. (2007) The chemotherapeutic agent DMXAA potently and specifi-cally activates the TBK1-IRF-3 signaling axis. J Exp Med 204(7):1559–1569.

33. Stetson DB, Ko JS, Heidmann T, Medzhitov R (2008) Trex1 prevents cell-intrinsic ini-tiation of autoimmunity. Cell 134(4):587–598.

34. Xiao J, Liu CC, Chen PL, Lee WH (2001) RINT-1, a novel Rad50-interacting protein,participates in radiation-induced G(2)/M checkpoint control. J Biol Chem 276(9):6105–6111.

35. Arasaki K, Taniguchi M, Tani K, Tagaya M (2006) RINT-1 regulates the localization andentry of ZW10 to the syntaxin 18 complex. Mol Biol Cell 17(6):2780–2788.

36. You Z, Chahwan C, Bailis J, Hunter T, Russell P (2005) ATM activation and its re-cruitment to damaged DNA require binding to the C terminus of Nbs1. Mol Cell Biol25(13):5363–5379.

37. Kerur N, et al. (2011) IFI16 acts as a nuclear pathogen sensor to induce the in-flammasome in response to Kaposi Sarcoma-associated herpesvirus infection. CellHost Microbe 9(5):363–375.

38. Rathinam VA, et al. (2010) The AIM2 inflammasome is essential for host defenseagainst cytosolic bacteria and DNA viruses. Nat Immunol 11(5):395–402.

39. McWhirter SM, et al. (2009) A host type I interferon response is induced by cytosolicsensing of the bacterial second messenger cyclic-di-GMP. J Exp Med 206(9):1899–1911.

40. Burdette DL, et al. (2011) STING is a direct innate immune sensor of cyclic di-GMP.Nature 478(7370):515–518.

41. Guo Z, Kozlov S, Lavin MF, Person MD, Paull TT (2010) ATM activation byoxidative stress. Science 330(6003):517–521.

2974 | www.pnas.org/cgi/doi/10.1073/pnas.1222694110 Kondo et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0