DNA single-strand break-induced DNA damage response causes heart failure

The DNA damage response (DDR) plays a pivotal role in maintaining genome integrity. DNA damage and DDR activation are observed in the failing heart, however, the type of DNA damage and its role in the pathogenesis of heart failure remain elusive. Here we show the critical role of DNA single-strand break (SSB) in the pathogenesis of pressure overload-induced heart failure. Accumulation of unrepaired SSB is observed in cardiomyocytes of the failing heart. Unrepaired SSB activates DDR and increases the expression of inflammatory cytokines through NF-κB signalling. Pressure overload-induced heart failure is more severe in the mice lacking XRCC1, an essential protein for SSB repair, which is rescued by blocking DDR activation through genetic deletion of ATM, suggesting the causative role of SSB accumulation and DDR activation in the pathogenesis of heart failure. Prevention of SSB accumulation or persistent DDR activation may become a new therapeutic strategy against heart failure.

V arious internal and external stresses induce many types of DNA damage such as chemical change of bases, single-strand break (SSB) and double-strand break (DSB). On DNA damage, the DNA damage response (DDR) signalling is activated to repair damaged DNA. Recruitment and autophosphorylation of ataxia telangiectasia mutated (ATM) kinase are one of the most well characterized DDR, which are usually activated after DNA DSB 1 . Phosphorylated ATM in turn phosphorylates multiple effector proteins including the histone variant H2AX and tumor suppressor p53, and triggers downstream signalling pathways that stop the cell cycle and repair damaged DNA 2,3 . When the DNA damage is too extensive to be repaired, cells usually undergo apoptosis or stop the cell cycle permanently, which is termed cellular senescence 2,4 . Activation of DDR is observed not only in mitotic cells but also in post-mitotic cells including cardiomyocytes [5][6][7] . Various types of DNA damage including DNA oxidation, SSB and DSB are observed in the infarcted heart and activation of DDR plays an important role in cardiac remodelling after myocardial infarction through inducing cardiomyocyte apoptosis [7][8][9] . Activation of DDR is also observed in cardiomyocytes of the patients with end-stage heart failure 5 and the pressure-overload induced heart failure model mice 6 , however, the type of DNA damage and the role of DDR in the pathophysiology of heart failure remain unclear.
In the present study, we show that DNA SSB, but not DNA DSB, is accumulated in cardiomyocytes of pressure-overload induced heart failure model mice. Accumulation of DNA SSB activates persistent DDR and induces inflammatory gene expression in an NF-kB-dependent manner. Cardiomyocytespecific gene deletion of Xrcc1, an essential protein for SSB repair, leads to more severe cardiac inflammation and heart failure after pressure-overload, which is rescued by simultaneous deletion of Atm gene. These results suggest the unrecognized role of DNA SSB in the pathogenesis of pressure overload-induced heart failure.

Results
DNA single-strand break accumulation in heart failure. We performed transverse aortic constriction (TAC) surgery on mice to induce pressure overload-induced heart failure. Left ventricular hypertrophy was observed in 2 weeks, and progressive cardiac dysfunction was observed in 8-10 weeks after the TAC surgery ( Supplementary Fig. 1). To elucidate the type of DNA damage in the failing heart, we isolated cardiomyocytes after the TAC surgery and analysed the type of DNA damage by comet assay. Comet tail moment in the alkaline condition (alkaline comet) reflects both DNA DSB and SSB whereas comet tail moment in the neutral condition (neutral comet) reflects only DNA DSB 10 . To optimize the experimental condition of the comet assays, we used cardiomyocytes isolated from doxorubicin-treated mice and confirmed that neutral comet assay could indeed detect DNA DSB in the heart ( Supplementary Fig. 2a). We found that the alkaline comet tail moment of cardiomyocytes was increased time dependently after TAC operation whereas neutral comet tail moment was comparable between Sham-and TAC-operated mice (Fig. 1a,b), suggesting that the number of cardiomyocytes with more SSB, but not DSB, is increased in the heart after pressure overload. To further analyse the type of DNA damage in the failing heart, we performed in situ oligo ligation (ISOL), which enables in situ detection of DNA DSB 11 , and immunofluorescent staining for NBS1, which is a component of MRE11-RAD50-NBS1 DSB repair complex and used as a DNA DSB marker 12 . The number of ISOL-and NBS1-positive cardiomyocytes was increased in the heart after myocardial infarction ( Supplementary Fig. 2b,c) whereas they were not increased in the heart after TAC operation (Fig. 1c-f), underscoring the results of the comet assay. Notably, the number of ISOL-positive non-cardiomyocytes was increased after TAC operation ( Supplementary Fig. 2d). We cannot totally exclude the possibility that the cells with DNA DSB are cleared by phagocytes too quickly to be detected by these assays (neutral comet assay, ISOL and NBS1 staining), or the sensitivity of these assays were not enough to detect DNA DSB in cardiomyocytes, however, it is very likely that the major type of DNA damage that accumulates in cardiomyocytes after pressure-overload is DNA SSB, but not DSB.
Despite our observation that DNA SSB was increased in cardiomyocytes after TAC operation, we found that the level of protein poly(ADP-ribosylation), which is induced by PARP1/2 as an initial process of SSB repair 13 , and the expression levels of SSB repair-related genes were rather decreased in the TAC-operated heart ( Fig. 1g-i). Together with the findings that the levels of reactive oxygen species (ROS) were significantly increased ( Fig. 1j-l), decreased SSB repair activity as well as increased ROS production may play a role in SSB accumulation in the heart after TAC operation ( Supplementary Fig. 3).
Xrcc1 deficiency exacerbates heart failure. In mitotic cells, unrepaired SSB usually does not accumulate but develops into DNA DSB and induces catastrophic cell death during mitotic phase 14 . Therefore, accumulation of SSB is supposed to be a unique phenomenon observed only in post-mitotic cells 15 . To examine the causal relation between SSB accumulation and heart failure, we generated the mice with defective SSB repair. X-ray repair complementing defective repair in Chinese hamster cells 1 (XRCC1) is a scaffold protein that interacts with various SSB repair enzymes and essential for SSB repair 16 . Since Xrcc1 knockout mice exhibit early embryonic lethality due to massive DNA damage and cellular apoptosis 17 , we crossed the mice homozygous for an Xrcc1 flox allele with transgenic mice expressing Cre-recombinase under the control of the a-myosin heavy chain promoter (a-myosin MHC-Cre tg ) 18,19 and obtained Xrcc1 flox/flox ; aMHC-Cre tg (Xrcc1 aMHC-Cre ) mice as cardiomyocyte-specific Xrcc1 knockout mice. Xrcc1 flox/flox ; aMHC-Cre À mice (Xrcc1 f/f ) were used as controls. Deletion of Xrcc1 in the heart tissue of Xrcc1 aMHC-Cre mice was confirmed at the DNA, mRNA and protein levels ( Supplementary Fig. 4a-c). Xrcc1 aMHC-Cre mice were born in expected Mendelian ratios, and were viable and fertile. Echocardiographic analysis revealed mild cardiac dysfunction in Xrcc1 aMHC-Cre mice (Supplementary Fig. 4d), but the mice showed no overt signs of heart failure such as tachypnea, weight loss and pleural effusions beyond 1-year follow-up ( Supplementary Fig. 4e), suggesting that deletion of Xrcc1 in cardiomyocytes is not sufficient to induce heart failure. After the TAC surgery, however, Xrcc1 aMHC-Cre mice exhibited more severe left ventricular dilatation and dysfunction (Fig. 2a,b), more severe signs of heart failure ( Fig. 2c) and higher mortality (Fig. 2d) compared with Xrcc1 f/f mice. Comet assay revealed that the level of SSB in cardiomyocytes tended to increase, but statistically comparable between shamoperated Xrcc1 aMHC-Cre and Xrcc1 f/f mice (Fig. 2e), suggesting that cardiomyocyte-specific deletion of Xrcc1 by itself is not sufficient to induce massive SSB accumulation in the non-stressed heart. After the TAC operation, however, the level of SSB in cardiomyocytes was more increased in the heart of Xrcc1 aMHC-Cre mice compared with Xrcc1 f/f mice (Fig. 2e), whereas the level of DSB in cardiomyocytes was comparable between Xrcc1 aMHC-Cre and Xrcc1 f/f mice ( Supplementary Fig. 5). These results collectively suggest that insufficient SSB repair due to deletion of Xrcc1 induces more SSB accumulation and exacerbates cardiac dysfunction in response to pressure overload and also indicate that SSB accumulation is tightly associated with the progression of pressure overload-induced heart failure.
SSB activates DDR and induce inflammation through NF-jB. To elucidate the molecular mechanism by which SSB accumulation in cardiomyocytes exacerbates pressure overload-induced heart failure, we established an in vitro model of cardiomyocytes  with SSB accumulation by using alkylating agent methyl methanesulfonate (MMS), which induces only DNA SSB 20 , and small interfering RNA (siRNA)-mediated knockdown of Xrcc1. To optimize the experimental condition of the comet assays and ISOL staining in in vitro cultured cardiomyocytes, we used irradiation to induce DNA DSB and confirmed that DNA DSB can be detected in both assays ( Supplementary Fig. 6a,b). We observed concentration-dependent, temporal SSB after 10-min treatment with MMS ( Fig. 3a,b) and repetitive treatment with MMS generated unrepaired SSB accumulation in cultured cardiomyocytes (Fig. 3c,d). Knockdown of Xrcc1 (Fig. 3e) also generated a time-dependent accumulation of SSB in cultured cardiomyocytes (Fig. 3f). DNA DSB was not induced either by MMS treatment or by knockdown of Xrcc1 (Fig. 3a,b,d,f and Supplementary Fig. 6a-d).
We then analysed how DDR is activated after SSB induction. Single treatment with MMS induced phosphorylation of ATM and H2AX whereas phosphorylation of p53 was not observed (Fig. 4a). Repetitive treatment with MMS or knockdown of Xrcc1, however, induced phosphorylation of p53 as well as ATM and H2AX (Fig. 4b,c) that resembles to the DDR after ionizing radiation or doxorubicin, which causes DNA DSB (Supplementary Fig. 7a,b). Persistent activation of DDR usually induces apoptotic cell death or cellular senescence 21 . Recent reports suggest that senescent cells show inflammatory phenotype, which is termed senescence-associated secretory phenotype (SASP) 22,23 . Therefore, we tested whether SSB accumulation and persistent activation of DDR induce inflammatory phenotype in cardiomyocytes. Knockdown of Xrcc1 increased the expression of inflammatory cytokines such as Il6, Cxcl1, Ccl2 and Vcam1 in cardiomyocytes suggesting that SSB accumulation leads to acquisition of inflammatory phenotype in cardiomyocytes. Persistent phosphorylation/activation of ATM plays a central role in acquisition of SASP in senescent cells 4 . Persistent activation of DDR and increased expression of inflammatory cytokines after knockdown of Xrcc1 were abolished by simultaneous knockdown of Atm ( Fig. 4d and Supplementary Fig. 7c), suggesting that ATM is essential for SSB accumulation-induced acquisition of inflammatory phenotype in cardiomyocytes.
NF-kB plays a key role during inflammation. Previous report also suggests the involvement of NF-kB in DDR [24][25][26] . We also observed that the number of cardiomyocytes with nuclear NF-kB staining was increased after knockdown of  )  SSB accumulation-induced NF-kB activation and inflammatory gene expression in cardiomyocytes ( Fig. 4g and Supplementary  Fig. 7d,e). These results collectively suggest that SSB accumulation induces inflammatory cytokine expression through persistent activation of DDR and subsequent activation of NF-kB pathway in cardiomyocytes.

Xrcc1 deficiency exacerbates DDR and cardiac inflammation.
We then investigated whether our in vitro findings are also observed in vivo. The number of cardiomyocytes positive for phosphorylated H2AX (gH2AX) was increased after TAC operation ( Supplementary Fig. 8a,b) and the number was more increased in the heart of Xrcc1 aMHC-Cre mice compared with Xrcc1 f/f mice (Fig. 5a,b). The amount of ROS in the heart was comparable between Xrcc1 aMHC-Cre and Xrcc1 f/f mice ( Fig. 5c and Supplementary Fig. 9a,b), suggesting that increased number of gH2AX-positive cardiomyocytes in Xrcc1 aMHC-Cre mice was due to defective SSB repair and SSB accumulation but not overproduction of ROS in Xrcc1 aMHC-Cre mice 27 . Chromatin immunoprecipitation (ChIP) analyses revealed that binding of NF-kB to the Vcam1 promoter was increased twofold in TAC-operated Xrcc1 aMHC-Cre cardiomyocytes compared with Xrcc1 f/f cardiomyocytes (Fig. 5d). Expression levels of inflammatory cytokines and infiltration of inflammatory cells were increased in cardiomyocytes after TAC operation, and were more increased in TAC-operated Xrcc1 aMHC-Cre mice compared with Xrcc1 f/f mice ( Fig. 5e-g). These in vivo results are consistent with our in vitro observations and strongly support our hypothesis that accumulation of SSB in cardiomyocytes exacerbates heart failure through the activation of persistent DDR and promoting cardiac inflammation.  ATM deletion rescues heart failure in Xrcc1-deficient mice. To further investigate whether SSB accumulation in cardiomyocytes plays a causative role in the pathogenesis of heart failure through activating DDR, we crossed Xrcc1 aMHC-Cre mice with Atm þ / À mutants 28,29 to block persistent DDR and obtained Xrcc1 aMHC-Cre ; Atm þ / À mice ( Supplementary Fig. 10a-d). Mild cardiac dysfunction observed in Xrcc1 aMHC-Cre mice was restored in Xrcc1 aMHC-Cre ; Atm þ / À mice (Fig. 6a,b). The level of SSB in cardiomyocytes which tended to be increased in Xrcc1 aMHC-Cre mice remained increased in Xrcc1 aMHC-Cre ; Atm þ / À mice ( Fig. 6c and Supplementary Fig. 11a,b), however, the number of gH2AX-positve cardiomyocytes that was increased in Xrcc1 aMHC-Cre mice was reduced to the level of Xrcc1 f/f mice in the heart of Xrcc1 aMHC-Cre ; Atm þ / À mice (Fig. 6d,e), suggesting that heterozygous Atm gene deletion indeed suppressed DDR activation which was induced by DNA SSB. Expression levels of   inflammatory cytokines in cardiomyocytes (Fig. 6f) and infiltration of inflammatory cells (Fig. 6g,h) were mildly increased in the heart of Xrcc1 aMHC-Cre mice and were attenuated in Xrcc1 aMHC-Cre ; Atm þ / À mice with the exception of Cxcl10 and Vcam1 gene expression. There was an apparent discrepancy between the level of SSB, which tended to increase but did not reach statistical significance, and the other cardiac phenotypes in Xrcc1 aMHC-Cre mice. This discrepancy may be explained by the presence of threshold between the level of DNA SSB and the downstream events in cardiomyocytes. Just like there is a threshold between the level of DNA damage and cell death (cell death is triggered when the level of DNA damage is above the threshold), it is conceivable to think that persistent DDR and the following inflammatory program are triggered when the level of unrepaired SSB in each cell is above the threshold.

Sham
TAC4w Sham  H2AX (a, gH2AX, green). Immunostaining for alpha-actinin (red) was used to label cardiomyocytes. Arrowheads indicate gH2AX-positive cardiomyocytes and arrows indicate gH2AX-positive non-cardiomyocytes. Scale bar, 50 mm. The number of gH2AX-positive cardiomyocytes was counted (b, n ¼ 4 each). Statistical significance was determined by one-way analysis of variance (ANOVA) followed by the Tukey-Kramer HSD test. **Po0.01 between arbitrary two groups. (c) Heart tissue sections were stained with dihydroethidium and mean fluorescence intensity relative to Sham-operated Xrcc1 f/f mice was measured (n ¼ 5 each). Statistical significance was determined by one-way ANOVA followed by the Tukey-Kramer HSD test. **Po0.01 between arbitrary two groups. (d) ChIP-qPCR analysis of binding of NF-kB to the Vcam1 promoter region in Sham-or TAC-operated Xrcc1 f/f and Xrcc1 aMHC-Cre mice. Data are presented as fold enrichment relative to Sham-operated Xrcc1 f/f mice (n ¼ 4, 4, 5, 5, respectively). Statistical significance was determined by one-way ANOVA followed by the Tukey-Kramer HSD test. *Po0.05 between arbitrary two groups. (e) The expression levels of inflammatory cytokines in the isolated cardiomyocytes of Sham-or TAC-operated Xrcc1 f/f and Xrcc1 aMHC-Cre mice was assessed by real-time PCR (n ¼ 18, 18, 18, 28, respectively, technical duplicates). Statistical significance was determined by one-way ANOVA followed by the Tukey-Kramer HSD test. **Po0.01 between arbitrary two groups. (f,g) Heart tissues of Sham-or TAC-operated Xrcc1 f/f and Xrcc1 aMHC-Cre mice were immunostained for CD45 or CD68 (f, green). Immunostaining for alpha-actinin (red) was used to label cardiomyocytes. Arrowheads indicate CD45-or CD68-positive cells. Scale bar, 50 mm. The number of CD45-and CD68-positive cells was counted (g, n ¼ 5 each). Statistical significance was determined by one-way ANOVA followed by the Tukey-Kramer HSD test. **Po0.01 between arbitrary two groups. Column and error bars show mean and s.e.m., respectively. We finally performed the TAC surgery to these mice and found that heterozygous Atm gene deletion attenuated the progression of severe left ventricular dysfunction (Fig. 7a,b), alleviated the signs of heart failure (Fig. 7c) and improved the mortality in Xrcc1 aMHC-Cre mice (Fig. 7d). The level of SSB in cardiomyocytes remained high in Xrcc1 aMHC-Cre ; Atm þ / À mice, however, the field     H2AX (f, gH2AX, green, arrowheads). Arrowheads indicate gH2AX-positive cardiomyocytes and arrows indicate gH2AX-positive non-cardiomyocytes. Scale bar, 50 mm. The number of gH2AX-positive cardiomyocytes was counted (g, n ¼ 4 each). Expression levels of inflammatory cytokines in the isolated cardiomyocytes were assessed by real-time PCR (h, n ¼ 10, 16, 12 for each genotype, respectively, technical duplicates). ChIP-qPCR analysis of binding of NF-kB to the Vcam1 promoter region. Data is presented as fold enrichment relative to TAC-operated Xrcc1 f/f mice (i, n ¼ 4, 5, 5, respectively). Heart tissues were immunostained for CD45 or CD68 (j, green, arrowheads). Arrowheads indicate CD45-or CD68-positive cells. Scale bar, 50 mm. The number of CD45-and CD68-positive cells was counted (k, n ¼ 4 each). Statistical significance was determined by one-way analysis of variance followed by the number of gH2AX-positve cardiomyocytes was reduced to the level of Xrcc1 f/f mice ( Fig. 7e-g), suggesting that heterozygous Atm gene deletion suppressed DDR activation also after the TAC operation. The expression level of inflammatory cytokines in cardiomyocytes (Fig. 7h), binding of NF-kB to the Vcam1 promoter (Fig. 7i) and infiltration of inflammatory cells (Fig. 7j,k) were also attenuated in Xrcc1 aMHC-Cre ; Atm þ / À mice compared with Xrcc1 aMHC-Cre mice. These results collectively suggest that SSB-induced activation of DDR plays a causal role in cardiac inflammation and heart failure in response to pressure overload.

Discussion
Accumulation of unrepaired oxidative DNA damage and activation of DDR have been observed in the failing heart 5,6,30 , however, their roles in the pathogenesis of heart failure remain elusive. In the present study, we identified the type of DNA damage which triggers DDR in cardiomyocytes and demonstrated that activation of DDR contributes to, at least in part, cardiac inflammation in the failing heart. We found that unrepaired DNA SSB accumulates in cardiomyocytes of pressure overload-induced heart failure model mice. Accumulation of DNA SSB induced persistent DDR and upregulated the expression of inflammatory cytokines through NF-kB pathway. Accumulation of DNA SSB was more increased in the mice lacking XRCC1, an essential protein for SSB repair. Increased level of SSB accumulation was associated with more activation of DDR, expression of inflammatory cytokines and infiltration of inflammatory cells, and also with exacerbation of cardiac dysfunction after pressure overload. Worse cardiac phenotype that was observed in the mice with defective SSB repair was rescued by blocking DDR activation through Atm gene deletion collectively suggesting the important role of SSB accumulation and subsequent activation of DDR in the pathogenesis of heart failure (Fig. 8).
In the present manuscript, we showed that accumulation of unrepaired SSB plays a causative role in the pathogenesis of heart failure. Unrepaired SSB develops into DNA DSB during mitosis 14 in mitotic cells, therefore, accumulation of SSB is supposed to be specific to post-mitotic cells such as neurons or cardiomyocytes 15 . Accumulation of unrepaired SSB in neural tissue is observed in spinocerebellar ataxia with axonal neuropathy 1 (SCAN1) 15 . Mutation of SSB repair enzyme, Tyrosyl-DNA phosphodiesterase 1, is observed in the patients of SCAN1 (refs 31,32) and DNA SSB is accumulated in the neural tissue of Tdp1 À / À mice 33 . Our findings indicate that SSB accumulation also occurs in stressed, post-mitotic cells and plays pathogenic role in non-genetic disorders. Compared with the pathogenic role of DSB, knowledge about the roles of SSB in human diseases is limited. Future study may reveal the role of SSB in the pathogenesis of various genetic or non-genetic diseases.
ATM becomes activated in response to DNA damage to stop the cell cycle and promote DNA damage repair. Its activation is essentially protective for the organism to maintain homeostasis. In the present study, however, we show that ATM may also play a detrimental role in pressure overload-induced heart failure through NF-kB-dependent induction of inflammatory gene expression. Recent reports also suggested that aberrant activation of ATM plays detrimental roles in myocardial infarction model mice, that is, ATM promotes cardiac inflammation during myocardial ischemia 34 and cardiac remodelling after myocardial infarction are attenuated in ATM knockout mice 34,35 . Further investigations for the precise molecular action of ATM against chronic DNA damage may provide a novel therapeutic target for heart failure.
On DNA SSB, PARP-1 is rapidly recruited to the damaged site, where it synthesizes branched ADP-ribose polymers (pADPr) on SSB repair proteins including PARP-1, XRCC1 and DNA ligases and facilitates SSB repair 13,36 . It was surprising for us to see that protein poly(ADP-ribosylation) and the expression levels of the genes that work for SSB repair were decreased in the failing heart ( Fig. 1g-i), despite the production of ROS and the amount of DNA damage was increased (Fig. 1a,b,j-l), suggesting that decreased SSB repair activity as well as increased ROS production may play a role in SSB accumulation in the heart after TAC operation (Supplementary Fig. 3). We do not know, at the moment, why SSB repair system become downregulated in the failing heart, but downregulation of SSB repair system may have a role in preventing cardiomyocyte death against massive DNA damage. Since the source of ADP-ribose to form pADPr is nicotinamide adenine dinucleotide (NAD þ ), hyperactivation of PARP-1 on excess DNA damage consumes the intracellular NAD þ pool, leading to exhaustion of ATP and necrotic death 13 . Hyperactivation of PARP-1 is also reported to induce caspaseindependent cell death termed parthanatos 37,38 . Although the system for eliminating the cells with massive DNA damage is beneficial to prevent cancer formation, it might become critical for the tissue consisted of post-mitotic cells such as cardiomyocytes. Cardiomyocytes may have intrinsic mechanism to downregulate DNA damage repair system for cell (and organ) survival, at the expense of DNA damage accumulation.
In summary, we identified a previously unrecognized role of DNA SSB in the pathogenesis of heart failure. Accumulation of SSB in cardiomyocytes activated DDR and persistent DDR promoted cardiac inflammation in pressure overload-induced heart failure. From a clinical perspective, approaches that promote efficient SSB repair or suppress aberrant DDR may become a novel therapeutic strategy against heart failure.
To generate cardiomyocyte-specific Xrcc1 knockout mice with an Atm þ / À background, Xrcc1 aMHC-Cre mice were crossed with Atm þ / À mice (Jackson Laboratory; stock#0008536, stock name: B6.129S6-Atm tm1Awb /J) and cardiomyocyte-specific Xrcc1 knockout mice with systemic loss of one allele of ATM (Xrcc1 aMHC-Cre ; Atm þ / À mice) were obtained. Genotyping PCR was performed according to the protocol from Jackson Laboratory. Expected PCR product sizes of Atm wild-type and mutant allele were 147 and 280 bp, respectively.
TAC operation and echocardiography. TAC surgery was performed to 10-to 12-week-old male mice 39 . The transverse aorta was constricted with a 7-0 silk suture paralleled with a 27-gauge blunt needle, which was removed after constriction. Sham-operated animals, which underwent a similar surgical procedure without aortic constriction, were used as controls. TAC surgery was performed to the mice randomly selected from those weighing 25-30 g at the age of 10-12 weeks. Surgeon was not informed about the genotypes of the mice. The mice which died within 1 day after the operation were excluded from the analysis. Transthoracic echocardiography was performed on conscious mice with a Vevo 770 imaging system (Visualsonics, Inc.). M-mode echocardiographic images were obtained from a longitudinal view to measure the LV size and LV function.
Measurement of ROS production. To measure ROS production, heart samples were immediately embedded in Tissue-Tek OCT cryo-embedding compound (Miles Laboratories) and stored at À 80°C. Unfixed frozen samples were cut into 5 mm-thick sections and placed on glass slides. Tissue section was covered with dihydroethidium (DHE, Sigma, 10 mM) and incubated in a light-protected humidified chamber for 30 min at 37°C. Ethidium fluorescence (excitation at 490 nm, emission at 610 nm) was examined by fluorescent microscopy and mean fluorescence intensity was measured by using ImageJ software.
To measure the level of hydrogen peroxide, mice hearts were excised and cut into small pieces. After the incubation in Krebs-HEPES buffer (128 mM NaCl, 2.5 mM KCl, 1 mM MgSO 4 , 2.7 mM CaCl 2 , 16 mM D-glucose, 20 mM HEPES), hydrogen peroxide levels per tissue weight were measured using Amplex red assay (Invitrogen) according to the manufacturer's instructions.
ChIP-qPCR. Isolated cardiomyocytes were crosslinked with 0.3% formaldehyde at room temperature for 10 min. Samples were sonicated using Bioruptor Plus (Diagenode). After sonication, soluble chromatin was incubated with an anti-NF-kB antibody (5 mg, sc-372X, Santa Cruz). Specific immunocomplexes were precipitated with Dynabeads Protein G (Life Technologies). Immunoprecipitates were washed, reverse-crosslinked and purified by Wizard SV Gel and PCR Clean-Up System (Promega). Extracted DNA was used for quantitative PCR (qPCR). Fold enrichment was determined as the fold change in per cent input (ChIP signal/input signal) at the target region compared to the control region. Primers were designed to amplify the Vcam1 promoter region containing putative NF-kB-binding sites. The control primers were designed against non-conserved and non-repetitive sequences at 25 kbp upstream of Vcam1 gene. Quantitative real-time PCR was performed using Universal Probe Library (UPL, Roche) and Light Cycler TaqMan Master kit (Roche). Primer sequences were as follows: target region: 5 0 -TAGAGAGGCGGAGGGAAATC-3 0 and 5 0 -GCTTCATGATGGC AAGTGG-3 0 ; the control region: 5 0 -GAAATCATCATTTGTGTATCTCGAA-3 0 and 5 0 -AGGGCTTGTTGATTCTGCTC-3 0 .
Statistical analysis. All values are presented as mean ± s.e.m. Two-group comparison was analysed by unpaired two-tailed Student's t-test or Mann-Whitney U-test. Multiple group comparison was performed with one-way analysis of variance followed by the Tukey-Kramer HSD test or Steel-Dwass test for comparison of arbitrary two groups. Survival curves after the TAC surgery were analysed by Kaplan-Meier method and comparison of the two groups was performed with Wilcoxon test. Significant differences were defined as Po0.05. The optimal sample size (n ¼ 26) to detect a difference of survival rate after pressure overload between Xrcc1 f/f and Xrcc1 aMHC-Cre mice was calculated based on our preliminary data that indicate the median survival time of TAC-operated Xrcc1 aMHC-Cre mice is 60 days. Since the actual median survival time of TAC-operated Xrcc1 aMHC-Cre mice in the present study was 54 days, we reduced the optimal sample size of Xrcc1 aMHC-Cre ; Atm þ / À mice (n ¼ 23). We used a power of 80% and a type I error probability of 0.05. Sample size calculations were performed using the software Power and Sample Size Calculations. All the statistical analyses were reviewed by Professor Ayumi Shintani, a statistical expert at Department of Clinical Epidemiology and Biostatistics, Osaka University, Japan.
Data availability. The data that support the findings of this study are available from the corresponding author on reasonable request.