DNA repair in cardiomyocytes is critical for maintaining cardiac function in mice

Abstract Heart failure has reached epidemic proportions in a progressively ageing population. The molecular mechanisms underlying heart failure remain elusive, but evidence indicates that DNA damage is enhanced in failing hearts. Here, we tested the hypothesis that endogenous DNA repair in cardiomyocytes is critical for maintaining normal cardiac function, so that perturbed repair of spontaneous DNA damage drives early onset of heart failure. To increase the burden of spontaneous DNA damage, we knocked out the DNA repair endonucleases xeroderma pigmentosum complementation group G (XPG) and excision repair cross‐complementation group 1 (ERCC1), either systemically or cardiomyocyte‐restricted, and studied the effects on cardiac function and structure. Loss of DNA repair permitted normal heart development but subsequently caused progressive deterioration of cardiac function, resulting in overt congestive heart failure and premature death within 6 months. Cardiac biopsies revealed increased oxidative stress associated with increased fibrosis and apoptosis. Moreover, gene set enrichment analysis showed enrichment of pathways associated with impaired DNA repair and apoptosis, and identified TP53 as one of the top active upstream transcription regulators. In support of the observed cardiac phenotype in mutant mice, several genetic variants in the ERCC1 and XPG gene in human GWAS data were found to be associated with cardiac remodelling and dysfunction. In conclusion, unrepaired spontaneous DNA damage in differentiated cardiomyocytes drives early onset of cardiac failure. These observations implicate DNA damage as a potential novel therapeutic target and highlight systemic and cardiomyocyte‐restricted DNA repair‐deficient mouse mutants as bona fide models of heart failure.


| INTRODUC TI ON
Heart failure has become a global disease epidemic, particularly in the elderly (Bui et al., 2011;Lloyd-Jones et al., 2002).
Notwithstanding major advances in treatment, the exact molecular mechanisms underlying the pathogenesis of heart failure remain incompletely understood hampering a more mechanism-based approach to effective treatment and prevention. Experimental and clinical evidence indicates that DNA damage, for example, due to oxidative stress or following chemotherapy (Octavia et al., 2012), is associated with heart failure (Bartunek et al., 2002;Higo et al., 2017;Shukla et al., 2010). In addition, accumulation of unrepaired DNA damage is linked with segmental premature ageing phenotypes in several organ systems (Hoeijmakers, 2009;Lopez-Otin et al., 2013;Vermeij, Hoeijmakers, & Pothof, 2016).
However, the precise effects of unrepaired endogenous DNA damage, and its role in the pathogenesis of heart failure, remain elusive. In the present study, we tested the hypothesis that DNA repair in cardiomyocytes is critical for maintaining normal cardiac function. For this purpose, we generated mice with genetically depleted DNA repair mechanisms in either all cell types or restricted to cardiomyocytes. XPG and ERCC1-XPF, two major players in DNA repair, are structure-specific endonucleases responsible for excision of DNA lesions in nucleotide excision repair (NER) and transcription-coupled repair (TCR). Moreover, XPG is implicated in promoting base excision repair of oxidative DNA damage and double-strand break (DSB) repair (Marteijn et al., 2014;Trego et al., 2016). ERCC1-XPF is additionally involved in interstrand cross-link repair, and sub-pathways of DSB repair (Gregg et al., 2011). Xpg −/− , Ercc1 −/− and Ercc1 Δ/− mice (and corresponding human syndromes) show many progressive progeroid characteristics (e.g., cachexia, neurodegeneration, osteoporosis and impaired growth) and a severely shortened lifespan (Barnhoorn et al., 2014;Weeda et al., 1997). In addition to these systemic mouse mutants, we generated mice with genetically depleted Xpg and Ercc1 in differentiated cardiomyocytes using an alpha-myosin heavy chain (αMHC)-promoter-driven Cre transgene (mice referred as αMHC-Xpg c/− and αMHC-Ercc1 c/− respectively). The cardiomyocyterestricted mutants were employed to circumvent the potentially confounding influence on the heart of premature ageing in other organs associated with systemic loss of Xpg and Ercc1. Both systemic and cardiomyocyte-restricted mutants were investigated to elucidate the cell-autonomous effect of endogenous DNA damage on cardiac function.

| Deficient DNA repair results in timedependent deterioration of global LV function and severely shortened lifespan in all mutants
Ercc1 −/− mice showed severe growth retardation and a dramatically reduced lifespan. Consequently, these mice could only be studied at 8 weeks at which time point survival rate was 39% ( Figure S1).
We therefore also included Ercc1 Δ/− mice, in which the Δ allele only partially inactivates the function of Ercc1 due to a seven amino-acid carboxy-terminal truncation; consequently, lifespan, compared with Ercc1 −/− mice, is extended to 4-6 months and Ercc1 Δ/− mice exhibit a milder degenerative phenotype, consistent with a milder mutation (Weeda et al., 1997). To achieve cardiomyocyte-restricted Xpg and S3b), indicating that Xpg or Ercc1 heterozygosity in all cell types did not affect cardiac function. This is compatible with the autosomal recessive nature of these genetic defects and the general absence of overt phenotypes in obligate carriers of defects in most human nucleotide excision repair disorders, such as Xeroderma Pigmentosum and Cockayne syndrome (Hoeijmakers, 2009;Vermeij, Hoeijmakers, & Pothof, 2016).
Xpg −/− and Ercc1 Δ/− mice exhibited reduced growth, followed by body weight loss after 8 weeks of age, and premature death, in striking contrast to apparent normal body growth of αMHC-Xpg c/− and αMHC-Ercc1 c/− mice ( Figure 1a and Figure S4b,c).
Cardiomyocyte-restricted inactivation of Xpg and Ercc1 also reduced lifespan in both male and female mice, albeit less than in Xpg −/− and Ercc1 Δ/− mice, respectively ( Figure S4c,d), suggesting that defective DNA repair in the heart is not lifespan limiting in the systemic mutants. As reported previously (Vermeij, Dollé, et al., 2016;Vermeij, Hoeijmakers, & Pothof, 2016), the more severe repair defect in the absence of Ercc1 relative to Xpg correlates with a shorter lifespan compared with Xpg inactivation, consistent with an inverse dose-response relationship between unrepaired DNA damage and lifespan ( Figure S4c). Importantly, echocardiography at 4 weeks of age showed normal cardiac function in all mutants, indicating unaffected heart development ( Figure 1a). Over time, all DNA repair-deficient mutants gradually developed bradycardia and contractile dysfunction compared with corresponding control (Figure 1a

| Increased cardiac expression of foetal genes in Xpg mutants
Based on the survival curves ( Figure S4c Figure S7). In addition, increased protein levels of β-MHC and decreased maximal rate of force redevelopment (max K tr ; changes typically associated with LV hypertrophy and failure (Hamdani et al., 2008)) were observed in Xpg mutants, reaching statistical significance in Xpg −/− mice ( Figure 3c,d). These changes explain, at least partly, the diastolic dysfunction in the Xpg mutants.

| Enhanced extracellular matrix turnover and increased levels of cell loss in Xpg mutants
Histological analysis revealed a small decrease in cardiomyocyte cross-sectional area in Xpg −/− , but not in αMHC-Xpg c/− mice, paralleling changes in heart weight ( Figure 4a). Collagen fraction tended to increase ( Figure 4a) and extracellular matrix turnover was enhanced ( Figure S8a,b), which may have contributed not only to the cardiac remodelling but also to the fractionated QRS pattern observed on surface ECG analysis in vivo ( Figure S8c), suggestive of cardiac conduction delay (Rizzo et al., 2012). Moreover, αMHC-Xpg c/− mice displayed bradycardia and significantly increased PR interval, P duration and QRS interval ( Figure S8d).
These ECG parameters were already affected at the age of 8 weeks, while these parameters were unaffected at the age of 4 weeks in αMHC-Xpg c/− mice ( Figure S8e). Collagen content was still unchanged in hearts of 8-week-old αMHC-Xpg c/− mice (αMHC-Xpg c/− 1.54 ± 0.13%, n = 5 vs. corresponding control 1.39 ± 0.24%, n = 5; p = 0.60), indicating that cardiac conduction abnormalities preceded the development of structural alterations and were not the consequence of the latter. While (atrio-)ventricular conduction was not affected in Xpg −/− mice, significant QT prolongation was observed. This was likely the consequence of the low heart rate in these mice since no differences were observed in QT interval corrected for heart rate (QTc; Figure S8d).
To assess the purported role of oxidative stress in heart failure, basal superoxide generation was measured in homogenates using lucigenin-enhanced chemiluminescence. Both NOX activity (using NADPH as substrate) and NOX-dependent superoxide production (using NOX inhibitor VAS2870) were increased, particularly in Xpg −/− mice, resulting in elevations in total superoxide production ( Figure S9). Since oxidative stress is a major cause of apoptosis (Kannan & Jain, 2000), molecular imaging was performed to determine the level of apoptosis in the in vivo heart using the near-infrared fluorescent Annexin-Vivo™ 750 probe, combined F I G U R E 1 Impaired DNA repair resulted in time-dependent deterioration of global LV function in all mutants (Study I, lifespan studies). (a) Effect of Xpg and Ercc1 deficiency on body weight, LV mass and geometry and hemodynamic parameters during age. LV mass, left ventricular mass, calculated using the VisualSonics Cardiac Measurements Package; LVEDD, LV end-diastolic lumen diameter. The number of animals is indicated in the body weight graph. All mutants have their own corresponding control littermates. Data are presented as mean ± SEM. (b) Representative LV short axis M-mode images of the different DNA repair-deficient mutants and corresponding control at age 16 weeks. *p < 0.05 vs. corresponding control; †p < 0.05 genotype × age using mixed linear model-repeated measures. 1.82 ± 0.32‰, n = 6 vs. corresponding control 1.73 ± 0.29‰, n = 5; p = 0.85) and collagen content (as mentioned previously) were observed between αMHC-Xpg c/− and corresponding control, indicating that apoptosis and fibrosis start at a later age, further underscoring the degenerative nature of the disease process. Cellular stress, such as DNA damage, activates the intrinsic apoptotic pathway (Giam et al., 2008). Indeed, apoptosis in advanced age αMHC-Xpg c/− mice was accompanied by increased expression of phorbol-12-myristate-13-acetate-induced protein 1 (Pmaip1), Bcl-associated X protein (Bax), Bcl2-related ovarian killer protein (Bok) and Bcl2-modifying factor (Bmf) ( Figure S11b). αMHC-Xpg c/− . TP53 is known to trigger cell-cycle arrest, apoptosis or DNA repair in response to DNA damage and thereby activates specific genes (Lakin & Jackson, 1999). In advanced age αMHC-Xpg c/− , 438 TP53-target genes were activated (Table S2)

| Association of genetic variants in ERCC1 and XPG with cardiac remodelling and dysfunction
To study the effect of genetic variation in the ERCC1 and XPG gene on cardiac function in humans, a genome-wide association study (GWAS) of LVEDD and fractional shortening was performed in up to 7653 individuals from the population-based Rotterdam Study (Ikram et al., 2020). The effect estimates for the common genetic variants located in the genes and those 50kB up-and downstream were extracted. A total of 26 genetic variants in ERCC1 were associated with LVEDD (p-value <0.05), of which two single-nucleotide polymorphisms (SNPs; 19:45967369 and 19:45976718) showed an opposite effect for LVEDD and fractional shortening (Table S3) F I G U R E 2 LV-remodeling and LV dysfunction in DNA repair-deficient mutants at a specific time point (study II). (a) Effect of Xpg and Ercc1 deficiency on LV mass and geometry and (b), hemodynamic parameters in 16-week-old Xpg −/− and αMHC-Xpg c/− , 8-week-old Ercc1 −/− , 16-week-old Ercc1 Δ/− and αMHC-Ercc1 c/− mice and corresponding control (n = 7-16 animals/group). LV weight, left ventricular weight; BW, body weight; LVEDD, LV end-diastolic lumen diameter; LVdP/dt max , maximum rate of rise of LV pressure; LVdP/dt min , maximum rate of fall of LV pressure; tau, relaxation time constant; LVEDP, LV end-diastolic pressure. Data are presented as mean ± SEM. *p < 0.05 vs. corresponding control by two-way ANOVA followed by SNK post hoc testing.

F I G U R E 3
Altered cardiomyocyte contractile properties in 16-week-old Xpg −/− mice, but not in 16-week-old αMHC-Xpg c/− mice (Study II). (a) Schematic experimental myocyte set-up. (b) Isometric force measurements in single permeabilized cardiomyocytes (n = 6 animals/group; 1-4 cardiomyocytes/animal). F max , maximal force; F pas , passive force. (c) Determination of Ca 2+ sensitivity of force (pCa 50 ) and maximal rate of force redevelopment (K tr ) (n = 6 animals/group; 1-4 cardiomyocytes/animal). (d) Myosin heavy chain isoform composition was determined by protein levels of myosin heavy chain isoforms (n = 6 animals/group). Beta-myosin heavy chain (β-MHC) content is expressed as % of total MHC. (e) Phos-tag analysis of cardiac troponin I (cTnI) species, expressed as % of total phosphorylated cTnI (n = 6-7 animals/ group). Phosphorylated cTnI species, of similar molecular weight, were separated into three forms: Non-phosphorylated (nonP), monophosphorylated (1P) and bis-phosphorylated (2P). These forms were visualized with a specific antibody against troponin I, which recognizes non-phosphorylated and phosphorylated forms. Data are presented as mean ± SEM. *p < 0.05 vs. corresponding control by two-way ANOVA followed by SNK post hoc testing.
showed an opposite effect for fractional shortening and LVEDD (Table S4). For XPG, a total of 14 SNPs were associated with LVEDD, of which four SNPs (13:103467184, 13:103457497, 13:103527113 and 13:103548761) showed an opposite effect for LVEDD and fractional shortening (Table S3). Conversely, five SNPs were associated with fractional shortening (p-value <0.05), none of which showed an effect with LVEDD (Table S4). Taken together, these findings reveal genetic variants that may be implicated in the development of heart failure in humans. Of notice, the observation that some of these genetic variants have opposite effects on LVEDD and fractional shortening makes these variants interesting candidates for further study.

| DISCUSS ION
Despite major advances in treatment, the exact molecular mechanisms underlying the pathogenesis of heart failure remain incompletely understood hampering a more mechanism-based approach to effective treatment and prevention. Experimental and clinical evidence indicates that DNA damage-due to oxidative stress (e.g., ischemia-reperfusion injury) or after chemotherapy (Octavia et al., 2012)-is associated with heart failure (Bartunek et al., 2002;Higo et al., 2017;Shukla et al., 2010). However, the precise effects of unrepaired DNA damage, and its role in the pathogenesis of heart failure, are still poorly understood. In the present study, we tested the hypothesis that DNA repair in cardiomyocytes is critical for in human GWAS data were found to be associated with cardiac remodelling and dysfunction. The implications of these findings will be discussed.

| Importance of sufficient DNA repair for the maintenance of cardiac function
Unrepaired DNA damage can lead to pronounced cellular dysfunction and promote disease development (Lopez-Otin et al., 2013). The structure-specific endonucleases XPG and ERCC1-XPF, two major players in DNA repair, are responsible for the excision of DNA lesions and are involved in several DNA repair mechanisms (Gregg et al., 2011;Marteijn et al., 2014;Trego et al., 2016). Previously, using cardiomyocyte-restricted Ercc1 knockout mice, genome instability in the heart, measured by variant calling in RNA and genomic DNA, was found to be strongly elevated compared with wildtype mice, and several other prematurely ageing mouse models (De Majo et al., 2021).
The finding of enhanced mutations is consistent with the repair deficiency. However, small-scale mutations are now being recognized not to be a major driver of ageing (Robinson et al., 2021), in line with the growing notion that DNA damage rather than mutations is relevant for ageing-associated functional decline in post-mitotic organs and tissues (Schumacher et al., 2021) and consistent with our finding of DNA damage-driven transcription stress, which we also found to be enhanced as apparent from gene expression data sets of ageing in the heart (Gyenis, Chang, et al., Nat. Genetics in press). In the pre- Importantly, we observed normal cardiac function in all DNA repair-deficient mutants at 4 weeks of age, after which all mutants gradually developed contractile dysfunction. These observations indicate that loss of DNA repair induces a degenerative rather than a developmental process in the heart. These results are consistent with observations in mice in which deficient DNA repair is induced by depletion of Xrcc1, an essential protein for single-strand break repair. In pressure overload-induced heart failure, accumulation of single-strand breaks was associated with more activated DNA damage response and deteriorated cardiac dysfunction in Xrcc1 deficient mutants (Higo et al., 2017).
In the present study, cardiac dysfunction in both systemic and cardiomyocyte-restricted mutants was associated with a markedly reduced lifespan. Interestingly, cardiomyocyte-restricted loss of Xpg and Ercc1 resulted in a slightly longer lifespan compared with mice with systemic loss of Xpg and Ercc1, suggesting that the heart is not lifespan limiting in the systemic mutants. This is also supported by the relatively milder degree of cardiac dysfunction in the systemic (fractional shortening ~30%) versus the cardiomyocyte-restricted (~10%) mutants just prior to death. Conversely, the cardiomyocyterestricted mutants displayed severe signs of congestive heart failure in the final week, suggesting that mortality in these mice was related to the severe loss of cardiac function. It has been shown that survival rates of DNA repair-deficient mouse models depend on the extent to which DNA repair mechanisms are affected, with Ercc1 mutants showing an even more severe phenotype than Xpg mutants (Vermeij, Dollé, et al., 2016). Indeed, we observed a slightly worse survival in the cardiomyocyte-restricted Ercc1 mutants compared with the Xpg mutants. From the present study, we cannot determine the exact cause of death, that is progressive pump failure or fatal ventricular arrhythmias, in the cardiomyocyte-restricted mutants. However, the observation that the surface ECG analysis did not show any sign of ventricular arrhythmias, together with the severe signs of congestive heart failure, including dyspnoea, pulmonary oedema and ascites, suggests that the cause of death most likely was progressive pump failure.

| Mechanism of cardiac dysfunction
The mechanisms underlying the observed cardiac dysfunction in the Xpg and Ercc1 mutants could be several-fold, including altered cardiomyocyte contractile properties, enhanced extracellular matrix deposition and increased levels of senescence and myocardial cell loss.
DNA repair-deficient mutants demonstrated severe systolic LV dysfunction, which could not be explained by a loss of maximum force development of individual cardiomyocytes, as F max was maintained in both Xpg mutants. We also observed diastolic LV dysfunction in Xpg mutants, which could be explained, at least in the systemic Xpg mutants, by significant elevations in F pas , myofilament Ca 2+ sensitivity, and a decrease in max K tr . The increased Ca 2+ sensitivity was likely due to the lower PKA-mediated cTnI phosphorylation, while reduced max K tr is explained by the shift towards the β-MHC isoform. The maintained force development, the increased Ca 2+ sensitivity and lower cTnI phosphorylation are in excellent agreement with the observed alterations in cardiomyocyte function in patients with end-stage dilated cardiomyopathy (DCM;Hamdani et al., 2008Hamdani et al., , 2010. We also found evidence of extracellular matrix remodelling, as we observed a 50%-80% increase in LV myocardial collagen content at 16 weeks of age, with an additional 240% increase by the end of the cardiomyocyte-restricted Xpg mutants' lifespan. These observations are similar to the elevated collagen volume fraction in ventricular biopsies of end-stage DCM patients (Hamdani et al., 2010) and may have contributed to the progressive deterioration of cardiac function.
DNA damage can interfere with the vital process of transcription or induce replication arrest, which can trigger cellular senescence or cell death (Hoeijmakers, 2009). Several studies with different DNA repair-deficient mouse mutants concerning other organ systems showed elevated levels of senescence markers, including increased expression of p21 and p16, and increased senescence-associated β galactosidase activity (Barnhoorn et al., 2014;Vermeij, Dollé, et al., 2016). In line with those previous studies, we observed increased expression of p21 and p16 in the liver of 16-week-old systemic Xpg mutants. Interestingly, expression of p21 and p16 was not elevated in the Xpg mutant heart at 16 weeks of age, which could be interpreted to suggest that differentiated cardiomyocytes with depleted Xpg appear may be highly sensitive to DNA damage, responding with early apoptotic cell death. Indeed, examination of apoptosis showed significantly elevated levels of apoptotic cell death in both Xpg mutants, already at 16 weeks, which was markedly (fivefold) further increased in the cardiomyocyte-restricted Xpg mutants at the end of their lifespan. These observations support the concept that the transition from cardiac dysfunction to overt heart failure is due to increased apoptosis (Li et al., 1997;Marin-Garcia, 2016).
There is evidence that DNA damage can drive apoptosis in cardiomyocytes (Higo et al., 2017;Shukla et al., 2011). The mechanisms underlying DNA damage-induced apoptosis may involve activation of TP53, a key regulator of apoptosis. Thus, in the accompanying manuscript by Henpita et al. (2023), unrepaired DNA damage in differentiated cardiomyocytes resulted in activation of p53, while genetic depletion of p53 attenuated the level of apoptosis. In line with their observations, the dramatic loss of cardiomyocytes at the end of the cardiomyocyte-restricted Xpg mutants' lifespan was associated with activation of TP53 and elevated expression of the pro-apoptotic genes Bax, Pmaip1, Bok and Bmf. Oxidative stress has been implicated in the pathogenesis of heart failure. In the failing human heart, due to dilated cardiomyopathy and acute myocardial infarction, elevated levels of the oxidative DNA damage marker 8-oxo-7,8-dihydro-2′deoxyguanosine and DNA repair enzymes were detected in serum and in cardiomyocytes (Bartunek et al., 2002;Kono et al., 2006).

Interestingly, both Xpg mutants showed NOX-dependent increases
in superoxide production, which may have contributed to apoptosis, possibly via activation of TP53, and the progressive deterioration of cardiac pump function in the DNA repair-deficient mutants.
Gene set enrichment analysis showed that the subnetworks of apoptosis and DNA damage were enriched. The NER pathway plays a critical role in neurodegenerative diseases, such as Alzheimer's, Parkinson's and Huntington's disease. Impaired DNA repair, due to defects in the NER pathway, causes severe neurodevelopmental F I G U R E 4 Characterization of LV phenotype in 16-week-old Xpg mutants (Study II). (a) Representative images of whole hearts and haematoxylin and eosin-stained two-chamber view sections and examination of total heart weight (n = 13-19 animals/group). Representative gomori-and picro-sirius red-stained LV sections and quantification of, respectively, cardiomyocyte cross-sectional area and myocardial collagen content (n = 8 animals/group). (b) Representative images and quantification of the in vivo visualization of the membranebound phospholipid phosphatidylserine to detect early apoptosis using FMT combined with μCT (n = 3-6 animals/group). (c) Representative TUNEL-stained LV sections and quantification of TUNEL-positive nuclei (indicated by arrows) to detect late apoptotic cells (n = 8 animals/ group). Data are presented as mean ± SEM. *p < 0.05, §p = 0.062, #p = 0.087 vs. corresponding control by two-way ANOVA followed by SNK post hoc testing.
abnormalities (Sepe et al., 2013). Identification of diseases related to the differentially expressed genes in our dataset revealed that cardiomyocyte-restricted loss of Xpg, affected genes that are involved in these DNA damage-related neurodegenerative diseases, consistent with the concept that DNA damage is present in our Xpgmutant mice.

| Association of genetic variants in ERCC1 and XPG with cardiac remodelling and dysfunction
To study the effect of genetic variation in the ERCC1 and XPG gene on cardiac remodelling and dysfunction in humans, a genome-wide association study of LVEDD and fractional shortening was performed in the Rotterdam Study. These results revealed that some of the genetic variants in ERCC1 and XPG genes were associated with an increase in LVEDD and a decrease in fractional shortening (corresponding with the cardiac phenotype in our mutant mice), suggesting that these variants might be implicated in the development of heart failure in humans.

| Conclusions
The present study, in conjunction with the accompanying study by Henpita et al. (2023), demonstrates that unrepaired endogenously generated DNA damage in differentiated cardiomyocytes drives early onset of cardiac failure. Since Xpg and Ercc1 are both involved in NER, TCR and pathways of DSB repair, these genome stability mechanisms protect against cardiac failure. Importantly, even in the presence of these potent repair mechanisms, DNA damage inevitably accumulates over time even in normal cells, as some types of DNA damage are not recognized by global genome repair systems, while others are irreparable (Hoeijmakers, 2009). Moreover, agerelated decline in repair also occurs (Moriwaki & Takahashi, 2008;Vermeij, Dollé, et al., 2016). Hence, time-dependent accumulation of DNA lesions may well contribute to the aetiology of heart failure, implicating DNA damage as a novel therapeutic target. Moreover, our work highlights systemic and cardiomyocyte-restricted DNA repair-deficient mutants as bona fide models of heart failure well suited for identifying novel therapeutic interventions.

| E XPERIMENTAL PROCEDURE S
For detailed experimental procedures, see Extended Methods in Appendix S2.
These controls were wildtype in all cell types, except for cardiomyocytes, which were heterozygous. All animals described in this study were born at a predicted Mendelian frequency and have the same genetic C57BL6/FVB F1 hybrid background to minimize background-specific effects. All animals were bred and maintained with ad libitum access to water and AIN93G synthetic pellets

| Experimental design
Two study protocols were performed. Study I consisted of a lifespan study, based on which we designed Study II, in which animals were studied in great detail and sacrificed after a pre-determined followup period.

| Study I
Lifespan studies were performed to characterize body growth, as well as global cardiac geometry and function in the various DNA repair-deficient mutants described above. For this purpose, animals were weighed and subsequently underwent transthoracic echocardiography, every 2-4 weeks, until death (DNA repair-deficient mutants) or 30 weeks of age (control mice).

| Study II
To study cardiac geometry, function and structure in more detail, Xpg −/− , αMHC-Xpg c/− , Ercc1 Δ/− , αMHC-Ercc1 c/− mice and corresponding control littermates were sacrificed at the age of 16 weeks. Since αMHC-Ercc1 c/− mice are deficient for Ercc1 in their cardiomyocytes, we added an additional group of Ercc1 −/− animals to Study II. However, as these animals have a much shorter lifespan (~8-10 weeks), we studied this additional group at 8 weeks of age.

| Statistical analysis animal studies
Data from study I (lifespan study) were compared using mixed linear model-repeated measures (SPSS Statistics 21.0). Data from study II were compared using one-way or two-way ANOVA, followed by Student-Newman-Keuls (SNK) post hoc testing when appropriate (SigmaPlot 11.0). Comparison of variables between two groups at a single time point was performed by unpaired Student's t-test

| Study cohort
In accordance with the observed phenotype in Xpg and in Ercc1 mutants, the association of common single-nucleotide polymorphisms (SNPs) in both genes with enlarged LV lumen diameter and deteriorated fractional shortening were investigated. To this end, a genome-wide association study (GWAS) of these two measures was performed in the population-based Rotterdam Study (RS). The RS is a prospective cohort study, initially comprising 7983 individuals (78% of 10,215 invitees), 55 years of age or over, living in the well-defined Ommoord district in the city of Rotterdam, The Netherlands (RS-I, recruitment 1990(RS-I, recruitment -1993. The RS aims to examine determinants of disease and health in older subjects focussing on neurogeriatric, cardiovascular, bone and eye diseases (Ikram et al., 2020). The RS has been extended several times. In 2000, RS was extended to include individuals who had become 55 years of age or moved into the study district since the start of the study (RS-II, including 3011 participants

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on reasonable request to the corresponding author.