DNA repair‐deficient premature aging models display accelerated epigenetic age

Abstract Several premature aging mouse models have been developed to study aging and identify interventions that can delay age‐related diseases. Yet, it is still unclear whether these models truly recapitulate natural aging. Here, we analyzed DNA methylation in multiple tissues of four previously reported mouse models of premature aging (Ercc1, LAKI, Polg, and Xpg). We estimated DNA methylation (DNAm) age of these samples using the Horvath clock. The most pronounced increase in DNAm age could be observed in Ercc1 mice, a strain which exhibits a deficit in DNA nucleotide excision repair. Similarly, we detected an increase in epigenetic age in fibroblasts isolated from patients with progeroid syndromes associated with mutations in DNA excision repair genes. These findings highlight that mouse models with deficiencies in DNA repair, unlike other premature aging models, display accelerated epigenetic age, suggesting a strong connection between DNA damage and epigenetic dysregulation during aging.


| DNA REPAIR DEFI CIENT MOUS E MODEL S E XHIB IT ACCELER ATED EPI G ENE TI C AG E
Here, to assess the relevance of several premature aging mouse models (Ercc1, Xpg, LAKI and Polg mice), we analyzed the epigenetic age of multiple tissues using a DNA methylation clock, known as the "Horvath Pan Tissue clock" (Mozhui et al., 2022).In these mice, various biological mechanisms are thought to cause premature aging; ERCC1 (Weeda et al., 1997) and XPG (Barnhoorn et al., 2014) mice exhibit a deficit in nucleotide excision repair (NER) of the nuclear DNA, POLG mice show accumulation of mitochondrial DNA mutations (Kujoth et al., 2005;Trifunovic et al., 2004) and LMNA knock-in (LAKI) mice suffer from nuclear lamina defects (Osorio et al., 2011;Varga et al., 2006).Here, we assessed DNA methylation age (DNAm) in Ercc1 −/Δ , Xpg −/− , Laki TG/TG , and Polg TG/TG mice at different ages, including post-natal development, and middle and old age.Both proliferative (blood and skin) and more terminally differentiated tissues (liver, cerebral cortex and skeletal muscle) were analyzed (Figure 1a).
During the generation of experimental mice, we noticed that while LAKI TG/TG and Polg TG/TG mice were born at a predicted Mendelian frequency, Ercc1 −/Δ and Xpg −/− showed a perinatal lethality (Figure S1a).Furthermore, all premature aging animals exhibited reduced body weight compared to their control littermates as expected (Figure 1b).As a quality check, we first looked at the clock performance in the control WT mice in multiple tissues.The chronological age prediction was highly accurate in blood in C57BL6J and C57BL6J-FVB backgrounds (r = 0.99 and r = 0.95, respectively) and provided sufficient accuracy in the other tissues (r = 0.89 to 0.98) (Figure S1b and Table S1), confirming the precision of DNAm clocks.
Next, we determined DNAm age in Ercc1 −/Δ , LAKI TG/TG , and Xpg −/− at 8 weeks, and Polg TG/TG at 30 weeks of age corresponding to the relative middle age of the strain.Importantly, the biological age of Ercc1 −/Δ mice was mainly increased in blood but also significantly increased in brain, liver, skeletal muscle, and skin according to the pan-tissue or tissue-specific clocks (Figure 1c and Table S4 respectively).Additionally, Xpg −/− mice showed increased age in blood and brain (Figure 1c).Conversely, we did not detect systemic DNAm age acceleration in LAKI or Polg mice in any of the tissues analyzed (Figure 1c).Additionally, we performed the "Tissue specific clock" analysis in mouse samples and we found similar results (Table S4).
Subsequently, we analyzed the methylation age at different ages including, Ercc1 −/Δ (2, 8 and 20 weeks), LAKI TG/TG (8 and 23 weeks), and Polg TG/TG (30 and 47 weeks).Interestingly, in the Ercc1 −/Δ mice, biological age was increased mildly at 2 weeks old in blood, and significantly accelerated in liver, and skin at 20 weeks (Figure 1d and Table S2).Conversely, DNAm age was not changed in aged LAKI TG/TG   or Polg TG/TG mice (Figure S1c).Together, our results further confirm that biological age is increased only in Ercc1 mice, at multiple ages.
Next, to determine if the rate of accelerated aging in Ercc1 mice was constant or increasing with age, we calculated the slope of biological vs. chronological age in both WT and KO mice.Importantly, the rate was significantly increased in blood, skeletal muscle, and brain (Figure 1e), demonstrating that the delta between biological and chronological age increased with age in Ercc1 −/Δ mice.All together, these results suggest DNA repair deficient mice as perhaps as one of the most promising mouse models of premature aging.

| IN CRE A S ED DNA m AG E IN H UMAN CELL S FROM PROG EROID SYNDROME S
At last, to determine whether these findings also apply to humans, we analyzed the DNAm age of human fibroblasts obtained from patients affected by diseases caused by mutations in DNA excision repair genes associated with aging phenotypes: Xeroderma Pigmentosum (XP) (Rizza et al., 2021), and Cockayne Syndrome type A (CSA) and B (CSB) (Laugel, 2000) (Table S3).DNAm age was significantly higher in the affected patients (Figure 2a), and the difference between DNAm age and chronological age was significantly increased (Figure 2b).These results indicate that human progeroid syndromes associated with mutations in DNA excision repair genes also display accelerated epigenetic age.
Although premature aging models have been widely used, their physiological relevance from the perspective of aging clocks has not been deeply investigated until now.Here, we observed accelerated epigenetic aging in Ercc1/5-deficient mice, and in related human diseases.Depletion of these proteins results in a defect in DNA repair, leading to an accumulation of DNA mutations.Our results indicate that defective DNA repair, leading to unrepaired persistent DNA damage, results in accelerated epigenetic aging, strongly suggesting a link between DNA damage and epigenetic dysregulation.In this line, DNA damage has been proposed as one of the central hallmarks of aging, as well as a potential driver of the aging process (Schumacher et al., 2021).We noted that even though DNAm age was increased in Ercc1 mice already at 2 weeks, greater changes were observed in older animals indicating a progressive age acceleration during aging.In this line, we propose that a higher DNA repair capacity during development (Mitchell & Hartman, 1990)    have a preventing impact on aging, dramatically extends lifespan of DNA repair-deficient mice (Vermeij et al., 2016) and Ercc1 depletion in blood specifically causes premature aging (Yousefzadeh et al., 2021).
Importantly, the methylation clock was more accurate in blood, a rapidly proliferative tissue that undergoes constant regeneration, in which the most significant and strongest differences were observed in DNA repair deficient mice.Therefore, due to its simple

| IN SUMMARY
Our survey of mouse models of premature aging may be expanded to alternative models (Koks et al., 2016), additional tissues, and ages.
Moreover, since one limitation of our study might be the use of the , might prevent potential epigenetic dysregulation as consequence of DNA damage, and could protect these mutant mice during gestation.Interestingly, dietary restriction, shown to F I G U R E 1 (a) Schematic representation of premature mouse strains, tissues collected, and timepoints taken.(b) Evolution of body weight until the euthanize point.(c) Methylation biological age of Ercc1 −/Δ , Xpg −/− , LAKI TG/TG , and Polg TG/TG mice.(d) Methylation biological age of Ercc1 −/Δ mice at 2, 8, and 20 weeks in multiple organs/tissues.Data are represented as box plots (center line shows median, box shows 25th and 75th percentiles and whiskers show minimum and maximum values), statistical significance was assessed by two-sided unpaired t-test.(e) Slope of aging in Ercc1 −/Δ and controls mice from 2 to 20 weeks old.Significance of the interaction term in the linear regression was analyzed.

Methylation biological age in Ercc1 mice at different ages
(a) DNAm age versus chronological age and (b) difference between biological and chronological age in control versus Cockayne Syndrome (CS) and Xeroderma Pigmentosum fibroblasts isolated from human samples.Data are represented as box plots (center line shows median, box shows 25th and 75th percentiles and whiskers show minimum and maximum values), statistical significance was assessed by two-sided unpaired t-test.
(Lehallier et al., 2020;Meyer & Schumacher, 2021;Rechsteiner et al., 2022)ological age using other epigenetic clocks, as well as other clocks built on different types of omics data, such as transcriptomic, proteomic or chromatin accessibility clocks(Lehallier et al., 2020;Meyer & Schumacher, 2021;Rechsteiner et al., 2022).AUTH O R CONTR I B UTI O N S K.P. and S.P. performed data and statistical analysis.A.P. generated mouse strains and collected tissues.C.M and L.S. were involved in culture and DNA extraction from human cells.C.R. extracted DNA from mice.S.H., A.H. analyzed data and made a critically revision.A.O. directed and supervised the study.K.P., A.P, and S.P. wrote the manuscript with input from all authors.F I G U R E 2