Loss of H3K9 trimethylation leads to premature aging

Summary paragraph

Aging is the major risk factor for most human diseases and represents a major socioeconomical challenge for modern societies. Despite its importance, the process of aging remains poorly understood. Epigenetic dysregulation has been proposed as a key driver of the aging process. Alterations in transcriptional networks and chromatin structure might be central to age-related functional decline. A prevalent feature described during aging is the overall reduction in heterochromatin, specifically marked by the loss of the repressive histone modification, histone 3 lysine 9 trimethylation (H3K9me3). However, the role of H3K9me3 in aging, especially in mammals, remains unclear. Here we show using a novel mouse strain, “TKOc”, carrying a triple knockout of three methyltransferases responsible for H3K9me3 deposition, that the inducible loss of H3K9me3 in adulthood results in premature aging. TKOc mice exhibit reduced lifespan, lower body weight, increased frailty index, multi-organ degeneration, transcriptional changes with significant upregulation of transposable elements, and accelerated epigenetic age. Our data strongly supports the concept that the loss of epigenetic information might directly drives the aging process. These findings reveal the importance of epigenetic regulation in aging and suggest that interventions targeting epigenetic modifications could potentially slow down or reverse age-related decline. Understanding the molecular mechanisms underlying the process of aging will be crucial for developing novel therapeutic strategies that can delay the onset of age-associated diseases and preserve human health at old age specially in rapidly aging societies.

In the last few years, the field of epigenetics has gained great importance, holding significant implications for various aspects of human health. During aging, epigenetic dysregulation is observed leading to changes in gene expression¹. In this context, aging is associated with a global loss and a local increase in DNA methylation. Consequently, novel epigenetic clocks based on these age-associated changes in DNA methylation have been developed by multiple groups^2,3. These clocks provide a valuable tool for understanding the complex interplay between epigenetic modifications and the aging process. In addition to changes in DNA methylation, alterations in the tri-methylation of H3 lysine 9 (H3K9me3) associated with repressive heterochromatin^4–6 occur during aging in model organisms⁷, in human samples of individuals at an advanced age, and in patients suffering from premature aging syndromes including Hutchinson–Gilford progeria syndrome and Werner syndrome^8–10. In addition, the levels of the H3K9me3 methyltransferase Suv39h1 and the heterochromatin protein 1 (HP1) decrease during normal aging due to alterations in other hallmarks of aging such as DNA damage, telomere shortening, and mitochondrial dysfunction¹¹.

Importantly, the decrease in heterochromatin results in increased transcriptional activity in non-coding regions of the genome, including repetitive regions containing transposable elements (TEs), which are generally repressed by H3K9me3^12–16. TEs can be broadly divided into DNA transposons or retrotransposons, which make cDNA copies through reverse transcription. Retrotransposons are further classified as long terminal repeats (LTR)-containing endogenous retroviruses (ERVs) or non-LTR retrotransposons such as long interspersed nuclear elements (LINEs) or short interspersed nuclear elements (SINEs). With age, transcriptional activation of retrotransposons can have detrimental consequences such as activation of innate immunity, DNA damage, and various diseases such as cancer and autoimmune diseases^16,17. Based on these evidences, the global loss of H3K9me3 observed in multiple species during aging has led to the “Heterochromatin loss theory of aging”^18,19. Although these observations suggest a central role of the age-associated loss of heterochromatin and H3K9me3 as drivers of the aging process, this role has not been demonstrated yet in mammals, where current data only shows a correlation between loss of H3K9me3 and aging.

To investigate the potential role of epigenetic dysregulation in the aging process, we aimed to generate an experimental mouse model featuring the inducible loss of H3K9me3 during adulthood. To achieve this goal, we selected a genetic approach involving the inducible knockout of the three methyltransferases, Suv39h1, Suv39h2, and Setdb1, known for establishing the H3K9me3 mark. Building on prior evidence highlighting the essential role of H3K9me3 during development, we induced H3K9me3 loss in adult mice^20–23. Importantly, loss of H3K9me3 resulted in reduced lifespan and was associated with multiple age-associated phenotypic alterations, indicating that the loss of the epigenetic mark H3K9me3 might contribute to the aging process.

Results

Novel mouse model for the inducible loss of H3K9me3 at adult stage

The generation of a triple knock out strain (TKO) for the three H3K9me3 methyltransferases Suv39h1, Suv39h2 and, Setdb1 has been previously reported for investigating the role of H3K9me3 during embryologic development²³. In this study, the disruption of these three methyltransferases during development demonstrated the crucial role of H3K9me3 during the initiation of organogenesis, as well as in the preservation of lineage fidelity²³.

In order to investigate the role of H3K9me3 during aging and bypass the developmental problems associated with H3K9me3 loss^24,21–23, we designed a transgenic strategy to conditionally induce H3K9me3 depletion at adult stage by using a tamoxifen-inducible Cre-mediated recombination system. To do so, TKO mice²³ were crossed with the constitutive CAG-CreER mouse²⁵. This new strain, TKOCAGCre, allows the whole-body Cre-mediated recombination of Setdb1 and Suv39h1 by tamoxifen administration while Suv39h2 is constitutively knocked-out (Fig. 1a and Extended Data Fig. 1a).

Fig. 1: Inducible loss of H3K9me3 in-vitro.

a, Schematic representation of the genetic strategy to generate the quadruple transgenic TKOCAGCre mouse strain carrying the insertion of CAG-CreER (Chr.3) for the tamoxifen-inducible CRE-mediated recombination system (Loxp sites), Setdb1 gene insertion of LoxP sites in the intron 14 and 16 (Chr.3), Suv39h1 gene insertion of LoxP sites in the intron 2 and 5 (Chr.X), Suv39h2 gene deletion at Exon2 (Chr. 10). b, Setdb1 and Suv39h1 mRNA levels in CTRL and TKOc tail-tip fibroblasts after 6 days of 4-OH Tamoxifen treatment. (n = 1 to 2, with 3 technical replicate). c, Immunofluorescence of H3K9me3 in CTRL and TKOc tail-tip fibroblasts upon tamoxifen treatment for 6 days. Scale bar, 50 μm. d, Quantification of H3K9me3 positive cells in CTRL and TKOc tail-tip fibroblasts upon tamoxifen treatment. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 according to one-way ANOVA data are median with interquartile range. Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test. *p < 0.05; **p < 0.01; ***p < 0.001.

First, to validate the genetic strategy, tail tip fibroblast (TTFs) from TKO CAG-Cre mice both positive (TKOc) and negative (CTRL) for CAG-Cre were isolated and cultured. After 2 days in culture, TTFs were treated with 4-hydroxy tamoxifen (4-OHT) for 6 days and mRNA levels of methyltransferases were analyzed. As expected, the results showed a downregulation of Setdb1 and Suv39h1 (Fig. 1b). In addition, a decrease in the protein levels of H3K9me3 was detected by immunofluorescence and Western blot upon 4-OHT treatment in TKOc TTFs compared to CTRLs (Fig. 1c, d and Extended Data Fig. 1b).

To further investigate the cellular effects of H3K9me3 loss, we assessed multiple aging-associated markers, including DNA damage, mitochondrial generation of reactive oxygen species (ROS), and nuclear envelope integrity. Importantly, loss of H3K9me3 in TKOc cells led to a significant elevation in γ-H2AX foci intensity, a well-established marker of DNA double-strand breaks associated with aging²⁶ (Fig. 2a). Mitochondrial ROS levels were also significantly increased in TKOc cells, contributing to oxidative stress²⁷ (Fig. 2b). Furthermore, the loss of H3K9me3 disrupted nuclear envelope integrity, resulting in a higher percentage of cells with nuclear abnormalities (Fig. 2c), consistent with reports linking such defects to loss of heterochromatin and aging^26,28. Additionally, depletion of H3K9me3 in TKOc TTFs upon 4-OHT treatment resulted in a significant reduction in cell proliferation and cell cycle arrest (Fig. 2d)

Fig. 2: Loss of H3K9me3 in-vitro induces cellular aging.

a-b, Immunofluorescence and quantification of γH2AX (a) and ROS (b) in CTRL and TKOc tail-tip fibroblasts upon tamoxifen treatment for 6 days. Scale bar, 50 μm. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 according to one-way ANOVA data are median with interquartile range. c, Immunofluorescence of Lamin A/C and quantification of nuclear abnormality in CTRL and TKOc tail-tip fibroblasts upon tamoxifen treatment for 6 days. Scale bar, 50 μm. *p < 0.05; **p < 0.01; ***p < 0.001 according to one-way ANOVA data are mean ± SEM d, Cumulative population doubling curves of CTRL and TKOc tail-tip fibroblasts with or without TAM treatment. Population doublings were calculated by the formula log [(number of cells harvested)/(number of cells seeded)]/ log2. e, Relative mRNA levels of genes related to senescence markers and senescence-associated secretory phenotype in CTRL and TKOc tail-tip fibroblasts after TAM treatment. (n = 1 to 2, with 3 technical replicate). Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test. *p < 0.05; **p < 0.01; ***p < 0.001. f, Senescence-associated β-galactosidase (SA-β-gal) staining in CTRL and TKOc tail-tip fibroblasts with TAM treatment. Quantification of SA-β-gal positive cells for CTRL and TKOc after treatment. Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test *p < 0.05; **p < 0.01; ***p < 0.001 of 3 or 5 replicates.

Next, to determine whether cell cycle arrest was due to senescence, we measured the expression of markers of cellular senescence and senescence-associated secretory phenotype (SASP) in TKOc cells. Interestingly, we observed an increase in cellular senescence in TKOc treated cells with the upregulation of Cdkn1a (cyclin dependent kinase inhibitor 1A)²⁹ and Stat1 (signal transducer and activator of transcription) mRNA³⁰. In addition, the expression of genes associated with SASP, including Il-6 (interleukin-6) and Mcp-1 (monocyte chemoattractant protein-1), were upregulated in TKOc cells due to the loss of H3K9me3 (Fig. 2e). In addition, a significant increase in SA-beta-galactosidase activity was detected upon 4-OHT treatment of TKOc TTFs compared to CTRLs (Fig. 2f). Together, these data indicate that our genetic strategy allows the inducible depletion of H3K9me3, which results in the manifestation of multiple molecular and cellular hallmarks of aging, including DNA damage, increased ROS, and nuclear envelope defects, ultimately leading to cell cycle arrest and cellular senescence.

Loss of H3K9me3 leads to premature aging

To investigate whether H3K9me3 reduction in adult mice could drive premature aging, we designed an induction protocol were 3-month-old mice were treated with 5 consecutive daily intraperitoneal injections of Tamoxifen (TAM) (Extended Data Fig. 2a). First, we validated the system in vivo by confirming the recombination of Setdb1 and Suv39h1 in DNA isolated from peripheral blood DNA of TKOc and control three months after treatment with TAM for 5 days (Extended Data Fig. 2b). Importantly, low levels of recombination were detected in blood upon a single treatment with tamoxifen. Nevertheless, TKOc mice started to exhibit a moderate change in their physical appearance at 6 months of age, prompting us to perform health status and behavioral assessments. We used a composite frailty index (FI) score to assess health measures including body and coat condition, kyphosis, cataract, and tail stiffening³¹. TKOc mice presented higher FI scores than CTRL mice (Extended Data Fig. 2c). In addition, we monitored changes in body weight and did not detect remarkable changes between treated TKOc and their CTRL groups over a period of 42 weeks (Extended Data Fig. 2d). On the other hand, analysis of hematological parameters by complete blood count (CBC) showed a significant decrease in red blood cell (RBC) and hemoglobin (HGB) levels, indicating signs of anemia in TKOc treated mice (Extended Data Fig. 2e). Moreover, analysis of activity by open field test indicated that TKOc mice exhibited hypoactivity and slower movements together with an increase in peripheral distance travelled compared to the CTRL group (Extended Data Fig. 2f). Contrary, grip strength analysis used to measure forelimb neuromuscular function showed no significant differences between the treated TKOc and CTRLs (Extended Data Fig. 2g). Finally, we did not detect significant differences in the lifespan of TKOc and CTRL mice (Extended Data Fig. 2h). Overall, these results show that our 5-day TAM treatment protocol induces a mild premature aging phenotype, maybe due to insufficient recombination efficiency. For this reason, we decided to perform an additional series of 5-day intraperitoneal TAM injections at 5.5-months of age (Fig. 3a).

Fig. 3: Loss of H3K9me3 in vivo leads to premature aging and decrease in lifespan.

a, Experimental design. b, Image of CTRL and TKOc mice (6-month-old mice). c, Frailty index of CTRL and TKOc mice (n = 13 to 21, 6-month-old mice). d-f, Behavioral characterization of CTRL and TKOc mice (6 months). d, Open field exploration (15 minutes), quantification of distance travelled, mean speed, time spent and distance traveled in the center with a representative tracking trace (n = 12 to 14–17, 6-month-old mice). e, Elevated plus maze test, quantification of the time spent on the open arms f, Grip strength measured as average grip strength of three different trials CTRL and TKOc mice. (n = 11 to 17, 6-month-old mice). g, Changes in body weight of CTRL and TKOc mice upon administration of tamoxifen. Arrows show the TAM injection timepoint h, Kaplan-Meier survival curves for CTRL and TKOc mice (n = 9 to 22 mice, including males and females). Dots in all panels represent individual sample data. Survival curve data were analyzed by log-rank (Mantel-Cox test). Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test. *p < 0.05; **p < 0.01; ***p < 0.001.

After the second round of TAM injections, we first assessed the percentage of recombination across various tissues. Our analyses revealed a high recombination rate in proliferative tissues, including the skin, small intestine, and spleen (Extended Data Fig. 3a). Additionally, we observed substantial recombination in skeletal muscle and brain, with the lowest recombination rate in the liver (Extended Data Fig. 3a). Subsequently, we examined H3K9me3 protein levels in highly proliferative tissues of TKOc mice, specifically the skin and small intestine. Our findings indicated a significant reduction in H3K9me3 levels in TKOc mice compared to CTRLs (Extended Data Fig. 3b, c). Importantly, after the second round of TAM injections, TKOc mice displayed a severe aged appearance (Fig. 3b). This premature aging phenotype was confirmed by a significant increase in FI scores in TKOc mice compared CTRL littermates (Fig. 3c). Next, we evaluated activity by open field test and observed that TKOc mice had decreased activity and were slower than the CTRL mice (Fig. 3d). Additionally, we noted that the TKOc mice spent substantially less time exploring the center of the arena, and traveled less, as compared to CTRL littermates (Fig. 3d). Building on this observation, we further assessed anxiety-like behavior using the elevated plus maze paradigm, a well-established test for evaluating anxiety in rodents. In this test, mice typically avoid the open arms and prefer to spend time in the closed arms. Our analysis revealed that TKOc mice spent significantly less time in the open arms compared to CTRL mice, indicating increased anxiety levels in the TKOc group (Fig. 3e). Moreover, we found that treated TKOc mice had reduced grip strength compared to controls (Fig. 3f). This was combined with the appearance of frailty, driven by an increased incidence of kyphosis, cataracts, alopecia, tail stiffening, and a decline in body condition, whisker density, and coat quality. In addition to these markers of frailty, a significant reduction in body weight was observed in the treated TKOc mice, which was not recovered over time (Fig. 3g). Most importantly, we observed a significant reduction in lifespan, with a media lifespan of only 30 weeks (Fig. 3h). Altogether, our data demonstrates that loss of H3K9me3 in TKOc mice leads to premature aging.

Loss of H3K9me3 is associated with the age-associated degeneration of multiple tissues and organs

Given the premature aging phenotype observed in TKOc mice, we decided to conduct an in-depth characterization of the model at 6–8 months of age, a time point corresponding to the relative median survival of the TKOc strain. For this reason, we examined age-related features across various tissues, including the hematopoietic system, small intestine, skin, kidney, muscle, spleen, bone, and liver^32–37. Interestingly, analysis of blood parameters revealed significant alterations, including a marked reduction in red blood cells (RBC), white blood cells (WBC), and hemoglobin (HGB) levels, indicative of anemia compared to CTRL mice (Fig. 4a and Extended Data Fig. 4a). TKOc mice exhibited a tendency toward premature inflammaging, characterized by elevated circulating levels of cytokines associated with inflammaging, including CCL11, CXCL1, IFN-γ, and IL-18. (Extended Data Fig. 4b). In the intestine, TKOc mice displayed notable structural changes compared to age-matched CTRLs, including reduced villus number and length, crypt deepening, and ballooning (Fig. 4b). Histological examination of TKOc skin demonstrated pronounced subepidermal thinning and a significant loss of hypodermal fat, consistent with aging (Fig. 4c). In the kidney, histological analyses revealed a decreased glomerular capillary volume-to-Bowman capsule volume ratio, indicative of glomerulosclerosis, alongside evidence of fibrosis (Fig. 4d and Extended Data Fig. 4b). Similarly, spleen histology not only showed an increase in white pulp area, but also revealed significant fibrosis, reflecting tissue remodeling associated with aging (Fig. 4f and Extended Data Fig. 4c). Fibrosis was also observed in the liver, further emphasizing the progressive deterioration of these organs in TKOc mice (Extended Data Fig. 4b). In addition, skeletal muscle in TKOc mice showed a reduction in fiber diameter, a hallmark of muscle atrophy with aging (Fig. 4e). Lastly, TKOc mice exhibited a significant reduction in bone thickness, mirroring physiological aging processes (Fig. 4g). Bone microarchitecture analysis confirmed substantial bone loss, including decreased bone volume fraction (BV/TV), cortical thickness, and cortical area, along with an increased bone marrow area. Collectively, these findings highlight the impact of age-associated epigenetic alterations, such as the loss of H3K9me3, in driving aging-associated changes and pathological remodeling across multiple tissues and organs.

Fig. 4: Age-associated organ degeneration results from H3K9me3 loss.

a, Hematological parameters in CTRL and TKOc mice (n = 11–12 CTRL and 16–17 TKOc, 6-month-old mice). b, Representative small intestine sections stained with hematoxylin and eosin (H&E) (right) and quantification of crypt depth, number and height of the villus (left) in CTRL and TKOc mice. c, Representative skin sections stained with H&E (right) and quantification of hypodermal fat and epidermal thickness (left) in CTRL and TKOc mice. Including EP, Epidermis. DE, Dermis. HP, Hypodermis. ML, Muscle. d Representative kidney sections stained with H&E (right) and quantification of glomerular capillary volume (GV) to Bowman capsule volume (BV) (left) in CTRL and TKOc mice. e, Representative quadriceps muscle sections stained with H&E (right) and quantification of myofiber cross-sectional area (left) in CTRL and TKOc mice. f, Representative spleen sections stained with H&E (right) and quantification of the white pulp area (left) in CTRL and TKOc mice. For the small intestine, skin, kidney, muscle and spleen at least 8 measurements were performed per animal. The graph shows mean values for n = 3 mice at 7–8 months of age. Scale bar, 200, 100 or 50μm. g, Representative micro-CT images (right) and quantitation of femur cortical bone (bottom; scale bar, 300μm) in CTRL and TKOc mice (n = 8 to12, 7–8-month-old mice). Including BV/TV, Bone Volume over Total Volume. Th, Thickness. Ar, Area. Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test. *p < 0.05; **p < 0.01; ***p < 0.001.

Loss of H3K9me3 leads to age-associated epigenetic and transcriptional changes

To further confirm that the loss of H3K9me3 loss results in premature aging at the molecular level, we used tissue-specific epigenetic clocks to assess DNA methylation age (DNAm) in TKOc and CTRL mice. Specifically, we analyzed both proliferative (skin, spleen, and small intestine) and post-mitotic tissues (skeletal muscle, liver, and brain) to account for potential differences in the levels of H3K9me3 recombination. Interestingly, accelerated DNAm age was detected in proliferative tissues of TKOc mice including the skin and small intestine as well as the spleen (Fig. 5a). In addition, a tendency for accelerated aging was also observed in muscle, while no significant differences were detected in the liver and brain (Extended Data Fig. 5a). Moreover, the mean DNA methylation levels of TKOc mice was reduced in most proliferative tissues, most prominently in the small intestine and spleen, in line with the role of H3K9me3 in recruiting the DNA methylation maintenance machinery to the DNA replication fork during cell division. Loss of H3K9me3 therefore correlates with global DNA methylation loss and might contribute to the deregulation of related transcripts and de-repression of heterochromatin. (Extended Data Fig. 5b) Overall, our results indicate that the TKOc mouse model exhibits an accelerated epigenetic age and global loss of DNA methylation compared to age-matched controls.

Fig. 5: Accelerated epigenetic age and transcriptional dysregulation in TKOc mice.

a, Epigenetic age of skin, spleen and small intestine of mice TKOc mice compare their CTRL (n = 6 to 5–6, 7–8-month-old mice). Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test *p < 0.05; **p < 0.01; ***p < 0.001. b, Bubble plots of the top significant (up and down) GO terms in spleen, small intestine, kidney, and liver. c, Volcano plots showing the TE transcripts that are significantly upregulated (in orange), downregulated (in blue) or unchanged (in gray) in skin, spleen, small intestine, kidney, and liver. Transcripts were determined to be significant if the adjusted p-value was < 0.05 and log₂fold change was > 1. A few top upregulated transcripts are labeled in each tissue.

To perform a comprehensive analysis of transcriptomic changes resulting from the loss of H3K9me3, we generated total RNA-seq libraries from 7 different tissues from TKOc and CTRL mice and livers of 3-month-old young and 18-month-old aged C57BL6/JN mice (n=6 per group, equally distributed males and females, total n=96 libraries). The 7 different tissues were derived from both proliferative (spleen, small intestine, skin) and non-proliferative (liver, muscle, brain, and kidney) tissues. Subsequently, we processed the raw FASTQ reads to enable both unique and multiple alignment with the mouse reference genome, which allowed us to generate counts from both genes and TEs.

Principal Component Analysis (PCA) of the transcriptomic data from TKOc and CTRL mice showed that the primary source of variation was the tissue type (Extended Data Fig. 5c). We then focused on the liver and performed a separate PCA analysis with TKOc, CTRL, young, and aged mice. Interestingly, the samples segregated by age on PC1 and by sex on PC2 with the TKOc samples showing a pro-aging transcriptome (Extended Data Fig. 5d, note shift to left similar to aged samples). Next, we used conventional DESeq2 analysis with uniquely mapped reads to identify differentially abundant mRNAs between TKOc and CTRL mice. For each tissue type, we then performed a Gene Ontology (GO) analysis³⁸ to derive insight into functional pathways activated by H3K9me3 depletion (Fig. 5b). Interestingly, downregulated mRNAs were related to tissues-specific function including glucuronidation in the small intestine, metabolic processes in the liver, and ion transport in the kidney. Notably, in our previous study of the aging liver from naturally aged mice, we observed similar GO terms up or downregulated³⁹. In contrast, several biological processes showed significant enrichment across tissues. These included, multicellular organism development, cell differentiation, response to virus and, immune/inflammatory response. Given this observation, we investigated the enrichment of senescence-associated and SASP-related genes using SenMayo⁴⁰ signatures within our RNA-seq cohort. Remarkably, these gene sets were enriched in the small intestine, kidney, liver and marginally spleen tissues derived from TKOc mice, further supporting the induction of senescence pathways upon loss of H3K9me3 (Extended Data Fig. 5e).

Lastly, we used DESeq2⁴¹ to identify differentially abundant TE transcripts between TKOc and CTRL tissues. (Fig. 5c and Extended Data Fig. 5e, note n). More importantly, in all tissues (except liver) derived from TKOc mice, there were more TEs that were significantly upregulated than downregulated (Fig. 5c). We were curious to note that the top identified TEs in all tissues were ERVs (Fig. 5c and Extended Data Fig. 5f, note labels). This observation prompted us to categorize all upregulated TEs to identify whether there was a preference for any specific family of TEs that are upregulated. Importantly, we found that among the top 50 most significantly upregulated TEs, ERVs were indeed the dominant family followed by LINEs showing the largest fold changes (Supplementary Table 1 for details). This pattern of robust upregulation of TEs and the preference for ERVs was replicated in the livers of young and old C57BL/6JN mice (Extended Data Fig. 5e last panel and Supplementary Table 1 for details), albeit to a lower extent. Overall, our comprehensive analyses of the transcriptome across both coding and non-coding regions of the genome highlight that H3K9me3 deregulation activates canonical age-related transcriptional pathways including upregulation of TEs. Thus, demonstrating that TKOc mice show features of premature aging at the molecular and cellular level.

Discussion

The molecular mechanisms and causal relationships underlying age-related transcriptional changes and the loss of heterochromatin during aging has largely remained elusive. According to the “‘Heterochromatin loss” theory of aging, the degradation of constitutive heterochromatin at the nuclear periphery, marked by H3K9me3, is believed to play a pivotal role during the aging processes driving pro-aging cellular phenotypes^42,39. Previous investigations have documented the age-associated decline of H3K9me3 in multiple models including Drosophila, C. elegans, as well as human cells^8,11,43,44. Consequently, the age-related reduction in H3K9me3 has been postulated as a key contributing factor to the aging process.

H3K9me3 constitutes a post-translational modification recognized for its involvement in regulating diverse biological processes, particularly in the establishment of transcriptionally silent heterochromatin. In this line, the proposed models align with the idea that H3K9me3 loss might aggravate the aging process through the dysregulation of chromatin organization. Through the simultaneous depletion of Setdb1 and Suv39h1/2, methyltransferases are crucial to the formation of constitutive heterochromatin, our mouse model analyzes consequential changes in transcription changes including a potential source of genomic instability by the activation of endogenous mobile genetic elements, specifically transposable elements^15,16.

In this study, we aim to demonstrate the causative role of H3K9me3 as driver of aging. Towards this goal, we circumvent the detrimental consequences of the constitutive loss of H3K9me3 during the embryonic development by generating a TAM inducible model that enables the depletion of H3K9me3 in adulthood²³. Our findings demonstrate that the induced loss of H3K9me3 during adulthood leads to the manifestation of premature aging phenotypes, characterized by an increased frailty, accelerated aging across diverse tissues (particularly those proliferative), and a reduction in lifespan. At the molecular level, loss of H3K9me3 results in de-repression of TEs, upregulation of lineage-inappropriate developmental genes, and downregulation of mRNAs specifying tissue function. Through the direct targeting of H3K9me3, we provide substantiating evidence that loss of epigenetic information might drive mammalian aging. Importantly, in alignment with the interconnectivity between different hallmarks of aging, DNA damage globally impacts chromatin structure. In this line, a recent study demonstrated that the repair of DNA breaks can lead to epigenetic erosion and a global reduction in chromatin compaction, ultimately resulting in premature aging⁴⁵. Similarly, we have recently shown that DNA-repair deficient premature aging models display accelerated epigenetic age⁴⁶.

While the decline in H3K9me3 is evident across various species during aging, this pattern is also contingent on the specific tissue and cell type^47–49. Consistent with this observation, we observed a more pronounced increase in biological age, particularly in proliferative tissues: spleen, skin, and small intestine, which might undergo a faster depletion of H3K9me3 by the lack of methyltransferases that can restore this mark after cell division. However, irrespective of proliferative potential, we found that loss of H3K9me3 increases the expression of TEs, particularly ERVs and less so LINEs, in all 7 tissues examined from TKOc mice. Interestingly, ERVs have been recently shown to be upregulated in aged murine, primate and human cells and organs, as well as serum from older individuals⁵⁰. Moreover, ERVs have been recently associated with cellular senescence and tissue aging⁵⁰. Similarly, several reports suggest that deregulation of LINE-1 elements drive aging in mice^17,51. These results were also recapitulated when analyzing RNA-seq data from aged mice livers. Importantly, ERVs and LINE-1 both result in activation of the innate immune system with the release of senescence-associated secretory phenotype (SASP) factors. Indeed, our conventional analysis of differential mRNAs from annotated genes show a strong immune activation signature. Conversely, downregulated mRNAs are mostly related to the tissue type indicating a loss of tissue function. Lastly, we hypothesize that interventions known to slow the process of aging process leading to healthy lifespan might mitigate age-associated epigenetic dysregulation. In this line, restoration of H3K9me3 levels is an early event observed during the rejuvenation of age-associated phenotypes by cellular reprogramming, and preventing H3K9me3 restoration using an H3K9 methyltransferase inhibitor is sufficient to block the amelioration of additional age-associated phenotypes⁴⁷. Similarly, therapeutic interventions that promote healthy longevity in mice have been shown to reduce the expression of repetitive elements⁵².

In conclusion, our data strongly support the notion that changes in the epigenome, in this case the age-associated loss of H3K9me3, might be sufficient to drive the aging processes in mammals, and therefore reinforce the role of epigenetic dysregulation as an important hallmark of aging. The TKOc mice generated in this study could potentially serve as a valuable experimental model for in-depth exploration of the molecular mechanisms of aging. Moreover, although this study has focused mainly on the study of H3K9me3, as a future direction, it would be very interesting to evaluate the role of additional histone marks, well-known to change during aging. Lastly, the loss of H3K9me3 might represent a novel target for the potential development of therapeutic interventions aiming at the prevention of age-associated diseases and extension of healthy lifespan.

Methods

Animal procedures

Animal housing

All the experimental procedures were performed in accordance with Swiss legislation after the approval from the local authorities (Cantonal veterinary office, Canton de Vaud, Switzerland). Animals were housed in groups of five mice per cage with a 12hr light/dark cycle between 06:00 and 18:00 in a temperature-controlled environment at 25°C and humidity between 40 and 70% (55% in average), with free access to water and food. Transgenic mouse models used in this project were generated by breeding and maintained at the Animal Facility of Epalinges and the Animal Facility of the Department of Biomedical Science of the University of Lausanne.

Mouse strains

Transgenic mice were used in hybrid C57BL/6J, B6C3F1 background. The TKOCAGCre mouse strains were generated by breeding the TKO strain, triple conditional knockout for the three H3K9me3 methyltransferases (Suv39h2, Suv39h1, Setdb1), previously generated by Professor Kenneth Zaret⁵³ with CAG-CreER^™ mice Stock No 004682. The final Homo-TKOCAGCre mouse strain is a quadruple transgenic conditional knockout mouse carrying: Setdb1 Flox/Flox: Loxp sites flanking exon 15–16 of Setdb1 gene (Chr.3). Suv39h1 Flox/Flox: Loxp sites flanking exon 3–5 of Suv39h1 gene (Chr.X). Suv39h2 KO/KO: deletion in the SUV39H2 gene (Chr. 10). CAG-CreER Cre/+: Insertion of CAG-CreER (Chr.3), for the tamoxifen-inducible CRE-mediated recombination system (Loxp sites). TKOCAGCre littermates not expressing Cre were used as control mice along the study.

Tamoxifen administration

Tamoxifen (Sigma, T5648) of TKOCAGCre mice was performed at 3 months of age and repeated at 5.5 months. Tamoxifen was administrated intraperitoneally at 67mg/kg for 5 consecutive days.

Mouse monitoring and euthanasia

All mice were monitored at least three times per week. Upon Tamoxifen injection, mice were monitored twice a week to evaluate their activity, posture, alertness, body weight and presence of tumors or wound. Mice were euthanized according to the criteria established in the scoresheet. We defined lack of movement and alertness, presence of visible tumors larger than 1cm³ or opened wounds and body weight loss of over 30% as imminent death points. Both genders were used for survival, body weight experiments, tissue and organ collection. Animals were sacrificed by CO₂ inhalation (6 min, flow rate 20% volume/min). Subsequently, before perfusing the mice with saline, blood was collected from the heart. Finally, multiple organs and tissues were collected in liquid nitrogen and used for DNA, RNA, and protein extraction, or placed in 4% formalin for histological analysis.

Behavior

Behavioral characterization was performed on both males and females, at the age of 6 and 12 months. Open field (OF:) Locomotor activity and anxiety-like behavior of adult mice were evaluated in an open field arena. Briefly, mice were individually placed in the center of a Plexiglas boxes (sides 45 cm, height 40 cm, Harvard Apparatus, 76–0439). Mice movements were recorded for 15 minutes. Recording was done with a USB camera (Stoelting Europe, 60516) and then analyzed using ANY-maze video tracking software (ANY-maze V7.11, Stoeling). Elevated plus maze (EP): To assess anxiety levels in adult mice, the experiment utilized a Plexiglas apparatus elevated 50 cm above the ground. The setup consisted of two closed arms (30 × 5 × 15 cm) and two open arms (30 × 5 × 0.5 cm), arranged in an alternating pattern around a central platform. Each mouse was placed on the central platform and allowed to explore the maze freely for 5 minutes. Movements were recorded throughout the test using a video tracking system (Stoelting Europe, 60516) and analyzed with ANY-maze software (version 7.11; Stoelting). The proportion of time spent in the open arms was calculated using the formula: (time in open arms / total time in open and closed arms). Grip strength test (GS): to measure muscular strength, a mouse was held by the tail and allowed to grip a mesh grip with the front paws (Harvard Apparatus, 76–1068). Three measurements minimum per trial were performed for each animal, with a few seconds resting period between measurements.

Frailty Index assessment

The Frailty Index (FI) was adapted from the previously described score³¹. For each mouse 28 health-related deficits were assessed going across the integument, physical/musculoskeletal, ocular/nasal, digestive/urogenital and respiratory systems were scored as 0, 0.5 and 1 based on the severity of the deficit. Total score across the items was divided by the number of items measured to give a frailty index score between 0 and 1.

System	Parameter
Integument	Alopecia
	Loss of fur color
	Dermatitis
	Loss of whiskers
	Coat condition
Physical/musculoskeletal	Tumors
	Distended abdomen
	Kyphosis
	Tail stiffening
	Gait disorders
	Tremor
	Body condition score
Vestibulocochlear/auditory	Vestibular disturbance
	Hearing loss
Ocular/nasal	Cataracts
	Corneal opacity
	Eye discharge/swelling
	Microphthalmia
	Vision loss
	Menace reflex
	Nasal discharge
Digestive/urogenital	Malocclusions
	Rectal prolapse
	Vaginal/uterine/penile prolapse
	Diarrhea
Respiratory	Breathing rate/depth
Discomfort	Mouse Grimace Scale
	Piloerection

Hematological analysis

Blood was collected from the temporal vein in potassium EDTA microtrainer tubes. Complete blood count was performed in Heska Element HT5 hematology analyzer.

Bone analysis

Bone microarchitecture was evaluate using a SkyScanner 1276 (Bruker, Belgium). 0.25 mm Al filter was used with a voltage of 200 kV and a current of 55 mA. To avoid drying samples were wrapped in paper towels soaked in PBS and scanned inside a drinking straw sealed on both ends. Voxel size for both applications was set at 10×10×10 μm3.

Bone microarchitecture was evaluated according to the ASBMR guidelines⁵⁴ using a custom CTan (Bruker, Belgium) script for automatic segmentation of trabecular bone in the distal femoral VOI, which was set 100 slices proximal to the distal growth plate and extended 200 slices towards the femoral diaphysis (slice thickness of 0.010 mm). The threshold used to binarize the calcified tissue was 40 on a 0–255 scale. Reconstruction of the scans was performed using NRecon (Bruker, Belgium) and further analysis were performed using CTan (Bruker, Belgium) with the minimum for CS to image conversion set at 0 and maximum set at 0.14. For the analysis of cortical parameters, the midpoint of the femur was determined, and the VOI was defined as the bone 50 slices (slice thickness of 0.010 mm) distal and proximal of the slice corresponding to the midpoint of the bone. All other parameters were kept the same as for the analysis of trabecular bone.

Recombination analysis by semiquantitative RT-PCR

DNA was amplified using DreamTaq Green PCR Master Mix 2X (Thermofisher, K1081) following the amplification protocol: 3 min at 95°C + 33 cycles (30 s at 95°C + 30 s at 56 or 60 °C + 1 min at 72°C) + 5 min at 72°C. PCR products were loaded and run in an agarose (1.6%) gel containing ethidium bromide (Carlroth, 2218.1). Images were scanned with a gel imaging system (Genetic, FastGene FAS-DIGI PRO, GP-07LED). Setdb1 and Suv39h1 recombination were detected using the following primers: Setdb1forward: 5’- CAGCTTGGAGGAATTGGTTC-3’ Setdb1 reverse 1: 5’- TTTCTTTGCCTTTGAGATGGA-3’ Setdb1 reverse 2: 5’- TACCATACCCACTAACACTTTGC-3’, Suv39h1 forward: 5’- GGAGCCCACTGAAAGTAGCA-3’, Suv39h1 reverse 1: 5’- ACTCCAGCCCCTCCTTTTT-3’ Suv39h1 reverse 2: 5’- GGTCAGGCTAGAAAACACAAGG-3’.

Cell culture

Mouse tail-tip fibroblasts (TTFs) were freshly extracted using Collagenase I (Sigma, C0130) and Dispase II (Sigma, D4693) and cultured in DMEM (Gibco, 11960085) containing non-essential amino acids (Gibco, 11140035), GlutaMax (Gibco, 35050061), Sodium Pyruvate (Gibco, 11360039) and 10% fetal bovine serum (FBS, Hyclone, SH30088.03) at 37°C in hypoxic conditions (3% O2). Subsequently, fibroblasts were passaged and cultured according to standard protocols. For activation of the Cre recombination TTFs were treated with 0.1 μM 4-Hydroxytamoxifen (4-OHT) for 6 days and subsequently cultured in medium was switched to one without 4-OHT.

Senescence-associated β-galactosidase assay

The senescence-associated beta-galactosidase (SA-βgal) assay was carried out following the method outlined by Debacq-Chainiaux et al. (2009). In summary, cells plated on glass coverslips underwent light fixation using a 3% paraformaldehyde and 0.2% glutaraldehyde solution in PBS buffer for 5 minutes. After removal of the fixation solution, the wells were washed multiple times and stained overnight at 37°C in a CO2-free incubator. The staining solution consisted of 40 mM citric acid/Na phosphate buffer, 5 mM K4[Fe(CN)6]·3H2O, 5 mM K3[Fe(CN)6], 150 mM sodium chloride, 2 mM magnesium chloride, and 1 mg/mL X-gal (Roth, 2315.1), with a pH range of 5.9–6.0. Subsequently, the coverslips were stained with DAPI and subjected to a standard immunofluorescence protocol. Bright-field microscopy was used to capture images, and the proportion of β-Gal-positive cells was quantified.

Reactive oxygen species assay

Mitochondrial reactive oxygen species (ROS) were assessed using the superoxide indicator dihydroethidium (DHE). To summarize, cells were first incubated in fresh serum-free media containing 5 μM DHE at 37°C under a humidified atmosphere with 5% CO2 for 30 minutes. After incubation, the wells were rinsed with PBS at room temperature, fixed with 4% paraformaldehyde for 15 minutes, and subsequently stained with DAPI. Images were then captured at 554 nm, following standard immunofluorescence imaging procedures.

DNA extraction and DNA methylation analysis

DNA extractions

Total DNA was extracted from tissues using Monarch Genomic DNA Purification Kit (New England Biolab, T3010L). Tissues were cut into small pieces to ensure rapid lysis. Total DNA concentrations were determined using the Qubit DNA BR Assay Kit (Thermofisher, Q10211).

DNA methylation clock

Analysis of epigenetic age was done in collaboration with the Clock Foundation. The mouse clock was developed in ref.⁵⁵. To analyze the epigenetic age, we used for skin, spleen, small intestine, skeletal muscle, liver and brain the following mouse clocks: “UniversalClock3Skin”, “UniversalClock2Blood”, “UniversalClock2Pan”, “DNAmAgeMuscleFinal”, “DNAmAgeLiverFinal”, “DNAmAgeCortexFinal”. The mouse methylation data were generated on the small and the extended version of HorvathMammalMethylChip⁵⁶. We used the SeSaMe normalization method⁵⁷. We used the noob normalization method implemented in the R function preprocessNoob.

Overall DNA methylation

Analysis was done by first removing probes with only background signal in a high proportion of samples. This was done by retaining only those probes that have a detection p-value of 0.05 or greater in more than 90% of the samples. Afterwards, we eliminated probes located on the X and Y chromosomes. Subsequently, for each sample, we calculated the median of the beta-values from the remaining probes to estimate the overall DNA methylation level.

Immunohistochemistry and immunofluorescence

Immunohistochemistry

Mice were euthanized with CO2 and multiple tissues and organs were collected, placed in 4% formalin (Sigma, 252549) overnight, and then immersed in 30% sucrose in phosphate buffered saline (PBS) for 72 h. Subsequently, samples were paraffin-embedded with a Leica ASP300S tissue processor (Leica, Heerbrug, Switzerland), sections prepared with a Microm HM 335 E microtome (Thermo Scientific,Walldorf, Germany) and mounted on Superfrost Plus slides (Thermo Scientific). Next, slides were deparaffinized and rehydrated with xylol and alcohol. Each section was routinely stained with hematoxylin and eosin or Masson’s trichrome, mounted on glass slides, and examined. For the analysis of H&E staining, at least eight distinct regions per animal were examined, and the average of these eight regions was calculated for the various tissues analyzed. For H3K9me3 immunostaining, the intensity was quantified in the small intestine and skin in four different regions per animal. Antibody used was rabbit Cell Signaling: Tri-Methyl-Histone H3 (Lys9) (D4W1U).

Immunofluorescence staining

Cells were washed with fresh PBS and then fixed with 4% paraformaldehyde (Roth, 0964.1) in PBS at room temperature (RT) for 15 minutes. After fixation, cells were washed 3 times, followed by a blocking and permeabilization step in 1% bovine serum albumin (Sigma, A9647–50G) in PBST (0.2% Triton X-100 in PBS) for 60 min (Roth, 3051.3). Cells were then incubated at 4°C overnight with appropriate primary antibody, washed in PBS, followed by secondary antibody incubation with DAPI staining at RT for 60 min. Coverslips were mounted using Fluoromount-G (Thermofisher, 00-4958-02), dried at RT in the dark for several hours, stored at 4°C until ready to image and −20°C for long-term.

Immunofluorescence imaging

Confocal image was acquired using the Ti2 Yokogawa CSU-W1 Spinning Disk (Nikon), using the 100X objective and with 15 z-sections of 0.3 μm intervals. The following lasers were used (405 nm and 488 nm) with a typical laser intensity set to 5–10% transmission of the maximum intensity for H3K9me3. Exposure time and binning were established separately to assure avoidance of signal saturation.

RNA extraction and quantitative RT-PCR analysis

RNA extraction

For cells total RNA extracted using Monarch Total RNA Miniprep Kit (New England Biolab, T2010s). Total DNA concentrations were determined using the Qubit RNA BR Assay Kit (Thermofisher, Q10210).

cDNA synthesis

cDNA synthesis was performed by adding 4 μL of iScript^™ gDNA Clear cDNA Synthesis (Biorad, 1725035BUN) to 500ng of RNA sample and run in a Thermocycler (Biorad, 1861086) with the following protocol: 5 min at 25°C for priming, 20 min at 46°C for reverse transcription, and 1 min at 95°C for enzyme inactivation.

qRT-PCR

cDNA was diluted 1:5 using nuclease free water and stored at − 20°C. qRT-PCR was performed in a Quantstudio 12K Flex Real-time PCR System instrument (Thermofisher) using SsoAdvanced SYBR Green Supermix (Bio-Rad, 1725274) in a 384-well PCR plate (Thermofisher, AB1384). Forward and reverse primers were used at a ratio 1:1 and final concentration of 5 μM with 1ul of cDNA.

Mouse Gene	Sequence (5’→3’)
Setdb1	Forward	GAGGAACTTCGTCAGTACATTGATG
	Reverse	ATCCTCAGAGCTACTGTCATGATACTG
Suv39h1	Forward	CCTGCCCTTGGTGTTTCTAA
	Reverse	CACGCCACTTAACCAGGTAATA
Cdkn1a	Forward	CGGTGTCAGAGTCTAGGGGA
	Reverse	ATCACCAGGATTGGACATGG
Stat1	Forward	GCTTGACAATAAGAGAAAGGAG
	Reverse	CTCGTCATTAATCAGAGTGTTC
IL6	Forward	CTGGGAAATCGTGGAAT
	Reverse	CCAGTTTGGTAGCATCCATC
Mcp1	Forward	GCATCCACGTGTTGGCTCA
	Reverse	CTCCAGCCTACTCATTGGGATCA
Gapdh	Forward	GGCAAATTCAACGGCACAGT
	Reverse	GTCTCGCTCCTGGAAGATGG

Antibodies and compounds

Antibodies were provided from the following companies. Cell Signaling: Tri-Methyl-Histone H3 (Lys9) (D4W1U), anti-γH2AX (9718), Santa Cruz Biotechnology: Lamin A/C Antibody (E-1) (sc-376248), Sigma: anti-β-Actin (A2228); Thermofisher: anti-Rabbit (A32790); Agilent: anti-Rabbit Immunoglobulins/HRP (P0448), anti-Mouse Immunoglobulins/HRP (P0447); Roth: DAPI (6843.1)

Western blot

Cell and tissue were lysed in RIPA buffer (50 mM Tris pH 7.5, 0.5 mM EDTA, 150 mM NaCl, 1% NP40, 0.1% SDS), protease inhibitors and ceramic beads using a MagNa lyser instrument (Roche). To the lysate 10% SDS was added to bring up the SDS concentration to 1%. The homogenate was sonicated for 10 minutes with Bioruptor, 30s on and 30s off then centrifuged at 21,000 g for 10 min at 4°C. The resulting supernatants were collected, and protein content determined by Quick Start Bradford kit assay (Bio-Rad, 500–0203). 5–15 μg of total protein was electrophoresed on 10% SDS– polyacrylamide gel, transferred to a nitrocellulose blotting membrane (Amersham Protran 0.45 μm, GE Healthcare Life Sciences, 10600002) and blocked in TBS-T (150 mM NaCl, 20 mM Tris–HCl, pH 7.5, 0.1% Tween 20) supplemented with 5% Bovine Serum Albumin (BSA). Membranes were incubated overnight at 4°C with the H3K9me3 primary antibody in TBS-T supplemented with 5% BSA, washed with TBS-T and next incubated with secondary HRP-conjugated anti-rabbit IgG (1:2,000, DAKO, P0448) or HRP-conjugated anti-mouse IgG (1:2,000, DAKO, P0447) for 1 hour at room temperature and developed using the ECL detection kit (Perkin Elmer, NEL105001EA). Antibodies: rabbit Tri-Methyl-Histone H3 (Lys9) (D4W1U) (1:3000); mouse anti-β-Actin (A2228) (1:10,000).

RNA-sequencing and analysis

RNA-sequencing

Total RNA was extracted by using the RNeasy Fibrous Tissue Mini Kit (Qiagen) with DNase treatment from 7 different tissues (liver, muscle, skin, spleen, small intestine, brain, and kidney) from TKOc and CTRL mice and livers of 3-month-old young and 18-month-old aged C57BL6/JN mice (n=6 per group, equally distributed males and females, total n=96 libraries) using a QIACube (Qiagen). Briefly, frozen tissue samples were transferred to 2 mL tubes (Sarstedt) containing 480 μL of RLT (RNA lysis buffer) (Qiagen) plus dithiothreitol (DTT, Teknova) buffer from the Qiagen kit, and 1 mm diameter Zirconia beads (Biospec) were added to make a total volume of 800 μL. The tube was processed using the Precellys 24 homogenizer, followed by centrifugation at 17,000 RCF (g) for 5 min at 4°C. The resulting homogenate was transferred to 1.5 mL Eppendorf tubes for Proteinase K treatment, and subsequently used for RNA isolation. A DNase digestion step was incorporated to remove contaminant genomic DNA. The quality and quantity of RNA were assessed using the Tapestation and the RNA Screen Tape (Agilent). RNA integrity numbers (RINs) ranged between 2.2–9.1 with the spleen having the lowest RINs due to high amounts of RNase.

~250 ng of total RNA from each sample was used as input to prepare libraries for total RNA-seq with Zymo-Seq RiboFree Total RNA Library Kit (Zymo Research) following the manufacturer’s protocol. The quality and quantity of the libraries were assessed using the High Sensitivity DNA 1000 Screen Tape (Agilent) on a Tapestation (Agilent). The RNA-seq libraries were pooled and paired-end sequenced on the NovaSeq 6000 platform (Illumina) using an S2 200 cycle kit (100 paired-end). We obtained ~45 million paired-end reads per sample.

RNA-seq analysis

For conventional gene-based analysis, Illumina sequencing reads (∼45 million paired-end reads per sample) were de-multiplexed using bcl2fastq/2.20.0.422. Reads were trimmed to remove adapter sequences using trimmomatic/0.39 (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). The quality of the resulting FASTQs was assessed using FASTQC/0.11.9 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and reads were aligned to the mouse reference genome (assembly GRCm38/mm10) using STAR/2.7.10b (PMID: 23104886). BAM files were sorted and indexed using samtools/1.17 (PMID: 19505943) and duplicates were removed using picard/3.1.0 (https://broadinstitute.github.io/picard/). The BAM files were then filtered to retain alignments with a minimum mapping quality of 10 using samtools/1.17 (PMID: 19505943). The featureCounts function of the Rsubread R package/2.16.0 was used to estimate counts for all transcripts. Differential gene expression analysis was performed using the R Bioconductor package, DESeq2/1.42.0.

For transposable element (TE) expression, trimmed FASTQs were aligned allowing for multimapping (--outFilterMultimapNmax 100 and --winAnchorMultimapNmax 100) using STAR/2.7.10b (PMID: 23104886) with unsorted BAM files as output. This was followed by assessment of significantly altered transcripts with the TEtranscripts function (--mode multi) in TEToolkit/2.2.3 (PMID: 29508296).

Gene Ontology and plots

Gene Ontology (GO) analysis and bubble plots

GO analysis was performed using DAVID/2021 (PMID: 19131956) with expressed genes as background for each tissue. The top 10 significant GO terms for upregulated and downregulated mRNAs for the Biological Process category are reported. The bubble sizes are scaled on normalized counts and colored on p-value.

GSEA (http://www.gsea-msigdb.org/gsea) was performed for each tissue using the mouse ortholog hallmark gene sets. Bubble plots were drawn with the top 10 and bottom 10 gene sets with the bubble size representing gene set size, and bubble color the nominal p-value. Additionally, GSEA was performed using clusterProfiler/4.14.3 for 2 gene sets related to senescence (SenMayo, PMID: 35974106) and aging (CellAge, PMID: 32264951). Running enrichment scores were plotted in R/4.4.1 using enrichplot/1.26.2.

Volcano plots

Volcano plots of TE expression changes were plotted in R/4.4.1 using the R Bioconductor packages ggplot2/3.5.1, dplyr/1.1.4, and ggrepel/0.9.6.

PCA plots

PCA plot was generated in R/4.4.1 with the TEtranscripts output using DESeq2 and then plotted with ggplot2/3.5.1.

Statistical analysis

Statistical analyses were conducted using GraphPad Prism (version 10). Outliers were identified and excluded using the ROUT method (5% threshold). For comparisons between two groups, unpaired two-tailed Student’s t-tests were used for data that met normal distribution criteria, and the Mann-Whitney U test was used for data that did not meet normal distribution criteria, as determined by the Shapiro–Wilk normality test. Data are presented as mean ± SEM (standard error of the mean). For comparisons involving more than three groups, One-Way ANOVA was used for data that met normal distribution criteria, and the Kruskal-Wallis test was used for data that did not meet normal distribution criteria, as determined by the Shapiro–Wilk normality test. Data are presented as median with interquartile range. Unless otherwise specified, ‘n’ represents the number of individual biological replicates, and each sample is represented as a dot on the graphs.

Extended Data

Extended Data Fig. 1: Genetic strategy for the generation of TKOc mice and loss of H3K9me3 in-vitro.

a, Detailed schematic representation of the genetic strategy to generate the quadruple transgenic TKOCAGCre mouse strain carrying the insertion of CAG-CreER (Chr.3), for the tamoxifen-inducible CRE-mediated recombination system (Loxp sites), Setdb1 gene insertion of LoxP sites in the intron 14 and 16 (Chr.3), Suv39h1 gene insertion of LoxP sites in the intron 2 and 5 (Chr.X), Suv39h2 gene deletion at Exon2 (Chr. 10). b, Western blot of H3K9me3 in CTRL and TKOc tail-tip fibroblasts with or without TAM treatment. H3K9me3 quantification in CTRL and TKOc after treatment.

Extended Data Fig. 2: TKOc characterization upon single tamoxifen treatment.

a, Experimental design. b, PCR analysis of Suv39h1 and Setdb1 recombination in blood from CTRL and TKOc mice upon Tamoxifen treatment. (n = 3 to 6). c, Frailty index of CTRL and TKOc mice (n = 6 to 6, 6-month-old mice). d, Changes in body weight of CTRL and TKOc mice, upon administration of tamoxifen. Arrow show the TAM injection timepoint e, Hematological parameters in CTRL and TKOc mice (n = 11 to 12, 6-month-old mice). f, Behavioral characterization of CTRL and TKOc mice (6 months). Open field exploration (15 minutes), quantification of distance travelled, mean speed, time spent and distance traveled in the center with a representative tracking trace (n = 12 to 11–12, 6-month-old mice). g, Grip strength measured as average grip strength of three different trials CTRL and TKOc mice. (n = 11 to 12, 6-month-old mice). h, Kaplan-Meier survival curves for CTRL and TKOc mice (n =13 to 13 mice, including males and females). Dots in all panels represent individual sample data. Survival curve data were analyzed by log-rank (Mantel-Cox test). Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test. *p < 0.05; **p < 0.01; ***p < 0.001.

Extended Data Fig. 3: Analysis of recombination and H3K9me3 levels in tissues isolated from TKOc and CTRL mice tissues.

a, PCR analysis of Suv39h1 and Setdb1 recombination in brain, skin, skeletal muscle, liver, small intestine and spleen from CTRL and TKOc mice upon Tamoxifen treatment. (n = 5 to 6, 7–8-month-old mice). b, Immunostaining and quantification of H3K9me3 intensity in small intestine and skin of treated, CTRL and TKOc mice. (n = 2 to 3, 7–8-month-old mice). c, Western blot of H3K9me3 in CTRL and TKOc small intestine and skin with TAM treatment. H3K9me3 quantification in CTRL and TKOc after treatment. (n = 3 to 3, 7–8-month-old mice). Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test. *p < 0.05; **p < 0.01; ***p < 0.001.

Extended Data Fig. 4: Additional age-associated organ changes in TKOc and CTRL mice.

a, Hematological parameters in CTRL and TKOc mice (n = 11 to 16, 6-month-old mice). Including HCT, Hematocrit. MCV, Mean Corpuscular Volume. RDW-CV, Red Cell Distribution Width-Coefficient of Variation. MCH, Mean Corpuscular Hemoglobin. MCHc, Mean corpuscular hemoglobin concentration. PLT, Platelet. MPV, Mean Platelet Volume. Eos, Eosinophil. Lym, Lymphocyte. Mon, Monocyte. Neu, Neutrophil. Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test *p < 0.05; **p < 0.01; ***p < 0.001. b, Plasma levels of inflammatory cytokines Eotaxin (CCL11), GRO alpha (CXCL1), Interferon-gamma (IFN-γ) and, IL-18 detected by multiplex in 7–8-month-old CTRL and TKOc mice (n = 5 to 6). c, Analysis of fibrosis in kidney, liver and spleen by Masson’s trichrome staining of CTRL and TKOc mice at 7–8 months of age.

Extended Data Fig. 5: Comprehensive epigenetic and transcriptomic analysis from multiple tissues of TKOc and CTRL mice.

a, Epigenetic age of skeletal muscle, liver and brain of mice TKOc mice compare their CTRL (n = 6 to 6, 7–8-month-old mice). Data are mean ± SEM. Statistical significance was assessed by Two-tailed Student’s t test *p < 0.05; **p < 0.01; ***p < 0.001. b, Overall DNA methylation levels estimated using the median beta-values. Each dot is a biological sample, and they are grouped by tissue, condition and sex. c, PCA plot from total RNA-seq data derived from TKOc and CTRL mice tissues. d, PCA plot from liver total RNA-seq data from TKOc, CTRL, young (3-months-old), and aged (18-months-old) C57BL/6JN mice. e, GSEA enrichment analysis of SenMayo gene set in TKOc mice tissues. f, Volcano plots showing TE transcripts that are significantly upregulated (in orange), downregulated (in blue) or unchanged (in gray) in skeletal muscle, brain and liver from young and old WT mice. Transcripts were determined to be significant if the adjusted p-value was < 0.05 and log₂fold change was > 1. A few top upregulated transcripts are labeled in each tissue.

Funding Statement

This work was supported by the Milky Way Research Foundation (MWRF), the Eccellenza grant PCEGP3_186969 from the Swiss National Science Foundation (SNSF), the University of Lausanne, and the Canton Vaud. GD-M was supported by the EMBO postdoctoral fellowship (EMBO ALTF 444–2021 to GD-M). PS was supported by the National Institute of Health (NIH ZIA AG000679 to PS).

Data availability:

The authors confirm that data supporting the findings of this study are available within the article and its Supplementary Information, or are available from the corresponding author upon reasonable request. Materials: The mouse model described in this work will be made available to investigators through an institutional or third-party Material Transfer Agreement (MTA) upon reasonable request. Data: Raw sequencing reads related to tissue analysis were deposited in the Gene Expression Omnibus (GEO) under the project number (GSE262109).