Dietary Restriction Slows Blood Vessel Aging, Affects Immune Signaling, and Reverses Cell Changes in DNA Repair-Deficient Aging Mice

Ercc1^Δ/− mice were generated by crossing Ercc1^Δ/+ (in a pure C57BL6J or FVB background) with Ercc1^+/− mice (in a pure FVB or C57BL6J background respectively), resulting in Ercc1^Δ/− offspring with a genetically uniform F1 C57BL6J/FVB hybrid background. Wildtype (WT) F1 littermates served as controls. Animals were housed at the Animal Resource Center (Erasmus University Medical Centre), and all experiments were conducted in accordance with the Principles of Laboratory Animal Care and the guidelines approved by the Dutch Ethical Committee in full compliance with European legislation.

All animals were bred and maintained on AIN93G synthetic pellets (Research Diet Services B.V.; gross energy content 4.9 kcal/g dry mass, digestible energy 3.97 kcal/g). Dietary restriction (DR) was initiated at 7 weeks of age by 10% food reduction (g/day). DR was increased weekly by 10% (g/day), until reaching 30% DR from 9 weeks of age onward. Mice in the ad libitum (AL) group were allowed unlimited access to AIN93G pellets. Mice were kept on an adjusted light–dark schedule (light from 12 pm to 12 am and dark from 12 am to 12 pm) to facilitate feeding just before the start of the dark (active) period, in order to avoid disruptions in the biological clock. Mice were individually housed to control food intake. At 6 weeks and 18–22 weeks of age, mice were sacrificed by CO₂ inhalation for the described experiments.

To assess the effect of DR on aortic and cardiac function and geometry, echocardiographic measurements were conducted on 6‐week‐old and 20–22‐week‐old Ercc1^Δ/− and Ercc1^+/+ mice that had either been on DR or an AL diet (n = 8–10 per group). Mice were anesthetized with 2.5% isoflurane and ventilated with 35% O₂. Echocardiography of the ascending aorta and left ventricle (LV) was performed using a Vevo3100 (VisualSonics Inc., Toronto, Canada). Aortic and LV diameters, ejection fraction, stroke volume, and fractional shortening were obtained from M‐mode images. Aortic strain was calculated using the following formula: (aorta diameter in systole—aorta diameter in diastole)/aorta diameter in systole * 100%.

Mice were intravenously injected with matrix metalloprotease (MMP) Sense680 (Perkin Elmer, Waltham, MA, USA) at a concentration of 2 nmol per 25 g of body weight. After 24 h, aortas were collected and imaged using the Odyssey CLx imaging system (LI‐COR Biosciences, Lincoln, NE, USA). Probe intensity was quantified with Image Studio Lite version 5.20 (LI‐COR Biosciences, Lincoln, NE, USA). Probe intensity was normalized by dividing the intensity by the shape area, resulting in intensity/mm² per aorta. Ercc1^Δ/− aortas were normalized to the average of the WT littermate aortas that were scanned simultaneously. Measurements were performed on aortas of Ercc1^Δ/− and Ercc1^+/+ mice at 18–20 weeks of age subjected to either DR or an AL diet (n = 5–7 per group).

Ercc1^Δ/− mice and their WT littermates were sacrificed at ages of 6 weeks or 22 weeks (n = 4–5 per group). Table S1 indicates the number of samples per group, including the sex of the mice used. Mice were perfused through the left ventricle with phosphate buffered saline (PBS) and the aortic arch was isolated. Aortic tissue was homogenized using the TissueLyser 24 (Shanghai Jing Xin). Total RNA and microRNA were isolated from the aortic arch using the miRNeasy Mini Kit (Qiagen). Library preparation and mRNA sequencing were performed at GenomeScan B.V. (Genomescan B.V., Leiden, The Netherlands). Libraries were sequenced on an Illumina NovaSeq 6000 platform, generating paired‐end reads (150 bp read length) with a sequencing depth of 20 million reads, including Unique Molecular Identifiers (UMIs).

We used UMI‐Tools v1.1.2 to move the UMI barcodes from separate files to the read pair headers (Smith et al. 2017). The raw fastq files were pre‐processed with fastp v0.23.2 (Chen et al. 2018). Samples were then aligned using STAR v2.7.10a against the mouse reference genome mm39 with ENSEMBL v107 annotation (Dobin et al. 2012). After alignment, we applied UMI‐Tools to deduplicate reads originating from the same molecule and minimize gene‐length biases from PCR amplification. Quality Control (QC) was assessed using fastqc v0.11.9 and RSeQC v5.0.1 (Andrews 2010; Wang et al. 2012; Wang et al. 2016). MultiQC v1.15.dev0 was used to summarize the QC of these two programs and the different steps of the data processing (Ewels et al. 2016). Read counts overlapping the entire gene body were generated using htseq‐count v2.0.2 (Putri et al. 2022).

The resulting gene‐count matrices were analyzed using R v4.1.2 (Team RC 2021). One sample (003, WT, female, 22 weeks, AL) showed too low read counts and diverging expression profiles, and it was therefore excluded from subsequent analyses.

Differential gene expression between groups and conditions was analyzed using DESeq2 v1.34.0, with an adjusted p‐value threshold of 0.05 and a |log2FC| ≥ log2(1.2) (~ ± 0.26) (Love et al. 2014). Analyses were restricted to genes with > 10 read counts in more than 10 samples. The design model used was “design = ~group + sex”, where group represents the combination of age, genotype and diet, allowing the decoupling of the effect of sex from the comparisons of age, genotype and diet. Comparisons between samples at 6 weeks and 22 weeks of age were performed only for AL samples, excluding DR samples to avoid biases in gene expression levels. Conversely, analyses on the effect of DR used only 22‐week‐old samples.

NOTE: Part of this data has been used in a previously published study by our lab (van der Linden et al. 2024). This concerns only the data from AL mice, and the dataset itself has not been previously published.

Gene Set Enrichment Analysis (GSEA) was performed using the R package fGSEA v1.20.0, and MSigDB (v2022.1) mouse Canonical Pathways were used as a reference (Subramanian et al. 2005; Castanza et al. 2023; Nishimura 2001; Milacic et al. 2023; Agrawal et al. 2023; Korotkevich et al. 2021). Each comparison between groups and conditions was analyzed separately using the log2 Fold Change of all detected genes. Genes not present in the data were excluded from the background.

For biological interpretation of differential expression data, normalized expression values were uploaded into Ingenuity Pathway Analysis (IPA software version 94302991, QIAGEN, Redwood City, California, USA). An IPA core analysis was performed with significantly differentially expressed genes. Canonical pathways and upstream regulators were investigated, with upstream regulators selected based on z‐score cut‐offs (−2.0 ≤ z‐score ≥ 2.0) and p‐value ≤ 0.05. A z‐score ≥ 2 indicates predicted activation of the upstream regulator, and a z‐score ≤ −2.0 indicates predicted inhibition of the upstream regulator. Top upstream regulators were ranked based on their z‐scores.

For histological analysis, mice were sacrificed by CO₂ inhalation (number of mice used per analysis is indicated in the corresponding figure legends). Mice were perfused through the left ventricle with PBS. Aortas were fixed in formalin, segmented, and embedded in 3% Bacto agar. Embedded aortas were dehydrated through the histokinette processor (Microm) and subsequently paraffin embedded, after which 4 μm sections were prepared and placed on slides. Resorcin‐Fuchsin (RF) staining was performed to visualize the elastin structure of the vessel wall. Elastin fragmentation was counted manually and corrected for surface size. Alcian Blue (AB) staining was performed to visualize proteoglycan deposition in the aorta. A macro in Fiji (ImageJ) was designed to determine the mean percentage of AB positive area. In short, eosin counterstaining was removed from the image and an overlay mask was created to measure the staining intensity relative to the total section surface. The percentage of stained pixels was determined by setting a threshold to distinguish stained pixels from the background.

For immunohistochemical analysis, aorta sections were deparaffinized in xylene, rehydrated, and washed in washing buffer (PBS with 0.025% Triton X‐100). Antigen retrieval was performed by boiling in antigen retrieval buffers EDTA buffer (10 mM Tris base, 1 mM EDTA solution, pH 9.0) for MAC2 and citrate buffer (10 mM Sodium citrate, 0.05% Tween‐20, pH 6.0) for VCAM1 at 300 W for 15 min. Sections were treated with 3% H₂O₂ in methanol for 10 min to block endogenous peroxidase activity and subsequently blocked with 5% Protifar in washing buffer for 1 h. Sections were incubated overnight at 4°C with primary antibodies diluted in antibody buffer (1% Protifar in washing buffer) (Table S2). The following day, sections were washed with washing buffer and incubated with biotinylated secondary antibody diluted in antibody buffer (1:200, DAKO) for 30 min at room temperature (RT). Subsequently, sections were washed with washing buffer and incubated with avidin‐biotinylated complex diluted in washing buffer (Vectastain Universal Elite ABC kit Vector Laboratories) for 30 min at RT. After a final washing step, sections were incubated with DAB chromogen (DAKO Liquid Dab substrate‐chromogen system) as a substrate and subsequently incubated in CuSO₄ for 5 min to boost the signal. Sections were counterstained with eosin. After dehydrating the sections and clearing them with xylene, they were mounted with Pertex mounting medium. Positive cells were counted manually and corrected for surface area.

Ercc1^Δ/− (n = 3) and Ercc1^+/+ (n = 3) mice were sacrificed at 6 weeks of age and perfused through the left ventricle with PBS. Aortas were isolated and incubated in 2 mg/mL collagenase type II (LS004176↗, Worthington) at 37°C, 5% CO₂ for 1–6 h until the tissue was dissociated. Primary VSMCs were centrifuged, resuspended in Dulbecco's Modified Eagle's Medium (DMEM) (Gibco) supplemented with 10% fetal calf serum (FCS) and 1% penicillin–streptomycin (PS) and cultured on dishes coated with 0.1% gelatin at 37°C, 5% CO₂.

To investigate cGAS‐STING pathway activation in vitro, Ercc1^Δ/− and Ercc1^+/+ VSMCs (passage numbers 9–12) were seeded in 12‐well plates (50,000 cells/well) on coverslips coated with 0.1% gelatin. The next day, cells were treated with 2 μg/mL cisplatin (Accord; infusion concentrate used in the clinic) for 48 h. After incubation, VSMCs were fixed with ice cold 100% methanol for 10 min and permeabilized with ice cold acetone for 1 min (for STING staining) or fixed with 2% paraformaldehyde in PBS for 15 min (for cGAS and ssDNA staining). Fixed cells were washed with PBS/0.1% Triton X‐100 and blocked with PBS+ (PBS supplemented with 0.5% bovine serum albumin (BSA) and 0.15% glycine) for 20 min. Next, cells were incubated overnight at 4°C with primary antibody diluted in PBS+ (Table S2). After washing, cells were incubated for 1 h at RT with secondary antibodies diluted in PBS+ (Table S2) followed by Hoechst 33342 staining (1:1000 in PBS, ThermoFisher). Images were recorded on a Stellaris 5 confocal microscope (Leica) with a 40x oil objective. Staining intensities of STING and cGAS in the cytoplasm were quantified using an in‐house developed macro in Fiji (ImageJ). Nuclei were selected based on Hoechst staining, and the mean intensity of STING or cGAS staining was measured outside the nuclei mask. The number of ssDNA spots was quantified by counting spots in Fiji and correcting for the number of nuclei. Experiments were performed in duplicates.

For immunoblot analysis, mice were sacrificed at an age of 22 weeks (n = 3 per group) and perfused with PBS. Aortic tissue was isolated, snap frozen in liquid nitrogen, and homogenized in 1× Laemmli buffer (2% SDS, 10% glycerol, 60 mM Tris pH 6.8) using the TissueLyser 24 (Shanghai Jing Xin). Samples were heated for 10 min at 65°C and pulled through a 25G syringe to make them less viscous. The protein concentration was determined using the Lowry assay. Proteins (10 μg per sample) were separated by SDS‐PAGE and transferred to PVDF membranes (1.5 h, 100 V, 4°C, Immobilon). Total protein was measured using the Revert 700 Total Protein Stain kit (LI‐COR biosciences). Membranes were blocked for 1 h at RT with 3% milk in PBST (PBS + 0.1% Tween‐20) and incubated with primary antibodies overnight at 4°C (Table S2). After washing with PBST, membranes were incubated with horseradish peroxidase‐conjugated secondary antibodies (Table S2) for 1 h at RT. After washing, secondary antibodies were detected with the Amersham Imager 600 (GE Healthcare Life Sciences) using chemiluminescence. The intensity of the signal was measured using Fiji image analyzing software.

Statistical analysis was performed using GraphPad Prism 10.2.0 (GraphPad Software Inc., La Jolla, California, USA). The D'Agostino‐Pearson (omnibus K2) test was used to test for normal distribution of the data. For normally distributed data, a two‐way ANOVA was applied to evaluate the effects of age and genotype or diet and genotype. For non‐normally distributed data, the Kruskal–Wallis test was used to determine significant differences between groups. The Sidak test was employed to correct for multiple comparisons. A p‐value < 0.05 was considered significant. Results are expressed as mean ± standard deviation (SD). In the figures, statistical significance is indicated as follows: one asterisk (*) indicates a p‐value < 0.05, two asterisks (**) indicate a p‐value < 0.01, three asterisks (***) indicate a p‐value < 0.001, and four asterisks (****) indicate a p‐value < 0.0001.

To investigate the aorta and left ventricle (LV) diameter and function in Ercc1^Δ/− mice and WT littermates and the impact of DR on these parameters, we used ultrasound imaging. Measurements were performed at 22 weeks of age because all Ercc1^Δ/− mice on an AL diet show advanced age‐related symptoms, including vascular aging, at this timepoint (van der Linden et al. 2024). Aorta diameter was significantly smaller in Ercc1^Δ/− mice compared to WT littermates at 22 weeks of age, in both the AL‐fed and DR conditions (Figure 1A). Since Ercc1^Δ/− mice experience growth retardation, they are significantly smaller in terms of total body size and bodyweight compared to WT littermates, and therefore aorta diameter is expected to be smaller (Figure S1) (Dollé et al. 2011; Waard et al. 2010). DR did not affect aorta diameter in either genotype. However, aortic strain, which measures the stiffness of the aorta, was significantly increased in AL‐fed Ercc1^Δ/− mice compared to WT littermates (Figure 1B). It should be noted that alterations in blood pressure can influence these measurements, and Ercc1^Δ/− mice were reported to have higher blood pressure compared to WT littermates (Durik et al. 2012). Although not significant, DR appeared to increase aortic strain, indicating a decreased stiffness in both genotypes (Figure 1B). Similarly, LV diameter in diastole was significantly smaller in AL‐fed Ercc1^Δ/− mice compared to WT littermates (Figure 1C). Fractional shortening, a measure of LV contractility, and ejection fraction, a measure of LV pumping efficiency, were both significantly reduced in AL‐fed Ercc1^Δ/− mice compared to WT littermates (Figure S2). Both parameters appeared to increase with DR in Ercc1^Δ/− mice, although this effect did not reach statistical significance. Stroke volume, the volume of blood ejected from the LV per cardiac cycle, was significantly lower in AL‐fed Ercc1^Δ/− mice compared to WT littermates. While not significant, DR tended to increase stroke volume in Ercc1^Δ/− mice (Figure 1D). In WT mice, DR significantly reduced stroke volume, indicating that DR can have differential effects depending on genotype and health status.

Next, we examined whether these functional changes were related to structural changes in the aortic wall. Since vascular aging is characterized by elastin fragmentation, we performed a Resorcin‐Fuchsin staining to analyze elastin structure (Figure 1E). Elastin fragmentation was significantly increased in the Ercc1^Δ/− aorta compared to the WT aorta of mice on an AL diet, consistent with previous reports (Figure 1F) (van der Linden et al. 2024). Notably, elastin fragmentation was significantly decreased in Ercc1^Δ/− mice on DR compared to AL‐fed Ercc1^Δ/− mice (Figure 1F). Another indication of extracellular matrix (ECM) remodeling is the deposition of proteoglycans, visualized by Alcian Blue staining. Alcian Blue staining was non‐significantly increased in the aorta of Ercc1^Δ/− mice compared to the aorta of WT mice on an AL diet, consistent with previous findings (Figure 1G). Remarkably, DR significantly reduced Alcian Blue staining in the aortic wall of Ercc1^Δ/− mice. Together, these results suggest that DR can partially reverse the structural and functional abnormalities observed in the aging aorta and heart of Ercc1^Δ/− mice.

To further explore vascular aging in Ercc1^Δ/− mice, bulk mRNA sequencing was conducted on the aorta of Ercc1^Δ/− and WT mice using the Illumina NovaSeq6000 platform. Differential expression analysis of the aortic arch of 6‐week‐old Ercc1^Δ/− mice compared to their WT littermates revealed 373 (Table 1) significant differentially expressed genes (DEGs). At 22 weeks of age, differential expression analysis revealed 1719 significant DEGs between these two groups. This 4.6‐fold increase in the total number of DEGs between Ercc1^Δ/− and WT mice in a time span of 18 weeks reflects the accelerated aging phenotype of Ercc1^Δ/− mice. Principal Component Analysis (PCA) showed a full separation between Ercc1^Δ/− and WT samples, indicating that genotype is the most prominent difference. Furthermore, samples of 6‐week‐old Ercc1^Δ/− mice clustered more closely with those of WT mice than samples of 22‐week‐old Ercc1^Δ/− mice did, indicating a progressive shift in RNA expression pattern upon aging (Figure 2A). Differential expression data confirmed significantly decreased Ercc1 gene expression in the Ercc1^Δ/− aorta compared to WT aorta (Figure 2B).

The top 20 most significantly changed genes in each direction in 22‐week‐old aortas from AL‐fed Ercc1^Δ/− compared to AL‐fed WT mice contained ECM remodeling genes such as Mmp3, Mmp17 and Timp4 (Figure 2C). Additionally, the gene encoding stress/senescence marker p21 (Cdkn1a) was significantly upregulated, confirming prior findings of increased ECM remodeling and p21 expression in Ercc1^Δ/− aortas (van der Linden et al. 2024). Gene Set Enrichment Analysis (GSEA) further indicated significant enrichment of pathways involved in ECM remodeling, including “elastic fibre formation”, “collagen chain trimerization” and “collagen formation” (Figure 2D). The “muscle contraction” pathway had a negative enrichment score, indicating downregulation of contraction‐related genes in the Ercc1^Δ/− aorta. This was consistent with previous studies revealing decreased protein expression of contractile VSMC markers (van der Linden et al. 2024). Additionally, pathways related to oxidative stress and redox responses had a positive enrichment score, suggesting upregulation of these processes (Figure 2D). This was consistent with previous findings showing the increased accumulation of oxidative DNA damage and elevated reactive oxygen species (ROS) levels in Ercc1^Δ/− liver and kidney (Robinson et al. 2018). Further analysis with Ingenuity Pathway Analysis (IPA) revealed downregulation of transcription factors Jun and Srf, which are downstream of the IGF‐1 signaling pathway, and predicted inhibition of this pathway in the 22‐week‐old Ercc1^Δ/− aorta of AL‐fed mice (Figure S3). Furthermore, GSEA analysis highlighted immune response pathways including “microglia pathogen phagocytosis pathway”, “complement cascade” and “type II interferon signaling”, and categorization of enriched pathways revealed that “immune system and inflammation” was the largest category containing 14 of 61 pathways (Figure 2E). These findings suggest that Ercc1 deficiency triggers a pro‐inflammatory immune response in the aorta, similar to effects seen in other tissues such as adipose tissue, the kidney and the pancreas of Ercc1^Δ/− mice (Karakasilioti et al. 2013; Chatzidoukaki et al. 2021; Schermer et al. 2013). Altogether, these findings revealed that mechanisms previously observed in Ercc1^Δ/− mice in different tissues were also visible at the RNA expression level in the aorta of AL‐fed Ercc1^Δ/− mice.

To assess the effect of DR on vascular aging, we compared RNA expression in the aorta of DR‐fed 22‐week‐old mice to AL‐fed 22‐week‐old mice. Differential expression analysis revealed 330 significant DEGs in the aortic arch of Ercc1^Δ/− mice on DR compared to the same group on an AL diet (Table 1). In WT mice, 201 DEGs were identified when comparing mice on DR to AL‐fed mice. When matching these two gene sets, only 38 genes were found to have a shared response upon DR in Ercc1^Δ/− mice and WT mice (26 downregulated genes, 12 upregulated genes), with 1 DEG switching direction (Figure 3A). This suggests that DR affects Ercc1^Δ/− mice at least in part differently than WT mice. GSEA analysis revealed that DR downregulated the upregulated “oxidative damage response” pathway in Ercc1^Δ/− mice (Figure 3B). Specifically, antioxidant genes Gpx1, Srxn1 and Txnrd1 were upregulated in the 22‐week‐old Ercc1^Δ/− aorta compared to WT aorta and DR restored their expression to baseline levels in Ercc1^Δ/− mice (Figure 3C). Additionally, immune response pathways like “MHC class II antigen presentation” and “microglia pathogen phagocytosis pathway” were downregulated by DR (Figure 3B). Furthermore, pathways involved in energy metabolism, such as “the citric acid cycle and respiratory electron transport”, were altered by DR. These results suggest that DR mitigates processes that contribute to impaired structure and function in the aging Ercc1^Δ/− aorta.

To explore the effects of DR on accelerated aging in Ercc1^Δ/− mice, we focused on significant DEGs in aging (22 weeks vs. 6 weeks, AL) that were also changed upon DR (DR vs. AL, 22 weeks) but switched directions (upregulated in aging but downregulated by DR, and vice‐versa). Among the 71 genes that switched direction, 52 were upregulated in aging but downregulated by DR, while 19 were downregulated in aging but upregulated by DR (Figure 4A). GSEA pathway analysis showed enrichment in pathways involved in ECM remodeling, oxidative damage response, and immune response (Figure 4B,C) further supporting that DR alleviates these processes in the Ercc1^Δ/− aorta.

RNAseq data suggested increased ECM remodeling in the aging Ercc1^Δ/− aorta, which is alleviated upon DR, underlined by the expression of Mmp3, Col8a1, and Timp1 (Figure 5A). These genes were upregulated in the aging Ercc1^Δ/− aorta and downregulated by DR. Immunoblot analysis showed that MMP3 protein levels were not elevated in AL‐fed Ercc1^Δ/− mice (Figure S4). Furthermore, a trend of increased TIMP1 protein expression was observed in the Ercc1^Δ/− aorta of AL‐fed mice, which seemed to be non‐significantly decreased by DR (Figure S4). The discrepancy between RNA and protein levels of these markers could be attributed to post‐transcriptional regulation, which is tightly controlled given the critical role of MMPs in physiological processes. To assess functionally relevant changes, we used the protease‐activatable MMPSense680 probe to measure enzymatic activity of MMPs. MMP activity was significantly higher in the Ercc1^Δ/− aorta compared to WT, and DR significantly reduced this activity (Figure 5B,C). Furthermore, DR significantly decreased fragmentation of elastin fibers, which is one of the substrates of MMP12, in the Ercc1^Δ/− aorta (Figure 1H). These results suggest that DR reduces ECM remodeling in the Ercc1^Δ/− aorta.

As mentioned above, our RNAseq data suggests the presence of an immune response in the aging Ercc1^Δ/− aorta, which is downregulated by DR. To further investigate this, an upstream regulator (UR) analysis was performed using IPA, which predicts whether a protein, gene, drug, or chemical is activated or inhibited based on the differential expression of downstream genes. For drugs or chemicals, which are not endogenously present, IPA bases its findings on the expression of genes that are known to be altered by the drug or chemical from experimental findings. This analysis revealed that immunostimulatory agents, including poly rl:rC‐RNA and lipopolysaccharide, were predicted to be activated in the 22‐week‐old Ercc1^Δ/− aorta compared to both the 6‐week‐old Ercc1^Δ/− aorta and 22‐week‐old WT aorta (Figure S5). Conversely, these URs were predicted to be inhibited by DR specifically in the Ercc1^Δ/− aorta, suggesting the presence of an immune response in the Ercc1^Δ/− mouse aorta that may be alleviated by DR.

In response to stress or damage, contractile VSMCs can dedifferentiate and adapt their phenotype to address stressful conditions, such as high blood pressure or damage. One such phenotype is the macrophage‐like phenotype (Cao et al. 2022). These VSMCs acquire phagocytic properties and interact with both resident and non‐resident immune cells, which can trigger a chronic inflammatory state in the aortic wall and promote vascular aging (Sorokin et al. 2020). This macrophage‐like phenotype is characterized by the expression of marker genes Vcam1, Cd68, and Lgals3/Mac2. All three genes were upregulated in the Ercc1^Δ/− aorta when compared between 22 and 6 weeks, as well as when compared to the WT aorta at 22 weeks (Figure 6A). Conversely, DR downregulated the expression of these genes in the Ercc1^Δ/− aorta (Figure 6A). Immunoblot analysis confirmed increased expression of VCAM1 protein in the Ercc1^Δ/− aorta compared to the WT aorta in AL‐fed mice (Figure 6B). DR resulted in a non‐significant trend toward decreased VCAM1 protein expression in the Ercc1^Δ/− aorta (p = 0.1) (Figure 6B). Unlike our RNAseq findings, no decrease in MAC2 protein expression was observed upon DR in the Ercc1^Δ/− aorta. To assess localization, immunohistochemical staining for VCAM1 and MAC2 was performed to identify cells positive for these markers (Figure 6C,D). Quantification of these stainings revealed a significantly increased number of VCAM1‐ and MAC2‐positive cells in the medial layer of the 22‐week‐old Ercc1^Δ/− aorta compared to the 22‐week‐old WT aorta in AL‐fed mice (Figure 6C,D). DR significantly reduced the number of MAC2‐positive cells in the Ercc1^Δ/− aorta and, although not significant, also reduced the number of VCAM1‐positive cells. Staining for CD3, a marker for T‐cells, confirmed that these were not immune cells infiltrating the aortic wall (Figure S6). Collectively, these results indicate that DR reduces the phenotypic switch of VSMCs to a macrophage‐like phenotype in the Ercc1^Δ/− aorta.

Defects in DNA damage repair are frequently linked to activation of the cGAS‐STING pathway (Chatzidoukaki et al. 2021). This activation, in turn, causes downstream expression of type I interferons and pro‐inflammatory cytokines including IL‐6 and TNF, which can trigger a DNA‐driven immune response. We therefore investigated whether activation of the cGAS‐STING pathway could underlie the observed immune response in the Ercc1^Δ/− aorta. UR analysis revealed predicted activation of a number of proteins involved in the cGAS‐STING pathway in the 22‐week‐old Ercc1^Δ/− aorta compared to the WT aorta in AL‐fed mice, such as interferon alpha, interferon beta, STING1, IRF7, and IRF3 (Figure 7A). Since STING1 is a key signal transducer in this pathway, we focused on its role. IPA generated mechanistic networks that connected URs to visualize how they might collaborate to induce the observed gene expression changes in our dataset. The mechanistic network for STING1 showed connections to several URs involved in the cGAS‐STING pathway, such as interferon beta, IRF3, IRF7, and NF‐κB, all of which were predicted to be activated (Figure 7B). Moreover, the canonical pathways downstream of cGAS‐STING, “interferon signaling” and “IL‐6 signaling” were both predicted to be activated (Figure 7C). Conversely, STING1 was predicted to be inhibited by DR in the aorta of Ercc1^Δ/− mice (Figure 7D). The resulting mechanistic network again contained URs involved in the cGAS‐STING pathway such as IRF3 and NF‐κB, but in this comparison, they were predicted to be inhibited. Specifically, expression of the transcription factors Irf7 and Irf9, which induce downstream expression of type I interferon genes, was significantly increased in the 22‐week‐old Ercc1^Δ/− aorta compared to the WT aorta in AL‐fed mice (Figure 7E). DR reduced expression of these genes in the Ercc1^Δ/− aorta, which was significant for Irf7. These findings suggest that the cGAS‐STING pathway is activated in the 22‐week‐old Ercc1^Δ/− aorta and that DR inhibits this pathway. Immunoblot analysis confirmed significantly increased expression of STING1 protein in the 22‐week‐old Ercc1^Δ/− mouse aorta compared to the WT aorta, with a trend toward decreased STING1 protein in the Ercc1^Δ/− aorta following DR (p = 0.07) (Figure 8A).

Next, we validated in vitro that impaired DNA damage repair could cause activation of the cGAS‐STING pathway. UR analysis focused on drugs and chemicals and identified gene expression changes typically induced by cisplatin, a drug that causes DNA damage by forming intrastrand and interstrand DNA cross‐links, in the 22‐week‐old Ercc1^Δ/− aorta of AL‐fed mice (Rocha et al. 2018) (Figure S5). These gene expression changes were reversed by DR in the Ercc1^Δ/− aorta (Figure S5). The DNA damage induced by cisplatin is typically repaired by nucleotide excision repair and interstrand crosslink repair (Rocha et al. 2018), which are impaired in Ercc1^Δ/− mice. These results suggest that DNA damage accumulation in the Ercc1^Δ/− aorta with aging may cause cGAS‐STING activation, which can be reduced by DR. Therefore, we investigated whether cisplatin‐induced DNA damage could activate the cGAS‐STING pathway in Ercc1^Δ/− VSMCs in vitro. After isolating VSMCs from mouse aorta, immunohistochemical staining for MYH11 and αSMA confirmed the VSMC phenotype (Figure S7). Treatment of VSMCs with 2 μg/mL cisplatin for 48 h resulted in a trend toward increased STING1 staining intensity (p = 0.1) in Ercc1^Δ/− VSMCs, but not in WT VSMCs (Figure 8B). Consistently, cisplatin treatment induced a significant increase in cGAS staining intensity in the cytoplasm of Ercc1^Δ/− VSMCs, but not in WT VSMCs (Figure 8C). Furthermore, cisplatin treatment induced the release of single‐stranded DNA (ssDNA) fragments into the cytoplasm of Ercc1^Δ/− VSMCs (Figure 8D). Quantification showed a significant increase in the average number of ssDNA spots per cell following cisplatin treatment in Ercc1^Δ/− VSMCs, but not in WT VSMCs (Figure 8D). These results suggest that cisplatin‐induced DNA damage, which cannot be adequately repaired by ERCC1‐deficient VSMCs, leads to the release of ssDNA fragments into the cytoplasm, which subsequently activates the cGAS‐STING pathway.

Despite its therapeutic potential, DR has several limitations that make its clinical application challenging, including difficulties with adherence, potential impacts on quality of life, and adverse effects such as reduced bone and muscle mass. Therefore, our objective was to identify compounds that could mimic the beneficial effects of DR. To achieve this, we conducted an UR analysis focused on drugs and chemicals predicted to be activated by DR in the Ercc1^Δ/− aorta. This analysis is based on the differential expression of genes in our dataset that, according to the literature, are affected by these compounds. Interestingly, this revealed predicted activation of sirolimus (rapamycin) and resveratrol, which are both known DR mimetics (Figure S8). Rapamycin is an inhibitor of the mTOR pathway and increases maximum lifespan in invertebrates and mice. More specifically, rapamycin treatment reverses age‐related vascular dysfunction and oxidative stress in naturally aged mice (Lesniewski et al. 2017). Resveratrol is a polyphenolic compound and has potential antioxidant and anti‐inflammatory activity. Resveratrol supplementation reduces atherosclerotic lesion size and promotes lipid metabolism in ApoE^−/− mice (Cheng et al. 2023). These previous findings suggested that these compounds, at least in part, mimic the beneficial effects of DR and have the potential to alleviate vascular aging. Furthermore, losartan, an Angiotensin II type 1 receptor inhibitor, was predicted to be activated. Inhibition of the renin‐angiotensin system increases survival and prevents age‐associated diseases in mice (de Cavanagh et al. 2011). Interestingly, losartan has already shown potential as a therapy for vascular aging in the clinic as it reduces aortic stiffness in patients with age‐related hypertension (Zhang et al. 2018). UR analysis also showed predicted activation of two peroxisome proliferator‐activated receptor (PPAR) agonists: pirinixic acid and troglitazone. Through their anti‐inflammatory, anti‐atherogenic, and antioxidant effects, activation of PPARs protects against the vascular complications of hypertension, atherosclerosis, and myocardial infarction in animal models of metabolic diseases (Cheang et al. 2015). Thus, UR analysis identified compounds that may mimic the therapeutic effect of DR in the Ercc1^Δ/− aorta, as suggested by previous studies in animal models and patients. Notably, the analysis also predicted activation of the MAC2 inhibitor TD139, which is relevant given that MAC2 is a marker for the macrophage‐like VSMC phenotype, and our results suggested that DR reversed the switch to this phenotype in the Ercc1^Δ/− aorta. Thus, this UR analysis identified several potential pharmacological alternatives to DR that could mitigate vascular aging.

The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE294678↗ (Edgar 2002).