2.1 Human tissue study

Aortic tissue from the patient was obtained from the Department of Cardiovascular Surgery at Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China. Written informed consent was obtained from all patients and organ donors in compliance with the Declaration of Helsinki. Details of this section are provided in Supporting Information.

2.2 Animal studies

Mice were housed in a pathogen-free facility at Tongji Medical College, Huazhong University of Science and Technology. A mouse model for sporadic AAD was developed by using ATF3 VSMC-specific conditional knockout mice (C57BL/6J-Atf3 fl/fl, Tagln Cre⁺) and their littermate controls (C57BL/6J-Atf3 fl/fl, Tagln Cre⁻) which were constructed by Saiye Inc. (Suzhou, China). Additionally, ATF3 VSMC-specific conditional knockout mice combined with Tmem173 knockdown (C57BL/6J-Atf3 fl/fl, Tmem173 fl/+, Tagln Cre⁺) were generated by Model Organisms (Shanghai, China). Comprehensive procedural details can be found in Supporting Information.

In the angiotensin II (Ang II) infusion assay, mice received subcutaneous infusions of angiotensin II or saline via Alzet osmotic minipumps, which were implanted under 2% isoflurane anaesthesia.

In the porcine pancreatic elastase (PPE)-induced AAA model, mice were anaesthetised with an intraperitoneal injection of 1% pentobarbital sodium and positioned supine on the control panel.

At the end of the experiments, animals were euthanised using CO₂ or 2% isoflurane anaesthesia.

2.3 Statistical analysis

Representative figures and images are used to illustrate typical outcomes of each experiment. The Shapiro–Wilk test for normality and the Brown–Forsythe test for equality of group variance were performed on all data using Prism 9 software (GraphPad Software Inc., La Jolla, CA). Differences between two groups were evaluated using an unpaired, two-tailed Student's t-test for normally distributed data or a Mann–Whitney test for non-normally distributed data. One-way ANOVA was used to assess differences across groups for a single independent variable in multiple comparisons, while a two-way ANOVA was used to examine differences across groups for two independent variables. p Values for pairwise comparisons were adjusted using the Bonferroni method when indicated. The figure legends specify the statistical tests used. Each experiment was conducted a minimum of three times. All data points represent independent samples, not technical replicates. Data are expressed as mean ± standard deviation.

3.1 ATF3 expression is elevated in both a mouse model of sporadic AAD and in human sporadic ATAAD tissues

We found that there was no significant immune cell infiltration in the aorta after 7 days of Ang II infusion, but there was a large amount of immune cell infiltration after 14 days (Figure S1). To assess the dynamic changes in aortic VSMCs during the early stages of AAD without the influence of immune cells, we performed single-cell transcriptomic analysis on aortic tissues from C57 mice treated with either saline or Ang II (2000 ng/kg/min) for 7 days. Nineteen clusters were identified and categorised into VSMCs, endothelial cells, fibroblasts, mono/macrophages, T cells, B lymphocytes and neurons based on specific marker gene expression (Figures S2A and 1A). We extracted VSMC clusters for further analysis. We characterised VSMC subclusters by identifying differentially expressed genes (DEGs) for each cluster and conducting Gene Ontology (GO) analysis on these cluster-specific DEGs to determine overrepresented functions (Figures S3, S4 and 1B). Clusters 4, 3 and 6 were identified as primary contractile VSMC clusters, demonstrating enhanced functions in contraction and extracellular matrix (ECM) organisation. Moreover, they were abundant in saline-infused mice, indicating that they are unchallenged VSMCs. Cluster 9 was defined as ECM organisation VSMCs, which exhibited functions related to ECM organisation. Cluster 7 was characterised by contractile and ECM organisation attributes akin to clusters 3, 4 and 6, while also displaying stress-related gene features such as Nr4a1, Atf3, Klf2 and Egr1. Cluster 5 was defined as contractile/stressed/migration VSMCs, which showed enriched functions in response to mechanical stress, smooth muscle cell contraction and migration. Cluster 8 was defined as contractile/metabolically reprogrammed VSMCs, which exhibited features of contraction and increased glucose intake. Cluster 10 was characterised by inflammatory and pro-apoptotic VSMCs, showing enhanced activity in inflammation and cell death processes. Cluster 1 was defined as stressed/proliferative VSMCs, which exhibited features of stress-related genes and proliferative functions. Cluster 2 was defined as stressed/biogenesis VSMCs, which exhibited features of stress-related genes and biogenesis functions.

Pseudotime analysis using Monocle2 revealed two directions of VSMC transition: from contractile VSMC4 to inflammatory/pro-death VSMC10 and from contractile VSMC4 to stressed/proliferative VSMC1 (Figures 1C,D and S5A,B). The VSMC10 cluster progressively down-regulated contractile gene expression and up-regulated genes associated with inflammation and apoptosis, suggesting it represents the terminal stage of VSMC differentiation following Ang II exposure. Additionally, we performed branch-dependent expression analysis to identify DEGs driving VSMC differentiation (Figure S5A). The context-specific regulation of gene expression is mainly achieved through TFs. Further analysis revealed that four differentially expressed TF genes, Atf3, Tcf15, Nr4a1 and Cebpb, overlapped between node1 DEGs and the mouse TF library. Analysis of the mRNA levels of these four TFs after H₂O₂ stimulation in vitro showed that ATF3 exhibited the most significant changes (Figure 1E). Additionally, the mRNA level of ATF3 was significantly up-regulated in the Ang II-challenged Apoe−/− or C57 mouse aortas, especially in VSMCs (Figures 1F and S2B,C). Importantly, among all VSMC clusters, VSMC10 showed the highest expression of ATF3 (Figure 1G). Increased ATF3 levels significantly decreased in cells transitioning to VSMC1 but remained elevated in those transitioning to VSMC10, indicating ATF3 as a potential driver influencing VSMC fate (Figure S5A). To avoid the interference of immune cells, we selected adjacent tissue from the AAD lesion for detection. Consistent with the transcriptomic data, ATF3 protein levels were significantly increased in both the sporadic AAD mouse model and human sporadic AAD tissues (Figure 1H,I). In primary mouse VSMCs cultured in vitro, ATF3 was highly induced after stimulation with Ang II or H₂O₂. Its expression peaked at 6 h, was maintained for 2 h and then down-regulated (Figure 1I). These findings indicate that ATF3 is significantly up-regulated in AAD and may contribute to arterial remodelling.

3.2 VSMC-specific ATF3 knockout aggravates AAD development

We created mice with a conditional ATF3 knockout in VSMCs (Atf3 cKO) to study the role of ATF3 in smooth muscle lesion formation. These mice underwent high-dose Ang II infusion (2000 ng/min/kg) without being fed a high-fat diet. Blood pressure was similarly increased in both the control and Atf3 cKO groups (Figure S6A). However, Atf3 cKO mice exhibited significant aortic degeneration and diameter enlargement upon challenge (Figure 2A–C). In addition, 25% of Atf3 cKO mice died (Figure 2D). Atf3 cKO mice exposed to Ang II showed an increased incidence of AAD (55%, p = .0001, encompassing both AAD) and rupture (25%, p = .0471), especially in the ascending aorta, suprarenal aorta and renal ostia (Figure 2E). Interestingly, without any intervention, five out of 20 aged Atf3 cKO mice (18 months old) exhibited spontaneous AAD formation (Figure 2F,G).

To further validate ATF3's involvement in AAD progression, we assessed aortic diameter differences between Atf3 cKO mice and their control littermates using a PPE-induced abdominal aortic aneurysm (AAA) model, in which PPE degrades elastic fibres, causing aortic damage and aneurysm development. Our study demonstrated that the application of PPE to the infrarenal aorta resulted in significant aortic enlargement in Atf3 cKO mice compared with the littermate control group (Figure S6B). Results from repeated modelling (Figure S7) and modelling in female Atf3 cKO mice (Figure S8) also supported the previous conclusions. The absence of Atf3 significantly exacerbated AAD development, and no gender differences were observed in the role of ATF3 in AAD.

3.3 ATF3 deficiency aggravates aortic degeneration, VSMC phenotypic switch and apoptosis

Histologic analysis revealed that the aortas from Ang II-challenged Atf3 cKO mice exhibited damaged aortic architecture, including a reduction in the VSMC layer and increased elastic fibre fragmentation, compared with aortas from littermate control mice (Figure 3A). In aged mice, Atf3 cKO mice showed thinner elastic lamellae and more severe elastic fibre fragmentation (Figure 3B). In Atf3 cKO mice, aortic challenge led to an increased number of  terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL)-positive VSMCs in the aortic lesion and decreased expression of contractile proteins (Figure 3C,D). Consistently, in vitro experiments showed that knocking down ATF3 led to a significant increase in apoptosis when exposed to oxidative stress (Figure 3E,F). Western blot analysis of aortas from Atf3 cKO mice revealed significantly elevated levels of cleaved Caspase-3 and cleaved poly ADP- ribose polymerase (PARP)-1 (Figure 3G). In vitro experiments further demonstrated that silencing ATF3 in VSMCs resulted in decreased expression of VSMC contractile genes and increased levels of cleaved Caspase-3 and cleaved PARP-1 (Figure 3H). In vivo experiments revealed that ATF3 deficiency notably elevated osteopontin(OPN) expression in VSMCs following Ang II exposure, whereas in vitro experiments showed no significant change (Figure S9A–C). Meanwhile, the deficiency of ATF3 aggravated the production of matrix metalloproteinase (MMP)-9, whereas MMP-2 showed no significant difference in Ang II challenged VSMCs (Figure S9D). These results collectively suggest that the up-regulation of ATF3 during AAD development acts as a self-protective mechanism against arterial remodelling.

3.4 ATF3 negatively regulates the cGAS–STING pathway in VSMCs

We explored the mechanisms through which ATF3 regulates VSMC homeostasis. We performed bulk RNA sequencing on aortas from Atf3 cKO and littermate control mice 3 days post-Ang II infusion to examine the initial phase of the aortic stress response (Figure 4A). GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses revealed that aortic challenge in Atf3 cKO mice led to increased expression of genes associated with biological processes such as the inflammatory response, cytosolic DNA-sensing pathway and cellular response to type I interferon, indicating activation of the cGAS–STING pathway (Figure 4B–D). To avoid the influence of immune cells, descending aortas were collected for immunostaining of the cGAS–STING pathway. Our transcriptomic study showed that aortic challenge in Atf3 cKO mice led to a significant increase in the protein levels of phospho-interferon regulatory factor 3 (p-IRF3), p-STING and phospho-TANK- binding kinase 1 (p-TBK1) in the aortic wall (Figure 4E–G), and elevated interferon (IFN)-β levels in serum (Figure S10). Additionally, in vitro experiments demonstrated that after 6 h of 500 µM H₂O₂ stimulation, cultured primary mouse VSMCs with silenced Atf3 exhibited significantly higher levels of p-STING, p-IRF3 and p-TBK1 compared with controls (Figure 4H). Meanwhile, as a non-canonical downstream molecule of STING, p65 was more significantly activated in ATF3 knockout VSMCs compared with the control group (Figure S11A). Overexpression of ATF3 suppressed the cGAS–STING pathway (Figures 5A–C and S12). Moreover, ATF3 also directly regulated STING mRNA levels (Figure 5D,E). Taken together, these results support the notion that ATF3 negatively regulates the cGAS–STING pathway in VSMCs.

3.5 Loss of ATF3 leads to severe DNA damage and increased micronuclei formation

To further investigate the mechanism through which Atf3 knockout promotes cGAS–STING pathway activation, cultured primary VSMCs with or without Atf3 knockdown were subjected to bulk RNA sequencing after 6 h of stimulation with 500 µM H₂O₂. GO enrichment analysis revealed the activation of DNA double-strand break repair-related pathways and cell cycle phase transition pathways in Atf3 knockdown cells (Figure 6A). Consistent with the sequencing results, the level of γ-H2AX was up-regulated in aortic tissue of challenged Atf3 cKO mice as well as in H₂O₂-treated VSMCs with ATF3 knockdown, indicating that Atf3 insufficiency aggravates genomic DNA damage (Figure 6B,C). In addition, these results were further demonstrated through immunofluorescence staining in ATF3-knockout primary VSMCs (Figure 6D). Meanwhile, Atf3 insufficiency facilitated the production of micronuclei in challenged VSMCs (Figure 6E). Furthermore, confocal immunofluorescence showed that cGAS co-localised with micronuclei in challenged ATF3-knockout primary VSMCs (Figure 6F), revealing increased recruitment of cGAS to micronuclei. These findings suggest that ATF3 plays a critical role in maintaining genomic DNA stability.

3.6 ATF3 maintains the stability of P21

Next, the effects of ATF3 on DNA double-strand break repair-related pathways and cell cycle phase transition pathways, particularly the G1/S phase transition, were investigated (Figure 6A). P21, the key regulator of the transition from G1 phase to S phase, was examined after ATF3 knockdown in cultured primary VSMCs. ATF3 knockdown significantly down-regulated P21 (Figure 7A). Similarly, aortic tissues from challenged Atf3 cKO mice showed decreased P21 levels (Figure 7B). Flow cytometric analysis indicated that ATF3 deficiency eliminated G1 phase cell cycle arrest, promoting progression to the S and G2/M phases (Figure 7C). To further evaluate the effects of ATF3 on P21 expression, both loss- and gain-of-function approaches were carried out in P53-null H1299 cells. Importantly, the expression pattern of P21 aligned with that of ATF3, indicating that ATF3 regulates P21 in a P53-independent manner (Figure 7D,E). These effects are unlikely to be attributed to the inhibition of P21 transcription, as ATF3 siRNA did not reduce P21 mRNA levels (Figure 7F), suggesting post-transcriptional regulation of P21. Consistent with our hypothesis, ATF3 was found to increase the half-life of endogenous P21 protein, while the loss of ATF3 led to increased degradation of P21 (Figure 7G). In summary, the findings indicate that ATF3 enhances P21 protein stability.

We then investigated whether ATF3 binds to P21 and affects its stability. Reciprocal co-immunoprecipitation (co-IP) experiments indicated that ATF3 and P21 were absent from the same immune complex (Figure 7H), implying that ATF3 does not directly interact with P21. We hypothesised that ATF3's effect on P21 degradation may be mediated by other proteins. We utilised IP-MS to identify proteins interacting with ATF3 and P21 to test this hypothesis. Among these overlapping proteins, we identified MDM2 as a potential intermediary molecule (Figure 7I). Additional co-IP experiments demonstrated that both ATF3 and P21 could bind to murine double minute 2 (MDM2) (Figure 7J,K). Furthermore, ATF3 was found to reduce the binding of MDM2 to P21, as well as decreasing the ubiquitination of P21 by MDM2 (Figure 7L,M). The same conclusion was validated in VSMCs after H₂O₂ stimulation (Figure S13). Knockdown of MDM2 in ATF3-knockout primary VSMCs elevated P21 protein levels, reduced DNA damage and decreased cGAS–STING pathway activation, suggesting MDM2's role in enhancing P21 degradation, promoting DNA damage and activating the cGAS–STING pathway in the absence of ATF3 (Figure S14). These findings suggest that ATF3 may stabilise P21 by reducing MDM2-mediated ubiquitination.

3.7 Roscovitine treatment mitigates the adverse effects of ATF3 deficiency on VSMCs

After confirming the significance of the ATF3–P21–cGAS–STING axis in AAD development, we explored the potential of pharmacologically restoring P21 function to prevent aortic degeneration and AAD progression in mice. We investigated the impact of roscovitine, a reversible CDK2/4 inhibitor, on mitigating AAD development and progression in a mouse model of sporadic AAD. Atf3 cKO mice were challenged with Ang II infusion and also received either roscovitine (50 mg/kg) or corn oil (control) during Ang II infusion. Atf3 cKO mice treated with roscovitine exhibited superior aortic structure preservation, reduced aortic enlargement, significantly lower AAD incidence (13.3%, p = .0078) and rupture (0%, p = .0421), and enhanced survival rates compared with control-treated mice (Figure 8A–C). In Ang II-challenged Atf3 cKO mice, roscovitine treatment resulted in better preservation of elastic fibre architecture in the aortas compared with those without the treatment (Figure 8D). Furthermore, roscovitine treatment in Atf3 cKO mice led to decreased STING and IRF3 phosphorylation (Figure S16A,B) and a reduction in TUNEL-positive cells in the aortic wall (Figure 8E). Conversely, roscovitine treatment restored the reduced contractile proteins SM22-α and Calponin in the aortic wall of Atf3 cKO mice (Figure S16E,F).

In challenged VSMCs with ATF3 knockout, roscovitine restored the function of P21, leading to G1 phase arrest (Figure 8F). Meanwhile, roscovitine partially alleviated H₂O₂-induced VSMC apoptosis and γ-H2AX protein level elevation (Figure 8G,H). Moreover, roscovitine mitigated the formation of micronuclei and their co-localisation with cGAS (Figure 8I,J). Western blotting showed that roscovitine partially prevented the effects of ATF3 deficiency, including IRF3 and TBK1 phosphorylation, DNA damage, caspase-3 cleavage and the decrease in contractile proteins (SM22-α and Calponin) (Figure 8K).

3.8 STING knockdown mitigates the adverse effects of ATF3 loss on VSMCs

To further determine the effect of ATF3 deletion on VSMCs mediated by the cGAS–STING pathway, we generated VSMC-specific STING conditional knockout mice based on VSMC ATF3 cKO mice. We found that knocking out one of the STING alleles had a significant rescue effect. In ATF3 fl/fl; Tmem173 fl/+; Tagln-Cre⁺ mice, there was a significant reduction in AAD incidence (20%, p = .0253). Moreover, all mice survived until the end of the experiments (Figure 9A–C). Histological analysis showed that ATF3 fl/fl; Tmem173 fl/+; Tagln-Cre⁺ mice maintained normal aortic architecture, an intact SMC layer, and exhibited less elastic fibre fragmentation compared with Atf3 cKO mice (Figure 9D). Additionally, the number of TUNEL-positive VSMCs and the levels of p-STING and p-IRF3 in the aortic lesions were markedly down-regulated in ATF3 fl/fl; Tmem173 fl/+; Tagln-Cre⁺ mice (Figures 9E and S16C,D). Importantly, VSMC contractile proteins (SM22 and Calponin) were restored in ATF3 fl/fl; Tmem173 fl/+; Tagln-Cre⁺ mice (Figure S16G,H). Western blot analysis from in vitro experiments indicated that C-176 administration reduced cGAS–STING pathway activation and caspase-3 cleavage, while restoring contractile proteins in VSMCs with ATF3 knockdown (Figure 9F). Additionally, the aggravated VSMC apoptosis caused by ATF3 silencing was alleviated with the administration of C-176 (Figure 9G).

This study indicates that inhibiting the cGAS–STING pathway can partially alleviate the negative impact of ATF3 knockdown on VSMCs.