The activator protein-1 complex governs a vascular degenerative transcriptional programme in smooth muscle cells to trigger aortic dissection and rupture

Abstract

Background and Aims

Stanford type A aortic dissection (AD) is a degenerative aortic remodelling disease marked by an exceedingly high mortality without effective pharmacologic therapies. Smooth muscle cells (SMCs) lining tunica media adopt a range of states, and their transformation from contractile to synthetic phenotypes fundamentally triggers AD. However, the underlying pathomechanisms governing this population shift and subsequent AD, particularly at distinct disease temporal stages, remain elusive.

Methods

Ascending aortas from nine patients undergoing ascending aorta replacement and five individuals undergoing heart transplantation were subjected to single-cell RNA sequencing. The pathogenic targets governing the phenotypic switch of SMCs were identified by trajectory inference, functional scoring, single-cell regulatory network inference and clustering, regulon, and interactome analyses and confirmed using human ascending aortas, primary SMCs, and a β-aminopropionitrile monofumarate–induced AD model.

Results

The transcriptional profiles of 93 397 cells revealed a dynamic temporal-specific phenotypic transition and marked elevation of the activator protein-1 (AP-1) complex, actively enabling synthetic SMC expansion. Mechanistically, tumour necrosis factor signalling enhanced AP-1 transcriptional activity by dampening mitochondrial oxidative phosphorylation (OXPHOS). Targeting this axis with the OXPHOS enhancer coenzyme Q10 or AP-1-specific inhibitor T-5224 impedes phenotypic transition and aortic degeneration while improving survival by 42.88% (58.3%–83.3% for coenzyme Q10 treatment), 150.15% (33.3%–83.3% for 2-week T-5224), and 175.38% (33.3%–91.7% for 3-week T-5224) in the β-aminopropionitrile monofumarate–induced AD model.

Conclusions

This cross-sectional compendium of cellular atlas of human ascending aortas during AD progression provides previously unappreciated insights into a transcriptional programme permitting aortic degeneration, highlighting a translational proof of concept for an anti-remodelling intervention as an attractive strategy to manage temporal-specific AD by modulating the tumour necrosis factor–OXPHOS–AP-1 axis.

Structured Graphical Abstract

The TNF–OXPHOS–AP-1 axis serves as a potential therapeutic target for aortic dissection. OXPHOS, oxidative phosphorylation; AP-1, activator protein-1; TNF, tumour necrosis factor; SMC, smooth muscle cell; AD, aortic dissection; BAPN, β-aminopropionitrile monofumarate.

Open in new tabDownload slide

See the editorial comment for this article ‘On target inhibition of vascular smooth muscle cell phenotypic transition underpins TNF–OXPHOS–AP–1 as a promising avenue for anti–remodelling interventions in aortic dissection and rupture’, by K. Miteva, https://doi.org/10.1093/eurheartj/ehad679.

Introduction

Aortic dissection (AD) is a vascular remodelling disease marked by an exceedingly high risk of mortality.^1,2 On the basis of anatomic patterns, dissections are classified as well-known Stanford type A and type B dissection.^3,4 The key pathological hallmark is aortic degeneration, manifested by smooth muscle cell (SMC) loss, extracellular matrix (ECM) fragmentation, and inflammation.⁵ These cellular alterations, potentiated by both environmental and genetic risk factors, contribute to the irreversible weakening, dissection, and rupture of the aortic tree, ultimately resulting in circulatory failure. Despite substantial advances in AD treatment in the last decade, the overall survival is still limited, and AD diagnosis is associated with extremely poor prognosis.^6,7 Although the optimal interventions predominantly employ surgical repair, there is still need for other therapies, including non-invasive types.⁸ The lack of effective therapies underscores the urgent need for a more complete understanding of the cellular properties of dissected human aorta as a first step towards development of suitable anti-AD therapies.

Smooth muscle cells are the predominant cell type lining the tunica media and play a crucial role in maintaining aortic morphology, integrity, and elasticity.⁹ Unlike cardiomyocytes and skeletal muscles, SMCs are not terminally differentiated and adopt a wide range of phenotypic states.¹⁰ By reversibly switching from a contractile to a synthetic phenotype in response to diverse external stimuli, SMCs exhibit loss of contractile proteins, accumulation and degradation of the ECM, and increased inflammation, all of which contribute to aortic remodelling.^11,12 In dissected aorta, expression of synthetic markers is increased with a concomitant decrease of contractile genes, suggesting an alteration in SMC phenotype during AD pathogenesis.^13–16 Manipulation of the SMC phenotypic switch has been reported to rewire the development of multiple aortic diseases.^17,18 However, the precise SMC subtypes, their functional features, the transition trajectories, and the mechanisms leading to pathophysiologic features in AD are far from clear.^19–21 This might be attributable to the high heterogeneity and functional diversity of SMC populations, key factors that are often masked by bulk genetic analysis. Fortunately, single-cell RNA sequencing (scRNA-seq) can robustly reveal cell heterogeneity and diverse cellular landscapes of human aortic SMCs (aoSMCs) and identify potential cell type–specific mechanisms of phenotypic transition.

Dissections undergo temporal evolution and can be divided into acute (within 2 weeks), subacute (from 2 weeks to 3 months), and chronic (over 3 months) stages based on the time interval between the onset of initial symptoms and presentations.^5,6 Acute Stanford type A AD (STAAD) can be rapidly fatal. Patients with delayed diagnosis of acute AD experience natural selection, surviving the acute phase, and ultimately enter a chronic course.²² Therefore, the distinct temporal patterns of dissections reflect a natural history of disease transition over time. Clinical studies have indicated that compared with acute AD, chronic forms differ in a number of ways in terms of aortic expansion rate, organ malperfusion, and disease outcomes.^23,24 Additionally, histological evaluations have revealed gradual changes in elastin fragmentation and medionecrosis over time, suggesting a tight association between the SMC phenotypes and temporal AD.²⁵ Therefore, deciphering the specific patterns of phenotypic transition across the different forms of AD would profoundly improve our understanding of the temporal evolution of the disease.

Here, we constructed a cross-sectional single-cell transcriptomic atlas of the human dissected ascending aorta among distinct temporal stages of AD and identified a vascular degenerative transcriptional programme that triggers SMC phenotypic transition during pathogenesis. Specifically, this study uncovered a tumour necrosis factor (TNF)–oxidative phosphorylation (OXPHOS)–activator protein-1 (AP-1) axis that triggers the phenotypic defects in dissected aorta. Consequently, enhancing OXPHOS or inhibiting AP-1 ameliorated AD development while increasing survival in a relevant AD mouse model. Thus, this work provides a comprehensive understanding of degenerative aortic remodelling, paving the way for potential therapies against human AD.

Methods

The detailed methods have been described in the online supplementary material.

Results

Single-cell RNA sequencing analysis identifies synthetic smooth muscle cells as an emerging cell type in human dissected aorta

To identify the cell types and pathologic mechanisms driving the phenotypic switch of SMCs during temporal AD, and to delineate the population shifts in a disease stage-specific manner, we performed scRNA-seq on ascending aorta from nine patients with STAAD (four with acute AD, three with subacute AD, and two with chronic AD) undergoing ascending aorta replacement and five control individuals undergoing heart transplantation (three donors and two recipients; Figure 1A and Table 1). In total, 28 libraries were generated, and 93 397 high-quality cells passed stringent quality control, reaching a median depth of 58 928 reads/cell and 1966 genes/cell, with expression of the housekeeping gene ACTB and the percentage of mitochondrial genes being comparable across samples (see Supplementary data online, Figure S1A and B and Excel Files S1 and S2).

After combining data of all cells, we performed integrative unsupervised clustering using Seurat.²⁶ Global transcriptomic profiling showed no differences in aortic cell type distribution among individuals, indicating that these cell types reflected physiological differences between cells rather than the effect of genetic backgrounds (see Supplementary data online, Figure S1C). Global, disease-, or vascular-specific distributions of cell clusters were represented by uniform manifold approximation and projection (UMAP, Figure 1B, Supplementary data online, Figure S1C–E). The cells from aortic walls were grouped into 12 major clusters (six non-immune and six immune). Guided by canonical cell type–specific markers, we were able to define endothelial cells (ECs), SMCs, and fibroblasts from different tunica layers. In addition to classical vascular cells, we obtained clusters corresponding to synthetic SMCs, mesenchymal cells, proliferative cells, myeloid cells, fibro-like macrophages, T cells, B cells, mast cells, and NK cells (Figure 1B and C, Supplementary data online, Figure S1F–H). Annotating the biological functions of each cluster by Gene Ontology (GO) analysis of differentially expressed genes (DEGs) further verified the unique characteristics of these aortic cells (Figure 1C). The consistency of both sexes and ages from all participants was evaluated by cell distribution pattern and correlation analysis of transcriptomes (see Supplementary data online, Figure S2A–F). To identify potential population shifts, we performed cell distribution comparisons by splitting the UMAP into control and AD groups. We identified synthetic SMCs as an emerging cluster in AD (Figure 1D). In addition, comparing the proportion of synthetic SMCs against all aortic cells further confirmed the expansion of this population in AD (Figure 1E, Supplementary data online, Figures S2G and H and S3 and Excel File S3). These results suggest that synthetic SMCs are an AD-specific emerging cluster.

The potential origin and characteristics of the emerging synthetic smooth muscle cells in human aortic dissection

The switch of SMCs from a contractile to a synthetic phenotype is the key event driving aortic degeneration and dissection.^5,27 To address the potential origins and pathologic features of the emerging synthetic SMCs, we performed reclustering of all non-immune cells and identified 17 subtypes (Figure 2A). Aside from ECs (EC-1 and EC-2), fibroblasts (fibroblasts 1–5), and a few small clusters corresponding to mesenchymal, inflammatory, and proliferative cells, we identified four clusters of SMCs, among which contractile SMCs were specifically enriched for classical contractile genes, including MYH11, ACTA2, MYL9, and TAGLN (Figure 2A and B, Supplementary data online, Figure S4).²⁷ Another three SMC clusters share the expression of contraction markers at relatively lower levels, with proliferating SMCs enriched for the cell cycle–related gene CCND1, while macrophage-like SMC-1 and SMC-2 preferentially expressed macrophage-related genes. Synthetic SMCs, enriched for the canonical synthetic markers COL1A1, COL1A2, and COL3A1, were partitioned into three clusters: fibro-like SMCs, lipo-SMCs, and fibromyocytes. Lipo-smooth muscle cells were specifically enriched for the lipid metabolic gene APOE.²⁸ Fibromyocytes were enriched for both fibroblast and contractile markers.²⁹ Therefore, the high enrichment of synthetic markers and the lower preservation of contractile genes in synthetic SMCs might reflect a disease-specific population shift from contraction to synthesis.

To understand the basis for such a phenotypic transition, we performed the following analysis. Firstly, by analysing cell proportions, we observed that contractile SMCs were dramatically lower in AD, with a concomitant expansion of synthetic SMCs (Figure 2C). Secondly, similarity analysis of the enrichment of gene signature (top 50 upregulated DEGs) showed that genes upregulated in synthetic SMCs mostly resembled signatures of SMCs among all vascular populations (Figure 2D). Thirdly, we applied the unsupervised inference method Monocle³⁰ to construct a potential developmental trajectory of individual SMC clusters to synthetic SMCs. We identified a clear directional trajectory flow, with contractile SMCs positioned at the opposite end of the other three synthetic clusters (Figure 2E), suggesting the potential contractile SMC origin of these synthetic populations. However, there was no clear trajectory from the other three SMC clusters to synthetic SMCs (see Supplementary data online, Figure S5). To better understand the trajectories, we defined scores of contraction and synthesis based on previously defined gene signatures and top expressed genes in each cluster (see Supplementary data online, Excel File S4).²⁷ Analysing the trajectory in the context of these functional scores revealed that Component 1 was negatively associated with contraction while positively associated with synthesis (Figure 2F). In addition, DEGs and pathway analysis further confirmed the enrichment of pathways involved in contraction in contractile SMCs, and inflammatory-, protein digestion-, and ECM-related pathways were enriched in synthetic SMCs (Figure 2G–I). Thus, the future fate of contractile SMCs appeared to switch to three distinct populations, all of which featured functional synthetic phenotypes.

Temporal stage-specific phenotypic transition in human aortic dissection

Dissections across temporal evolution show changing pathologic features.^23,24 We next delineated the phenotypic switch at distinct temporal stages. The expression of contraction score in contractile SMCs isolated from both the acute and the subacute groups was lower than those from SMCs from the chronic group. Conversely, the levels of synthetic scores in each synthetic cluster were higher in acute and subacute stages (Figure 3A and B). Furthermore, both functional scorings of the phenotypic switch and the relative cell number of synthetic SMCs compared with contractile SMCs were notably higher in acute and subacute AD than chronic AD, suggesting a disease temporal-specific phenotypic switch (Figure 3C and D).

Extracellular matrix remodelling and degradation are key synthetic features in AD.⁵ Therefore, we compared ECM and protease modules in each group (see Supplementary data online, Excel File S4). The fitting curve showed higher expression of both scores in the acute and subacute groups compared with the chronic group, suggesting disease temporal-specific ECM remodelling (Figure 3E and F). In addition, the preferential downregulation of contractile markers and upregulation of synthetic markers (protease, ECM, and inflammatory factors) in the acute and subacute groups were further confirmed using medial samples of the human ascending aorta (Figure 3G–I). All the analysis above supports a disease temporal-specific phenotypic switch in human AD.

The AP-1 transcriptional complex mediates the transition of contractile smooth muscle cells to both fibro-like smooth muscle cells and lipo-smooth muscle cells

To identify master regulators driving the phenotypic switch, we first screened DEGs belonging to transcription factors (TFs) by comparing gene expression in contractile SMCs against synthetic SMC clusters. By charting the levels of differentially expressed TFs, we found that multiple subunits of AP-1, including FOS, FOSB, JUN, JUNB, and JUND, were consistently elevated in fibro-like and lipo-SMCs compared with that in contractile SMCs (Figure 4A). Single-cell regulatory network inference and clustering (SCENIC) analysis indicated the AP-1 regulon was active in fibro-like and lipo-SMC clusters (Figure 4B). Moreover, AP-1 elevation along the developmental trajectory was observed in the acute and subacute AD groups (Figure 4C and D). We further found that hypertension, but not smoking or aortic valve regurgitation, contributes to AP-1 expression in synthetic SMCs in AD patients (see Supplementary data online, Figure S6A–C), consistent with the finding that hypertension is an important risk factor for AD.^5,8,31,32

The main AP-1 proteins in mammalian cells are FOS and JUN, and phosphorylation of their N-terminal domain crucially enhances stability and transactivation potential.³³ To determine whether AP-1 activation occurs in the medial layer, we dissected the tunica media of ascending aortas from control and AD (human) and β-aminopropionitrile monofumarate (BAPN)-induced AD (mouse) tissues and analysed them for FOS, JUN, and p-JUN expression or their mouse equivalents. We observed elevated AP-1 at both the mRNA and protein levels, with a concomitant greater degree of phosphorylation at Ser73 in aortas from both acute and subacute groups (Figure 4E and F). Notably, greater nuclear accumulation of p-JUN was observed in Col1a1⁺ but not α-SMA⁺ cells from dissected aorta (Figure 4G, Supplementary data online, Figure S6D and E), suggesting the nuclear translocation and activation of AP-1 in synthetic SMCs of AD.

To further infer the role of AP-1 in the phenotypic switch, we performed correlation analysis, showing that AP-1 negatively correlated with contractile SMC score while positively correlated with fibro-like and lipo-SMC scores (see Supplementary data online, Figure S6F). This correlation was confirmed in tunica media samples from both human and mice (Figure 4H, Supplementary data online, Figure S6G and H). A further prediction of AP-1 target genes in three distinct TF databases yielded multiple fibro-like- and lipo-SMC-specific genes as targets of AP-1 (see Supplementary data online, Figure S6I). We next performed chromatin immunoprecipitation (ChIP) analysis to examine whether AP-1 directly binds to promoters of selected synthetic genes. We observed the enrichment of p-JUN-specific antibody binding in putative sites of COL5A1, COL1A1, COL1A2, and ApoE (Figure 4I). We further verified the role of AP-1 in the phenotypic switch in primary aoSMCs. Treating cells with the AP-1 activator 12-O-tetradecanoylphorbol-13-acetate (TPA)^34,35 dramatically upregulated synthetic SMC genes, with a concomitant downregulation of contractile markers (Figure 4J). Conversely, treatment with the AP-1 inhibitor T-5224^36,37 showed the opposite effect (Figure 4K), suggesting that the elevation and activation of the AP-1 facilitate synthetic SMC expansion in AD.

A defect in oxidative phosphorylation acts upstream of activator protein-1 to promote synthetic smooth muscle cell expansion

We next explored upstream regulators promoting AP-1-mediated phenotypic switching. Differentially expressed genes and pathway analysis revealed enrichment of adenosine 5'-triphosphate (ATP) metabolic process and OXPHOS in contractile SMCs compared with fibro-like and lipo-SMCs (Figure 2G and H). Further analysis of six typical mitochondria-related features and OXPHOS module expression along developmental trajectories confirmed this enrichment, especially in acute and subacute AD (Figure 5A–D, Supplementary data online, Figure S7A–D), suggesting OXPHOS might negatively regulate synthetic SMC expansion in a disease subtype-specific manner.

To validate whether OXPHOS regulates AP-1 activation, we employed primary aoSMCs. We observed that treatment of the cells with the mitochondria complex 1–specific inhibitor rotenone actively elevated AP-1 expression, phosphorylation, and nuclear translocation (Figure 5E–I, Supplementary data online, Figure S7E). Conversely, the addition of coenzyme Q10 (CoQ10), an essential OXPHOS cofactor, showed the opposite effect, further confirming the requirement of compromised OXPHOS for AP-1 activation.

We next directly examined the involvement of OXPHOS in the AP-1-mediated phenotypic switch of SMCs. We performed ChIP in human aoSMCs and found that the binding of a p-JUN antibody to putative sites within promoter regions of selected synthetic genes was enhanced by rotenone while being diminished by CoQ10 (Figure 5J). Consistent with these findings, AP-1 inhibition blocked rotenone-induced expression of synthetic genes while it downregulated contractile marker expression (Figure 5K and L). Conversely, AP-1 activation showed the opposite effect (see Supplementary data online, Figure S7F). Together, these results indicate that defects in OXPHOS act upstream of AP-1 to preferentially promote synthetic SMC expansion (Figure 5M).

Tumour necrosis factor triggers oxidative phosphorylation–activator protein-1 axis to promote the phenotypic switch

Activator protein-1 is responsive to external mitogenic stimuli.³⁸ We, therefore, further explored extracellular signals triggering the OXPHOS–AP-1-mediated phenotypic switch. Gene Ontology analysis of AP-1 co-expressed genes indicated strong enrichment for multiple TNF signalling and ECM-related pathways in both fibro-like and lipo-SMCs compared with contractile SMCs (Figure 6A). Tumour necrosis factor pathway genes, including NFKB1A, IRF1, TNFAIP3, CCL2, CXCL2, and SOCS3, were co-expressed with AP-1 (Figure 6B). Moreover, correlation analysis showed that the TNF module was positively correlated with both the fibro-like + lipo-SMC score and the AP-1 score (Figure 6C), suggesting that TNF signalling might be potentially involved in the phenotypic switch of SMCs.

We next explored whether TNF signalling might differentially impact SMC transition in AD subgroups. Although we found higher levels of TNF-α in the aorta from chronic AD (see Supplementary data online, Figure S8A–D), contractile SMCs from either acute or subacute AD showed higher level of multiple TNF receptors (see Supplementary data online, Figure S8E). As myeloid cells and T cells are the major immune clusters (Figure 1B, Supplementary data online, Figure S9 and Excel Files S6 and S7) and TNF sources (see Supplementary data online, Figure S8F and G) in AD pathogenesis,^39–41 the contradictory expression pattern between TNF-α and TNF receptors further prompted us to figure out how TNF ligands might differentially regulate TNF signalling in contractile SMCs from distinct temporal AD. By analysing the TNF ligand–receptor interactions of immune populations (expressing TNF ligands) with contractile SMCs (expressing TNF receptors) using CellPhoneDB, we found a preferential interaction of TNF signalling in both acute and subacute AD (Figure 6D and E).

We next explored the potential role of TNF on OXPHOS–AP-1 axis. We found that treatment of aoSMCs with TNF-α reduced ATP levels, membrane potential, and nicotinamide adenine dinucleotide (NAD⁺)/nicotinamide adenine dinucleotide hydrogen (NADH), while CoQ10 restored these effects (Figure 6F). These results suggest that TNF might act upstream of the OXPHOS–AP-1 axis. To test that notion directly, we treated aoSMCs with TNF-α and found that the expression of AP-1, as well as its phosphorylation and nuclear translocation, was increased, while this effect was abrogated upon CoQ10 treatment (Figure 6G–I). Consistently, TNF-α-mediated downregulation of contractile genes and elevation of synthetic markers were inhibited by co-treatment with either CoQ10 or the AP-1 inhibitor (Figure 6J). These results indicate that TNF signalling dampens OXPHOS and subsequent AP-1 activation to promote synthetic SMC expansion (Figure 6K).

Rewiring of the oxidative phosphorylation–activator protein-1 axis modulates β-aminopropionitrile monofumarate–induced thoracic aortic dissection in mice

As the OXPHOS–AP-1 axis preferentially promotes the phenotypic transition of SMCs in thoracic aorta, we further determined the role of this axis in thoracic aortic dissection (TAD) progression in vivo by employing a classical BAPN-induced TAD model in mice.^39,42,43 We treated BAPN-induced mice with CoQ10 or TPA (Figure 7A). We found that CoQ10 dramatically reduced the percentage of lethal AD compared with BAPN group (Figure 7B and C). Consistently, CoQ10 significantly reduced the maximal diameter of the aortic ring after 28 days of BAPN treatment as measured by ultrasonography (Figure 7D and E). Further histologic analysis confirmed the reduction of dissecting aneurysm formation and medial degeneration, featured by elastic fibre fragmentation and disarray, excessive collagen deposition, and glycosaminoglycan accumulation, upon CoQ10 treatment (Figure 7F and G). Moreover, CoQ10 elevated contractile markers while downregulating synthetic markers in thoracic aorta (Figure 7H–K, Supplementary data online, Figures S10 and S11), suggesting that enhancing OXPHOS suppressed the phenotypic switch of contractile SMCs. Conversely, treatment of TPA accelerated BAPN-induced death, promoted dissecting aneurysm formation, medial degeneration, and the expansion of synthetic SMCs (Figure 7B–K, Supplementary data online, Figure S10). These data confirm that rewiring of the OXPHOS–AP-1 axis modulates BAPN-induced TAD in mice by regulating the phenotypic transition of SMCs.

Targeting activator protein-1 alleviates β-aminopropionitrile monofumarate–induced thoracic aortic dissection in mice by suppressing the phenotypic switch of smooth muscle cells

Because AP-1 contributes to synthetic SMC expansion, and targeting AP-1 suppressed the SMC phenotypic switch in vitro, we hypothesized that blockade of AP-1 might effectively alleviate TAD progression in vivo. To test this scenario, C57BL/6J mice were therapeutically treated with T-5224 for 2 weeks (starting from Day 14 of BAPN administration) or 3 weeks (starting from Day 7 of BAPN administration; Figure 8A). Administration of mice with BAPN for 28 days results in 67% lethal AD in mice (Figure 8B and C). In contrast, both 2-week and 3-week treatments of T-5224 significantly reduced lethal AD (to 25% and 17%, respectively). Moreover, we noted an increasing trend in both the survival curve and AD incidence by comparing 3-week to 2-week treatment (Figure 8B and C). In line with this observation, ultrasonographic measurement of the aortic ring on Day 28 showed decreased maximal aortic diameter upon inhibitor treatment (Figure 8D and E). The histologic analysis further confirmed the reduction of medial degeneration upon T-5224 treatment (Figure 8F and G). Additionally, T-5224 notably reduced the expression of synthetic markers while elevating that of contractile markers in thoracic aortic tissues (Figure 8H–K, Supplementary data online, Figures S12 and S13), suggesting suppression of the phenotypic switch of aoSMCs after AP-1 blockade. Together, these results suggest that targeting AP-1 represents a feasible strategy to treat TAD by attenuating the phenotypic switch of SMCs.

Discussion

Current efforts to understand AD pathogenesis have primarily focused on vascular cell–mediated degenerative responses.⁵ In this regard, the phenotypic switch of aoSMCs is a central event in degenerative aortic remodelling.⁴⁴ However, given the heterogeneity and functional diversity of aortic cells and emerging evidence that dissections manifest distinct histopathologies during temporal transition,²⁵ the precise cell types pertaining to population shifts during disease evolution and the mechanisms governing the switch to fuel AD progression remain largely unknown. Here, we leveraged the advantages of deep scRNA-seq to quantitatively delineate the phenotypic transition of aoSMCs in a disease temporal-specific manner, which provides four noteworthy contributions. Firstly, we constructed a full repertoire of human ascending aortic cell lineages and elucidated their functional signatures. Secondly, it allowed for modelling of disease temporal-specific phenotypic transitions of aoSMCs. Thirdly, it uncovered a role for a TNF–OXPHOS–AP-1 axis in mediating the phenotypic switch. Fourthly, targeting this axis robustly reduces AD and rupture. Altogether, these observations identify potential targets for human AD (Structured Graphical Abstract).

Despite major progress in AD therapy within the last decade, there is still no cure for this devastating disease. At present, few effective therapeutic options are available other than surgical repair, which shows high perioperative mortality.⁴⁵ Existing standard pharmacologic treatments predominantly employ pain-relieving drugs coupled with vasodilators to control blood pressure.^46,47 Although these medications provide symptomatic relief, they only modestly improve survival given their inability to impede the phenotypic switch of SMCs to prevent aortic degeneration. As this phenotypic switch is tightly correlated with vascular degeneration, one possible solution to this challenge is to directly target the switch. This study provides evidence of a TNF–OXPHOS–AP-1 axis in governing the phenotypic switch of aoSMCs. Consistent with our findings, a report showed an elevation of multiple subunits of AP-1 in SMCs of a BAPN-induced AD model.³⁹ Additionally, one vascular population termed ‘stressed SMC’ was also enriched for FOS and JUN in human thoracic aortic aneurysm tissues.²⁷ These results suggest AP-1 upregulation in non-contractile SMCs might be a common feature of degenerative vascular remodelling. We further show that TNF-α stimulation, OXPHOS dysfunction, or AP-1 activation together promote this switch by enhancing the AP-1 complex. In contrast, boosting OXPHOS by CoQ10 or blocking AP-1 can inhibit this phenotypic transition. These results support a potential anti-remodelling approach for AD intervention by targeting the TNF–OXPHOS–AP-1 axis.

CoQ10 is a naturally occurring compound essential for cellular bioenergetics, a feature critical to its application to affect tissues with high metabolic demands, such as the heart and aorta. Therefore, it offers potential benefits in managing patients with cardiovascular diseases.⁴⁸ We observed that CoQ10 restored the contractile phenotype of aoSMCs, normalized aortic function, and improved survival in BAPN-induced AD in mice through rapidly improving mitochondrial function. Consistent with these findings, CoQ10 has emerged as an effective approach to boost mitochondrial dysfunction in different vasculopathies, including heart failure and hypertension, although it is not approved by the US FDA for the treatment of any medical condition.⁴⁸ Further consistent with our findings, another mitochondrial booster, use of a NAD precursor, effectively reverses synthetic signature of SMCs and aortic dilation in a mouse model of aortic aneurysm.⁴⁹ These studies indicate that mitochondrial dysfunction is a common pathological hallmark of aortic remodelling, and enhancement of mitochondrial function might have far-reaching physiological implications in a spectrum of vasculopathies associated with dysfunctional mitochondria.

We observed that targeting the OXPHOS–AP-1 axis robustly alleviated AD development in a BAPN-induced AD model in mice during a 2–3-week treatment period. These results provide a rationale for translating such a treatment to humans. However, human AD manifests a much longer disease course and more complicated temporal stages compared with mouse AD.²⁵ Therefore, the effectiveness and safety of this therapy, especially in the long term, must be verified before translating this approach for human AD. Although CoQ10 has been widely used in various cardiovascular diseases with an excellent safety record, even at high, chronic doses, the feasibility of using AP-1 inhibitory drugs and their potential long-term side effects are still lacking.^50,51 Further efforts to determine the long-term effects of this therapy, as well as understanding potential side effects, should lead to new therapeutic options for human AD.

We made several interesting discoveries regarding temporal-specific phenotypic transitions. Functional quantitative scoring of both contractile and synthetic scores revealed the phenotypic switch is most prominent in acute/subacute stages in response to dissection. Concomitantly, experimental validation of the expression pattern of contraction, ECM remodelling, and inflammation further reflects a disease stage–preferential occurrence of this population shift. Consistently, the TNF–OXPHOS–AP-1 axis that drives the phenotypic switch also showed a higher degree of regulation in both acute and subacute AD. Therefore, the disease stage–specific alterations of this axis fit the preferential pattern of population shifts and more severe clinical presentations and worse outcomes observed in acute and subacute STAAD.^23,24 Therefore, the temporal-specific patterns might further deepen our understanding of phenotypic transition for better design of precise anti-AD therapies.

There is a tight connection between inflammation and the onset and rupture of the aorta.^52–54 However, it is unclear how inflammation initiates vascular degeneration in terms of the phenotypic switch of aoSMCs. Here, we show that TNF, a potent pro-inflammatory factor, functions upstream of OXPHOS to trigger a stress-responsive transcriptional programme leading to synthetic SMC expansion. Likewise, multiple pathways have been proposed for TNF-mediated inhibition of OXPHOS as well as mitochondrial dysfunction–induced AP-1 activation at distinct cellular background.^55–57 Therefore, TNF–OXPHOS-mediated activation of this transcriptional complex establishes a pro-inflammatory microenvironment to permit vascular degeneration, further highlighting the importance of anti-inflammation therapy for AD.

Because of limited material availability, few studies concerning aortic cellular compositions have been reported for human AD. Thus, it has been particularly difficult to discover mechanisms underlying the shifts of specific SMC populations. One recent study used scRNA-seq to generate cell compositions of human aorta with AD.⁵⁸ However, because of the limited sample size and input of cell number, the study only generated 12 clusters and was restricted to fibroblast–SMC interactions. Another study used only five ascending aortas for scRNA-seq of immune cells, providing information on immune heterogeneity.⁴⁰ Here, we collected 14 human ascending aortas and deeply sequenced up to 93 397 single cells. This enabled us to successfully map a transcriptomic atlas of the human ascending aorta, providing high-quality data to uncover in-depth temporal AD-related alterations at a single-cell resolution. Notably, by comparing major aortic cells between ours and that in aortic aneurysm,²⁷ we identified an additional cluster ‘proliferation’. As other studies have reported,^59,60 this cluster specifically expressed proliferating and cell cycle genes, with enrichment of nuclear division, organelle fission, and DNA replication. The potential factors attributing to this discrepancy might be due to distinct aortic diseases (AD vs. aneurysm) or the number of individual cells analysed (93 397 vs. 48 128). As both studies used human ascending aorta, all other major cell types were similarly identified.

Here, the single-cell data were generated using aortas from 14 subjects. The low number of samples in each AD subgroup might prevent a full assessment of changes in gene expression and identification of rare populations. Indeed, the low cell number in some aortic subsets was due to a small sample size, which might reduce the robustness of these data. In addition to ethnicity and race that were kept consistent across all samples, two control samples were from heart transplant recipients manifesting heart disease, which might introduce cellular and molecular alterations in the aorta. Although the participants differ in terms of sex (2 females vs. 12 males) and age, we found no apparent bias in cellular distribution and global gene expression patterns between different sexes or ages, suggesting the reliability of control samples.

The aortic wall is a compact tissue and hard to digest. We combined an enzyme cocktail with subsequent sorting to obtain high-quality viable single cells for scRNA-seq. Variations in tissue processing conditions (e.g. temperature and enzyme) have been reported to trigger stress pathways in cancer single-cell studies.⁶¹ Consistently, a longer processing time in aortic tissues also induced stress pathways, making it difficult to discriminate stress responses in disease groups.²⁷ Here, the tissue processing conditions, including time, temperature, and enzymes, for all aortas were strictly followed to avoid any influence or bias in comparison between AD and control tissues.

Conclusions

In conclusion, the previously unappreciated mechanistic insights that guide the phenotypic switch of SMCs to fuel AD pathogenesis establish new avenues for developing pharmacologic anti-remodelling interventions to protect against dilation, dissection, or even rupture of aortic wall.

Acknowledgements

We thank Analytical BioSciences for performing scRNA-seq and data analysis, Beijing Fuyu Biotechnology Co., Ltd, for bulk transcriptome, and Dr Dehua Liu from Xiang An Hospital in assisting the collection of aortic samples.

Supplementary data

Supplementary data are available at European Heart Journal online.

Declarations

Disclosure of Interest

All authors declare no disclosure of interest for this contribution.

Data Availability

The scRNA-seq and bulk RNA-seq data are available at GEO:GSE222318 and GEO:GSE222382, respectively.

Funding

This work was supported by the Beijing Advanced Innovation Center for Food Nutrition and Human Health, the Beijing Advanced Innovation Center for Big Data-Based Precision Medicine Capital Medical University, the Beijing Lab for Cardiovascular Precision Medicine Capital Medical University (PXM2020_014226_000054) to F.R. and H.Z., the National Natural Science Foundation of China (31970717, 82170487, 82170429, 81800404, and 32371229) to Y.L., W.J. and P.A., the Beijing Municipal Natural Science Foundation (7222111) to Y.L., the National High Level Hospital Clinical Research Funding (bj-2023-72) to P.A., and the Pinduoduo-China Agricultural University Research Fund (PC2023B01014) to Y.L.

Ethical Approval

All animal experiments were performed in compliance with the guidelines for the care and use of laboratory animals and were approved by the ethics committee of the China Agricultural University. The protocol for collecting human aortic tissue samples was approved by the Ethics Committee of Beijing Anzhen Hospital. All experiments involving human aortic tissue samples were performed in accordance with the guidelines approved by the committee. Informed consent was obtained from all participants or donor/recipient families.

Pre-registered Clinical Trial Number

None supplied.

References

Author notes