Single-cell RNAseq of Angiotensin II-induced abdominal aortic tissue identifies aneurysm-associated cell clusters in C57BL/6J mice Open Access

Aortic aneurysm (AA) ranks as the 13th leading cause of death in the United States [1,2]. It has been reported that abdominal aortic aneurysm (AAA) causes approximately 0.35% of all deaths in the United States, which is mainly in individuals older than 55 years old with the male/female ratio of 1.40 [3]. AAA is usually asymptomatic until a sudden rupture. Due to the challenge of obtaining pathological specimens of human patients, animal models are crucial tools for studying the pathogenesis and progression of diseases. Angiotensin II (Ang II) infusion in mice has become the most frequently used method to induce AAA to study the molecular mechanisms and potential therapeutic strategies [4,5]. At the very beginning, it was reported that Ang II can only induce AAA in apoE^-/- mice, presumably the presence of hyperlipidemia was required because infusion of Ang II into apoE^+/+ mice failed to generate aneurysm [6]. However, later studies have shown that after 6 or 10 days of Ang II infusion into 7- to 12-month-old C57BL/6 J mice at a dose of 2500 ng/kg/minute, aortic hematomas/dissections were observed in 35% (8 of 23) mice [7]. Although Ang II can induce AAA in normolipidemic mice at an older age, the incidence was significantly lower compared with that in apoE^-/- mice, which was 50% (8 of 16 mice) at the age of 12 weeks [8].

The incidence of AAA in C57BL/6 J mice treated with Ang II infusion varies widely among different studies [9,10]. Considering the low incidence of AAA in Ang II-treated C57BL/6 J mice, one study pointed out that a combination of Ang II and β-aminopropionitrile monofumarate (BAPN), an inhibitor of lysyl oxidase, can induce both thoracic and AAAs with a high incidence [11]. The loss of elastic fibers is an early event during aneurysm formation. The adventitious tissue, in which collagen and fibroblasts are rich, is responsible for the resistance of aneurysm in the absence of medial elastin [12]. BAPN can break the crosslinks of collagen fibers, leading to a high incidence of aneurysm formation.

The disruption of elastic fibers, degradation of collagen fibers, and destruction of vascular smooth muscle cells (SMCs) are hallmarks of AAA pathogenesis. Compelling evidence suggests that endothelial dysfunction is an early pathological episode during AAA formation, contributing to inflammation in the degenerating arterial wall [13]. However, the exact function of endothelial cells (ECs) in the development of AAA is largely unknown. Here, by using single-cell RNA sequencing (scRNAseq), we discovered a distinct cluster of heterogeneous EC subpopulations, referred to as malignant EC, showing elevated expression of genes in lipid metabolism, such as Cd36, Lpl, Gpihbp1, and Fabp4. We also found that ECs, SMCs, and fibroblasts were significantly decreased in the AAA group, whereas immune cells were notably increased. Furthermore, our data suggest that the Cxcl12–Cxcr4/Ackr3 axis functions in cell–cell interactions, which may promote inflammatory cell infiltration. These new findings reveal the key features in AAA formation, which are important for intervening AAA.

In this study, C57BL/6J male mice were used. Male mice aged 14, 20, or 32 weeks were selected for pump implantation. All mice were maintained under the specific pathogen-free facility of Institute of Genetics and Developmental Biology (IGDB), Chinese Academy of Sciences (CAS). ALZET osmotic mini-pumps (Model 1004, Durect Corporation, U.S.A.) were placed subcutaneously in mice to deliver Ang II at a concentration of 1.44 mg/kg/day or 2.16 mg/kg/day, as described previously [14,15]. Pumps filled with PBS were used as normal controls (NCs) for age-matched littermates of mice. Mice were anesthetized with 2% pentobarbital sodium (45 mg/kg), and pumps were implanted through a small incision in the back of the neck, then the incision was sutured with 5–0 suture thread. All incision sites healed within seven days. All experiments were conducted in accordance with the guidelines of the Animal Care and Use Committee of IGDB, CAS.

After 28 days of Ang II infusion, mice were anesthetized with 1% isoflurane. Their thorax and abdomen were depilated and coated with a developer solution. The Vevo 2100 Imaging System (FUJIFILM VisualSonics, Canada) was utilized, equipped with a 30 MHz transducer, to conduct ultrasound scans in B-mode. These scans included the longitudinal images of the ascending aorta, aortic arch, abdominal aorta, and measurements of internal diameter of the aorta. The branch of the cephalic trunk and approximately 1 cm above the branch of the coeliac trunk were chosen as the anatomical landmarks for measuring the diameter of aortic arch and abdominal aorta, respectively. AAA could be diagnosed when the ratio of the diameter of the suprarenal aorta and the non-dilated adjacent aorta is ≥1.5 [2].

Mice were euthanized following an intraperitoneal injection of excessive 2% pentobarbital sodium after monitoring aortic ultrasonography. The aorta was perfused through the heart with cold PBS, and segments of the aorta from the aortic root to the suprarenal abdominal aorta were isolated, with the surrounding connective tissue stripped off under a stereoscope. Mouse aortic tissue was fixed in 4% paraformaldehyde. The adventitial tissue was carefully removed, and the tissue was placed on a black gel plate for imaging of the aorta. Aortic tissue was embedded in paraffin and then sectioned. Aortic sections (5 μm) were stained with Masson’s trichrome, hematoxylin and eosin (H&E), and elastic staining reagents, respectively. Images were captured by the Nikon microscope Eclipse H550S (Nikon, Japan).

In order to confirm the cell cluster changes, the first antibodies used for immunofluorescence staining, including mouse anti-osteopontin (OPN, also named as SPP1) antibody (1:200, AF808, R&D system, U.S.A.), rabbit anti-CD68 antibody (1:100, 97778, Cell Signaling Technology, U.S.A.), mouse anti-lipoprotein lipase (LPL) antibody (1:200, ab93898, Abcam, U.K.), rabbit anti-GPIHBP1 antibody (1:20, NB110-41539, Novus Biologicals, U.S.A.), rabbit anti-CXCR4 antibody (1:100, T55380S, ABmart, China), and rabbit anti-ACKR3 antibody (1:100, A12712, ABclonal, China). The sections were fixed with 4% paraformaldehyde for 10 min and permeabilized with 1% Triton X-100 for 15 min. Then, the sections were blocked with 1% bovine serum albumin for 30 minutes and incubated with the aforementioned primary antibodies overnight at 4°C. After washing with PBS containing 0.1% (v/v) Tween 20 three times, sections were incubated with Alexa Fluor 488-conjugated goat anti-rabbit IgG (H + L) secondary antibody (A-32731, Invitrogen, U.S.A., 1:2,000) or Alexa Fluor 594-conjugated donkey anti-goat IgG (H + L) secondary antibody (A-32758, Invitrogen, U.S.A., 1:2,000) or Alexa Fluor 594-conjugated donkey anti-mouse IgG secondary antibody (A-21203, Invitrogen, U.S.A., 1:2000 dilution) or Alexa Fluor 594-conjugated goat anti-rabbit IgG (H + L) secondary antibody (A-11012, Invitrogen, U.S.A., 1:2000 dilution) for 60 min at 37°C in the dark. Finally, the sections were mounted with ProLong Gold Antifade reagent containing DAPI (Invitrogen, U.S.A.). Images were captured using the Zeiss LSM 880 Confocal Microscope (Zeiss, Germany).

We conducted scRNAseq on three groups of mice: 20-week-old mice receiving Ang II (1.44 mg/kg/day) or PBS infusion. The mice receiving Ang II were further divided into two groups (based on ultrasonography monitoring, with or without AAA). After conducting aortic ultrasonography monitoring, the mice were killed. The aortas were perfused through the heart with cold PBS, and the abdominal aortas were isolated and pooled from four mice in each group for single-cell suspension. The adventitial tissue was carefully removed, and the abdominal aortas were washed three times with cold PBS, then transferred to a 2 ml centrifuge tube containing 500 µl RPMI 1640 (11875093, Thermo Fisher Scientific, U.S.A.). The tissue was cut with scissors as small as possible. 1 ml RPMI 1640 containing 0.125% trypsin (25200056, Invitrogen, U.S.A.) was added to the tube, which was then placed on ice and gently mixed. The tubes were incubated at 37°C with slow rotation for 30 min. Following this, the tubes were centrifuged at 4°C, 300 g for 5 min, completely discard the supernatant, and 1.5 ml PB (BSA in PBS, final concentration is 0.5%) buffer was added to resuspend the tissue. The supernatant was discarded again as described above. 1.5 ml RPMI 1640 containing 1 mg/ml collagenase II (1148090, Sigma-Aldrich, Germany) was added into the centrifuge tube, and the tubes were rotated slowly at 37°C for 30 min. The samples were filtered with a 70 µM filter. The strainer and 2 ml centrifuge tube were washed with RPMI 1640, respectively, and the final volume was 10 ml. The samples were centrifuged at 4°C, 300 g for 5 min, then completely discarded the supernatant, and the cells were resuspended with 400 µl PB buffer. Acridine Orange (AO)/Propidium Iodide (PI) staining was conducted for quality inspection. After erythrocyte lysis, cell debris was removed.

Analyses were performed using R software (https://www.r-project.org/, version 3.6) and primarily utilizing the ‘Seurat’ packages. To process the data, cells were filtered. A total of 29,435 cells (before-treatment specimens) were retained based on the filtering criteria: 1) the number of genes detected per cell was >300 and <4000 and 2) the percent of mitochondrial genes was <0.10. Then, we selected the expression matrix of cells from pre-treatment samples as the input. Expression profiles were log-normalized and scaled, with the percentage of mitochondrial genes regressed out. Clusters were determined using the first 15 principal components and visualized through the uniform manifold approximation and projection (UMAP) dimensional reduction method for each sample. The identified cell clusters could be readily assigned to known cell lineages through conventional marker genes. The cell–cell interactions and receptor–ligand pairs between all major cell types were estimated using CellPhoneDB (version 2.1.2) (www.cellphonedb.org). Potential interactions between the two cell types were inferred through gene expression levels using 1000 permutation tests. Then, the resulting adjacency matrices were generated for all cell–cell interactions and visualized on heatmaps. Cell–cell interactions within identical cellular lineages were excluded. Only gene pairs involved in receptor-ligand interactions in cell types of interest were visualized if the combined P value was <0.001 (calculated by multiplying all P values within each gene pair).

Data were presented as mean ± SEM. All statistical tests were two-sided, a significant difference was considered when the P value was <0.05. Comparisons of the aortic diameters were analyzed using the student’s t-test and one-way ANOVA. Comparisons of the AAA incidence rates were analyzed using Fisher’s exact test. Comparisons of differences in the composition ratios of cell clusters were analyzed using chi-square test. GraphPad Prism 9 was used for statistical analyses. Differential gene expression testing was conducted in Seurat as outlined in the scRNAseq section.

To investigate the impact of age and dosage on Ang II-induced AAA, we observed no AAA formation at doses of 1.44 mg/kg/day or 2.16 mg/kg/day in 14-week-old mice. However, a minor suprarenal aortic dilation was noted at the 2.16 mg/kg/day dose compared with the NC group (P<0.05) (Figure 1A–C, Table 1). At the age of 20 weeks and 32 weeks, both at a concentration of 1.44 mg/kg/day, the incidence of AAA was 18.2% (2/11) and 57.1% (4/7) (Figure 1D–G). Although it appears that the incidence increased with age, the difference in the occurrence of aortic dissections between these two age groups was not significant (P>0.05), which may be due to the relatively small number of mice used. To precisely define cell-specific transcriptomic changes associated with Ang II-induced AAA, we performed scRNAseq of isolated abdominal aortic tissues from 20-week-old mice, as illustrated in Figure 1H.

After alignment and quality control, 8,716, 8,443, and 12,276 cells from three groups of abdominal aortas of NC, without AAA (no-AAA) and with AAA (AAA), respectively, were combined and subjected to further analysis. Unsupervised clustering using the Seurat R package revealed 26 distinct clusters, representing 12 cell lineages, including all major aortic cell types (Figure 2A–C, Table 2). In accordance with the canonical markers [16], the major clusters were ECs [Cluster 2, 9, 16, 19, 23, 24; Cdh5⁺ (encoding a calcium-dependent glycoproteins)], SMCs [Cluster 5, 10, 13, 18, 20; Tagln⁺ (encoding an SMC-specific cytoskeletal protein)], fibroblasts [Cluster 0, 1, 7, 11, 12; Dcn⁺ (encoding a protein that regulates collagen fibril assembly)], and immune cells [Cluster 3, 4, 6, 8, 14, 15, 17, 21, 22, 25; Ptprc⁺ (encoding Cd45)] (Figure 2D).

In the AAA tissues, the proportions of three major aortic cell types, including SMCs, fibroblasts, ECs were significantly reduced. This reduction may explain why the thickness of the aortic aneurysm wall decreased and its mechanical function was compromised. Consistent with previous findings indicating that inflammation is a fundamental element in the development of AAA [17], we observed a significant increase in immune cells such as monocytes_ neutrophils, macrophages, and T and B lymphocytes in the AAA group. Notably, T and B lymphocytes were the predominant cell populations in AAA tissues (29.9% and 22.6%, respectively). These finding suggest that inflammation in the vasculature is a hallmark of AAA. The differences in composition ratios among the three groups were statistically significant (P<0.0001) (Figure 2E).

Both IL-1β and TNF-α have been associated with the progressive inflammatory process that promotes aneurysm progression [18,19]. These two factors have long been considered among the most important factors promoting the inflammatory process in diseases. In our study, the expression of Il1b and Tnf was up-regulated significantly in the clusters of macrophages and mono_neu, which may promote the inflammatory process (Supplementary Figure S1A and B). IL-6 is produced by multiple cell types and mediates a diverse set of pathophysiologic functions. In our study, the expression of Il6 was relatively high in vascular cells including four fibroblast clusters, endothelium, and muscle (Supplementary Figure S1C). Il6 is higher in no-AAA group than in NC group, which may indicate that IL-6, as an important cytokine, was involved in the formation of AAA.

Compared with NC, the proportions of ECs and SMCs decreased slightly in the no-AAA group, while the proportion of immune cells slightly increased. In contrast, we found that the proportion of fibroblasts increased significantly in the no-AAA group, accounting for almost half of the total cells (17.6% in NC, 53.9% in no-AAA, 14.1% in AAA, P<0.0001). Taken together, the decreased number of three major aortic cell types and the expansion of immune cells may represent typical features in AAA.

Endothelial dysfunction plays a crucial role in the cellular transformation and inflammation of the media and adventitia [20]. In our study, ECs were classified based on cellular markers [21,22] into three groups: malignant (cluster 16, with markers Cd36, Lpl, Gpihbp1, and Fabp4), endothelium_EMT (epithelial–mesenchymal transition) (cluster 23, with markers Lrg1, Ctla2a, Aqp1, Plvap, and Selp), and endothelium (clusters 2, 9, 19 and 24, with markers Cytl1, Pecam1, Clu, Ptprb, and Cldn5). We found that the proportion of ECs significantly decreased in the AAA group, but there was no significant decrease in the no-AAA group (25.64% in NC, 22.18% in no-AAA, and 9.92% in AAA). The cluster of malignant cells mainly presented in AAA tissues (5% of all cells) and no-AAA tissues (4.22% of all cells), while occurring at a low frequency in NC tissues (0.56% of all cells). The proportion of malignant ECs increased significantly (2.17% in NC, 19.04% in no-AAA, 50.41% in AAA), while the proportion of endothelium ECs decreased significantly (96.40% in NC, 74.08% in no-AAA, and 39.41% in AAA). The differences in composition ratio of these three clusters among the three groups were statistically significant (Figure 3A). The EC marker Cdh5 was expressed in the clusters of endothelium, endothelium_EMT, and malignant. Our results suggest that the distributional pattern of EC subpopulations in the three groups differs, allowing for differentiation between the NC, no-AAA, and AAA groups (Figure 3B).

We identified that Cd36, Lpl, Gpihbp1, and Fabp4 were markers of the malignant cluster (Figure 3C). Cd36 is extensively expressed on the surface of ECs throughout the aorta [23], where it functions as a transporter of fatty acids (FAs) [24]. EC-specific Cd36 knockout mice showed diminished transfer of long-chain FAs across ECs [25]. Fabp4 is a canonical PPARγ target gene involved in FA transport. Cd36 and Fabp4 are specialized transporters that facilitate the exogenous uptake of FAs from the surrounding microenvironment [26]. Our results showed the highest co-expression of Lpl and Gpihbp1 (Figure 3D and E), confirming that the cells expressing Lpl and its partner Gpihbp1 transport FA to the capillary lumen [27].

The 1601 differentially expressed genes (including 1168 up-regulated genes and 433 down-regulated genes) in the malignant cluster were used for functional annotation. By conducting Gene Ontology (GO) analysis, it was found that functions related to angiogenesis and blood vessel morphogenesis were up-regulated in the cluster of malignant (Figure 3F). In addition, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed enrichment in the pathways of peroxisome proliferator-activated receptor (PPAR) signaling, relaxin signaling, and ECM-receptor interaction (adjusted P value <0.05 for all pathways) (Figure 3F). Pparg (which encodes PPARγ), a gene that regulates glucose and lipid metabolism, endothelial function, and inflammation, was highly expressed in the cluster of malignant (Figure 3G). Loss of PPARγ function in the vascular endothelium enhances atherosclerosis in high-fat diet-fed apoE^-/- mice [28]. Aortic inflammation during AAA formation is dynamic. Protective anti-inflammatory cytokines are up-regulated in the early compensatory stage. However, pro-inflammatory cytokines are dominant in the late decompensatory stage. PPARγ is likely to continue to up-regulate the expression of anti-inflammatory cytokines, extend the compensatory stage, and decelerate the process of AAA development and rupture [29]. By using the Ang II-induced murine model of aortic aneurysm, Jones et al. found that treatment with PPARγ agonist rosiglitazone had ameliorative effects on aneurysm development and rupture, which exerted a protective effect on aneurysm development and rupture [30]. PPARγ is also important in lipid metabolism as it regulates genes involved in the release, transport, and storage of FAs, including Lpl and the FA transporter Cd36 [31]. In ECs, FAs transport is facilitated by PPARγ, which stimulates the expression of Cd36 and Fabp4 [32]. In mice which lack PPARγ in the endothelium, the Cd36 and Fabp4 expression levels were decreased significantly in the endothelium, serum-free fatty acid and triglyceride (TG) levels were increased [32].

Ang II can induce AAA in normolipidemic mice at an older age, although the incidence was significantly lower compared with that in apoE^-/- mice. The development of AAA is age-dependent, and aging is a major risk factor for cardiovascular disease [33]. By single-cell transcriptomic analysis, Zhang et al. compared aortas and coronary arteries in young and old monkeys. They found differentially expressed genes between young and old monkeys, and the core pathways annotated for these genes are involved in ‘response to lipids’ [34]. Therefore, we speculated that advancing age accompanied with abnormal lipid metabolism may contribute to the development of AAA. The cluster of malignant EC appeared after Ang II infusion, in which Cd36, Lpl, Gpihbp1, Fabp4, and Pparg were highly expressed. Thus, in the cluster of malignant, we speculated that the high expression of Cd36, Lpl, Gpihbp1, Fabp4, and Pparg was important in the metabolism of FAs and may contribute to endothelial dysfunction.

Monocytes and macrophages play a crucial role in aneurysm formation by influencing the migration of leukocytes from the circulating bloodstream to the arterial subendothelial space. We found that the proportion of the mono_neu cluster (representing monocytes and neutrophils) significantly increased in the AAA group (17.24%) compared with 1.05% and 0.87% in the NC group and no-AAA group, respectively. Compared with monocytes, there was no significant difference in the proportion of macrophages between AAA and no-AAA, which were 6.76% and 5.60%, respectively. It has been reported that Spp1-deficient leukocytes provide substantial protection against Ang II-accelerated atherosclerosis and AAA formation [35]. In our study, compared with macrophages in NC and no-AAA, we identified a subpopulation of macrophages in AAA that exhibited significantly up-regulated expression of Spp1 and notably down-regulated expression of Ccl4 and Cxcl10 (Figure 4A). This indicates that these cells may have a crucial role in AAA pathogenesis. In order to validate the existence and localization of Spp1, our results showed that Spp1 was present in the AAA tissues and colocalized with Cd68 through co-immunofluorescence staining. In contrast, Spp1 was not observed in the NC and no-AAA groups (Figure 4B).

We next explored why about two thirds of mice receiving Ang II infusion did not develop AAA. Fibroblasts were identified as the largest cell population in no-AAA tissues. These fibroblasts have been differentiated into four major subpopulations: fibroblast, fibroblast_Il6, fibroblast_Ccn5, and fibroblast_Cd34 [36,37]. Il-6 is a member of the interleukin family. Camk2d (Ca²⁺/calmodulin-dependent protein kinase II) is known to play an important role during cardiac hypertrophy [38]. Signal transducer and activator of transcription 3 (Stat3) is a member of the STAT protein family, which acts as a transcription activator, and response to cytokines and growth factors. Il-6 can directly activate Camk2d, the latter then activate Stat3 [39]. The Il6–Camk2d–Stat3 pathway is involved in fibroblast activation, leading to cardiac fibrosis [40]. In the cluster of Il6⁺ fibroblasts, the expression of Camk2d and Stat3 was up-regulated (Figure 5A), suggesting that this cluster may promote vascular fibrosis.

Cd34 is one of the markers for myofibroblast (MF). MFs are specialized cells that originate in pathophysiological conditions and contribute to tissue repair during wound healing [41]. Cellular communication network 5 (Ccn5), which is known as a secreted protein, mediates anti-fibrotic activity by inhibiting fibroblast-to-myofibroblast transition [42,43]. In consideration of the increased proportions of both Cd34⁺ fibroblasts and Ccn5⁺ fibroblasts in no-AAA compared with NC, we speculate that under the stimulation of Ang II, Cd34⁺ fibroblasts proliferated, which may contribute to tissue repair. Additionally, Ccn5⁺ fibroblasts increased to balance the increase of Cd34⁺ fibroblasts due to their anti-fibrotic activity [42,43]. Our data showed that although the total number of fibroblasts differed significantly between no-AAA and AAA tissues (52.21% vs 10.04%, P<0.0001), the changes in fibroblast proportions exhibited a similar trend in these two groups when compared with NC. The cell proportions of fibroblast and fibroblast_Il6 were lower in no-AAA and AAA tissues compared with the NC group, while the cell proportions of fibroblast_Cd34 and fibroblast_Ccn5 were higher in no-AAA and AAA tissues compared with NC (Figure 5B). The differences in composition ratios of these four clusters among the three groups were statistically significant (P<0.0001). Thus, changes in fibroblast subtypes may be associated with AAA development.

The loss of elastic fibers is an early event during aneurysm formation. The adventitious tissue, in which collagen and fibroblasts are rich, is responsible for the resistance of aneurysm in the absence of medial elastin [12]. In our study, the expression of Eln was high in the cluster of endothelium, fibroblast and fibroblast_Ccn5, and muscle, and the proportions of these clusters were reduced significantly in the AAA group (Figure 5C).

To visualize the intercellular communication network, Figure 6 illustrates the number and strength of interactions between 12 cell clusters. All four types of fibroblast clusters are strongly associated with mono_neu and macrophages. Additionally, fibroblast_Ccn5 is strongly linked to the muscle cluster, and fibroblast_Cd34 is strongly connected to the endothelium_EMT cluster. By analyzing cell communication with CellphoneDB, we identified specific ligand–receptor interaction pairs, such as Cxcl12–Cxcr4 and Cxcl12–Ackr3, which consistently occurred between different clusters (Figure 7A). Cxcr4 expression was significantly increased in the cluster of immune cells, including macrophages, monocytes, and T and B lymphocytes, while Ackr3 expression was significantly increased in the cluster of non-immune cells (Figure 7B). By immunostaining, we confirmed the increased expression of Cxcr4 mostly in the aneurysm of the AAA group (Supplementary Figure S2A) and increased expression of Ackr3 mostly in the vascular wall and adventitia of the AAA group (Supplementary Figure S2B).

As a highly conserved seven transmembrane regions protein, Cxcl12 is the only ligand for Cxcr4, which induces the activation of PI3K–Akt signaling pathway and regulates the phosphorylation of ERK1/2 to activate NF–κB and mTOR signaling pathways, thereby regulating cell growth and proliferation [44]. Cxcr4 plays a role in leukocyte chemotaxis during inflammatory conditions in various autoimmune diseases, and elevated levels of Cxcr4 have been observed in various types of human cancers [45]. In our study, we found ubiquitous expression of Cxcr4 in macrophages, monocytes, and T and B lymphocytes, which might be recruited by Cxcl12 signaling and promote the progression of inflammation, which play a progressive role in the AAA group. Meanwhile, Ackr3 has 10 times more binding affinity for Cxcl12 than the classical receptor Cxcr4. The primary role of Ackr3 is to internalize and deliver its ligand Cxcl12 for lysosomal degradation, thus negatively regulating the Cxcl12-Cxcr4 signaling cascade [46]. The Cxcl12–Ackr3 activates MAPK/ERK signaling pathway to regulate cell survival, migration and differentiation [47]. The cell–cell interaction via Cxcl12–Ackr3 was identified between all clusters of endothelium and fibroblasts by the CellPhone database, which may play a protective role in AAA formation. The differentially distributional expression of Cxcr4 and Ackr3 as shown in Supplementary Figure S2 may imply their antergic roles in different cell types during AAA formation. For example, the higher expression of Cxcr4 and lower expression of Ackr3 in the aneurysm may contribute to the progression of AAA. Thus, the Cxcl12–Cxcr4/Ackr3 axis may play pivotal roles in the initiation, development, and progression of aneurysm.

In this study, we investigated the incidence of AAA in normolipidemic C57BL/6 J mice at different ages. We found that at the age of 20 weeks and 32 weeks with an Ang II concentration of 1.44 mg/kg/day, the incidence of AAA reached 18.2% and 57.1%, respectively. This suggests that the development of AAA is age-dependent. We also found that the diameters of the suprarenal aortic lumen slightly increased with a dose effect of Ang II at the age of 14 weeks. Compared with the apoE^-/- mice, the incidence of AAA in the normolipidemic mice is relatively lower. Therefore, the C57BL/6 J mouse model would be more appropriate for testing induction factors rather than protection factors. As advancing age is a risk factor for AAA development in humans [48], which affecting up to 9% of adults older than 65 years of age [49], the incidence of Ang II-induced AAA in mice increased with age, as demonstrated in this study. Therefore, aged C57BL/6 J mice with AAA are better to mimic human AAA.

In human, the most accessible AAA tissue is from advanced lesions that have been dissected during open surgical repair, which cannot provide sufficient evidence of the molecular events at earlier stages. We investigated the cellular landscape of the no-AAA group to mimic the earlier stages of AAA. We found that fibroblasts were the most abundant cell type in the no-AAA tissues. We speculated that the expansion of fibroblasts may be a compensatory reaction to protect the mice from the development of aneurysm. It has been reported that the protective effect may be achieved through modulating extracellular matrix production, thereby enhancing aortic stability [50]. Hong et al. also reported that enhancing collagen synthesis, which increases aortic stiffness, can be a promising approach for treating AAA [51], but further studies are needed to evaluate the benefits and side-effects. The reduction in fibroblasts may be related to the formation of AAA. In addition, C57BL/6 J mice are widely used in studying AAA. We have depicted the difference between the no-AAA and AAA groups induced by Ang II. The cellular landscape can be used to assess the impact of genes or chemicals on alleviating or exacerbating the AAA condition in this mouse models.

The two types of vascular diseases, atherosclerosis and AAA, are strongly interrelated and associated [52]. Although these two diseases develop through different pathogenic mechanisms, they are both closely associated with endothelial dysfunction [20]. The earliest changes that eventually result in the formation of atherosclerotic lesions occur in the endothelium [53]. Endothelium can first sense excessive hemodynamic stimulation, then release and activate various vasoactive substances, disrupt the dynamic balance between proliferation and apoptosis of various cells including ECs, SMCs and fibroblasts, and ultimately alter the morphology, structure, and function of blood vessels. In our study, we identified a unique cluster of malignant arisen after Ang II infusion, and these cells are implicated in lipid uptake. Consistent with previous studies, our results suggest that dyslipidemia and disordered lipid metabolism may be a significant risk factor for AAA development.

The Cxcl12–Cxcr4/Ackr3 axis is involved in both proatherogenic and atherogenesis-protective functions in various cell types [46]. We found that this axis played a role in cell–cell interactions, which is confirmed by the increased expression of Cxcr4 in the aneurysm and increased expression of Ackr3 in the vascular wall and adventitia. Similarly, He et al. found that VSMCs interacted with neutrophils and fibroblasts via Cxcl12–Cxcr4/Ackr3 axis, thereby deteriorating the progression of acute thoracic aortic dissection [54]. We speculate that Cxcl12–Ackr3 plays a protective role, while Cxcl12–Cxcr4 plays a promoting role in AAA formation. Due to the functional heterogeneity and cross-talk between Cxcr4 and Ackr3 at the receptor level and downstream pathways, it will be interesting to explore the clinical implications and modulators of Cxcl12–Cxcr4/Ackr3.

The developmental origin, anatomical location, and other factors contribute to aortic heterogeneity in a physiological state. In recent years, several studies utilizing scRNAseq have focused on the changes in cell clusters within AAA tissues, consistently revealing the heterogenous nature of aortic cells [55]. Chen et al. reported that the reprogramming of normal SMCs into a mesenchymal stem cell-like state led to the formation of aortic aneurysms [56]. Therefore, it has been revealed that various cell types contribute to the formation of AAA, and the roles of different vascular and immune cells in AAA pathogenesis require further characterization. However, since the transcriptional changes observed in this study are based on descriptive analyses, the changes in immune cells may reflect the consequences of the inflammatory response following aortic wall disruption rather than the causes of AAA initiation or progression.

In summary, we have characterized the cellular landscape of the NC, no-AAA, and AAA group. We speculated that upon Ang II infusion, ECs first sense excessive hemodynamic stimulation, then subsequently releasing and activating various vasoactive substances. The dynamic balance may be disrupted between proliferation and apoptosis of various cells including ECs, SMCs, fibroblasts, and immune cells and ultimately to alter the morphology, structure, and function of blood vessels. Although mice were subjected to the same infusion of Ang II, there were interindividual differences in endothelial shear stress and blood flow. This resulted in two distinct groups (no-AAA and AAA) despite receiving the same treatment. Under the influence of Ang II, clusters of endothelium_EMT and malignant EC may appear, which contributes to endothelial dysfunction and triggers a series of cascades in the immune response, leading to SMC apoptosis, monocyte recruitment, and macrophage accumulation. This provides new clues for the intervention and prevention of AAA formation.

The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2021) in National Genomics Data Center (Nucleic Acids Res 2022), China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA012412) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa/s/mv05GM5r [57, 58].

The authors declare that there are no competing interests associated with the manuscript.

This work was equally supported by grants from the Ministry of Science and Technology of China [2019YFA0802104], and the National Natural Science Foundation of China [32293204; 81770234].

H.L., P.Z., and W.L.: Methodology and Writing—original draft. H.L.: Formal analysis and Data curation. X.H., P.Z., and J.G.: Data curation.

We thank Dr. Zhiqing Li and Dr. Yiting Jia from Peking University for providing AAA and control samples used in this study.

AAA

abdominal aortic aneurysm

Ang II

angiotensin II

BAPN

β-aminopropionitrile monofumarate

ECs

endothelial cells

NC

negative control

SMCs

smooth muscle cells

scRNAseq

single-cell RNA sequencing