All cell types in aortic aneurysm

A total of 50,690 cells with 37,335 gene expression profiles were used to construct the Seurat object. An unsupervised clustering algorithm in Seurat was used for filtered cells, resulting in 25 distinct clusters. Regarding canonical markers and automated reference-based annotation tools (SingleR), the clusters were monocytes/macrophages (CD14 + CD68 + CD163 +), smooth muscle cells (ACTA2 + MYH11 + CNN1 +), endothelial cells (ECs) (PECAM1 + VWF + ECSCR +), CD8 + T cells (CD3D + CD3E + CD8A +), B cells (CD37 + CD79A + CD79B +), natural killer cells (CD160 + KLRC1 + XCL2 +), fibroblasts (COL1A1 + COL1A2 + PDGFRA +), and haematopoietic stem cells (HSCs) (KIT + CD44 + GATA2 +) (Fig. 1A-C). The proportion of immune cells (macrophages, B cells, T cells, NK cells) was increased in aortic aneurysm, while the proportions of VSMCs and fibroblasts were decreased (Fig. 1D).

Fig. 1

General analysis and annotation of all cells from the normal aorta and aortic aneurysm. Uniform manifold approximation and projection (UMAP) representation of the aligned gene expression data from the normal aorta and aortic aneurysm, showing the partitioning of 9 cell types A and groupings by sample source B. The expression of marker genes is shown in the dot plot C. The origin distribution of various cell types and the proportions of various cell types in aortic aneurysm and the normal aorta D

Identifying VSMCs phenotypes

A total of 7150 cells were preliminarily identified as VSMCs (3255 from aortic aneurysm, 3895 from the normal aorta). These cells were clustered into 14 Seurat clusters after being scaled and normalized. By labelling canonical markers and calculating differential genes, potential VSMC phenotypes were identified as follows: (i) Contractile VSMCs highly expressed the VSMCs marker genes ACTA2 and MYH11, possessed the function of contraction and expressed no other specific markers (CD68-CD34-FABP4- CD3D-). (ii) Fibroblast VSMCs, which showed lower expression of the contraction genes ACTA2 and MYH11, robustly secreted extracellular matrix (COL1A1, COLIA2) and showed no expression of other specific markers (CD68-CD34-FABP4-CD3D-). (iii) T-cell-like VSMCs mildly expressed VSMC markers (ACTA2 + MYH11 +) and strongly expressed T-cell markers (CD3D + CD3G +), and they also exhibited the highest levels of inflammatory gene expression. (iv) Adipocyte-like VSMCs expressed adipocyte markers (EBF2 + FABP4 +) and no other specific markers (CD68-VWF-CD34-KLRB1-). (v) Macrophage-like VSMCs had high expression of macrophage markers (CD14 + CD68 +). (vi) Mesenchymal-like VSMCs highly expressed stem cell markers (CD34 + ENG +) (Fig. 2A-B).

Fig. 2

Specific analysis and annotation of all VSMCs phenotypes. Uniform manifold approximation and projection (UMAP) representation of the aligned gene expression data of all VSMCs showing the partitioning of 6 annotated VSMCs phenotypes A. The expression of marker genes is shown on the dot plot B (gene expression log-normalized by Seurat). The proportion of VSMCs phenotypes in aortic aneurysm and the normal aorta C and the origin distribution of various VSMCs phenotypes D

Functional analysisContractile VSMCs

Contractile VSMCs were present in aortic aneurysm and normal aortic tissue, but their proportion in AA was significantly decreased (Fig. 2C-D). This result suggested that VSMCs were less capable of contracting during aortic aneurysm and dedifferentiating into other potential phenotypes. Contractile VSMCs showed high expression of contractile proteins and exhibited high cell‒cell junction scores and cell–matrix junction scores, which are important for maintaining the structural integrity and rigidity of the aortic wall (Fig. 3A-B). The expression of inflammatory chemokines was lowest in contractile VSMCs, which was consistent with normal aorta physiology. Some ECM genes (such as COL14A1) were upregulated in contractile VSMCs to maintain the ECM structure and repair the ECM. Moreover, lower protease gene expression in contractile VSMCs was associated with less ECM remodelling in normal aortas (Fig. 3C-D). Contractile VSMCs primarily offered contractile force and constituted the muscle structure (Fig. S1A).

Fig. 3

Functional analysis of various VSMCs phenotypes. The cell‒cell junction scores between any two VSMCs phenotypes A and cell–matrix scores of each VSMCs phenotype B. Heatmap showing chemokine genes C, collagen genes D, and proteinase genes E in various VSMCs phenotypes

Fibroblast-like VSMCs

The proportion of fibroblast-like VSMCs was higher in aortic aneurysm than in the normal aorta (Fig. 2E-F). Fibroblast-like VSMCs also had the highest cell‒cell junction score and cell–matrix junction score (Fig. 3A-B). Compared with those in contractile VSMCs, the contractile genes in fibroblast-like VSMCs were decreased, but overall ECM gene expression was further increased (Fig. 3D). Thus, fibroblast-like VSMCs possess a stronger ability to secrete ECM proteins, which means that fibroblast-like VSMCs may mediate repair after vascular injury. The overall level of proteases in fibroblast-like VSMCs was low, but the expression of MMP2 was increased, indicating that fibroblast-like VSMCs are involved in ECM remodelling to a certain extent, which is consistent with previous studies (Fig. 3E). Interestingly, the expression levels of inflammatory chemokines (CXCL10, CXCL12, CCL2) in fibroblast-like VSMCs were slightly higher than those in contractile VSMCs (Fig. 3C). The upregulated genes in fibroblast-like VSMCs were enriched in extracellular matrix organization and collagen-containing extracellular matrix (Fig. S1B).

Mesenchymal-like VSMCs

An increased proportion of mesenchymal-like VSMCs was present in aortic aneurysms (Fig. 2C-D). We observed downregulated expression of contraction genes in mesenchymal-like VSMCs, and no other specific markers were expressed except for stem cell markers (CD34 + ENG +). Mesenchymal-like VSMCs obtained a moderate cell–matrix junction score and a low cell‒cell junction score (Fig. 3A-B). In addition, mesenchymal-like VSMCs exhibited moderate chemokine levels, moderate ECM secretion, and relatively high proteinase levels (Fig. 3C-E). The enriched GO terms of mesenchymal-like VSMCs were similar to those of secretory VSMCs and included extracellular matrix organization, collagen-containing extracellular matrix, and collagen binding (Fig. S1C).

Adipocyte-like VSMCs

Approximately 70% of adipocyte-like VSMCs were present in aortic aneurysms (Fig. 2C-D). Adipocyte-like VSMCs obtained low cell‒cell junction scores and the lowest cell–matrix junction scores (Fig. 3A-B). Adipocyte-like VSMCs were characterized by low chemokine levels but expressed CCL2, CCL8, CCL19, and CXCL12 (Fig. 3C). In addition, adipocyte-like VSMCs exhibited moderate ECM protein levels (Fig. 3D). The general expression of proteases was low except for ADAMTS1 and ADAMTS4 (Fig. 3E). Regarding gene enrichment analysis, adipocyte-like VSMCs were involved in ribosomal metabolism and peptide hormone metabolism (Fig. S1D).

Macrophage-like VSMCs

The proportion of macrophage-like VSMCs was significantly increased in aortic aneurysms (Fig. 2C-D). Macrophage-like VSMCs did not form strong cell‒cell junctions with any VSMC phenotype, and a moderate cell–matrix junction score was obtained (Fig. 3A-B). The most significant characteristic of macrophage-like VSMCs was that these cells had the highest level of chemokines such as CCL3, CCL4, and CXCL8 (Fig. 3C). No significant expression of ECM or protease genes was observed (Fig. 3D-E). Functional enrichment analysis indicated that macrophage-like VSMCs exerted effects on antigen processing and presentation (Fig. S1E).

T-cell-like VSMCs

The characteristics of T-cell-like VSMCs are similar to those of macrophage-like VSMCs. T-cell-like VSMCs were primarily present in aortic aneurysms (Fig. 2C-D). Among all VSMC phenotypes, T-cell-like VSMCs obtained the lowest cell‒cell junction scores than other VSMCs phenotypes and a low cell–matrix junction score (Fig. 3A-B). T-cell-like VSMCs were characterized by low ECM protein levels, low protease levels, and high chemokine levels (Fig. 3C-E). T-cell-like VSMCs were involved in the positive regulation of leukocytes according to GO functional enrichment analysis (Fig. S1F).

Single-cell trajectory analysis

To identify the origin, trajectory, and timing of differentiation in the VSMC phenotypes, we performed trajectory analysis. From the pseudotime plot and the phenotype-based trajectory plot, we learned that all VSMCs phenotypes were initially derived from contractile VSMCs and fibroblast-like VSMCs (Fig. 4A). During the mid-to-late timeline, contractile VSMCs and fibroblast-like VSMCs gradually differentiated into two groups (Fig. 4B-C). One group differentiated into T-cell-like VSMCs, macrophage-like VSMCs, and mesenchymal-like VSMCs. Fibroblast-like VSMCs appeared in the early stage of aortic aneurysm formation, which may be a response to vascular injury (Fig. 4D). T-cell-like VSMCs and macrophage-like VSMCs, which are present in the late stage, may be related to the progression of aortic aneurysm.

Fig. 4

Cell trajectory analysis of various VSMCs phenotypes. Pseudotime plot showing trajectory analysis of all VSMCs A and various VSMCs phenotypes B-C. Cell density is plotted along the timeline D

Regulon network analysis by SCENIC

Previous studies have shown that TFs such as KLF4 [21], RUNX2 [29], and SRF [30] play a pivotal role in regulating VSMC phenotypes. We further investigated the correlation between regulon (TFs and their target genes) activity and VSMC phenotypes using SCENIC. A total of 144 regulons were constructed, and 31 regulons scored high activity (Fig. 5A). MAF (32 g) and MAFB (29 g) had specific regulatory effects on macrophage-like VSMCs (Fig. 5B-C). RUNX3 (39 g) specifically regulated T-cell-like VSMC differentiation (Fig. S2A). NR2F2_extended (14 g) demonstrated high regulatory activity on adipocyte-like VSMCs (Fig. S2B). Fibroblast-like VSMCs were positively regulated by PLAGL1 (14 g), and mesenchymal-like VSMCs were positively regulated by AR (14 g) (Fig. S2C-D).

Fig. 5

SCENIC analysis indicated significant regulons for each VSMCs phenotype A. The regulatory activity of MAF (32 g) B and MAFB (29 g) C was projected with a UMAP plot

Assessment of cell‒cell communication among 6 VSMCs phenotypes

As previously mentioned, there is no understanding of how VSMCs phenotypes occur in aortic aneurysms. We believe that intercellular communication among distinct VSMCs phenotypes may accelerate this process. Therefore, the R package ‘CellChat’ was used to investigate the communication and interaction among identified VSMC phenotypes. We first assessed the general interactions between various VSMC phenotypes. Figure 6A shows the number of cell‒cell communications and the interaction strength among 6 VSMCs phenotypes. Macrophage-like VSMCs possessed the highest interaction numbers and strength. Pseudotime analysis showed significant signalling pathways from early-emerging VSMCs to late-occurring VSMCs (Fig. 6B). Macrophage migration inhibitory factor (MIF) signalling showed the highest level of communication probability. Figure 6C shows the incoming/outgoing communication patterns of VSMCs populations. Macrophage-like VSMCs mediated the highest incoming communication, while mesenchymal-like VSMCs mediated the highest outgoing communication.

Fig. 6

Global cell‒cell communication patterns involve multiple signalling pathways. The line width represents the interaction quantity and interaction strength among various VSMCs phenotypes A. Communication strength of all significant signalling pathways from contractile/fibroblast-like VSMCs to macrophage-like/T-cell-like/adipocyte-like/mesenchymal-like VSMCs B. Significant signalling of each VSMCs phenotype in incoming/outgoing communication patterns C

Regarding the network central plot, contractile VSMCs, fibroblast-like VSMCs, and adipocyte-like VSMCs were more likely to act as senders in the MIF signalling-mediated network (Fig. S3A). On the other hand, macrophage-like VSMCs, T-cell-like VSMCs, and mesenchymal-like VSMCs were suggested to play sophisticated roles. In the MIF signalling network, fibroblast-like VSMCs send the most signals, which were mainly received by macrophage-like VSMCs (Fig. S3B-C). The ligand‒receptor pairs MIF-(CD74 + CXCR4), MIF-(CD74 + CD44), and MIF-ACKR3 contributed to MIF signalling (Fig. S3D).

Quantification of VSMCs markers and the spatial localization of VSMCs

Mesenchymal-like VSMCs, macrophage-like VSMCs, and T-cell-like VSMCs were selected for further validation by qPCR and dual immunofluorescence analysis, due to their potential vital roles in aortic aneurysms. A mouse model of aortic aneurysm was successfully established (Fig. 7A). The relative expression level of the contractile marker ACTA2 decreased to 0.307 (p < 0.001) in the aortic aneurysm model, while the expression level of CD68 increased to 54.85 (p < 0.001) (Fig. 7B). Additionally, CD3D and CD34 were overexpressed (CD3D = 7.59 (P < 0.001), CD34 = 8.025 (p < 0.001)). RNA FISH indicated that a spot of cells coexpressing αSMA and CD68 (Fig. 8A) and a spot of cells coexpressing αSMA and CD3D were present in the tunica media of aneurysm tissue (Fig. 8C), verifying the presence of macrophage-like VSMCs and T-cell-like VSMCs. No macrophage-like VSMCs or T-cell-like VSMCs were detected in normal aortas. A few cells coexpressing αSMA and CD34 were present in normal aortas (Fig. 8B). However, the number of αSMA + /CD34 + cells and the fluorescence intensity were dramatically increased in full-thickness aortic aneurysms, indicating the transition of normal VSMCs into mesenchymal-like VSMCs.

Fig. 7

Successful construction of the mouse abdominal aortic aneurysm model A. The relative expression levels of ACTA2. CD68, CD3D, and CD34 were quantified by qPCR B

Fig. 8

RNA FISH of the normal aorta and aortic aneurysm. Dual hybridization of αSMA and CD68 to identify macrophage-like VSMCs A. Dual hybridization of αSMA and CD34 to identify mesenchymal-like VSMCs B. Dual hybridization of αSMA and CD3D to identify T-cell-like VSMCs C