The GEO database (https://www.ncbi.nlm.nih.gov/geo/) was consulted in order to obtain the DaExpression matrix of TAAD and associated clinical data. All of the data was collected up to January 31, 2024. The inclusion criteria: (I) samples in the dataset were obtained from the aorta; (II) the samples contained TAAD patients and Controls (patients without TAAD); (III) the dataset was based on human gene expression profiles. The inclusion criteria: samples without clinical information. Hereby, Ten TAAD tissues and ten normal Controls were included in the GSE153434 dataset [18], which was found on Illumina HiSeq X Ten Homo sapiens platform GPL20795. The R.DESeq2 package (version 1.30.1) was used to select differentially expressed genes (DEGs), with cutoffs of $|$log2 fold change (FC)$|$ $>$1 and p 0.05, and genes associated with differentially enriched immune cells were filtered by least absolute shrinkage and selection operator (LASSO). Volcano plots and heatmaps of DEGs were generated using the ggplot2 R package (Version R3.2.3) and heatmap R package (version R2.2.1) [19]. The RMA algorithm R.limma program was used to normalize data from five normal tissues and seven TAAD in the microarray dataset GSE52093, which acted as the external validation cohort [20].

Immune cell infiltration and target gene expressions in TAAD was assessed using R (Version 3.6.3, The R Foundation for Statistical Computing, Vienna, Austria).

Targeting the DEGs, a co-expression network was built with the R.WGCNA package (Version 4.1.3). The weighted adjacency matrix was constructed using a power function based on the soft threshold parameter $\beta$, and the scale-free network was ensured by setting $\beta$ = 18 (scale-free R² = 0.8). In order to measure network connectivity, the adjacency matrix was transformed into a topological overlap matrix (TOM). Genes with similar expression profiles were then clustered based on TOM dissimilarity, and a gene dendrogram was used to divide the modules into minimum sizes of 50. After calculating the correlations between Pearson’s module eigengenes (ME), modules with strong correlations were integrated at r $>$ 0.7. Gene significance (GS) and module membership (MM) were used to identify the clinical trait-associated significant models. The immunophenotypic score of infiltrating immune cells was the clinical trait that was determined in this study.

The R package clusterProfiler (Version 3.6.2) was utilized to perform enrichment studies for Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) in order to explore the possible pathways and functions of candidate hub genes in important modules. Cell components (CC), biological processes (BP), and molecular functions (MF) make up GO enrichment. It was decided to set statistical significances for only terms with false discovery rate (FDR) 0.05. The R packages ggplot2 and GO plot (Version 1.0.2 ) are used to display the enrichment findings.

The Least Absolute Shrinkage and Selection Operator (LASSO) algorithm, a logistic regression method, was utilized to filter variables to improve the predictive performance in order to find the candidate biomarkers and create a diagnostic model for TAAD. The “glmnet” software package (Version 0.2.0) was used to screen the potential biomarkers. Next, to improve the accuracy, ensemble learning is employed to merge the Random Forest (RF) algorithm with numerous trees, and the Random Forest software package (Version 6.3.1) is utilized to reduce the pool of potential biomarkers. The hub genes for creating TAAD diagnostic models are the overlapping genes of the LASSO model and the MeanReduced Gini $>$2 genes of the RF model.

Based on the clinical trait relevance and module connectivity, hub genes were selected from module genes using absolute value of GS $>$0.6 and MM $>$0.8 as thresholds. Next, we examined the connection between 24 immune cells and hub genes.

We evaluated the discriminatory power of the hub genes as TAAD biomarkers using the GSE52093 dataset, which included five normal samples and seven TAAD. Results were assessed using receiver operator characteristic (ROC) curves, where the area under the curve (AUC) was computed.

Six clinical samples were obtained from Shaanxi Provincial People’s Hospital; written informed consent forms were submitted by the patients or their families/legal guardians. The samples included three normal samples (Control group) from donors without TAAD and three TAAD samples (TAAD group) from patients with TAAD. Total RNA was extracted from aorta samples using the Total RNA Kit I (Omega Bio-tec, Inc., Norcross, GA, USA), then transformed into cDNA PrimeScript™ RT reagent kit (TaKaRa, Dalian, China). The Green PCR Master Mix System (Thermo Fisher Scientific, Waltham, MA, USA) was used in quantitative polymerase chain reaction (qPCR) using an ABI7500 Real-Time PCR System (Thermo Fisher Scientific). GAPDH served as the internal point of reference. The 2^{-Δ⁢Δ⁢CT} method was utilized to evaluate the relative fold change. Likewise, the aorta in the rat model is operated by the RT-qPCR experiment in the same manner as previously said. Supplementary Table 1 contains a list of primer sequences that were employed.

All animal experiments were in accordance with the guide for the care and use of laboratory animals established by United States National Institutes of Health (Bethesda, MD, USA). Male Sprague-Dawley (SD) rats, weighing between 90 and 120 grams, 4–6 weeks old, were obtained from the Laboratory Animal Center of the Medical Department of Xi’an Jiaotong University (Xi’an, China). The rats were randomly assigned to two groups: the Control group and the TAAD group. Each group contains six rats. Control group: Oral administration of 0.9% saline was given to the rats. For four weeks, once a day. TAAD group: For four weeks, rats were given 1 mg/kg/day of $\beta$-Aminopropionitrile (BAPN) dissolved in 0.9% saline orally. At the sixth week, tissue samples were collected, and all of the rats were put to death by breathing in an excessive amount of isoflurane. Prior to sacrifice, an echocardiogram was done, and the body weight was recorded during modeling.

An intraperitoneal dose of 30 mg/kg pentobarbital was used to anesthetize the rats. Echocardiography was performed with ultra-high-resolution small animal ultrasound imaging system Vevo3100. Measurements were made of the ejection fraction (EF), fraction shortening (FS), aortic diameter (AO), and left coronary artery diameter (LCA). Three cardiac cycles’ worth of data were analyzed to assess cardiovascular function.

The whole protein was extracted from rat aortic tissue, separated using polyacrylamide gel electrophoresis containing sodium dodecyl sulfate, and then transferred onto nitrocellulose membranes (MilliporeSigma). The membranes were blocked for two hours at room temperature with 5% non-fat milk, and then incubated overnight at 4 °C, with the corresponding primary antibodies: Rabbit anti-Human ATP-binding cassette superfamily G menber 2 (ABCG2) (Dilution 1:1000, Abcam, Cambridge, UK), Rabbit anti-Rat ABCG2 (Dilution 1:1000, Abcam, Cambridge, UK), Rabbit anti-Human family with sequence similarity 20, member C (FAM20C) (Dilution 1:500, Abcam), Rabbit anti-Rat FAM20C (Dilution 1:500, Abcam), Rabbit anti-Human metabolism enzymemethylenetetrahydrofolate dehydrogenase 2 (MTHFD2) (Dilution 1:1000, Abcam), Rabbit anti-Rat MTHFD2 (Dilution 1:1000, Abcam), Rabbit anti-Human CUB domain containing protein 1 (CDCP1) (Dilution 1:1000, Abcam), Rabbit anti-Rat CDCP1 (Dilution 1:1000, Abcam), Rabbit anti-Human $\beta$-actin (Dilution 1:1000, Abcam), Rabbit anti-Rat $\beta$-actin (Dilution 1:1000, Abcam). The membranes were treated with the secondary antibody (Goat anti-Rabbit) for one hour at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) kit (Pierce, Thermo Fisher Scientific, Inc.).

Tissue sections fixed in paraffin, measuring 4 µm in thickness, were rehydrated and dewaxed. Antigen was extracted, then incubated respectively with the same monoclonal antibody as Western blotting at 4 °C overnight, then treated with Horseradish Peroxidase (HRP)-conjugated secondary antibody at room temperature for half an hour, followed by staining in DAB detection kit and restained with hematoxylin for 30 minutes. Ultimately, they were examined using a BX41 fluorescent microscope (Olympus Corporation, Tokyo, Japan), and visualized the immunoblot bands in a pre-cooled LAS4000 luminescent imager (GE, Boston, MA, USA).

The Pearson algorithm was used to assess the parameter (EF%, FS%, and AO), and the R package “ggplot2” was used to illustrate the results [21].

GraphPad Prism 8.0 (GraphPad Software, Inc., San Diego, CA, USA) was used to analyze all of the data, which were reported as the mean $\pm$ standard deviation (SD) of three separate experiments. Dunnett’s post-hoc test was used after a one-way analysis of variance or Student’s t-test to ascertain the statistical differences between the groups. Indicating a statistically significant difference was a p value less than 0.05.

1253 upregulated and 1486 downregulated DEGs were tested using the GSE153434 dataset. Fig. 1 illustrated the expression features of the DEGs in volcano plots (A) and heatmaps (B) for the ImmuCellAI, which indicated notable variations in the infiltrative abundance of 24 immune cells types between TAAD and normal samples. Among these, six cell types with significant differences were found in TAAD and the Control group, as shown in Fig. 2. More specifically, the TAAD group had higher concentrations of macrophages, T helper cell 17 (Th17), CD8 naive cells, dendritic cells (DC) and mucosal-associated invariant T (MAIT). By comparison, the Control group showed a significant enrichment of natural killer cells (NK), induced regulatory T cells (iTregs), and B cells (Fig. 2A–C). Moreover, macrophages showed the strongest negative correlation with iTregs cells and the strongest positive correlation with Th17 cells, according to the correlation matrix data (Fig. 2D).

Fig. 1.

Volcano plot (A) and heatmap (B) clustering for differentially expressed genes identified in type A aortic dissection (TAAD).

Fig. 2.

The immunological cell infiltration landscape across all TAAD samples. (A) The violin plot displays the proportions of each immune cell. (B) The bar chart of immune cells. (C) The heatmap depicts the proportions of all 24 types of immune cells. (D) Correlations between the proportions of individual immune cells. Th17, T helper cell 17; DC, dendritic cells; NK, natural killer cells; MAIT, mucosal-associated invariant T.

The WGCNA network was built from these DEGs. Based on the scale-free topology model and average connectivity, the gene regulation network’s ideal soft threshold power is 12 (Fig. 3A). The hierarchical clustering tree was structured using dynamic hybrid cutting to create five modules (brown, red, blue, black, and pink) (Fig. 3B,C). The network heat map was plotted based on the five modules and 24 immune cell types (Fig. 3D). The module-trait relationships revealed that all five modules were linked to macrophages and iTreg, while four modules (brown, red, blue, and black) were linked to Th17 cells, four modules with NK cells, and none with CD8 naive cells. Based on MM $>$0.8 and GS $>$0.6 standards, 432 genes in all modules were chosen as hub genes for further analysis.

Fig. 3.

Determination of key modules linked to immunophenotypic scores using weighted gene co-expression network analysis (WGCNA). (A) Selection of the soft-thresholding powers ($\beta$) based on the scale-free fit index and mean connectivity. (B) Hierarchical cluster analysis was adopted to identify co-expression modules. (C) The heatmap displayed selected module genes’ topological overlap matrix (TOM). (D) The heatmap shows the correlations between modules and immune traits.

GO and KEGG enrichment analyses were carried out to identify 432 genes’ putative activities and pathways. Consequently, 461 associated BPs and 22 KEGG pathways with statistical significance were found. Cell junction assembly, cell migration, platelet degranulation, angiogenesis, and vascular development were found to be enriched in BP, according to GO analysis. Additionally, we found that CC was associated with an enrichment of platelet alpha granules, cell-cell junctions, focal adhesion, and cell-substrate junctions. Furthermore, MF was discovered to have an enrichment of extracellular matrix binding, growth factor binding, integrin binding, and chemorepellent activity (Fig. 4A). More significantly, the KEGG data showed that the hub genes had a higher probability of enrichment in terms of PI3K-Akt signaling pathway, HIF-1 signaling pathway, lipid and atherosclerosis (Fig. 4B).

Fig. 4.

Gene Ontology (GO) functions (A) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (B) enrichment analyses of candidate genes.

Seven genes—ABCG2, FAM20C, ELL2, MTHFD2, ANKRD6, GLRX, and CDCP1—among the candidate 432 genes were found to be hub genes and potential immune-related diagnostic biomarkers of TAAD using the LASSO algorithm (Fig. 5A,B). Seven hub genes were shown to be strongly correlated with macrophages, Th17, B cells, iTregs, and natural killer cells (Fig. 5C). Immune cells were also found to differ significantly in TAAD. Using the GSE52093 dataset, the discriminating ability of all seven genes was further confirmed. Then the seven optimal genes were used to construct the diagnostic model. The diagnostic model in TAAD diagnosis with the area under ROC (AUC) was 0.996, and the sensitivity was 99.21%, the specificity was 98.67%. The ROC curve analysis demonstrated a strong discriminatory ability for TAAD (Fig. 5D), indicating that all seven genes could serve as potential biomarkers and are implicated in inflammation and immune cell infiltration in TAAD.

Fig. 5.

Identification and verification of hub-genes. (A) The coefficient profile plot is generated based on the log (lambda) sequence. (B) Least absolute shrinkage and selection operator (LASSO) coefficients for seven hub genes. (C) Correlation between hub genes and 24 types of immune cells. (D) External validation of the diagnostic potential of each hub gene based on the GSE52093 dataset. AUC, the area under receiver operating characteristic.

Three hub genes (ABCG2, FAM20C, MTHFD2) had significantly higher mRNA expression levels in the TAAD group compared to the Control group (p 0.05), according to RT-qPCR data. Nevertheless, there was a significant downregulation of one hub gene (CDCP1) in the TAAD group (p 0.05). The two groups’ changes were particularly pronounced in ABCG2, FAM20C, MTHFD2, and CDCP1 (Fig. 6A). Thus, in the tests that followed, we concentrated on confirming the connection between these four hub genes and TAAD. By using Western blotting and immunohistochemistry, we were able to further elucidate the differences in the protein expression levels of ABCG2, FAM20C, MTHFD2, and CDCP1 between the Control and TAAD groups. Western blotting results indicated that CDCP1 was lowered (p 0.05), whereas the levels of protein expression of ABCG2, FAM20C, and MTHFD2 (Fig. 6B,C). The results of immunohistochemistry were consistent with those of the Western blot (Fig. 6D).

Fig. 6.

The relationship between hub genes and TAAD in clinical samples is validated. (A) The mRNA level of seven hub genes in the Control and TAAD groups. (B,C) Western blotting and quantitative analysis display ABCG2, FAM20C, MTHFD2, and CDCP1 protein levels in aortic tissues. (D) Immunohistochemistry analysis displaying ABCG2, FAM20C, MTHFD2, and CDCP1 protein levels in aortic tissues. Scale bars = 100 μm. Compared with the Control group, *p 0.05, **p 0.01, ***p 0.001.

We created TAAD rat models with $\beta$-Aminopropionitrile to confirm the link between these four hub genes and TAAD. During modeling, the TAAD group had considerably lower body weight than the Control group (p 0.05) (Fig. 7A). qPCR analysis of aorta samples revealed higher mRNA expression levels for ABCG2, FAM20C, and MTHFD2 (p 0.05) and decreased CDCP1 (p 0.05) in the TAAD group compared to the Control group (Fig. 7B). Compared to the Control group, the TAAD group had significantly lower EF% and FS% (p 0.001) and higher AO (p 0.001). LCA was increased in the TAAD group, although not significantly compared to the Control group (Fig. 7C,D). We further analyzed the correlation between four hub genes (ABCG2, FAM20C, MTHFD2, and CDCP1) and cardiac function in the TAAD group. The results showed the number of PCR cycles of ABCG2, FAM20C, MTHFD2 and CDCP1 both had significant positive correlations with EF%, FS% and AO (p 0.001) (Fig. 7E). In summary, the upregulation of ABCG2, FAM20C, and MTHFD2 and the downregulation of CDCP1 in aorta tissues were closely linked to cardiac and vascular function decline.

Fig. 7.

The relationship between ABCG2, FAM20C, MTHFD2, CDCP1 mRNA levels, and TAAD in rats model. (A) The body weight in the Control and TAAD group. (B) ABCG2, FAM20C, MTHFD2, CDCP1 mRNA levels in the Control and TAAD group. (C) The echocardiography features in the Control and TAAD groups. (D) The correlations between four optimal hub genes ABCG2, FAM20C, MTHFD2, CDCP1 mRNA levels, and coronary artery functional parameters in the Control and TAAD group. (E) Correlation analysis between four hub genes and cardiovascular function. Compared with the Control group, *p 0.05, **p 0.01, ***p 0.001.

The Western blotting results indicated that the protein expression levels of ABCG2, FAM20C, and MTHFD2 were increased, and CDCP1 was decreased in the TAAD group compared to the Control group. The findings were in line with the results obtained from mRNA analysis, with a statistical significance of p 0.05 (Fig. 8A,B). The immunohistochemistry results revealed an irregular arrangement of endothelial cells in the vascular intima, partial damage to the intima, and an increase in the thickness of the vascular media. Additionally, the protein expression levels of ABCG2, FAM20C, MTHFD2, and CDCP1 were found to be in agreement with those observed in western blotting (Fig. 8C).

Fig. 8.

The relationship between ABCG2, FAM20C, MTHFD2, CDCP1 protein levels, and TAAD in rats model. (A,B) Protein levels of ABCG2, FAM20C, MTHFD2, and CDCP1 by western blotting and quantitative analysis in aortic tissues in Control and TAAD groups. (C) Immunostaining of ABCG2, FAM20C, MTHFD2, CDCP1 protein expression. Scale bars = 50 μm. Compared with Control group, *p 0.05, **p 0.01, ***p 0.001.