Abstract
Atherosclerosis (AS) is the pathologic basis of various cardiovascular and cerebrovascular events, with a high degree of heterogeneity among different arterial beds. However, mechanistic differences between arterial beds remain unexplored. The aim of this study was to explore key genes and potential mechanistic differences between AS in different arterial beds through bioinformatics analysis. Carotid atherosclerosis (CAS), femoral atherosclerosis (FAS), infrapopliteal atherosclerosis (IPAS), abdominal aortic atherosclerosis (AAS), and AS-specific differentially expressed genes (DEGs) were screened from the GSE100927 and GSE57691 datasets. Immune infiltration analysis was used to identify AS immune cell infiltration differences. Unsupervised cluster analysis of AS samples from different regions based on macrophage polarization gene expression profiles. Weighted gene co-expression network analysis (WGCNA) was performed to identify the most relevant module genes with AS. Hub genes were then screened by LASSO regression, SVM-REF, and single-gene differential analysis, and a nomogram was constructed to predict the risk of AS development. The results showed that differential expression analysis identified 5, 4, 121, and 62 CAS, FAS, IPAS, AAS-specific DEGs, and 42 AS-common DEGs, respectively. Immune infiltration analysis demonstrated that the degree of macrophage and mast cell enrichment differed significantly in different regions of AS. The CAS, FAS, IPAS, and AAS could be distinguished into two different biologically functional and stable molecular clusters based on macrophage polarization gene expression profiles, especially for cardiomyopathy and glycolipid metabolic processes. Hub genes for 6 AS (ADAP2, CSF3R, FABP5, ITGAX, MYOC, and SPP1), 4 IPAS (CLECL1, DIO2, F2RL2, and GUCY1A2), and 3 AAS (RPL21, RPL26, and RPL10A) were obtained based on module gene, gender stratification, machine learning algorithms, and single-gene difference analysis, respectively, and these genes were effective in differentiating between different regions of AS. This study demonstrates that there are similarities and heterogeneities in the pathogenesis of AS between different arterial beds.
Similar content being viewed by others
Introduction
Atherosclerosis (AS) is a common clinical vascular pathology, which starts from the intima and is characterized by localized lipid accumulation, fibrous tissue proliferation, and calcium deposits, and ultimately plaque formation. The development of AS primarily occurs in large and medium-sized arteries, such as carotid arteries, femoral arteries, and abdominal aorta1. However, studies have indicated that deceased patients without cardiovascular events are more susceptible to AS in the abdominal aorta2; smoking accelerated AS and plaque formation in the abdominal aorta in adolescents without affecting right coronary artery3; diabetes significantly accelerated plaque formation in the lower limb arteries, and hypertension appeared to occur preferentially occur in the carotid arteries4; and superficial femoral atherosclerosis develops more slowly than coronary and carotid arteries5. In addition, carotid atherosclerosis (CAS) is more likely to cause ischemic cerebrovascular events; coronary atherosclerosis is associated with an increased risk of ischemic heart disease; lower extremity atherosclerosis tends to result in lower extremity arterial occlusion; whereas abdominal aortic atherosclerosis (AAS) seems to contribute to abdominal aortic aneurysms as well as renal and lower extremity blood supply deficiencies. These findings suggest significant differences in the progression of AS between individuals, with different arterial beds exhibiting heterogeneity6,7. The formation of AS involves the mechanisms of lipid deposition, inflammation, and immune response. Nevertheless, femoral atherosclerosis (FAS) shows higher levels of ossification and calcification but lower cholesterol concentrations than carotid atherosclerosis8. Superficial femoral artery plaques have a more fibrous component and less lipid compared to coronary arteries9. This also demonstrates the heterogeneity of AS mechanisms between different arterial beds.
To better understand the expression differences of AS in various arterial beds, this study focused on CAS, FAS, infrapopliteal atherosclerosis (IPAS), and AAS. First, differential expression analysis was conducted to identify genes that distinguish the AS groups from normal samples. Subsequently, differential expression analysis was performed to identify genes that differentiate AS in different arterial beds. Finally, machine learning and single-gene differential analysis were employed to screen for hub genes, and a nomogram was constructed to predict the risk of morbidity.
Materials and methods
Subjects and dataset acquisition
The entire study process is depicted in Fig. 1. Two gene expression profiles (GSE10092710 and GSE5769111) were retrieved from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). Among them, GSE100927 included 12 carotid normal and 29 CAS tissues; 12 femoral normal and 26 FAS tissues; 11 infrapopliteal normal and 14 IPAS tissues; and GSE57691 included 10 abdominal aortic normal and 9 AAS tissues.
Identification of differentially expressed genes (DEGs)
The GSE100927 and GSE57691 datasets were merged and standardized using the “Affy” R package12, while batch effects were removed with the “SVA” R package13. DEGs were identified by comparing disease and control groups using the “limma” R package, with the criteria set at P < 0.05 and |logFC| > 1.0 for DEG selection14. Macrophage polarization-related genes were obtained from the GeneCards (http://www.genecards.org/) database with a “relevance score” > 1015.
Gene Set Enrichment Analysis (GSEA)
The “GSEA” package was used to explore the related pathways of specific genes and to calculate the correlation between specific genes and other genes in R. All genes were then sorted from highest to lowest according to their correlation, and these sorted genes were the set of genes to be tested. The KEGG signaling pathway set was called a “predefined set” to detect its enrichment in the gene set.
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis
Genes were imported into the David database (https://david.abcc.ncifcrf.gov/) for GO and KEGG analysis, including biological processes, cellular components, molecular functions, and signaling pathways, and the species was limited to “Homo sapiens,” with a P < 0.05 as the screening condition16,17.
Immune infiltration analysis and correlation analysis
The degree of infiltration of 22 immune cells was quantified using the CIBERSORT deconvolution algorithm based on gene microarray data18. CIBERSORTx (https://cibersortx.stanford.edu/) was used to analyze immune cell infiltration for each sample19. Differences between the two groups were compared using the Wilcox test.
Unsupervised clustering analysis of AS samples from different regions
Unsupervised clustering analysis of AS samples based on macrophage polarization expression profiles was performed using the “ConsensusClusterPlus” R package20. The AS samples from different regions were grouped by applying the k-means algorithm with 1000 iterations and k = 9. The appropriate number of clusters was determined based on the cumulative distribution function curve, consensus matrix, and consistent cluster score (> 0.9)20.
Gene Set Variation Analysis (GSVA)
The “GSVA” package was used to conduct a GSVA enrichment analysis for different clusters, considering a significant change if the |t value of the GSVA score| was greater than two21.
Weighted Gene Co-expression Network Analysis (WGCNA)
WGCNA was employed to identify co-expression modules by clustering the samples using the “WGCNA” R package22. Outlier samples were removed, and a co-expression network for the gene expression matrices of the remaining samples was constructed. Appropriate soft thresholds were determined to identify gene modules and select the most relevant ones.
Machine learning models screening for Hub genes
The candidate Hub genes were screened with the Support Vector Machine-Recursive Feature Elimination (SVM-RFE) algorithm and Least Absolute Shrinkage and Selection Operator Regression (LASSO) regression analysis. The LASSO algorithm was used to select the variables and adjust the complexity23. The SVM-RFE algorithm analyzes the most appropriate key genes24.
Single-gene differential analysis
Hub genes were identified by comparing the expression differences of candidate Hub genes in different groups by single gene differential analysis. The one-way analysis of variance (ANOVA) was performed between multiple groups using the “car” R package. One-way ANOVA was used for normal distribution with homogeneous variance, Welch's ANOVA for heterogeneous variance, and Bonferroni's method was used to correct for pairwise comparisons. The Kruskal-Walli’s test was used for non-normal distributions, and the Wilcox rank sum test was used for pairwise comparisons.
Construction and validation of a nomogram model
A nomogram model was established using the “rms” R package to predict the probability of the occurrence of AS in different regions, and its predictive power was estimated by using calibration curves and decision curve analysis. The area under the curve (AUC) of the subject operating characteristic (ROC) was then calculated for each key gene to test the diagnostic efficacy of the Hub genes.
Statistical analysis
All statistical analyses were performed using R software, and P < 0.05 was considered significant.
Results
Identification of DEGs
The differential expression analysis of CAS, FAS, IPAS, AAS, and AS identified 229, 104, 35, 119, and 159 DEGs, respectively (Fig. 2A,B). Firstly, the relevant pathogenesis of AS was explored, and functional enrichment analysis showed that its pathogenesis is closely related to inflammatory response, cell migration and chemotaxis (chemotaxis of neutrophils, monocytes, lymphocytes, and macrophages), cell death, lipid metabolism, and immune response (Fig. 2C). Then, enrichment analysis showed that CAS was associated with cell differentiation (osteoclast, dendritic, and myeloid cell differentiation), ion transfer (membrane repolarization during atrial cardiac muscle cell action potential, potassium ion transport, regulation of cellular calcium homeostasis); FAS was enriched in cell chemotaxis-related pathways; IPAS was involved in the regulation of various interleukin factors (IL-10, IL-12, IL-1β, IL-8); and AAS was mainly involved in cellular redox reactions (Fig. 2D). In order to explore the potential specific pathogenesis of AS in different regions, 5 CAS-specific DEGs, 4 FAS-specific DEGs, 121 IPAS-specific DEGs, 62 AAS-specific DEGs, and 42 common DEGs for AS were obtained by Venn diagram, respectively (Fig. 2E). The enrichment analysis of these genes showed that IPAS is closely associated with interleukin factor regulation and osteoclast differentiation; AAS is closely related to inflammatory response (IL-17 signaling pathway, chemokine signaling pathway) and cell migration (neutrophil chemotaxis, focal adhesion, actin filament binding) (Fig. 2F).
To further understand the immune microenvironment of AS in different regions, the results showed that AS had a higher abundance of memory B cells, M0 macrophages, and activated mast cells, and lower abundance of naive B cells, resting memory CD4 T cells, and resting mast cells (P < 0.05); CAS had a higher abundance of memory B cells, M0 macrophages, and activated mast cells, and lower abundance of naive B cells, resting memory CD4 T cells, monocytes, and resting mast cells (P < 0.05); FAS had a higher abundance of memory B cells and activated mast cells, and lower abundance of resting memory CD4 T cells and resting mast cells (P < 0.05); IPAS had a higher abundance of M0 macrophages and activated mast cells, had lower abundance of resting memory CD4 T cells, resting mast cells, and neutrophils (P < 0.05); and AAS had a lower abundance of M1 macrophages (P < 0.05) (Fig. 3A). It is apparent that macrophages and mast cells are differently enriched in different regions of the AS, and we believe that macrophage polarization plays an important role.
Identification of macrophage polarization molecular clusters
The GeneCards database was searched to obtain 182 macrophage polarization-related genes with a “relevance score” > 10, and 23, 17, 37, and 19 intersected genes were obtained by intersecting with the specific DEGs of CAS, FAS, IPAS, and AAS, respectively. To explore the role of macrophage polarization on AS in different regions, unsupervised cluster analysis was performed on 29 CAS samples, 26 FAS samples, 14 IPAS samples, and 9 AAS samples based on the expression profiles of these genes, respectively (Fig. 3B). The results showed that CAS, FAS, IPAS, and AAS all resulted in two distinct and stable groups, with CAS divided into the C1 cluster (n = 10) and the C2 cluster (n = 19); FAS divided into the C3 cluster (n = 14) and the C4 cluster (n = 12); IPAS divided into the C5 cluster (n = 10) and the C6 cluster (n = 4); and AAS into the C7 cluster (n = 6) and the C8 cluster (n = 3) (Fig. 3C). The immune infiltration analysis showed that the C1 cluster of CAS had a higher abundance of resting memory CD4 T cells, activated NK cells, and neutrophils, while the C2 cluster had a higher abundance of γ-δ T cells and M0 macrophages; the C3 cluster of FAS had a higher abundance of plasma cells, resting memory CD4 T cells, and M1 macrophages, while the C4 cluster had a higher abundance of follicular helper T cells and M0 macrophages; the C6 cluster of IPAS had a higher abundance of resting mast cells; and the C8 cluster of AAS had a higher abundance of M2 macrophages (Fig. 3D).
The GSVA results showed that the C1, C3, and C7 clusters are mainly involved in various cardiomyopathies and cancers and amino acid metabolism-related pathways; the C2, C4, C5, and C8 clusters are mainly involved in various glycolipid metabolism, biosynthesis, metabolic processes, and immune disorders; whereas the C6 cluster is involved in cell adhesion-related pathways (Fig. 3E). These findings suggest that, based on macrophage polarization, AS samples from different regions can be distinguished into two subgroups with significantly different biological functions, particularly biosynthetic and metabolic processes, immune responses, and cardiomyopathies.
WGCNA and screening of gene modules
WGCNA was used to screen for co-expressed modules most relevant to AS and macrophage polarization. WGCNA was first performed on different regional AS samples and control samples to obtain the most relevant modules for AS. The results showed that 3, 4, 4, 4, and 3 co-expression modules with CAS, FAS, IPAS, AAS, and AS were obtained at soft thresholds of 8, 8, 4, 4, and 6, respectively. The “turquoise” module showed a significantly positive correlation with all of them; thus, 3671, 3728, 3873, 3187, and 815 module genes related to CAS, FAS, IPAS, AAS, and AS were obtained, respectively (Fig. 4A–C).
Identification of Hub genes
First, AS-DEGs, module genes of AS, and common DEGs were intersected to obtain 28 candidate Hub genes for AS. Then the intersection of CAS, FAS, IPAS, and AAS-specific DEGs with module genes to obtain candidate Hub genes. In order to generalize the diagnostic efficacy of Hub genes, the samples were stratified by gender to explore whether there were differences in the expression of specific DEGs. The results showed that 939 DEGs for CAS, 463 DEGs for FAS, and 300 DEGs for IPAS were identified in males (P < 0.05 & |logFC| > 1.0), and 792 DEGs for CAS, 529 DEGs for FAS, and 417 DEGs for IPAS were identified in females (P < 0.05 & |logFC| > 1.0), and that 22, 1, 11, and 53 candidate Hub genes for CAS, FAS, IPAS, and AAS were identified by intersecting with the above genes, respectively (Fig. 5A). Therefore, Hub genes from different regions of AS and AS were used for subsequent screening.
To further screen Hub genes and reduce the bias of diagnostic models, LASSO regression and SVM-REF were used to screen the characterized key genes. The results showed that LASSO regression screened 10, 6, 6, 3, and 6 featured genes of AS, CAS, FAS, IPAS, and AAS (Fig. 5B); SVM-REF screened 16, 11, 9, 4, and 2 key genes of AS, CAS, FAS, IPAS, and AAS (Fig. 5C), and ultimately obtained 6, 6, 6, 3, and 2 Hub genes of AS (ADAP2, CSF3R, FABP5, ITGAX, MYOC, and SPP1), CAS (GSTM5, ADAP2, C16orf89, FABP5, LONRF2, and SOST), FAS (DOK7, CSF3R, FABP5, ITLN1, SAA1, and SPP1), IPAS (ADAP2, FABP5, and MYOC), and AAS (RPL21 and RPL26) (Fig. 5D).
Then, the expression differences of these genes among different AS were verified by single-gene differential analysis, which showed that ADAP2, CSF3R, FABP5, ITGAX, MYOC, and SPP1 of the AS samples were significantly different from the other AS; RPL21 and RPL26 of the AAS samples were significantly lower than those of the other groups (Fig. 5E). Thus, we obtained 6 Hub genes for AS (ADAP2, CSF3R, FABP5, ITGAX, MYOC, and SPP1) and 2 Hub genes for AAS (RPL21 and RPL26).
In order to obtain more potential Hub genes, differential expression analysis was also performed on different AS samples, and 1, 4, 19, and 689 potential Hub genes (P < 0.05 & |logFC|> 0.25) were identified for CAS, FAS, IPAS, and AAS, respectively (Fig. 6A). To minimize overfitting, LASSO regression and SVM-REF were performed on the potential Hub genes of IPAS and AAS to obtain 6 potential Hub genes of IPAS and 6 potential Hub genes of AAS, respectively (Fig. 6B,D). Then, four Hub genes for IPAS (CLECL1, DIO2, F2RL2, and GUCY1A2) and one Hub gene for AAS (RPL10A) were obtained by single-gene difference analysis (Fig. 6E).
Finally, six Hub genes for AS (ADAP2, CSF3R, FABP5, ITGAX, MYOC, and SPP1) and four Hub genes for IPAS (CLECL1, DIO2, F2RL2, and GUCY1A2) and three Hub genes for AAS (RPL21, RPL26, and RPL10A) were obtained by differential expression analysis between AS and normal samples and between samples with different AS (Fig. 6F).
Construction and evaluation of models
To further assess the predictive efficiency of the SVM model, a nomogram was constructed to estimate the risk of occurrence in AS and different regions of AS (Fig. 7A), and then the predictive efficiency of the nomogram was assessed using calibration curve and decision curve analysis (Fig. 7B,C). The ROC curves demonstrated that all of these genes had satisfactory diagnostic efficacy in distinguishing between normal and AS (ADAP2, AUC = 0.8601; CSF3R, AUC = 0.8145; FABP5, AUC = 0.8581; ITGAX, AUC = 0.8239; MYOC, AUC = 0.8464; SPP1, AUC = 0.7923), IPAS (CLECL1, AUC = 0.8429; DIO2, AUC = 0.8683; F2RL2, AUC = 0.8190; GUCY1A2, AUC = 0.8937), and AAS (RPL21, AUC = 1; RPL26, AUC = 1; RPL10A, AUC = 1) samples from different regions (Fig. 7D).
Discussion
AS commonly occurs in systemic arteries and exhibits certain similarities across different arterial beds. In our study, we initially conducted a differential expression analysis comparing AS samples with control samples, and the results demonstrated significant associations between AS pathogenesis and inflammatory response, cell migration and chemotaxis, cell death, lipid metabolism, and immune response (Fig. 2C), all of which were confirmed25. Subsequently, six hub genes for AS were identified, which effectively distinguished between normal and AS samples, among which ADAP2, CSF3R, FABP5, ITGAX, and SPP1 were highly expressed in AS, whereas MYOC expression was downregulated (Fig. 5E).
ADAP2 encodes ArfGAP with the double PH domain 2 protein, which plays an important role in cardiac development. Cardiovascular malformations caused by patients with NF1 microdeletions syndrome are associated with ADAP2 dysfunction26, and its high expression is seen in patients with myocardial infarction27.
CSF3R is essential for granulocyte maturation and is primarily involved in the regulation of inflammation. The expression of colony-stimulating factor and its receptor is up-regulated in patients with multiple stages of acute myocardial infarction and stable angina pectoris, suggesting that adhesion, chemotaxis, and phagocytosis of neutrophils and monocyte macrophages are enhanced with the development of AS and acute myocardial infarction28. In addition, the CSF3/CSF3R axis is involved in lipid metabolism and promotes the development of nonalcoholic fatty liver disease and AS29,30.
FABP5 plays a role in the transportation, regulation, and metabolism of lipids. Its high expression in AS has been demonstrated, which is not only considered a potential biomarker for residual risk of AS but also a sensitive marker for lipid-rich macrophages on the luminal side of AS lesions31,32,33.
ITGAX is involved in early AS lesion formation and is highly expressed in AS tissues, and several bioinformatic analyses have suggested it as a potential biomarker for AS34,35. The mice with double knockouts of ITGAX and Apoe demonstrated lower monocyte adhesion and AS plaque macrophage content36. ITGAX was found to promote AS progression by regulating B differentiation through the TNF-α signaling pathway37.
MYOC is closely associated with glaucoma and has high expression levels in skeletal muscle and hearts in addition to the eye38. However, the relationship between MYOC and AS remains unclear, and its secreted glycoproteins may influence AS by regulating processes such as cell adhesion, cytoskeleton, and migration39,40.
SPP1 exhibits high expression in endothelial cells, macrophages, and vascular smooth muscle cells present in atherosclerotic plaques41,42. The level of SPP1 expression is closely associated with the development and severity of coronary artery disease43. Additionally, it has been identified as a potential marker for monitoring the severity of AS and predicting cardiovascular event-related mortality44.
Currently, there are several serum markers used to identify early AS, such as lipoproteins, ceramides, C-reactive proteins, high-sensitivity C-reactive proteins, matrix metalloproteinases, and miRNAs, among which high-sensitivity C-reactive proteins are comprehensively validated in the prediction of atherosclerotic cardiovascular events45,46,47. In our study, we identified six Hub genes for AS, and the nomogram demonstrated satisfactory diagnostic efficacy. These genes may supplement the diagnostic biomarkers for early AS.
Then, we conducted differential expression analysis on different types of AS and control groups to obtain DEGs and specific DEGs, and enrichment analyses showed that CAS was related to cellular differentiation, calcium and potassium homeostasis; FAS with cellular chemotaxis; IPAS with osteoclast differentiation and regulation of interleukin factors; and AAS with inflammation and cell migration (Fig. 2D–F). Previous studies reported higher foam cell lesions and lipid plaques in CAS and higher calcification and lower cholesterol concentrations in FAS8,48. In our study, we found that CAS is involved in the regulation of pathways related to calcium homeostasis, which may explain why CAS has a lower degree of calcification.
In addition, we identified four Hub genes for IPAS (CLECL1, DIO2, F2RL2, and GUCY1A2) and three Hub genes for AAS (RPL21, RPL26, and RPL10A) after high-intensity screening. The CLECL1 and F2RL2 expressions were significantly up-regulated in IPAS compared to control and other types of AS, while DIO2 and GUCY1A2 expressions were down-regulated; the RPL21, RPL26, and RPL10A expressions were significantly down-regulated in AAS compared to control and other AS.
CLECL1 plays an important role in the regulation of the immune response, inducing phosphorylation of p44/42 and JNK MAPK as well as up-regulation of MHC class II expression, resulting in a partially mature dendritic cell phenotype49; it also serves as a T-cell co-stimulatory molecule to bias the Th2 response by enhancing IL-4 secretion from CD4(+) T cells50.
F2RL2 mediates thrombin-induced proliferation, migration, and matrix biosynthesis, and the production of inflammatory and pro-growth mediators, which contribute to the progression of inflammatory diseases51,52.
DOI2 is mainly present in brown adipose tissue, placenta, pituitary and muscle and is responsible for deiodinating T4 to T353. The upregulation of DIO2 gene expression was observed in ApoE mice with hyperlipidemia and AS in Western diets, indicating that DIO2 plays a certain role in the development of AS54.
GUCY1A2 is the active subunit of guanylate cyclase, and deficiency of guanylate cyclase function contributes to the formation of AS plaques in LDL-/- mice, which may be associated with increased adhesion of leukocytes to the endothelium55,56. GUCY1A2 also protects against cerebral ischemia perfusion injury by mitigating apoptosis and mitochondrial damage57.
RPL21, RPL26, and RPL10A are important components of the large 60S ribosomal subunit with functions involved in cell proliferation, differentiation, apoptosis, and DNA repair, which have been associated with metabolism and cardiovascular disease progression58. However, the mechanism of its specific involvement in AAS remains undefined, with some studies suggesting a possible association with inflammation and mitochondrial apoptosis59.
The immune infiltration analysis revealed significant differences between macrophages and mast cells between AS and control samples. Therefore, we concluded that macrophage polarization plays an important role in the progression of AS. We then classified AS samples based on macrophage polarization-related genes and obtained different molecular clusters, which showed significant biofunctional differences between biosynthesis metabolism and cardiomyopathy. M1 macrophages produce pro-inflammatory factors that are involved in the clearance of pathogens, while M2 macrophages produce anti-inflammatory factors that promote tissue repair and fibrosis. In the early stages, M2 macrophages are predominantly present in the plaques and stabilize them. As the lesions progress, the number of M2 macrophages decreases, the number of M1 macrophages gradually increases, and the secretion of pro-inflammatory factors increases, which promotes the development of AS and renders the plaques susceptible to rupture60. In addition, we found that AAS has a lower abundance of M1 macrophages, which may be related to the involvement of anti-inflammatory-related pathways (Th1 and Th2 cell differentiation, Th17 cell differentiation, IL-17 signaling pathway).
We found that AS has a higher abundance of activated mast cells, which were found to promote AS and plaque development by stimulating endothelial cells to express adhesion molecules, recruiting neutrophils and circulating leukocytes to plaques, secreting tryptase to disrupt endothelial integrity, and secreting histamine to alter microvascular permeability61.
To summarize, we obtained DEGs and specific DEGs by differential expression analysis among different samples, and then, after screening by multiple methods of gender stratification, key module genes, machine learning models, and single-gene differential analysis, we finally identified 6 Hub genes for AS, 4 Hub genes for IPAS, and 3 Hub genes for AAS. Immune infiltration analysis revealed that AS had a higher abundance of M0 macrophages and activated mast cells, while AAS had a lower abundance of M1 macrophages. Functional enrichment analysis indicated that CAS was involved in the regulation of calcium homeostasis; IPAS was involved in ossification; and AAS was involved in inflammation-related pathways.
However, there are still some limitations to our study. First, we did not investigate other large arteries such as cerebral arteries, coronary arteries, and thoracic aortas due to missing datasets, and we were unable to merge more datasets for our study. Secondly, we did not perform prognostic analysis. AS as a non-oncologic disease, the GEO database seldom contains the basic information and prognostic information of the patients. Due to the disease selection and database limitations, we were unable to perform the follow-up study for prognostic analysis. Finally, we identified the Hub genes of IPAS and AAS, which show satisfactory diagnostic efficacy, but the specific mechanisms of Hub genes in IPAS and AAS remain unclear.
Conclusions
There are similarities and heterogeneities in the pathogenesis of AS between different arterial beds.
Data availability
The following information was supplied regarding data availability: Data is available at NCBI GEO: GSE100927 and GSE57691. Additionally, any analytic technology-related questions can be directly contacted by the corresponding author.
Abbreviations
- AS:
-
Atherosclerosis
- AAS:
-
Abdominal aortic atherosclerosis
- AUC:
-
Area under curve
- CAS:
-
Carotid Atherosclerosis
- DEGs:
-
Differentially expressed genes
- FAS:
-
Femoral atherosclerosis
- GEO:
-
Gene Expression Omnibus Database
- GO:
-
Gene ontology
- GSEA:
-
Gene set enrichment analysis
- GSVA:
-
Gene set variation analysis
- KEGG:
-
Kyoto Encyclopedia of Genes and Genomes
- IPAS:
-
Infrapopliteal Atherosclerosis
- LASSO:
-
Least Absolute Shrinkage and Selection Operator Regression
- ROC:
-
Receiver Operating Characteristic
- SVM-RFE:
-
Support Vector Machine-Recursive Feature Elimination
- WGCNA:
-
Weighted gene co-expression network analysis
References
Libby, P. The changing landscape of atherosclerosis. Nature. 592(7855), 524–533 (2021).
Roberts, J. C., Moses, C. & Wilkins, R. H. Autopsy studies in atherosclerosis: I. Distribution and severity of atherosclerosis in patients dying without morphologic evidence of atherosclerotic catastrophe. Circulation. 20(4), 511–519 (1959).
McGill, H. C. J. et al. Associations of coronary heart disease risk factors with the intermediate lesion of atherosclerosis in youth. The Pathobiological Determinants of Atherosclerosis in Youth (PDAY) Research Group. Arterioscler. Thromb. Vasc. Biol. 20(8), 1998–2004 (2000).
Danese, C. et al. Do hypertension and diabetes mellitus influence the site of atherosclerotic plaques?. Clin. Ter. 157(1), 9–13 (2006).
Dalager, S., Falk, E., Kristensen, I. B. & Paaske, W. P. Plaque in superficial femoral arteries indicates generalized atherosclerosis and vulnerability to coronary death: An autopsy study. J. Vasc. Surg. 47(2), 296–302 (2008).
Sulkava, M. et al. Differentially expressed genes and canonical pathway expression in human atherosclerotic plaques—Tampere Vascular Study. Sci. Rep. 7, 41483 (2017).
Levula, M. et al. Genes involved in systemic and arterial bed dependent atherosclerosis–Tampere Vascular study. PLoS One. 7(4), e33787 (2012).
Herisson, F. et al. Carotid and femoral atherosclerotic plaques show different morphology. Atherosclerosis. 216(2), 348–354 (2011).
Poredoš, P., Cevc, M. & Blinc, A. Characteristics of atherosclerosis in femoropopliteal artery and its clinical relevance. Atherosclerosis. 335, 31–40 (2021).
Steenman, M. et al. Identification of genomic differences among peripheral arterial beds in atherosclerotic and healthy arteries. Sci. Rep. 8(1), 3940 (2018).
Biros, E. et al. Differential gene expression in human abdominal aortic aneurysm and aortic occlusive disease. Oncotarget. 6(15), 12984–12996 (2015).
Gautier, L., Cope, L., Bolstad, B. M. & Irizarry, R. A. Affy-analysis of Affymetrix GeneChip data at the probe level. Bioinformatics. 20(3), 307–315 (2004).
Leek, J. T., Johnson, W. E., Parker, H. S., Jaffe, A. E. & Storey, J. D. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 28(6), 882–883 (2012).
Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 43(7), e47 (2015).
Stelzer, G. et al. The GeneCards Suite: From gene data mining to disease genome sequence analyses. Curr. Protoc. Bioinform. 54, 1.30.31-31.30.33 (2016).
Sherman, B. T. et al. DAVID: A web server for functional enrichment analysis and functional annotation of gene lists (2021 update). Nucleic Acids Res. 50(W1), W216–W221 (2022).
Kanehisa, M., Furumichi, M., Sato, Y., Kawashima, M. & Ishiguro-Watanabe, M. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. 51(D1), D587–D592 (2023).
Chen, B., Khodadoust, M. S., Liu, C. L., Newman, A. M. & Alizadeh, A. A. Profiling tumor infiltrating immune cells with CIBERSORT. Methods Mol. Biol. 1711, 243–259 (2018).
Steen, C. B., Liu, C. L., Alizadeh, A. A. & Newman, A. M. Profiling cell type abundance and expression in bulk tissues with CIBERSORTx. Methods Mol. Biol. 2117, 135–157 (2020).
Wilkerson, M. D. & Hayes, D. N. ConsensusClusterPlus: A class discovery tool with confidence assessments and item tracking. Bioinformatics. 26(12), 1572–1573 (2010).
Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: Gene set variation analysis for microarray and RNA-seq data. BMC Bioinform. 14, 7 (2013).
Langfelder, P. & Horvath, S. WGCNA: An R package for weighted correlation network analysis. BMC Bioinform. 9, 559 (2008).
Huang, H. Controlling the false discoveries in LASSO. Biometrics. 73(4), 1102–1110 (2017).
Lin, X. et al. Selecting feature subsets based on SVM-RFE and the overlapping ratio with applications in bioinformatics. Molecules. 23(1), 52 (2017).
Libby, P. Inflammation and the pathogenesis of atherosclerosis. Vascul. Pharmacol. 154, 107255 (2023).
Venturin, M. et al. ADAP2 in heart development: A candidate gene for the occurrence of cardiovascular malformations in NF1 microdeletion syndrome. J. Med. Genet. 51(7), 436–443 (2014).
Osmak, G., Baulina, N., Koshkin, P. & Favorova, O. Collapsing the list of myocardial infarction-related differentially expressed genes into a diagnostic signature. J. Transl. Med. 18(1), 231 (2020).
Li, C. R. et al. Expression characteristics of neutrophil and mononuclear-phagocyte related genes mRNA in the stable angina pectoris and acute myocardial infarction stages of coronary artery disease. J. Geriatr. Cardiol. 12(3), 279–286 (2015).
Zhang, Y. et al. GCSF deficiency attenuates nonalcoholic fatty liver disease through regulating GCSFR-SOCS3-JAK-STAT3 pathway and immune cells infiltration. Am. J. Physiol. Gastrointest. Liver Physiol. 320(4), G531–G542 (2021).
Cimato, T. R., Palka, B. A., Lang, J. K. & Young, R. F. LDL cholesterol modulates human CD34+ HSPCs through effects on proliferation and the IL-17 G-CSF axis. PLoS One. 8(8), e73861 (2013).
Zheng, X. et al. Transcriptome-wide N(6)-methyladenosine methylation profile of atherosclerosis in mice. BMC Genom. 24(1), 774 (2023).
Xu, B. et al. The biological functions and regulatory mechanisms of fatty acid binding protein 5 in various diseases. Front. Cell Dev. Biol. 10, 857919 (2022).
Furuhashi, M. et al. Serum FABP5 concentration is a potential biomarker for residual risk of atherosclerosis in relation to cholesterol efflux from macrophages. Sci. Rep. 7(1), 217 (2017).
Yan, S. et al. Identification of ITGAX and CCR1 as potential biomarkers of atherosclerosis via Gene Set Enrichment Analysis. J. Int. Med. Res. 50(3), 3000605211039480 (2022).
Meng, Y. et al. Identification of potential key genes involved in the carotid atherosclerosis. Clin. Interv. Aging. 16, 1071–1084 (2021).
Wu, H. et al. Functional role of CD11c+ monocytes in atherogenesis associated with hypercholesterolemia. Circulation. 119(20), 2708–2717 (2009).
Wang, X. et al. PAX5/ITGAX contributed to the progression of atherosclerosis by regulation of B differentiation via TNF-α signaling pathway. DNA Cell Biol. 42(2), 97–104 (2023).
Fingert, J. H. et al. Characterization and comparison of the human and mouse GLC1A glaucoma genes. Genome Res. 8(4), 377–384 (1998).
Shen, X., Koga, T., Park, B. C., SundarRaj, N. & Yue, B. Rho GTPase and cAMP/protein kinase A signaling mediates myocilin-induced alterations in cultured human trabecular meshwork cells. J. Biol. Chem. 283(1), 603–612 (2008).
Goldwich, A., Scholz, M. & Tamm, E. R. Myocilin promotes substrate adhesion, spreading and formation of focal contacts in podocytes and mesangial cells. Histochem. Cell Biol. 131(2), 167–180 (2009).
Wolak, T. Osteopontin—A multi-modal marker and mediator in atherosclerotic vascular disease. Atherosclerosis. 236(2), 327–337 (2014).
O’Brien, E. R. et al. Osteopontin is synthesized by macrophage, smooth muscle, and endothelial cells in primary and restenotic human coronary atherosclerotic plaques. Arterioscler. Thromb. 14(10), 1648–1656 (1994).
Ohmori, R. et al. Plasma osteopontin levels are associated with the presence and extent of coronary artery disease. Atherosclerosis. 170(2), 333–337 (2003).
Momiyama, Y. et al. Associations between plasma osteopontin levels and the severities of coronary and aortic atherosclerosis. Atherosclerosis. 210(2), 668–670 (2010).
Bordeianu, G. et al. Circulating biomarkers for laboratory diagnostics of atherosclerosis-literature review. Diagnostics (Basel). 12(12), 3141 (2022).
Tibaut, M. et al. Markers of atherosclerosis: Part 1—Serological markers. Heart Lung Circ. 28(5), 667–677 (2019).
Della Corte, V., Todaro, F., Cataldi, M. & Tuttolomondo, A. Atherosclerosis and its related laboratory biomarkers. Int. J. Mol. Sci. 24(21), 15546 (2023).
Dalager, S., Paaske, W. P., Kristensen, I. B., Laurberg, J. M. & Falk, E. Artery-related differences in atherosclerosis expression: Implications for atherogenesis and dynamics in intima-media thickness. Stroke. 38(10), 2698–2705 (2007).
Ryan, E. J., Magaletti, D., Draves, K. E. & Clark, E. A. Ligation of dendritic cell-associated lectin-1 induces partial maturation of human monocyte derived dendritic cells. Hum. Immunol. 70(1), 1–5 (2009).
Ryan, E. J. et al. Dendritic cell-associated lectin-1: A novel dendritic cell-associated, C-type lectin-like molecule enhances T cell secretion of IL-4. J. Immunol. 169(10), 5638–5648 (2002).
Schrör, K. et al. Thrombin receptors in vascular smooth muscle cells—Function and regulation by vasodilatory prostaglandins. Thromb. Haemost. 103(5), 884–890 (2010).
Cupit, L. D., Schmidt, V. A. & Bahou, W. F. Proteolytically activated receptor-3. A member of an emerging gene family of protease receptors expressed on vascular endothelial cells and platelets. Trends Cardiovasc. Med. 9(1–2), 42–48 (1999).
Bates, J. M., St Germain, D. L. & Galton, V. A. Expression profiles of the three iodothyronine deiodinases, D1, D2, and D3, in the developing rat. Endocrinology. 140(2), 844–851 (1999).
Min, B. K. et al. Therapeutic effect of dichloroacetate against atherosclerosis via hepatic FGF21 induction mediated by acute AMPK activation. Exp. Mol. Med. 51(10), 1–12 (2019).
Mauersberger, C. et al. Loss of soluble guanylyl cyclase in platelets contributes to atherosclerotic plaque formation and vascular inflammation. Nat. Cardiovasc. Res. 1(12), 1174–1186 (2022).
Jamaiyar, A., Chen, J. & Feinberg, M. W. Soluble guanylyl cyclase in platelets keeps atherosclerosis at bay. Nat. Cardiovasc. Res. 1, 1124–1126 (2022).
Wang, F., Ma, Q., Dong, X., Wang, T. & Ma, C. circ-Gucy1a2 protects mice from cerebral ischemia-reperfusion injury by attenuating neuronal apoptosis and mitochondrial membrane potential loss. J. Invest. Surg. 36(1), 2152509 (2023).
Wang, W. et al. Ribosomal proteins and human diseases: Pathogenesis, molecular mechanisms, and therapeutic implications. Med. Res. Rev. 35(2), 225–285 (2015).
Xie, X. et al. Bioinformatics analysis reveals the potential diagnostic biomarkers for abdominal aortic aneurysm. Front. Cardiovasc. Med. 8, 656263 (2021).
Wu, J. et al. Macrophage polarization states in atherosclerosis. Front. Immunol. 14, 1185587 (2023).
Razeghian-Jahromi, I., Karimi Akhormeh, A., Razmkhah, M. & Zibaeenezhad, M. J. Immune system and atherosclerosis: Hostile or friendly relationship. Int. J. Immunopathol. Pharmacol. 36, 3946320221092188 (2022).
Funding
Science and Technology Research Project of Education Department of Liaoning Province (No. L202048).
Author information
Authors and Affiliations
Contributions
The paper's writing and content are the sole responsibility of W.X.Z. This study's design, data analysis, paper preparation, and revision all benefited greatly from the efforts of P.X., Z.Y., and P.J.X. K.J., Y.J.J. and C.Y.Y. were involved in the raw data analysis and original draft writing of this study. Q.C., G.L.H., and L.Y. proposed the study conceptualization and reviewed/edited the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Wu, X., Pan, X., Zhou, Y. et al. Identification of key genes for atherosclerosis in different arterial beds. Sci Rep 14, 6543 (2024). https://doi.org/10.1038/s41598-024-55575-8
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-55575-8
Keywords
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.