Integrative Analysis of Single-Cell Transcriptomics and Genetic Associations Identify Cell States Associated with Vascular Disease

Mark E Pepin; William Schwartzman; Shi Fang; Shamsudheen K Vellarikkal; Deepak S Atri; Ankith Reddy; Qiaohan Xu; Andrew R Hamel; Marie Billaud; Ayellet V Segrè; Rajat M Gupta

doi:10.1016/j.atherosclerosis.2025.119108

. Author manuscript; available in PMC: 2026 Mar 14.

Published in final edited form as: Atherosclerosis. 2025 Feb 3;403:119108. doi: 10.1016/j.atherosclerosis.2025.119108

Integrative Analysis of Single-Cell Transcriptomics and Genetic Associations Identify Cell States Associated with Vascular Disease

Mark E Pepin ^1,^2,³, William Schwartzman ^1,², Shi Fang ^1,², Shamsudheen K Vellarikkal ^1,², Deepak S Atri ^1,², Ankith Reddy ², Qiaohan Xu ⁴, Andrew R Hamel ^1,^4,⁵, Marie Billaud ⁶, Ayellet V Segrè ^1,^4,⁵, Rajat M Gupta ^1,²

PMCID: PMC12984030 NIHMSID: NIHMS2118170 PMID: 40120433

Abstract

Background:

Vascular diseases are accompanied by alterations in cellular phenotypes which underly disease pathogenesis, with single-cell technologies aiding in the discovery of cellular heterogeneity among endothelial cell (EC) and vascular smooth muscle cell (VSMC) populations. VSMCs are theorized to switch between contractile and synthetic phenotypes in atherosclerotic disease, though it is not known which vascular subpopulations and/or intermediate cell states contribute to early vascular dysfunction. In atherosclerotic disease, VSMCs are hypothesized to transition between contractile and synthetic states; however, the specific vascular subpopulations and intermediate cell states responsible for early vascular dysfunction remain unclear.

Methods:

We integrated newly generated and published single-nuclear RNA-sequencing (snRNA-seq) to analyze normal (n = 7), aneurysmal (n = 9), and atherosclerotic (n = 2) flash-frozen human ascending thoracic aortas. Cell types and subtypes were defined using both top marker genes and canonical gene markers. Disease enrichment and relevant cell types were identified using newly developed computational tools to integrate GWAS data from multiple vascular disease-relevant studies with the single nuclei aortic expression profiles.

Results:

Nuclear dissociation and snRNA-seq identified ten distinct transcriptomic clusters from the integrated analysis representing all major vascular cell populations. Three distinct VSMC populations emerged that exhibited differential expression of extracellular matrix, contractile and pro-proliferative genes. Aneurysmal specimens were enriched for one of the fibroblasts and one of the VSMC subpopulations compared to healthy tissue. RNA-trajectory analysis inferred a phenotypic continuum of gene expression between VSMC A and VSMC B or C and between two fibroblast types. VSMCs and Fibroblast C exhibited the greatest cell type-specific enrichment of genes mapped to GWAS loci for coronary artery disease (CAD), blood pressure, and migraine. Cell type-specific enrichment scores were more robust among the transcriptional profiles from non-diseased vascular tissue.

Conclusions:

Our use of single-cell isolation and new computational methods prioritizes the cell types that most contribute to vascular disease pathogenesis. Specifically, tissue dissociation and single-nuclear transcriptomics better represent all vascular cell types, from which we demonstrate enrichment of pro-proliferative VSMCs in TAA and further implicate phenotypic switching as a likely pathologic mechanism. Integrated analysis of cell-specific gene expression and vascular disease GWAS data implicate genes and pathways associated with fibroblast and VSMC cell-state transitions.

Graphical Abstract:

Overview of experimental approach to single cell RNA-sequencing and identification of cells prioritized for contribution to the genetic risk of vascular disease.

Introduction

The pathogenesis of vascular disease can be attributed to cellular dysfunction in the arterial wall, mediating several pathologic processes including neointimal formation, impaired angiogenesis, and inflammation.^1,2 Endothelial dysfunction and inflammatory cell recruitment are linked to the early stages of vascular disease,¹ whereas later stages involve vascular smooth muscle cell migration to the cell-rich atherosclerotic plaque.³ Cell fate transitions have also been implicated in atherosclerosis, though the exact causal cell-types and/or cell-states contributing to early vascular dysfunction and vascular disease in human cells remain undefined^4,5.

The integration of genetic risk profiles for vascular disease with single cell transcriptional profiles represents a promising avenue to identify causal cell types and cell states⁶. Genome-wide association studies (GWAS) have implicated thousands of common variants associated with vascular diseases and traits, and larger meta-analyses have linked single nucleotide polymorphisms (SNPs) to risk of CAD,^7–9 hypertension,^10,11 migraine headache,^12–15 aortic size,¹⁶ and other common vascular diseases. However, very few of these associated SNPs have been functionally characterized to explain their role in vascular disease pathogenesis owing largely to the small individual effect size of common variants on overall disease risk, their location within non-coding genomic regions, or unclear contributory cell-types. To begin to address these challenges, recent studies have begun leveraging the combination of GWAS disease-causing gene identification with single-cell or single-nucleus RNA-sequencing (sc/snRNA-seq) to prioritize novel cell-types for common diseases.^5,17,18 For common vascular diseases, however, progress has remained limited by a scarcity of scRNA-seq data from human vascular tissue and the need for computational tools that map genes to GWAS loci and bridge genetic risk loci with cell-type specific gene expression.

In this study, we address both challenges and present a novel approach that prioritizes specific vascular cell subpopulations relevant for vascular disease risk. First, we integrated from own and previously published scRNA-seq from freshly harvested human vascular tissue to mitigate the over-representation of immune cells and generate enough data for analysis^19–22. We then applied a new computational tool called ECLIPSER^23,24 that tests for cell type-specific enrichment of genes mapped to GWAS loci based on expression and splicing quantitative trait loci (e/sQTLs) to link gene expression profiles of vascular cells to cardiovascular disease risk. Taken together, these methods identify robust markers of vascular cell heterogeneity and gene expression changes in healthy and diseased tissue to implicate cell-types that likely drive the early stages of vascular disease.

Methods

Sample collection.

Aortic samples were collected from the Brigham & Women’s Hospital cardiovascular center. Additionally, we conducted a search for similar published datasets of single-nuclei RNA-sequencing of human aortic tissue using the NCBI GEO database; only a single dataset was identified, published by Chou & Chaffin et al.²⁵ The study protocol was reviewed and approved by the Institutional Review Board (IRB, #2021P001077). Informed consent was obtained from all participants involved in the study, and all patient data and samples were de-identified to maintain privacy and confidentiality. Control aortic tissue was obtained from the proximal aorta at the aortic root during coronary artery bypass graft surgery. For ascending thoracic aortic aneurysmal (TAA) specimens, aortic tissue was obtained from the aortic root of patients undergoing elective surgical repair. All tissue was collected directly from the operating room and immediately immersed in liquid nitrogen. Upon arrival in the laboratory, the tissues were stored at −80°C until processing for the snRNA-seq sample preparation.

Nuclei isolation, library construction and sequencing.

The frozen tissue was homogenized using a liquid nitrogen cooled mortar (Bel-Art). Tissue samples were pulverized via mortar and pestle cooled in liquid nitrogen. The homogenate was then suspended in 1 mL Nuclei EZ lysis buffer (Sigma Aldrich Incl.) containing 1 U/μl RNAse inhibitor (NEB). After incubating for 5 minutes on ice, equal amounts of wash buffer (1% BSA in PBS supplemented with 1 U/μL RNAse inhibitor) were added to stop the lysis reaction. This suspension was then filtered of cellular debris by passing through a 70μm filter. Nuclei were centrifuged at 600x g and washed twice with 5 mL of wash buffer, preserving the pellet of nuclei. Before the second wash, nuclei were passed through a 30μm filter, and nuclei were diluted in 100μL of wash buffer and counted to assess yield. Approximately 20,000 particles were used to create droplets and sequencing libraries using 10X Genomics v3 kits according to manufacturer’s instructions (10X Genomics). 150bp paired-read sequencing was generated using Novaseq platform (Illumina, San Diego, CA) according to the manufacturer’s instructions. Sequencing reads were processed for single-cell assignment and transcript counting using Cellranger (10X Genomics, version 5.0).

Single cell data analysis and integration.

A detailed description of the bioinformatic workflow, including all coding scripts and quality control metrics, are provided as an open-source GitHub repository: https://github.com/mepepin/TAA_snrnaseq. Briefly, raw Cellranger output data were filtered to remove ambient RNA using CellBender in ‘full’ running mode. These output files, which constituted feature-barcode matrices and cluster assignments, were then imported into the R (v4.4.1) environment using the Seurat package (v5.0.1). To ensure data quality, cells with fewer than 200 detected genes, or those containing more than 5% mitochondrial gene content, were excluded. Sample-wise quality benchmarks are compiled as a data supplement (Table S1).

Normalization and scaling were performed using a regularized negative binomial regression to account for technical noise, stabilize variance, and improve the identification of variable features via the SCTransform (v0.4.1) method. The 3,000 most variable genes were selected for integration via Harmony (v1.2.1), which also employed batch correction across both samples and studies (Chou.et.al. and Pepin.et.al.). The DoubletFinder (v2.0.4) R package was used to identify and remove doublets from the integrated dataset by running a parametric sweep, estimating the expected doublet proportion, and subsetting to include only sequenced nuclei meeting “Singlet” classification.

Unsupervised clustering via Uniform Manifold Approximation and Projection (UMAP) was performed on the scaled and variable gene expression data, and the top 30 principal components were used to define Louvian clustering with the “FindClusters” function. Uniform Manifold Approximation and Projection (UMAP) was used for dimensionality reduction and visualization of the clustered data, wherein we used an iterative process to identify the clustering resolution threshold capable of delineating individual clusters seen on the integrated UMAP; we used a range of arbitrary resolutions (0.2, 0.4, 0.6, 0.8, 1.0), from which we selected resolution = 0.4 for subsequent cell type identification (Supplemental Figure S1A). Inter-cluster differential gene expression analysis was carried out using the “FindMarkers” function using the Wilcoxon rank sum test, to identify genes that were differentially expressed between clusters determined at FDR<0.05. Gene-set enrichment analysis for all comparisons was performed using EnrichR (v3.2).²⁶

GWAS cell type enrichment analysis.

Cell type-specific disease enrichment was performed using ECLIPSER: Enrichment of Causal Loci and Identification of Pathogenic cells in Single-Cell Expression and Regulation data^23,24. ECLIPSER tests whether genes mapped to the GWAS loci of a given complex disease or trait based on functional genomics data are enriched for cell type-specific expression compared to a null distribution of thousands of loci associated with unrelated traits. Significantly enriched cell types are thus implicated in the disease or trait under consideration. In this tool, the putative causal genes were assigned to GWAS loci based on cis-eQTLs and cis-sQTLs from 49 GTEx v8 tissues²⁷ that were in linkage disequilibrium (r² > 0.8) with the lead GWAS variant using GWASvar2gene (https://github.com/segrelabgenomics/GWASvar2gene) and based on Open Targets Genetics causal gene prioritization that utilizes additional functional information, such as Hi-C data and predicted deleterious protein coding variants.²⁸ Cell type specificity scores were computed for each GWAS locus as the fraction of cell type-specific genes (Fold-change > 1.4 and false discovery rate (FDR) < 0.1) mapped to the locus. Cell type specificity fold-enrichment and significance (p-value) were estimated per GWAS locus set using a Bayesian Fisher’s exact test and the 95th percentile of the null locus scores was used as the cell type specificity cutoff (https://github.com/segrelabgenomics/ECLIPSER). Cell types were considered significant at a tissue-wide Benjamini-Hochberg FDR below 0.1.

Results

Integrated single-nuclear RNA-sequencing (snRNA-Seq) of flash-frozen tissue recapitulates established vascular cell types.

To generate single nuclear RNA-seq (snRNA-seq) profiles from normal and ascending thoracic aneurysmal human aortas, we obtained full-thickness intraoperative sections of the proximal ascending aorta from five patients undergoing elective aortic root replacement of thoracic aortic aneurysms (TAA) or coronary artery bypass graft surgery (CABG) at Brigham and Women’s Hospital (summarized in Graphical Abstract). Once sequenced, these data were integrated with snRNA-sequencing data provided by Chou and Chaffin et al.²⁵ to yield a total of 7 non-diseased (CON), 9 aneurysmal (TAA), and 2 atherosclerotic (CABG) samples for inter-comparison.

To reduce technical confounding from sample processing, Harmony (v1.2.0) in R (v4.4.1) was used to reduce sequencing batch effects, providing robust separation of 10 nuclear clusters via UMAP-based deconvolution that was conserved across dataset origins (Fig. 1A) and disease conditions (Fig. 1B); after quality filtering, this analysis yielded 36,345 TAA, 37,648 CON, and 2,301 CABG nuclei. Annotation of nuclei with known cellular identities included both unbiased (Supplemental Fig. S1) and canonical gene marker-based approaches, as previously published (Fig. 1C).²⁹ From the 10 UMAP clusters found in the integrated UMAP (Fig. 1C), all of the main vascular cell types clustered distinctly according to the following gene markers: VSMCs (MYH11, ACTA2, ITGA8, PRKG1, CRISPLD1), Fibroblast sub-type 1 (ABCA10, C3, ADGRD1, FBLN1, DCN), Fibroblast sub-type 2 (NFASC, SMAD5, PRSS23), Endothelial Cells (DIPK2B, ARHGEF15, STC1, FLT1), Macrophages (MRC1, LGMN, F13A1, RBM47), Natural Killer T cells (SKAP1, RIPOR2, ITGAL, CD96), and Neuronal cells (NRXN1, CADM2, SORCS1, ARHGAP15). Ultimately, 3 distinct clusters emerged from within the nuclei enriched for VSMC marker gene expression, as previously reported.²⁵

Figure 1: — Disease specific cell-type heterogeneity of the ascending aorta. **(A)** UMAP-based dimensionality reduction of vascular smooth muscle, demonstrating integration of data published by Chou & Chaffin et al. (GSE207784). **(B)** Distinct proportional representation is seen among samples obtained from aneurysmal and atherosclerotic relative to non-diseased aortic biopsies. **(C)** Annotation of integrated UMAP demonstrating 3 VSMC clusters (VSMC_A/B/C), 3 fibroblast clusters (Fibroblast_A/B/C), endothelial cells (EC), Macrophage, Natural Killer T cells (NKT), and Neuronal cells. **(D)** Proportional bar plot of annotated clusters, separated by aortic disease. **(E)** Bar plot of average proportions (+/− SD) of each cell type for healthy donor (CON), ascending aortic aneurysm (TAA), and CABG aortic button (CABG).* *Statistical significance is defined by 2-way ANOVA with multiple comparisons t-test. *P < 0.05, ** P < 0.01

VSMC and Fibroblast subpopulations are either enriched or diminished in diseased aortic tissue.

Robust heterogeneity was seen in the VSMC and fibroblast subpopulations (Fig. 1D), with distinct subpopulations for each cell type and an intermediate population, here termed “Fibroblast C” that expressed markers of both, additionally defined by expression of ADAMTS1, RGS6, TNC, ANGPT2, DGKG, and GRIP2. To determine whether alterations in cellular composition exist between diseased and healthy states in our dataset, we performed a proportional comparison of annotated cellular identities among CON, TAA, and CABG (Fig. 1E). We found decreased representation of Fibroblast A cells in both the aortic aneurysm (P = 0.020) and CABG (P = 0.033) samples compared with normal aorta, whereas Fibroblast B cells exhibited a trending enrichment among aneurysmal (P = 0.078) – but not CABG – samples relative to CON. Conversely, we noted a significant increase in VSMC B cell population in TAA (P = 0.004) and CABG (P < 0.0001) disease states; other cell types showed no significant differences among disease.

Transcriptional profiles and biologic pathways underlying VSMC heterogeneity.

Since the VSMC B subpopulation showed enrichment in both TAA and CABG samples, we characterized the differential gene expression and pathway enrichment for these cells relative to other VSMC subpopulations. The VSMCs were separately clustered and re-analyzed to identify the markers that differentiated the subpopulations. In total, 23,008 (63%) TAA, 19,696 (52%) CON, and 1,738 (76%) CABG nuclei annotated as VSMCs were selected for differential analysis and re-clustered via UMAP (Fig. 2A). Differential expression between the two predominant sub-types – VSMC A and VSMC B – revealed 2,079 differentially-expressed genes (DEGs) (Q<0.05). Of these, serine protease inhibitor serpin family E member 1 (SERPINE1) was among the most differentially increased in VSMC B, with 2.8-fold higher expression (P=3.2 × 10⁻²⁴³) relative to VSMC A (Table S3). Hierarchical clustering of the three VSMC subtypes based on the top VSMC A versus VSMC B DEGs highlighted more similarity between VSMC B and VSMC C sub-types (Fig. 2C). To identify signaling and pathway-based differences between VSMC B and VSMC A, gene-set enrichment revealed that, among the 166 up-regulated genes in VSMC B relative to VSMC A (Q<0.05, Log₂FC > 0.5), a significant number were associated with “PI3K-AKT Signaling” (q=0.0011) and “Focal Adhesion” (q= 0.0035) pathways (Fig. 2D,E). Conversely, among the 313 genes with higher expression in VSMC A (Q<0.05, Log₂FC < −0.5), “Myometrial Relaxation and Contraction” (q=1.1×10⁻⁶) and “TGF-beta Signaling” (q=7.5×10⁻⁶) were most enriched (Fig. 2D,E), further highlighting a divergent expression pattern of VSMC B with it being associated with a de-differentiated state (Table S4).

Figure 2: — Identifying transcriptional heterogeneity in ascending aortic smooth muscle. **(A)** UMAP-based dimensionality reduction of vascular smooth muscle, revealing 3 distinct clusters. **(B)** Volcano plot illustrating differentially-expressed genes between VSMC_A and VSMC_B, labelling the top 15 DEGs by P-value, which are also shown in **(C)** via hierarchical clustering and heatmap visualization. **(D)** Gene-set enrichment analysis using the Wikipathways database for the up-regulated (orange) and down-regulated (cyan) DEGs, highlighting the normalized expression (z-score) of genes responsible for enrichment of the top pathways **(E)** “PI3K AKT Signaling” and “Myometrial Relaxation and Contraction,” respectively. **(F)** Upstream regulator identification via ChEA gene-set enrichment of transcription factors over-represented within the proximal promoter of DEGs. **(G)** Circular genome plot highlighting differentially expressed targets of TCF21 based on ChIP-sequencing analysis published by Nagaoet al. (GSE124011). **(H)** Trajectory analysis via Monocle3 (v1.3.4) to infer the cluster-specific transcriptional shift from VSMC_A towards VSMC_B. **(I)** Heatmap and hierarchical clustering of genes identified via trajectory mapping, highlighting genes enriching the top-most enriched pathway within each cluster, including “vascular smooth muscle contractility” (P = 6.7 × 10-9), “TGFβ Signaling” (P = 6.0 × 10-5), and “Focal Adhesion” (P = 5.6 × 10-12) the top 25 DEGs according to FDR-adjusted P (Q) value). **(J)** Pseudotime plot of genes differentially expressed at the transition between VSMC_A and VSMC_B/C.

VSMC Subtype-specific Transitional and Upstream regulators.

To identify putative regulatory mechanisms that accompany the cell state transition from VSMC A to VSMC B, we analyzed the transcription factor binding motifs in the promoters of differentially regulated genes (Fig. 2F). All genes differentially expressed in VSMC B vs. A were analyzed for TF binding in gene promoters using chromatin immunoprecipitation sequencing (ChIP-Seq) data (available through the ChIP Enrichment Analysis (ChEA) database). Genes that were upregulated in VSMC B enriched for the TCF21 promoter binding sites (35/1629 genes; enrichment q=1.0×10⁻⁴) compared with genes upregulated in VSMC A, supporting prior work highlighting TCF21 as a regulator of VSMC phenotyping switching. Of the genes upregulated in VSMC B relative to VSMC A, 35 possessed at least one TCF21 binding site at their promoter (Fig. 2G), including several collagen genes (COL1A1, COL4A4, COL6A1, COL16A1, COL27A1) and fibroblast marker gene DCN. This analysis for transcription factor binding in upregulated genes also identified EGR1, KLF6, DMRT1, and BACH1 TFs as enriched for potential target genes in VSMC B versus VSMC A (Fig. 2F). Taken together, our data support prior studies identifying TCF21 as a transcriptional regulator that promotes the transition towards fibroblast-like VSMCs.

Trajectory analysis identifies the gene expression changes that drive VSMC state transitions.

To identify transcriptional profile accompanying cell state shifts towards the VSMC B subpopulation, we performed trajectory analysis within the VSMC subpopulations by selecting the origin at VSMC A; this approach identified a single trajectory uniting the VSMC states (Fig 2H). Though the VSMC A and VSMC B subpopulations had clustered distinctly in the UMAP, trajectory analysis identified a smooth continuum of intermediate cell states possessing distinct gene expression profiles (Fig. 2I, heatmap). Specifically, we identified the following differentially-expressed genes that may underlie the transition in developmental trajectory connecting VSMC A and B subtypes (Fig. 2I, 2J): NOX4, INSR, OGT, and SETD5.

Fibroblast cell state shift underlies phenotypic transition towards a pro-contractile state.

Similar to our approach in VSMC subpopulations, we shifted our focus onto the expression differences between the 3 identified Fibroblast clusters, represented by 5,838 (8.0%) TAA, 8,276 (11.0%) CON, and 165 (3.6%) CABG nuclei annotated as fibroblasts. Based on the observation that Fibroblast A exhibited a relative reduction in the CABG and aortic aneurysm samples relative to healthy controls, we again performed pseudotime-based trajectory analysis beginning in the Fibroblast A pool to determine the genes that underly a possible transition away from it (Fig. 3A, 3B). We found that, among the 348 genes that reflect transitional state expression (FDR < 0.05, Moran’s I > 0.1), 3 distinct transcriptional patterns of gene expression emerged (Fig. 3C). Gene set enrichment analysis for these genes identified differing pathways, where the 126 early activated genes (i.e. higher in Fibroblast A) were disproportionately enriched in TGFβ signaling (25/565, q = 5.7 × 10⁻¹²); the 76 genes activated in Fibroblast B nuclei were enriched in “β1 integrin cell surface interactions” (5/66, q= 0.0013); lastly, the 146 genes activated within “Fibroblast C” were enriched in “contractile” and “smooth muscle contraction” (11/22, q = 4.67 × 10⁻¹⁶) and, more specifically, “vascular smooth muscle contraction” (16/116, q = 2.89 × 10⁻¹⁴). Plotting the differential expression both CABG vs. CON and Aneurysm vs. CON revealed a high-degree of concordance in genes (Fig. 3D), with modest overlaps in KEGG pathways enriched in either comparison (Fig. 3F, Table S5, Table S6)

Figure 3: — Identifying fibroblast-specific gene expression clusters. **(A)** UMAP-based clustering of fibroblasts, identifying three distinct transcriptional phenotypes (0.2 resolution). **(B)** Trajectory analysis via Monocle3 (v1.3.4) was used to infer disease-associated phenotypic shift from Fibroblast A towards adjacent clusters. **(C)** Heatmap and hierarchical clustering of genes identified via trajectory mapping, highlighting the top DEGs (by Q-value). **(D)** Scatterplot of differentially co-expressed genes within the “Fibroblast A” cluster, highlighting the top 15 most robustly conserved DEGs between aneurysmal and atherosclerotic aortic tissue. **(E)** genes populating the most enriched pathway “Hallmark Hypoxia” and “Hallmark UV Response” for aneurysmal and atherosclerotic tissue, respectively. *significance defined at bonferroni-adjusted P (Q) < 0.05.

VSMCs in vascular disease possess distinct transcriptional profiles.

To better understand both common and disease-specific gene expression changes in VSMCs, we performed differential expression analysis among both VSMC A and B clusters comparing aneurysmal (TAA) and atherosclerotic (CABG) samples relative to non-diseased “control” (CON) aortas. Comparison of these differentially expressed genes (DEGs) between the diseased samples revealed 3,011 DEGs distinctly within CABG vs. CON, 5,341 DEGs distinct in TAA vs. CON, and 3,250 DEGs similarly changing in both CABG and aneurysmal aortic VSMCs vs. CON (Fig. 4A, Table S7). Comparing the fold-change between CABG and CON to that of TAA vs. CON for the 3,250 common DEGs highlighted PHACTR1 (CABG: 372-fold decrease, q = 1.75 × 10⁻⁹¹; TAA: 2.41-fold decrease, q = 5.48 × 10⁻¹²⁹) among the most differentially suppressed genes in VSMCs relative to non-diseased tissues, whereas SERPINE1 was among the most robustly increased in both TAA (3.1-fold, q = 3.1 × 10⁻⁵⁶) and CABG (1.6-fold, q = 2.9 × 10⁻⁶) samples (Fig. 4B). Pathway enrichment of these DEGs identified 7 genes associated with VEGF signaling (q = 0.004) and 3 genes involved in regulation of actin cytoskeletal dynamics (q = 0.15) (Fig. 4C). Examination of the genes distinctly changing in CABG relative to CON (Fig. 4D-E) identified a disproportionate number involved in angiotensin (q= 0.015) and inflammatory signaling (q = 0.015). By contrast, genes distinctly changing in aneurysmal aortic tissue (Fig. 4F-G) were enriched in hypoxia-related (q= 1.4 × 10⁻⁶) and TGFβ signaling (q= 2.64 × 10⁻⁶) pathways. Altogether, these observations suggest that disease-relevant shared and etiology-specific transcriptional alterations exist within VSMCs.

Figure 4: — Identifying disease-associated transcriptional programs in vascular smooth muscle. **(A)** Venn diagram illustrating differentially-expressed genes (DEGs) in CABG (red) and aneurysmal (gray) tissues relative to non-diseased control aorta (Q < 0.05). **(B)** Scatterplot of differentially co-expressed genes within the “VSMC_B” cluster, highlighting the top 15 most robustly expressed and conserved DEGs between aneurysmal and atherosclerotic aortic tissue. **(C)** Table of the most disproportionately enriched gene ontology (GO)-term pathways as determined by bonferroni-adjusted p-value. **(D)** Volcano plot illustrating differentially-expressed genes in atherosclerotic aorta obtained from CABG aortic buttons relative to healthy aorta (CON), highlighting the top 10 DEGs (P-value). **(E)** Enriched GO-term pathways found in CABG vs. CON (Q < 0.05), and within the VSMC_B sub-cluster. **(F)** Volcano plot illustrating differentially-expressed genes in atherosclerotic aorta obtained from CABG aortic buttons relative to healthy aorta (CON), highlighting the top 10 DEGs (P-value). **(G)** Enriched GO-term pathways found in CABG vs. CON (Q < 0.05) within the VSMC_B sub-cluster. **(H)** GWAS inferred cell type-specific disease enrichment using Enrichment of Causal Loci and Identification of Pathogenic cells in Single-Cell Expression and Regulation data (ECLIPSER). Significance (circle size, -log10(P-value)) and fold-enrichment (circle color) of the cell type specificity of GWAS locus sets (rows) of three vascular diseases and inflammatory bowel disease (IBD), as negative control, are shown for 10 cell types found in the healthy aorta snRNA-seq data. Red rings: experiment-wide significant (Benjamini-Hochberg (BH) FDR < 0.1). Yellow rings: tissue-wide significant (BH FDR < 0.1). Grey rings: nominal significant (P < 0.05). **(I,J)** Differential gene expression (log2(Fold-change), y axis) in VSMC C **(I)** and Fibroblast C **(J)** compared to all other cell types is shown for the set of genes (x axis) driving the enrichment signal of CAD GWAS loci in these most strongly enriched cell types. The horizontal dashed line represents log2(Fold-change) of 0.5 (FC=1.4) and FDR < 0.1 used as the cell type-specificity enrichment cutoff.

VSMC and fibroblast cell populations show greatest enrichment for multiple CVD GWAS traits.

We sought to determine the contribution of all the vascular cell subpopulations from healthy and diseased aorta tissue to common vascular traits and diseases. To accomplish this, we performed disease association enrichment analysis in our single cell with ECLIPSER, which uses expression data by combining human population genetic association data from GWAS with single cell transcriptomics to identify disease relevant cell types and states. Expression and splicing QTLs (e/sQTLs) along with other functional genomics data from Open Targets Genetics were used to identify putative causal genes in GWAS loci. For example, for CAD the ECLIPSER method proposed 35 predicted causal genes, 21 specifically expressed in VSMCs and 18 in Fibroblasts, in 34 associated genomic loci (Supplementary Table 9)”. Enrichment of cell type-specific expression of genes mapped to the GWAS loci per vascular trait was estimated in each cell subpopulations using ECLIPSER (Methods). Cell types were assigned a fold-enrichment of trait associations and p-value based on the fraction of GWAS loci with cell type-specific expression compared to a null set of GWAS loci of unrelated traits.

Three vascular traits exhibiting strong association signals in GWAS were used to calculate enrichment (CAD, systolic and diastolic blood pressure, and migraine headache) along with one non-vascular trait used as a negative control (inflammatory bowel disease) (Table S8). Analysis of the CON snRNA-seq data demonstrated enrichment for all three VSMC subtypes and fibroblasts C and B in CAD and blood pressure (FDR<0.05; Fig. 4H, Supplemental Figure S3). By contrast, only the VSMC and the fibroblast C populations were enriched for migraine headache loci (Fig. 4H). Inflammatory bowel disease (IBD) only showed enrichment with NKT cells, as expected and as a sort of negative control. The complete list of genes and GWAS loci found with cell type-specific enrichment for the vascular diseases are listed in Table S9. For example, nine genes drive the 2.25 fold-enrichment of CAD GWAS loci in VSMC C (Fig. 4I) that are enrich forcell-substrate adhesion (Gene Ontology GO:0031589; gProfiler Benjamini-Hochberg (BH) Q<0.05), whereas 14 genes drive enrichment in Fibroblast C (Fig. 4J) that are enriched in blood vessel development (GO:0001568; BH Q<0.05).

To determine if diseased vascular cells demonstrate more enrichment for GWAS genes compared to healthy tissue we next compared ECLIPSER enrichment using cell type-specific gene expression from snRNA-seq data in the CABG and TAA tissues. The CABG tissue only showed significant enrichment for one VSMC subpopulation (VSMC A) in blood pressure (FDR<0.1), and nominal enrichment in two VSMC subpopulations (VSMC A and C) in migraine, which likely reflects a lower number of DEGs in these tissue samples (Supplemental Figure 4A, Table S9). The TAA tissue showed a similar pattern of cell type enrichment as CON samples, albeit less robustly for each of cell type (Supplemental Figure 4B). This supports the study of GWAS genes and risk loci in cells from healthy tissues, where the biological effects primarily mediate the initiation of disease.

Discussion

The technological advancements in single-cell techniques has enabled the implication of cell types and cell states that are likely responsible for conferring vascular disease risk. In the current study, we use the latest computational methods to integrate single-cell data from multiple sources at different stages of disease to identify enriched cellular identifies. Lastly, we traced transcriptomic trajectories of vascular cell populations to prioritize specific subpopulations for their contribution to the genetic risk for vascular disease using the ECLIPSER enrichment algorithm. Together, these analyses identify cell populations relevant to multiple vascular diseases and prioritize the cell-types in which GWAS variants can be functionally characterized.

Our high-depth transcriptomics confirmed earlier reports of distinct VSMC subpopulations that depict a cell state transition trajectory. In our collection of aortic samples from control, atherosclerotic, and aneurysmal aortas we support the presence of a VSMC subpopulation (“VSMC B”) that expresses many fibroblast marker genes. Our trajectory analysis demonstrated that contractile VSMC A cells exist within a cell state continuum alongside VSMC B cells, which again aligns with a recent analysis from reporter-labelled VSMCs extracted from murine aortic tissue.³⁰ Therefore, we provide evidence using human samples to support an emerging paradigm that VSMC cell-state transitions can directly contribute to the progression of vascular disease.

We identified and traced lineage trajectories for multiple vascular cell subpopulations to facilitate a novel cell prioritization strategy of vascular disease. The ECLIPSER method was originally developed using multi-tissue single-nuclear atlas that lacked high quality data from vascular tissue^24,31. Harsh dissociation methods traditionally used for single-cell isolation often deplete ECs and likely alter the transcriptional profile of VSMCs. ECLIPSER also expands the definition of disease-associated genes by incorporating expression and splicing quantitative trait loci (e/sQTLs) that associate variants with proximal and distal target genes up to 1Mb away.³² Using this prioritized gene list, ECLIPSER tests whether the mapped genes to all GWAS loci for a given complex trait are enriched for cell type-specific expression within cell types. In control vascular tissue from subjects without known vascular disease, we found enrichment for VSMC and fibroblast subpopulations for vascular diseases of CAD, blood pressure, and migraine headache. The multiple VSMC and fibroblast subpopulations have different associated GWAS risk genes, and therefore functional study of these loci should be performed in cells that reproduce this biological heterogeneity.

We also used snRNA-seq data to directly compare the cell-type specific enrichment differences in healthy and diseased transcriptomic profiles from vascular tissue. Most published analyses have used healthy human tissue to prioritize causal cell populations to show enrichment for GWAS genes of vascular disease, since GWAS variants are present in the germline and are therefore assumed to affect healthy cells. In vascular diseases, however, the early phases of disease initiation often occur in dysfunctional vascular cells that have altered gene expression profiles after years of direct exposure to atherogenic risk factors including hyperlipidemia, hypertension, and reactive oxygen species.² Recent studies from the STARNET consortium showcase that the transcriptional signatures of diseased vascular tissue exhibit different co-expression networks compared with healthy vascular tissue.^22,33 Our study offers a first glimpse using single-cell data to compare the enrichment of transcriptional pathways in both healthy and diseased tissue. The majority of GWAS loci show stronger enrichment in control cells, consistent with assumptions about the role of these risk variants in disease initiation.

A notable limitation of single-cell analysis is the associated cost of sample processing that limits sample size. In our analysis, we generated data from over 76,000 cells and validated our findings by integrating data from a published aortic snRNA-seq study.²⁵ And yet, our analysis was limited to 16 aortic samples collected at different times from patients with varying baseline characteristics. The ability to identify cell-state transitions enriched for disease risk will hopefully improve with new technologies to reduce single cell RNA-seq costs and access to bio-banked vascular tissue from healthy and diseased subjects. Additionally, our computational methods rely on existing functional genomic data to prioritize causal genes at each GWAS locus. Despite its computational advantages, ECLIPSER still relies on e/sQTL data from bulk tissues rather from single-cell expression data, warranting additional variant-to-function studies in vascular tissue to highlight true causal genes. As the cost of single cell RNA-seq decreases and multiplexing technologies improve, detecting e/sQTLs at the cell type level in vascular tissues will become feasible. With a better set of gold-standard genes among the 300+ CAD GWAS genes, these computational programs will improve and provide greater power to identify pathogenic cell types.

Despite the limitations our analysis of snRNA-seq data allows us to identify disease-relevant vascular cell heterogeneity in vascular cell subpopulations. We provide in-depth analysis of VSMC and fibroblast subpopulations, and the developmental trajectories that link these two cell-types. Enrichment of these transitional states for vascular disease risk provides new insight into disease, validates new methods to analyze vascular tissue, and prioritizes cell types in which to study the functional effects of vascular disease-associated genetic risk loci.

Supplementary Material

Supplemental figures

Supplementary Figure S1: Data comparison and cell type composition. (A) UMAP illustrating resolution threshold and cluster discrimination. (B) Sample-specific UMAP demonstrating uniform population of clusters at resolution = 0.4. (C) Principal component analysis, showing top-5 upregulated and downregulated genes within the 15 top principal components. (D) Top most differentially enriched genes within each cluster. (E) Density plots of classical gene markers of cell types.

Supplementary Figure S2: Dot-plot showing relative expression (z-score) of genes across all cell-types which were identified via trajectory analysis across VSMC sub-types, specifically (A) top-15 up-regulated and (B) top-15 downregulated DEGs from VSMC A/B/C.

Supplementary Figure S3. Cell type specificity fold-enrichment (x-axis) in healthy aorta (CON) cell types ranked in descending order for e/sQTL-mapped genes to GWAS loci of coronary artery disease (CAD) (A), migraine (B), blood pressure (C), and the negative control, inflammatory bowel disease (IBD) (D). Points: fold-enrichment estimates from ECLIPSER. Error bars: 95% confidence intervals. Red: tissue-wide significant (BH FDR < 0.1). Grey: nominal significant (P < 0.05). Blue: non-significant (P ≥ 0.05).

Supplementary Figure S4: ECLIPSER enrichment analysis of vascular disease loci genes in diseased aorta cell types. Each phenotype was tested against all the vascular and non-vascular cell types obtained from (A) CABG and (B) TAA samples. Cell type specificity significance is proportional to circle size (-log10(P-value)) and the fold-enrichment with circle color for each trait GWAS locus sets (rows) and cell type (columns). Traits (rows) and cell types (columns) were clustered based on hierarchical clustering of the Euclidean distance between the GWAS locus set cell type-specificity enrichment scores. Red rings: experiment-wide significant (Benjamini-Hochberg (BH) FDR < 0.1). Yellow rings: tissue-wide significant (BH FDR < 0.1). Grey rings: nominal significant (P < 0.05). CABG, atherosclerosis; TAA, thoracic aortic aneurysm.

NIHMS2118170-supplement-Supplemental_figures.pdf^{(10.4MB, pdf)}

Supplemental_tables

Supplemental Table S1: Table summarizing patient characteristics and quality control parameters for single-nuclei RNA-sequencing data used in the current study.

Supplemental Table S2: Differentially expressed genes identified across cell-types within the integrated dataset.

Supplemental Table S3: Table summarizing the gene set enrichment analysis for genes differentially expressed between VSMC_B vs. VSMC_A (Q < 0.05).

Supplemental Table S4: Pathway enrichment results for vascular smooth muscle cels.

Supplemental Table S5: Gene set enrichment results of KEGG pathways for Fibroblasts using differentially expressed genes in Aneurysm vs. Control.

Supplemental Table S6: Gene set enrichment results of KEGG pathways for Fibroblasts using differentially expressed genes in CABG vs. Control.

Supplemental Table S7: Table summarizing the overlapping differentially expressed genes among VSMC clusters to compare aneurysmal (TAA) and atherosclerotic (CABG) samples relative to non-diseased “control” (CON) aortas.

Supplemental Table S8: List of vascular traits GWAS and number of loci used in ECLIPSER enrichment analysis.

Supplemental Table S9: ECLIPSER cell type enrichment analysis of vascular disease loci in healthy and diseased aorta tissue.

NIHMS2118170-supplement-Supplemental_tables.xlsx^{(5.8MB, xlsx)}

Acknowledgements:

This work was funded by National Institutes of Health Grants U01-HL166060 (R.M.G. and A.V.S.), R01-HL164811 (R.M.G.), DP2-HL152423 (R.M.G.) and the Eugene Braunwald Junior Faculty Scholar Award (R.M.G.).

Footnotes

Conflicts of Interest: None

References

1.Libby P. The changing landscape of atherosclerosis. Nature. 2021;592(7855):524–533. [DOI] [PubMed] [Google Scholar]
2.Gimbrone MA, Garcia-Cardeña G. Endothelial Cell Dysfunction and the Pathobiology of Atherosclerosis. Circ Res. 2016;118(4):620–636. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Basatemur GL, Jørgensen HF, Clarke MCH, Bennett MR, Mallat Z. Vascular smooth muscle cells in atherosclerosis. Nat Rev Cardiol. 2019;16(12):727–744. [DOI] [PubMed] [Google Scholar]
4.Sharma D, Worssam MD, Pedroza AJ, Dalal AR, Alemany H, Kim HJ, Kundu R, Fischbein MP, Cheng P, Wirka R, Quertermous T. Comprehensive integration of multiple single-cell transcriptomic data sets defines distinct cell populations and their phenotypic changes in Murine atherosclerosis. Arterioscler Thromb Vasc Biol. 2024;44(2):391–408. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Turner AW, Hu SS, Mosquera JV, Ma WF, Hodonsky CJ, Wong D, Auguste G, Song Y, Sol-Church K, Farber E, Kundu S, Kundaje A, Lopez NG, Ma L, Ghosh SKB, Onengut-Gumuscu S, Ashley EA, Quertermous T, Finn AV, Leeper NJ, Kovacic JC, Björkegren JLM, Zang C, Miller CL. Single-nucleus chromatin accessibility profiling highlights regulatory mechanisms of coronary artery disease risk. Nat Genet. 2022;54(6):804–816. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Quertermous T, Li DY, Weldy CS, Ramste M, Sharma D, Monteiro JP, Gu W, Worssam MD, Palmisano BT, Park CY, Cheng P. Genome-wide genetic associations prioritize evaluation of causal mechanisms of atherosclerotic disease risk. Arterioscler Thromb Vasc Biol. 2024;44(2):323–327. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Walsh R, Jurgens SJ, Erdmann J, Bezzina CR. Genome-wide association studies of cardiovascular disease. Physiol Rev. 2023;103(3):2039–2055. [DOI] [PubMed] [Google Scholar]
8.Aragam KG, Jiang T, Goel A, Kanoni S, Wolford BN, Atri DS, Weeks EM, Wang M, Hindy G, Zhou W, Grace C, Roselli C, Marston NA, Kamanu FK, Surakka I, Venegas LM, Sherliker P, Koyama S, Ishigaki K, Åsvold BO, Brown MR, Brumpton B, de Vries PS, Giannakopoulou O, Giardoglou P, Gudbjartsson DF, Güldener U, Haider SMI, Helgadottir A, Ibrahim M, Kastrati A, Kessler T, Kyriakou T, Konopka T, Li L, Ma L, Meitinger T, Mucha S, Munz M, Murgia F, Nielsen JB, Nöthen MM, Pang S, Reinberger T, Schnitzler G, Smedley D, Thorleifsson G, von Scheidt M, Ulirsch JC, Biobank Japan, EPIC-CVD, Arnar DO, Burtt NP, Costanzo MC, Flannick J, Ito K, Jang DK, Kamatani Y, Khera AV, Komuro I, Kullo IJ, Lotta LA, Nelson CP, Roberts R, Thorgeirsson G, Thorsteinsdottir U, Webb TR, Baras A, Björkegren JLM, Boerwinkle E, Dedoussis G, Holm H, Hveem K, Melander O, Morrison AC, Orho-Melander M, Rallidis LS, Ruusalepp A, Sabatine MS, Stefansson K, Zalloua P, Ellinor PT, Farrall M, Danesh J, Ruff CT, Finucane HK, Hopewell JC, Clarke R, Gupta RM, Erdmann J, Samani NJ, Schunkert H, Watkins H, Willer CJ, Deloukas P, Kathiresan S, Butterworth AS, CARDIoGRAMplusC4D Consortium. Discovery and systematic characterization of risk variants and genes for coronary artery disease in over a million participants. Nat Genet. 2022;54(12):1803–1815. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Tcheandjieu C, Zhu X, Hilliard AT, Clarke SL, Napolioni V, Ma S, Lee KM, Fang H, Chen F, Lu Y, Tsao NL, Raghavan S, Koyama S, Gorman BR, Vujkovic M, Klarin D, Levin MG, Sinnott-Armstrong N, Wojcik GL, Plomondon ME, Maddox TM, Waldo SW, Bick AG, Pyarajan S, Huang J, Song R, Ho YL, Buyske S, Kooperberg C, Haessler J, Loos RJF, Do R, Verbanck M, Chaudhary K, North KE, Avery CL, Graff M, Haiman CA, Le Marchand L, Wilkens LR, Bis JC, Leonard H, Shen B, Lange LA, Giri A, Dikilitas O, Kullo IJ, Stanaway IB, Jarvik GP, Gordon AS, Hebbring S, Namjou B, Kaufman KM, Ito K, Ishigaki K, Kamatani Y, Verma SS, Ritchie MD, Kember RL, Baras A, Lotta LA, Regeneron Genetics Center, CARDIoGRAMplusC4D Consortium, Biobank Japan, Million Veteran Program, Kathiresan S, Hauser ER, Miller DR, Lee JS, Saleheen D, Reaven PD, Cho K, Gaziano JM, Natarajan P, Huffman JE, Voight BF, Rader DJ, Chang KM, Lynch JA, Damrauer SM, Wilson PWF, Tang H, Sun YV, Tsao PS, O’Donnell CJ, Assimes TL. Large-scale genome-wide association study of coronary artery disease in genetically diverse populations. Nat Med. 2022;28(8):1679–1692. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Hiura Y, Tabara Y, Kokubo Y, Okamura T, Miki T, Tomoike H, Iwai N. A genome-wide association study of hypertension-related phenotypes in a Japanese population. Circ J. 2010;74(11):2353–2359. [DOI] [PubMed] [Google Scholar]
11.Giri A, Hellwege JN, Keaton JM, Park J, Qiu C, Warren HR, Torstenson ES, Kovesdy CP, Sun YV, Wilson OD, Robinson-Cohen C, Roumie CL, Chung CP, Birdwell KA, Damrauer SM, DuVall SL, Klarin D, Cho K, Wang Y, Evangelou E, Cabrera CP, Wain LV, Shrestha R, Mautz BS, Akwo EA, Sargurupremraj M, Debette S, Boehnke M, Scott LJ, Luan J, Zhao JH, Willems SM, Thériault S, Shah N, Oldmeadow C, Almgren P, Li-Gao R, Verweij N, Boutin TS, Mangino M, Ntalla I, Feofanova E, Surendran P, Cook JP, Karthikeyan S, Lahrouchi N, Liu C, Sepúlveda N, Richardson TG, Kraja A, Amouyel P, Farrall M, Poulter NR, Understanding Society Scientific Group, International Consortium for Blood Pressure, Blood Pressure-International Consortium of Exome Chip Studies, Laakso M, Zeggini E, Sever P, Scott RA, Langenberg C, Wareham NJ, Conen D, Palmer CNA, Attia J, Chasman DI, Ridker PM, Melander O, Mook-Kanamori DO, Harst P van der, Cucca F, Schlessinger D, Hayward C, Spector TD, Jarvelin MR, Hennig BJ, Timpson NJ, Wei WQ, Smith JC, Xu Y, Matheny ME, Siew EE, Lindgren C, Herzig KH, Dedoussis G, Denny JC, Psaty BM, Howson JMM, Munroe PB, Newton-Cheh C, Caulfield MJ, Elliott P, Gaziano JM, Concato J, Wilson PWF, Tsao PS, Velez Edwards DR, Susztak K, et al. Trans-ethnic association study of blood pressure determinants in over 750,000 individuals. Nat Genet. 2019;51(1):51–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Ghaffar A, International Headache Genetics Consortium, Nyholt DR. Integrating eQTL and GWAS data characterises established and identifies novel migraine risk loci. Hum Genet. 2023;142(8):1113–1137. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Gormley P, Anttila V, Winsvold BS, Palta P, Esko T, Pers TH, Farh KH, Cuenca-Leon E, Muona M, Furlotte NA, Kurth T, Ingason A, McMahon G, Ligthart L, Terwindt GM, Kallela M, Freilinger TM, Ran C, Gordon SG, Stam AH, Steinberg S, Borck G, Koiranen M, Quaye L, Adams HHH, Lehtimäki T, Sarin AP, Wedenoja J, Hinds DA, Buring JE, Schürks M, Ridker PM, Hrafnsdottir MG, Stefansson H, Ring SM, Hottenga JJ, Penninx BW, Färkkilä M, Artto V, Kaunisto M, Vepsäläinen S, Malik R, Heath AC, Madden PAF, Martin NG, Montgomery GW, Kurki MI, Kals M, Mägi R, Pärn K, Hämäläinen E, Huang H, Byrnes AE, Franke L, Huang J, Stergiakouli E, Lee PH, Sandor C, Webber C, Cader Z, Muller-Myhsok B, Schreiber S, Meitinger T, Eriksson JG, Salomaa V, Heikkilä K, Loehrer E, Uitterlinden AG, Hofman A, van Duijn CM, Cherkas L, Pedersen LM, Stubhaug A, Nielsen CS, Männikkä M, Mihailov E, Milani L, Göbel H, Esserlind AL, Christensen AF, Hansen TF, Werge T, Consortium, International Headache Genetics, Kaprio J, Aromaa AJ, Raitakari O, Ikram MA, Spector T, Jarvelin MR, Metspalu A, Kubisch C, Strachan DP, Ferrari MD, Belin AC, Dichgans M, Wessman M, van den Maagdenberg AM, Zwart JA, et al. Meta-analysis of 375,000 individuals identifies 38 susceptibility loci for migraine. Nat Genet. 2016;48(8):856–866. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Islam MR, International Headache Genetics Consortium (IHGC), Nyholt DR . Cross-trait analyses identify shared genetics between migraine, headache, and glycemic traits, and a causal relationship with fasting proinsulin. Hum Genet. 2023;142(8):1149–1172. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Hautakangas H, Winsvold BS, Ruotsalainen SE, Bjornsdottir G, Harder AVE, Kogelman LJA, Thomas LF, Noordam R, Benner C, Gormley P, Artto V, Banasik K, Bjornsdottir A, Boomsma DI, Brumpton BM, Burgdorf KS, Buring JE, Chalmer MA, de Boer I, Dichgans M, Erikstrup C, Färkkilä M, Garbrielsen ME, Ghanbari M, Hagen K, Häppölä P, Hottenga JJ, Hrafnsdottir MG, Hveem K, Johnsen MB, Kähönen M, Kristoffersen ES, Kurth T, Lehtimäki T, Lighart L, Magnusson SH, Malik R, Pedersen OB, Pelzer N, Penninx BWJH, Ran C, Ridker PM, Rosendaal FR, Sigurdardottir GR, Skogholt AH, Sveinsson OA, Thorgeirsson TE, Ullum H, Vijfhuizen LS, Widén E, van Dijk KW, International Headache Genetics Consortium, HUNT All-in Headache, Danish Blood Donor Study Genomic Cohort, Aromaa A, Belin AC, Freilinger T, Ikram MA, Järvelin MR, Raitakari OT, Terwindt GM, Kallela M, Wessman M, Olesen J, Chasman DI, Nyholt DR, Stefánsson H, Stefansson K, van den Maagdenberg AMJM, Hansen TF, Ripatti S, Zwart JA, Palotie A, Pirinen M . Genome-wide analysis of 102,084 migraine cases identifies 123 risk loci and subtype-specific risk alleles. Nat Genet. 2022;54(2):152–160. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Klarin D, Devineni P, Sendamarai AK, Angueira AR, Graham SE, Shen YH, Levin MG, Pirruccello JP, Surakka I, Karnam PR, Roychowdhury T, Li Y, Wang M, Aragam KG, Paruchuri K, Zuber V, Shakt GE, Tsao NL, Judy RL, Vy HMT, Verma SS, Rader DJ, Do R, Bavaria JE, Nadkarni GN, Ritchie MD, VA Million Veteran Program, Burgess S, Guo DC, Ellinor PT, LeMaire SA, Milewicz DM, Willer CJ, Natarajan P, Tsao PS, Pyarajan S, Damrauer SM. Genome-wide association study of thoracic aortic aneurysm and dissection in the Million Veteran Program. Nat Genet. 2023;55(7):1106–1115. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Montoro DT, Haber AL, Biton M, Vinarsky V, Lin B, Birket SE, Yuan F, Chen S, Leung HM, Villoria J, Rogel N, Burgin G, Tsankov AM, Waghray A, Slyper M, Waldman J, Nguyen L, Dionne D, Rozenblatt-Rosen O, Tata PR, Mou H, Shivaraju M, Bihler H, Mense M, Tearney GJ, Rowe SM, Engelhardt JF, Regev A, Rajagopal J. A revised airway epithelial hierarchy includes CFTR-expressing ionocytes. Nature. 2018;560(7718):319–324. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Örd T, Lönnberg T, Nurminen V, Ravindran A, Niskanen H, Kiema M, Õunap K, Maria M, Moreau PR, Mishra PP, Palani S, Virta J, Liljenbäck H, Aavik E, Roivainen A, Ylä-Herttuala S, Laakkonen JP, Lehtimäki T, Kaikkonen MU. Dissecting the polygenic basis of atherosclerosis via disease-associated cell state signatures. Am J Hum Genet. 2023;110(5):722–740. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Li Y, Ren P, Dawson A, Vasquez HG, Ageedi W, Zhang C, Luo W, Chen R, Li Y, Kim S, Lu HS, Cassis LA, Coselli JS, Daugherty A, Shen YH, LeMaire SA. Single-cell transcriptome analysis reveals dynamic cell populations and differential gene expression patterns in control and aneurysmal human aortic tissue. Circulation. 2020;142(14):1374–1388. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Fernandez DM, Rahman AH, Fernandez NF, Chudnovskiy A, Amir EAD, Amadori L, Khan NS, Wong CK, Shamailova R, Hill CA, Wang Z, Remark R, Li JR, Pina C, Faries C, Awad AJ, Moss N, Björkegren JLM, Kim-Schulze S, Gnjatic S, Ma’ayan A, Mocco J, Faries P, Merad M, Giannarelli C. Single-cell immune landscape of human atherosclerotic plaques. Nat Med. 2019;25(10):1576–1588. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Lutgens E, Faber B, Schapira K, Evelo CTA, van Haaften R, Heeneman S, Cleutjens KBJM, Bijnens AP, Beckers L, Porter JG, Mackay CR, Rennert P, Bailly V, Jarpe M, Dolinski B, Koteliansky V, de Fougerolles T, Daemen MJAP. Gene profiling in atherosclerosis reveals a key role for small inducible cytokines: validation using a novel monocyte chemoattractant protein monoclonal antibody. Circulation. 2005;111(25):3443–3452. [DOI] [PubMed] [Google Scholar]
22.Franzén O, Ermel R, Cohain A, Akers NK, Di Narzo A, Talukdar HA, Foroughi-Asl H, Giambartolomei C, Fullard JF, Sukhavasi K, Köks S, Gan LM, Giannarelli C, Kovacic JC, Betsholtz C, Losic B, Michoel T, Hao K, Roussos P, Skogsberg J, Ruusalepp A, Schadt EE, Björkegren JLM. Cardiometabolic risk loci share downstream cis- and trans-gene regulation across tissues and diseases. Science. 2016;353(6301):827–830. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Rouhana J, Wang J, Eraslan G, Anand S, Hamel A, Cole B, Regev A, Aguet F, Ardlie K, and Segrè A. ECLIPSER: identifying causal cell types and genes for complex traits through single cell enrichment of e/sQTL-mapped genes in GWAS loci. BioRxiv. Published online 2021. [Google Scholar]
24.Eraslan G, Drokhlyansky E, Anand S, Fiskin E, Subramanian A, Slyper M, Wang J, Van Wittenberghe N, Rouhana JM, Waldman J, Ashenberg O, Lek M, Dionne D, Win TS, Cuoco MS, Kuksenko O, Tsankov AM, Branton PA, Marshall JL, Greka A, Getz G, Segrè AV, Aguet F, Rozenblatt-Rosen O, Ardlie KG, Regev A. Single-nucleus cross-tissue molecular reference maps toward understanding disease gene function. Science. 2022;376(6594):eabl4290. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Chou EL, Chaffin M, Simonson B, Pirruccello JP, Akkad AD, Nekoui M, Lino Cardenas CL, Bedi KC Jr, Nash C, Juric D, Stone JR, Isselbacher EM, Margulies KB, Klattenhoff C, Ellinor PT, Lindsay ME. Aortic cellular diversity and quantitative genome-wide association study trait prioritization through single-nuclear RNA sequencing of the aneurysmal human aorta. Arterioscler Thromb Vasc Biol. 2022;42(11):1355–1374. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Kuleshov MV, Jones MR, Rouillard AD, Fernandez NF, Duan Q, Wang Z, Koplev S, Jenkins SL, Jagodnik KM, Lachmann A, McDermott MG, Monteiro CD, Gundersen GW, Ma’ayan A. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 2016;44(W1):W90–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Aguet F, Barbeira AN, Bonazzola R, Brown A, Castel SE, Jo B, Kasela S, Kim-Hellmuth S, Liang Y, Oliva M, Parsana PE, Flynn E, Fresard L, Gaamzon ER, Hamel AR, He Y, Hormozdiari F, Mohammadi P, Muñoz-Aguirre M, Park Y, Saha A, Segrć AV, Strober BJ, Wen X, Wucher V, Das S, Garrido-Martín D, Gay NR, Handsaker RE, Hoffman PJ, Kashin S, Kwong A, Li X, MacArthur D, Rouhana JM, Stephens M, Todres E, Viñuela A, Wang G, Zou Y, Brown CD, Cox N, Dermitzakis E, Engelhardt BE, Getz G, Guigo R, Montgomery SB, Stranger BE, Im HK, Battle A, Ardlie KG, Lappalainen T, The GTEx Consortium. The GTEx Consortium atlas of genetic regulatory effects across human tissues. bioRxiv. Published online October 3, 2019. doi: 10.1101/787903 [DOI] [Google Scholar]
28.Ghoussaini M, Mountjoy E, Carmona M, Peat G, Schmidt EM, Hercules A, Fumis L, Miranda A, Carvalho-Silva D, Buniello A, Burdett T, Hayhurst J, Baker J, Ferrer J, Gonzalez-Uriarte A, Jupp S, Karim MA, Koscielny G, Machlitt-Northen S, Malangone C, Pendlington ZM, Roncaglia P, Suveges D, Wright D, Vrousgou O, Papa E, Parkinson H, MacArthur JAL, Todd JA, Barrett JC, Schwartzentruber J, Hulcoop DG, Ochoa D, McDonagh EM, Dunham I. Open Targets Genetics: systematic identification of trait-associated genes using large-scale genetics and functional genomics. Nucleic Acids Res. 2021;49(D1):D1311–D1320. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Kalluri AS, Vellarikkal SK, Edelman ER, Nguyen L, Subramanian A, Ellinor PT, Regev A, Kathiresan S, Gupta RM. Single-Cell Analysis of the Normal Mouse Aorta Reveals Functionally Distinct Endothelial Cell Populations. Circulation. 2019;140(2):147–163. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Wirka RC, Wagh D, Paik DT, Pjanic M, Nguyen T, Miller CL, Kundu R, Nagao M, Coller J, Koyano TK, Fong R, Woo YJ, Liu B, Montgomery SB, Wu JC, Zhu K, Chang R, Alamprese M, Tallquist MD, Kim JB, Quertermous T. Atheroprotective roles of smooth muscle cell phenotypic modulation and the TCF21 disease gene as revealed by single-cell analysis. Nat Med. 2019;25(8):1280–1289. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Hamel AR, Yan W, Rouhana JM, Monovarfeshani A, Jiang X, Mehta PA, Advani J, Luo Y, Liang Q, Rajasundaram S, Shrivastava A, Duchinski K, Mantena S, Wang J, van Zyl T, Pasquale LR, Swaroop A, Gharahkhani P, Khawaja AP, MacGregor S, International Glaucoma Genetics Consortium (IGGC), Chen R, Vitart V, Sanes JR, Wiggs JL, Segrè AV. Integrating genetic regulation and single-cell expression with GWAS prioritizes causal genes and cell types for glaucoma. Nat Commun. 2024;15(1):396. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.GTEx Consortium. The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science. 2020;369(6509):1318–1330. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Hao K, Ermel R, Sukhavasi K, Cheng H, Ma L, Li L, Amadori L, Koplev S, Franzén O, d’Escamard V, Chandel N, Wolhuter K, Bryce NS, Venkata VRM, Miller CL, Ruusalepp A, Schunkert H, Björkegren JLM, Kovacic JC. Integrative Prioritization of Causal Genes for Coronary Artery Disease. Circ Genom Precis Med. 2022;15(1):e003365. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials