Mutational landscape of atherosclerotic plaques reveals large clonal cell populations

Abstract

The notion of clonal cell populations in human atherosclerosis has been suggested but not demonstrated. Somatic mutations are used to define cellular clones in tumors. Here, we characterized the mutational landscape of human carotid plaques through whole-exome sequencing to explore the presence of clonal cell populations. Somatic mutations were identified in 12 of 13 investigated plaques, while no mutations were detected in 11 non-atherosclerotic arteries. Mutated clones often constituted over 10% of the sample cell population, with genes related to the contractile apparatus enriched for mutations. In carriers of clonal hematopoiesis of indeterminate potential (CHIP), hematopoietic clones had infiltrated the plaque tissue and constituted substantial fractions of the plaque cell population alongside locally expanded clones. Our findings establish somatic mutations as a common feature of human atherosclerosis and demonstrate the existence of mutated clones expanding locally, as well as CHIP clones invading from the circulation. While our data do not support plaque monoclonality, we observed a pattern suggesting the coexistence of multiple mutated clones of considerable size spanning different regions of plaques. Mutated clones are likely to be relevant to disease development, and somatic mutations will serve as a convenient tool to uncover novel pathological processes of atherosclerosis in future studies.

Keywords: Atherosclerosis; Clonal selection; Genetic variation; Genetics; Vascular biology.

Figure 1. Somatic mutations and locally expanded clonal cell populations are inherent features of atherosclerosis.

(A) Carotid atherosclerotic plaques from 13 patients undergoing carotid endarterectomy were segmented into 1–4 samples. DNA extracted from these segments was analyzed by whole-exome sequencing, with patient-matched buffy coat DNA serving as reference. (B) Non-atherosclerotic arterial tissue samples were obtained from ascending thoracic aortas (ATAs, n = 5) and internal thoracic arteries (ITAs, n = 6) of 11 patients undergoing coronary artery bypass surgery. ITA samples were subdivided into an average of 3 samples each, leading to a total of 18 ITA samples. (C) Bar plot showing the number of plaque- or arterial tissue–confined somatic mutations (i.e., mutations not identified in patient-matched buffy coats) for each patient. (D) Frequency of clonal cells carrying a specific somatic mutation in plaque samples for each patient. Mutation effect is indicated by dot color, and variant reads is indicated by dot size. For each patient, the mutated gene representing the highest clonal cell frequency is shown. (E) Density plot showing the distribution of clonal cell frequencies for each patient in which mutations were detected. (F) The number of mutations detected per sample is plotted against the estimated number of sample cells, which was calculated based on the sample’s DNA yield and assumes an average of 6.6 pg DNA per cell. (G) The median clonal cell frequency of mutations detected in each sample plotted against the estimated number of sample cells. Vertical bars indicate the interquartile range, while dot sizes show the number of mutations per sample. (H) Correlation between VAFs obtained from ddPCR and VAFs obtained from whole-exome sequencing of the same plaque samples was analyzed for 6 specific mutations by linear regression analysis. Extended analyses are provided in Supplemental Figure 2.

Figure 2. Mutated clones span several regions of the plaque.

(A) In 5 patients, somatic mutations were detected in multiple plaque segments from the same individual. In the top panel, mutations shared across more than one segment are represented as interconnected colored dots. The colors of the dots correspond to the associated gene symbols, and the size of the dots reflects the frequency of the clonal cells. The mutation effect is indicated by smaller dots within the main mutation dots. Intersegment distances are shown, providing an estimate of the physical extent of clonal populations. The bottom panel illustrates how each segment was subdivided into the samples analyzed by whole-exome sequencing. One plaque segment from patients 1, 2, 4, and 6 exhibited sufficient morphological integrity, allowing for dissection while maintaining the morphological context (colored). (B–E) Depicted are images of the plaque segments, illustrating the division process into distinct samples. (F–I) The number of mutations identified in each sample is represented by white numbers, while the number of shared mutations between samples is denoted by black numbers and the thickness of the intersample connection lines. (J–M) Mutations detected in each sample are shown as dots, with specific mutations shared among samples connected by lines. The size of each dot corresponds to the clonal cell fraction it represents, as indicated in the key. Of note, the sample area does not correspond to 100%, as it was increased to fit the dots. (N–Q) Possible interpretations of the distribution of clones harboring selected mutations are depicted. The dot color and size mirror the key in J–M. Additionally, a shaded area has been added for which the size corresponds to the clonal cell frequency for each sample, setting the sample area to 100%. The genes that are mutated are indicated.

Figure 3. Non-random distribution of genes mutated in plaque tissue.

(A) Dot plot showing the enriched gene ontology terms of the 334 genes having a loss-of-function or missense mutation across all patients. Dot sizes indicate number of mutated genes for each term. The ratio shows the coverage of a given term by mutated genes, and dot colors indicate significance level of the enrichment (Kolmogorov-Smirnov–like running sum statistic using Benjamini-Hochberg to control for false discovery rate). (B) Network plot showing enriched gene ontology terms and the mutated genes belonging to each term. (C) To evaluate the expression pattern of mutated genes, 3 single-cell RNA sequencing datasets of human plaques were used. The mean expression level of genes belonging to the contractile apparatus, in which we found a mutation, is plotted for each cell population. The mean expression level of all genes of the atherosclerosis transcriptome was plotted for each dataset in gray for comparison. Asterisks indicate that contractile genes have significantly higher expression as compared with the background gene population. *P < 0.05, **P < 0.01 by Mann-Whitney U test. fib, fibroblast; mSMC, modulated SMC; peri, pericyte; mac, macrophage; neu, neutrophil; EC, endothelial cell, T, T cell; B, B cell; pla, plasma cell; NK, natural killer cell; mast, mast cell; other, non-annotated clusters in original publications. (D) Differences in the expression of genes associated with the 3 enriched ontology terms (contractile apparatus, desmosome, and mitotic spindle pole) were analyzed between carotid plaques and non-lesioned control arteries (left) and between carotid plaques from symptomatic and asymptomatic patients (right). Expression levels for individual microarray probes are displayed for each gene. Red data points indicate significant differences between conditions (Student’s t test), while the size of the data points reflects the mean expression level for each probe across all plaque samples.

Figure 4. Contribution of clonal hematopoiesis of indeterminate potential (CHIP) clones to the carotid plaque cell population.

(A) Six of the 13 plaque donors were CHIP carriers. The plot shows clonal cell frequencies of CHIP mutations detected in buffy coat (blood leukocytes) DNA. For each patient, the identity of the mutated gene with the highest clonal cell frequency is shown. Dot sizes indicate variant reads, and colors indicate the type of mutation. The dashed line shows the defined limit of detection. (B) CHIP clonal cell frequencies in buffy coats (yellow bars) and in subdivided plaque segments (gray bars). Clonal cell frequencies are represented by the proportion of colored area within yellow bars or gray subsamples (relative to the subsample bar area). Mutation type is indicated by dot color. (C and D) Validation of VAFs for a TET2 loss-of-function mutation, identified in patient 8, was conducted by comparing VAFs derived from ddPCR with those from whole-exome sequencing of both buffy coat and plaque samples from patient 8 (by linear regression analysis), as well as from samples of other patients (serving as negative controls).