Background: Cardiovascular events, such as ischemic events and stroke, are still the major cause of death in western societies. Atherosclerosis, also of the carotid artery, is the underlying mechanism. Although the general risk factors are well known and studied, the exact molecular mechanisms of plaque destabilization and rupture remain unclear. Long non-coding RNAs (lncRNAs) have great potential as therapeutic targets, as their expression is highly cell-type and disease specific and they exert key regulatory functions by directly interacting with other RNAs, DNA, as well as proteins.

Materials and Methods: We utilized combined total (bulk) RNA and single cell (sc) RNA to study the transcriptome of advanced carotid artery lesions from patients undergoing carotid endarterectomy in our vascular surgery clinic. In addition to both sequencing tools, we performed hybridization-based RNA in situ sequencing (HybRISS) to indicate where cluster-defining genes (of the single cell and bulk datasets) are located within the plaques. All three methods allowed us to build a 3D (single cell, bulk, spatial) transcriptome map of advanced human plaques.

Results: In this current study, four sequencing datasets were investigated. Dataset 1 contains total RNA data from 13 early vs. 13 late lesions from the same individual patient. Dataset 2 contains total RNA data from 8 unstable vs. 6 stable lesions from individual patients. In both datasets, significantly de-regulated lncRNAs were isolated and integrated into two separate scRNA-seq datasets. Dataset 3 consists of 4 carotid artery plaques, applied to 10x Chromium System by our institute and Dataset 4, consisting of 3 carotid artery plaques, which was kindly provided by collaborators from Columbia University, New York, USA.

Dataset 1 showed 745 lncRNAs with p < 0.05 in late vs. early lesions, whereas 863 significantly de-regulated lncRNAs were found in dataset 2 (unstable vs. stable lesions). From these lncRNAs, 37 and 26 respectively could also be detected in both scRNA-seq datasets. Finally, 16 lncRNAs were cross-referenced between all four datasets. These 16 lncRNAs presented a cell-type specific expression pattern. Interestingly, 11 lncRNAs were significantly enriched in different smooth muscle cell (SMC) clusters.

Further were all newly identified lncRNAs conserved in pigs and mice, and could therefore be detected in two scRNA-seq datasets of carotids from genetically mutated (LDLR-/-) Yucatan mini-pigs, as well as in the inducible carotid artery plaque rupture model in ApoE-deficient mice. In addition, several of these targets could be observed in abdominal aortic aneurysm sequencing datasets of human, mini-pig and mouse, confirming their overall importance during vascular disease progression.

The cluster-defining genes from the human scRNA-seq data as well as newly identified and well studies SMC markers were then located in human carotid artery tissue sections using the HybRISS method to unravel their distinct location within advanced and destabilized human plaques.

Discussion: Taken together, our datasets, methods and animal-models demonstrate that combining bulk with scRNA-seq data and spatially resolved sequencing methods are powerful tools to identify and characterize novel lncRNAs being expressed by a certain cell-type in the disease process. This approach therefore leads to the identification and validation of novel regulators of human and experimental atherosclerotic plaque destabilization.