RNA Sequencing Reveals the Differentially Expressed circRNAs between Stable and Unstable Carotid Atherosclerotic Plaques

Objective This study aimed to identify circular RNA profiles (circRNAs) via high-throughput RNA sequencing and distinguish the differentially expressed (DE) circRNAs between stable and unstable plaques. Methods RNA sequencing was performed on unstable and stable carotid plaque samples obtained from patients with carotid artery stenosis. DE circRNAs were screened, and six DE circRNAs were verified using quantitative real-time PCR (qRT-PCR). Functional evaluation of the DE circRNAs was conducted via Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses. Results We screened 344 DE circRNAs in unstable plaques, consisting of 342 upregulated and 2 downregulated circRNAs. GO analysis showed that the host genes of the upregulated circRNAs were related to ER to Golgi transport vesicle membrane, endocytic vesicle membrane, and Ran GTPase binding. KEGG analysis revealed that the host genes of the upregulated circRNAs were primarily associated with protein processing in endoplasmic reticulum, lysine degradation, homologous recombination, epithelial cell signaling in Helicobacter pylori infection, and yersinia infection. The results of qRT-PCR verified three upregulated DE circRNAs and two downregulated DE circRNAs in unstable plaques. Conclusion Hsa-circ-0001523, hsa-circ-0008950, hsa-circ-0000571, hsa-circ-0001946, and hsa-circ-0000745 may be involved in regulating the stability of atherosclerotic plaques and serves as a therapeutic target for unstable plaques.


Introduction
Atherosclerosis is a chronic systemic infammatory disease of the arterial wall and is one of the foremost reasons for the morbidity and mortality of cardiovascular diseases worldwide. Te distinguishing feature of this disease is the formation of atherosclerotic plaques in the arteries, which are attributed to lipid accumulation, infammatory cell infltration, cell apoptosis, and increased extracellular matrix secretion [1]. Diferent from stable plaques, vulnerable/ unstable plaques have an active infammatory response that contributes to the thinning of the fbrous cap and to plaque rupture, which is a major cause of fatal cardiovascular diseases [2,3]. Studies have demonstrated that the dysregulation of genes that play critical roles in the stability of atherosclerotic plaques predisposes them to rupture and thrombosis formation [4,5]. Terefore, identifying the dysregulation of active molecules that afect the instability of plaques is important for the diagnosis, treatment, and prognosis of unstable atherosclerotic plaques.
Circular RNAs (circRNAs) are characteristic of unique covalently closed loops and constitute a new family of endogenous noncoding RNAs (ncRNAs). Owing to their regulatory activity in many biological processes, circRNAs have been recently recognized as potential diagnostic and therapeutic targets in various diseases [6,7]. As reported in the literature, biological activities associated with the presence of atherosclerotic plaques, including infammation, apoptosis, and lipid metabolism, are activated or deactivated by circRNAs [8,9]. Tere is more direct evidence that circRNAs are associated with atherosclerotic stroke [10,11] and afect atherosclerotic plaque stability [5]. However, there is a lack of comprehensive understanding of the infuence of circRNAs in plaque stability.
In this study, circRNAs expression patterns in human stable and unstable plaques were analyzed using RNA sequencing, and the diferentially expressed (DE) circRNAs and their potential functions in unstable plaques were further explored. Our fndings revealed a comprehensive understanding of the circRNA expression pattern associated with atherosclerotic plaque instability.

Samples from Patients with Carotid Stenosis.
Atherosclerotic plaques were collected from patients with carotid stenosis who underwent carotid endarterectomy in our hospital. Plaques were classifed based on criteria from the American Heart Association [12] by two independent investigators. Plaques in types I-VIII were categorized as stable, and plaques in types IV-VI as unstable. In this study, three patients with unstable plaques were enrolled, and three patients with stable plaques were enrolled, with their clinical information as shown in Table 1. A total of three stable atherosclerotic plaque samples and three unstable atherosclerotic plaque samples were obtained and placed immediately in liquid nitrogen. Plaques were stored in a −80°C fridge until RNA sequencing and quantitative real-time PCR (qRT-PCR) assay were performed. Our research protocols have been approved by the Ethics Committee of Shanghai Pudong Hospital. All participants provided informed consent before the study procedure began.

Total RNA Extraction and Library Construction. Te
TRIzol reagent (Termo, USA) was employed to extract total RNA from the stable and unstable groups. RNA samples were strictly controlled in three aspects: I. RNA integrity and presence of DNA contamination detected using agarose gel electrophoresis; II. RNA purity detected using a NanoDrop spectrophotometer (Termo Scientifc, USA); III. accurate detection of RNA integrity via an Agilent 2100 bioanalyzer (Agilent Technologies, USA).
To perform library sequencing, ribosomal RNA was removed from the total RNA using a Ribo-Zero Magnetic Kit (EpiCentre, Beijing, China). Te generated RNAs were interrupted randomly and then used for library preparation with the Illumina Truseq ™ RNA Sample Prep Kit with Ribo-Zero (Illumina, USA). Synthesis of the frst-strand cDNA was catalyzed using SuperScript II reverse transcriptase, and the second-strand cDNA synthesis was catalyzed via DNA polymerase I and RNase H. Te generated cDNAs were adenylated at the 3′ end and ligated using Illumina PE adapter oligonucleotides. Te library was obtained using DNA fragments enriched via 15 cycles of PCR with Illumina PCR Primer Cocktail.

RNA Sequencing.
All the samples were sequenced via sequencing-by-synthesis technology on a NovaSeq 6000 platform. Te produced raw data in FASTQ format were subjected to qualitative fltering to remove low-quality reads (10% of reads containing N; small fragments less than 25 bp in length after quality clip) using Cutadapt (v1.15) software. Te generated clean data were mapped to the reference genome (GRCh38) on HISAT2 v2.0.5 software, and circRNAs were annotated using CIRI2 software.

DE circRNA Identifcation.
Te intersample Pearson correlation of gene expression levels was analyzed to evaluate the rationality of sample selection. For DE circRNA analysis, the read count of circRNAs (back-spliced junction read) was normalized, and the resulting P values were adjusted to control the false discovery rate (FDR). DE circRNAs between the stable and unstable plaque tissues were analyzed using DESeq2 based on the screening criteria of |log 2 fold change| ≥ 1 and FDR < 0.05 [13].

Gene Ontology (GO) Terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway
Analysis of the Upregulated circRNAs. Te potential functions of the upregulated circRNAs in unstable plaques were predicated by the annotation of GO terms and analysis of KEGG pathways using clusterProfler in R package. Tere were three GO categories for enriched host genes, namely, biological process (BP), cell component (CC), and molecular function (MF) [14]. P value <0.05 represented the signifcance of enriched terms or KEGG pathways.
2.6. qRT-PCR. Six DE circRNAs, i.e., four upregulated circRNAs and two downregulated circRNAs, were selected for qRT-PCR verifcation (n � 3 for each group). Briefy, total RNA was obtained from three stable and three unstable homogenized plaque tissues using RNAiso Plus (TAKARA 9190; Takara Biomedical Technology (Beijing) Co., Ltd, China). Te purifed RNA precipitates were dissolved in DEPC H 2 O, and a spectrophotometer was used to measure the purity and concentration of the isolated total RNA. Subsequently, total RNA (1000 ng) was reverse transcribed  Table 2. Te relative expression levels of the related DE circRNAs were determined using the 2 −ΔΔCT method, with GAPDH as the internal control.

Statistical Analysis.
Te data were presented as mean-± standard deviation. A comparison between the two groups was conducted via Student's t test in Graphpad Prism 5 software (USA). A P value <0.05 was considered statistically signifcant.

circRNA Profles in Plaques by RNA Sequencing and Identifcation of DE circRNAs.
A total of 2450 circRNAs were annotated by RNA sequencing in 3 stable plaques and 3 unstable plaques. Based on the |log 2 FC| ≥ 1 and FDR < 0.05 criteria and comparing the stable and unstable plaques, we identifed 344 DE circRNAs in the unstable plaques, comprising 342 upregulated and 2 downregulated DE circRNAs ( Figure 1(a)). Te most evident DE circRNAs are shown in volcano plots in Figure 1(b). Table 3 displays the top 15 upregulated circRNAs from the unstable plaques.

GO Terms and KEGG Pathway Enrichment of the Upregulated DE circRNAs.
Given that most DE circRNAs were upregulated, we predicted the function of the upregulated DE circRNAs based on the GO and KEGG databases. GO enrichment analysis focused on the three categories (BP, CC, and MF). According to the enriched host gene of upregulated circRNAs, we ranked the top 20 GO terms in the three categories in Figure 2. Notably, the most enriched GO terms in each of the three categories were isotype switching to IgG isotypes, mitochondrial calcium uptake, antigen processing and presentation in BP (Figure 2(a)), endocytic vesicle, ER to Golgi transport vesicle membrane, and MHC class II protein complex in CC (Figure 2(b)), as well as vinculin binding, Ran GTPase binding, and potassium channel regulator activity in MF (Figures 2(c)). Additionally, KEGG pathway enrichment analysis showed that the host genes of the identifed DE circRNAs were mainly related to protein processing in the endoplasmic reticulum, other glycan degradation, lysine degradation, homologous recombination, epithelial cell signaling in Helicobacter pylori infection, neurotrophin signaling pathways, biosynthesis of nucleotide sugars, aminoacyl-tRNA biosynthesis, endocytosis, and yersinia infection (Figure 3(a)). Based on the number of pathway-related genes, those pathways were sorted as endocytosis, protein processing in the endoplasmic reticulum, yersinia infection, neurotrophin signaling pathways, epithelial cell signaling in H. pylori infection, lysine degradation, aminoacyl-tRNA biosynthesis, homologous recombination, biosynthesis of nucleotide sugars, and other glycan degradation (Figure 3(b)).

Validation of RNA Sequencing Using qRT-PCR.
Using the qRT-PCR assay, the RNA sequencing results were verifed in four upregulated DE circRNAs (hsa-circ-0001523, hsa-circ-0008950, hsa-circ-0000571, and hsa-circ-008267) and two downregulated DE circRNAs (hsa-circ-0001946 and hsa-circ-0000745). Te results showed that compared with the stable plaques, the expression levels of hsa-circ-0001523, hsa-circ-0008950, and hsa-circ-0000571 were signifcantly increased in the unstable plaques (P < 0.05), whereas the expression levels of hsa-circ-0001946 and hsa-circ-0000745 were signifcantly decreased (P < 0.05, Figure 4), consistent with the expression patterns of the sequencing. For hsa-circ-0008267, there was no signifcant diference in its expression between the stable and unstable plaques (P > 0.05, Figure 4). Te results indicated that the concordance rate of the qRT-PCR results and sequencing data was approximately 83.33%, which implies the relatively high reliability of RNA sequencing.

Discussion
RNA sequencing supported by biochemical enrichment strategies and in-depth bioinformatic approaches allows comprehensive studies of ncRNAs in the pathogenesis of diseases [6]. Te dysregulation of ncRNA expression profles was found in a diferent stage of atherosclerosis progression [15]. Nevertheless, a comprehensive understanding of the expression and function of circRNAs in unstable plaques is lacking. Tis study identifed DE circRNAs between unstable and stable plaques and found 344 DE circRNAs in the unstable plaques. Ultimately, the dysregulation of the DE circRNAs was validated in the unstable plaques by qRT-PCR.
CircRNAs have attracted increasing attention for their role and function in the manipulation of multiple metabolic and signaling pathways in the pathogenesis of diseases through their efect on a variety of biological functions, such as their absorbance of microRNAs or proteins and direct control of protein function [16]. Te characteristics of the conserved expression of circRNAs and their infuence on gene expression indicate that the dysregulation of circRNAs plays an important role in physiological and pathological changes. A previous review summarized the studies that have demonstrated the value of circRNAs as diagnostic biomarkers and therapeutic targets for atherosclerosis [17]. One study by Yu et al. indicated that circRNAs play a role in protecting atherosclerotic plaque stability [5]. Te efect of circRNAs on plaque stability indicates a promising direction for the management of plaque rupture. Tis led us to explore abnormally expressed circRNAs in unstable plaques as compared to that in stable plaques. Our RNA sequencing showed that of a total of 2450 annotated circRNAs, there were 342 upregulated circRNAs and 2 downregulated Genetics Research 3 circRNAs; this is a large volume of data regarding DE circRNAs in unstable plaques. Our RT-qPCR results also showed that hsa-circ-0001523, hsa-circ-0008950, and hsacirc-0000571 were upregulated, while hsa-circ-0001946 and hsa-circ-0000745 were downregulated in the unstable plaques. Reportedly, some DE circRNAs are important participants in atherosclerotic diseases. For instance, hsa-circ-0001946 was identifed in peripheral blood, demonstrating an intimate relationship with coronary atherosclerotic heart disease [18]. A case-control study by Sun et al. found that the correlation of hsa_circ_0001946 to coronary heart disease is independent of other common environmental risk factors [19]. Tese data indicate the critical role of hsa_circ_0001946 in vascular pathological change; however, the mechanism of action is largely unknown and needs further exploration. Besides hsa_circ_0001946, other DE circRNAs are not functionally understood, especially in the progression of atherosclerosis and plaque stability. Nevertheless, our fnding is encouraging, but the function of those dysregulated circRNAs needs further explanation. We further analyzed the potential efects of the 342 upregulated circRNAs on plaque stability by GO term analysis of their host genes and concluded that the host genes of the upregulated circRNAs are, in the basic    Genetics Research 5 pathological processes, indispensable to plaque stability. Remarkably, the GO terms enriched most by upregulated circRNAs in biological process and molecular function were ER to Golgi transport vesicle membrane and Ran GTPase binding. Te Golgi apparatus is the biosynthetic center of the secretory pathway and uses vesicle transport to deliver substances to a specifc part of the cell or plasma membrane [20]. Golgi is responsible for modifcations of the nascent very low-density lipoprotein (VLDL) and the vesicle transport of the VLDL, the precursor for low-density lipoprotein (LDL), to the plasma membrane [21]. A high concentration of plasma LDL cholesterol is one of the critical risk factors that accelerate the progression of atherosclerosis. Golgi vesicle transport plays a role in cholesterol efux, and with the export of excess cellular cholesterol by cells, Golgi vesicle transport to the plasma membrane increased twofold [22]. Cholesterol in the dense LDL fractions afects carotid plaque cellular composition [23], whereby low LDL cholesterol accompanies a signifcantly lower prevalence of plaque rupture [24]. Another enriched GO term was Ran GTPase binding, a contributor to GTPase activation. Activated GTPase is generally involved in the pathological process of atherosclerosis [25,26]. Tus, we deduce that circRNAs may be involved in controlling plaque stability by afecting cholesterol efux and GTPase activation. Te mechanism of plaque formation and instability is a complex process involving multiple metabolic and signaling pathways [27]. We further performed KEGG analysis to reveal the host genes of signifcantly upregulated circRNA-related KEGG pathways and found that those dysregulated circRNAs were enriched in protein processing in the endoplasmic reticulum and epithelial cell signaling in H. pylori infection. Te endoplasmic reticulum is a type of eukaryotic membranous system that provides a venue for some protein processing and lipid synthesis. Te function and dysfunction of protein processing in the endoplasmic reticulum play an important role in plaque stability [28,29]. H. pylori infection activates multiple signal transduction pathways that promote inappropriate infammatory responses and contributes to pathological changes in lipid metabolism and epithelial cell proliferation, survival, and function [30]. Tose events associated with H. pylori are closely related to plaque formation and stability [31,32]. Overall, functional analysis by GO and KEGG annotation revealed the important roles of upregulated circRNAs in regulating the plaque stability of the arterial wall. Nevertheless, further studies are warranted to confrm the function of those dysregulated circRNAs in unstable plaques.
Notwithstanding the fndings, this study has some limitations. First, the sample size was small, and our fndings need to be further confrmed with a larger sample size. Moreover, the specifc roles and mechanisms of the identifed dysregulated circRNAs in the progression of the unstable plaques should be further investigated in vitro and in vivo.
In summary, this study identifed circRNA expression profles in stable and unstable plaques and identifed plaquestability-associated dysregulated circRNAs in unstable plaques. Our results provide a comprehensive understanding of circRNAs involved in plaque stability of carotid stenosis and provide a theoretical basis for circRNAs (such as hsa-circ-  0001523, hsa-circ-0008950, hsa-circ-0000571, hsa-circ-0001946, and hsa-circ-0000745) to be used as potential targets to prevent or diagnose plaque rupture of carotid stenosis patients in clinical settings.

Data Availability
Te data used to support the fndings of this study are included within the article.
(3) KEGG analysis suggests that DE circRNAs were mainly associated with protein processing in endoplasmic reticulum.

Disclosure
Xueguang Lin and Ying Deng are the co-frst authors.

Conflicts of Interest
Te authors declare that they have no conficts of interest.