Human umbilical vein endothelial cells (HUVECs) were obtained from the Chinese Academy of Medical Sciences, Shanghai Institute Cell Bank (Shanghai, China). HUVECs were cultivated using Dulbecco's modified Eagle's medium (Invitrogen, Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS, Sigma, St. Louis, MO, USA) and 1% antibiotic mixture (Sangon Biotech, Shanghai, China) at 37°C with 5% CO₂. HUVECs at passages 3–8 were utilized in all experiments. The utilization of HUVECs has gotten the permission of the Ethics Committee of Weifang Hospital of Traditional Chinese Medicine.

The AS cell model was established through exposing HUVECs to 40 mg/L ox-LDL (Solarbio, Beijing, China) for 24 h.

Cell viability was analyzed by MTT assay. Transfected HUVECs in 96-well plates were incubated with 20 μL MTT reagent (Sigma), and the culture supernatant was discarded 2 h later. A total of 150 μL dimethylsulfoxide (DMSO; Sigma) solution was added into each well, and cell culture plates were shaken for 0.5 h. The optical density at 570 nm was determined using the microplate reader (Bio-Rad, Hercules, CA, USA).

RNA samples were isolated using Trizol reagent (Invitrogen). Isolated RNAs were used to synthesize complementary DNA (cDNA) using a MicroRNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) and TaqMan Reverse Transcription Reagents (Invitrogen). qPCR reaction was performed using a mirVana^TM RT-qPCR miRNA Detection Kit (Ambion, Austin, TX, USA) and SYBR^TM Green PCR Master Mix (Invitrogen). The relative expression levels were calculated by the 2^−ΔΔCt method. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as the internal reference for circ_0068087 and ROBO1, and U6 functioned as the reference for miR-186-5p. The primer sequences are shown in Table 1.

The cytoplasmic and nuclear RNAs were extracted using the PARIS^TM Kit Protein and RNA Isolation system (Thermo Fisher Scientific, Waltham, MA, USA) in accordance with the manufacturer's instructions.

To silence circ_0068087, small interfering (si)RNA targeting circ_0068087 (si-circ_0068087; 5'-GTGCTTTTAACCAGAATTACA-3') was synthesized to target the junction sites of circ_0068087, and negative control of siRNA (si-NC; 5'-GTGTGCAGTAGCAGTAA-3') was utilized as the control. MiR-186-5p mimic (5'-TCGGGTTTTCCTCTTAAGAAAC-3') or miR-186-5p inhibitor (5'-ACGTAGGACTGGACAAAC-3') was synthesized to upregulate or silence miR-186-5p, and miRNA NC (5'-CTGATTAGCATACAGTGG-3') and inhibitor NC (5'-GTGCGTAGGCATTACAGTA-3') were utilized as the controls. To upregulate ROBO1, the ectopic expression plasmid of ROBO1 using pcDNA vector (pc-ROBO1) was constructed, and pc-NC was utilized as the control. RNAs and plasmids provided by GenePharma (Shanghai, china) and Sangon Biotech were transfected into HUVECs using the Lipofectamine 3000 (Invitrogen) reagent when cell confluence reached about 70%.

HUVECs were collected and resuspended in 200 μL binding buffer (Qiagen, Valencia, CA, USA). Afterward, Annexin V-fluorescein isothiocyanate (Annexin V-FITC; Qiagen) and propidium iodide (PI; Qiagen) were pipetted to incubate with HUVECs. The apoptosis rate of HUVECs was analyzed by the BD FACSCalibur flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA).

HUVECs were washed using ice-cold phosphate buffer saline (PBS; Sangon Biotech) and disrupted using whole cell lysate buffer (Beyotime, Shanghai, China). Equal amounts of protein samples were electrophoretically separated on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and were transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore, Billerica, MA, USA). Afterward, the membrane was blocked through incubating using 5% milk and was incubated with the primary antibodies overnight. Primary antibodies including antiproliferating cell nuclear antigen (anti-PCNA; ab92552), anti-B cell leukemia/lymphoma 2 (anti-Bcl-2; ab182858), anti-ROBO1 (ab7279), anti-Ki67 (ab16667), anti-Bcl-2 associated X, apoptosis regulator (anti-Bax; ab32503), and anti-GAPDH (ab8245), were purchased from Abcam (Cambridge, MA, USA). The secondary antibody (Abcam; ab205718) labeled with horseradish peroxidase was then incubated with the membrane for 2 h. The protein signals were visualized using an enhanced chemiluminescent (ECL) chromogenic substrate (Beyotime).

The levels of interleukin 6 (IL-6; D6050), IL-1β (201-LB) and tumor necrosis factor α (TNF-α; MTA00B) in the culture supernatant were analyzed using the Human IL-6/IL-1β/TNF-α Quantikine ELISA Kit (R&D Systems, Minneapolis, MN, USA) according to the manufacturer's instructions.

The levels of MDA (A003-4) and ROS (E004) and the activity of SOD (A001-3) were analyzed by their corresponding kits (Jiancheng Biotech, Nanjing, China) in accordance with the manufacturer's instructions.

Bioinformatic database circinteractome (https://circinteractome.irp.nia.nih.gov) was utilized to predict the circ_0068087-miRNAs interactions, and the StarBase database (http://starbase.sysu.edu.cn) was utilized to predict the miR-186-5p/mRNA interactions.

The WT-circ_0068087 and WT-ROBO1-3'UTR reporter plasmids were constructed through inserting the partial fragment of circ_0068087 or ROBO1 3'UTR into the psiCHECK2 (Promega, Madison, WI, USA) vector. The GeneArt^TM Site-Directed Mutagenesis System (Invitrogen) was utilized to amplify the mutant fragment. The mutant fragment of circ_0068087 or ROBO1 3'UTR, including the mutant binding sites with miR-186-5p, was also inserted into the psiCHECK2 vector (Promega) to generate MUT-circ_0068087 and MUT-ROBO1-3'UTR. HUVECs were cotransfected with luciferase plasmids and miR-186-5p mimic or miRNA NC. The luciferase intensity was determined using a commercial Dual-Luciferase Reporter Assay Kit (Promega).

The Magna RIP^TM RNA-Binding Protein Immunoprecipitation Kit (Millipore) was utilized to test whether miR-186-5p was a target of circ_0068087 in HUVECs. HUVECs were disrupted in RIP lysis buffer, and cell extracts were mixed with RIP buffer containing Argonaute2 (Ago2) antibody (Abcam, ab186733)- or immunoglobulin G (IgG) antibody (Abcam, ab172730)-precoated magnetic beads. RNA enrichment was determined by RT-qPCR.

Data were analyzed using GraphPad Prism 7.0 software (GraphPad, La Jolla, CA, USA) and were expressed in the form of mean ± standard deviation (SD). The unpaired Student t-test was utilized to assess the differences in two groups. The differences in multiple groups were assessed using the one-way analysis of variance (ANOVA) followed by Tukey's post hoc test. P < 0.05 was considered statistically significant.

We find that ox-LDL exposure reduced the viability of HUVECs in a dose- and time-dependent manner (Figures 1A,B). ox-LDL (40 mg/L; 24 h) was chosen for further experiments. ox-LDL treatment significantly upregulated circ_0068087 expression in HUVECs (Figure 1C). We assessed the subcellular localization of circ_0068087 in HUVECs prior to exploring its biological function. Circ_0068087 was mainly localized in the cytoplasmic fraction of HUVECs (Figure 1D), which endowed circ_0068087 the potential to serve as a miRNA sponge.

To explore whether ox-LDL-induced effects on HUVECs were partly attributed to the upregulation of circ_0068087, HUVECs were transfected with si-circ_0068087 prior to ox-LDL exposure. High transfection efficiency of si-circ_0068087 was confirmed via RT-qPCR assay in HUVECs (Figure 2A). ox-LDL exposure suppressed cell viability, which was largely counteracted by circ_0068087 silencing in HUVECs (Figure 2B). Cell apoptosis was triggered by ox-LDL stimulation, and the interference of circ_0068087 restrained the apoptosis of ox-LDL-induced HUVECs (Figure 2C). To further explore whether ox-LDL restrained cell viability and induced cell apoptosis via upregulating circ_0068087, we measured the protein levels of proliferation markers (PCNA and Ki67), anti-apoptotic protein (Bcl-2), and pro-apoptotic protein (Bax). Circ_0068087 silencing rescued the levels of PCNA, Ki67, and Bcl-2 and decreased the expression of Bax in ox-LDL-induced HUVECs (Figures 2D,E, Supplementary Figure 1A). ox-LDL exposure induced the release of inflammatory cytokines (IL-6, IL-1β, and TNF-α), and circ_0068087 silencing largely attenuated ox-LDL-induced influences (Figures 2F–H), demonstrating that ox-LDL-induced inflammation of HUVECs was partly based on the upregulation of circ_0068087. Oxidative stress contributes to the progression of AS. Vascular oxidative stress induces multiple molecular events in AS progression, including oxidative modification of lipoproteins and phospholipids, macrophage infiltration, and foam cell formation (17). We determined the levels or activity of oxidative stress-associated markers (MDA, ROS, and SOD) to analyze cellular oxidative stress status. ox-LDL exposure increased the levels of MDA and ROS, whereas it inhibited the activity of SOD, and these effects were largely counteracted by the silence of circ_0068087 in HUVECs (Figures 2I–K). These findings suggest that ox-LDL-induced dysfunction in HUVECs was partly dependent on the upregulation of circ_0068087.

CircRNAs have shown their “miRNA sponge” role in regulating cellular physiological and pathological processes (18, 19). We performed bioinformatic analysis using circinteractome and circbank databases to predict the possible targets of circ_0068087. There were 43 candidate miRNA targets of circ_0068087 predicted by both databases (Supplementary Figure 2A). Among these miRNAs, we screened out five miRNAs that were implicated in AS progression, including miR-186-5p (13), miR-140-3p (20), miR-640 (21), miR-515-5p (22), and miR-665 (23). MiR-186-5p was selected for further experiments due to it having the most significant negative regulatory relationship with circ_0068087 (Supplementary Figure 2A). The putative binding sites with miR-186-5p in circ_0068087 (3916 bp-3921 bp) are shown in Figure 3A. RT-qPCR verified the high overexpression efficiency of miR-186-5p mimic in HUVECs (Figure 3B). Transfection with miR-186-5p mimic significantly reduced the luciferase activity of wild-type luciferase reporter plasmid (WT-circ_0068087) rather than its mutant plasmid (MUT-circ_0068087) (Figure 3C), suggesting that miR-186-5p was a target of circ_0068087 in HUVECs. The RIP assay was employed to further confirm the target relationship between circ_0068087 and miR-186-5p. MiRNAs can form the ribonucleoprotein complexes (miRNPs) that include Ago2, the core component of the RNA-induced silencing complex (RISC) (24). As shown in Figure 3D, circ_0068087 and miR-186-5p were both enriched in the Ago2 antibody group. MiR-186-5p was enriched when using biotinylated circ_0068087 (biotin-circ_0068087) (Figure 3E). The results of the RIP assay and RNA pulldown assay further validated the interaction between circ_0068087 and miR-186-5p. ox-LDL exposure downregulated the expression of miR-186-5p in HUVECs (Figure 3F). The results of RT-qPCR confirmed the high knockdown efficiency of the miR-186-5p inhibitor in HUVECs (Figure 3G). Circ_0068087 silencing upregulated the expression of miR-186-5p, and the expression of miR-186-5p was reduced by the introduction of the miR-186-5p inhibitor in HUVECs (Figure 3H), demonstrating the negative regulatory relation between circ_0068087 and miR-186-5p. These results suggest that circ_0068087 negatively regulates miR-186-5p expression by sponging it in HUVECs.

Rescue experiments were performed through transfecting HUVECs with si-circ_0068087 alone or together with an miR-186-5p inhibitor prior to ox-LDL exposure. MiR-186-5p knockdown suppressed cell viability in circ_0068087-silenced HUVECs upon ox-LDL exposure (Figure 4A). MiR-186-5p silencing triggered cell apoptosis in circ_0068087-silenced HUVECs again upon ox-LDL treatment (Figure 4B). Circ_0068087 silencing mediated the promoting effect on the expression of PCNA, Ki67, and Bcl-2, and the suppressive effects on the level of Bax were largely reversed by the introduction of the miR-186-5p inhibitor in ox-LDL-induced HUVECs (Figures 4C,D, Supplementary Figure 1B). These results demonstrate that circ_0068087 knockdown attenuates ox-LDL-induced effects on the viability and apoptosis of HUVECs largely by upregulating miR-186-5p. Circ_0068087 silencing suppressed the cell inflammatory response, which was largely counteracted by the knockdown of miR-186-5p (Figures 4E–G). Circ_0068087 knockdown decreased the levels of MDA and ROS and increased the activity of SOD in ox-LDL-induced HUVECs, and these effects were largely alleviated by the addition of the miR-186-5p inhibitor (Figures 4H–J). Circ_0068087 silencing promoted the proliferation of ox-LDL-induced HUVECs, and the addition of the miR-186-5p inhibitor restrained cell proliferation again (Supplementary Figure 3A). In addition, we found that circ_0068087 knockdown facilitated cell migration, and cell migration was suppressed in the si-circ_0068087 and miR-186-5p inhibitor co-transfected group in ox-LDL-induced HUVECs (Supplementary Figure 3B). Taken together, ox-LDL induced the injury of HUVECs partly through upregulating circ_0068087 and downregulating miR-186-5p.

We used bioinformatic software StarBase to predict the downstream targets of miR-186-5p. Among all the predicted mRNA targets of miR-186-5p, we focused on five mRNAs due to their vital roles in AS progression, including ROBO1 (16), KLF4 (25), STAT3 (26), AURKA (27), and IFI44L (28). ROBO1 was selected for further analysis due to it having the most significant negative regulatory relationship with miR-186-5p (Supplementary Figure 2B). The putative binding sites between miR-186-5p and ROBO1 are shown in Figure 5A. Subsequently, the target interaction between miR-186-5p and ROBO1 was validated by the dual-luciferase reporter assay. Luciferase activity of wild-type plasmid (WT-ROBO1-3'UTR) was markedly decreased by the transfection of miR-186-5p mimic rather than miRNA NC (Figure 5B), suggesting that ROBO1 was a target of miR-186-5p in HUVECs. The miR-186-5p-binding sites in ROBO1 3'UTR were mutated to “AAGAAA,” and the luciferase activity of mutant plasmid (MUT-ROBO1-3'UTR) was unchanged by the transfection of miR-186-5p mimic or miRNA NC (Figure 5B), demonstrating that miR-186-5p bound to the 3'UTR of ROBO1 via the predicted sites. The binding relation between miR-186-5p and ROBO1 was also confirmed by RIP assay (Figure 5C). ox-LDL treatment increased the protein level of ROBO1 in HUVECs (Figure 5D). High transfection efficiency of ROBO1 plasmid (pc-ROBO1) was verified by Western blot assay (Figure 5E). MiR-186-5p overexpression reduced ROBO1 protein expression, and the protein level of ROBO1 was largely rescued by the addition of ROBO1 plasmid in HUVECs (Figure 5F). HUVECs were cotransfected with si-circ_0068087 and miR-186-5p inhibitor to analyze the regulation among circ_0068087, miR-186-5p, and ROBO1. As shown in Figure 5G, circ_0068087 knockdown reduced the protein expression of ROBO1, and the silence of miR-186-5p largely rescued the protein level of ROBO1 in HUVECs. These results suggest that ROBO1 was a target of miR-186-5p, and ROBO1 was regulated by circ_0068087/miR-186-5p axis in HUVECs.

HUVECs were transfected with miR-186-5p mimic alone or together with pc-ROBO1 prior to ox-LDL exposure to conduct rescue experiments. MiR-186-5p overexpression elevated cell viability and suppressed cell apoptosis, inflammation, and oxidative stress (Figures 6A–J), which further demonstrated that miR-186-5p protected HUVECs against ox-LDL-induced injury. The overexpression of ROBO1 suppressed cell viability in miR-186-5p-overexpressed HUVECs upon ox-LDL exposure (Figure 6A). Cell apoptosis was induced again in miR-186-5p and pc-ROBO1 cotransfected group in ox-LDL-exposed HUVECs (Figure 6B). MiR-186-5p overexpression-mediated upregulation in the expression of PCNA, Ki67, and Bcl-2 and downregulation in the level of Bax were largely reversed by the accumulation of ROBO1 in ox-LDL-stimulated HUVECs (Figures 6C,D, Supplementary Figure 1C), demonstrating that miR-186-5p protected HUVECs from an ox-LDL-induced suppressive effect on cell viability and promoting effect on cell apoptosis partly by reducing ROBO1 expression. The overexpression of ROBO1 largely rescued the release of inflammatory cytokines (IL-6, IL-1β, and TNF-α) in miR-186-5p-overexpressed HUVECs upon ox-LDL exposure (Figures 6E–G). MiR-186-5p overexpression reduced the levels of MDA and ROS and increased the activity of SOD although these effects were all counteracted by the overexpression of ROBO1 in ox-LDL-induced HUVECs (Figures 6H–J). Taken together, miR-186-5p protected HUVECs from ox-LDL-induced damage partly by reducing the ROBO1 level.

Vascular cellular adhesion molecule-1 (VCAM-1) and intracellular adhesion molecule-1 (ICAM-1) exert vital roles in recruiting inflammatory monocytes into the vascular wall and initiating AS (29). Based on former articles (30), ox-LDL induced the production of ICAM-1 and VCAM-1. We explored the role of the circ_0068087/miR-186-5p/ROBO1 axis in regulating the levels of VCAM-1 and ICAM-1 in HUVECs. Circ_0068087 interference reduced the levels of VCAM-1 and ICAM-1, and the overexpression of ROBO1 rescued the expression of VCAM-1 and ICAM-1 in ox-LDL-induced HUVECs (Supplementary Figures 4A,B).