Thirty-one patients with AMI diagnosed by angiography were recruited from The First Affiliated Hospital of University of South China (patients with severe liver and kidney dysfunction, blood system diseases, cancer, and infectious diseases were excluded). In addition, 25 healthy control volunteers were recruited in the same hospital. The blood samples were centrifuged and the serum was collected for later use. All subjects signed a written consent. This study was approved by The First Affiliated Hospital of University of South China.

Human cardiomyocytes (AC16) (Biovector National Typical Culture Center, Beijing, China) were cultured in DMEM medium (Gibco, Grand Island, NY, United States) containing 10% fetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin (Invitrogen, Carlsbad, CA, United States). Under normoxia condition, cells were cultured at 37°C in a normoxia incubator containing 21% O₂ and 5% CO₂. For hypoxia treatment, the cells were grown at 37°C in a hypoxia incubator containing 1% O₂ and 5% CO₂ for 24 h.

Lentivirus short hairpin RNA (shRNA) targteing circ_0124644 (sh-circ_0124644) and its controls (sh-NC), miR-590-3p mimic and inhibitor (miR-590-3p and anti-miR-590-3p) and their controls (miR-NC and anti-NC), pcDNA SRY-box transcription factor 4 (SOX4) overexpression vector (SOX4) and its control (pcDNA) were synthesized from Ribobio (Guangzhou). All oligonucleotides (50 nM) and vectors (4.0 μg) were transfected into AC16 cells with Lipofectamine 3000 (Invitrogen).

RNA was extracted using trizol reagents (Invitrogen), and cDNA was synthesized with Reverse Transcription Kit (Qiagen, Duesseldorf, Germany). qRT-PCR was performed with SYBR Green Kit (Vazyme, Nanjing, China). Relative expression was calculated by 2^{–ΔΔ CT} method with β-actin or U6 as endogenous control. All primer sequences were shown as follows: circ_0124644, F 5′-AAACTGCCTTTTGGATTTGG-3′, R 5′-CTGGCCGGTGAGATACAAGT-3′; ROBO2, F 5′-CGAGC CCACGACTCTGAAC-3′, R 5′-ACACAAACGTAGCTTCCT TCATC-3′; miR-590-3p, F 5′-GCCGAGTAATTTTATGTATAA-3′, R 5′-ATCCAGTGCAGGGTCCGAGG-3′; SOX4, F 5′-AGCG ACAAGATCCCTTTCATTC-3′, R 5′-CGTTGCCGGACTTCAC CTT-3′; β-actin, F 5′-CTCCATCCTGGCCTCGCTGT-3′, R 5′-G CTGTCACCTTCACCGTTCC-3′; U6, F 5′-ATTGGAACGAT ACAGAGAAGATT-3′, R 5′-GGAACGCTTCACGAATTTG-3′.

After RNA was extracted from AC16 cells, the RNA was treated with RNase R (Geneseed, Guangzhou, China) for 30 min. The expression of circ_0124644 and linear ROBO2 was then measured by qRT-PCR. Non-treated RNA was used as control (Mock).

AC16 cells were incubated with Act D solution (Bioss, Beijing, China) for 1 h. Then, the cells were further cultured for indicated times. qRT-PCR was utilized for measuring circ_0124644 and linear ROBO2 expression.

The cytoplasm RNA and nuclear RNA of AC16 cells were isolated using PARIS Kit (Ambion, Austin, TX, United States). After that, qRT-PCR was used to detect circ_0124644 expression in the cytoplasm RNA and nuclear RNA of AC16 cells. U6 and GAPDH expression were used as nuclear control and cytoplasm control.

After transfection and treatment, AC16 cells were collected and then re-seeded into 96 well plates. Then, 48 h later, the cells were incubated with cell counting kit 8 (CCK8) solution (Apexbio, Houston, TX, United States) for 4 h. The absorbance at 450 nm was observed under a microplate reader (Bio-Rad, Hercules, CA, United States).

AC16 cells were harvested and collected into centrifuge tubes. Cell Cycle Detection Kit (KeyGen, Nanjing, China) was used for detecting cell cycle process. Briefly, cells were treated with 70% ethanol, and then incubated with propidium iodide and RNase A. The cell cycle distribution was analyzed by flow cytometer (Beckman Coulter, Miami, FL, United States). Annexin V-FITC Apoptosis Detection Kit (Beyotime, Shanghai, China) was used to measure cell apoptosis. In brief, the cell suspensions were incubated with Annexin V-FITC and propidium iodide. Finally, cell apoptosis rate was detected by flow cytometer (Beckman Coulter).

RIPA buffer (Beyotime) was utilized for extracting total protein from serum and cells. Then, the protein samples were separated by 10% SDS-PAGE gel and transferred to PVDF membranes (Millipore, Billerica, MA, United States). After that, the membranes were incubated with anti-CyclinD1 (1:2,000, bs-20596R, Bioss), anti-Cleaved-caspase3 (anti-Cleaved-casp3, 1:2,000, bs-20364R, Bioss), anti-SOX4 (1:1,000, bs-11208R, Bioss), or anti-β-actin (1:20,000, bs-0061R, Bioss) at 4°C overnight. Goat Anti-rabbit IgG antibody (1:2,000, bs-0295G, Bioss) was used to incubate with the membranes for 2 h. After that, protein signal was detected by BeyoECL Plus reagent (Beyotime).

Lactate dehydrogenase (LDH), malondialdehyde (MDA), superoxide dismutase (SOD), and catalase (CAT) commercial Assay Kits were bought from Nanjing Jiancheng Bioengineering Institute (Nanjing, China). The culture medium of AC16 cells was collected and the supernatants were obtained after centrifuged. According to the kit instructions, the levels of LDH, MDA, SOD, and CAT were examined.

The wild-type (wt) and mutate-type (mut) sequences of circ_0124644 or SOX4 3′UTR were cloned into psiCHECK-2 vector (Promega, Madison, WI, United States) and synthesized by General Biosystems (Anhui, China). In addition, 293T cells (Biovector National Typical Culture Center) were transfected with circ_0124644 wt/mut vector (or SOX4 3′UTR wt/mut vector) and miR-590-3p mimic (or miR-NC). Then, relative luciferase activity was measured by Dual Luciferase Reporter Gene Assay Kit (Beyotime).

Magna RIP Kit (Millipore) was utilized for this assay. AC16 cells were lysed by RIP buffer, and then the cell lysates were incubated with magnetic beads conjugated with human Ago2 antibody or IgG antibody at 4°C overnight. After the RNAs were purified, the circ_0124644 and miR-590-3p expression was analyzed by qRT-PCR.

According to the binding sites of circ_0124644 on miR-590-3p, the wt and mut sequences of miR-590-3p were designed. Biotin-labeled miR-590-3p-wt/mut probes (Bio-miR-590-3p-wt/mut) and control probe (Bio-NC) were synthesized by Sangon Biotech (Shanghai, China). These probes were transfected into AC16 cells for 24 h. After that, cell lysates were collected and then streptavidin-labeled magnetic beads (Pierce, Rockford, IL, United States) were added into cell lysates overnight at 4°C. Finally, circ_0124644 enrichment was detected by qRT-PCR.

All data were expressed as mean ± standard deviation from 3 independent experiments. GraphPad Prism 8.0 (GraphPad, San Diego, CA, United States) was utilized for data analysis. Differences were calculated using Student’s t-test or analyses of variance followed by Tukey post hoc test. The correlations among circ_0124644, miR-590-3p, and SOX4 were adopted using Pearson correlation analysis. P < 0.05 was considered significant difference.

In the serum of AMI patients, circ_0124644 was found to be highly expressed compared with that in the serum of healthy control volunteers (Figure 1A). Under the condition of hypoxia, circ_0124644 expression was significantly increased in AC16 cells (Figure 1B). RNase R assay and Act D assay were used to confirm the circular characteristic of circ_0124644 and the results suggested that circ_0124644 could resist the digestion of RNase R, and its stability was higher than linear ROBO2 mRNA (Figures 1C,D). In addition, subcellular localization analysis showed that circ_0124644 was mainly distributed in the cytoplasm of AC16 cells (Figure 1E). All data confirmed that circ_0124644 was a stable and highly expressed circRNA in AMI patients and hypoxia-induced cardiomyocytes injury models, and it might be mainly involved in post-transcriptional regulation.

To explore the role of circ_0124644 in hypoxia-induced cardiomyocytes injury, the shRNA of circ_0124644 was constructed. After transfected with sh-circ_0124644 into AC16 cells, circ_0124644 expression was markedly reduced, indicating that the transfection efficiency of sh-circ_0124644 was good and could be used for further experiments (Figure 2A). Subsequently, sh-NC or sh-circ_0124644 was transfected into AC16 cells and then treated with hypoxia. By detecting cell viability, cell cycle process, and apoptosis, we discovered that hypoxia could inhibit cell viability, induce cell cycle arrest in G0/G1 phase, and enhance cell apoptosis rate. However, the cell phenotype induced by hypoxia could be reversed by circ_0124644 knockdown (Figures 2B–D). Also, hypoxia reduced the cell cycle marker CyclinD1 protein expression and promoted apoptosis marker Cleaved-casp3 protein expression in AC16 cells, while these effects could be abolished by silencing circ_0124644 (Figure 2E). Additionally, we also evaluated the levels of oxidative stress markers, and the results showed that the promotion effect of hypoxia on LDH and MDA levels and the inhibition effect on SOD and CAT levels in AC16 cells also could be overturned by circ_0124644 downregulation (Figures 2F–I). These results showed that hypoxia indeed induced cardiomyocytes injury, while silenced circ_0124644 could alleviate hypoxia-induced cardiomyocytes injury.

For investigating the new mechanism of circ_0124644 regulated cardiomyocytes injury, the circBank software¹ was used to predict the targeted miRNA of circ_0124644. MiR-590-3p was discovered to have targeted binding sites with circ_0124644 (Figure 3A). Then, miR-590-3p mimic was established and its transfection efficiency was confirmed by detecting its high expression after transfection (Figure 3B). To further confirm the interaction between circ_0124644 and miR-590-3p, dual-luciferase reporter assay, RIP assay, and RNA pull-down assay were performed. The results suggested that the luciferase activity of the circ_0124644 wt vector rather than that of the circ_0124644 mut vector could be reduced by miR-590-3p overexpression (Figure 3C). RIP assay results manifested that both circ_0124644 and miR-590-3p were abundant in the RNA-induced silencing complex immunoprecipitated by Ago2 antibody instead of IgG antibody, implying that circ_0124644 and miR-590-3p co-existed in RNA-induced silencing complex (Figure 3D). Furthermore, the enrichment of circ_0124644 was markedly increased in the Bio-miR-590-3p-wt probe (Figure 3E). In addition, we found that hypoxia could decrease miR-590-3p expression, while this effect could be reversed by silencing circ_0124644 (Figure 3F). Compared to healthy control volunteers, miR-590-3p was remarkably downregulated in the serum of AMI patients (Figure 3G), and its expression also was negatively correlated with circ_0124644 expression in AMI patients (Figure 3H). These data confirmed that miR-590-3p could be sponged by circ_0124644.

To confirm the function of miR-590-3p on hypoxia-induced cardiomyocytes injury, we transfected with miR-590-3p mimic into hypoxia-induced AC16 cells. Function experiments showed that overexpressed miR-590-3p could reverse the inhibitory effect of hypoxia on cell viability and cell cycle, as well as the promoting effect on cell apoptosis (Figures 4A–C). Moreover, the decreased CyclinD1 protein expression and the increased Cleaved-casp3 protein level in AC16 cells under hypoxia treatment also could be reversed by miR-590-3p overexpression (Figure 4D). Furthermore, the addition of miR-590-3p mimic also abolished the enhancing effect of hypoxia on LDH and MDA levels and the reducing effect on SOD and CAT levels in AC16 cells (Figures 4E–H). Our data showed that miR-590-3p could relieve hypoxia-induced cardiomyocytes injury.

In order to determine that circ_0124644 regulated cardiomyocytes injury by sponging miR-590-3p, the miR-590-3p inhibitor was constructed. After transfection with anti-miR-590-3p into AC16 cells, the expression of miR-590-3p was markedly reduced, indicating that the transfection of anti-miR-590-3p was successful (Figure 5A). In hypoxia-induced AC16 cells transfected with sh-circ_0124644 and anti-miR-590-3p, we found that the promotion effect of circ_0124644 silencing on cell viability and cell cycle as well as the suppressive effect on cell apoptosis rate in hypoxia-induced AC16 cells could be reversed by miR-590-3p inhibitor (Figures 5B–D). Also, miR-590-3p inhibitor could abolish the positive regulation of circ_0124644 knockdown on CyclinD1 protein expression and the negative regulation on Cleaved-casp3 protein expression in hypoxia-induced AC16 cells (Figure 5E). In addition, the inhibiting effect of circ_0124644 silencing on LDH and MDA levels and the enhancing effect on SOD and CAT levels in hypoxia-induced AC16 cells also could be reversed by miR-590-3p inhibitor (Figures 5F–I). Therefore, we confirmed that circ_0124644 sponged miR-590-3p to regulate hypoxia-induced cardiomyocytes injury.

The downstream target of miR-590-3p was predicted using targetscan software². It was found that the 3′UTR of SOX4 could interact with miR-590-3p (Figure 6A). Then, the results of dual-luciferase reporter assay showed that miR-590-3p mimic could inhibit the luciferase activity of SOX4 3′UTR wt vector, which had no effect on that of the SOX4 3′UTR mut vector (Figure 6B). Moreover, hypoxia could promote SOX4 expression, while this effect could be abolished by miR-590-3p overexpression (Figure 6C). Also, circ_0124644 silencing could reduce SOX4 protein expression in hypoxia-induced AC16 cells, and the addition of anti-miR-590-3p also could reverse this effect (Figure 6D). In addition, SOX4 protein and mRNA expression levels were found to be markedly enhanced in the serum of AMI patients (Figures 6E,F). Correlation analysis showed that SOX4 mRNA expression also was negatively correlated with miR-590-3p expression and positively correlated with circ_0124644 expression in AMI patients (Figures 6G,H). All data suggested that circ_0124644 sponged miR-590-3p to regulate SOX4.

Subsequently, pcDNA SOX4 overexpression vector was constructed. After transfection with pcDNA SOX4 overexpression vector into AC16 cells, SOX4 protein expression was significantly increased (Figure 7A). Then, miR-590-3p mimic and pcDNA SOX4 overexpression vector were co-transfected into AC16 cells, and then the cells were treated with hypoxia. Analysis results showed that ectopic SOX4 expression could reverse the enhancing effect of miR-590-3p overexpression on cell viability and cell cycle and the decreasing effect on cell apoptosis in hypoxia-induced AC16 cells (Figures 7B–D). Meanwhile, the promotion of miR-590-3p on CyclinD1 protein expression and the repression on Cleaved-casp3 protein expression also could be reversed by SOX4 upregulation in hypoxia-induced AC16 cells (Figure 7E). At the same time, overexpressed SOX4 also could abolish the miR-590-3p inhibiting LDH and MDA levels and miR-590-3p accelerating SOD and CAT levels in hypoxia-induced AC16 cells (Figures 7F–I). Hence, our results illuminated that miR-590-3p targeted SOX4 to regulate hypoxia-induced cardiomyocytes injury.