Circular RNA RBM33 contributes to extracellular matrix degradation via miR-4268/EPHB2 axis in abdominal aortic aneurysm

Tissue samples

A total of 23 AAA tissues were collected from the AAA patients (age 48–65, mean age 56.3, 12 male, 11 female) receiving surgical procedures at the Second Affiliated Hospital of Nanchang University. The inclusion criteria were as follows: (a) with confirmed AAA diagnosis; (b) no radiotherapy and chemotherapy prior to surgery; (c) available clinical information. Patients would be excluded if suffered from other tumors and serious diseases and their clinical data were incomplete. Normal abdominal aortic tissues were obtained from age- and gender-matched organ donors without aortic diseases. All the patients were informed and signed the informed consent. This study was approved by the Ethics Committee of the Second Affiliated Hospital of Nanchang University (Ethical Application Ref: Study Clinical Review (2020) No. 044). Tissue samples were sliced into small segments in liquid nitrogen, and further stored at −80 °C.

RNA extraction, library construction and sequencing

Total RNA was extracted from AAA tissues (N = 3) and normal abdominal aortic tissues (N = 3) with a Trizol RNA extraction reagent (Sigma–Aldrich, Shanghai, China). The quantity and purity of RNA were detected by the NanoDrop spectrophotometer (Thermo Fisher Scientific Inc., Shanghai, China). The RNA was electrophoresed on 1.2% agarose gel to analyze the integrity. For RNA-seq library construction, 1 μg of total RNA was used. The RNA libraries were constructed using the Total RNA-seq (HMR) Library Prep Kit (Azyme Biotech Co., Ltd., Nanjing, China). In brief, ribosomal RNA was removed with rRNA probes and RNase H, RNA was fragmented to ∼300 bp by metal ion, reverse transcribed into cDNA, and connected with adapters. The fragments were sorted by DNA Clean Beads (Beckman, USA), and subjected to PCR amplification and purification and libraries were validated using Agilent 2200 (Agilent, Santa Clara, CA, USA). Illumina platform (Illumina, Inc., CA, USA) was used for RNA sequencing.

RT-qPCR

Total RNA from tissues (N = 3) and cells (N = 3) were extracted using TRIzol reagent (Invitrogen, CA, USA) according to the manufacture’ introduction. After the detection of RNA quality, purity, and content, 1 μg of RNA was reverse-transcribed into cDNA with a PrimeScript™ RT reagent kit (K1622; Thermo Fisher Scientific Inc., Shanghai, China). The PCR was performed with the SYBR Green PCR kit (Roche, Basel, Switzerland) on an ABI Q6 Real-time PCR system (Applied Biosystems Inc., Foster City, CA, USA). The reaction conditions were: 10 min of pre-denaturation at 95 °C, 45 cycles of denaturation at 95 °C for 15 s, and annealing at 60 °C for 60 s. The internal references in RT-qPCR were GAPDH and U6. Three independent experiments were conducted. The relative expression of each factor was evaluated using the 2^−ΔΔCt method. The RT primer and special PCR primers were listed in Table S1.

Western Blot

Protein was extracted from cells with ice-cold RIPA buffer (Cell Signaling Technology, Danvers, MA, USA) containing a cocktail of protease inhibitors. Protein concentration was measured by BCA assay kit (Thermo Fisher Scientific Inc., Shanghai, China). Protein samples (50 µg) were separated in 10% SDS-PAGE gels and transferred to polyvinylidene difluoride (PVDF) membranes (Sigma-Aldrich, Shanghai, China). Membranes were blocked in 5% skim milk powder for 2 h at room temperature and incubated overnight at 4 °C with primary antibodies. The primary antibodies applied are as follows: rabbit anti-human tissue inhibitor of matrix metalloproteinases-1 (TIMP-1; ab109125, 1:1000, abcam, Shanghai, China), rabbit anti-human EPH receptor B2 (EPHB2; ab216629, 1:2000, abcam, Shanghai, China), and rabbit anti-human GAPDH (ab9485, 1:2000, abcam, Shanghai, China). The membranes were then washed three times with TBST for 10 min each and incubated with goat anti-rabbit IgG-HRP antibody (ab205718, 1:20,000, abcam, Shanghai, China). After three washes with TBST, membranes were subjected to ECL chemiluminescence reagent (Thermo Fisher Scientific Inc., Shanghai, China). Images were obtained using a ChemiDoc Gel Imaging system (Thermo Fisher Scientific Inc., Shanghai, China), and the optical density was quantified with Quantity One image analysis software (Bio-rad, Hercules, CA, USA). Experiments were performed in triplicate.

Luciferase reporter assay

The circRBM33 wild-type (WT, containing miR-4268 binding sites) and mutant 3′-UTR fragments (Mut, mutant miR-4268 binding sites) were cloned into a psicheck-2 vector. The 293T cells were seeded into 48-well plates and cultured for 24 h. Then, the reporter gene vectors were co-transfected with miR-4268 mimics or NC mimics into cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). After 48 h transfection, the luciferase activity in each group was examined by a Dual-Luciferase Assay Kit (Promega, Madison, WI, USA). The experiment was repeated thrice.

Statistical analysis

All of these experiments were repeated three times. Statistical analyses were assessed with the SPSS v.21.0 software (IBM Corp, Armonk, NY, USA). Data were expressed as mean ± SD. The comparisons of means among the two groups were evaluated by Student’s t-test. One-way ANOVAs were performed for multiple comparisons, and the Tukey-HSD post hoc test was used. For all tests, p < 0.05 was considered to be statistically significant.

Analysis of DEcircRNAs

To investigate the expression profiles of circRNAs implicated in AAA, RNA sequencing was performed on samples from AAA tissues and normal abdominal aortic tissues. After filtering low-quality and duplicated reads (filtration rate > 89%), approximately 9.1–11.2 million of clean reads were obtained in each sample (Table S3). The clean reads were then aligned onto the human (GRCh38) reference genome, the alignment rates of clean reads ranged from 89% to 96%, indicating that the sequencing data was reliable. Approximately 0.9–1.8 million junction reads were selected for circRNA prediction and quantification, and 5621 (AAA-1), 7190 (AAA-2), 5547 (AAA-3), 7402 (Con-1), 8233 (Con-2), 8480 (Con-3) were generated in six tissue samples, respectively. After filtering of lowly-expressed circRNAs, 1,081, and 1,372 were respectively obtained in the AAA group and control group, and 200 circRNAs were co-expressed in these two groups (Fig. S1).

To identify circRNA expression signatures in AAA, the DEcircRNAs were identified by the selection criteria of FDR <0.05, log2 fold change >1 or <−1. In total, 65 significant DEcircRNAs were recognized, including 30 up-regulated DEcircRNAs and 35 down-regulated DEcircRNAs in the AAA tissues compared with the normal control group (Fig. 1A). The six samples were clustered closely into two groups, namely the AAA group and the control group, indicating a difference of circRNA expression between the AAA group and the control group (Fig. 1B).

Functional and pathway analysis of DEcircRNAs

To explore wthether the DEcircRNAs were associated with ECM degradation, the GO and KEGG analysis was performed. The target genes modulated by DEcircRNAs were enriched in biological process (BP) of “cell adhesion”, “inflammatory response”, “ECM organization”, “ECM disassembly”, “regulation of ECM disassembly”, and “regulation of ECM assembly”; the top terms in cellular component (CC) were “ECM” “cell junction”, “proteinaceous ECM”, and “extracellular region”; these DEcircRNAs were found to be involved in molecular function (MF) of “receptor activity”, “cell adhesion molecule binding”, “transmembrane signaling receptor activity”, and “ECM structural constituent”(Fig. 2A). KEGG enrichment analysis for DEcircRNAs were assigned into ECM-related pathways, including TNF signaling pathway (Mei et al., 2019), NF-κB signaling pathway (Guo et al., 2021), ECM-receptor interaction (Lu et al., 2011), and PI3K-Akt signaling pathway (Gao et al., 2020) (Fig. 2B).

circRBM33 is highly expressed in AAA

To seek for key circRNAs in the progression of AAA, five circRNAs (circHLA-DRB6, circCFLAR, circEPSTII, circRBM33, circHLA-IGLJ3) with higher expressed abundance, more fold change and higher significance were selected for RT-qPCR examination in six sequencing samples (Fig. 3A). Obviously, the expression of circHLA-DRB6, circCFLAR and circRBM33 had statistical differences between the AAA samples and normal samples. Two studies have showed that circRBM33 (chr7_155680908_155672867_+8041-RBM33, generated from human gene RBM33, termed circRBM33) is the pro-oncogenic factor in cervical cancer and gastric cancer (Wang et al., 2020; Ding, Yuan & Gu, 2021). Moreover, the expression of circRBM33 in AAA tissues was significantly up-regulated compared with that in the normal abdominal aortic tissues (Fig. 3B). Divergent and convergent primers were designed to amplify circRBM33 in gDNA and cDNA of AAA tissues, and found only the divergent primers amplified circRBM33 from cDNA, whereas, no amplification product was visualized from gDNA (Fig. 3C). The PCR products were purified and subjected to Sanger sequencing to detect the back-splicing exon junction regions of circRBM33. The sequencing data showed that circRBM33 was generated from exon 3 and exon 5 of RBM33 gene by back-splicing (Fig. 3D).

Overexpression of circRBM33 induces the ECM degradation

Subsequently, we evaluated the roles of circRBM33 in AAA progression. AAA model was mimicked in human VSMCs via inducing by Ang II to measure the expression. Compared with the blank control group, the circRBM33 expression in Ang II-induced group was significantly increased (Fig. 4A). Overexpression of circRBM33 in VSMCs was mediated by circRBM33 overexpression vector. RT-qPCR analysis found that transfection with circRBM33 overexpression vector resulted in a significant up-regulation of circRBM33 in VSMCs (P < 0.01, Fig. 4B). Numerous enzymes have been reported to be involved in regulating ECM degradation, including MMP-2 and TIMP-1 (Yuan & Wu, 2018; Guo et al., 2019). MMP-2 is well documented to promote the degradation of ECM (Miyakawa et al., 2019), and TIMP-1 attenuates the ECM degradation by inhibiting nearby MMPs (Nordgaard et al., 2019). Thus, we determined MMP-2 and TIMP-1 expression in the circRBM33 overexpressing VSMCs by western blot to evaluate the effect of circRBM33 on ECM degradation. As expected, the level of MMP-2 in VSMCs with circRBM33 overexpression was significantly increased relative to the control cells, and the TIMP-1 expression was reduced in VSMCs with circRBM33 overexpression (Fig. 4C).

circRBM33 mediates the ECM degradation by directly binding to miR-4268

To explore the mechanism of circRBM33 in ECM degradation and AAA progression, we selected the potential targets of circRBM33 and created a circRNA-miRNA-mRNA interaction network (Fig. 5A). The miR-4268, which targeted several oncogenic genes (L1TD1, LYN, SOCS3, SRCIN1, AQP3, EPHB2), was detected in the VSMCs. We found that the overexpression of circRBM33 caused a significant reduction in the miR-4268 expression (Fig. 5B). Furthermore, the binding sites of miR-4268 and circRBM33 were predicted by bioinformatics analysis, and the binding sites was exhibited in Fig. 5C. For luciferase reporter gene assay, the mutant luciferase reporters and the WT luciferase reporters were constructed. Luciferase activity assay showed that miR-4628 mimic significantly decreased the luciferase activity of the WT group but not the Mut group (Fig. 5D), suggesting the direct binding of circRBM33 with miR-4628. The miR-4268 mimics was transfected into VSMCs, and induced the miR-4268 expression in VSCMs as compared to the control cells (Fig. 5E). Moreover, the decreasing TIMP-1 expression in the VSMCs caused by circRBM33 overexpression was rescued by miR-4268 mimics (Fig. 5F), indicating that circRBM33 mediates the ECM degradation by directly binding to miR-4268 in AAA.

circRBM33 regulates EPHB2 expression

Previous studies on circRNAs have indicated that circRNAs can regulate genes by acting as miRNA sponges. Here, we screened two predicted target genes (SOCS3, EPHB2) of miR-4268 for further study. In addition, the expression of EPHB2 was significantly decreased, and that of SOCS3 had no significant change in VSMCs with miR-4268 overexpression (Fig. 6A). To verify whether circRBM33 exerts its role by sponging miR-4268, we carried out the rescue experiment. As shown in Fig. 6B and 6C, the mRNA and protein levels of EPHB2 were increased in the circRBM33 overexpressing VSMCs, and the increase was partly reversed by miR-4268 mimics.