CircRNA_103765 acts as a proinflammatory factor via sponging miR-30 family in Crohn’s disease

Increasing evidence suggests that circular RNAs (circRNAs) play critical roles in various pathophysiological activities. However, the role of circRNAs in inflammatory bowel disease (IBD) remains unclear. Here we report the potential roles of hsa_circRNA_103765 in regulating cell apoptosis induced by TNF-α in Crohn’s disease (CD). We identify that CircRNA_103765 expression was significantly upregulated in peripheral blood mononuclear cells (PBMCs) of patients with active IBD. A positive correlation with TNF-α significantly enhanced circRNA_103765 expression in CD, which was significantly reversed by anti-TNF-α mAb (infliximab) treatment. In vitro experiments showed that TNF-α could induce the expression of circRNA_103765, which was cell apoptosis dependent, while silencing of circRNA_103765 could protect human intestinal epithelial cells (IECs) from TNF-α-induced apoptosis. In addition, circRNA_103765 acted as a molecular sponge to adsorb the miR-30 family and impair the negative regulation of Delta-like ligand 4 (DLL4). Collectively, CircRNA_103765 is a novel important regulator of the pathogenesis of IBD via sponging miR-30 family-mediated DLL4 expression changes. Blockade of circRNA_103765 could serve as a novel approach for the treatment of IBD patients.

Inflammatory bowel disease (IBD), which mainly includes ulcerative colitis (UC) and Crohn's disease (CD), is a multifactorial autoimmune disease that is characterized by chronic and recurrent digestive tract inflammatory disorder 1 . In mainland China, with the development of the economy, the locals 's life styles are increasingly westernized, such as higher sugar diet and lower fiber diet, intake of more fast food, which bring about rising incidence and prevalence of IBD 2 . Although the etiology of IBD remains unclear, growing evidence has shown that the pathogenesis of CD is a complex interplay of immunology, genetic predisposition, gut microbiota content and environmental risk factors 1,3,4 .
Traditionally, proinflammatory cytokines produced by immune cells are considered to be crucial in the pathogenesis of IBD. They not only recruit the immune cells and perpetuate inflammation but also have a direct impact on intestinal epithelial cells (IECs), leading to disruption of tight junctions [5][6][7] and induction of cell apoptosis 8,9 . Among the various cytokines, the significant role of TNF-α has been illustrated by the dramatic improvement in the management of moderate to severe IBD since the development of anti-TNF-α agents such as infliximab (IFX) 10 .
Circular RNAs (circRNAs) are a newly discovered class of endogenous noncoding RNAs that play important roles in the development and progression of a variety of cancers 11 , autoimmune disease 12 , and nervous system disorders 13 , among others. Owing to their distinct circular covalently closed structure, circRNAs are evolutionally conserved and appear highly stable in the cytoplasm. Recent study has proposed a theory called the competing endogenous RNA (ceRNA) hypothesis that lncRNAs and mRNAs could communicate with and modulate each other through competitively sharing miRNA response elements (MREs), which provides a novel mechanism of gene regulation 14 . It has been shown that circRNAs could also act as ceRNAs to sequester away miRNAs from the target genes. For example, Hansen TB et al. first identified that ciRS-7 may serve as a critical factor in the function of neurons by sponging miR-7 15 . CircHIPK3 is considered a prognostic biomarker in colorectal cancer (CRC) by targeting the c-Myb/circHIPK3/miR-7 axis. However, as the roles of most circRNAs in the initiation and progression of IBD remain unclear and further research is still needed.
The Notch signaling pathway plays a critical role in the development of digestive tract diseases. The Notch pathway contains membrane-bound ligands (Delta-like ligand [DLL]1, DLL3, DLL4, Jagged1, or Jagged2) and receptors (Notch1-4). Direct cell-to-cell interactions by binding a ligand to a receptor causes the release of Notch intracellular domain (NICD) for transcriptional induction of Notch target genes, such as HES1 16 . Moreover, previous studies have shown that DLL4 is the direct binding target of microRNA-30 (miR-30) family members [17][18][19] . The miR-30 family is a significant member of miRNA family, which is composed of five members (miR-30a, miR-30b, miR-30c, miR-30d, and miR-30e). Recent studies have shown that miR-30 family controls intestinal epithelial cell proliferation and differentiation by targeting a broad gene expression program 20 . MiR-30a-5p ameliorates inflammatory responses through MAPK/ERK signaling 21 . MiR-30e contributed to abnormal small intestinal epithelial differentiation by negatively regulating the Dll4/NICD/Hes1 signaling pathway 19 .
We analyzed the expression profile of circRNAs by microarray analysis in peripheral blood mononuclear cells (PBMCs) from CD patients versus healthy controls (HCs) and identified 384 significantly dysregulated circRNAs 22 . In this study, we investigated the expression of circRNA_103765 in the PBMCs of IBD patients by quantitative polymerase chain reaction (qPCR). Then, the functions and mechanisms of circRNA_103765 were deeply explored in vitro. We found that circRNA_103765 was significantly increased in IBD patients. Moreover, enhanced circRNA_103765 expression was distinctly reversed after anti-TNF-α mAb treatment. Knockdown of circRNA_103765 could inhibit cell apoptosis, promote proliferation and downregulate the expression of Deltalike ligand 4 (DLL4). In addition, we performed a luciferase reporter assay and confocal visualization of dual fluorophore RNA-fluorescence in situ hybridization (FISH) and observed that circRNA_103765 could sponge miR-30 family members, including miR-30a-5p and miR-30e-5p. Overall, we determined that circRNA_103765 may regulate DLL4 expression by binding to the miR-30 family. Collectively, these findings indicate that cir-cRNA_103765 might exert regulatory functions in IBD and could be a potential target for therapeutic intervention in IBD therapy.

Results
CircRNA_103765 is highly increased in PBMCs of patients with IBD. Our previous study based on a microarray analysis identified 384 dysregulated circRNAs in PBMCs from CD patients compared with HCs. Herein, we focused on circRNA_103765, which is located at chr4:166141085-166184511 and spliced from Kelch-like2 (KLHL2). The kelch-like protein was identified as the genes responsible for inherited diseases, such as a hereditary hypertensive disease-pseudo hypoaldosteronism type II and autosomal dominant retinitis pigmentosa 23,24 . We determined that the expression of circRNA_103765 is upregulated in patients with IBD compared with healthy donors (P < 0.05). However, there was no significant difference between the UC and CD groups (Fig. 1A). Moreover, circRNA_103765 expression levels in active CD and UC were distinctly higher than those in remission CD and UC (Fig. 1B).
ROC analysis revealed that circRNA_103765 could differentiate the CD and UC groups from the HCs group (Fig. 1C,D, Supplementary Table S1). The correlations between circRNA_103765 levels and laboratory test results or disease activity of IBD patients are listed in Fig. 1E,F and Supplementary Fig. S1, Supplementary Table S2. We observed a positive correlation of circRNA_103765 with TNF α (CD: R 2 = 0.5122, P < 0.001 and UC: R 2 = 0.4694, P < 0.001).

TNF-α upregulates circRNA_103765 expression in PBMCs of CD patients. To investigate whether
the increased TNF-α in PBMCs contributes to the upregulation of circRNA_103765 in IBD patients, we detected circRNA_103765 expression in PBMCs from CD patients prior to and after treatment with IFX. To this end, 21 patients (46.7%) achieved clinical remission (CDAI < 150), 13 patients (28.8%) did not achieve clinical remission, but they achieved a standard clinical response with a drop in CDAI of ≥ 70 but still with a CDAI ≥ 150. The other 11 participants (24.4%) were unfortunately classified as a failure of IFX due to a CDAI ≥ 150 and a drop in CDAI of ≤ 70 or a changed or increased CDAI from the baseline. Interestingly, we found that circRNA_103765 was considerable decreased in CD patients from the clinical remission and response groups after IFX therapy in comparison with that before IFX induction therapy ( Fig. 2A-C). In addition, no change in the expression of circRNA_103765 was observed in the failure group (Fig. 2D).
Moreover, we also observed that the TNF-α level was greatly positively correlated with circRNA_103765 expression in the remission group after IFX therapy in CD patients, as well as before IFX induction therapy ( Fig. 2E-G). Nevertheless, no important correlation between circRNA_103765 and TNF-α was found in the response or failure group (Fig. 2H,I). Collectively, these data showed that TNF-α may be able to upregulate circRNA_103765 expression in CD patients and that IFX therapy could markedly reverse the expression of circRNA_103765.

TNF-α indirectly induces circRNA_103765 transcription by promoting cell apoptosis in a human intestinal epithelial cell line.
To verify the above findings in vitro, we then detected the time course changes in circRNA_103765 expression induced by TNF-α in Caco2 cells and HIECs. As shown in Fig. 3A, TNF-α stimulated circRNA_103765 expression in a dose-and time-dependent manner. The expression level of circRNA_103765 was considerable upregulated in Caco2 and HIECs compared with that in the controls after exposure to TNF-α (P < 0.001). Flow cytometry-based Annexin V/PI staining showed that TNF-α significantly increased the apoptotic rate compared with the negative control (P < 0.05, Fig. 3B,C). Furthermore, TNF-α increased the levels of the proapoptotic proteins Bax and cleaved caspase-3 and decreased the expression of Bcl-2 ( Fig. 3D), which is in line with a previous study 25,26 . Collectively, the data suggest that TNF-α induces circRNA_103765 following IEC apoptosis. www.nature.com/scientificreports/ To further determine whether the TNF-α-upregulated circRNA_103765 expression was due to the induction of apoptosis in intestinal epithelial cells, the effect of z-VAD-FMK (a broad-spectrum caspase inhibitor) was determined. Caco2 cells and HIECs were pretreated for 2 h with z-VAD-FMK and then induced with TNF-α. The results showed that z-VAD-FMK induced suppression of circRNA_103765 expression in a concentration-and time-dependent manner (Fig. 3E). The addition of z-VAD-FMK at 25 μM after 24 h had little effect on the expression level of circRNA_103765. However, at 100 μM z-VAD-FMK, the circRNA_103765 level was significantly decreased (P < 0.001).The carrier solvent DMSO (< 0.1%) had no effect on circRNA_103765 expression (results not shown). Z-VAD-FMK inhibited the cell apoptotic rate, downregulated Bax and cleaved caspase-3 protein, and increased the expression of Bcl-2, which was induced by TNF-α ( Fig. 3B-D).
Taken together, these data suggest that TNF-α-induced circRNA_103765 expression is cell apoptosis dependent.

Knockdown of circRNA_103765 protects human intestinal epithelial cells from TNF-α-induced apoptosis in vitro.
To explore the biological functions of circRNA_103765 in CD, we synthesized a special siRNA (si-circRNA_103765) targeting the back-splice junction site of circRNA_103765 to knock down cir-cRNA_103765 in a TNF-α-treated Caco2 and HIEC. As illustrated in Fig. 4, the siRNA greatly decreased the expression level of circRNA_103765 (P < 0.05, Fig. 4A). Knockdown of circRNA_103765 greatly promoted cell proliferation in both Caco2 cells and HIECs (P < 0.05, Fig. 4B). Furthermore, si-circRNA_103765 substantial reduced the apoptotic rate compared with that in TNF-α-induced cells (P < 0.05, Fig. 4C,D). Western blot experiments further confirmed this result (Fig. 4E). The above in vitro experiments collectively confirmed that knock-   www.nature.com/scientificreports/ Hsa_circRNA_103765 functions as a sponge for the miR-30 family. To elucidate the molecular mechanism of circRNA_103765, we first predicted the targets of circRNA_103765 according to the TargetScan and miRanda databases. The results showed that circRNA_103765 contains a conserved target site of the miR-30 family (miR-30a-5p, miR-30b-5p, miR-30d-5p, miR-30e-5p) ( Supplementary Fig. S2). Thus, we investigated the corresponding miR-30 expression in PBMCs from CD patients prior to and after treatment with IFX. Interestingly, the expression level of circRNA_103765 was negatively correlated with miR-30a-5p and miR-30e-5p expression (P < 0.001) (Fig. 5A,B). However, there was no significant correlation between circRNA_103765 and miR-30b-5p or miR-30d-5p ( Supplementary Fig. S3). On this basis, a dual luciferase reporter assay, was conducted to validate the sponging relationship between circRNA_103765 and miR-30a-5p and 30e-5p in 293T cells. The results demonstrated that miR-30a-5p and miR-30e-5p mimics could markedly decrease the luciferase activity of the circRNA_103765 WT group but not the mutant group (P < 0.05 for miR-30a-5p; P < 0.001 for miR-30e-5p, Fig. 5C,D). We performed a FISH assay to observe the subcellular localization of circRNA_103765, miR-30a-5p, and miR-30e-5p. It was found that most circRNA_103765 (red) and miR-30a-5p/30e-5p (green) were mainly colocalized in the cytoplasm (Fig. 5E).
Together, these data demonstrated that circRNA_103765 could act as a sponge of miR-30a and miR-30e in a human intestinal epithelial cell line.
In addition, western blot and qRT-PCR demonstrated that the TNF-α-induced upregulation of cir-cRNA_103765 enhanced the protein and mRNA levels of DLL4 and the downstream proteins NICD and Hes1, while knockdown of circRNA_103765 decreased the levels of these proteins (Fig. 6G,H), and the effects caused by upregulation or silencing of circRNA_103765 could be reversed by miR-30a-5p and miR-30e-5p mimics or inhibitors, respectively. Interestingly, co-expression of miR-30a-5p and miR-30e-5p mimics or inhibitors at an equivalent total concentration did not further increase or decrease DLL4 expression, suggesting that there were no synergistic or additive effects between miR-30a-5p and miR-30e-5p (Fig. 6I-L).
In summary, these data demonstrated that circRNA_103765 might serve as a ceRNA for miR-30a-5p/miR-30e-5p to regulate the expression level of DLL4 (Fig. 6M).

Discussion
In recent years, a larger number of circRNAs have been discovered in multiple disease and cell lines by high-density microarrays and next-generation sequencing technology. Due to their disease-specific expression, molecular stability and conservatism, circRNAs can regulate various biological processes to act as miRNA sponges, and they are considered ideal diagnostic or therapeutic candidates. To the best of our knowledge, the biological function of most circRNAs remains largely unclear in IBD studies.
Here, our previous research analyzed the expression profile of circRNAs by microarray analysis in CD patients versus healthy controls (HCs) and identified 384 significantly dysregulated circRNAs 22 . Subsequently, we identified a novel circRNA termed circRNA_103765 that was obviously upregulated in active IBD patients and significantly correlated with the level of TNF α in CD. To further investigate whether increased TNF-α contributes to the upregulation of circRNA_103765 in IBD patients, we detected circRNA_103765 expression in CD patients prior to and after treatment with IFX. Interestingly, we determined that TNF-α could upregulate circRNA_103765 expression in CD patients. Moreover, our data also show that effective anti-TNF-α treatment resulted in a rapid reduction in circRNA_103765 expression, which may be able to 'predict' whether treatment with IFX is effective in IBD patients. Furthermore, in vitro cytology experiments also confirmed this result. Moreover, our research further showed that TNF-α-induced circRNA_103765 expression was cell apoptosis dependent, while knockdown of circRNA_103765 could protect human intestinal epithelial cells from TNF-α-induced apoptosis.
Growing evidence has demonstrated that some circRNAs can serve as functional miRNA sponges to regulate the expression of miRNA target genes in various human diseases. In our study, we found that circRNA_103765 contained the MRE of the miR-30 family through bioinformatics analyses. We also found that the expression level of circRNA_103765 was negatively correlated with miR-30a-5p and miR-30e-5p in CD patients prior to and after treatment with IFX. Furthermore, a dual-luciferase reporter assay confirmed that circRNA_103765 could interact with miR-30a-5p and miR-30e-5p directly. FISH assays showed that circRNA_103765 and miR-30a-5p/miR-30e-5p were colocalized in the cytoplasm. Herein, circRNA_103765 could act as a sponge of miR-30a and miR-30e in IBD. Moreover, previous studies have confirmed that DLL4 is the direct binding target of miR-30 family members [17][18][19] . Notch Delta-like ligand 4 (DLL4) is a well-known transmembrane protein that can activate Notch receptors on adjacent cells. Furthermore, Notch signaling is a key signaling pathway that is involved in cell-cell communication and development 27 . It regulates a variety of cellular processes, such as differentiation, proliferation, apoptosis, and inflammation in IECs [28][29][30] . In this study, to validate the crosstalk between circRNA_103765 and DLL4, we revealed that the TNF-α-induced upregulation of circRNA_103765 enhanced the protein and mRNA levels of DLL4 and the expression of the downstream proteins NICD and www.nature.com/scientificreports/ Hes1, while knockdown of circRNA_103765 decreased the levels of these proteins. Moreover, the effects caused by upregulation or silencing of circRNA_103765 could be reversed by miR-30a-5p and miR-30e-5p mimics or inhibitors, respectively, which supports our hypothesis that circRNA_103765 functions as a ceRNA to promote DLL4-mediated cell apoptosis by decoying the miR-30 family in CD. Significant crosstalk between TNF-α and Notch signaling has been observed. For example, TNF-α activates Notch signal-related proteins in rheumatoid arthritis 31 , but TNF-α suppresses the Notch signaling pathway in skeletal myogenesis 32 . Moreover, Notch ligand Delta-1 reduces TNF-α-induced apoptosis by decreasing the activation of caspases 33 . It seems that there would be certain reciprocal regulation between the two signals, which is cell-and context-dependent. Our newly found evidence indicates that the upregulation of circRNA_103765 associated with TNF-α is under the control of the Notch signaling pathway by sponging the miR-30 family, which is a proinflammatory and proapoptotic factor in IECs. However, how DLL4/Notch signaling interferes with TNF-α signaling during the apoptosis process in IBD requires further study.
In summary, our study indicates that circRNA_103765 is a novel important regulator of the pathogenesis of IBD by regulating the apoptosis of IECs via miR-30 family-mediated DLL4 expression changes. CircRNA_103765 knockdown was able to ameliorate cell apoptosis in vitro, further confirming the above findings. Our study provides new insight into the pathogenic role of circRNA_103765 in IBD and suggests that targeted therapy directly against circRNA_103765 could be a novel approach for the treatment of IBD patients.
Because our experiment is only based on clinical samples and cell lines, further in vivo studies are needed to evaluate its utility. Herein, the newly discovered link between circRNA_103765 and IBD will hopefully provide a better understanding of the disease, with potential for improved diagnosis or therapy in the future.

Methods
Patients. IBD patients (60 with CD and 60 with UC) were prospectively enrolled at the Department of Gastroenterology of the First Affiliated Hospital of Soochow University and the North District of the Affiliated Suzhou Hospital of Nanjing Medical University (Jiangsu, China). Forty healthy controls were included. The demographic and clinical information are listed in Table 1. The diagnosis of IBD was based on clinical manifestations, radiological findings, endoscopic examination and histological findings. The severity of diseases was scored according to the CD Activity Index (CDAI) for the diagnosis of patients with CD and the Mayo score for UC. Active IBD patients were defined as CDAI ≥ 150 or Mayo score > 2. Table 1. Clinical characteristics of IBD patients and controls. The data are presented as the mean ± SD or medians (1/4-3/4 quarters). Patients with isolated upper disease (L4) were excluded. L1 terminal ileum, L2 colon, L3 ileocolon, B1 non-stricture and non-penetrating, B2 stricturing, B3 penetrating, E1 rectum, E2 left side, E3 extensive, CD Crohn's disease, UC ulcerative colitis, HCs healthy controls, CRP C-reactive protein, TNF-α tumor necrosis factor α, CDAI CD Activity Index, A/R active/remission. www.nature.com/scientificreports/ PBMCs were promptly isolated after blood sample collection from all subjects according to the manufacturer's protocol (GE Healthcare) and then frozen at − 80 °C.

Characteristic CD (A/R) UC (A/R) HCs
Anti-TNF-α mAb treatment of patients with CD. Forty-five active CD patients were treated with anti-TNF-α monoclonal antibody (infliximab (IFX), 5 mg/kg; Switzerland) at 0, 2 and 6 weeks 34 (Table 2). Clinical response to IFX was assessed at week 12 after initial infusion. Clinical remission was defined as CDAI < 150, and clinical response was defined as a decrease of 70 or more points in the CDAI score compared with the baseline index. The failure of IFX therapy included patients whose CDAI was not greatly changed or increased 35,36 . Cell culture and inflammatory model. Human intestinal epithelial cell lines (Caco-2 and HIEC) were purchased from Shanghai Suer Biological Technology Co., Ltd. (Shanghai). The human embryonic kidney 293T (HEK293T) cell line was donated by Dr Chen at the First Affiliated Hospital of Soochow University (Suzhou). The Caco-2 and 293T cell lines were cultured in high-glucose Dulbecco's modified essential medium (HyClone) containing 10% fetal bovine serum (Gibco) and 1% penicillin and streptomycin (Gibco) in a 5% CO2 atmosphere at 37 °C. HIECs were cultured in RPMI 1640 medium (HyClone) according to the recommended protocols.
TNF-α (MedChemExpress) was added to the medium to mimic an inflammatory background as previously described 37 .

Ethics statement.
All the studies were carried out in accordance with the approved guidelines. The study was approved by the Ethics Committees of the First Affiliated Hospital of Soochow University and the Affiliated Suzhou Hospital of Nanjing Medical University. The informed consent forms from all the participants were received.
Statistical analysis. The results are presented as mean ± SD (standard deviation) or medians (1/4-3/4 quarters) and were obtained from three independent experiments. Normality (Kolmogorov-Smirnov) test was initially carried out and since the normality assumption was satisfied for the comparison of means between two groups, the paired t test or unpaired t test were used. For comparisons between multiple groups, one-way ANOVA was applied. For variables are skew distribution, Mann-Whitney's U test was applied. Spearman's analysis was used to analyze the linear correlation between groups. The clinical diagnostic value of the candidate circRNA_103765 was assessed with a receiver operating characteristic (ROC) curve. A P value < 0.05 was considered statistically significant. All statistical analyses were performed using GraphPad Prism 7.04 (GraphPad Software, La Jolla, CA) and SPSS 19.0 (SPSS Inc., Chicago, IL, USA).