Chao Chen, Dina Zhu, Shuai Zhang, Wensheng Zhang*

a Zhuhai Branch of State Key Laboratory of Earth Surface Processes and Resource Ecology, Advanced Institute of Natural Sciences,Beijing Normal University, Zhuhai 519087, China

b Engineering Research Center of Natural Medicine, Ministry of Education, Advanced Insti-tute of Natural Sciences, Beijing Normal University, Zhuhai 519087, China

c International Cooperation Laboratory of Molecular Medicine, Academy of Chinese Medical Sciences, Zhejiang Chinese Medical University,Hangzhou 310053, China

d Beijing Key Laboratory of Traditional Chinese Medicine Protection and Utilization, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, China

Keywords:

Catechin

Diabetic nephropathy

CircRNA

miRNA

ceRNA

A B S T R A C T

(+)-Catechin (CE) is mainly found in green and black tea and has many biological effects, such as antiinflammatory, anti-cancer, anti-viral effects, protecting human organs, especially the kidney.This study aims to identify the circRNAs induced by CE in db/db mice and their roles in diabetic nephropathy progression.After the db/db mice were treated with CE, RNA-seq was performed to identify the differentially expressed circRNA and mRNAs.The ceRNA regulatory network was constructed and analyzed using bioinformatics software and public databases (Cytoscape, ClueGO, MiRWalk, STRING, et al.).Our results revealed that 6 differentially expressed circRNAs are most associated with the cholinergic synapse, neurotrophin signaling pathway, and insulin signaling pathway.Among these, circRNA.5549 and circRNA.4712 might regulate Cd36, Cyp26b1, C8a, Cyp2j13, Grem2 genes through ceRNA regulatory mechanism in the presence of CE treatment.The expanded network of proteins interacting with these 5 genes shows that the TGF-β signaling pathway, signaling pathways regulating pluripotency of stem cell, fat digestion and absorption, and PPAR signaling pathway was highly enriched.Overall, circRNA.5549 and circRNA.4712 exhibit a promotive function in CE-treated db/db mice, especially in circRNA.5549/miR-29a-5P/Cd36 regulatory network, and this evidence suggest that their ceRNA regulatory network might be a therapeutic target for DN in humans.

1.Introduction

Diabetes mellitus (DM) is an enormous and growing public health problem.It was estimated that in 2017 there are 451 million adults with diabetes worldwide, and by 2040, the number will increase to 693 million [1].Diabetic nephropathy (DN) is the most common microvascular clinical complication among diabetic patients and one of the leading causes of end-stage kidney disease.Until recently, there were few treatment options for patients with poor performance status but depended on early detection, glycemic control, and strict blood pressure management [2,3].Therefore, early prevention and treatment of DN are particularly important to slow down DN progression,improving the patients’ lives.

The catechins in tea have been reported to protect against chronic diseases such as cardiovascular complications, diabetes, and cancer [4].Al-Attar et al.[5].reported that some crude extracts of tea leaves,including CE, effectively alleviate diabetic-mediated kidney dysfunctions in STZ-induced diabetes mice.Similarly, our earlier study found that CE significantly lowered plasma creatinine and urea levels in diabeticdb/dbmice and ameliorated DN by trapping methylglyoxal [6].Cheng et al.[7]provide direct evidence that catechin inhibits the activation of NF-κB, AMPK, FOXO3a, and SIRT1 in adipocytes.At the chemical molecular and protein levels,ROS-related mechanisms [8]and catechin-protein interactions [9]are considered two fundamental ways catechin performs its biological functions.At the nucleic acid level, epigenetic changes induced by green tea catechin have been widely reported [10,11].We also note that accumulating evidence has shown that non-coding RNA (ncRNA)are in DN progression [12], but few are described as inducing by CE.

Circular RNA (circRNA) is an endogenous ncRNA with a covalently closed cyclic structure [13].Intriguingly, circRNAs are increasingly recognized for their microRNA (miRNA) sponge activity roles in regulating the biological process and gene expression because they act as competing endogenous RNAs (ceRNAs), which disrupt normal siRNA: mRNA interactions and rescue mRNA.These complex ceRNA networks are always found in the initiation and progression of diabetes [14].For example, the cricRNA CDR1as could increase insulin content and secretion by acting as a miR-7 sponge [15].The circHIPK3 exhibits a promotive function in DN by sponging miR-185 [16].Although the details of the ceRNA regulatory networks may not be clear, the circRNA target genes regulating the DN progression are possible.

To our knowledge, we are the first to identify the ceRNAs indb/dbmice involving in CE ameliorates DN.Brie fly, we isolated circRNAs that were differentially expressed (DE) in mice treated with CE,predicted their miRNAs and corresponding target genes, then identified potential cellular pathways that these DE mRNAs could affect.Finally,screening with DE genes, we constructed 6 possible ceRNA pathways and assessed the plausibility of their presence in the cells.

2.Materials and methods

2.1 Animals and treatments

All BKS.Cg-Dock7m+/+Leprdb/J (db/db) mice (Cavens Lab.,Changzhou, Jiangsu, China) were housed under standard controlled conditions at (20 ± 2) °C and 12 h light/dark cycle.The 16-weekold mice were divided into 2 groups (n= 8 per group), in which the treatment group was fed CE (25 mg/kg·day, Nanjing Zelang Pharmaceutical Technology Co., Ltd.Jiangsu, China) for 8 weeks.CE was freshly dissolved in sterile distilled water before feeding via gastric gavage.The mouse kidneys were stored in liquid nitrogen before use.The Beijing Normal University Laboratory Animals Care and Use Committee approved all animal studies under the NIH Guide for the Care and Use of Laboratory Animals.

2.2 RNA library construction and sequencing

The total RNA was prepared from the whole kidney with TRIzol reagent (Invitrogen, Carlsbad, CA, USA).When the DNA was removed by RNase-free DNaseI (Qiagen, Valencia, CA, USA), the rRNA was then removed from the RNA sample using Ribo-Zero Gold rRNA kit (Illumina.San Diego, CA, USA).About 5 μg of mouse RNA were fragmented at high temperature using divalent cations to prepare the sequence cDNA library with a final insert size of about 350 bp.RNA-seq (Paired-end RNA sequencing) was performed on Illumina Hiseq2500 (Illumina, San Diego, CA, USA).

2.3 Sequence and primary analysis

About 1 g of sequence data per sample was used for mRNA and circRNA identification.For mRNA data, after obtaining clean data, the fragments were mapped into GRCm38.p4 (mm10) genome using the BWA-MEM method [17], and the expression of genes was assessed by FPKM-based statistical methods in edgeR [18].For circRNA, the CIRI method was used for projections [19].Since sequencing data are not available to cover the complete sequences of circRNAs, the junction reads at the back-splicing sites of circRNAs were used for calculating the circRNAs expression.The SRPBM(spliced reads per billion mapping) was used to normalize the junction reads.The circRNAs location in the genome was visualized with CIRCOS [20].The heatmap, volcano, and bubble diagrams were created with R.The KEGG analysis was performed with ClueGO [20]plug-in Cytoscape.

2.4 CircRNA target miRNA and miRNA target mRNA prediction

miRDB [21]was used to predict the DE circRNA target miRNAs.To consider which cellular pathways DE circRNAs may interfere with miRNAs, we retrieved experimentally validated miRNAs target mRNA from miRTarBase [22]in MiRWalk 3.0 website tool (http://mirwalk.umm.uni-heidelberg.de/).In CE-treated mice, we used predicting miRNAs to understand further which DE genes may be affected by DE circRNAs.The miRDB or TargeScan database was used in MiRWalk 3.0 to predict the miRNAs that bind to DE mRNAs’ 3’UTR region.Overlapping miRNA obtained from miRDB by circRNAs and predicted by 3’UTR region of DE mRNAs are assumed to be bridge miRNAs for sponge regulation.All possible ceRNA networks were visualized with Cytoscape 3.8.0(https://cytoscape.org/).

2.5 Construction of a protein-protein interaction (PPI)network and identification of core pathways

Using STRING (https://string-db.org/), we obtained the proteins that may interact with ceRNA regulating genes, constructed a PPI network, and analyzed its KEGG pathway.The core pathways were visualization by adobe illustration (AI, Adobe Systems Incorporated,San Jose, CA, USA).

2.6 RT-qPCR analysis

RT-qPCR was performed with 7500 real-time PCR system.The PCR temperature setting parameters were 60, 95, and 72 °C, 30 cycles.Three biological replicates and 3 technical replicates were used (Additional file S1 and S2).The qPCR data were analyzed using the delta-delta Ct method.

2.7 Statistical analysis

The multiplet-test was used for RT-qPCR measurements.In addition, all the software’s statistical methods used for corresponding biometric analysis were presented in the results.

3.Results

3.1 Overview of circRNAs in db/db mice genome.

CE (Fig.1a) might ameliorate renal dysfunction indb/dbmice as a consequence of inhibiting AGEs formation [6].Using RNA-seq,6 284 circRNAs were obtained from CE treated or untreateddb/dbmice.Fig.1b represents all locations of circRNA on the mice genome,which shows circRNAs are almost distributed on all chromosomes,but few on chromosome Y.

Fig.1 CE and the distribution of circRNAs in mice genome.(a) The CE structure; (b) the count numbers (RNA-seq) of circRNA in mice genome.The treated or untreated mice are marked with different colors.

3.2 CE Regulate circRNAs expression profiles in db/db mice.

A total of 30 DE circRNAs were identified from CE-treated mice.The circRNAs fold change |log2FC| ≥ 1.5 andP＜ 0.05 were selected to construct the heatmap (Fig.2a) or mark in the volcano diagram (Fig.2b).The co-expression relationships between genes are clustered together in the heatmap diagram.The details of all DE circRNAs characterization are summarized in Additional file S3.Since the ceRNA sponge regulatory mechanism is predicated on the presence of a rich circRNA gene expressed, the top 5% highexpressed circRNAs were selected for further analysis (Table 1).All sequence details of these 6 target circRNAs are shown in Additional file S4.

Fig.2 The DE circRNAs in CE-treated db/db mice.(a) Clustered heatmap for CE-induced circRNAs from mouse kidneys.(b) Volcano plot of DE circRNAs.Red represents DE circRNAs with P ＜ 0.05 and log2FC ≥ 1.5 or ≤ -1.5.Green and black indicate no significant DE lncRNAs.log2FC value marked with 10 or–10 shows the zero expression of circRNAs in CE treated or untreated mice, respectively.

Table 1Top 5% high-expressed circRNA.

3.3 Prediction of miRNA interacting with top 5% high expressed circRNAs

We predicted the sponge target miRNA of the top 5% highexpressed DE circRNAs.Six circRNAs sequences were set as seeds for executing the miRDB search program, and 69 miRNAs were identified interacting with these circRNAs (Additional file S5).

3.4 The miRNAs experimentally target genes and their KEGG analysis

To understand which cellular pathways DE circRNAs may interfere with miRNAs, we first retrieved experimentally validated miRNAs target mRNAs from miRTarBase.In our results, 251 genes bind to the miRNAs were identified in the MiRWalk web (Additional file S6).In the ClueGO KEGG enrichment analysis (Fig.3a), we find that 36.84% of the pathway are associated with cholinergic synapse,and 26.32% are the neurotrophin signaling pathway.However,21.05% of the pathway are about insulin signaling.Some possible DN-associated pathways were also enriched, such as the AMPK signaling pathway and apelin signaling pathway (Fig.3b).

Fig.3 The KEGG enrichment analysis of the 6 DE circRNAs-miRNA target genes.(a) Bar plot.*P ＜ 0.05, **P < 0.01.(b) Pie plot.The color indicates the integration of similar items in the bar chart.

Fig.3 (Continued)

3.5 The DE genes and their predicted interacted miRNA.

To find as accurate DE genes involved in the ceRNA regulatory network as possible, we selected |log2FC| ≥ 1.5 andq-value ＜ 0.05 as the threshold to screen DE genes.In the CE-treateddb/dbmice, 56 DE genes were identified, including 30 up-regulated and 26 downregulated genes (Fig.4).The miRDB or TargeScan database was used in the MiRWalk web to predict the miRNAs that bind to DE mRNAs’3’UTR region.

Fig.4 The DE genes in CE treated db/db mice.(a) Clustered heatmap for CE-induced mRNAs from mouse kidneys.(b) Volcano plot of DE circRNAs.Red represents DE mRNAs with q-value ＜ 0.05 and log2FC ≥ 1.5 or ≤ -1.5.Green and black indicate no significant DE mRNAs.

3.6 Construction of a circRNA/miRNA/mRNA network

Venn diagram analysis was performed to find which miRNA are bridges of DE circRNAs and DE mRNAs (Fig.5a).A total of 11 miRNAs are identified, including: mmu-miR-34b-5p, mmu-miR-449a-5p,mmu-miR-7094-1-5p, mmu-miR-29a-5p, mmu-miR-6925-5p,mmu-miR-7041-5p, mmu-miR-7683-3p, mmu-miR-6338, mmumiR-669c-3p, mmu-miR-7212-3p, and mmu-miR-134-5p.All these miRNAs are predicted to bind to the 3’UTR region of target mRNAs.In Fig.5b, circRNA/miRNA/mRNA networks were established with circRNA target to miRNA and miRNA target to mRNA relationship.By reason that the circRNA and gene are always co-expressed in the ceRNA regulatory network, theCd36,Cyp26b1,C8a,Cyp2j13(human isoform isCYP2J2),Grem2were considered to be the ceRNA target DE genes.

Fig.5 Construction of a circRNA–miRNA–mRNA network.(a) The intersection of miRNA between circRNA targets and predicted by DE genes with miRDB and TargeScan.miRNAs were combining 3’UTR regions in genes.(b) The possible ceRNA network.

3.7 Construction of a PPI network and identification of core signaling pathway

Using the STRING database, we retrieved about 55 proteins from the human that might interact withCd36,Cyp26b1,C8a,Cyp2j13(CYP2J2),Grem2.Visualization of the network enriched in the KEGG pathway was accomplished using the R program (Fig.6a).The significantly enriched core pathways were marked out with circles(Fig.6b).In all of enriched 23 KEGG pathways, TGF-β signaling pathway, signaling pathways regulating pluripotency of stem cell, fat digestion and absorption, and PPAR signaling pathway was highly enriched.

Fig.6 The KEGG enrichment analysis of the proteins that interacted with ceRNA regulated genes.(a) 23 KEGG pathways enrichment in all interacting proteins.(b) The top 4 significantly enriched the KEGG pathway in interacting proteins.

4.Discussion

Thedb/dbmouse has significant type 2 diabetes characteristics such as hyperglycemia, hyperlipidemia, and hyperinsulinemia.DN is one of the most severe chronic complications in patients with diabetes.There are few reports on the role of circRNA in the treatment of DN by CE, and no circRNA regulatory network has been reported indb/dbmouse model.Our results revealed that circRNAs expression is covered the whole genome indb/dbmice (Fig.1b).When CE was used as a DN drug to treat mice, 30 DE circRNAs were identified.Since the “miRNA sponge” mechanism requires a “l(fā)arge” amount of circRNA, the top 5% high-expressed circRNAs were selected in subsequent studies.Furthermore, for circRNAs to regulate other RNA transcripts by competing for shared microRNAs as ceRNAs, reliable binding of miRNAs to other RNAs is key to analyzing the ceRNA regulatory network.Hence, we used predictive and experimental databases for the analysis of mRNA-binding miRNAs, respectively.However, we have to use a predictive approach to analyze the circRNA target RNAs based on the fact that they are new circRNAs.

First, we studied the data that miRNAs and mRNAs can bind to each other experimentally to observe the potential ability of circRNAs to regulate the KEGG pathway.These 6 DE circRNAs(Table 1) interaction miRNAs and their target genes are enriched in the cholinergic synapse, neurotrophin signaling pathway and insulin signaling.A previous epidemiological study in Japan has indicated that green tea CE consumption improves cognitive impairment [23].Furthermore, Ci et al.[24]argued that cholinergic synapse serves an essential role in the cognitive function of type 2 diabetic mice.Hence,the 6 circRNAs may have the ability to change the cholinergic system.Much more interesting, the neurotrophin signaling pathway has been reported as necessary for glucose-induced insulin secretion [25],and the AMP-dependent protein kinase, a key molecule regulating biological energy metabolism, has been argued as a central protein to study diabetes and other metabolism-related diseases [26,27].Additionally, apelin has been reported involved in the progression of DN by inhibiting autophagy in podocytes [28].In summary, our study firstly revealed the potential role of 6 circRNAs in a systemic understanding of the CE function in DN progression.

Second, we studied predictive data of miRNA and mRNA binding to discover potential regulatory networks of ceRNAs.Coexpression of ceRNAs and its target RNAs is an essential parameter in constructing ceRNA regulatory networks, predicted miRNAs that can bind to the DE gene should also be fully considered.In this work, 5 DE genes are identified as the ceRNA network targets:Cd36,Cyp26b1,C8a,Cyp2j13(human isoform isCYP2J2),Grem2.Two new circRNAs, circRNA.5549 and circRNA.4712 are at the center of these ceRNA networks (Fig.5b).Based on these 5 DE genes’interaction proteins, KEGG analysis revealed that target genes were significantly associated with the TGF-beta signaling pathway.The signaling pathways regulating pluripotency of stem cells, fat digestion and absorption, and PPAR signaling pathway were high enriched(Fig.6).Some previous works suggested kidney lipotoxicity due to disorders of lipid metabolism may be a causal mechanism leading to DN and renal dysfunction [29], and stem cells can mediate the regulation of the immune response balanced and ultimately contribute to the down-regulation of the local inflammatory response and the reduction of tissue inflammation [30].We suggest CE might stabilizing lipid metabolism and regulating stem cell development through DE genes.To our interest, the TGF-β signaling pathway is the most significantly enriched in the PPI network, and it can be interfered with by the PPAR signaling pathway.Moreover, these two pathways have some genes that belong to the signaling pathways regulating pluripotency of stem cell, and fat digestion and absorption pathways.It suggested that even though the DE ceRNAs target DE gene effects DN are systemic effects, the core pathway may be the PPAR signaling pathway because the hub DE gene is the Cd36 (Fig.6b).

Third, we investigated the direct relationship between circRNA target DE genes and DN to support the ceRNA regulatory network’s credibility.Griffin et al.[31]reported that glucose regulates the expression ofCd36at the level of translation.Undoubtedly, fasting and postprandial glucose could be decreased by CE [32].Moreover,Pucha?owicz et al.[33]reported the detrimental effects ofCd36signaling mediated by oxidative stress, inflammation, apoptosis, and fibrosis leading to the progressive are confirmed in DM patients.In our results, we showed circRNA.5549 andCd36were also remarkably down-regulated by CE.We suggested that the circRNA.5549 might act as a ceRNA to mediateCd36down-regulated and affect DN progress through the PPAR signaling pathway.All 5 DE ceRNA target genes are associated with diabetes:Cyp26b1andCyp2j13belong to the cytochromeP450gene family, which generate a series of bioactive products attributed to protective/anti-inflammatory and insulin-sensitizing effects in animal models [34].TheC8agene was reported to relate to the innate immune response [35].TheGrem2gene was reported mediated podocyte apoptosis in high glucose milieu [36].According to the above reports, we suggested the circRNA.5549 and circRNA.4712 function as a ceRNA to regulateCd36,Cyp26b1,C8a,Cyp2j13(CYP2J2),Grem2expression in CE treateddb/dbmice.

Finally, among the 6 circRNAs, we suggest that the circRNA.5549/miR-29a-5P/Cd36regulatory network is most likely to improve the condition of CE-treated DN mice.To begin with, only circRNA.5549 and circRNA.4712 had co-expressed genes (Fig.5b),implying that the processes of these two circRNA regulating genes can be detected at the mRNA level.For another thing, of the 5 coexpressed genes (Cd36,Cyp26b1,C8a,Cyp2j13(CYP2J2),Grem2),onlyCd36was partitioned into two diabetes-related pathways in the PPI analysis (Fig.6b).As a target protein of circRNA.5549/miR-29a-5P,Cd36can interact with fat digestion and absorption or PPAR signaling pathway.And thirdly, miRNAs in the circRNA.5549/miR-29a-5P/Cd36regulatory networks are highly associated with diabetes.It has been reported that miR-29a plays a crucial role in reducing glucose-stimulated insulin secretion [37], while miR-29a-5P can be significantly up-regulated after weight loss intervention [38].If ceRNA regulation is available, a reduction in circRNA.5549 might increases functional miR-29a-5P.Therefore, the biological effect of circRNA.5549 reduction may have a resemblance to the weight loss intervention.The conclusion can be drawn that all components of this regulatory pathway are more relevant to DN than other ceRNAs.

5.Conclusions

In conclusion, we found that 6 DE circRNA has the potential to influence DN pathways via miRNA.CircRNA.5549 and circRNA.4712 might regulateCd36,Cyp26b1,C8a,Cyp2j13,Grem2genes through ceRNA regulatory mechanism.KEGG pathway analysis revealed that these two circRNAs are mainly involved in the TGF-β signaling pathway, signaling pathways regulating pluripotency of stem cell, fat digestion and absorption, and PPAR signaling pathway.Considering the critical position ofCd36in the PPI map and the role of miR-29a-5P in the insulin secretory pathway, circRNA.5549/miR-29a-5P/Cd36has the potential to be an attractive target for therapeutic intervention.However, further confirmation is needed.

Declaration of competing interest

The authors declare that they have no conflict of interest.

Acknowledgements

This work was supported by the National Nature Science Foundation of China (81771152), National Key R&D Plan (No.2017YFC1702500), and the Beijing Joint Project for the Central-Affiliated University (2017-01).

Appendix A.Supplementary data

Supplementary data associated with this article can be found in the online version, at http://doi.org/10.1016/j.fshw.2021.12.023.