The HK-2 cell line (CRL-2190, ATCC) was grown in complete growth medium, keratinocyte serum-free medium (K-SFM, 17005-042, Gibco; Thermo Fisher Scientific, Inc.) supplemented with 0.05 mg/ml bovine pituitary extract and 5 ng/ml epidermal growth factor. The HK-2 cells were incubated at 37°C in a suitable incubator containing a 5% CO₂ in air atmosphere. To induce H/R, the HK-2 cells were treated with K-SFM medium and subjected to 12 h of hypoxia (5% CO₂, 1% O₂ and 94% N₂), followed by cultivation in complete medium for 0/3/6/9/12 h of reoxygenation (5% CO₂, 21% O₂ and 74% N₂).

The expression of mRNAs, circRNAs and miRNAs was analyzed following the cell treatment or transfection. Total RNA was isolated from the HK-2 cells using Trizol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Total RNA (1.5 μg) was used for cDNA synthesis at 42°C with SuperScript™ IV revertase (Invitrogen; Thermo Fisher Scientific, Inc.). qPCR reactions were performed with SYBR^®-Green Mix (Applied Biosystems; Thermo Fisher Scientific, Inc.) for mRNA and circRNA expression analysis, with the TaqMan™ MicroRNA Assay kit (Applied Biosystems; Thermo Fisher Scientific, Inc.) for miRNA expression analysis. The specific primer pairs of ITGB1 (NCBI Reference Sequence NM_002211), PIM1 (NM_002648), neurofibromin 2 (merlin) (NF-2; NM_181833), GAPDH (NM_001357943), GATA1 (NM_0020nf-49) and ubiquitin protein ligase E3 component n-recognin 4 (UBR4; NM_020765.3) were designed using NCBI Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/index.cgi?LINK_LOC=BlastHome). The stem-loop primer of miR-328-3p (miRbase ID: MIMAT0000752) was designed using the Primer3 (https://primer3.ut.ee/) online tool. In addition, Primer3 was utilized to design the divergent primers of circITGB1 (CircBank ID hsa_circ_0018148), circARL6IP1 (CircBank ID hsa_circ_0038229) and circH-NRNPF (CircBank ID hsa_circ_0018263). After obtaining the divergent primer sets as aforementioned, CircPrimer 2.0 software (downloaded from https://www.bio-inf.cn/; DOI: 10.1186/s12859-018-2304-1) was utilized to examine the specificity of the circRNA primers. The lists of primers for each mRNA, circRNA and miRNA are presented in Table I. Relative expression was analyzed using the 2^−ΔΔCq method (16) and normalization with internal controls (GAPDH or U6). As for RNase R treatment, total RNA (2 μg per group) of HK-2 cells were incubated with RNase R (a final concentration of 10 μg/ml) for 3 min at 37°C (Beijing Solarbio Science & Technology Co., Ltd.). RT-qPCR was performed to determine the expression levels of ITGB1 mRNA and circITGB1. To identify the circular feature of circITGB1, Oligo(dT) 18 primers (Beijing Solarbio Science & Technology Co., Ltd.) and random primers (Beijing Solarbio Science & Technology Co., Ltd.) were employed to perform cDNA synthesis, respectively. The resulting cDNAs were then detected by qPCR using the primers targeting circITGB1 or ITGB1, respectively.

The knockdown of circITGB1, PIM1 and GATA1 was achieved by transfecting the HK-2 cells with small interfering RNAs (siRNAs). siRNAs targeting PIM1 (NM_002648), circITGB1 (circbase ID: hsa_circ_0018148) and GATA1 (NM_002049) were designated as siPIM1, si-circITGB1 and si-GATA1, respectively. siPIM1 and si-GATA1 was designed using BLOCK-iT™ RNAi Online Designer (http://rnaidesigner.thermofisher.com/rnaiex-press/design.do). In addition, the Circinteractome online database (https://circinteractome.irp.nia.nih.gov/index.html) (17) was utilized to design the siRNA specific to circITGB1. All siRNA sequences were synthesized by Shanghai Sango Biotechnology Co., Ltd. and were as follows: siPIM1 sense, 5′-CGGGUUUCUCCGGCGUCAUUAGGdTdT-3′ and antisense, 5′-CCUAAUGACGCCGGAGAAACCCGdTdT-3′; si-circITGB1 sense, 5′-GUGGAGAAUCCAGAUGAAAAUdTdT-3′ and antisense, 5′-AUUUUCAUCUGGAUUCUCCACdTd T-3′; si-GATA1 sense, 5′-CCCAGUCUUUCAGGUGUACCCAUUGdTdT-3′ and antisense, 5′-CAAUGGGUACACCUGAAAGACUGGGdTdT-3′. The miR-328-3p mimics, inhibitors and negative controls were obtained from Guangzhou RiboBio Co., Ltd. and the sequences were as follows: miR-328-3p mimic (miR10000752-1-5) sense, 5′-CUGGCCCUCUCUGCCCUUCCGU-3′ and antisense, 5′-GGAAGGGCAGAGAGGGCCAGUU-3′; NC mimic (miR1N0000001-1-5) sense, 5′-UUCUCCGAACGUGUCACGUTT-3′ and antisense, 5′-ACGUGAACAGUUCGGAGAATT-3′; miR-328-3p inhibitor (miR20000752-1-5) sense, 5′-ACGGAAGGGCAGAGAGGGCCAG-3′; NC inhibitor (miR2N0000001-1-5), 5′-CAGUACUUUUGUGUAGUACAA-3′. The GATA1 coding sequence fragment (~1.4 kb) was amplified with a primer set (forward, 5′-ACTGAGCTTGCCACATCCCC-3′ and reverse, 5′-GGGCCCAGAGACTTGGGTTG-3′) and inserted into the pcDNA3.1 vector (Invitrogen; Thermo Fisher Scientific, Inc.) between the BamH1 and XhoI sites. The plasmid construct was transformed in E. coli, and selected colonies were then cultured. Plasmid DNA was purified using the Plasmid Maxi Preparation kit (Beyotime Institute of Biotechnology). To perform cell transfection, the HK-2 cells were plated in 96-well plates and incubated for 24 h at 37°C. Each 1 μg of plasmids (or 50 nM miRNA mimic or 100 nM miRNA inhibitor) were transfected into the HK-2 cells using Lipofectamine™ 3000^® Transfection Reagent (Invitrogen; Thermo Fisher Scientific, Inc.). After 48 h, the HK-2 cells were subjected to H/R (12 h of hypoxia and 6 h of reoxygenation). Additionally, the transfection efficiency was examined using RT-qPCR.

The HK-2 cells were transfected with the siRNAs or miRNA mimic or miRNA inhibitor for 48 h as described above and then subjected to H/R. The cells were then collected and plated in a 96-well plate (1,000 cells/well). Following cell adherence, Cell Counting Kit-8 (CCK-8) solutions (Beijing Solarbio Science & Technology Co., Ltd.) were added to the cells for detecting cell viability. At 2 h following the addition of CCK-8, the absorbance at 450 nm was measured using a microplate reader (DNM-9606, Beijing Pulangxin technology Co., Ltd.) and cell the viability rate was analyzed compared to the control group (100%).

The HK-2 cells were subjected to H/R as described above. For cell apoptosis analysis, 1×10⁶ cells were harvested using trypsase solution (Beijing Solarbio Science & Technology Co., Ltd.) and resuspended in binding buffer (Beijing Solarbio Science & Technology Co., Ltd.). Following centrifugation, the pellets were incubated with 5 μl Annexin V-Fluorescein isothiocyanate isomer (FITC) solutions (Beijing Solarbio Science & Technology Co., Ltd.) for 10 min and then with with 5 μl propidium iodide (PI) solutions (Beijing Solarbio Science & Technology Co., Ltd.). Flow cytometry was performed using a Guava EasyCyte Mini System (Luminex, Inc.).

Enzyme linked immunosorbent assay (ELISA) interleukin-1β (IL-1β), interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α) kits (Beijing Solarbio Science & Technology Co., Ltd.) were utilized for the quantitative determination of concentrations in the cell culture supernatants. Firstly, 100 μl samples were added to each well and incubated for 90 min at 37°C. The cells were then incubated with 100 μl working solution (a component of the ELISA kits) containing biotin-conjugated anti-human IL-1β/IL-6/TNF-α antibodies (a component of the ELISA kits) for 60 min at 37°C. Subsequently, 100 μl streptavidin-horseradish peroxidase (HRP) working solutions (a component of ELISA kit) were utilized for a further 30 min of incubation at 37°C. Following incubation with 100 μl substrate solution (a component of the ELISA kits) away from light at 37°C for 15 min, the reactions were terminated using stop solutions (a component of the ELISA kits) and the absorbance at 450 nm was measured using a microplate reader (DNM-9606, Beijing Pulangxin technology Co., Ltd.).

Using the keywords 'renal ischemia reperfusion' and 'microRNA', datasets were searched in the GEO database (https://www.ncbi.nlm.nih.gov/gds/). The selection criteria were as follows: i) Datasets with non-coding RNA profiling types; ii) datasets containing a sham operation group and a renal IRI group; iii) datasets with reasonable biological replications. The GSE124669 dataset contains samples from multiple time points post-IRI. The GSE124669 dataset was used for RNA-seq analysis. In brief, Sonoda et al (18) constructed a bilateral IRI model. Sprague-Dawley rats (male, 10 weeks) were randomly divided into the sham operation group and IRI group. The left and right renal vascular pedicles were subjected to clamped operation for 25 min and then reperfused with blood. The RNA used for sequencing was isolated from urinary exosomes at days 1, 2, 3, 7 and 14 post-IRI. The RNA samples were then sequenced using the Illumina HiSeq 2500 platform. Raw sequence files were obtained from the SRA (SRP175286, https://www.ncbi.nlm.nih.gov/sra). The RNA sequencing data were obtained from Gene Expression Omnibus (GSE124669). The reads containing adapter sequences and low-quality reads (Phred-scale quality score <20) were removed from the raw data. The sequence quality was assessed using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/, released Version 0.11.9). Clean data were subjected to analysis using the IDEG6.2 online tool (http://telethon.bio.unipd.it/bioinfo/IDEG6_form/) (19). The differentially expressed miRNAs with ±1.5-fold change (FC) and an adjusted P-value <0.05 were obtained using RNA sequencing analysis. The differentially expressed miRNA (rno-miR-328a-3p) was then subjected to an analysis on the UCSC Genome Browser (https://blast.ncbi.nlm.nih.gov/Blast.cgi) to confirm the conservation between rno-miR-328a-3p and hsa-miR-328-3p.

The mRNAs targeting miR-328-3p were predicted using three different databases: TargetScan (http://www.targetscan.org/vert_72/), Pictar (https://pictar.mdc-berlin.de/cgi-bin/new_PicTar_mouse.cgi) and miRDB (http://mirdb.org/). The overlapping interactions were calculated utilizing a web tool (http://bioinformatics.psb.ugent.be/webtools/Venn/). A graphical output in the form of venn diagram was produced to illustrate the overlapping mRNA interactions.

The binding sites between circRNAs and miR-328-3p were predicted using StarBase v2.0 (http://starbase.sysu.edu.cn/). StarBase perform miRNA-circRNA interactions by intersecting the predicting target sites of miRNAs with binding sites of Ago protein, which were derived from CLIP-seq data. The Ago clipExpNum is defined as the number of supporting AGO CLIP-seq experiments. An additional screening parameter (8mer class, clipExpNum strict stringency ≥5) was changed to refine the results. The 51 candidate circRNAs of miR-328-3p are listed in Table II. The 51 candidate circRNAs were sorted using Ago clipExpNum in descending numerical order. The top three circRNAs with a high clipExpNum were subjected to further RT-qPCR analysis in the cells subjected to H/R.

RIP assay was carried out to confirm the RNA-Ago2 protein interaction by utilizing the Dynabeads™ Protein A Immunoprecipitation kit (Invitrogen; Thermo Fisher Scientific, Inc.). Briefly, the HK-2 cells overexpressing circITGB1 or miR-328-3p were incubated in a 25 cm² cell culture flask for 48 h. A total of 5×10⁶ cells were lysed using cell extraction buffer. Dynabeads magnetic beads were resuspended on a roller for 5 min. Subsequently, 50 μl (1.5 mg) beads were placed on the magnet to separate the beads from the solution. Typically, 1 μg Anti-Ago2 (ab186733, Abcam) or IgG (ab172730) antibodies diluted in 200 μl Ab Binding Buffer were incubated with the beads for 10 min at room temperature. The tube was placed on the magnet to remove the supernatant. The magnetic bead-Ab complex was resuspended in 200 μl Ab Binding and washing buffer (a component of the immunoprecipitation kit). To allow the antigen (Ag) to bind to the bead-Ab complex, the cell sample (100 μl) containing RNAs were incubated with the bead-Ab complex in a shaker for 10 min. The tube containing the bead-Ab-Ag complex was placed on the magnet to remove the supernatant. The bead-Ab-Ag complex were washed with 200 μl washing buffer three times. The bead-Ab-Ag complex was then separated on the magnet between each wash. The resulting complex was resuspended in 100 μl washing buffer and was transferred to a clean tube. The tube was placed on the magnet to remove the supernatant. A total of 20 μl elution buffer (a component of the immunoprecipitation kit) was then incubated with the beads-Ab-Ag complex for 2 min to dissociate the complex. The tube was placed on the magnet. The supernatant containing eluted Ab and Ag was transferred to a clean tube. The coprecipitated RNAs were isolated from the supernatant by using RNeasy MinElute Cleanup kit (Qiagen, Inc.) and reverse transcribed into cDNA using the Prime-Script RT reagent kit (Takara Biotechnology Co., Ltd.) according to the manufacturer's instructions. Subsequently, the mRNA levels of circITGB1 and miR-328-3p were measured using RT-qPCR assay as described above.

The in silico analysis of the ITGB1 promoter (2,000 bp upstream of the transcriptional start site) was performed to identify the binding sites for GATA1. The sequence of the ITGB1 promoter was downloaded from NCBI (https://www.ncbi.nlm.nih.gov/). The JASPAR database (http://jaspar.genereg.net/) provided the frequency matrix of transcription factor GATA1 (Matrix ID: MA0035.4). A FASTA-formatted sequence (ITGB1 promoter) was input to scan with the GATA1 matrix model. A total of four putative sites were predicted with a relative profile score threshold of 80%.

Four fragments in the ITGB1 promoter were amplified using PCR primers (Table I). The fragments of the ITGB1 promoter were inserted into the pGL3 luciferase vector (Promega Corporation) and the resulting vectors were named as ITGB1 promoter-luc or site1/2/3/4-WT, respectively. The QuickMutation™Plus Site-Directed Mutagenesis kit (Beyotime Institute of Biotechnology) was utilized to construct mutated ITGB1 promoter luciferase reporter vectors. Cells in which GATA1 was overexpressed or silenced (1×10⁵) were seeded in 96-well plates and co-transfected with the luciferase plasmid (2 μl/μg) using Lipofectamine™ 3000^® (Invitrogen; Thermo Fisher Scientific, Inc.). After 48 h, a luciferase reporter assay kit (Promega Corporation) was utilized for the detection of the luciferase activities. The relative luciferase activities were analyzed via normalization against the Renilla luciferase values.

Cells were collected by incubation with trypsin and lysed with radio immunoprecipitation assay (RIPA) buffer (Beijing Solarbio Science & Technology Co., Ltd.) for 30 min. Following centrifugation at 800 × g for 5 min at 4°C (Beckman Coulter Life Sciences), the concentration of total protein were determined using BCA Protein Assay kit (Beijing Solarbio Science & Technology Co., Ltd.). The proteins were then denatured at 95°C and then subjected to separation on 12% SDS-PAGE gels. A total of 20 μg proteins were then transferred onto polyvinylidene fluoride (PVDF) membranes and blocked with bovine serum albumin (BSA) buffer for 2 h at room temperature. The membranes were incubated with primary antibodies that were diluted in BSA buffer overnight at 4°C. The primary antibodies used were anti-PIM1 (1:2,000, ab54503), anti-NF-2 (1:10,000, ab109244) and anti-GAPDH (1:1,000, ab8245) (all from Abcam). The PVDF membranes were then incubated with HRP-conjugated secondary antibodies (1:1,000, ab7090; 1:5,000, ab97080, Abcam) for 2 h at room temperature. Finally, immunoreactive proteins were detected using enhanced chemiluminescence (ECL) Western Substrate (Thermo Fisher Scientific, Inc.) and visualized using a Chemiluminescence Imager (Tanon Science & Technology).

Data are presented as the mean ± standard deviation (SD) from three independent experiments. Statistical analysis between multiple groups was performed with one-way analysis of variance (ANOVA) and Bonferroni's post hoc comparisons test using GraphPad Prism 8.0 software (GraphPad, Inc.). A P-value <0.05 was considered to indicate a statistically significant difference.

miRNAs are one of the most genetically expressed non-coding RNAs in AKI (20) and consequently potentiate or inhibit the pathogenicity of inflammation (21). With an objective to explore the miRNAs regulated in AKI, a RNA sequencing dataset (GSE124669) was obtained and data analysis was performed. miRNA expression profiles in rats were clustered based on the treatment conditions (IR surgery group and sham group). The statistical analysis of the RNA-sequencing data with ±1.5 FC and adjusted P-value <0.05 generated a list that contained 36 miRNAs that were differentially expressed in the IR surgery groups at each time points compared with the sham groups. The heatmap displayed a total of eight miRNAs (rno-miR-378a-3p, rno-miR-328a-3p, rno-miR-200c-3p, rno-miR-23b-3p, rno-miR-9a-5p, rno-miR-21-5p, rno-miR-351-5p and rno-miR-3473) that were differentially expressed at two time points (Fig. 1A). It was previously reported that miR-378a-3p plays a crucial role in ferroptosis in AKI (22). The present study focused on the second highest differentially expressed miRNA (rno-miR-328a-3p). rno-miR-328a-3p was downregulated in the IR group on days 1/2/3. To explore the potential function of rno-miR-328a-3p in cell development and regulation, a conservation analysis we performed using UCSC Genome Browser (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The results indicated that the precursor sequence for both rat rno-miR-328a-3p and human hsa-miR-328-3p was highly conserved with a 94% sequence identity and only a four nucleotide difference (Fig. 1B). In order to elucidate whether miR-328-3p expression has any effect on cells damaged by H/R, HK-2 cells were subjected to hypoxia for 12 h and then cultured in a normoxic incubator for 0/3/6/9/12 h. The validation of miR-328-3p expression using RT-qPCR analysis was performed and the results revealed a reoxygenation-dependent inhibition of miR-328-3p levels in HK-2 cells (P<0.01 and P<0.001, Fig. 1C). The 9-h and 12-h reoxygenation decreased the survival rate of the HK-2 cells by ~90%. The cell amount was not sufficient to perform subsequent experiments. As the 9-h and 12-h reoxygenation may exert cytotoxic effects on HK-2 cells, the 6-h reoxygenation time point was selected for use in further experiments. Thus, it was concluded that miR-328-3p downregulation was induced by H/R in HK-2 cells.

Since miR-328-3p was expressed at low levels following H/R, it was overexpressed by miR-328-3p mimic transfection in HK-2 cells and its effects on various parameters related to renal injury were examined. The overexpression efficiency was confirmed using RT-qPCR assay (P<0.01, Fig. S1A). As shown in Fig. 1D, mimic transfection indeed led to a high level of miR-328-3p in the H/R-exposed cells compared with the H/R group (P<0.01 and P<0.001). It was observed that H/R induced a significant decrease in the viability of the HK-2 cells (P<0.001, Fig. 1E). Notably, the re-expression of miR-328-3p significantly increased cell viability when compared to H/R treatment alone (P<0.01, Fig. 1E). Moreover, while H/R treatment increased the cell apoptotic rate, the overexpression of miR-328-3p reversed the promoting effects of H/R on HK-2 cell apoptosis (P<0.001, Fig. 1F and G). The concentrations of inflammation-related cytokines were also significantly increased in the H/R group, as shown using ELISA (P<0.001, Fig. 1H-J). However, the overexpression of miR-328-3p inhibited the inflammatory response triggered by H/R (P<0.001, Fig. 1H-J). These results indicated that the pro-inflammatory effects of H/R on HK-2 cells were partly reversed by the overexpression of miR-328-3p.

One of the mechanisms through which miRNAs function in cell regulation is competitively targeting the 3′UTR region of target genes, which leads to the loss or gain of gene function and controls cellular biological behaviors (23). In the present study, bioinformatics analysis integrated with some online databases predicted a total of 13 genes targeting miR-328-3p (Fig. 2A). Furthermore, the expression levels of the predicted targets were screened in HK-2 cells following H/R. The results revealed that the exposure of HK-2 cells to H/R resulted in a >1.5-fold increase in PIM1 and NF-2 levels compared to the control group (P<0.01 and P<0.001, Fig. 2B). Subsequently, the present study examined whether PIM1 or NF-2 expression was regulated by miR-328-3p in H/R-e HK-2 cells. H/R led to the increased expression of PIM1 and NF-2, as determined using RT-qPCR and western blot analysis (P<0.001, Fig. 2C). Moreover, PIM1 and NF-2 expression was considerably lower in the H/R + miR-328-3p mimic group than in the H/R group (P<0.01 and P<0.001, Fig. 2C). The scanning of miR-328-3p for human PIM1 and NF2 gene recognition sites, revealed an 8mer complementary binding site in the 3′UTR region of PIM1 mRNA and 7mer-8 sequence complementary to the 3′UTR region of NF2 mRNA, as shown in Fig. 2D. To elucidate whether miR-328-3p targets PIM1 and NF2, the putative miR-328-3p recognition sequences of targets were cloned in the pMIR reporter vector. Co-transfection with PIM1 3′UTR-WT and miR-328-3p mimic in HK-2 cells significantly reduced the luciferase activity (P<0.001, Fig. 2E). In addition, the mutant PIM1 sequence in the putative miR-328-3p binding sites abolished this effect (Fig. 2E). However, luciferase reporter gene assay indicated that miR-328-3p failed to target the NF-2 3′UTR (Fig. 2F).

To examine the effects of PIM1 regulation by miR-328-3p in H/R-exposed cells, siPIM1, miR-328-3p inhibitor and the corresponding control plasmid were generated and transfected into HK-2 cells. The knockdown of miR-328-3p was further confirmed (P<0.001, Fig. S1B). RT-qPCR analysis was also performed to confirm the inhibitory degree of silencing PIM1 (P<0.001, Fig. S1C). Furthermore, the viability, apoptosis and inflammatory potential of H/R-exposed HK-2 cells were assessed. An ~20% decrease in cell viability was observed in the cells in which miR-328-3p was silenced, compared to the H/R-exposed cells (P<0.001, Fig. 2G). However, PIM1 knockdown in the miR-328-3p-silencing cells resulted in a significant increase in cell viability compared to the H/R + miR-328-3p inhibitor group (P<0.05, Fig. 2G). Moreover, it was observed that the re-expression of miR-328-3p induced a significant increase in the apoptosis of H/R-exposed HK-2 cells; it also further increased the levels of pro-inflammatory factors (P<0.001, Fig. 2H-L). In addition, the silencing of PIM1 reversed the cell damage triggered by H/R and miR-328-3p knockdown in HK-2 cells (P<0.001, Fig. 2H-L). These data indicated that PIM1 upregulation by miR-328-3p exerts pro-inflammatory effects in H/R-exposed HK-2 cells.

Since circRNAs are aberrantly regulated in human diseases, the present study explored whether the level of miR-328-3p in HK-2 cells could be regulated by specific circRNAs. To this end, circRNAs that have potential binding sites to miRNAs were first examined using StarBase (http://starbase.sysu.edu.cn/). By applying an additional screening parameter (8mer class, clipExpNum strict stringency ≥5), the results were narrowed down to three circRNAs candidates that target miR-328-3p (Fig. 3A). The present study then examined the expression levels of these circRNAs (hsa_circ_0038229, hsa_circ_0018263 and hsa_circ_0018148) in H/R-exposed HK-2 cells using RT-qPCR. The results revealed that H/R resulted in an upregulation of hsa_circ_0018148 (P<0.001, Fig. 3B). To confirm the effect of hsa_circ_0018148 (also known as circITGB1) on the miR-328-3p level in HK-2 cells, siRNA targeting circITGB1 was transfected into the HK-2 cells. RT-qPCR analysis was performed to confirm the inhibitory effect of si-circITGB1 on the circITGB1 expression level (P<0.001, Fig. S1D) and to exclude an off-target effect of the siRNA. miR-328-3p expression was also examined in ITGB1-silenced cells and it was found that the H/R-exposed cells that were transfected with si-circITGB1 expressed a higher level of miR-328-3p when compared to the cells that exposed to H/R (P<0.001, Fig. 3C). hsa_circ_0038229 (2,097 bp sequence in length), is generated from exons 2-14 of the ITGB1 gene located on chr10:33199150-33221528 (Fig. 3D). It was found that circITGB1 was resistant to RNase R digestion, while the linear mRNA of ITGB1 in HK-2 cells was markedly decreased by RNase R treatment (P<0.001, Fig. 3E). Moreover, RT-qPCR was utilized to detect the levels of circITGB1 and ITGB1 mRNA following reverse transcription with random primers and oligo(dt)18 primers (P<0.001, Fig. 3F). Additionally, as shown in Fig. 3G, a target sequence of circITGB1 was found in miR-328-3p, suggesting that circITGB1 directly targeted miR-328-3p. To further confirm this, a fragment of wild-type circITGB1 containing the putative miR-328-3p target sequence was first cloned into a luciferase reporter vector, and it was found that miR-328-3p mimic transfection resulted in a significant decrease in luciferase activity (P<0.001, Fig. 3H). However, the luciferase activity of the circITGB1-MUT + miR-328-3p mimic group had no significant change compared to the circITGB1-MUT + NC mimic group. In addition, RIP assay was performed in H/R-exposed HK-2 cells to precipitate circITGB1 and miR-328-3p with an anti-Ago2 antibody. The data suggested that both circITGB1 and miR-328-3p were evidently pulled down by anti-Aso2 antibody compared with the IgG group (P<0.001, Fig. 3I). Collectively, these results indicated that circITGB1 functioned as a sponge of miR-328-3p in HK-2 cells.

As it was found that circITGB1 functions as a regulator of miR-328-3p expression which modulates cell inflammation, the present study then investigated whether circITGB1 affects the inflammatory context by binding miR-328-3p. The knockdown of circITGB1 alone increased the growth of HK-2 cells compared to the H/R group (P<0.001, Fig. 4A). This was consistent with the reduction of cell apoptosis in the H/R + si-circITGB1 + NC inhibitor group (P<0.001, Fig. 4B and C). Furthermore, the combination of si-circITGB1 and miR-328-3p inhibitor transfection decreased cell viability compared to the co-transfection of si-circITGB1 and NC inhibitor (P<0.01, Fig. 4A). However, the apoptotic percentage was increased in the H/R + si-circITGB1 + miR-328-3p inhibitor group compared with the H/R + si-circITGB1 + NC inhibitor group (P<0.001, Fig. 4B and C). Moreover, the concentrations of pro-inflammatory factors (TNF-α, IL-6 and IL-1β) were decreased following si-circITGB1 transfection compared with the H/R group (P<0.001, Fig. 4D-F). However, the addition of miR-328-3p inhibitor reversed the inhibitory effects of si-circITGB1 on the secretion of inflammatory factors in H/R-exposed cells (P<0.001, Fig. 4D-F).

Accumulating evidence indicates that transcription factors can bind to the promoter of circRNAs and then regulate the effects of circRNAs in human diseases (24,25). To investigate the mechanisms through which circITGB1 is regulated in HK-2 cells, the JASPAR database (http://jaspar.genereg.net) was used to predict potential transcription factors that can bind to the promoter of ITGB1. The predication results indicated four putative GATA1-binding sites upstream of the transcription start site (TSS) (−1,252~−1,242, −1,184~−1,174, −1,144~−1,134 and −913~−903 bp) in the ITGB1 promoter (Fig. 5E). GATA1 is a member of the GATA family of essential transcription factors (26) and modulates gene expression at transcriptional level by binding to specific promoters. Firstly, the present study examined the effect of H/R on GATA1 expression. As shown in Fig. 5A and B, H/R markedly enhanced the GATA1 mRNA and protein expression level in HK-2 cells (P<0.001). To elucidate the effects of GATA1 on the level of circITGB1, GATA1 overexpression vectors were constructed and transfected into HK-2 cells. The successful overexpression of GATA1 in HK-2 cells was confirmed using RT-qPCR (P<0.01, Fig. S1D). In addition, transfecting HK-2 cells with GATA1-specific siRNA leed to a 50% decrease in the GATA1 expression level in the si-GATA1 group compared to that in the si-NC group (P<0.01, Fig. S1G). Subsequently, to verify the transcriptional regulation of GATA1 on ITGB1, the expression of circITGB1, TGB1 mRNA and pre-mRNA was examined in GATA1-overexpressing cells. The results indicated that the exogenous expression of GATA1 increased the expression of circITGB1 (P<0.001, Fig. 5C). Of note, GATA1 overexpression increased the ITGB1 pre-mRNA level (P<0.001), but did not affect the expression of ITGB1 mRNA (Fig. 5C). To confirm the binding of GATA1 to the ITGB1 promoter, the 1.5-kb promoter region and the ITGB1 fragments containing potential binding sites were cloned into a luciferase reporter vector, respectively (Fig. 5D and E). The dual-luciferase reporter assay indicated that GATA1 overexpression increased the ITGB1 promoter activity, whereas GATA1 knockdown decreased ITGB1 promoter activity (P<0.001, Fig. 5F). In addition, the present study further explored which site GATA1 binds to the ITGB1 promoter. The results revealed that the sequences of −1,252~−1,242 and −913~−903 bp upstream of the ITGB1 TSS were essential for the GATA1-mediated transcriptional activation of the ITGB1 gene (P<0.001, Fig. 5G-N). Taken together, these results suggested that ITGB1 is the target gene of transcription factor GATA1.