HPA-v (cat. no. 7210; ScienCell Research Laboratories, Inc.) were isolated from human visceral fat tissue and cultured in preadipocyte medium (cat.no 7211; ScienCell Research Laboratories, Inc.) containing 5% fetal bovine serum, 100 IU/ml penicillin-streptomycin, and 1% preadipocyte growth supplement (cat.no. 7252; ScienCell Research Laboratories, Inc.). After reaching confluence, the HPA-v were induced to differentiate for 3 days in DMEM containing 0.1 mM 3-isobutyl-1-methylxanthine, 1 µM dexamethasone, and 5 µg/ml insulin. The differentiated HPA-v were then maintained in DMEM containing 5 µg/ml insulin for 6 days.

Cellular lipids were detected using Oil Red O staining. Briefly, upon reaching 100% confluence or differentiation, the preadipocytes were washed three times with phosphate-buffered saline and fixed in 10% formalin for 15 min at room temperature. After fixation, the cells were stained with Oil Red O for 20 min at room temperature. Stained cells were visualized using a Leica DMI 4000 B fluorescent microscope on the white light setting (magnification, ×100).

Total RNA was isolated from 5×10⁶ HPA-v cells and adipocytes, which were differentiated from HPA-v cells, using TRIzol^® reagent (cat. no. 15596018; Invitrogen; Thermo Fisher Scientific, Inc.) and reverse transcribed into cDNA using the HiScript III 1st Strand cDNA Synthesis kit (cat. no. R312-01; Vazyme Biotech Co., Ltd.), according to the manufacturer's protocol. RNA integrity was evaluated by electrophoresis on 2% (w/v) denaturing agarose gels. The concentration and purity of RNA were determined according to the OD₂₆₀/OD₂₈₀ values using the NanoDrop1000 Spectrophotometer (Thermo Fisher Scientific, Inc.).

A human circRNA microarray (Agilent Technologies, Inc.) containing 170,340 human circRNA probes was used. Six samples (three HPA-v and three adipocyte samples) were detected by CapitalBio Corporation using circRNA microarrays. CircRNAs were purified, amplified, labeled with Cy3-dCTP, and hybridized onto the circRNA array according to the manufacturer's protocol. The circRNA expression data were normalized using the GeneSpring GX software version 13.0 (https://www.agilent.com). Differentially expressed circRNAs between HPA-v and adipocytes were selected according to the following thresholds: |fold change| ≥5 and P-value <0.01. Volcano plots were generated to visualize the circRNAs differentially expressed between HPA-v and adipocytes. Hierarchical cluster analysis was used to evaluate differential circRNA expression patterns across the six samples. The parental genes of the differentially expressed circRNAs were analyzed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database within the Database for Annotation, Visualization and Integrated Discovery (https://david.ncifcrf.gov). The homology of circRNAs between human and mice was analyzed using CIRCpedia version 2 software (http://circatlas.biols.ac.cn). The miRNA response elements (MERs) within circRNAs were predicted using MiRanda version 3.3 software (http://www.microrna.org).

The expression levels of peroxisome proliferator-activated receptor gamma 2 (PPARG2), CCAAT enhancer binding protein alpha (CEBPA), fatty acid binding protein 4 (FABP4), hsa_circ_0136134, hsa_circ_0017650, and hsa-circRNA9227-1 were detected by qPCR. Ribosomal protein lateral stalk subunit P0 (RPLP0) was used as an invariant control. qPCR was performed using the ChamQ SYBR^® qPCR Master mix (cat. no. Q311-02; Vazyme Biotech Co., Ltd.), according to the manufacturer's instructions, on an ABI 7300 instrument (ABI; Thermo Fisher Scientific, Inc.). The primers used for qPCR were as follows: PPARG2 forward, 5′-CGGATTGATCTTTTGCTA-3′ and reverse, 5′-CTTTCTGGGTCAATAGGAG-3′; CEBPA forward, 5′-CGTGGAGACGCAGCAGAA-3′ and reverse, 5′-GGCCTTGACCAAGGAGCT-3′; FABP4 forward, 5′-CAGCACCCTCCTGAAAAC-3′ and reverse, 5′-GCAAAGCCCACTCCTACT-3′; RPLP0 forward, 5′-CTCTGCATTCTCGCTTCC-3′ and reverse, 5′-GACTCGTTTGTACCCGTTG-3′; hsa_circ_0136134 forward, 5′-AAGGCACCTGCGGTATTT-3′ and reverse, 5′-AGCCACGGACTCTGCTACT-3′; hsa_circ_0017650 forward, 5′-AAGACCTTCCTCCTTTACCC-3′ and reverse, 5′-GCAACAGTCTGACTTGCCTC-3′; and hsa-circRNA9227-1 forward, 5′-CCGACGCACCATCAGTTT-3′ and reverse, 5′-GAGCGAGGCACAGAAAGG-3′. The thermocycling conditions for the qPCR were as follows: Initial denaturation at 95°C for 30 sec; 45 cycles of denaturation at 95°C for 5 sec; and annealing and extension at 60°C for 30 sec. The relative expression levels of RNA were analyzed with the 2^−ΔΔCq method (17) and normalized to the loading control RPLP0.

The data are presented as the means ± standard deviation. The significance of the differences was analyzed using Student's t-test. P<0.05 was considered to indicate a statistically significant difference.

To obtain mature visceral adipocytes, HPA-v were induced to differentiate in medium containing 0.1 mM 3-isobutyl-1-methylxanthine, 1 µM dexamethasone, and 5 µg/ml insulin. The characteristics of HPA-v differentiation were confirmed by Oil Red O staining and evaluation of adipogenic marker gene expression (Fig. 1). Oil Red O staining revealed substantial lipid deposition in the cytoplasm after differentiation (Fig. 1A). In addition, PPARG2, CEBPA and FABP4 mRNA expression levels were significantly increased in mature visceral adipocytes (Fig. 1B).

To investigate whether circRNAs are associated with lipid deposition, a human circRNA microarray (version 2.0; Agilent Technologies, Inc.) was used to assess circRNA expression profiles in HPA-v and adipocytes. The distribution of circRNA expression was illustrated in a box plot after normalization using the GeneSpring GX software (Fig. 2A), which revealed that the distribution of log₂ ratios was similar among all samples. Volcano plots were generated to compare the circRNA expression profiles between HPA-v and adipocytes (Fig. 2B). The circRNAs differentially expressed between HPA-v and adipocytes were identified as those with a fold change ≥5.0 and a P-value ≤0.01. In total, 2,215 up- and 1,865 downregulated circRNAs were identified in adipocytes compared with HPA-v (Table S1). Table I lists the top 10 up- and downregulated circRNAs, and the homologous circRNAs of hsa_circ_0094183, hsa_circ_0116913 in mice were MMU_CIRCpedia_216382, MMU_CIRCpedia_14213, respectively, while the other 18 circRNAs did not find their homologous circRNAs in mice (Table I). The expression patterns of the differentially expressed circRNAs were visualized by hierarchical cluster analysis (Fig. 2C), which indicated different expression circRNA patterns between visceral adipocytes and HPA-v.

To examine the reliability of the circRNA microarray data, three circRNAs were selected for validation by qPCR. According to the qPCR results, hsa_circ_0136134 was detected in adipocytes exclusively, hsa_circ_0017650 and hsa-circRNA9227-1 were upregulated 30.0- and 2.3-fold in adipocytes compared with HPA-v, respectively (Fig. 2D). The qPCR results were consistent with those of the circRNA microarrays (Fig. 2D and E).

The distribution of the differentially expressed circRNAs on human chromosomes was analyzed, and the that 4,080 differentially expressed circRNAs were derived from genes located on all chromosomes, although rarely on chromosomes 13, 18, 21, and Y (Fig. 3A). When the distribution of these circRNAs was analyzed among the parental genes, it was revealed that 3,968 (97.25%) circRNAs were mapped to 971 parental genes, with 437 (45.01%) parental genes generating one circRNA and 179 (18.43%) parental genes generating more than five circRNAs (Fig. 3B).

To investigate the potential functions of the differently expressed circRNAs, 971 parental genes were analyzed using the KEGG database. The top 15 most significantly enriched pathways, which included fatty acid metabolism, fatty acid degradation, fatty acid biosynthesis, and PPAR signaling pathways are presented in Fig. 4. The most significantly enriched pathway was fatty acid metabolism (P=7.83E-08), and most genes were involved in metabolic pathways (Gene count=104).

To dissect the potential functions of differentially expressed circRNAs, the MERs of top 10 up- and downregulated circRNAs were predicted. Table II lists the miRNAs with more than two MERs targeting the top 10 up- and downregulated circRNAs. hsa_circ_0136134, hsa_circ_0067409, hsa-circRNA9227-1, hsa_circ_0060971, hsa-circRNA9333-2, hsa_circ_0052586, and hsa-circRNA2910-9 potentially interact with 1, 1, 1, 1, 2, 3, and 5 MERs, respectively. The interaction between circRNAs and miRNAs requires further study.