Determining the influence of high glucose on exosomal lncRNAs, mRNAs, circRNAs and miRNAs derived from human renal tubular epithelial cells

Exosome enrichment and identification

We first extracted exosomes from the culture supernatant of both the high-glucose-induced group and the control group. The characterization of exosomes, including their shape and size, was presented using TEM. The image indicates a round shape of exosomal vesicles and including size ranging from 47 to 154 nm (Figure 1A). We investigated the expression of the exosomal markers CD63, CD9, and CD81 in the exosomes isolated from these two groups and found them expressed in all of our samples (Figure 1B). Next, Exosome size was also identified by NTA, which indicated a mean size of 116.8nm in exosomes isolated from high glucose-induced HK-2 cells and 119.3nm in those isolated from normal HK-2 cells (Figure 1C). Taken together, our results indicated that the vesicles derived from high glucose-induced and normal HK-2 cells in line with the trend of exosome distribution.

Differential expression analysis of exosomal lncRNAs, mRNAs, circRNAs and miRNAs

The whole-transcriptome sequencing data (lncRNA, circRNA, miRNA, mRNA) of five high-glucose induced samples and five control samples were obtained by hierarchical clustering. A total of 169 significantly expressed exo-lncRNAs were identified, including 93 upregulated lncRNAs and 76 downregulated lncRNAs (Figure 2A). The top five upregulated lncRNAs were RP11-178L8.9, CTD-2530H12.2, RP11-503N18.4, RP11-20B24.7 and RP11-256I23.1, and the top five downregulated lncRNAs were RP11-517A5.7, RN7SL870P, CTD-2298J14.2, ANKRD10-IT1 and AP000442.1. In addition, a total of 885 exo-mRNAs were differentially expressed (Figure 2B). Among these exo-mRNAs, 403 were upregulated and 482 were downregulated in the HG groups compared with the control groups. Meanwhile, the top five upregulated mRNAs included CASP7, ZNF766, AL603965.1, ELK3 and MRPL36, and downregulated mRNAs included FASLG, MRPL42, AL137002.1, CLTC and MT-ND6. In the same way, 152 exo-microRNAs were identified as being differential expressed (Figure 2C). Of these exo-microRNAs, 113 were identified as upregulated, and 39 were identified as downregulated. The top five upregulated microRNAs were hsa-miR-6724-5p, hsa-miR-6716-3p, hsa-miR-2355-3p, hsa-miR-135b-3p and hsa-miR-3180. The top 5 downregulated microRNAs were hsa-miR-5008-3p, hsa-miR-6785-5p, hsa-miR-3654, hsa-miR-335-3p and hsa-miR-3074-3p. Finally, only 3 circRNAs were specifically dysregulated, and they were all downregulated in the HG groups, including circRNA_164, circRNA_225 and circRNA_57. Information on these identified circRNAs is presented in Table 1. To further predict the function of differentially expressed exo-circRNAs, we established the circRNA/microRNA interaction based on the Circular RNA Interactome database (https://circinteractome.nia.nih.gov/).

Functional enrichment analysis

GO enrichment analyses were conducted to explore the potential function of differentially expressed exo-lncRNAs, mRNAs, circRNAs and miRNAs. GO classification showed that 2 biological processes (BPs), such as transmembrane transport, 4 molecular functions (MFs), such as transcription factor activity-protein binding, and 17 cellular components (CCs), such as macromolecular complex, were the major functions of these upregulated exo-mRNAs (Figure 3A). Twenty biological processes (BPs), such as cellular metabolic process, 6 molecular functions (MFs), such as methyltransferase activity, and 27 cellular components (CCs), such as cell part, were the major functions of these downregulated exo-mRNAs (Figure 3B). In addition, under the hypothesis that exo-lncRNA function could be associated with the known function of their targeted mRNAs, including colocated and coexpressed mRNAs, lncRNA-targeted mRNAs were further mapped with GO terms. The results show that GO items of upregulated exo-lncRNA-targeted mRNAs (coexpressed) were involved in 1 biological process (BP) called cell junction organization, 3 molecular functions (MFs), such as transporter activity, and 4 cellular components (CCs), such as extracellular matrix (Figure 3C). GO items of downregulated exo-lncRNA-targeted mRNAs (coexpressed) were involved in 30 biological processes (BPs), such as signal transduction and regulation of biological process, 6 molecular functions (MFs), such as pyrophosphatase activity and RNA binding, and 13 cellular components (CCs), such as protein complex (Figure 3D).

Moreover, GO items of upregulated exo-lncRNA-targeted mRNAs (colocated) were involved in 8 biological processes (BPs), such as homeostatic processes, and 2 molecular functions (MFs), such as nucleic acid binding transcription factor activity (Figure 3E). GO items of downregulated exo-lncRNA-targeted mRNAs (colocated) were involved in 10 biological processes (BPs), such as cellular catabolic process and transmembrane transport, 4 molecular functions (MFs), such as translation factor activity and RNA binding, and 5 cellular components (CCs), such as nuclear envelope (Figure 3F).

Functional enrichment analysis based on the differentially expressed exo-miRNAs indicated that the mRNAs targeted by upregulated and downregulated miRNAs were related to 18 biological processes (BPs), such as cell-cell signaling and homeostatic process, 7 molecular functions (MFs), such as DNA binding, and 6 cellular components (CCs), such as the plasma membrane (Figure 3G).

According to the relationship between circRNAs and source genes, GO analyses were also performed to investigate the potential function of differentially expressed exo-circRNAs. The mRNAs producing differentially expressed exo-circRNAs were related to 4 BPs, such as single-organism membrane organization, 2 MFs, such as enzyme binding, and 4 CCs, such as macromolecular complexes (Figure 3H).

To further explore the potential biological role of the identified exo-lncRNAs, mRNAs, circRNAs and miRNAs, we utilized KEGG to provide functional annotations of genes based on their associated biochemical pathways. Enrichment analysis for significantly differentially expressed mRNAs showed that upregulated mRNAs were related to 7 pathways, including insulin secretion, protein digestion and absorption (Figure 4A), and downregulated mRNAs were associated with 15 pathways, including fatty acid metabolism and fatty acid degradation (Figure 4B).

With respect to differentially expressed lncRNA-targeted mRNAs (colocated and coexpressed mRNAs), up- and downregulated lncRNA-targeted mRNAs (coexpressed) were related to such pathways as biotin metabolism, type II diabetes mellitus, amino sugar and nucleotide sugar metabolism, and the GnRH signaling pathway (Figure 4C–4D). Up- and downregulated lncRNA-targeted mRNAs (colocated) were related to such pathways as the chemokine signaling pathway, fat digestion and absorption, N-glycan biosynthesis and HIF-1 signaling pathway. In addition, mRNAs targeted by the up- and downregulated exo-miRNAs were associated with such pathways as maturity onset diabetes of the young, type II diabetes mellitus, and insulin signaling pathway (Figure 4G). Furthermore, mRNAs producing DE exo-circRNAs were associated with such pathways as the Hippo signaling pathway and T cell receptor signaling pathway (Figure 4H).

Most importantly, the defined DE exo-RNAs were frequently enriched in protein export pathway, which can provide strong evidence for the imperative role of exosomes in transporting proteins. These proteins may be secreted by exosomes and participate in the progression of DN occurrence and development.

CeRNA network of DE exo-mRNAs, lncRNAs, circRNAs and miRNAs

To investigate the interaction between the exo-lncRNAs, circRNAs and miRNAs and mRNAs, the circRNA-lncRNA-miRNA-mRNA network was constructed according to the ceRNA hypothesis by integrating expression profile data and their regulatory relationships. We first utilized miRanda and TargetScan software to predict miRNA binding sites of the DE exo-circRNA and exo-lncRNAs. Then, we extracted differentially expressed miRNAs from the intersection of these predicted miRNAs targeted by DE exo-circRNAs and exo-lncRNAs. Following this step, we forecasted DE exo-mRNAs that were potential targets of these DE miRNAs. Finally, a network of DE exo-circRNA-lncRNAs-miRNAs-mRNAs, including a total of 3 circRNAs, 137 lncRNAs, 102 miRNAs and 207 mRNAs, was visualized by Cytoscape (Figure 5).

Functional enrichment analysis indicated that DE lncRNAs/circRNAs-miRNAs-mRNAs with an up-down-up expression pattern were related to BPs, such as oxoacid metabolic process, MFs, such as DNA binding, and CCs, such as the mitochondria (Figure 6A). Additionally, DE lncRNAs-miRNAs-mRNAs with a mode of down-up-down were related to BPs, such as biosynthetic processes, and MFs, such as nucleic acid binding (Figure 6B). DE circRNAs-miRNAs-mRNAs with a mode of down-up-down were associated with BPs, such as cell wall organization or biogenesis, MFs, such as methyltransferase activity, and CCs, such as cytoskeletal part (Figure 6C).

Moreover, KEGG pathway enrichment analysis for lncRNAs/circRNAs-miRNAs-mRNAs with an up-down-up expression pattern indicated that such pathways as the PPAR signaling pathway and protein export (Figure 6D) were enriched. However, lncRNAs-miRNAs-mRNAs with a mode of down-up-down were enriched in such pathways as the TNF signaling pathway and cell adhesion molecules (CAMs) (Figure 6E). Moreover, circRNAs-miRNAs-mRNAs with a mode of down-up-down were enriched in such pathways as the NF-kappa B signaling pathway and fatty acid metabolism (Figure 6F).

Previous study had certificated some molecules could aggravate renal injury in diabetic nephropathy through the TNF-α pathway [16, 17]. In our study, we found that mRNAs in the sub-ceRNA network all enriched in TNF-α signaling pathway, which indicated the exosomal RNAs derived by high glucose-challenged HK-2 cells may contribute to DN progression via activating TNF-α pathway.

Cell culture

The human proximal tubular epithelial cell line HK2 was cultured in DMEM plus F12 (1:1) medium (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% FBS (Gibco, USA) and 1% penicillin (100 U/ml)-streptomycin (100 μg/ml) (Gibco, USA) at 37° C in a humidified atmosphere of 5% CO₂. For glucose stimulation experiments, the cells were cultured in high glucose (HG) medium containing 30 mmol/l D glucose for 48 h at 37° C. For the control group, HK2 cells were cultured in ordinary medium. Fresh medium was supplied every 3 days. Five duplicates of each group were prepared. Culture supernatants of both the high glucose-induced group and the control group were collected for further exosome isolation.

Exosome extraction

Cell supernatant from each sample was collected for exosome enrichment using the ultracentrifugation method. The cell supernatant was centrifuged at 300 g for 10 min to eliminate cell components. The supernatant was further centrifuged at 2,000 g for 10 min to eliminate dead cells, and then we eliminated debris and shed microvesicles by centrifuging at 10000 g for 30 min and at 100000 g for 70 min sequentially. Then, the supernatant was further centrifuged at 100000 g for 70 min to eliminate contaminating proteins. Finally, the pellet was resuspended in PBS and stored at -80° C for further use.

Transmission electron microscopy (TEM)

The exosome suspension was diluted to 0.5 mg/ml suspension with 1X PBS at 4° C. The suspension was then added onto the carbon-coated nickel mesh placed on filter paper. After 30 min of fixing under infrared light, the samples were stained with one drop of phosphotungstic acid (1% aqueous solution) and dried again for 5 min. Finally, the morphology of exosomes was visualized by a transmission electron microscope (HT7700, Hitachi, Tokyo, Japan).

Nanoparticle tracking analysis (NTA)

Exosomes were diluted and analyzed with a NanoSight NS300 instrument according to the manufacturer's instructions. NTA 3.2 software records the particle trajectory and computes the concentration and diameter distribution of all samples. The exosome concentration of the original solution was calculated based on the dilution.

Western blot analysis

Exosomes were further identified by western blotting. We selected blood monocytes as non-exosome controls. First, ice-cold 1X RIPA buffer (Thermo Scientific, Pittsburgh, PA) was utilized to lyse exosome proteins, and the protein concentration of exosomes was quantified using a BCA protein assay kit (Beyotime, China). Protein samples were separated using 8% SDS-PAGE and transferred to PVDF membranes. The PVDF membranes were blocked with 5% BSA for 1 h, washed 3 times (10 min each time) with 1X TBST and probed overnight with primary antibodies against the most common exosomal proteins CD63 (Abcam, USA), CD9 (Abcam, USA), CD81 (Cell Signaling Technology, USA) and non-exosomal protein markers β-Actin (Abcam, USA) at 4° C. Membranes were then washed and incubated with a secondary antibody at room temperature for 1 h. Finally, the membrane was covered with an enhanced chemoluminescence detection kit (GE Healthcare, Piscataway, NJ) to display protein bands.

Exosomal RNA isolation, RNA library preparation and sequencing

Total exosomal RNA from each sample was extracted by TRIzol reagent (Invitrogen, USA) according to the manufacturer’s instructions. The quality and purity of the extracted RNAs were measured using an Agilent 2100 Bioanalyzer (Agilent Technologies, USA). The sequencing library was performed using NEXTflex®Small RNA-Seq Kit v3 (Bio Scientific Corporation, NOVA-5132-05) or Ovation Human FFPE RNA-Seq Library System (NUGEN, 0340-32), and we used a sample input of 20 ng of each total RNA. Finally, we profiled the expression of the sequencing library using HiSeq 2500 (Illumina, Inc., San Diego, CA, USA). Cutadapt [37] software was utilized to remove low-quality reads and acquire high-quality reads. The comparison software Tophat (http://ccb.jhu.edu/software/tophat/index.shtml) was used to map clean reads to a reference genome annotated with a gene location, and Bowtie2 (http://ccb.jhu.edu/software/tophat/index.shtml) was applied to build the index of the reference genome. CircRNA was identified by find_circ [38] and CIRCexplorer [39] through corresponding strict filtering. We used StringTie software to assemble the transcripts and utilized Perl scripts to screen known lncRNAs. Quantitative analysis of lncRNAs and mRNAs was performed using the ballgown R package.

Differential expression analysis

Differential exosomal mRNA, lncRNA, miRNA and circRNA expression analysis of the two groups was conducted by DEGseq [40]. Fold change (FC) was regarded as an indicator of differential expression between the HG-induced group and the control group. T-tests were utilized to evaluate the statistical significance of differences. P-values < 0.05 were considered to demonstrate differential expression.

Gene ontology (GO) and kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis

Gene Ontology (GO) enrichment analysis was conducted according to the target gene candidates of differentially expressed miRNAs, the source genes of differential circRNAs and colocated or coexpressed genes with differential lncRNAs. In addition, we applied the NCBI DAVID server (http://david.abcc.ncifcrf.gov) with default settings [41] to test the statistical enrichment of the target gene candidates in KEGG pathways.

Target gene prediction and competing endogenous RNA (ceRNA) network construction

Detailed circRNA information was downloaded from the circBase database (http://www.circbase.org/). The target mRNAs of miRNAs were theoretically evaluated by TargetScan (http://www.targetscan.org/vert_71/), RNAhybrid (http://bibiserv.techfak.uni-bielefeld.de/rnahybrid) and miRanda (https://regendbase.org/tools/miranda).