RNA-sequencing reveals genome-wide long non-coding RNAs profiling associated with early development of diabetic nephropathy

Figure 1. Workflow and phenotypes of animal models (A) Workflow of experiments. a. Construction of animal models; b. examination of phenotype; c. PTC isolation; d. libraries for preparation of RNA-sequencing; e. next-generation sequencing; f. bio-information analysis. (B) Fasting blood sugar over time. Data are represented as mean ±SEM. *p < 0.05 by t-test versus control at 0 week. #p < 0.05 by t-test versus control at the same week. n = 6 for each group. (C) ACR over time. Data are represented as mean ±SEM. *p < 0.05 by t-test versus control at 0 week. #p < 0.05 by t-test versus control at the same week. n = 6 for each group. (D) Left: PAS-stained sections in each group. Images display kidney sections of each group at 200× magnification. White arrow indicates normal tubule, and black arrow indicates injured tubule. Views with high magnification are shown at lower left. Time course are labeled as a, b, c, and d; letters correspond to 0, 2, 4, and 8 weeks. Right: Tubule injuries were evaluated for widened lumen, atrophy, or thickened basement membranes. Tubular injuries are indicated with arrows. Bar graph shows TDI for each group. Data are represented as mean ±SEM. *p < 0.05 by t-test…

Figure 2. Transcriptome profiles of mRNAs associated with disease progression (A) Number of differentially expressed mRNA genes at various time points compared with control (W0). P-value cutoff is 0.05. (B) Venn diagram of differentially expressed mRNA genes at different times. (C) Overlapping between differentially expressed mRNA genes and reported mRNA genes associated with kidney disease progression. (D) Gene ontology enrichment for week-specific and common differentially expressed mRNA genes, presented as -log10 hypergeometric P-value for enrichment. (E) Gene expression of some key genes that are highly related to kidney. y-axis denotes FPKM value, x-axis is for four time points, and error bar is standard deviation, * indicates p-value < 0.05 by comparison with W0. (F) Validation for differentially expressed mRNAs in human PTCs (HK2)Left portion: Results from RNA-seq in mouse DN models (*p<0.05 versus W0). Right portion: Results from qPCR in vitro diabetic models. NG, cultured with normal glucose; HG, cultured with 30 mM glucose for 48 h.*p<0.05 versus NG, n=3.

Figure 3. Transcriptome profiles of lncRNA genes associated with disease progression (A) Hierarchical clustering of lncRNA genes that are differentially expressed in at least one time point by comparison with that at W0; rows and columns correspond to samples and lncRNA genes, respectively. (B) Number of differentially expressed lncRNA genes at various time points compared with control (W0) with corrected P-value cutoff of 0.05. (C) Venn diagram of differentially expressed lncRNA genes at different times. (D) Gene ontology enrichment for week-specific and common differentially expressed lncRNA genes, presented as −log10 hypergeometric P-value for enrichment. (E) Validation for differentially expressed lncRNAs in PTCs (HK2)Left portion: Results from RNA-seq in mouse DN models (*p<0.05 versus W0). Right portion: Results from qPCR in vitro diabetic models. NG, cultured with normal glucose; HG, cultured with 30 mM glucose for 48 h. *p<0.05 versus NG, n=3. (F) Motifs searched for validated lncRNAs in PTCs (HK2). TF, transcriptional factor. Sites, number of possible motifs in one LncRNA sequence.

Figure 4. TSSs of differentially expressed lncRNA genes are enriched by active chromatin signal (A) ChiP-Seq analysis of histone-modification profiles (H3K4me1 and H3K4me3) at TSSs of lncRNA genes, which are differentially expressed in at least one time point. (B) Density functions for lncRNA genes of H3K4me1 and H3K4me3. (C) ChIP-Seq “metaplots” of histone-mark density at TSSs of protein-coding and lncRNA genes, which are stratified by gene expressions: differentially expressed genes, non-expressed genes, week-specific differentially expressed genes, expressed but not differentially expressed genes, and random genes.

Figure 5. Disease progression is characterized by global expression patterns of lncRNA genes rather than mRNA genes Sample clustering analysis based on gene expression of various combination of protein-coding and lncRNA genes, where dashed line is cut line for dendrogram, and samples surrounded by dashed circles are misclassified. (A) Protein-coding and lncRNA genes. (B) Differentially expressed protein-coding and lncRNA genes in at least one time point. (C) Differentially expressed protein-coding genes. (D) Differentially expressed lncRNA genes.

Figure 6. lncRNA genes are co-expressed with protein-coding genes during disease progression (A) Boxplot of pairwise Spearman expression correlations between genes of different classes in trans (left) and cis (right): lncRNA–to–lncRNA, lncRNA–mRNA, mRNA–mRNA, and random mRNA–lncRNA. (B) Gene ontology enrichment for protein-coding genes in trans and correlated with lncRNAs (first column, mRNA–lncRNA), and nearest protein-coding genes in trans and correlated with lncRNA genes (lncRNA–lncRNA) presented as −log10 hypergeometric P value for enrichment. (C) Model-based expression profiles (Profiles 1–22; Groups I–VI) of protein-coding and lncRNA genes expressed differentially during disease progression. Numbers above plots indicate profile number plus total genes and lncRNA genes in profile (in parentheses: total lncRNA). y-axis is Z-score of gene expression, and x-axis is time point. In each plot, gray line is expression level of gene in module, and red line is average expression level of genes within module. Red asterisks indicate profiles showing enrichment for lncRNA genes (P < 0.05 versus proportion of lncRNAs among all differentially expressed genes). (D) Validation for predicted…

Figure 7. Network analysis identifies co-expression modules of lncRNA and protein-coding genes (A) Heatmap of WGCNA topological overlap matrix, where each row (column) corresponds to a gene, and bar color indicates genes clustered as module. (B) Eigengene of selected modules represents gene expression level of genes within a module. (C-D) Screening strategies for identification of candidate lncRNA genes and lncRNA-gene–protein-coding gene co-expression associations in specific modules for further functional studies presented as Circos plots from analysis of HSPC module (broken lines in circumference indicate individual chromosomes).