Cell Type-Specific Expression of Long Noncoding RNAs in Human Diabetic Kidneys Identifies TARID as a Key Regulator of Podocyte Function

Abstract

Long noncoding RNAs (lncRNAs) play essential roles in cellular processes, often exhibiting cell type-specific expression and influencing kidney function. While single-cell RNA sequencing (scRNA-seq) has advanced our understanding of cellular specificity, past studies focus solely on protein-coding genes. We hypothesize that lncRNAs, due to their cell-specific nature, have crucial functions within particular renal cells and thereby play essential roles in renal cell function and disease. Using single-nucleus RNA-seq (snRNA-seq) data from kidney samples of five healthy individuals and six patients with diabetic kidney disease (DKD), we explored the noncoding transcriptome. Cell type-specific lncRNAs were identified, and their differential expression in DKD was assessed. Integrative analyses included expression quantitative trait loci (eQTL), genome-wide association studies (GWAS) for estimated glomerular filtration rate (eGFR), and gene regulatory networks. Functional studies focused on TCF21 antisense RNA inducing promoter demethylation (TARID), a lncRNA with podocyte-specific expression, to elucidate its role in podocyte health. We identified 174 lncRNAs with cell type-specific expression across kidney cell types. Of these, 54 lncRNAs were differentially expressed in DKD. Integrative analyses, including eQTL data, GWAS results for eGFR, and gene regulatory networks, pinpointed TARID, a podocyte-specific lncRNA, as a key candidate upregulated in DKD. Functional studies confirmed TARID's podocyte-specific expression and revealed its central role in actin cytoskeleton reorganization. Our study provides a comprehensive resource of single-cell lncRNA expression in the human kidney and highlights the importance of cell type-specific lncRNAs in kidney function and disease. Specifically, we demonstrate the functional relevance of TARID in podocyte health.

Article highlights: This study provides a resource for kidney (cell type-specific) long noncoding (lnc)RNA expression and demonstrates the importance of lncRNAs in renal health. We identified 174 cell type-specific lncRNAs in the human kidney, with 54 showing altered expression in diabetic kidney disease. TCF21 antisense RNA inducing promoter demethylation (TARID), a podocyte-specific lncRNA upregulated in diabetic kidney disease, is crucial for actin cytoskeleton reorganization in podocytes.

Graphical abstract

Figure 1

snRNA-seq of human kidney reveals cell type–specific lncRNAs. A: UMAP visualization of snRNA-seq data, highlighting overall cellular clustering. Cell types: proximal tubule (PT), VCAM1⁺ proximal tubule (PT_VCAM1), parietal epithelial cells (PEC), ascending thin limb (ATL), CLDN16⁻ thick ascending limb (TAL1), CLDN16⁺ thick ascending limb (TAL2), early distal convoluted tubule (DCT1), late distal convoluted tubule (DCT2), principal cells (PC), type A intercalated cells (ICA), type B intercalated cells (ICB), podocytes (PODO), endothelial cells (ENDO), mesangial cells and vascular smooth muscle cells (MES), fibroblasts (FIB), and leukocytes (LEUK). B: UMAP projection illustrating the distribution of lncRNAs within the snRNA-seq data set. C: Distribution of lncRNAs across different cell types. D: Identification of 349 cell type–specific lncRNAs, with the top 5 lncRNAs for each cell type displayed. E: Analysis of cell type–specific lncRNAs and their correlation with known protein-coding marker genes. Left: Violin plots showing expression patterns of representative lncRNAs in specific cell types. Right: Heat map depicting the correlation between these lncRNAs and cell type–specific protein-coding marker genes, with color intensity indicating the degree of correlation. F: Heat map of 190 lncRNAs uniquely associated with DKD in specific kidney cell types, identified from a total of 405 DKD-associated lncRNAs. Each red mark indicates a significant association between a lncRNA and DKD in that particular cell type. Several lncRNAs show significant associations in multiple cell types.

Figure 2

Cell type–specific and DKD-associated lncRNAs overlap with eGFR-associated loci. A: Dot plot showing 54 lncRNAs that are both cell type–specific and associated with DKD. B: Violin plot of four lncRNAs that are cell type–specific and DKD associated: PCAT19 in endothelial cells, AL355612.1 in early distal convoluted tubule cells, TARID in podocytes, and AC092813.2 in podocytes. C: Among 349 independent SNPs significantly associated with eGFR, only 1 lncRNA, TARID, was located within the 1-megabase region. The SNPs rs74379084-C and rs17643040-C were found in this region, with rs17643040-C identified as a cis-eQTL in multiple tissues, including the kidney cortex, for TARID. NES, normalized enrichment score.

Figure 3

TARID is a podocyte-specific lncRNA. A: UMAP projection showing TARID expression across kidney cell types. Expression of TARID in healthy kidneys and in kidneys of DKD patients analyzed by snRNA-seq (B), by qPCR analysis of additional RNA samples (n = 32) from glomerular kidney biopsy specimens from patients with DKD and healthy control participants (C) and correlation with Sirius red (D). E: In situ hybridization of TARID in human FFPE kidney tissue. The glomeruli is displayed within the dotted line, bar = 50 μm. Left and right display different glomeruli on different slides. DAPI was used to stain for nucleus. F: In situ hybridization of TARID and WT1 and their colocalization in human FFPE kidney tissue. The glomeruli is displayed within the dotted line, bar = 50 μm. DAPI was used to stain for nucleus.

Figure 4

Gene regulatory network of TARID. A: Correlation analysis of PARD3B and TARID in podocytes. B: Circos plot showing correlation coefficient (>0.5) for TARID with genes that are linked to podocyte dysfunction (red) and not previously linked to podocyte dysfunction (blue). Pathway enrichment analysis (genes = 101, Pearson correlation coefficient >0.4) with Reactome Pathway Analysis (PA) (C) or KEGG (D). Color of dots indicates P value, and size of dots indicates gene count. The x-axis shows the BgRatio (ratio of the number of genes within that distribution that are annotated to the total number of genes in the background distribution).

Figure 5

Bulk RNA-seq analysis of podocytes following TARID knockdown. A: RNA-seq analysis of TARID expression in CIHP-1 cells upon transfection with scrambled Gapmer and TARID Gapmer (design 2). CPM, counts per million. B: Volcano plot displaying differentially expressed genes in CIHP-1 cells following TARID knockdown. Blue dots indicate downregulated genes (n = 171), red dots indicate upregulated genes (n = 87), and gray dots indicate not significantly changed genes. Gene names of the top 20 associated genes and TARID are displayed in the volcano plot. C: Pathway enrichment analysis results of significantly differentially expressed genes. Left panel indicates activated pathways. Right panel indicates suppressed pathways. Color of dots indicates P value, and size of dots indicates GeneRatio. The x-axis shows the count.

Figure 6

Actin staining analysis of podocytes following TARID knockdown. A: Microscopic images of podocytes (CIHP-1) transfected with scrambled Gapmer (left) and TARID Gapmer (right) stained for actin and DAPI, bar = 500 pixels = 180 μm. B: Quantification of total length of the actin filaments in podocytes (CIHP-1) treated with scrambled Gapmer (n = 4) and TARID Gapmer (n = 4). C: Quantification of total number of actin filaments in podocytes (CIHP-1) treated with scrambled Gapmer (NC, normal control) (n = 4) and TARID Gapmer (n = 4).

Figure 7

Actin cytoskeleton analysis of podocytes following TARID overexpression. A: Podocytes (CIHP-1) were transduced with a lentiviral vector overexpressing TARID or a control vector. Gene expression was assessed by qPCR. TARID expression was robustly increased in TARID-transduced cells compared with control (P < 0.0001). The expression of the podocyte marker WT1 remained unchanged. Among the potentially downstream target genes analyzed (TCF21, CDC42BPG, ASB2, and NEDD9), only CDC42BPG showed a significant upregulation (P = 0.0007). Expression of TCF21, ASB2, and NEDD9 was not significantly affected. Data are presented as mean ± SD (n = 8 per group). Statistical analysis was performed using unpaired two-tailed Student t tests. ns, not significant. B: Microscopic images of CIHP-1 podocytes transduced with control vector (left) or TARID-overexpressing vector (right), stained for actin and nuclei (DAPI); scale bar = 500 pixels = 180 μm. Quantification of the total actin filament length (left) and total number of actin filaments (right), n = 4 per group.