Cell Type- and Age-Specific Expression of lncRNAs across Kidney Cell Types

Abstract

Key Points:

1. We constructed a single-cell long noncoding RNA atlas of various tissues, including normal and aged kidneys.

2. We identified age- and cell type–specific expression changes of long noncoding RNAs in kidney cells.

Background: Accumulated evidence demonstrates that long noncoding RNAs (lncRNAs) regulate cell differentiation and homeostasis, influencing kidney aging and disease. Despite their versatility, the function of lncRNA remains poorly understood because of the lack of a reference map of lncRNA transcriptome in various cell types.

Methods: In this study, we used a targeted single-cell RNA sequencing method to enrich and characterize lncRNAs in individual cells. We applied this method to various mouse tissues, including normal and aged kidneys.

Results: Through tissue-specific clustering analysis, we identified cell type–specific lncRNAs that showed a high correlation with known cell-type marker genes. Furthermore, we constructed gene regulatory networks to explore the functional roles of differentially expressed lncRNAs in each cell type. In the kidney, we observed dynamic expression changes of lncRNAs during aging, with specific changes in glomerular cells. These cell type– and age-specific expression patterns of lncRNAs suggest that lncRNAs may have a potential role in regulating cellular processes, such as immune response and energy metabolism, during kidney aging.

Conclusions: Our study sheds light on the comprehensive landscape of lncRNA expression and function and provides a valuable resource for future analysis of lncRNAs (https://gist-fgl.github.io/sc-lncrna-atlas/).

Graphical abstract

Figure 1

Targeted scRNA-seq enabled the detection of cell type–specific lncRNAs in six tissues. (A) Experimental procedure overview of the scRNA-seq and lncRNA-targeted sequencing were performed on six mouse tissues and aged kidneys enriched glomerulus and tubules. (B) The UMAP plot displays 32 cell clusters, with assigned cell types labeled below each cluster. Tissue-specific cell types are indicated by red circles. (C) The bubble plot illustrates differentially expressed lncRNAs in each cell type. The color scale represents expression level, and dot size indicates percentage of cells expressing each lncRNA. (D) The cellular connections between the clusters in (B) are shown on the basis of partition-based graph abstraction (PAGA) projection. Left: PAGA analysis using only mRNAs. Right: PAGA analysis using only lncRNAs. Each number corresponds to the cell type number in (B). (E) The UMAP plot represents the 32 cell clusters using only lncRNAs. The experimental workflow graphic was created using BioRender. AM, alveolar macrophages; AT1, alveolar type 1 cells; AT2, alveolar type 2 cells; DN, double-negative T cells; DP, double-positive T cells; GEC, glomerular endothelial cells; ISC, intestine stem cells; JGC, juxtaglomerular cells; LE, lymphatic endothelial; lncRNA, long noncoding RNA; macro/mono, macrophage/monocyte; neutro/basophil, neutrophil/basophil; NK, natural killer; PT, proximal tubule cells; RBC, red blood cells; sc library, single-cell library; scRNA-seq, single-cell RNA-sequencing; UMAP, uniform manifold approximation and projection.

Figure 2

Cell type–specific lncRNA expression and correlation with canonical markers in the kidney cells. (A) The UMAP plot represents kidney samples after merging the pre- and postcapture data. Each cell type is represented by a different color, and the corresponding labels are provided on the UMAP plot. (B) The bubble plot displays lncRNAs that are differentially expressed in each cluster. The x axis represents cell types, and the y axis indicates lncRNAs. The color represents expression level. Dot size represents the percentage of cells expressing each lncRNA. (C) ISH shows double staining of well-known marker genes (Umod, Pvalb) and lncRNAs (Gm17750, Pantr1) in the ascending loop of Henle (left) and DCT (right). (D) Differentially expressed lncRNAs and correlation with marker genes. Left: Violin plots depict representative lncRNAs in each cell type. Right: Heatmap displays the correlation between representative lncRNAs and marker genes. The color indicates the level of correlation. ALOH, ascending loop of Henle; CD IC, collecting duct intercalated cells; CD PC, collecting duct principal cells; DCT, distal convoluted tubule; ISH, in situ hybridization; PEC, parietal epithelial cells; PST, proximal straight tubule.

Figure 3

Gene regulatory network analysis of lncRNAs and transcription factors in the kidney. (A) Motif ranking by P value in kidney proximal tubule cells. (B) Identification of enriched motifs in open chromatin regions of kidney cells. The enrichment fold change was normalized to z-score, indicated by the color. (C) Expression levels of differentially expressed transcription factors in proximal tubule cells. Expression of Hnf4a and Ppara in kidney cell clusters. (D) Genome browser screenshots showing binding patterns of PPARA (GSE113157) and HNF4A (GSE144824) near 0610005C13Rik and Hnf4aos in kidney cells. (E) Gene regulatory networks between lncRNAs, transcription factors, and highly correlated genes in the kidney. Each network is represented by a different color. The biological functions associated with each network are shown below the corresponding gene regulatory network. (F) Highlighting of a gene regulatory network involved in kidney tubule functions (orange color network). The sky blue line indicates coexpression between lncRNAs and transcription factors. The blue line indicates the target lncRNAs of transcription factors. The red line indicates the target lncRNA of transcription factors along with coexpression with transcription factors. Red, sky blue, yellow, and white dots represent transcription factors, lncRNAs, downregulated genes in the 0610005C13Rik KD cells, and other interacting genes, respectively. (G) Downregulated (blue) and upregulated (red) genes after the 0610005C13Rik knockdown using siRNA. “No sig” indicates no significant values. The x axis indicates –log₂(fold change). The y axis indicates –log₁₀(P value). (H) Gene ontology enrichment analysis of downregulated genes after 0610005C13Rik knockdown. The x axis indicates –log₁₀(P value). ChIP-seq, chromatin immunoprecipitation sequencing; DAR, differential accessibility regions; ESRRA, estrogen-related receptor alpha; GRN, gene regulatory network; KD, knockdown; PPARA, peroxisome proliferator–activated receptor alpha; siRNA, small interfering RNA.

Figure 4

Exploring age- and cell type–specific expression of lncRNAs in glomerular cells. (A) UMAP visualization of 15 cell type clusters. (B) Changes in cell proportions with age and samples of origin. Upper panel shows the proportion change between young and old; lower panel shows proportion change between the glomerulus and the tubule. (C) Violin plot showing senescence and inflammatory scores in specific cell types from old samples. Each color represents a different cell type. (D) Violin plot showing senescence and inflammatory scores in four glomerular cell types (podocyte, mesangial cell, JGCs, and glomerular endothelial cells) within the glomerulus region. (E) Volcano plot representing age-specifically expressed lncRNAs between young and old in all cell types. Blue and red dots indicate downregulation and upregulation of lncRNA expression in old samples, respectively. X axis indicates log₂(fold change), and y axis indicates –log₁₀(P value). The “n” number represents the number of significantly increased and decreased lncRNAs in old samples. (F) Bubble plot showing age- and cell type–specific lncRNAs in each cell type. X axis indicates cell type, and the y axis indicates lncRNAs. Color represents average log₂(fold change). Dot size represents expression percentage. Blue and red dots represent genes decreased and increased in old samples, respectively. The gray dots indicate nonsignificant genes. (G) Bar plot showing fold change variation of lncRNA that is shared among various cell types and age-specifically expressed.

Figure 5

Characterization of age-specific gene expression and lncRNA dynamics in kidney aging. (A) Heatmap showing age-specific gene expression in mesangial cells. The lncRNAs on the left side of the heatmap indicate the cell type– and age-specific lncRNAs. Biological functions are indicated on the right. (B) Expression comparison between young and old samples of mesangial cell–specific and age-specific lncRNAs. (C) Violin plot indicating expression levels of each lncRNA according to high and low inflammatory scores in mesangial cells. (D) Line plot showing the correlation with Gm12840 in the mesangial cells. Genes with high correlation are displayed beside the line. X axis indicates each gene, and y axis indicates correlation value. (E) WGCNA module variation presenting significant differences (P value < 0.05) between young and old samples in mesangial cells. The lower panel indicates biological functions and representative genes of the turquoise and blue modules. (F and G) Heatmap showing age-specific gene expression in JGC (F) and glomerular endothelial cells (G). The left side of the heatmap indicates lncRNAs specifically expressed in the cell type and age. Biological functions are indicated below. (H) UMAP visualization of subclutering the immune cells. The cell type is labeled on the UMAP plot. (I) Bubble plot showing age- and cell type–specific lncRNAs in each cell type. X axis indicates cell type, and the y axis indicates lncRNAs. Color represents average log₂(fold change). Dot size represents expression percentage. Blue and red dots indicate genes increased in young and old, respectively. The gray dots indicate nonsignificant genes. (J) Bar plot showing fold change variation of AW112010 that are shared among various cell types and are age-specifically expressed. (K) Comparison of total read differences of AW112010 in all cells between young and old and between glomerulus and tubule-enriched samples. cDC, conventional dendritic cell; CM, classical monocyte; NCM, nonclassical monocyte; pDC, plasmacytoid dendritic cells; WGCNA, weighted gene coexpression network analysis.