doi: 10.1038/s41586-020-2941-1.
Epub 2020 Nov 11.
Decoding myofibroblast origins in human kidney fibrosis
Affiliations
- PMID: 33176333
- PMCID: PMC7611626
- DOI: 10.1038/s41586-020-2941-1
Item in Clipboard
Decoding myofibroblast origins in human kidney fibrosis
Nature.
2021 Jan.
Abstract
Kidney fibrosis is the hallmark of chronic kidney disease progression; however, at present no antifibrotic therapies exist1-3. The origin, functional heterogeneity and regulation of scar-forming cells that occur during human kidney fibrosis remain poorly understood1,2,4. Here, using single-cell RNA sequencing, we profiled the transcriptomes of cells from the proximal and non-proximal tubules of healthy and fibrotic human kidneys to map the entire human kidney. This analysis enabled us to map all matrix-producing cells at high resolution, and to identify distinct subpopulations of pericytes and fibroblasts as the main cellular sources of scar-forming myofibroblasts during human kidney fibrosis. We used genetic fate-tracing, time-course single-cell RNA sequencing and ATAC-seq (assay for transposase-accessible chromatin using sequencing) experiments in mice, and spatial transcriptomics in human kidney fibrosis, to shed light on the cellular origins and differentiation of human kidney myofibroblasts and their precursors at high resolution. Finally, we used this strategy to detect potential therapeutic targets, and identified NKD2 as a myofibroblast-specific target in human kidney fibrosis.
Conflict of interest statement
The authors have no competing interests.
Figures
a. A schematic of human nephrectomy kidneys. Kidney samples were sampled from the tumor-free kidney cortex distant from the tumor region. b. A schematic of the whole kidney sorting strategy. Single cell 10x Genomics RNA-Seq libraries were prepared from CD10 negative, living (DAPI-) and CD10 positive, living (DAPI-) cells separately. CD10 negative cells are enriched for mesenchymal cells. c. Immunofluorescence staining of CD10, LTA, CD45 and WT1. CD10 expression labels proximal tubule epithelial cells. d. Representative flow cytometric plots from the sorting and gating strategy described in c enriching for CD10- cells. e. Relationship between serum creatinine and age in patients included in the scRNA-seq experiments of Fig. 1. f. Relationship between serum creatinine and degree of interstitial fibrosis scored by a blinded nephropathologist for the same patients in e. g. Relationship between serum creatinine and tubular atrophy as scored by a blinded nephropathologist for the same patients in e and f. h. Representative images of PAS stained (left) and Collagen 3 immunostained kidneys of patients with eGFR>60 ml/min/1.73 m2 body surface area. i. Same as h but patient with eGFR<60 ml/min/1.73 m2 body surface area. j. Each patient visualized in the UMAP of Fig. 1b. k. The main 5 cell types found in the CD10- fraction, illustrated on the same UMAP embedding from Figure 1b. l-r. Expression of select marker genes visualized on the same UMAP embedding. s. Each cell state/type visualized in the UMAP of Fig.1b. t. Doublet Score (see Methods) for human CD10+ cells. u. Doublet Score (see Methods) for human CD10- cells. Scale bars: in c1-c3 50 μm, in c4-c6 30 μm, in h-i 75 μm.
a. Scaled gene expression of marker genes in mesenchymal cell clusters of the CD10- data depicted in Fig. 1b-e. Each 100 cells are averaged in one column. b. Same as a. but endothelial clusters c. Same as a. but immune clusters. d. Same as a. but epithelial clusters. e. The 7 cell clusters found in the CD10+ fraction cells visualized on the UMAP embedding from Fig. 1e. f. UMAP embedding of e. with colors representing the cell types/states. g. Scaled ECM score on UMAP from e. h. 8 patients visualized on UMAP embedding as e. i. Expression of RBP4 visualized on UMAP embedding from e. j. Scaled gene expression of the top 20 genes by specificity of each of the 7 cell clusters discovered in the CD10+ data depicted in e-h. k. Log fold change of cell cycle stage assignment frequencies in healthy and CKD epithelial cells relative to permuted frequencies. Positive numbers represent enrichment, negative numbers represent depletion. l. Percentage of cells per cell cluster in each cell cycle phase as predicted from gene expression data. m-n. KEGG and GEO Process terms enriched in cells belonging to healthy or CKD patients, according to differentially expressed genes between healthy and diseased patients (see Fig. 1f). Note Fatty Acid Catabolic Process and Lipid Oxidation consistent with KEGG pathway enrichment results o. ECM, collagen, proteoglycan and glycoprotein score of human diabetic kidney dataset (Fan Y. et. al. Diabetes 2019). Advanced DN (diabetic nephropathy) n=21, early DN n=6, control n=9. p. Distribution of single cell ECM scores for all cells in the CD10- cell fraction, colors indicate cell groups obtained by unsupervised mixture model clustering of ECM scores. q. ECM groups visualized on the UMAP of Fig. 1b. r. The same ECM scores as in p. but scaled and visualized on the UMAP embedding from Fig. 1b. s.-u. Scaled expression of gene groups summarized in the ECM score including collagens (r), glycoproteins (s) and proteoglycans (t). All 50 cell clusters are shown, all cells from each cluster are averaged in one column.
a. UMAP embedding of Fibroblast/Pericyte/Myofibroblast cells from 13 human kidneys (n=2,689). Colors represent the cell types. Lines refer to a lineage trajectory predicted by slingshot (see Methods). b. Expression of selected genes on the embedding of a. c. Gene Ontology Biological Process analysis for Pericyte (Pe), Myofibroblast (MF), Fibroblast cell clusters (Fib) and vascular smooth muscle cells (VSMCs) based on the top marker genes for each cluster (CD10- data, see Methods). d. ECM score and scaled expression of select genes visualized on the Mesenchymal cell Diffusion Map embedding of Figure 1o. e.-h. The distribution of ECM score, collagen score, glycoprotein score and proteoglycan scores stratified by epithelial cell clusters in the CD10-
-cell fraction. i. Scaled expression of select genes in proximal tubules and injured proximal tubule cell clusters. Each 100 cells are averaged in one column. j. Gene Ontology Biological Process analysis based on differential expression between proximal tubules and injured proximal tubules. k.-n. The distribution of ECM score, collagen score, glycoprotein score and proteoglycan scores for epithelial cells (CD10+ cell fraction) o. Percentage of cells expressing PDGFRb and Col1a1 in each main cell niche. Neuronal Schwann cells were excluded since they are represented by a small number of cells.
a. Patient samples (n=8) visualized on the UMAP from Figure 2a. Different cell clusters are indicated by different colors. Stratification of single cells according to patient clinical parameters (CKD=chronic kidney disease, eGFR=estimated glomerular filtration rate). b-c. Expression of select genes on the same UMAP embedding from a. d. Scaled gene expression of the top 10 genes in each cell type/state cluster. Gene ranking per cell cluster was determined by genesorteR. e. Correlation between cell clusters identified in CD10- data (Figure 1, columns) and PDGFRb+ data (Figure 2, rows). f. ECM score stratified by 4 main cell types in PDGFRb+ data. g. ECM score stratified by main mesenchymal cell types. h. ECM score stratified by 5 epithelial cells clusters. i. ECM score visualized on the UMAP embedding from a. j. Doublet Score (see Methods) for human PDGFRb+ cells. k. Representative image of combined immunofluorescent and multiplex RNA in-situ hybridization of LTA (proximal tubular marker), Col1a1 and PDGFRb+. Note Col1a1 and PDGFRb expression in LTA+ tubular cells (j’ arrows). l. Representative image of combined immunofluorescent and multiplex RNA in-situ hybridization of CD68 (macrophage marker), Col1a1 and PDGFRb. m. Representative image of multiplex RNA in-situ hybridization of Pecam1, Col1a1 and PDGFRb. Scale bars k-m 50 μm.
a. The mesenchymal cell clusters in Figure 2 here indicated on the Diffusion Map embedding from Figure 2c (left) and stratified by eGFR class (right) and the expression of selected genes on the same embedding. b. UMAP embedding of mesenchymal cell populations from Fig. 2a. Colors represent the cell types/states shown in Fig 2a. c. ECM score visualized on the UMAP in b. UMAP embedding indicates distinct and separate pericyte and fibroblast origins for myofibroblasts, consistent with Diffusion Map embedding of the same cells (Figure 2e) d. 3 main pericyte and (myo-) fibroblast cell types indicated on the same UMAP embedding. e. Pseudotime as predicted by the Slingshot algorithm on the same UMAP embedding from b. f.-g. Col1a1 and Notch3 expression on the UMAP embedding from b. h-i. Violin plots across mesenchymal cells types of Col1a1 and Postn of human PDGFRb+ dataset in Figure 2. j. Quantification of Meg- (Notch3/Postn-) cells in human kidneys (n=35) (see patient data Extended data table 2). n=17 (healthy), 10 (early) and 8(late); *p<0.05, **p < 0.01 by 1-way ANOVA followed by Bonferroni’s correction. Tukey box whisker plot. k-m. Representative image of multiplex RNA in-situ hybridization of Meg3, Notch3 and Postn. Note that triple positive cells (arrow with tails) or double positive cells (Notch3+Postn+, l magnification 2 arrow heads) can be detected in the kidney interstitium. n. Immunofluorescence staining of Cxcl12(SDF-1). Note expression in the kidney interstitium in PDGFRb+ cells (arrow with tails) and LTA- tubular cells (arrows). o.-q. Representative image of multiplex RNA in-situ hybridization of Ccl19, Ccl21 and PDGFRb. r. Quantification of Ccl19/Ccl21+ cells in human kidneys (n=35) (see Patient data Extended data table 2). n=17 (healthy), 10 (early) and 8(late); ***P < 0.001, ****P <0.0001 by 1-way ANOVA followed by Bonferroni’ post-hoc test. Tukey box whisker plot. For details on statistics and reproducibility, please see Methods.
a. KEGG pathway enrichment analysis along pseudotime for lineage 1 (see Figure 2c.) b. Top: Gene expression dynamics along pseudotime for lineage 2 (Fibroblasts to Myofibroblasts, see Figure 2c.). Cells (in columns) were ordered along pseudotime and genes (in rows) that correlate with pseudotime were selected and plotted along pseudotime (see Methods). Each 10 cells were averaged in one column. Genes were grouped signifying their pseudotime expression pattern. Selected example genes for each group are indicated. See Supplementary File 3 for gene cluster assignments. Bottom: Cell cycle stage along pseudotime as percent of each 500 cells along pseudotime. c. Same as in b. but for lineage 3 cells (see Figure 2c) d. PID signaling pathway enrichment analysis along pseudotime for lineage 2 cells ordered along pseudotime as in b. e. KEGG pathway enrichment analysis along pseudotime for lineage 2. f. Same as in d. but for lineage 3 cells (see Figure 2c). g. Same as in e. but for lineage 3 cells. h.-k. Violin plots across mesenchymal cells types of selected genes of the human PDGFRb+ dataset in Figure 2. l. TF scores for proximal promoter regions (l) and distal regions (m) obtained by TF sequence motif enrichment analysis for top marker genes for the mesenchymal cell clusters of the human PDGFRb+ dataset (see Methods). Note enrichment of Fos and Jun motifs in promoters of fibroblast marker genes. m. Schematic of human kidney PDGFRb+ cell generation and immortalization. n. Cell proliferation (WST-1) and expression of cFos, Col1a1, Postn and Ogn by RNA qPCR after AP-1 inhibitor treatment (T-5224) and/or TGFb treatment of immortalized human PDGFRb kidney cells. n=3 per group. *P < 0.05, **P<0.01, ***P < 0.001, ****P <0.0001 by 1-way ANOVA followed by Bonferroni’ post-hoc test. Mean± S.D. o. Expression of Ogn (Fib1+3) and Postn (MF1) visualized on the same UMAP embedding from Extended data Fig. 5b. p. AP-1 average TF expression (left) and average expression of putative AP-1-regulated genes (right) against Collagen scores stratified by fibroblast and myofibroblast cells. Interestingly, the expression of AP-1 anti-correlates with collagen score but the expression of its target genes positively correlates with collagen score, potentially pointing towards an inhibitory role for AP-1. q. The number of statistically significant receptor-ligand interactions between mesenchymal cells and all other cell types (CD10- fraction, Figure 1) according to CellphoneDB Analysis. Dendritic cells, monocytes, myofibroblasts, podocytes, arteriolar endothelial cells and injured tubules as major sources of signaling ligands to pericytes fibroblasts and myofibroblasts. r. Dot plot for significant ligand-receptor interactions from the selected signaling pathways EGFR, PDGF, WNT, TGFb, Notch and Hedgehog for pericytes, fibroblast and myofibroblasts. Interacting ligand-receptor and cell types are shown by pairs. The first cell type of the interacting pair expresses the ligand and the second cell type expresses the receptor (i.e. first and second proteins from the interaction, respectively). Ligand-receptor interactions are grouped by signaling pathways. Yellow: EGFR, pink: PDGF, green: WNT, red: TGFb, black: Notch, blue (light or dark): mixed of TGFb and EGFR. None of the hedgehog interactions were significant. For details on statistics and reproducibility, please see Methods.
a. Representative image of Col1a1 in-situ hybridization in a PdgfrbCreER;tdTomato kidney after UUO surgery. Scale bar 10 μm b-c. Quantification of aSMA+ cells in PDGFRbtdtom+ kidneys from UUO day 10. n=3. Mean± SD. d. Scaled expression of the top 10 genes by specificity in each cell cluster depicted in Figure 3e. All cells from each cell cluster are averaged in one column. e. Expression of select genes in all 10 cell clusters from Figure 3e. f. ECM score visualized on the same UMAP embedding from Figure 3e. g. Distribution of ECM score, collagen score, glycoprotein score and proteoglycan score per cell cluster. h. Immunofluorescence (IF) staining in sham and UUO (day 10) mouse kidney showing Pdgfra expression in a subset of PDGFRbCreER;tdTomato positive cells (arrows). i. RNA in-situ hybridization showing colocalization of Col1a1 expression in PDGFRa/PDGFRb double-positive cells. Col1a1/PDFGRa/PDFRb triple-positive cells (arrows) occur solely in the kidney interstitium. j. Left: Col1a1 expression and ECM score in CD10 negative cells (Figure 1b) stratified according to PDGFRa and PDGFRb expression. Right: Percent of Col1a1 positive and negative cells in the same data, stratified in the same way. Col1a1 negative cells occur mostly in PDGFRa/b double-negative cells while Col1a1 positive cells occur predominantly in PDGFRa/b double-positive cells (n=51,849). Group comparisons: (other genes) vs. (a/b): p~0, (a-) vs. (a/b): p~0, (b) vs. (a/b): p~0, (other genes) vs. (a): p~0, (b) vs. (a): p~0, (other genes) vs. (b):p~0. Bonferroni corrected p-values based on a two-sided Wilcoxon rank sum test. k. Distribution of IF/TA-Score over 62 patients and representative image of a trichrome stained human kidney tissue microarray (TMA) stained by multiplex RNA in-situ hybridization using PDGFRa, PDGFRb and Col1a1 probes with nuclear counterstain (DAPI) of 62 kidneys (patient data in Extended Data Table 2) (left), average scaled Col1a1 expression in the in-situ hybridization data stratified by PDGFRa/PDGFRb detection in the same data (middle) and percent of Col1a1 positive and negative cells in the same data stratified in the same way (right). Group comparison: (a/b) vs. (col1α1): p~0, (a/b) vs. (b): p~0, (a/-) vs. (a): p~0. Bonferroni corrected p-values based on a two-sided Wilcoxon rank sum test. l-p. A Diffusion Map embedding of pericytes and matrix producing cells with annotation of the different time points in m, cell cluster annotation in n and scaled expression of selected genes in o-q. q. The surgery type per cell (sham versus UUO) visualized on the same UMAP embedding from Figure 4c (top), or with colors representing the cell types/states (bottom). r. Expression of select genes on the same UMAP embedding from 3j. s. ECM and collagens score distribution for the 4 major cell types (top) and for mesenchymal clusters (bottom). Scale bars h+j 50 μm, in k 10 μm. For details on statistics and reproducibility, please see Methods.
a. Scaled expression of the top 10 genes by specificity in each of the mesenchymal cell clusters depicted in Figure 3d. Each 100 cells are averaged in one column. b. Cell doublet score (see Methods) of mouse PDGFRa/b+ dataset per cell cluster. c. A violin plot of Col15a1 expression per cell cluster. Only mesenchymal cells are shown. Bonferroni corrected p-values based on a two-sided Wilcoxon rank sum test in Supplemental File 4. d. A UMAP embedding of Meg3 as in Figure 2a and multiplex in-situ staining of Meg3 on human kidney tissue. Scale bars d1 30 μm, d2+3 40 μm e. Representative image of multiplex RNA in-situ hybridization for PDGFRa, PDGFRb and Meg3 in n=34 human kidneys (Patient Data in Extended Data Table 2). Meg3 colocalizes with PDGFRa and PDGFRb. Scale bar 10 μm f. Percent of Meg3-cells out of PDGFRa/b double-positive cells, quantified from RNA in-situ hybridization. n=34. Tukey box-whisker plot. g. Expression of select genes on the same UMAP embedding from Figure 3j. For details on statistics and reproducibility, please see Methods.
a. Classification tree of human PDGFRb dataset derived by the CHETAH algorithm based on single cell expression and clustering information. b. Supervised classification of mouse PDGFRa+/b+ cells using human PDGFRb+ cells as a reference (see classification tree in a.). Heatmap displays percentage of mouse PDGFRa+/b+ cells in each mouse cell cluster. Fibroblasts 1 in mice are largely classified as Fibroblasts 1 according to human data. Mouse myofibroblasts are classified as Node 15 and myofibroblasts 2b in humans indicating variability between mouse and human with myofibroblast states. c. Schematic of proposed cellular origin of fibrosis. d. Scaled gene expression of transcription factors discovered by ATAC-Seq (see Figure 3q) in six fibroblasts and myofibroblast cell populations. e. ATAC-Seq signal for motif matches inside open chromatin regions for five selected transcription factors. f. Genome browser snapshots for select genes. ATAC-Seq signal and motif matches in open chromatin regions are shown. Multiz Align is conservation scores between mouse and human, ClinVar lift is clinical variants lift to mouse genome coordinates. Nrf, Irf8, Elf/Ets and Klf motifs are located in promoter and enhancer open chromatin regions of myofibroblast associated genes such as Col1a1, Col15a1, Tgfb and Nkd2. Creb5_Atf3 is found in genes associated to Fib1. cluster, such as Tmeff2. g. Expression of some of the genes investigated in g-i. Visualized on the same UMAP embedding from Figure 4c. i. Scaled expression of genes that correlate or anti-correlate with injury time across matrix producing cells (mouse PDGFRb+ data). Note the expression of Ogn, Scara5 and Pcolce2 is largely specific to day0-day2 cells while the expression Nkd2, Fbn2 and Nkd1 is specific is increased in day 10 after UUO. h. Signaling pathway enrichment in the same mesenchymal cell clusters depicted in Figure 3q.
a-b. Gene ontology Biological Process terms for genes that correlate or anti-correlate with Nkd2+ expression across single cells in pericytes fibroblasts and myofibroblasts in mouse PDGFRa+/b+ data (a) and human PDGFRb+ data (b). Genes correlated with Nkd2+ expression are related to ECM expression, integrin signaling and focal adhesion. c. Pathway activity as estimated by the PROGENy algorithm in NKD2+ vs. NKD2- cells from the human PDGFRb+ dataset. p>0.05 n.s., *p<0.05, **p<0.01, ***p<0.001, p values were adjusted for multiple testing using Benjamin/Hochberg method (FDR) (c). d. Scaled gene expression of top 100 genes whose expression is correlated or anti-correlated with Nkd2 expression across single cells in human PDGFRb+ data (see also b.) e. Gene regulatory network predicted based on the expression of cells and genes depicted in l. using the GRNBoost2+ algorithm. Connection between genes indicate putative direct or indirect regulatory interactions. Colors indicate clustering of the gene regulatory network using the Louvain algorithm and highlights the regulatory network of ECM expression (module 2, Nkd2+) and fibroblast and pericyte maintenance (module 4 and 3) f. Module 2 from l. Depicted separately, connections of Nkd2 are highlighted in red. g. Expression of genes highlighted in e. and f. including Etv1 transcription factor and Lamp5 which are both directly connected to Nkd2 in e. and f. h. Expression of Col1a1, Fibronectin (Fn) and Acta2 (aSMA) by qPCR after Nkd2 over-expression in human immortalized PDGFRb+ cells treated with transforming growth factor beta (TGFb) or vehicle (PBS). n=3 per group. 1-way ANOVA followed by Bonferroni’ post-hoc test. Data represent the mean ± SD. i. Expression of NKD2 by RNA qPCR in NKD2 KO cells. ****P <0.0001 by 1-way ANOVA followed by Bonferroni’ post-hoc test. Data represent the mean ± SD. j. Expression of Col1a1, Fibronectin (Fn) and Acta2 by RNA qPCR after Nkd2 knock-out in the same clones depicted in h. n=3 per group. #p<0.05, ##p<0.01, ###p<0.001, ####p<0.0001 (vs. control NTG); ****p <0.0001 (vs. TGFb NTG) by 2-way ANOVA followed by Sidak’s post-hoc test. Data represent mean ± SD. k. PID signaling pathways enriched in PDGFRb+ NKD2-KO clones and overexpression (up indicates up-regulated genes in indicated condition, and down indicates down regulated genes). l. Gene ontology Biological Process terms enriched in PDGFRb+ NKD2-KO clones (up indicates up-regulated genes in KO condition, and down indicates down regulated genes). m. Scaled gene expression of WNT pathway receptors and ligands in Nkd2-perturbed human kidney PDGFRb+ cells.*p< 0.05, **p< 0.01, and ***p < 0.001 as determined by the empirical Bayes from the test for differential expression after adjusting p-values for multiple testing correction (Benjamini & Hochberg) n. Representative image of multiplex RNA in-situ hybridization of PDGFRa, PDGFRb and NKD2 in human iPSC derived kidney organoids. o. Immunofluorescence stainings of human iPSC derived kidney organoids (day 7+18). LTA and HNF4a mark proximal tubular like-cells. pan-CK (Cytokeratin) marks epithelial-like cells. ERG (ETS regulated-gene) marks endothelial-like cells. Dach1 and Nephs1 mark podocyte-like cells. Col1a1 marks fibroblast/myofibroblasts. p. Immunofluorescence stainings of Col1a1 in IL1b treated kidney organoids. Scale bar in n, o, p=50 μm. For details on statistics and reproducibility, please see Methods.
a. Scheme of the kidney b. UMAP embedding of 51,849 CD10- single cells from 15 human kidneys. Labels refer to 50 clusters identified, see Supplementary File 1. c. Scaled gene expression of the top 10 specific genes in each cluster (see Supplementary File 2 for detailed information). Each column is the average expression of all cells in a cluster. d. Stratification of cells by estimated glomerular filtration rate (eGFR). e. UMAP embedding of 31,875 CD10+ single cells stratified by eGFR f. KEGG pathway enrichment for CD10+ cells. g. CD10- clustering by ECM (extracellular matrix) score stratified by eGFR (see Extended Data Figure 2p). h. ECM score stratified by cell type and eGFR, Mesenchymal (p ~0), Immune (p ~0), Epithelial (p ~0), Endothelial (p ~0) i. Single cell ECM score for mesenchymal cells, stratified by major cell types and by eGFR. P-value of differences in eGFR categories: Fib1 (0.00015), Fib2 (1), Fib3 (0.54), MF1a (1), MF1b (0.59), Pe1 (0.096), Pe2 (1), SMC (0.162). (h.-i.) Bonferroni corrected p-values based on two-sided t-test. j. Number of cells per mesenchymal cell type and clinical parameter. Hypergeometric test, adjusted p value for fibroblast and myofibroblast = ~0 - for pericyte and vascular smooth muscle cells ~ 1. k. Diffusion mapping of mesenchymal cells, pseudotime indicates cell ordering along putative differentiation processes. For details on statistics and reproducibility, please see Methods.
a. UMAP embedding of 37,800 Pdgfrb+ single cells from 8 human kidneys. Labels refer to identified cell-types by unsupervised clustering (see Supplementary File 1). b. Expression of selected genes on the embedding from a. c. Diffusion Map embedding of Pdgfrb+ fibroblasts, myofibroblasts and pericytes (n=23,883) and the expression of selected genes on the same embedding. Red lines correspond to the three lineage trajectories (L1, L2 and L3) predicted by Slingshot given the Diffusion Map coordinates and the clusters depicted in Extended Data Figure 5b. d. Representative images of RNA-in-situ hybridization for Meg3, Notch3, Postn in 35 human kidneys (Patient Data in Extended Data Table 2, IFTA=interstitial fibrosis, tubular atrophy score). n=17 (a), 10 (b) and 8(c); ***p < 0.001 by 1-way ANOVA followed by Bonferroni’s correction. Tukey box whisker plot. Scale bar left 10 μm, right 25 μm. e. Top: Gene expression dynamics along pseudotime for Lineage 1 (see c., see Methods). Middle: Cell cycle stage as percent of each 2000 cells along pseudotime. Bottom: PID Signaling pathway enrichment along pseudotime. For details on statistics and reproducibility, please see Methods.
a. Fate tracing experiment design b. Col1a1 in-situ hybridization in a PdgfrbCreER;tdTomato kidney. Scale bar 1000 μm. c. Percentage of Col1a1-mRNA expressing cells that co-express tdTomato at day10 after (unilateral ureteral obstruction, UUO, n = 3, shown mean). d. Time-course UUO experiment design. e. UMAP embedding of the mouse Pdgfrb+ cells. Labels refer to a cell-types identified. f. Percent of cells per cell type and time-point. g. Expression of selected genes on the UMAP embedding from e. h. Scheme of the PDGFRa/PDGFRb isolation UUO experiment. i. Quantification of Pdgfra+/Pdgfrb+ cells by flow cytometry (n=5 per group). *p<0.05; **p<0.01 by one-way ANOVA with post-hoc Bonferroni correction. Data shown as mean ± s.e.m. j. Left: UMAP embedding of the Pdgfra+/Pdgfrb+ cells Right: Percent of cells per cluster. k. Expression of selected genes in each of the cell clusters from j. n., o. UMAP and diffusion map embedding of fibroblasts and myofibroblasts. p. Computational cell ordering (pseudotime) and expression of selected genes on the embedding in n. q. Enrichment of transcription factor motif occurrence in fibroblasts and myofibroblasts. TF motifs were identified from Pdgfra+/Pdgfrb+ day 10 UUO ATAC-Seq data (see Methods). For details on statistics and reproducibility, please see Methods.
a. Expression of Nkd2 visualized on the UMAP embedding from Figure 3j. b. Percent of Col1a1+/- cells in mouse Pdgfra+/Pdgfrb+ cells (Figure 3j, stratified by Pdgfra and Nkd2 expression). c. Scaled gene expression of Nkd2 correlating or anti-correlating genes in human Pdgfrb+ cells (Figure 2). d.-e. RNA in-situ hybridization (ISH) of PDGFRa, PDGFRb and NKD2 in human kidneys and quantification of triple positive cells (n=36, Patient data in Extended Data Table 2). n=20 and 16. Two-tailed Mann-Whitney test. Tukey box whisker plot. IF-score = interstitial fibrosis score. Scale bar 10μm. f.-g. Representative Western blots of Nkd2 overexpression and KO cells. For gel source data, see Extended Data Fig. 10e. h. GSEA (Gene set enrichment analysis) of ECM genes in Nkd2-perturbed PDGFRb- kidney cells. n=3 each. *P < 0.05, **p< 0.01, and ***p < 0.001 as determined by FGSEA-multilevel method after adjusting p-values for multiple testing correction (Benjamini & Hochberg). i. ISH of Pdgfra, Pdgfrb and Nkd2 in human iPSC derived kidney organoids. j. Quantification of Nkd2 RNA expression in organoids. n=4 each. Two-tailed unpaired t-test. k.-l. Immunofluorescence staining and quantification of Col1a1 in organoids. n=4 each. *P < 0.05, **p< 0.01, and ***p < 0.001 by 1-way ANOVA followed by Bonferroni’s correction. Scale bar in i+k 50 μm. Data shown as mean±SD. For details on statistics and reproducibility, please see Methods.
Comment in
-
Tracing the origins and differentiation of human kidney myofibroblasts.Nat Rev Nephrol. 2021 Mar;17(3):151. doi: 10.1038/s41581-020-00382-3. Nat Rev Nephrol. 2021. PMID: 33268843 No abstract available.
-
Searching for the origin of the myofibroblast one cell at a time.Kidney Int. 2021 Jun;99(6):1259-1261. doi: 10.1016/j.kint.2021.02.016. Epub 2021 Feb 26. Kidney Int. 2021. PMID: 33647325 No abstract available.
-
Uro-Science.J Urol. 2021 Aug;206(2):480-482. doi: 10.1097/JU.0000000000001849. Epub 2021 May 12. J Urol. 2021. PMID: 33975458 No abstract available.
References
-
- Falke LL, Gholizadeh S, Goldschmeding R, Kok RJ, Nguyen TQ. Diverse origins of the myofibroblast—implications for kidney fibrosis. Nat Rev Nephrol. 2015;11:233–244. - PubMed
Publication types
MeSH terms
Substances
Grants and funding
LinkOut - more resources
Full Text Sources
Other Literature Sources
Medical
Molecular Biology Databases
