Molecular characterization of the human kidney interstitium in health and disease

Abstract

The gene expression signature of the human kidney interstitium is incompletely understood. The cortical interstitium (excluding tubules, glomeruli, and vessels) in reference nephrectomies (N = 9) and diabetic kidney biopsy specimens (N = 6) was laser microdissected (LMD) and sequenced. Samples underwent RNA sequencing. Gene signatures were deconvolved using single nuclear RNA sequencing (snRNAseq) data derived from overlapping specimens. Interstitial LMD transcriptomics uncovered previously unidentified markers including KISS1, validated with in situ hybridization. LMD transcriptomics and snRNAseq revealed strong correlation of gene expression within corresponding kidney regions. Relevant enriched interstitial pathways included G-protein coupled receptor. binding and collagen biosynthesis. The diabetic interstitium was enriched for extracellular matrix organization and small-molecule catabolism. Cell type markers with unchanged expression (NOTCH3, EGFR, and HEG1) and those down-regulated in diabetic nephropathy (MYH11, LUM, and CCDC3) were identified. LMD transcriptomics complements snRNAseq; together, they facilitate mapping of interstitial marker genes to aid interpretation of pathophysiology in precision medicine studies.

Fig. 1. Known segmental nephron markers in the LMD interstitium compartment.

(A) Dissection of the renal interstitium revealed increased expression of collagen markers (COL1A1, COL6A2, and COL6A3) compared with the other subsegments. (B) The interstitial dissection lacked expression of glomerular markers such as nephrin (NPHS1), podocin (NPHS2), and podocalyxin (PODXL). (C) Proximal tubular expression of megalin (LRP2), cubilin (CUBN), and the sodium-hydrogen exchanger regulatory factor 3 (PDZK1) was higher as compared with the interstitium. (D) Uromodulin (UMOD), NKCC2 (SLC12A1), and calcium-sensing receptor (CASR) expression was higher in the thick ascending loop than the interstitium. (E) The interstitium had minimal expression of CD markers such as aquaporin 2 (AQP2), the amiloride-sensitive sodium channel subunit beta (SCNN1), and solute carrier family 4 member 9 (SLC4A9). Immunofluorescence staining with DAPI (nuclei), phalloidin (F-actin, brush border), Tamm-Horsfall protein (THP) (TAL), and peanut agglutinin (collecting duct) were used along with morphology to dissect each specific compartment. Dot plots are provided to visualize individual data points, but all P values correspond to the whole-transcriptome analysis provided in table S1 to maintain consistency throughout the study. ***P <0.05 after FDR correction.

Fig. 2. Known markers and top DEGs in nephron compartments.

(A) Heatmap illustrating relative gene expression of expected subsegmental markers in the subsegments of interest from reference tissue. (B and C) Heatmap depicting the expression of the top 15 protein-coding differentially expressed markers for each subsegment sorted by significance. The identity of each gene is provided in (C). (D to G) Scatter plots indicating differential expression of selected interstitial markers (COL6A2, collagen 6A2; AEBP1, adipocyte enhancer-binding protein 1; SYNM, synemin; TPSB2, tryptase beta 2) identified in the unbiased analysis with representative images of immunohistochemical staining; each marker is obtained from the Human Protein Atlas. Dot plots are provided to visualize individual data points, but all P values correspond to the whole-transcriptome analysis provided in table S1 to maintain consistency throughout the study. ***P <0.05 after FDR correction.

Fig. 3. Network and pathway analysis of gene expression enriched in the renal interstitium.

(A) A transcriptogram illustrating pathway enrichment of the interstitial compartment as compared with the mean expression of all tubular and glomerular compartments. Protein-coding genes are ordered along the x axis according to their network interactions. The top plot and middle plot illustrate the average expression and P value of the network region. The bottom plot provides pathway enrichment based on expression and significance. (B) S-curve plot of genes contributing to the GPCR ligand-binding pathway. (C) S-curve plot of genes contributing to the collagen biosynthesis and modifying enzyme pathway.

Fig. 4. Deconvolution and alignment of LMD expression with single nuclear RNA sequencing.

(A) Uniform Manifold Approximation and Projection (UMAP) of 30 snRNAseq clusters. (B) Cluster identities are provided for all 30 clusters, along with correlation plots between the snRNAseq clusters and compartmental transcriptomic signatures. Selected genes with increased expression in the interstitial compartment are displayed in the corresponding dot plot. In a separate analysis, bulk expression was used to calculate the expression correlation in the bulk heatmap column. (C) UMAP illustrating subclustering of the fibroblast population into three subpopulations. (D) Dot plot revealing the expression of marker genes associated with each of the three identified fibroblast subclusters. (E to H) Feature plots displaying selected genes associated in the fibroblast subclusters: ACTA2 as a marker of myofibroblasts, PDGFRA corresponding to fibroblast subcluster 3, and SPP1 and MYL12B corresponding to fibroblast subcluster 2.

Fig. 5. Localization of identified interstitial cell genes with in situ hybridization and immunofluorescence.

Colocalization of COL6A1 (collagen 6A1), KISS1 (kisspeptin 1), and CELA3A (elastase 3A) with known markers of endothelial cells, VSMCs, and fibroblasts. (A, F, and K) Immunohistochemical staining of COL6A1, KISS1, and CELA3A, as illustrated in the Human Protein Atlas. (B to E, G to J, and K to O) Single-molecule FISH colocalization. The 5-μm measurement bar of “E” corresponds to all FISH images unless otherwise noted. COL6A1 (B; purple) colocalized with fibroblast markers PDGFRA (C; red) and ACTA2 (D; green) and in a merged image of all channels (E; *colocalization of all three markers]; KISS1 (G; purple) colocalizing with VSMC marker TAGLN (H; red), and minimal expression of the endothelial cell marker CD31 (I; green) and with merged image in (J) (^#colocalization of purple and red); CELA3A (L; purple) colocalized with PDGFRA (M; red) and ACTA2 (N; green); merged image (O; white arrows indicate colocalization of markers, and the yellow arrowhead indicates an RNA clump artifact, which should not be mistaken for colocalization). (P to S) DAPI staining of tubular nuclei is seen in a tubular cross section (P and R) with selected nuclei at higher magnification (Q and S) illustrating the absence of targeted RNA molecules associated with the tubular cell. (T to W) Immunofluorescence staining illustrating the presence of CELA3A protein within interstitial cells and lack of colocalization with endothelial cells, further supporting CELA3A expression being associated with fibroblasts. (T and U) CELA3A staining in the interstitial compartment. BV, blood vessel; DT, distal tubule; G, glomerulis; PT, proximal tubule. Phalloidin was used to visualize F-actin, and DAPI was used to stain nuclei. (V and W) CELA3A (red) not colocalized with CD31 stain (in blue). Arrowheads indicate CELA3A-positive cells.

Fig. 6. Network and pathway analysis of DEGs in healthy and diseased renal interstitium.

(A) S-curve plot of DEGs between diabetic nephropathy and reference samples contributing to extracellular matrix organization. (B) S-curve plot of genes contributing to small-molecule catabolic processes. (C to H) Scatter plots of example genes with differential expression in reference and diabetic samples. Examples include up-regulation of plasminogen (PLG) and fatty acid binding protein 1 (FABP1) and down-regulation of collagen 11A2 (COL11A2), dimethylarginine dimethylaminohydrolase 2 (DDAH2), pyruvate kinase L/R (PKLR), and lumican (LUM). (*P < 0.05 uncorrected, **P < 0.1 FDR-corrected, and ***P < 0.05 FDR-corrected). (I) Transcriptogram illustrating pathway enrichment between the diabetic nephropathy and reference interstitial compartments.