Persistent decline in renal function and UCC

To study the effects of prolonged ureteric obstruction on the kidney, male BALB/c mice underwent R-UUO using a light vascular clamp, followed by contralateral nephrectomy (Nx) 10 days later (Figure 1A and Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.187008DS1). In 2 separate experiments, after 7 days of UUO, 30%–50% of mice died within 2 days of Nx, increasing to 80% after 8 days of UUO (Figure 1B and Supplemental Figure 1B). While exclusion of mice dying early after Nx creates a bias for long-term follow-up studies, by performing contralateral Nx 10 days after removing the ureteric clamps, we are excluding those mice in which there is permanent structural or functional loss of ureteric patency resulting in a persistently obstructed kidney. In survivors of 7 days of UUO, blood urea nitrogen (BUN) peaked at ~50–60 mg/dL 1 day after Nx but normalized over time (Figure 1C and Supplemental Figure 1, C and D). There was an increase in serum creatinine at 45 days (Supplemental Figure 1E), and tGFR decreased significantly at 28 and 45 days, improving partially by 84 days but remaining significantly below baseline (Figure 1, D and E, and Supplemental Figure 1F). UCC was markedly reduced at 28, 45, and 84 days after R-UUO (Figure 1, F and G, and Supplemental Figure 1G), indicating lasting renal medullary dysfunction independent of tGFR decline. Survivable UUO clamp times varied between experiments and mouse strains, with 50%–80% survival rates in different experiments (see below). However, all groups showed similar reductions in tGFR and UCC 84 days after R-UUO, suggesting that the severity of functional damage was comparable between strains and experiments.

Long-term decrease in renal function and UCC after R-UUO. Male BALB/c mice underwent R-UUO and contralateral Nx or Nx alone. (A) Study design; (B) survival, with numbers of mice indicated; (C) blood urea nitrogen (BUN) time course; (D) tGFR time course; (E) tGFR at day 28 and 84 after R-UUO. (F and G) Urinary osmolality after 18-hour water restriction. Time course (F) and 84 days after R-UUO (G). (C, D, and F) Mean ± SD. Two-way ANOVA. (E and G) Data points with mean ± SEM. Data points color coded to show relationship between individual tGFR and urine osmolality measurements. One-way ANOVA Nx vs. healthy control and R-UUO. If P < 0.05, q values for between-group comparisons.

Renal medullary injury: early repair and structural remodeling

Renal histology over 28 days after R-UUO revealed tubular damage and increased peritubular cellularity in the IM (Figure 2A). From day 0 (the time of ureteric clamp release) to 3, more than 50% of mice had papillary necrosis, characterized by denuded tubular basement membranes and loss of nuclei, mostly in the distal IM. Papillary urothelium thickened from 1 to 3 or more cell layers during this period. By day 7, there was evidence of tubular repair, with epithelial repopulation of denuded basement membranes progressing over the next 3 weeks. By day 28, most tubular structures were repopulated, with peaks in cell cycle and regeneration markers (Ki67 and SRY-box transcription factor 9, Sox9) at day 7 (Figure 2, B–D). However, at 28 days, the interstitial space remained disorganized, showing expanded myofibroblast-like cells (Figure 2A and Supplemental Figure 2), but by day 84, the IM recovered its full length and appeared histologically normal (Figure 2E and Supplemental Figure 3). Fibrosis, measured by Sirius red (SR) staining, showed early increase in the OM persisting to day 84. In contrast, SR staining in the IM peaked at 28 days and declined by day 84 (Figure 2, F and G, and Supplemental Figure 4). These findings suggest that while there is robust early repair and remodeling of the IM, subtle defects in IM structure and function, undetectable by standard histology, may underlie the long-term impairment in UCC after R-UUO.

Renal medullary injury, repair, and remodeling after R-UUO. Kidneys were collected at different time points after R-UUO. (A) Periodic acid–Schiff (PAS) staining, whole kidney (scale bars = 1 mm), ISOM, and proximal and distal IM (scale bars = 100 μm). *Papillary necrosis; black arrows show thickening of papillary urothelium. (B) Ki67, Sox9, and LTL in the distal IM after R-UUO (scale bars = 50 μm). (C and D) Time course of Ki67+ and Sox9+ cells (% of total nuclei). N = 5-Ctrl, 10-D0, 9-D3, 12-D7, 4-D14, 4-D28. (E and F) PAS (E) and SR (F) staining (scale bars = 1 mm). (G) Quantification of SR staining. Time course in the OSOM, ISOM, and IM. N = 5-Ctrl, 6-D0, 6-D7, 4-D28, 6-D84. Means ± SEM.

Permanent loss of AQP1 and transient loss of AQP2 expression in the IM

Immunostaining was used to assess markers of differentiated IM descending thin limb (DTL) and CD epithelia. Aquaporin 1 (AQP1), which stains DTLs and descending vasa recta (DVR), showed a significant reduction throughout the IM during the first 7 days after R-UUO (Figure 3, A and B). By day 28, AQP1 expression was still reduced throughout the IM, and at day 84, reduction persisted in the distal but not in the proximal IM. This lasting AQP1 loss may result from fewer differentiated DTLs, reduced LOH extension into the IM, or fewer AQP1-positive DVRs. To evaluate CD epithelia, AQP2, which is expressed by differentiated CD principal cells, was initially reduced throughout the IM for 7 days after R-UUO (Figure 3, C and D). AQP2 staining was still decreased by day 28 but returned to control levels by day 84. In contrast, LTL, which stains IM CDs (Supplemental Figure 5) (50), and V-ATPase–positive intercalated cells (ICs) (Supplemental Figure 6), showed a minor, nonsignificant reduction at early time points but normalized by day 28. By day 84, LTL staining in the distal IM increased significantly (Figure 3, C and E). The restoration of AQP2 expression suggests successful regeneration and differentiation of CD epithelium, while increased LTL staining at day 84 may reflect higher density or surface area of IM CD epithelium or nonspecific staining of other tubular structures.

Long-term changes to inner medullary epithelial cells after R-UUO. (A) AQP1 staining of the IM after R-UUO (scale bars = 200 μm). (B) Quantification of AQP1 staining in the proximal and distal IM time course. Day 28 and 84 Nx shown. (C) AQP2 and Lotus Tetroglobinous Lectin (LTL) staining of IM CDs after R-UUO (white scale bars = 200 μm, yellow = 400 μm). (D and E) Quantification of AQP2 (D) and LTL staining (E) as the percentage of the area of interest. Data points shown. Blue dots, proximal IM; red dots, distal IM. Means ± SEM. One-way ANOVA, HC or Nx vs. R-UUO time points. If P < 0.05, q values are shown.

Persistent loss of inner medullary cells and expansion of injured cell populations

To assess changes in IM cell populations following R-UUO, scRNA-Seq was performed on dissected renal medullas from BALB/c mice at 28 (n = 12) and 84 days (n = 6) after R-UUO, compared with healthy controls (n = 12). Analysis of 181,255 single-cell sequences identified 7 major cell types (Supplemental Figure 7, A and B). Two technical considerations should be noted, however.

Sample collection. Renal medullas were isolated to enrich IM cell populations as these cells are generally underrepresented in whole kidney scRNA-Seq datasets (44). Immunostaining verified intact IM structures but variable staining of the OM, including ISOM thick ascending limb (TAL; uromodulin staining) and outer stripe of the outer medulla (OSOM) PTECs (LTL staining) (Supplemental Figure 7C). This limits our ability to provide precise quantification of changes in ISOM and OSOM cell populations after R-UUO.

scRNA-Seq versus snRNA-Seq. While single nuclear RNA sequencing (snRNA-Seq) reduces dissociation bias and avoids stress-induced transcriptional artifacts (51), the 10x Genomics fixed-cell RNA-processing method used here offers significant advantages. It employs barcoding for cost-effective, multiplexed sequencing, enabling the analysis of large numbers of singlet cells (43,000–78,000 per pooled sample, Supplemental Figure 7A). This mitigates dissociation bias and enhances the detection of underrepresented cell populations. Additionally, the use of fixed samples prevents stress responses during dissociation, and scRNA-Seq captures both cytoplasmic and nuclear RNA, providing a more comprehensive view of cellular states and gene expression profiles compared with snRNA-Seq.

Using uniform manifold projection (UMAP) plots for visualization and anchor genes to identify cell clusters in the IM versus OM (see Supplemental Methods 1) (44, 52, 53), 25 cell clusters were identified, encompassing IM and OM CD cells, LOH segments, endothelial cells (ECs), fibroblasts, immune cells, and ICs (Supplemental Figure 7, D and E, and Supplemental Table 2). There was a sustained decrease in IM CD, LOH (DTL, ascending thin limb [ATL], and IM TL), and EC populations after R-UUO (Supplemental Figure 7, F and G). Fibroblasts decreased, and myofibroblasts expanded at day 28 and 84 after R-UUO. Populations of injured CD and LOH cells were also detected at day 28 and 84 after R-UUO. These injured cells had reduced expression of classical markers of ATL, DTL, TAL, and CD principal cells and increased expression of injury markers lipocalin 2 (Lcn2), secreted phosphoprotein 1 (Spp1), Ccn2, clusterin (Clu), and Igfbp2 (Supplemental Figure 7, E and H). These injured, partially dedifferentiated cells resemble maladaptive cell populations reported in mice after IRI-AKI and I-UUO and in patients with CKD and sepsis-associated AKI (35, 36, 38–41).

Sustained reduction of inner medullary LOH

Reclustering of LOH cells based on differentially expressed genes (DEGs) identified 3 IM TL populations: IM ATL-0, upper IM juxtamedullary nephron DTL (jDTL-1), and lower IM TL-5 cells (Figure 4, A and B, and Supplemental Table 2). All 3 populations showed a marked and persistent reduction 84 days after R-UUO (Figure 4, C and D). Lineage analysis using Six2-Cre R26R-LSL-tdTomato mice, which genetically label nephron epithelium including IM LOH cells but not CDs, validated these findings (Figure 4E). Mice optimized for a 5-day UUO (allowing >50% survival after Nx) exhibited similar long-term reductions in tGFR and UCC to those in the scRNA-Seq studies (Supplemental Figure 8). tdTomato quantification verified a generalized loss of LOH cells throughout the IM 84 days after R-UUO (Figure 4, F and G). To further evaluate LOH cell populations, Six2 lineage markers were combined with AQP1 antibody staining, identifying 3 IM LOH cell types: 1) Six2+AQP1+ DTL cells, 2) Six2–AQP1+ DVR cells, and 3) Six2+AQP1– ATL and TL cells. Results showed significant reductions in Six2+AQP1+ DTL cells in the distal IM and Six2– AQP1+ DVR cells throughout the IM 28 days after R-UUO (Figure 4, H–J, and Supplemental Figure 9, A–C). However, Six2+AQP1– ATL and TL cells showed no change at 28 days (Supplemental Figure 9D). This suggests that the reduction in Six2 lineage cells primarily involves IM DTL cells, while IM ATL and TL populations remain unchanged, contrasting with scRNA-Seq data indicating reductions across all IM LOH populations.

Permanent reduction in inner medullary LOH with expansion of injured cells. (A and C) Reclustering of LOH cells identified 10 distinct cell populations in the combined dataset (A) and after R-UUO (C); (B) top 5 DEGs for each LOH cluster; (D) fractional representation of LOH clusters after R-UUO; (E) genetics labeling of IM LOH cells using Six2-Cre mice; (F) fluorescence images of Six2 lineage cells in the IM after R-UUO (scale bars = 200 μm); (G) quantification of Six2 lineage cells in the IM 84 days after R-UUO; (H) fluorescence images with overlays showing AQP1 and Six2 lineage staining in the distal IM 28 days after R-UUO (scale bars = 200 μm); (I and J) quantification of AQP1+ and Six2+ cells in the distal IM as the percentage of cells (I); AQP1+Six2+ DTLs (% of Six2+ cells), and AQP1+Six2– DVRs (% of total cells) 28 days after R-UUO (J). Data points, means ± SEM. P values determined by t test. (K) Dot plot of the top 20 DEGs from injured LOH-4 cells in other LOH cell types after R-UUO; (L) UMAPs illustrating expression of injury markers Lcn2 and Ly6e after R-UUO.

Expansion and features of injured LOH cell populations

There was an expansion of injured LOH-4 cells after R-UUO. Injured LOH-4 cells exhibit characteristics of failed repair, including partial dedifferentiation and persistent cellular injury. They expressed injury markers (e.g., Lcn2, Spp1, Clu, and Mmp7) while showing reduced expression of segment-specific LOH markers (Figure 4B and Supplemental Table 2). Many of their DEGs, such as Sik1, Wfdc2, Kcnk1, and Krt8, were either unique to injured LOH-4 cells or also expressed in other LOH clusters in control mice, where they did not change with injury (Figure 4, K and L). Detailed analysis identified 4 subclusters of injured LOH-4 cells (Supplemental Figure 10A). Subclusters 4-1 and 4-3 closely resembled upper IM and ISOM LOH populations (e.g., jDTL-1, ATL-0, TL-5), while subclusters 4-0 and 4-2 were more similar to OSOM and ISOM TAL clusters (e.g., TAL-9 and TAL-5) (Supplemental Figure 10B). UMAP plots for cell type–specific genes, such as Aqp1 in injured LOH 4-3, and Psca in injured LOH 4-0, confirmed the origins of injured LOH-4 subclusters from different LOH cell types in the IM and OM (Supplemental Figure 10, C and D). Gene set enrichment analysis (GSEA) revealed that injured LOH-4 cells were metabolically quiescent, with downregulated pathways for respiration, metabolite processing, and oxidative phosphorylation, but enrichment for cell migration and wound healing processes (Supplemental Figure 11, A and B, and Supplemental Table 3). Hallmark pathways showed increased apoptosis and TNFa via Nfkb signaling. These findings align injured LOH-4 cells with FR-PTECs after I-UUO (40), showing overlap in DEGs and shared pathways for inflammation, migration, and wound healing (Supplemental Figure 12 and Supplemental Table 4). This suggests similar maladaptive repair mechanisms underlie both FR-PTECs and injured LOH-4 cells.

Minor long-term increase in inflammatory macrophages

Inflammatory macrophages typically expand around failed-repair cells after IRI-AKI (35), but scRNA-Seq revealed only a minor increase in renal medullary inflammatory cells 28 and 84 days after R-UUO (Big UMAP, Supplemental Figure 7, D and F). Reclustering of immune cells identified 3 populations: T cells and 2 macrophage clusters (Mac-0 and Mac-1). T cell numbers decreased, while Mac-1 cells increased after R-UUO (Supplemental Figure 13, A–D, and Supplemental Table 2). GSEA showed activated pathways for migration, adaptive immunity, and phagocytosis in Mac-0 and Mac-1 cells (Supplemental Figure 13E). Hallmark pathways in Mac-1 cells highlighted TNFa via Nfkb and IL-2/STAT signaling, with an overall inflammatory response (Supplemental Figure 13F). However, analysis of pro-inflammatory (M1) and anti-inflammatory (M2) gene sets used to characterize renal macrophage populations after AKI (54–56) revealed no evidence of M1 or M2 polarization in either macrophage cluster (Supplemental Figure 13G). F4/80 antibody staining verified an early increase in macrophages after R-UUO, which decreased over time but remained elevated in the cortex, OM, and proximal IM 84 days after R-UUO (Supplemental Figure 13, H–J). These data indicate that despite the inflammatory signature of injured LOH-4 cells, long-term increases in inflammatory macrophages were modest.

Permanent reduction in vasa recta throughout the IM

Reclustering of ECs identified 9 distinct populations, including ascending vasa recta (AVR)/papilla-1, DVR/papilla-6, AVR-0, DVR-2, and capillary angiogenic-3 and -7 cells (Figure 5, A and B, and Supplemental Table 2). After R-UUO, AVR/papilla-1 and DVR/papilla-6 cell numbers decreased at 28 and 84 days, while capillary angiogenic-3 and -7 cells increased (Figure 5, C and D). GSEA showed enriched Gene Ontology (GO) terms for cell migration in AVR/papilla-1 and DVR/papilla-6 cells, with AVR/papilla-1 cells also displaying ER stress responses (Supplemental Figure 11C). Capillary angiogenic-3 and -7 cells were enriched for angiogenesis and migration pathways, suggesting attempted repair (Supplemental Figure 11D). Despite this, CD31 antibody staining revealed reduced EC density throughout the IM at 84 days after R-UUO (Figure 5, E–G). This is consistent with the reduced Six2 lineage, AQP1+ DVR cells and the persistent scRNA-Seq reduction in AVR/papilla-1 and DVR/papilla-6 cells. These findings indicate a permanent loss of IM peritubular capillaries, despite repair efforts. This rarefaction likely contributes to worsening hypoxia, particularly in the distal IM, which may exacerbate cellular injury.

Reduced density of vasa recta in the renal medulla. (A and C) Reclustering of renal medullary ECs identified 9 populations in the combined dataset (A) and at various time points after R-UUO (C); (B) top 5 DEGs for each EC cluster; (D) fractional representation of EC clusters after R-UUO; (E) CD31 immunostaining of ECs in the IM 84 days after R-UUO (scale bars = 200 μm); (F) quantification of CD31 staining after R-UUO in the proximal (F) and distal (G) IM. Data are mean ± SEM, analyzed by 1-way ANOVA. P < 0.05; q values shown where applicable.

Temporal and spatial expansion of myofibroblasts

Reclustering of medullary fibroblasts identified 8 subclusters, including 4 fibroblast clusters (Fib-1, Fib-2, Fib-3, Fib-6), a pericyte cluster (Peri-5), and 2 myofibroblast clusters (Fib/Myofib-4, Myofib-0) (Figure 6, A and B, and Supplemental Table 2). After R-UUO, Myofib-0 cells markedly increased, with a smaller increase in Fib/Myofib-4 cells at 28 and 84 days. Concurrently, Fib-2 and Fib-6 cells decreased, Fib-1 cells showed a modest reduction, and no changes were observed in Fib/interferon-7, Fib-3, or Peri-5 cells (Figure 6, C and D). Immunostaining with α–smooth muscle actin (aSMA), a marker for myofibroblasts, revealed increased aSMA+ cells in the cortex and OM shortly after R-UUO, peaking in the IM at 28 days (Figure 6, E and F). By day 84, aSMA staining in the ISOM and IM was markedly reduced. Transcriptionally, Myofib-0 cells exhibited low Acta2 (aSMA) expression but high levels of Fn1 (fibronectin 1) and type 1 and 3 collagen genes (Col1a2, Col1a1, Col3a1), highlighting their role in extracellular matrix remodeling (Figure 6B and Supplemental Figure 14A). These findings indicate a temporally and spatially restricted expansion of myofibroblasts in response to R-UUO.

Spatially and temporally restricted myofibroblast expansion derived from RMICs. (A and C) Reclustering of fibroblasts identified 8 clusters in the combined dataset (A) and at various time points after R-UUO (C); (B) top 5 DEGs for each fibroblast cluster; (D) fractional representation of fibroblast populations after R-UUO; (E) immunostaining of aSMA and AQP1 after R-UUO (white lines indicating ISOM/IM boundaries, scale bars = 250 μm). (F) Quantification of aSMA staining at various time points after R-UUO. N = 4-Ctrl, 9-D0, 8-D3, 12-D7, 4-D28, 14-D84. Means ± SEM. (G) UMAPs showing RMIC marker distribution (Tnc and Dan/Nbl1); (H) aSMA staining in the distal IM after R-UUO; (I) TNC Cre-ERT2-IRES-GFP genetics used to track RMIC lineages; (J) Immunostaining of Tnc lineage (tdTomato) and LTL (white scale bars = 200 μm, yellow scale bars = 50 μm); (K) Tnc lineage (tdTomato) overlaid with GFP to detect active Tnc or FN (dotted lines marking IM/ISOM boundaries, scale bars = 250 μm).

Localization of fibroblast clusters in the renal medulla

Unlike LOH and EC populations, fibroblast anchor genes did not indicate spatial localization within the renal medulla. Hierarchical clustering of fibroblast DEGs showed that Myofib-0 cells were closely related to Fib-2, Fib-6, and Peri-5 cells, while Fib/Myofib-4 and Fib/interferon-7 cells were more related to Fib-1 and Fib-3 cells (Supplemental Figure 14B). Smoc2 is expressed by Fib/Myofib-4 cells in controls, and in Fib-1, Fib-3, and Myofib-0 cells after R-UUO (Supplemental Figure 14A). Since Smoc2 is expressed by interstitial fibroblasts in the OM (57), and is increased in OM myofibroblasts after IRI-AKI (58), these findings suggest that Fib-1, Fib-3, and Fib/Myofib-4 cells are likely to be localized in the OM and give rise to OM myofibroblasts. Meanwhile, Akr1b3, Lmo7, and CryAB were expressed in control Fib-2 and Fib-6 clusters and restricted to Fib-2 cells after R-UUO (Supplemental Figure 14A). Akr1b3 regulates osmolyte synthesis, Lmo7 is upregulated in hyperosmolar states, and both are increased in the IM (59, 60), but CryAB is a molecular chaperone that is increased in all IM cell types (Supplemental Figure 14C) (44). This suggests that Fib-2 and Fib-6 cell clusters are restricted to the IM.

Renal medullary myofibroblasts derive primarily from RMICs

Expansion of Myofib-0 and Fib/Myofib-4 cells was accompanied by a depletion of Fib-2 and Fib-6, and to a lesser extent Fib-1 cells, but there was no change in vascular pericyte (Peri-5) numbers after R-UUO (Figure 6, C and D). Since pericytes are depleted as they give rise to cortical and OSOM myofibroblasts’ expansion after IRI-AKI and I-UUO (61–63), these findings suggest that Fib-2 and Fib-6, and to a lesser extent Fib-1, contribute to the IM and ISOM myofibroblast expansion after R-UUO, while vascular pericytes play only a minor role. RMICs are specialized renal medullary fibroblasts that express Tenascin C (Tnc) and Dan/Nbl1 (64) and are also expressed by Fib-1, Fib-2, Fib-3, and Fib-6 cells (Figure 6G). In addition, aSMA staining at early time points after R-UUO showed localized ladder-like patterns characteristic of RMICs (Figure 6H). This suggests that Fib-1, Fib-2, Fib-3, and Fib-6 cells include RMICs and that some of the IM myofibroblasts maybe derived from RMICs. To verify this, we used a tamoxifen-inducible Tenascin C Cre-ERT2-IRES-GFP mouse model (64). These mice selectively target RMICs but not other fibroblast or pericyte populations in the kidney (48, 49, 64, 65). We evaluated Tnc expression using the GFP reporter, and the fate of RMICs by crossing with R26R-LSL-tdTomato reporter mice, which enables tracking RMIC lineage cells after tamoxifen activation (Figure 6I). These mice had 60% survival after 6 days of UUO, and while tGFR was unchanged, UCC was significantly reduced 28 days after R-UUO (Supplemental Figure 15). Tamoxifen-treated mice showed tdTomato+ RMIC-derived cells extending throughout the IM and into the ISOM, reaching just below the OSOM/ISOM junction (Figure 6J). However, 28 days after reversal, RMIC lineage cells in the IM appeared disorganized, lacking the characteristic laddered arrangement between tubules (46, 47), and by day 84, there was only partial reorganization of these cells in the IM, since many were no longer arrayed in columns along the length of IM CDs. Costaining with GFP antibodies showed a more limited distribution of Tnc expression within the RMIC lineage domains (Figure 6K and Supplemental Figure 16). FN staining, which is limited to a subset of RMIC lineage cells in controls, was expressed by the majority of these cells 28 days after R-UUO (Figure 6K and Supplemental Figure 16). aSMA staining showed a similar distribution to FN, and like FN, a few IM aSMA+ cells were Tnc lineage negative. FN+ and aSMA+ myofibroblasts in the renal cortex were also Tnc lineage negative. These findings indicate that 1) most myofibroblasts in the ISOM and IM are derived from RMICs; 2) there are at least 2 populations of RMICs, both of which express RMIC markers and contribute to the Tnc lineage, including Fib-1 and Fib-3 cells, which also express Smoc2 and are most likely localized in the OM, and Fib-2 and Fib-6 cells, which express CryAB, Akr1B, and Lmo7 and are most likely localized in the IM; and 3) the organization of RMICs in the IM is disrupted as the cells differentiate into myofibroblasts and remains abnormal 84 days after R-UUO.

Reduced numbers of enlarged CDs in the IM

Reclustering of CD principal cells revealed 2 CD populations that are spatially restricted to the IM, IM CD-2 and deep IM CD-0 cells (Figure 7, A and B, and Supplemental Table 2). By day 84 after R-UUO, expression of water channels (Aqp2, Aqp3) and the urea transporter Slc14a2 were significantly reduced in these cell populations (Supplemental Table 5). Notably, injured CD-5 cells appeared only in R-UUO samples but without significant loss of water channels (Aqp2, 3, 4, 6) or urea transporters (Slc12a1, Slc12a2) compared to other CD clusters. IM CD-2 and deep IM CD-0 cells were also reduced at day 84 after R-UUO (Figure 7, C and D). To investigate further, lineage tracing with HoxB7 Cre–knockin mice crossed with R26R-LSL-tdTomato mice was performed (Figure 7, E and F). A 6-day ureteric clamp caused a similar reduction in tGFR and UCC by day 84 as seen in scRNA-Seq studies (Supplemental Figure 17). Quantification showed an increased proportion of HoxB7 lineage CD cells in the distal IM but fewer total cells overall (Figure 7, G and H). This may reflect reduced LOH and EC populations. Detailed analysis revealed fewer CD tubules throughout the IM (Figure 7I) but increased tubule size and luminal surface areas (Figure 7, J and K). Consistent with the increase in proportion of IM HoxB7 lineage cells, the number of nuclei per CD section also rose (Figure 7L). These findings indicate that there is a permanent reduction in IM CD numbers accompanied by compensatory enlargement of the surviving CDs after R-UUO.

Reduced numbers and enlargement of CDs in the IM after R-UUO. (A and C) Reclustering of CD principal cells identified 6 populations in the combined dataset (A) and at various time points after R-UUO (C); (B) top 5 DEGs for each CD population; (D) fractional representation of CD populations after R-UUO; (E) genetic labeling of CD lineage using HoxB7 Cre; (F and G) fluorescence images and quantification of HoxB7 lineage cells 84 days after R-UUO (scale bars = 200 μm); (I–L) measurement of CD numbers (I), surface areas (J), luminal areas (K), and nuclei per CD section (L), using HoxB7-Cre tdTomato and AQP2 staining. Individual data points, means ± SEM. P values from t tests shown.

Persistent expression of injury markers in inner medullary CDs after R-UUO

Injured CD-5 cells displayed increased expression of injury markers, such as Lcn2, Spp1, Clu, Mmp7, and Ccn1 (Figure 7B and Supplemental Figure 18, A and B). Injured CD-5 cells expressed CD markers like Aqp2, 3, and 4, found across CD clusters, and Slc14a2, typically dominant in deep IM CD-0 cells, suggesting these cells originated from both IM and OM CD populations (Supplemental Figure 18A). To characterize these cells further, we analyzed the top 25 upregulated DEGs in injured CD-5 cells (Figure 8A). Some DEGs, such as Pdgf and Dcdc2a, were uniquely expressed in CD-5 cells, while others, such as Clu, Mmp7, Lcn2, Spp1, and Gdf15, were shared with other CD clusters (Figure 8, A and B, and Supplemental Figure 18B). Immunostaining revealed the spatial distribution of 2 injury markers, NGAL (encoded by Lcn2) and SPP1, in HoxB7 Cre lineage mice. NGAL was absent in controls but markedly increased after R-UUO in urothelium, AQP1+ DTLs, and HoxB7 lineage CD cells in the IM (Figure 8C). Big UMAP analysis showed Spp1 expression in ISOM TAL, urothelium, and intercalated cells (ICs) in uninjured mice, with increased expression in ICs, urothelium, injured CD cells, deep IM CD cells, ISOM TAL, and injured LOH cells after R-UUO (Figure 8D). Consistently, immunostaining detected SPP1 in TALs (Na+/K+-ATPase high) and in interspersed HoxB7 lineage cells in the ISOM and proximal IM of uninjured mice (Figure 8E). These cells, likely ICs, were localized in ISOM and proximal IM CDs. After R-UUO, SPP1 staining increased in TALs and presumed ICs by day 28 and in urothelium and distal IM HoxB7 lineage CDs by day 84 (Figure 8E). These findings corroborate scRNA-Seq data and highlight the widespread, persistent expression of injury markers, particularly in the distal IM, after R-UUO.

Persistent expression of injury markers in inner medullary CDs. (A) Dot plot of the top 25 DEGs from injured CD-5 cells in other CD populations after R-UUO; (B and D) UMAPs showing the expression of injury markers (Lcn2, Spp1, Ccn2) in CD clusters (B) and Spp1 across all renal medullary single-cell clusters (D). (C) NGAL, AQP1, and HoxB7 lineage (tdTomato) staining after R-UUO (scale bars = 100 μm); (E) SPP1, Na+/K+-ATPase, and HoxB7 lineage staining after R-UUO (white scale bars = 250 μm, yellow = 50 μm).

Injured inner medullary CDs exhibit hypoxic adaptation and features of failed-repair cells

GSEA of DEGs in injured CD-5 cells revealed enrichment for processes related to cell migration, wound healing, apoptosis, and TNFa signaling via Nfkb (Figure 9, A and B, and Supplemental Table 3). These align with injury profiles observed in injured LOH-4 cells after R-UUO and FR-PTECs after IRI-AKI and I-UUO (40). However, unlike injured LOH-4 cells and FR-PTECs, injured CD-5 cells displayed upregulation of hypoxia-related genes, including Ldha, Slc2a1/Glut1, Pfkl, and Pfkp, which drive anaerobic glycolysis and likely represent a compensatory response to hypoxia (Figure 9B, Supplemental Figure 12D, and Supplemental Table 3). DEG overlap analyses indicated that deep IM CD-0 and IM CD-2 cells also showed injury responses similar to injured CD-5 cells. Both cell types demonstrated increased cell migration, wound healing, apoptosis, inflammation, and hypoxia signatures at day 28 and 84 after R-UUO, with substantial overlap in DEGs between deep IM CD-0 and injured CD-5 cells (Supplemental Figure 19, A–C, and Supplemental Table 5). In contrast, while IM CD-2 cells exhibited hypoxia and inflammation signatures at day 28, their overlap with injured CD-5 cells diminished by day 84 (Supplemental Figure 19, F and G). These findings contrast with IM LOH clusters. Lower IM TL-5 cells displayed TNFa signaling but lacked evidence of wound healing, migration, or hypoxic responses at day 28 (Supplemental Figure 19, H and I, and Supplemental Table 6). IM ATL-0 cells showed wound healing responses at days 28 and 84 and TNF-α activation at day 28, but no hypoxia signatures were detected (Supplemental Figure 19, J and K, and Supplemental Table 7). These findings highlight widespread and persistent CD injury in the IM following R-UUO. CD injury is distinguished by a robust hypoxic response, likely reflecting adaptation to increased IM hypoxia.

Injured inner medullary CDs are hypoxic and have features of cell migration, wounding responses, apoptosis, and inflammation. GSEA of upregulated GO terms (A) and Hallmark terms (B) in injured CD-5 DEGs.

Long-term effects on cellular interactions in the renal medulla

To investigate how changes in CD, LOH, and fibroblasts affect cellular interactions in the IM after R-UUO, we analyzed ligand-receptor pairings between cell clusters (Supplemental Figure 20 and Supplemental Table 8) (66). The most notable finding was the de novo activation of the SPP1 pathway in injured LOH-4 and CD-5 cells. SPP1 signaling, mediated via integrins, promotes proinflammatory and profibrotic responses following AKI (67, 68). Other pathways that were affected included autocrine canonical Wnt signaling, which was observed in deep IM CD-0 cells, which express Wnt7b and 9b. These ligands were also expressed by injured LOH-4 and injured CD-5 cells after R-UUO. PdgfA signaling showed dynamic changes: in controls, PdgfA was restricted to IM ATL-0 cells, with its receptors (PdgfRA, PdgfRB) found in IM Fib-2 and Fib-6 clusters. After R-UUO, PdgfA expression shifted to injured LOH-4 cells at day 28, with receptors primarily in Myofib-0 cells. By 84 days, PdgfA expression returned to IM ATL-0 cells, but Myofib-0 remained the primary responder cells. Noncanonical Wnt signaling also showed altered cellular interactions. Wnt5A, secreted by Fib-1 and Fib-2 clusters in controls, targeted deep IM CD-0 responder cells. After R-UUO, responder cells shifted to Myofib-0 and injured LOH-4 clusters. These findings suggest that long-term activation of these pathways contributes to the inflammatory environment and promotes remodeling between days 28 and 84 after R-UUO. These altered interactions may disrupt normal tissue homeostasis, perpetuating cellular disorganization in the IM after R-UUO.

Common pathways are activated in IMs of patients with renal stone disease

To assess whether similar pathways are activated in patients with other IM diseases, we analyzed published snRNA-Seq data of renal papillary biopsies from patients with recurrent renal stone disease (69). Control samples were obtained from Nx patients or deceased donor kidneys unsuitable for transplantation. Two renal papillary principal cell populations, PapPC1 and PapPC2, were identified, resembling deep IM CD-0 cells in our studies. Significant overlap was found between DEGs in PapPC1 and -2 cells from patients with renal stone disease and DEGs in injured CD-5 and deep IM CD-0 cells 84 days after R-UUO (Supplemental Figure 21, A and B, and Supplemental Figure 22, A and B). GSEA showed enrichment for pathways related to wound healing, cell migration, apoptosis, and TNFa signaling in PapPC1 and -2 cells, consistent with findings in injured CD-5 and deep IM CD-0 cells after R-UUO (Supplemental Figure 21, C and D; Supplemental Figure 22, C and D; and Supplemental Tables 9 and 10). Notably, hypoxic pathway activation, a hallmark of IM CD injury following R-UUO, was absent in patients with renal stone disease. These findings indicate that shared pro-inflammatory, wound healing, and cell migration responses are activated in PapPC cells from patients with renal stone disease and IM CD cells after R-UUO. However, the hypoxic response observed in IM CD cells after R-UUO is not seen in patients with renal stone disease.