Tonicity drives collecting duct maturation in the mammalian kidney

Abstract

Organ maturation is a fundamental biological process and a major challenge for organoid-based regenerative medicine. During kidney maturation, osmolality increases in the renal medulla for urine concentration, yet whether this extreme environment conversely contributes to kidney maturation remains unclear. Here we show that high tonicity, i.e., salt, drives morphological, transcriptional, and functional maturation of medullary collecting ducts (CDs) primarily via nuclear factor of activated T cells 5 (NFAT5). This occurs both in vitro and in vivo. Combining high tonicity with maturation-promoting hormones, we establish a robust maturation protocol for human medullary CD organoids and a functional assay to measure water permeability in CD organoids, enabling modeling of three disease states: NFAT5 deletion, and lithium-induced and congenital nephrogenic diabetes insipidus. Collectively, high tonicity drives mammalian kidney maturation, offering a promising platform to maximize the potential of organoids for future therapies.

Organ maturation is an emerging focus in the fields of organoid and developmental biology. The authors show that high tonicity drives the maturation of renal collecting ducts in vitro and in vivo, improving morphology, gene expression, and function.

Introduction

How organs acquire their physiological functions, namely, organ maturation, is a fundamental question in biology¹. Drastic changes in organs and organisms occur during the perinatal period in mammals. For example, the fetus depends on the mother and the placenta for body fluid balance before birth, but the kidneys take over after birth. Newborns lack the capacity to produce concentrated urine; however, postnatal development enables more efficient water retention². The maximum urine osmolality in the adult is 4–10 times higher than the normal serum osmolality of 300 mOsm/kg³.

To concentrate urine, the kidney has a specialized region, the medulla, consisting of the medullary CDs, the loops of Henle, and an elaborate vascular network^3,4 (Fig. 1a). This unique structure allows salt (sodium chloride: NaCl) and urea to accumulate in the non-epithelial region, the medullary interstitium. The osmolality of kidney interstitial tissue gradually increases from the cortex to the inner medulla. While the osmolality is 300 mOsm/kg in the cortex, it potentially reaches levels similar to the most concentrated urine in the inner medulla: 1200 mOsm/kg in humans and 4000 mOsm/kg in mice^3,5. The urine reaching the cortical CDs is hypotonic because NaCl is actively reabsorbed in the preceding loops of Henle and distal tubules. The cortical CDs then reabsorb water, resulting in isotonic urine. As this urine flows through the medullary CDs in the hyperosmotic region, water is further reabsorbed osmotically, yielding the final concentrated urine. The medullary CDs represent the final gate that fine-tunes the body’s water balance. Water reabsorption is regulated by the hormone arginine vasopressin (AVP), also known as antidiuretic hormone. It acts on the AVP receptor 2 (AVPR2) in the CDs, inducing the translocation of the water channel aquaporin 2 (AQP2) to the luminal side of the CDs. This increases osmotic water reabsorption across the CD epithelia in coordination with AQP3 or AQP3/AQP4, which are located on the basolateral sides of human and mouse CDs, respectively^4,6. During renal maturation, the medulla develops and acquires these functional properties, enabling effective urine concentration. However, how this extreme hyperosmolality, in turn, contributes to kidney maturation is unclear.

Fig. 1. High tonicity matures the morphology of mCDs.

a Schematic diagram of a kidney and a close-up of the nephron and collecting duct. The collecting duct and loop of Henle run through the medulla. Osmolality is higher in the medulla than in the cortex. b The mCD maturation protocol. UBs isolated from E11.5 mouse kidneys were cultured for 9 days with the medium osmolality increased stepwise at a rate of 150 mOsm/kg per day. c Brightfield images. Lower panels: high magnification images of the parts enclosed by black rectangles in the upper panels. The lumen dilated at 750 mOsm/kg. Scale bars, 200 µm. d Hematoxylin and eosin staining. Scale bars, 50 µm. e–g Immunofluorescence staining of KRT8 and CDH16 (e), MUC1 (f), acetyl-alpha-tubulin (AcTUB), and ARL13B (g). White arrows indicate cilia positive for AcTUB and ARL13B (g). Scale bars, 10 µm (e), 20 µm (f), 6 µm (g). h, Cilia length in g (3 independent experiments; n = 16 for 300 mOsm/kg, n = 20 for 750 mOsm/kg). ^****P = 1.31 × 10⁻⁵; two-tailed t-test. Data are mean ± SEM.

Recapitulating early developmental events has enabled the in vitro generation of organoids from pluripotent stem cells (PSCs). While actively used for disease modeling and drug screening, their developmental immaturity¹ remains a major limitation. We and others have established differentiation protocols for nephron and CD organoids^7–15. Reportedly, hormones (AVP and aldosterone)^13–16 promote CD maturation, while shear stress¹⁷ and metabolic substrates¹⁸ mature the proximal tubules of the nephrons. Recently, spatially organized kidney assembloids have been developed by mixing different types of renal progenitors^19–21. However, their maturation status is still equivalent to the second trimester²¹ in humans and postnatal day (P) 0 in mice^19,21. In human organoids, AQP2^15,20 and AQP3¹³ were not clearly localized on the apical and basolateral sides, respectively. Another functional gene, AVPR2 was not detected in scRNA-seq analysis²⁰. Furthermore, the distinct genetic profiles of cortical and medullary CD cells^6,22,23 remain poorly characterized in both human and mouse CD organoids.

Here, we investigated the developmental role of high osmolality in mouse and human CDs. We provided evidence that salt drives medullary CD maturation in morphology, gene expression, and physiological function and that the maturation program works not only in vitro but also in vivo. We further leveraged mature human CD organoids to model three pathological conditions.

Results

Tonicity matures mouse CDs

To elucidate effects of high osmolality on CDs, we cultured ureteric buds (UBs; CD progenitors) isolated from embryonic day (E) 11.5 mouse kidneys in hyperosmotic medium supplemented with NaCl solution (750 mOsm/kg) (Fig. 1b, Supplementary Fig. 1a). Hyperosmolality resulted in luminal dilation and increased cell height and cytoplasmic area (Fig. 1c, d, Supplementary Fig. 1b–d). These features were prominent in the inner medullary CDs in P7 kidneys (Supplementary Fig. 1e). High osmolality upregulated polarized epithelial markers, including basolateral cadherin-16 and apical mucin 1, elongating apical primary cilia (Fig. 1e–h). Reported osmolality-induced genes^22,24 were upregulated in mouse CDs (mCDs) at high osmolality, reaching levels comparable to P7 CDs²⁵ (Supplementary Fig. 1f). However, Aqp2 was an exception with only a mild increase. Adding only water, instead of NaCl solution, did not elicit these changes, excluding the possibility that they were caused by dilution of the basal medium (Supplementary Fig. 2a–c). These changes were not exclusive to salt but were also evident with sucrose, while urea did not induce similar effects (Supplementary Fig. 2d–f). Unlike urea, which slowly diffuses across the cell membrane, non-permeating solutes (NaCl, glucose, sucrose) induce water movement, a property defined as effective osmolality or tonicity²⁶. Therefore, high tonicity is a developmental cue for mCD maturation. Reverting the medium tonicity, we demonstrated that the mature mCDs at high tonicity maintained a degree of plasticity (Supplementary Fig. 2g–j), suggesting that maturation is “neither terminal nor a one-way progression but a fluid continuum of reversible adaptive states¹”.

To explore transcriptional changes, we performed single-cell RNA sequencing (scRNA-seq) analysis of cultured mCDs (Fig. 2a, Supplementary Fig. 3a, b, Supplementary Data 1). In the uniform manifold approximation and projection (UMAP) plot, Cdh1+ epithelial cells had two subclusters: Krt19^-Ret⁺ UB tip (progenitors for CDs) and Krt19⁺Ret^- CDs. UB tip (Ret and Wnt11) and cell proliferation (Top2a and Mki67) markers were reduced at high tonicity; consistently, branching ceased at 750 mOsm/kg (Supplementary Fig. 3c–e), suggesting that normal tonicity is favorable for UB tip maintenance and proliferation. CDs consist primarily of Elf5⁺Aqp2⁺ principal cells (PCs) for water and salt homeostasis^27,28, alongside intermingled Foxi1⁺ intercalated cells (ICs) responsible for pH homeostasis^29,30. Since cultured CD cells were positive for Elf5, but not for Foxi1 (Supplementary Fig. 3f), they were mainly composed of PC precursors. Comparing the transcriptomes of CDs at normal and high tonicity (Fig. 2a), we identified a list of genes upregulated by high tonicity (3,942 genes), covering diverse functional categories: transcription factors, ligands/receptors, channels/transporters, metabolic enzymes, and extracellular matrix/cell adhesion molecules (Fig. 2b, Supplementary Fig. 4a–e, Supplementary Data 2). This list represents 43.1% (683 genes) of the medullary CD maturation markers obtained by re-analyzing the E15.5 and P7 kidney datasets (Fig. 2c, Supplementary Fig. 4f–i, Supplementary Data 3, 4)^31,32. Over-representation analysis on these 683 genes (GO terms and the Molecular Signature Database³³) revealed enrichment of body fluid regulation, CD development, and hypoxia (Supplementary Fig. 4j, k), indicating that high tonicity is a key developmental signal for mCDs. Additionally, hypoxia-inducible factor 1-alpha (Hif1a) and several hypoxia-related genes encoding glycolytic enzymes and a glucose transporter were upregulated by high tonicity (Fig. 2b). Given that hypoxia is another unique feature of the renal medulla³⁴, high tonicity is likely to endow mCDs with hypoxia-induced transcriptional activity even under normoxia. Furthermore, unbiased hierarchical clustering revealed that high tonicity shifted the transcriptional profile of mCDs toward that of P7 medullary PCs, whereas mCDs at normal tonicity remained closer to that of E15.5 immature PCs (Fig. 2d, Supplementary Data 5). Collectively, mCDs underwent global transcriptional maturation at high tonicity.

Fig. 2. High tonicity transcriptionally and functionally matures mCDs.

a UMAP plot of scRNA-seq analysis. Gene expression in CD clusters (arrows) was analyzed. b The genes upregulated by high tonicity covered broad biological categories, including transcription factors (TFs), signal ligands (Ligands) and receptors (Receptors), channels and transporters, metabolic enzymes, and extracellular matrix and cell adhesion molecules (ECM/CAM). For Aqp2 and Aqp4, rescaled plots are shown at the bottom right, surrounded by a black rectangle. c The contribution of high tonicity to medullary CD maturation markers. Of the 1583 maturation markers, 683 genes (43.1%, pink) were upregulated by high tonicity. d Unbiased hierarchical clustering of mCDs at 300 and 750 mOsm/kg (mCD 750 and mCD 300), E15.5 immature PCs (E15), and P7 medullary PCs (P7). e AVP-responsiveness. AQP2 (green) accumulated on the apical side of the epithelium in the presence of AVP at 450 mOsm/kg, not at 300 mOsm/kg. The protocol for 450 mOsm/kg is described in Supplementary Fig. 6c. Right panels: high magnification images of the parts enclosed by white rectangles in the left panels. Scale bars, 200 µm (low magnification) and 20 µm (high magnification). f Experimental procedure to study the urea tolerance. They were exposed to 450 mM urea after the maturation protocol and examined on Day 10. g Brightfield images of mCDs before and after exposure to urea. Scale bars, 200 µm. h The production of PGE₂ (n = 3 biological replicates). ^*P = 0.0346; two-tailed t-test. Data are mean ± SEM. IS internal standard.

In cultured CDs, stromal cells remained due to technical difficulty in their complete removal (Fig. 2a). These heterogenous populations included Wnt4+ medullary stroma-like cells³⁵, which exhibited higher levels of medullary stroma maturation markers at high tonicity (Supplementary Fig. 5, Supplementary Data 6–9), suggesting that medullary stroma is likely to mature at high tonicity as well.

One of the major physiological functions of CDs is water reabsorption, which is mediated by AVP-induced translocation of AQP2 to the apical membrane^3,4. Since in situ hybridization confirmed Avpr2 upregulation by high tonicity (Fig. 2b, Supplementary Figs. 1f, 6a), we added AVP to our maturation protocol, finding that AQP2 accumulated on the apical side at high tonicity, but not at normal tonicity; interestingly, AQP2 expression was more robust at 450 mOsm/kg than at 750 mOsm/kg (Fig. 2e, Supplementary Fig. 6b–d). CDs also transport urea via urea transporter A (UT-A)^3,4. In situ hybridization confirmed that high tonicity increased Slc14a2, encoding UT-A (Supplementary Fig. 1f, 4c, 6e). Given that our scRNA-seq datasets^31,32 revealed that Slc14a2 increased in medullary CDs during perinatal development (Supplementary Fig. 4h, i), high tonicity likely enhances UT-A expression, leading to medullary urea accumulation during kidney maturation. Medullary CDs tolerate high urea levels by accumulating organic osmolytes and heat shock proteins (HSPs)²⁶; consistently, high tonicity upregulated osmoprotective genes (Supplementary Fig. 4c). Then, we exposed mCDs to 450 mM urea (Fig. 2f), finding that mCDs at high tonicity maintained their structure, while those at normal tonicity did not (Fig. 2g). Additionally, CDs produce prostaglandin E₂ (PGE₂) for sodium and water homeostasis²⁸. High tonicity upregulated Ptgs1 and Ptges2, encoding PGE2 biosynthesis enzymes: cyclooxygenase-1 (COX-1) and PGE₂ synthase 2, respectively³⁶ (Supplementary Fig. 4d). Consistently, mass spectrometry revealed higher PGE₂ production in mCDs at high tonicity (Fig. 2h, Supplementary Fig. 6f). Taken together, mCDs cultured at high tonicity acquired key physiological functions: hormone response, osmo-adaptation, and metabolic activity.

NFAT5 regulates mCD maturation

NFAT5, also known as tonicity-responsive enhancer binding protein (TonEBP), is the only known transcription factor that transduces extracellular hypertonic stress into the downstream signaling pathway³⁷. Despite ubiquitous Nfat5 mRNA expression in our scRNA-seq (Supplementary Fig. 7a), NFAT5 protein was expressed in the nuclei of epithelial and stromal cells in the medulla, but not in the cortex, of P7 kidneys (Fig. 3a). The NFAT5+ epithelial cells expressed AQP4, a hallmark of medullary CDs (Fig. 3b). Consistently, NFAT5 accumulated in the nuclei of mCDs at high tonicity (Fig. 3c, d, Supplementary Fig. 7b, c). To investigate its role in the maturation process, we deleted Nfat5 specifically in UB/CDs^38,39, followed by the culture of E11.5 Nfat5 knockout (KO) UBs (Fig. 3e). Even at 600 mOsm/kg, the growth of Nfat5 KO mCDs was impaired with epithelial disorganization (Fig. 3f, Supplementary Fig. 7d–g), which is consistent with a previous report that Nfat5 KO mice have impaired medullary development in the kidney⁴⁰.

Fig. 3. NFAT5 regulates the mCD maturation.

a NFAT5 in the P7 cortex and medulla. The signal was found in the nuclei of the CDs (KRT8⁺) and the surrounding stromal cells (KRT8^-) in the medulla. The signal was negative in the cortex. Scale bars, 20 µm. b NFAT5 in the P7 medulla. The signal was found in the nuclei of the PCs (AQP4⁺). Scale bars, 20 µm. c, d NFAT5 in cultured mCDs. Scale bars, 20 µm. The nuclear-to-cytoplasmic (N/C) ratio of the signal intensity (n = 15 cells for each condition from 3 independent experiments). NFAT5 accumulated in the nuclei at 750 mOsm/kg. The signal was weak and widespread in both the nuclei and cytoplasm at 300 mOsm/kg. ^****P = 0.000003; two-tailed t-tests. Data are mean ± SEM. e Experimental procedure to study the role of NFAT5 in the maturation. E11.5 Nfat5 KO UBs were cultured with the final tonicity of 600 mOsm/kg. f Bright field images of control and Nfat5 KO mCDs at 600 mOsm/kg. The growth of KO mCDs was impaired. Scale bars, 400 µm. g, h scRNA-seq analysis of cultured mCDs. Cdh1⁺Pdgfra^- epithelial cells had Krt19⁺Ret^- CDs (arrows) and Krt19^-Ret⁺ UBs (g). CDs were further divided into Nfat5-deleted cells identified only in KO mCDs (arrowheads) and Nfat5-expressing cells identified in control and KO mCDs (arrows) (h). i Volcano plots illustrating differentially expressed genes in the comparison of Nfat5-expressing cells in control mCDs (arrows) and Nfat5-deleted cells in KO mCDs (arrowheads) in (h). The osmo-regulated genes shown in Supplementary Fig. 1f were labeled. Unadjusted P-values were used to maintain statistical sensitivity throughout this study. j Overview of the NFAT5 regulation. High tonicity activates NFAT5-dependent (red) and -independent (blue) pathways. Different transcription factors were activated in each pathway. Genes associated with ion and molecule transport and osmotic protection were mainly regulated by NFAT5, while the others were regulated by both pathways. k The contribution of NFAT5 to the tonicity-induced maturation. Of the 693 tonicity-induced medullary CD maturation markers (pink in Fig. 2c), 379 genes (55.5%, pink) were regulated by NFAT5.

We performed scRNA-seq on control and Nfat5 KO mCDs at high tonicity, using a customized reference genome to detect Cre and floxed alleles (Supplementary Fig. 7h, i). In the UMAP space, epithelial cells (Cdh1⁺Pdgfra⁻) had two Krt19⁺Ret⁻ CD subclusters: Nfat5-expressing (Cre⁻Nfat5-exon 5⁺) and Nfat5-deleted (Cre⁺Nfat5-exon 5⁻) cells (Fig. 3g, h, Supplementary Fig. 7j, Supplementary Data 10). We compared Nfat5-expressing cells of control mCDs and Nfat5-deleted cells of KO mCDs, identifying thousands of differentially expressed genes in the Nfat5-deleted cells (Fig. 3i, Supplementary Data 11, 12). NFAT5 regulated 35.3% (1390 of 3942) of the genes induced by high tonicity (Fig. 2c, Supplementary Data 2), suggesting that the tonicity-induced maturation is mediated by NFAT5-dependent and -independent pathways (Fig. 3j). In particular, NFAT5 regulates 55.5% (379 of 683) of the tonicity-induced maturation markers (Fig. 3k). At 300 mOsm/kg, most marker genes remained unaffected by NFAT5 deletion, except for Akr1b3 (Supplementary Fig. 7k), suggesting that NFAT5 mainly regulates gene expression at high tonicity. Taken together, NFAT5 is a major regulator of the tonicity-induced CD maturation.

Tonicity is involved in the CD maturation in vivo

High tonicity in the renal medulla is established by Na-K-2Cl cotransporter (NKCC2, encoded by Slc12a1) in the thick ascending limb (TAL), the distal part of the loop of Henle^3,4 (Fig. 4a). To investigate the physiological role of high tonicity in CD maturation in vivo, we reduced NaCl build-up by disrupting the function of SLC12A1 in neonatal mice. Genetic deletion of Slc12a1 led to severe hydronephrosis at P0 and P7 (Supplementary Fig. 8a, b), which was consistent with a previous report⁴¹. Treating neonatal mice with furosemide, an SLC12A1 inhibitor, we bypassed confounding secondary effects of hydronephrosis. Furosemide-treated P7 mice exhibited impaired development of the inner medulla, although the cellular morphology was largely unaffected (Fig. 4b, Supplementary Fig. 8c). Reduced NFAT5 signal in the inner medulla indicates that furosemide impaired NaCl accumulation (Fig. 4c, d). Maturation markers upregulated by high tonicity in our in vitro experiments were reduced in the inner medullary CDs of furosemide-treated mice (Fig. 4e–i). Wnt4, Hif1a, and Col4a3 were expressed in both the epithelia and the stroma; notably, Hif1a and Col4a3 levels were reduced across both compartments, suggesting that high tonicity is required for the maturation of both the medullary CDs and the stroma. Although Slc14a2 and Avpr2 were also reduced, Aqp2 was upregulated in furosemide-treated mice (Fig. 4j, k). AVP secretion may have been augmented to compensate for severe dehydration due to impaired urine concentration^41,42, leading to enhanced Aqp2 expression. Collectively, high tonicity contributes to the maturation of the CDs in the inner medulla in vivo.

Fig. 4. NaCl depletion impairs medullary maturation.

a Schematic diagram of the nephron and collecting duct. SLC12A1 in the thick ascending limb of the Loop of Henle uptakes NaCl into the interstitial space of the medulla, resulting in NaCl accumulation in the medulla. b Furosemide-treated mice had impaired medullary elongation. Scale bars, 1 mm. c Immunostaining of NFAT5 in P7 kidneys of mice treated with DMSO and furosemide. The signal intensity in the medulla was reduced in the furosemide-treated kidney, suggesting that the medication disrupted NaCl accumulation in vivo. Scale bar, 500 µm. d High magnification images of the parts enclosed by white rectangles in (c). Scale bar, 250 µm. e Immunostaining of MUC1 in the P7 kidney of mice treated with DMSO and furosemide. The signal intensity in the inner medulla was reduced in the furosemide-treated kidney. Scale bars, 1 mm. f In situ hybridization of Wnt4, Wnt7b, Hif1a, and Col4a3 in P7 kidney of mice treated with DMSO and furosemide. Their expression levels in the inner medulla were reduced in the furosemide-treated kidney. Dot lines: borders between the inner and outer medulla. Scale bars, 1 mm. g–i High magnification images of the parts enclosed by white squares in (f). Aqp2 expression is shown to distinguish CDs and stromal cells. Hif1a (h) and Col4a3 (i) were reduced in both the epithelial and stromal cells of the furosemide-treated kidneys. Wnt4 (g) was reduced only in the epithelial cells. Scale bars, 250 µm. In situ hybridization of Aqp2, Slc14a2, and Avpr2 in P7 kidney of DMSO-treated and furosemide-treated mice (j). High magnification images of the parts enclosed by white squares are in (k). Slc14a2 and Avpr2 were downregulated in the furosemide-treated kidney, whereas Aqp2 was upregulated. Scale bars, 1 mm in (j), 250 µm in (k). b–k At least three biologically independent experiments were performed, and 1–3 sections per condition were examined in each experiment.

Tonicity and hormones cooperatively mature mCDs to the postnatal stage

While high tonicity evoked global transcriptional changes, Aqp2 was an exception with only a mild increase (Supplementary Fig. 1f). Since AQP2 is involved in water reabsorption^3,4, it is essential to upregulate this gene to ensure functional maturation. Hormones (AVP and aldosterone) reportedly induce Aqp2 expression in mouse and human CD organoids^13–16. Consistently, hormone supplementation yielded slightly higher Aqp2 expression than the tonicity-based protocol, though the difference was not statistically significant; however, high tonicity, combined with the hormones, significantly upregulated Aqp2, almost reaching the level of P7 CDs (Fig. 5a). However, the hormones alone did not upregulate other maturation markers except for Slc14a2 (Supplementary Fig. 9a), suggesting that tonicity and hormones cooperatively mature mCDs. Interestingly, Aqp2 expression was higher at 450 mOsm/kg than at 750 mOsm/kg under this combined protocol (Supplementary Fig. 9b), which is consistent with higher AVP responsiveness at 450 mOsm/kg than at 750 mOsm/kg (Fig. 2e, Supplementary Fig. 6d). In contrast, other medullary CD maturation markers, such as Slc14a2, Hif1a, and Col4a3, exhibited higher levels at 750 mOsm/kg than at 450 mOsm/kg (Supplementary Fig. 9c). In P7 kidneys, Aqp2 expression was relatively high in the outer region of the inner medulla and weak in the innermost region (papilla), whereas Slc14a2 expression was strong in the papilla, where osmolality is highest (Supplementary Fig. 9d), which is consistent with a previous report²³. These data suggest that 450 and 750 mOsm/kg may induce different domains of medullary CDs.

Fig. 5. Tonicity and hormones cooperatively mature mCDs to the postnatal stage.

a Expression of Aqp2 in mCDs at 300 and 750 mOsm/kg in the absence and presence of hormones (H: AVP and aldosterone) (n = 6 biological replicates). P7: GFP+ CDs sorted from P7 Hoxb7-GFP mice²⁵, presented as a reference. Aqp2: ns (not significant), P = 0.9909 (300 vs. 300 + H); ns, P > 0.9999 (300 vs. 750); ^***P = 0.0005 (300 vs. 750 + H); ns, P = 0.9915 (300 + H vs. 750); ^**P = 0.0010 (300 + H vs. 750 + H); ^***P = 0.0005 (750 vs. 750 + H). P values were calculated using a two-sided one-way ANOVA followed by Tukey’s multiple comparison tests. Data are mean ± SEM. b, c scRNA-seq analysis of epithelial populations of mCDs cultured at 300 mOsm/kg and at 750 mOsm/kg in the absence and presence of the hormones. They had CD and UB tip clusters (b), with expression of Avpr2 and Ret, respectively (c). Genes used for cluster annotation are presented in Supplementary Fig. 9e and Supplementary Data 13. d Dot plots of the genes upregulated by high tonicity alone (T), hormones alone (H), and both high tonicity and hormones (T + H). e, f Integrated scRNA-seq analyses of CD cells in mCDs (750 + H) and P7 medullary CDs. The CD cluster of the organoids (750 + H) overlapped with P7 inner medullary PC (e) and was labeled as P7 inner medullary PC (f). g Feature plots of Aqp2, Avpr2, Slc14a2, and Akr1b3 in the scRNA-seq analysis in (f). h Violin plots of medullary CD maturation markers in the organoid CD cells (750 + H) and P7 inner medullary PC (P7 medulla).

We performed scRNA-seq of mCDs at 750 mOsm/kg with the hormones, and analyzed epithelial clusters (Fig. 5b, c, Supplementary Fig. 9e, Supplementary Data 13). Comparison with the dataset of mCDs at high tonicity (Fig. 2a) identified hormone-responsive genes (Supplementary Data 14). High tonicity and the hormones had different targets, upregulating 43.1% and 11.0% of the medullary PC maturation markers, respectively (Fig. 5d, Supplementary Data 15). Only 7.0% were shared, but water-related transporters/channels (Aqp2, Aqp4, and Slc14a2) were regulated by both high tonicity and the hormones. In contrast, medullary-associated genes involved in hypoxia (Hif1a³⁴) and development (Wnt4⁴³ and Wnt7b⁴⁴) were mainly regulated by high tonicity alone. Integrated with the P7 dataset³², CD populations cultured at 750 mOsm/kg with the hormones overlapped with medullary PC, not cortical PC (Supplementary Fig. 9f, g). For further characterization, we here performed scRNA-seq of P7 kidney medulla (Supplementary Fig. 9h). Cultured mCDs mapped to P7 inner medullary PCs, exhibiting comparable maturation marker expression (Fig. 5e–h, Supplementary Fig. 9i, Supplementary Data 16). Taken together, high tonicity and hormones cooperatively promoted the maturation status to the postnatal stage.

Tonicity and hormones create mature human CD organoids that respond to AVP

We have previously reported an induction protocol of UB organoids from human induced PSCs (iPSCs)¹². Here, we applied the mouse maturation protocol to the UB organoids (Fig. 6a). Because they were immature and sensitive to high tonicity, we added forskolin (FSK), an adenylyl cyclase activator, to elongate UB stalk⁴⁵, and increased the tonicity more slowly during the extended culture period (Fig. 6a, Supplementary Fig. 10a). FSK enabled the organoids to tolerate high tonicity and maintain better morphology (Supplementary Fig. 10b). CD organoids cultured at 900 mOsm/kg exhibited luminal dilation and increased cell height and cytoplasmic area (Fig. 6b–f), with KRT8 localized on the apical side (Supplementary Fig. 10c), as in vivo⁴⁶. High tonicity significantly upregulated PC markers, including AQP2 and AVPR2 (Supplementary Fig. 10d).

Fig. 6. Tonicity and hormones create mature human CD organoids that respond to AVP.

a A tonicity-based maturation protocol for human CD organoids. UB organoids were cultured with FSK, while the osmolality was increased stepwise to 900 mOsm/kg. b Brightfield images of CD organoids cultured with FSK at normal (left) and high (right) tonicity. Lower panels: high magnification images of the parts enclosed by black rectangles in the upper panels. The lumen dilated at high tonicity. Scale bars, 200 µm. c Luminal diameter in b (3 independent experiments; n = 27 for each condition). ^****P = 2.0 × 10⁻¹⁴; two-tailed t-test. Data are mean ± SEM. d Hematoxylin and eosin staining. Scale bars, 50 µm. Cell height (e) and cytoplasmic area (f) in d (3 independent experiments; n = 24 for FSK, n = 25 for FSK + Tonicity in e; n = 27 for each condition in f). ^****P = 6.8 × 10⁻⁹ (e), 3.2 × 10⁻⁹ (f); two-tailed t-tests. Data are mean ± SEM. g A hormone-based protocol (H) involved removal of branching factors (BFs) and addition of AVP and aldosterone. A combined protocol (T + H) integrated the tonicity-based protocol in a with the hormone-based protocol (H). h Gene expression of CD organoids cultured with the hormone-based (H) and the combined (T + H) protocols (n = 3 biological replicates). A bulk sample of adult human kidney was used as a reference (Adult). Asterisks indicate statistical significance. Exact P values are provided in the Source Data; two-tailed t-tests. Data are mean ± SEM. i, j AQP2 and AQP3 were localized on the apical and basolateral sides in CD organoids cultured with the combined protocol (T + H), respectively, but not with hormones alone (H). Scale bars, 25 µm. k Experimental procedure to study the responsiveness to ddAVP. After the combined maturation protocol, ddAVP was added, while the hormones (H) and FSK were withdrawn. l CD organoids induced with the combined protocol responded to ddAVP with AQP2 on the apical membrane, while those with hormones alone did not. Scale bars, 25 µm.

Existing CD protocols rely on removing branching factors such as GDNF and retinoic acid and adding AVP and aldosterone^13–16. We thus combined our tonicity-based maturation protocol with the hormones without branching factors (Fig. 6g). This combination increased CD maturation markers, to levels comparable to or exceeding bulk adult human kidney (Fig. 6h, Supplementary Fig. 10e). Whereas AQP2 protein was weakly expressed under the hormone-alone condition, its expression was robustly enhanced with clear localization on the apical membrane of the organoids cultured with the combined protocol (Fig. 6i, Supplementary Fig. 10f, g), resembling adult human kidney (Supplementary Fig. 10h). Basolateral AQP3 was also enhanced with the combined protocol (Fig. 6j, Supplementary Fig. 10i). Of note, AQP4 is not detectable in human CDs, at least according to the currently available datasets^6,47,48. Furthermore, AVPR2 was increased approximately 120-fold with the combined protocol (Fig. 6h). Since AVPR1A is expressed in our UB organoids⁴⁵, we used the AVPR2-selective synthetic analog 1-deamino-8-D-arginine vasopressin (ddAVP), rather than the non-selective native ligand AVP, to accurately assess AVPR2-mediated functionality, following the maturation culture (Fig. 6k). We simultaneously withdrew all of the factors (AVP, aldosterone, and FSK) because FSK and AVP stimulate cAMP signaling and may confound ddAVP responses. After washing out the factors, CD organoids lost AQP2 expression; however, ddAVP enhanced apical AQP2 in combined-protocol organoids, while hormone-protocol organoids did not (Fig. 6l, Supplementary Fig. 10j), indicating that combined-protocol organoids acquired responsiveness to ddAVP. Taken together, the tonicity-based protocol, when combined with hormones, further matures human CD organoids.

The mature human CD organoids potentially reach the adult stage

scRNA-seq analysis of CD organoids cultured in various conditions: hormones-alone (H: AVP and aldosterone), tonicity-alone (T), and combined protocols (T + H), integrating with the dataset of Shi et al., which utilized a hormone-based protocol (Shi)¹⁵, revealed four subclusters: CD cells (CDH1 + AQP2+), ureter-like cells (CDH1 + UPK2+), stromal cells (CDH1-PDGFRA+), and proliferating cells (TOP2A+) (Fig. 7a, b, Supplementary Fig. 11a–c, Supplementary Data 17). Based on ELF5⁺/AQP2⁺/FOXI1⁻ signatures, CD cells were primarily composed of PCs, rather than ICs. Combined-protocol organoids exhibited higher expression levels of CD maturation markers than the other three protocols (Fig. 7c). AQP2 was expressed in 97.0% of the CD cells in combined-protocol organoids, while the percentage of AQP2+ cells was much lower in Shi et al. dataset and under hormone-alone and tonicity-alone conditions (38.3%, 2.7%, 52.3%, respectively) (Supplementary Fig. 11d). Considering that our UB organoids expressed AQP2 after 10 days of branching culture⁴⁵ and that Shi et al. cultured their organoids for only one week¹⁵, weak AQP2 expression in our hormone-alone condition was likely due to the prolonged 20-day culture period rather than technical variation. AVPR2 expression was also highest in combined-protocol organoids, while it was almost undetectable in the other conditions (Supplementary Fig. 11e).

Fig. 7. The mature CD organoids potentially reach the adult stage.

a scRNA-seq analysis of CD organoids induced with the protocol by Shi et al.¹⁵ (Shi), the hormones alone (H), high tonicity alone (T), and high tonicity plus the hormones (T + H). In our protocols (H, T, and T + H), ureter-like cells were minor portions, while their number was relatively high in the Shi protocol. The number of stromal cells and PDGFRA and PDGFRB levels were reduced in the organoids generated by the combined protocol. The proliferating cells could not be defined as epithelium or stroma because they were positive for both CDH1 and PDGFRA. b Feature plots of ELF5, AQP2, and AVPR2. The combined protocol (T + H) significantly increased AQP2 and AVPR2. c Violin plots of medullary CD maturation markers in the CD clusters in (a). The combined protocol (T + H) had significantly higher expression levels of these markers than the other three conditions. d Integrated analysis of CD cells in the human CD organoids (T + H) and kidney assembloids of Huang et al.²¹ and Shi et al.²⁰ ELF5 + CD populations of our mature CD organoids and kidney assembloids were extracted from the original datasets and re-analyzed. e Feature plots of AQP2, AQP3, and PTGS1. The combined protocol (T + H) significantly increased these genes. f Violin plots of CD maturation markers in the CD clusters of our mature CD organoids (T + H) and the kidney assembloids^20,21. The combined protocol (T + H) had significantly higher expression levels of these markers than the assembloids. g Dot plots of the top 50 medullary PC-specific genes identified in the KPMP dataset⁶. This list included SLC14A2, PTGS1, and HIF1A. Our mature CD organoids (T + H) had significantly higher expression levels of these genes than the kidney assembloids^20,21. h Violin plots of CD maturation markers in the organoid CD cluster (T + H) and adult human medullary PC (Adult). The organoid CD cluster exhibited comparable expression levels of these markers to adult human medullary PC.

Kidney assembloids, recently developed by mixing several types of kidney progenitors^19–21, are still equivalent to a second-trimester kidney²¹. Compared with ELF5 + CD populations of these kidney assembloids^20,21, those of our combined-protocol organoids exhibited higher expression levels of representative CD genes (Fig. 7d-f, Supplementary Fig. 11f). However, the Kidney Precision Medicine Project (KPMP) datasets showed that many of these genes were not specific to medullary PCs⁶. Then, we selected the top 50 medullary PC-specific markers, including SLC14A2, PTGS1, and HIF1A, from the KPMP datasets⁶, revealing higher expression of these genes in our organoids (Fig. 7g).

In addition, integrated with a dataset of adult human kidney, CD cells in our combined-protocol organoids were classified as the same cluster as human medullary PC, which was positive with medullary PC-specific markers (SMIM22 and LAMC2)⁶ and a broader PC marker (AQP2) (Supplementary Fig. 11g–j, Supplementary Data 18). Marker expression levels in our organoids were comparable to those of adult medullary PCs (Fig. 7h).

Altogether, our tonicity-based approach plus hormones matures human CD organoids, potentially bringing them closer to an adult kidney. However, multistage in vivo data remain necessary to precisely define CD organoid maturation stages.

Mechanistic study and disease modeling using the mature human CD organoids

Finally, we utilized mature human CD organoids for mechanistic study and disease modeling. We generated NFAT5 KO and isogenic control iPSC lines, inducing combined-protocol organoids from these lines (Supplementary Fig. 12a–c). NFAT5 KO CD organoids could not tolerate 900 mOsm/kg (Fig. 8a), so medium osmolality was elevated to 700 mOsm/kg in the subsequent experiments (Supplementary Fig. 12d, e). NFAT5 protein was not detected in KO organoids (Fig. 8b, Supplementary Fig. 12f). Several human homologs of NFAT5-regulated medullary CD markers in mice were downregulated in KO organoids, whereas PAX2 and PAX8 were rather upregulated (Supplementary Fig. 12g), possibly reflecting the species or stage differences. KO organoids showed significantly reduced apical AQP2 protein and mRNA levels (Fig. 8c, Supplementary Fig. 12g, h). Variants in NFAT5 are reportedly associated with serum osmolality and hypertension in genome-wide association studies (GWAS)^49,50. Given that serum osmolality is determined by water reabsorption in CDs, variants in NFAT5 likely diminish AQP2 expression and thereby reduce water reabsorption, leading to increased serum osmolality. Additionally, SCNN1A and SCNN1G, encoding the epithelial sodium channel (ENaC) subunits, were significantly increased in KO organoids only at high tonicity (Fig. 8d, Supplementary Fig. 12i), suggesting that NFAT5 variants may increase ENaC expression, resulting in salt accumulation and hypertension²⁸. This is consistent with a previous report that renal epithelium-specific Nfat5-KO mice show salt-sensitive hypertension because of upregulated ENaCs⁵¹.

Fig. 8. Disease modeling in the mature CD organoids.

a Brightfield images of control (Ctrl) and NFAT5 KO organoids at 900 mOsm/kg. KO organoids degraded at 900 mOsm/kg. Scale bars, 200 µm. b, c NFAT5 was undetectable (b), and AQP2 was reduced in NFAT5 KO organoids at 700 mOsm/kg (c). Scale bars, 25 µm. d SCNN1A and SCNN1G were upregulated in NFAT5 KO organoids (n = 5 biological replicates). e and f Reduced AQP2 and AQP3 in LiCl-treated organoids. Scale bars, 25 µm. g Time-lapse phase-contrast imaging of control and LiCl-treated organoids at 0 and 9 min after the osmotic change. Scale bars, 200 µm. h and i Time-course of organoid volume changes following a relative hypotonic stimulus (h). Control organoids exhibited larger volume changes than LiCl-treated organoids. For each condition, 20 regions were selected from 4 independent experiments. Area under the curve (AUC) for each regional volume change was calculated (i). The cumulative volume response represented by AUC was higher in control organoids. j Position of an amino acid substitution (Q174R) that causes congenital NDI. The amino acid is located in one of the transmembrane domains. k Experimental procedure to study ddAVP responsiveness in AVPR2^WT and AVPR2^Q174R CD organoids. After the combined maturation protocol (T + H), ddAVP was added, while hormones (H) and FSK were withdrawn. l Impairment of ddAVP responsiveness in AVPR2^Q174R organoids. Note the absence of AQP2 (green) on the luminal membrane. Scale bars, 25 µm. m, n Time-course of organoid volume changes after a relative hypotonic stimulus. AVPR2^WT organoids exhibited larger volume changes than AVPR2^Q174R organoids. For each condition, 20 regions were selected from 3 independent experiments. AUC for each regional volume change in m was calculated in n. The cumulative volume response represented by AUC was significantly higher in AVPR2^WT organoids. In all panels, asterisks indicate statistical significance. Exact P values are provided in the Source Data. P values were calculated using a two-tailed t-test (i), two-way repeated measures ANOVA followed by Sidak’s (h) and Tukey’s (m) multiple comparisons test at each time point, and a two-sided one-way ANOVA followed by Tukey’s multiple comparison tests (d, n). Data are mean ± SEM.

Patients with nephrogenic diabetes insipidus (NDI) suffer from polyuria due to impaired urine concentration in the CDs, a condition that can be inherited or acquired⁵². Among the acquired forms, lithium, a medication commonly used to treat psychiatric disorders, is the most frequent causative agent; it disrupts water reabsorption in the CDs, primarily by reducing AQP2 expression⁵². We treated combined-protocol CD organoids with lithium chloride (LiCl) for the final 6 days of culture. This treatment significantly reduced AQP2 and AQP3, while CDH1 expression was maintained (Fig. 8e, f, Supplementary Fig. 13a, b), indicating that the downregulation of AQPs was unlikely due to non-specific cytotoxicity. To assess the functional impact of LiCl, we developed an assay to measure water permeability in human CD organoids. Water permeability via AQP channels has been quantified by measuring cellular volume changes in response to osmotic gradients^53,54. However, these methods utilize isolated single oocytes or two-dimensional (2D) culture systems and have not been applied to 3D organoids. We first exposed untreated organoids to a medium of 450 mOsm/kg and monitored them using time-lapse imaging, quantifying the volume changes over time. Given that the organoids were pre-cultured at 900 mOsm/kg, the introduction of a 450 mOsm/kg medium served as a relative hypotonic challenge. The organoids, which highly expressed AQP2 and AQP3, started to swell immediately following the osmotic change, peaking at 9 min with a 25.9% increase in volume (Fig. 8g–i, Supplementary Fig. 13c, d, Supplementary Movie 1). This indicates that water influx occurred into the epithelial cells along an osmotic gradient, leading to the swelling of the entire organoids. In contrast, LiCl-treated organoids exhibited a blunted swelling, peaking at 6 min with a 12.0% increase in volume, suggesting the reduced water permeability in LiCl-treated organoids. Collectively, LiCl treatment compromised the water permeability function of human CD organoids, possibly due to reduced AQP expression.

To further assess the functional importance of AQP2, we expanded our investigation to a genetic model: congenital NDI, which is primarily caused by AVPR2 mutations that lead to an impaired response to AVP⁵². AVPR2 is on the X chromosome, and this disease is inherited in an X-linked recessive manner. We generated male iPSC lines with a missense mutation (g.882A>G) in AVPR2, resulting in a pathogenic amino acid substitution of Q174R in the transmembrane domain of AVPR2 (AVPR2^Q174R), which has been reported to cause severe congenital NDI⁵⁵ (Fig. 8j, Supplementary Fig. 13e–g). Combined-protocol CD organoids induced from isogenic control (AVPR2^WT) and AVPR2^Q174R lines were exposed to ddAVP along with washout of the other factors (Fig. 8k). Control organoids showed AQP2 accumulation on the apical side of the epithelia, while AVPR2^Q174R organoids did not (Fig. 8l, Supplementary Fig. 13h). This finding is consistent with a previous case report showing reduced apical AQP2⁵⁶. However, available histological data are limited, because this disease is routinely diagnosed by genetic testing rather than renal biopsy⁵². Furthermore, we measured water permeability in these organoids (Fig. 8m, n). The volumes of both control and AVPR2^Q174R organoids peaked at 9 min following the osmotic stress. Consistently across different clones, AVPR2^Q174R organoids exhibited blunted responses compared with control organoids; AVPR2^WT organoids exhibited 30.7% and 29.5% increases at 9 min, while AVPR2^Q174R organoids showed only 18.8% and 17.2%. The reduced volume changes in AVPR2^Q174R organoids suggest that the absence of apical AQP2 led to lower water permeability. These data suggest that these organoid phenotypes likely mimic the pathophysiology in patients with congenital NDI.

Taken together, these data prove the robustness of our tonicity-induced maturation protocol and demonstrate that our mature CD organoids can provide a platform to gain valuable insights into a broader range of human kidney diseases.

Discussion

Our study demonstrated that high tonicity (salt) promotes the maturation of medullary CDs in both mice and humans. Our tonicity-based protocol matured CD cells to at least the postnatal stage, and potentially to the adult stage via NFAT5 activation, combined with hormones. High tonicity endowed CD organoids with distinct medullary characteristics, recapitulating the cellular morphology and gene expression profiles of the native medullary CDs. High tonicity upregulated both general PC markers (Aqp2 and Avpr2) and medullary PC-specific markers (Slc14a2, Hif1a, and Ptgs1). In scRNA-seq, mCDs cultured at high tonicity with hormones aligned with and exhibited expression levels comparable to P7 inner medullary PCs. Tonicity and hormones shared some targets while regulating others independently, cooperatively promoting renal maturation. Human CD organoids cultured at high tonicity with hormones also expressed higher levels of general PC markers (AQP2 and AQP3) and medullary PC-specific markers (SLC14A2 and PTGS1) than conventional CD organoids^13,15 and kidney assembloids^20,21. Although organ maturation requires multiple factors, the combination of high tonicity and hormones markedly promoted kidney maturation.

Crucially, our mature CD organoids exhibited two physiological functions of native CDs: AVP-dependent AQP2 translocation and water permeability. These features enabled the human functional modeling of lithium-induced and congenital NDI. Although lithium is known to reduce urinary AQP2 excretion in humans⁵⁷, studies utilizing human kidney samples from patients with lithium-induced NDI remain limited. A recent study showed that lithium downregulated AQP2 mRNA in adult kidney-derived tubuloids⁵⁸, but protein expression and functional water permeability remained unexplored. Here, we demonstrated that lithium exposure decreases AQP2 protein expression and significantly impairs water permeability in human CD organoids. We also showed that these mature organoids can model congenital NDI. More than 280 AVPR2 mutations and at least 70 AQP2 mutations have been reported⁵⁹, and this modeling can be extended to additional variants. In particular, our water permeability assay enables functional evaluation of AQP2 mutations, which has previously been challenging. Although autosomal dominant polycystic kidney disease (ADPKD) has been modeled in immature kidney organoids^{11,45,60–62}, our mature CD organoids may reproduce the disease more accurately because the axis of AVP-AVPR2-AQP2 is central to its pathophysiology⁶³. In addition, establishing large-scale systems of mature CD cells, for example, through 2D culture, may facilitate drug screening and accelerate therapeutic development.

High tonicity also contributed to in vivo kidney maturation. Furosemide-treated mice exhibited NaCl depletion in the medulla and, consequently, impaired medullary maturation. These findings suggest that medullary immaturity, not only sodium reduction, causes urinary dilution in Slc12a1 knockout mice⁴¹. In our scRNA-seq of developing mouse kidneys^31,32, Slc12a1 expression began at E15.5, followed by upregulation of Slc14a2 at P7, indicating that salt accumulation precedes urea accumulation. This sequence mirrors vertebrate kidney evolution⁶⁴. Reptile kidneys, which lack a medulla, cannot concentrate urine. Amphibian kidneys develop a primitive medulla with limited concentrating capacity that contains NaCl, but not urea. Mammalian kidneys subsequently evolved a complex medulla containing both NaCl and urea, greatly enhancing urine concentration. This parallel between kidney development and evolution provides another example of evolutionary developmental biology.

Native kidneys contain CDs and nephrons in an organized manner and maintain a tonicity gradient along the corticomedullary axis, whereas our organoids consisted only of CDs. Although recently developed assembloids have partial corticomedullary organization^19–21, direct application of our protocol to the assembloids is challenging because high tonicity is detrimental to the renal cortex, which is never exposed to hypertonicity in vivo. Furthermore, medullary development in these assembloids remains immature: SLC12A1 + TAL segments are short, and the loop of Henle fails to form. Developing elongated medullae containing loops of Henle and medullary CDs, combined with a secure urinary outlet, will enable more realistic kidney organoids with endogenous osmotic gradients.

Our findings also raise the question of how tonicity is sensed. Cells initially shrink upon hypertonic stress, but restore their volume within minutes through regulatory volume increase (RVI), a process driven by with-no-lysine (WNK) kinase-activated ion transporters²⁵. A recent study discovered that molecular crowding caused by cell shrinkage triggers phase separation of WNK1, inducing RVI⁶⁵. In contrast, kidney maturation represents a long-term adaptation to hypertonicity over a timescale of hours to days. As shown here, the majority of the tonicity-induced maturation process is mediated by NFAT5. Cytoskeletal remodeling after cellular shrinkage has long been proposed to activate NFAT5 signaling, but the causal relationship remained obscure²⁶. More recently, increased intracellular ionic strength, without cell volume change, was shown to activate NFAT5 via a C-terminal prion-like domain⁶⁶. Therefore, NFAT5 itself may function as a tonicity sensor, triggering renal maturation. Unlike NaCl, high urea concentrations do not stimulate NFAT5 transcriptional activity in promoter-reporter assays^26,67, although the basis for this difference remains unclear.

The concept of tonicity-driven maturation can be generalized to other biological settings because other organs and tissues can be subjected to hypertonic stress. Downstream of the kidneys, the ureter and the bladder are also exposed to hyperosmotic urine during maturation. In addition, while neonates ingest breast milk that is inherently isotonic, which maintains the isotonicity of their gastric contents⁶⁸, the osmolality of adult meals can reach 600 mOsm/kg⁶⁹. Consequently, adult gastric contents often hover between 400 and 500 mOsm/kg.

In summary, we present high tonicity as the environmental cue driving organ maturation, thereby providing a platform for broader disease modeling and therapeutic development. We expect that many more physical stimuli for organ maturation will be discovered. For the kidney, low pH and hypoxia are also promising because the medulla is exposed to them^34,70. Elucidating such physical cues will further advance the research field of organ maturation, which bridges the gap between developmental biology and physiology.

Methods

Ethical statement on the use of human iPS cells

All experiments using human iPS cells were performed in accordance with institutional guidelines and approved by the licensing committees at the Faculty of Life Science, Kumamoto University: Ethics Committee for Epidemiological and General Research (approval number: 1453) and Ethics Committee for Human Genome and Gene Analysis Research (approval number: 359).

Mice

Nfat5^flox/flox mice³⁸ were described previously. Hoxb7-Cre mice³⁹ were purchased from Jackson Laboratory (JAX stock #004692). Both were maintained on an inbred background (C57BL/6J; CLEA Japan, Inc.). Hoxb7-GFP mice²⁵ were purchased from Jackson Laboratory (JAX stock #016251) and maintained on an outbred background (Jcl:ICR; CLEA Japan Inc.). Slc12a1^tm1Tkh mice⁴¹ were purchased from the Mutant Mouse Resource and Research Center (MMRRC) (stock number 000017-UCD). Mice were housed under a 12 h light/dark cycle (lights on at 7:00 AM) in a temperature-controlled room at 22 ± 2 °C with a relative humidity of 50 ± 10%. They had ad libitum access to water and a standard chow diet. All animal experiments were performed in accordance with institutional ethical guidelines and approved by the licensing committee of Kumamoto University (approval numbers: A2021-008, A2023-009, and A2024-073).

Administration of furosemide into neonatal mice

We injected 30 mg/kg body weight of furosemide (Tokyo Chemical Industry, F0182), an SLC12A1 inhibitor, into neonatal mice to deplete NaCl accumulation in the medulla, as described previously⁴². Furosemide was dissolved in DMSO at 48 mg/mL and further diluted in 10% DMSO with 40% PEG300 (Selleck, S6704), 5% Tween 80 (Selleck, S6702), and 45% H₂O to 4.8 mg/mL. Neonatal mice (Jcl:ICR; Japan SLC, Inc.) were subjected to daily subcutaneous injections of furosemide or DMSO behind their necks (from P1 to P7). Kidneys of the female mice were collected and histologically analyzed at P7.

Maturation culture of mouse UBs/CDs

UBs were isolated from embryonic kidneys of E11.5 ICR mouse embryos (Jcl:ICR; Japan SLC, Inc.). After the kidneys were incubated in DMEM/10% fetal bovine serum (FBS) (Nichirei Biosciences Inc., #172012) for 5 min at 37 °C, UBs were manually isolated by removing the mesenchyme using 30 G needles (Dentronics). Isolated UBs were embedded in 150 µL mouse branching medium containing 50% growth factor-reduced Matrigel (Corning #354230) (upper medium) in 24-well transwell inserts (Costar #3421), which were placed on 500 µL of the medium without Matrigel (bottom medium). The mouse branching medium consisted of DMEM/F12 (Gibco # 11320-033) with 10% FBS, 1% penicillin/streptomycin (P/S) (Gibco #15070063), 0.1 µM retinoic acid (RA), 100 ng/mL human R-spondin 1, 2 ng/mL human GDNF, and 100 ng/mL mouse FGF1. From Day 6, the osmolality of the medium was increased stepwise (Day 6: 450 mOsm/kg, Day 7: 600 mOsm/kg, Day 8: 750 mOsm/kg). The tissues were harvested on Day 9, unless otherwise indicated. The tissues can be cultured for several additional days beyond Day 9. The cultured mCDs were compared with P7 CD cells, sorted from Hoxb7-GFP transgenic mice³³. Nfat5 KO mCDs were cultured as described above, but the total culture duration was 8 days with the final osmolality of 600 mOsm/kg because the growth of Nfat5-deleted mCDs was significantly impaired at 750 mOsm/kg, and we needed to obtain a sufficient number of cells for the downstream scRNA-seq analysis. To study the plasticity of the mature state, we reduced the osmolality of the media from 750 mOsm/kg back to 300 mOsm/kg stepwise (Day 9: 600 mOsm/kg, Day 10: 450 mOsm/kg, Day 11: 300 mOsm/kg). When the final osmolality of 450 mOsm/kg was used, the osmolality was maintained from Day 6. When we treated mCD with hormones, the media was supplemented with 100 nM AVP (and 100 nM aldosterone) from Day 6 to Day 9. To study the urea tolerance, 450 mM urea (Wako # 219-00175) was added on Day 9 and examined on Day 10. Detailed reagent information is provided in Supplementary Table 1.

Human iPSCs and maintenance

The 201B7 and RN7 iPSC lines were established from a healthy female donor⁷¹ and a healthy male volunteer⁷², respectively. Both were maintained on iMatrix-511 (Nippi) in StemFit AK03N medium (AJINOMOTO). The cells were cultured in a humidified atmosphere of 5% CO₂ at 37 °C, and passaged every 5 days by Accutase (ESGRO). The cell culture medium was changed on days 2 and 4.

Generation of NFAT5 KO and AVPR2^Q174R human iPSCs

Two NFAT5 KO human iPSC lines were generated from the 201B7 line. A single-guide RNA (sgRNA) was designed to target exon 4, upstream of the Rel-homology domain (target sequence, 5’-AGCCTTGCTGTCGGTGACTG-3’; PAM sequence, AGG). After preparing a single-cell suspension of the 201B7 line, the sgRNA and Cas9 protein (Integrated DNA Technologies, IDT) were electroporated into the dissociated cells, using 4D-Nucleofector (Lonza). The electroporated cells were subjected to single-cell cloning by dilution. The target region was sequenced, and the NFAT5 KO lines were selected based on the analysis using the Synthego ICE Analysis tool (v3)⁷³. The results were consistent between the two clones, and data from one of the clones were presented. AVPR2^Q174R human iPSC lines with a c.521A>G substitution were generated from the RN7 iPSC line. An sgRNA was designed to target the region surrounding the 521st base in the coding regions (target sequence, 5’-TGGGCGAAGATGAAGAGCTG-3’; PAM sequence, GGG). The sgRNA, Cas9 protein, and a single-stranded DNA oligo (5’-CGGCCGGTGCTAGTGGCTTGGGCCTTCTCGCTCCTTCTCAGCCTGCCCCGGCTCTTCATCTTCGCCCAGCGCAACGTGGAAGGTGGCAGCGGGGTCACTG-3’) for homologous recombination (IDT) were electroporated into a single-cell suspension of the RN7 line. The rest of the procedure was performed as for the NFAT5 KO line. The results were consistent between the two clones, and data from one of the clones were presented. Detailed primer information for sequencing is provided in Supplementary Table 2.

Maturation culture of human CD organoids

UB organoids were induced from human iPSCs based on our previous protocol¹². The basal differentiation medium was serum-free medium consisting of DMEM/F12 (Gibco #11320-033), supplemented with Insulin-Transferrin-Selenium (Gibco #41400-045), 2×B27 without retinoic acid (RA) (Gibco #12587-010), non-essential amino acids (Gibco #11140-050), sodium pyruvate (Gibco #11360-070), GlutaMAX (Gibco #35050-061), 0.5×penicillin/streptomycin (Gibco #15070-063), and 90 μM 2-mercaptoethanol (Nakarai #21438-82). At day 0, iPSCs were aggregated in the medium containing 10 ng/mL human activin A, 1 ng/mL human BMP4, and 10 µM Y27632 in V-bottom 96-well low cell-binding plates. At day 1, the spheres were stimulated with 10 μM CHIR and 1 ng/mL human BMP4 for 36 h and then transferred at day 2.5 to the medium containing 100 nM RA, 100 ng/mL human FGF9, 100 nM LDN193189, and 100 µM SB431542. At day 4.5, the resulting spheres were treated with 100 nM RA, 5 µM CHIR, 100 ng/mL human FGF9, and 100 nM LDN193189. At day 6.25, the spheres were dissociated with 0.25% trypsin/EDTA and sorted to isolate CXCR4+/ KIT+ Wolffian duct progenitors. 5000 sorted cells per well were aggregated in 96-well low cell-binding plates, and cultured for 7 days to UBs in the three types of medium containing 10 µM Y27632, 100 nM RA, 100 ng/mL human FGF1, 10 nM LDN193189, 10% Matrigel (growth factor reduced; Corning #356230), and varying concentrations of CHIR (Day 6–8: 1 µM, Day 8-13: 3 µM), FGF9 (Day 6–10: 5 ng/mL, Day 10–13: 0 ng/mL) and GDNF (Day 8–10: 1 ng/mL, Day 10-13: 2 ng/mL). On Day 13, which was regarded as Day 0 of the branching culture, UB organoids were embedded in 150 µL human branching medium containing 50% growth factor-reduced Matrigel (upper medium) in 24-well transwell inserts. The inserts are placed on 500 µL of the medium without Matrigel (bottom medium). The medium consisted of DMEM/F12 with 10% FBS, 1% GlutaMAX, 1% nonessential amino acids, 1% pyruvate, 1% P/S, 0.1 µM RA, 100 ng/mL human R-spondin 1, 2 ng/mL human GDNF, 100 ng/mL human FGF1, 30 ng/mL human FGF7, and 10 nM LDN193189. From Day 6, 10 nM FSK was added. The osmolality of the medium was increased stepwise from Day 12 (330 mOsm/kg on Day 12, 360 mOsm/kg on Day 13, 400 mOsm/kg on Day 14, 550 mOsm/kg on Day 16, 700 mOsm/kg on Day 17, and 900 mOsm/kg on Day 18). The medium was changed on Days 6, 9, 12, 13, 14, 16, 17, and 18. We optimized the protocol for human CD organoids to produce phenotypes similar to those of mature mouse CDs in terms of morphology, gene expression, and ddAVP responsiveness. For NFAT5 KO organoids, the organoids were harvested on Day 19 with the final osmolality of 700 mOsm/kg because NFAT5 KO organoids did not tolerate 900 mOsm/kg. For the hormone-based protocol, from Day 6, CD organoids were cultured in advanced RPMI 1640 Medium (Gibco #12633012) supplemented with 1% GlutaMax, 0.1 µM RA, 100 ng/mL human R-spondin 1, 2 ng/mL human GDNF, 100 ng/mL human FGF1, 30 ng/mL human FGF7, and 10 nM LDN193189. From Day 16, CD organoids were cultured in advanced RPMI 1640 Medium with 1% GlutaMax, 100 nM AVP, and 100 nM aldosterone. This protocol involved medium changes every 3 or 4 days. For the tonicity plus hormone protocol, 10 nM FSK from Day 6 and osmotic elevation from Day 12 were combined with the hormone-based protocol. Human CD organoids were compared with a sample of the adult human kidney. To examine the ddAVP responsiveness, the medium under the Transwell inserts was replaced with the fresh medium containing 1 µM ddAVP alone at 300 or 900 mOsm/kg every day from Day 20. On Day 25, the organoids were harvested for further study. To examine the effect of lithium on human CD organoids, we added 20 mM lithium chloride (Wako #123-00162) from Day 16 to Day 20 of the maturation protocol. The same concentration of sodium chloride instead of lithium chloride was further added for the control organoids. Detailed reagent information is provided in Supplementary Table 1.

Medium osmolality

The osmolality in the mouse and human branching medium was measured to be ~320 mOsm/kg using an osmometer (Fiske 210 Micro-Sample Osmometer), and was assumed to be 300 mOsm/kg for simplicity. To achieve the target osmolality, 5 M NaCl solution (Wako #195-01663), 8 M urea solution, or 2 M sucrose solution (Wako #196-00015) was added to the medium. The dilution by adding these solutions was approximately 95%. We added an equivalent amount of water instead of NaCl solution to study the effects of diluting the basal medium.

Immunohistochemistry

Specimens were fixed in 10% formalin, embedded in paraffin, and sectioned at 6 µm. Paraffin-embedded sections were deparaffinized and subjected to antigen retrieval in a citrate buffer. After washing with phosphate-buffered saline (PBS) twice, sections were blocked in PBS containing 1% bovine serum albumin (BSA) (Sigma #4503) for 1 h at room temperature. Sections were incubated with primary antibodies overnight at 4 °C, followed by incubation with secondary antibodies conjugated with Alexa Fluor dyes for 1.5 h at room temperature. For NFAT5 and AQP4 staining, the signal was amplified using an ImmPRESS HRP Reagent kit (Vector Laboratories) and a Tyramide Signal Amplification (TSA) kit (Thermo Fisher Scientific). When we stained NFAT5 and AQP4 simultaneously, the NFAT5 signal was detected by using biotin-streptavidin amplification. Fluorescence images were captured by confocal microscopy (Leica TCS SP8). For the N/C ratio of NFAT5, the line profile was used to obtain the maximal (nuclear value) and minimal values (cytoplasmic value) inside a cell, whose borders were defined by KRT8. At least 3 biologically independent experiments were performed, and 1–3 sections per condition were stained in each experiment. Detailed antibody information is provided in Supplementary Table 1.

Section in situ hybridization

RNAscope analysis of 10% formalin-fixed paraffin sections was performed using an RNAscope Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics; Cat# 323100). Signal amplification was performed using TSA kits. At least 3 biologically independent experiments were performed, and 1–3 sections per condition were examined in each experiment. Details of the RNAscope probes are provided in Supplementary Table 1.

Measurement of lumen diameter, cell height, cytoplasmic area, and cilia length

These parameters were measured using ImageJ software (1.54 g)⁷⁴. Lumen diameter was measured in brightfield images, cell height and cytoplasmic area in hematoxylin and eosin-stained sections, and cilia length in sections stained with both AcTUB and ARL13B.

Measurement of PGE₂ production by MS

Three mCDs, cultured at normal and high tonicity, were harvested on Day 9, collected into By Hand Style Simple Grinding Containers (ASONE #2-1825-01), and stored at −80 °C. The tissues were homogenized with 500 μL methanol:formic acid (100:0.2) containing d4-PGE₂ as an internal standard using microtip sonication. The samples were submitted to solid-phase extraction using an Oasis HLB cartridge (5 mg; Waters, Milford, MA) as described previously⁷⁵. Then, PGE₂ was quantified using a triple-quadrupole mass spectrometer equipped with an electrospray ionization (ESI) ion source (LCMS-8060, Shimadzu Corporation) in the negative mode ESI and multiple reaction monitoring (MRM) modes, as described previously⁷⁶. Briefly, the samples were resolved on the C8 column (2.1 mm ID × 150 mm, 2.6 µm, Phenomenex Kinetex), using a step gradient with mobile phase A (0.1% formate in water) and mobile phase B (acetonitrile). The gradient was programmed at 10% B at 0 min, 25% B at 5 min, 35% B at 10 min, 75% B at 20 min, 95% B from 20.1 to 28 min, and 10% B at 28.1 min, with the program ending at 31 min. The flow rate was 0.4 ml min⁻¹ and the column temperature was 40 °C. The injection volume was 10 µL. Chromatogram peaks obtained with compound-specific MRM channels were integrated and manually confirmed. The relative PGE₂ signal was calculated as (PGE₂ peak area/d4-PGE₂ peak area)×10³. The PGE₂ production of the mCDs was calculated by subtracting the amount of PGE₂ in Matrigel from that in cultured mCDs because cultured mCDs were embedded in Matrigel. We repeated three independent experiments.

Measurement of water permeability in human CD organoids

An organoid was manually harvested, with the surrounding 50% Matrigel layer minimized as much as possible. Then, the organoid was gently placed on top of 6 µL of 100% Matrigel in the center of a well of a 24-well plate. After a 2-min incubation at 37 °C for fixation, phase-contrast images were acquired at ∼25 °C, using a BZ-X1000 microscope (Keyence). Then, as a relative hypotonic stimulus, medium at 450 mOsm/kg was gently added into the well containing the CD organoid. After the osmotic change, images were captured at 3 s intervals for a total duration of 12 min. The first image, just after the relative hypotonic stress was counted as 0 min and images at 0, 3, 6, 9, and 12 min were analyzed for volume changes. The areas of clearly defined regions of organoids at different time points were measured using ImageJ software (1.54 g)⁷⁴. The volume changes were estimated by normalizing the area of clearly defined organoid regions at each time point to the initial area at 0 min. The area under the curve (AUC) was calculated for each region, defined as the area above the baseline of 1. For lithium-induced and congenital NDI, this procedure was performed within 2 and 5 days of the end of the 6-day LiCl treatment and the 5-day ddAVP treatment, respectively.

Human kidney sample

Adult human kidney tissue was obtained from the normal parts of excised kidney specimens from a 31-year-old female patient with kidney cancer. The procedure was performed in accordance with institutional guidelines and approved by the licensing committee at the Faculty of Life Science, Kumamoto University: Ethics Committee for Epidemiological and General Research (approval number: 2561). After explaining our study, the patient agreed to participate in the study and signed informed consent forms.

scRNA-seq analyses

We utilized previously described scRNA-seq data: E15.5, P0, and P7 mouse kidneys^31,32, induced nephron progenitors and stroma progenitors from embryonic stem cells (iNP and iSP)¹⁹, human CD organoids cultured with the hormones¹⁵, and human kidney assembloids²⁰. We also analyzed single-nucleus RNA sequence data of human kidney assembloids cultured in vitro²¹. Cultured mCDs, P7 medullary kidneys, human CD organoids, and adult human kidneys were newly analyzed in this study. We manually removed the cortex to obtain P7 medullary kidneys. They were incubated in 0.25% trypsin/EDTA for 10 min. The cells were washed with HEPES-buffered saline solution (HBSS) (Gibco #14185052) containing 2% FBS, 50 µg/mL DNase I (Worthington #LS002139), 1 mM CaCl₂ (Wako #031-00435), and 0.035% NaHCO₃ (Wako # 191-01305). The cells were resuspended in 0.04% BSA/PBS and filtered through a 40 µm-pore strainer (Falcon #352340). The cell number and viability were measured using Countess Automated Cell Counter (Invitrogen). Aliquots containing 5000 dissociated cells from each sample were applied to Chromium Controller (10x Genomics). Chromium Single Cell 3’ Library & Gel Beads Kit v3 (10x Genomics) was used to generate cDNA libraries, which were then sequenced by Illumina HiSeq X Ten (422,086,013 reads for the mCDs at 300 mOsm/kg; 439,376,373 reads for the mCDs at 750 mOsm/kg; 324,539,522 reads for the mCDs at 750 mOsm/kg with the hormones; 549,830,752 reads for the control mCDs at 600 mOsm/kg; 444,505,977 reads for the Nfat5 KO mCDs at 600 mOsm/kg; 757,548,299 reads for the E15.5; 955,496,397 reads for the P0; 370,196,520 reads for the P7; 375,856,018 reads for the P7 medulla-1; 395,311,654 reads for the P7 medulla-2; 387,644,117 reads for the iNP; 450,799,847 reads for the iSP; 557,712,745 reads for the human CD organoids at 300 mOsm/kg with the hormones; 382,321,150 reads for the human CD organoids at 900 mOsm/kg with FSK; 439,527,630 reads for the human CD organoids at 900 mOsm/kg with FSK and the hormones; 411,174,676 reads for the adult human kidney). The Q30 base RNA reads (Q-scores indicating sequencing quality) of the samples were 94.0% for the mCDs at 300 mOsm/kg; 93.1% for the mCDs at 750 mOsm/kg; 96.3% for the mCDs at 750 mOsm/kg with the hormones; 95.8% for the control mCDs at 600 mOsm/kg; 95.6% for the Nfat5 KO mCDs at 600 mOsm/kg; 86.2% for the E15.5; 93.6% for the P0; 91.9% for the P7; 97.0% for the P7 medulla-1; 96.9% for the P7 medulla-2; 81.0% for the iNP; 87.8% for the iSP; 96.0 % for the human CD organoids at 300 mOsm/kg with the hormones; 95.0% for the human CD organoids at 900 mOsm/kg with FSK; 96.1% for the human CD organoids at 900 mOsm/kg with FSK and the hormones; and 93.3% for the adult human kidney.

The raw sequence datasets were processed using the cell ranger count command in Cell Ranger (version 7.1.0 for the NFAT5 datasets and version 9.0.0 for the other datasets; GRCm38; 10x Genomics). These individually processed datasets were integrated as needed using the cell ranger aggr command (10× Genomics). Subsequent analyses were performed using the R statistical programming language (version 4.3.2)⁷⁷ and the RStudio software (version 2023.12.1 + 402)⁷⁸. The Seurat package (version 5.2.1) was used for analyses, including quality control, data normalization, data scaling, visualization, and differential analyses^79,80. For quality control, cells expressing <500 genes, >35% of mitochondrial genes, and >1% of hemoglobin genes were filtered out. Potential doublets were also removed by using DoubletFinder (version 2.0.4) before further analysis. Principal component analysis was applied for dimension reduction. Dimension values were determined by the JackStrawPlot function or elbow plots: 55 for the mCDs at 300 and 750 mOsm/kg, 53 for the mCDs at 300 and 750 mOsm/kg without or with the hormones, 20 for mCDs at 750 mOsm/kg with the hormones and in vivo mouse UB/CD, 26 for the E15, P0, and P7 UB/CD, 44 for the E15, P0, and P7 stroma, 44 for the control and Nfat5 KO mCDs, 83 for the human CD organoids, 54 for the human CD organoids and the kidney assembloids, and 50 for the human CD organoids and adult human kidney. Normalization was performed using SCTransform with the top 3000 highly variable genes selected by the SelectIntegrationFeatures function. Integrated data were generated by IntegrateData using anchor genes created by the FindIntegrationAnchors function, followed by scaling the data by regressing out cell cycle, high mitochondrial, ribosomal, and hemoglobin ratios. UMAP plots were generated using the uwot package⁸¹. Two datasets of P7 medullary kidneys were merged and demonstrated in UMAP plots. The UMAP coordinates and Seurat cluster coordinates were exported as csv files for demonstration in Loupe Cell Browser (version 7.0.0).

Identification of differentially expressed genes

We used the PrepSCTFindMarkers and FindMarkers functions to select genes that were differentially expressed between two clusters and to pick up representative genes in each cluster (log2 fold change < −0.58 or log2 fold change > 0.58; P-value < 0.05). We identified tonicity-regulated genes in mCDs by comparing CD clusters between normal and high tonicity (Supplementary Fig. 4a). In this analysis, unadjusted P-values were utilized to maintain statistical sensitivity for identifying tonicity-response genes, which were validated by RT-qPCR (Supplementary Fig. 1f); this statistical criterion was applied consistently throughout the study. We compared E15.5 CDs and P7 medullary CDs, and selected maturation markers of medullary CD and stroma by comparing E15.5 and P7 kidneys. We compared Nfat5-deleted cells in the Nfat5 KO mCDs and Nfat5-expressing cells in the control mCDs, and selected genes downregulated in the former cells as NFAT5-regulated genes. For these genes, volcano plots were described using the EnhancedVolcano function⁸². We also calculated the percentage of cells expressing marker genes such as Aqp2 in specific clusters. To obtain a list of human medullary PC-specific markers, we utilized lists of cluster marker genes provided by the KPMP project⁶. First, we generated a list of genes shared between the inner and outer medullary PCs. Because this list included general PC markers such as AQP2, we removed the cortical PC genes⁶ to obtain the medullary PC-specific genes. We used the top 50 highly expressed genes from the medullary PC-specific genes for the analysis.

Over-representation analysis

We performed over-representation analysis based on GO terms and the Molecular Signature Database using the RNAseqChef program³³.

Unbiased hierarchical clustering

We first re-integrated the datasets of the mCDs at 300 and 750 mOsm/kg, the iNP, the iSP, the E15.5, and the P7, and generated the UMAP plots as described above. Then, we utilized the AverageExpression function to get average gene expression data for CD/medullary CD clusters in each dataset. Then, a heatmap with unbiased hierarchical clustering analysis was obtained using the pheatmap function⁸³ with all differentially expressed genes between E15.5 CDs and P7 medullary CDs as a reference gene set (Supplementary Data 5).

Reference modification for scRNA-seq

Based on the genetic information of Nfat5^flox/flox mice³⁸, a sequence between the loxP sites including exon 5 was added to a reference for the mouse analysis as “Nfat5-exon 5”. Instead, the original reference for Nfat5 was removed. Sequences of Cre and the human GH mini gene that is originally located at the 3’ end of Hoxb7-Cre³⁹, were joined in tandem and added to the reference as “Cre”. This modification enabled us to distinguish Nfat5-deleted and -expressing cells (Supplementary Fig. 7h, i).

RNA extraction and RT-qPCR

Total RNA was isolated using RNeasy Plus Micro Kit (Qiagen #74034) and reverse-transcribed using Superscript VILO cDNA synthesis kit (Invitrogen #11754-250). Quantitative PCR was performed, using Real-Time PCR System (Takara Bio) and TB Green Fast qPCR Mix (Takara Bio #RR430A). Relative mRNA expression levels were normalized to β-actin gene expression. Detailed primer information is provided in Supplementary Table 2.

Statistics and Reproducibility

Each experiment was repeated at least three times independently. Representative images were presented. Quantification data, including RT-qPCR, were presented as mean ± SEM with plots. For statistical analysis, two-sided t-tests were performed between the two groups. For three or more groups, a two-sided one-way ANOVA was first performed, followed by Tukey’s or Dunnett’s multiple comparison tests. To evaluate overall differences in water permeability, a two-way repeated comparisons ANOVA was performed, followed by post-hoc analyses using Sidak’s or Tukey’s multiple comparisons test at specific times. All analyses were performed using Prism (version 10).

Figure construction

Graphs for quantification data were created using Prism (version 10). All schematics were created using Adobe Illustrator 2023.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

D.I. and R.N. conceived and designed the experiments. D.I. performed the cell culture experiments and bioinformatics analyses. M.M., J.A., C.L-R. and Y.I. provided Nfat5-floxed mice. T.O. performed RNAscope in situ hybridization. R.M. and Y.S. performed mass spectrometry. S.T. and K.Y. supported the sample preparation for scRNA-seq. K.M. processed the datasets in CellRanger. K.M. and K.Y. prepared the references for the scRNA-seq analysis. S.F. performed histological sectioning for the SLC12A1-inhibited mouse kidneys. T.K. provided adult human kidney samples. D.I. and R.N analyzed data and wrote the manuscript, with input from all authors. R.N. supervised the project and acquired funding.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported in part by JSPS KAKENHI Grant Number JP21H05050 (for R.N.), JSPS KAKENHI Grant Number JP24K23941 (International Leading Research: Creating a Kidney) (for R.N.) and the Acceleration program for regenerative medicine (JP26bm1423043) from the Japan Agency for Medical Research and Development (for R.N.). This work was also supported by MEXT-Supported Program for the Inter-University Research Network for High-Depth Omics, IMEG, Kumamoto University and MEXT Promotion of Development of a Joint Usage/Research System Project: Coalition of Universities for Research Excellence Program CURE (Grant Number JPMXP1323015486).

Data availability

The current study involves scRNA-seq analysis of several types of samples. The datasets generated during the current study have been uploaded to the National Center for Biotechnology Information Gene Expression Omnibus (GEO) (GSE275292 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE275292]). These include mCDs cultured at normal and high tonicity, mCDs cultured at high tonicity with the hormones, control and Nfat5 KO mCDs, P7 kidney medulla, and human CD organoids cultured with the hormones at normal and high tonicity. The datasets of E15.5, P0 (GSE149134 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE149134]), P7 (GSE271515 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE271515]), iNP, and iSP (GSE178263 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE178263]) were described previously^19,31,32. For quick reference, these mouse datasets have also been uploaded to GEO (GSE275292 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE275292]). The datasets of human CD organoids (GSE206153 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE206153])¹⁵ and kidney assembloids (GSE290442 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE290442] (Day 14 kidney assembloids cultured in CD medium with A83-01, U0126, and XAV939)²⁰; GSE272707 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE272707]²¹) were also analyzed. Source data are provided with this paper.

Competing interests

D.I. and R.N. are inventors on a patent application related to the protocol described in this study, which is currently pending. The remaining authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77903-4.