Plasticity of distal nephron epithelia from human kidney organoids enables the induction of ureteric tip and stalk.

SUMMARY

During development, distinct progenitors contribute to the nephrons versus the ureteric epithelium of the kidney. Indeed, previous human pluripotent stem cell-derived models of kidney tissue either contain nephrons or pattern specifically to the ureteric epithelium. By re-analysing the transcriptional distinction between distal nephron and ureteric epithelium in human fetal kidney, we show here that while existing nephron-containing kidney organoids contain distal nephron epithelium and no ureteric epithelium, this distal nephron segment alone displays significant in vitro plasticity and can adopt a ureteric epithelial tip identity when isolated and cultured in defined conditions. “Induced” ureteric epithelium cultures can be cryopreserved, serially passaged without loss of identity and transitioned towards a collecting duct fate. Cultures harbouring loss-of-function mutations in PKHD1 also recapitulate the cystic phenotype associated with autosomal recessive polycystic kidney disease.

Graphical Abstract

eTOC

Little and colleagues demonstrate the plasticity of the distal nephron epithelium present within kidney organoids showing that this can be induced to adopt a ureteric epithelial phenotype. Subsequent maturation of this epithelium generated collecting ducts, facilitating the accurate modelling of autosomal recessive polycystic kidney disease.

INTRODUCTION

The development of methods to direct the differentiation of human pluripotent stem cells into kidney organoids has facilitated the recreation of aspects of human kidney organogenesis in vitro, facilitating disease modelling (Forbes et al., 2018; Freedman et al., 2015; Hale et al., 2018; Subramanian et al., 2019; Tanigawa et al., 2018) and enhancing our understanding of nephrogenesis (Howden et al., 2019; Yoshimura et al., 2019). Human kidney organoids contain nephron progenitors, patterning and segmenting nephrons, surrounding stromal populations and even a forming vasculature, with strong transcriptional congruence to human fetal kidney (Combes et al., 2019b; Takasato et al., 2015). As in other directed differentiation protocols, the identification of component cell types has relied heavily on the presence of proteins or expression of genes previously regarded as marking a particular cell type in mouse or human. However, few markers are unique to a single cell type. Despite the growing application of transcriptional profiling, determination of cellular identity is frequently biased, leaning on the selection of recognisable genes rather than the unbiased prediction of cellular identity based on the global transcriptional profile. In the kidney, this has presented a particular challenge around the distinction between ureteric epithelium (UE) and elements of the distal nephron (DN) due to significant congruence in transcriptional profile of these epithelial elements in both human and mouse (Combes et al., 2019a; Ransick et al., 2019). The definitive identification of cellular identity within iPSC-derived kidney structures is critical to establish accurate models of human disease. A better understanding of cellular identity will also facilitate the development of cell-based products for regenerative medicine purposes.

The UE arises as a budding side-branch of the nephric duct that is attracted to the metanephric mesenchyme to form the metanephros, the third and final kidney in mammals. The invading ureteric bud (UB) gives rise to proliferating RET⁺ ureteric tips. These induce the surrounding mesenchyme to form nephrons via a mesenchyme to epithelial transition, with reciprocal signalling from this mesenchyme inducing the ureteric tip to dichotomously branch, thereby driving organ expansion and increasing nephron formation (Costantini and Kopan, 2010). The stalks of the branching epithelium lose RET expression and eventually display regionally specific cellular subtypes evident in the final collecting ducts of the kidney. The nephron epithelium does not branch but shows distinct segmentation from proximal (podocytes, proximal tubules) through medial (loop of Henle) to distal (distal straight, distal convoluted, connecting segment) nephron. All epithelial cell types within the nephron arise from a SIX2 expressing progenitor population. This was established via lineage tracing in mouse (Kobayashi et al., 2008) and is also evident in kidney organoids (Howden et al., 2019).

The nephron and ureteric epithelia have distinct origins in vivo. The ureteric tip and stalk arise as a side-branch of the nephric duct, which is itself initiated from a more anterior intermediate mesoderm than the metanephric nephron progenitors. As such, it has previously been proposed that the generation of UE from human pluripotent stem cells requires a distinct directed differentiation protocol to that used to derive nephron epithelia (Taguchi et al., 2014; Taguchi and Nishinakamura, 2017). Despite this ontogeny, there is a physical continuum between the DN and UE, together with substantial overlap in gene expression (Combes et al., 2019a; Ransick et al., 2019). In particular, segments of the DN express many genes that have previously been assumed to specifically mark the UE, including Hoxb7, Krt8, Gata3 and Calb1, and both DN and UE ultimately contain principal cells (PCs) and intercalated cells (ICs) (Combes et al., 2019b). This eventual congruence in cellular composition and transcriptional expression between DN and UE suggests that, rather than having to separately specify UE, it may be possible to induce the formation of UE from DN and vice versa.

In this study, we return to the developing human kidney to more carefully define the specific transcriptional profile of individual DN and UE segments during human development, thereby improving the capacity to benchmark the identity of in vitro iPSC-derived populations. Using this approach, we show that a GATA3⁺/EPCAM⁺ epithelial population present within our existing nephron-containing kidney organoids is transcriptionally most aligned with a distal straight /connecting segment identity. We then demonstrate that it is possible to shift the identity of this epithelium towards UE, including the induction of ureteric tips, cortical and medullary UE, by altering the in vitro culture conditions. The resulting UE cultures can be derived from iPSC lines with diverse genetic backgrounds, without the requirement for cell sorting, and expanded and stably cultured for several weeks without loss of cell identity. The phenotype of this induced UE can also be switched from ureteric tip to ureteric stalk in response to vasopressin and aldosterone. Finally, we show that UE cultures can be established from iPSCs harbouring loss-of-function mutations in PKHD1 and that when transitioned towards ureteric stalk, spontaneously form dilated structures that recapitulate the cyst formation seen in autosomal recessive polycystic kidney disease (ARPKD).

RESULTS

Unique transcriptional profiles distinguish human UE and DN cell phenotypes

Takasato et al defined a directed differentiation protocol that enables the generation of kidney organoids containing patterning and segmenting nephrons, surrounding stroma and endothelial cells (Takasato et al., 2015). A GATA3⁺CDH1⁺PAX2⁺ epithelium is also evident in organoids generated by this method, often present as a contiguous connecting structure linking nephrons together. Initially described as UE, this segment was separated from the proximal nephron segments via a distal tubule-like GATA3⁻CDH1⁺ epithelium. GATA3 expression, however, is enriched but not unique to the UE and has also been reported in the DN in mouse and human (Combes et al., 2019a; Wu et al., 2018). This has important implications for delineating the structures generated within organoid cultures in vitro, where a lack of anatomical and spatial information makes the interpretation of cell identity more challenging. To improve our capacity to distinguish between these GATA3-expressing epithelial components in vitro, a comprehensive single cell dataset from two human fetal kidney (HFK) samples (96 and 108 days gestation) (Holloway et al., 2020) comprising a total of 10,349 single cells post quality control was employed. Cluster analysis identified 29 distinct populations; including 18 nephron / nephron progenitor clusters, 5 stromal clusters, 4 UE clusters, immune cells and endothelium (Figure 1A, Figure S1A, Table S1 and S2). The clusters representing the DN and UE populations were subsequently analysed at higher resolution, with a particular focus around the identification of the component cellular signatures that define each epithelial subsegment (Figure 1B). Previously defined markers allowed the identification of 5 DN subclusters including the distal early nephron (N.Distal_EN), loop of Henle (N.LOH), distal straight tubule (N.DST), distal convoluted tubule (N.DCT) and connecting segment (N.CS) (Figure 1B and Table S2). Dimensional reduction of the UE cluster revealed four discernible subclusters (Figure 1B). A ureteric tip population (U.Tip) was identified based on co-expression of known tip markers such as RET, SOX9, WNT11 and MUC6. This cluster was also enriched for cell cycle genes. The largest UE subcluster, identified as cortical UE (U.Cortical), exhibited low RET and high AQP2 expression. The medullary UE, which exhibits highest TACSTD2 expression, could be further separated into two distinct clusters based on the presence (U.Med_Inner) or absence (U.Med_Outer) of UPK2, UPK3B and SNX31. As noted above, many of the common markers previously used to define UE were enriched but not restricted to this population (Figure 1C and Figure S1B), including HOXB7, GATA3, KRT8 and KRT18. Unique DN and UE markers recently defined in an analysis of embryonic mouse kidney (Combes et al., 2019a) did specifically mark either UE or DN clusters within our human dataset (Figure 1D, Figure S1C and S1D). Furthermore, by performing an unbiased differential gene expression analysis, we identified additional markers that were either highly enriched or restricted to specific cell types of the DN and UE lineages (Figure 1E). Genes that we identified as specifically expressed in the human UE but not DN included RET, WNT9B, SMIM22, KANK4 and KCNN4 while those specific to the human DN included BSND, KCNJ1, FXYD2 and CLCNKB (Figure S1D). This analysis established a more comprehensive framework for identifying cellular identity within an unknown population.

Figure 1. Identification of cell-specific transcriptional signatures within human fetal kidney.

(A) UMAP plot showing combined single cell data from 10349 cells collected from day 96 (4620 cells, male) and day 108 (5729 cells, female) gestation human fetal kidney annotated by lineage (N, nephron; S, stroma; U, ureteric epithelium) and cell type based upon previously identified markers. CS, connecting segment; DCT, distal convoluting tubule; DST, distal straight tubule; LOH, loop of Henle; Distal_EN, distal early nephron; PT_Mat, maturing proximal tubule; PT_Dev, developing proximal tubule; Medial_EN, medial early nephron; Pod_Mat, maturing podocyte; Pod_Dev, developing podocyte; PEC, parietal epithelial cell; RV, renal vesicle; NPC_PTA_CC, nephron progenitor / pretubular aggregate in cell cycle; NPC-PTA, pretubular aggregate; NPC_primed, primed nephron progenitors; NPC_CC, proliferating nephron progenitors; NP_STR, nephron progenitor/stromal phenotype; T.Tip, ureteric tip; U.Med_Outer, outer medullary ureteric epithelium; U,Med_Inner, inner medullary ureteric epithelium; U.Cortical, cortical ureteric epithelium; S.Med, medullary stroma; S.IC, inner cortical stroma; S.OC_NZ, outer cortical stroma / nephrogenic zone. (B) UMAP plot of re-clustered Distal Nephron (DN) and Ureteric Epithelial (UE) subclusters showing a distinct separation of the two lineages with dimensional reduction and clustering. (C) Gene expression analysis of commonly used UE markers. Many of these are also expressed in regions of the DN segments. (D) Transcriptional analysis of genes that are specifically expressed in DN or UE in human fetal kidney. These were previously defined in an analysis of fetal mouse kidney (Combes et al. 2019a). (E) Relative specificity of the top 10 differentially expressed genes for each of the 5 DN and 4 UE subclusters.

UE identity can be induced from the GATA3⁺ epithelium in kidney organoids

A recent study in mouse reported the reconstruction of ureteric tip-like structures from individual UE cells purified by fluorescent activated cell sorting (FACS) from transgenic E11.5 Hoxb7/Venus mouse kidneys (Yuri et al., 2017). These budding tree-like structures could be established from single, dispersed UE cells after several days of culture in a serum-free medium supplemented with CHIR99021 (CHIR), GDNF, FGF, retinoic acid and the Rho-kinase inhibitor, Y-27632. Armed with an improved capacity to distinguish between DN and UE populations, we sought to more definitively identify the GATA3⁺ epithelium within an existing kidney organoid protocol (Howden et al., 2019) and examine whether this cellular population could be propagated using this previously defined UE medium (Yuri et al., 2017). To facilitate isolation of the GATA3⁺ population from kidney organoids, we utilised a previously described dual fluorescent reporter cell iPSC line, MAFB^mTagBFP2:GATA3^mCherry (Vanslambrouck et al., 2019a). This line harbours the mTagBFP2 and mCherry fluorescent reporter genes under the control of the endogenous MAFB and GATA3 loci, marking the podocyte and UE / DN populations respectively (Figure 2A). Using our previously described differentiation protocol (Howden et al., 2019; Takasato et al., 2016), we generated kidney organoids containing both mCherry⁺ epithelium and mTagBFP2⁺ podocytes (Figures 2B and 2C). We have previously shown that the relative proportion of GATA3⁺ cells within kidney organoids is influenced by the duration of WNT signalling during the initial stage of differentiation, which we interpret to be influencing relative anterior versus posterior mesodermal patterning (Takasato and Little, 2015). As anticipated, the ratio of mCherry⁺ epithelium to mTagBFP2⁺ podocytes varied in response to the length of CHIR exposure during the initial stage of differentiation, with the highest proportion of mCherry⁺ epithelium attained with a shorter (3 d) duration of CHIR treatment (Figure 2B). In this study, we also replaced FGF9 (200 ng/ml) with FGF2 (400–600 ng/ml) which also results in maximal mCherry⁺ epithelium within kidney organoids (Figure S2A and S2B). This modified kidney organoid differentiation protocol (3 days of CHIR followed by 600 ng/ml FGF2) was used to promote specification of the GATA3⁺ epithelium in all subsequent experiments unless otherwise stated. Using an anti-EPCAM antibody to distinguish the GATA3⁺ epithelial and mesenchymal populations, the mCherry⁺/EpCAM⁺ (Ch⁺/Ep⁺-org) fraction was purified by fluorescence activated cell sorting (FACS) and transferred to a Transwell plate containing UE-medium (GDNF, CHIR, FGF2, ATRA, Y-27632) (Yuri et al., 2017) and mounted in 50% Matrigel (Figure 2C). These conditions promoted the proliferation and expansion of human GATA3⁺/EpCAM⁺ epithelium (Ch⁺/Ep⁺-UE), resulting in the formation of complex, budding structures which maintained GATA3 promoter-driven reporter gene expression even after weeks of continuous culture (Figure 2D, Supplementary Movies 1 and 2). After extended culture in UE medium, we also detected the presence of markers commonly used to identify UE, including KRT8 (Cytokeratin-8), PAX2 and GATA3 by whole mount immunofluorescence. These structures also displayed the presence of SOX9⁺ tips and CALB1⁺ stalks which, while suggestive of UE, is not definitive (Figure 2E).

Figure 2. Characterisation and culture of kidney organoid-derived GATA3⁺ epithelium.

(A) Schematic diagram of the targeting strategy used for generation of MAFB^mTagBFP2:GATA3^mCherry dual reporter iPSCs which mark the podocyte and collecting duct / distal nephron populations respectively (from Vanslambrouck et al. 2019). (B) Quantification of mCherry⁺ and mTagBFP2⁺ cells as determined by flow cytometry in kidney organoids exposed to 3, 4 or 5 days of CHIR during the first stage of differentiation. Data represent mean ± SD, n = 4. (C) Overview of the strategy used to isolate and expand mCherry⁺/EpCam⁺ epithelium from kidney organoids generated from MAFB^mTagBFP2:GATA3^mCherry iPSCs. Image is stitched. (D) Representative bright-field and fluorescent images of UE cultures established from MAFB^mTagBFP2:GATA3^mCherry kidney organoids. Scale bars, 100 μm. (E) Immunofluorescence analysis of day 17 UE cultures using commonly used markers of UE (GATA3, PAX2, KRT8, SOX9 and CALB1). Scale bars, 100 μm. (F) Comparison of the transcriptional profiles of populations derived from kidney organoids to human fetal kidney (HFK) using the “gene set scoring” method. While the mCherry⁺/GATA3⁺ epithelium within kidney organoids (Ch+/Ep+_org) displayed a more distal nephron identity, after culture in UE medium (Ch+/Ep+_UE), this transitioned towards a ureteric tip identity. (G) Heatmap showing expression of UE-specific and DN-specific genes (identified from HFK dataset) in Ch+/Ep+_org and Ch+/Ep+_UE populations.

Bulk RNAseq transcriptional profiling was performed on the mCherry⁺/EPCAM⁺ epithelium, both at the time of isolation from kidney organoids (Ch⁺/Ep⁺-org) and after culture in UE-promoting conditions (Ch⁺/Ep⁺-UE). For comparison, we also profiled the mCherry⁺/EPCAM⁻ mesenchymal (Ch⁺-org) and mCherry non-expressing epithelial (Ep⁺-org) populations directly after FACS-purification. To more definitively assign cellular identity to each of these populations in an unbiased fashion, we compared their transcriptional profiles to the HFK dataset using a ‘gene set’ scoring approach. Here, unique gene sets were determined for each of the 29 cellular subtypes present within the HFK dataset by comparing differentially expressed genes and identifying those most distinctly expressed within each cell type (see STAR methods). These gene sets were then compared to a pseudobulk of the HFK dataset to generate a “score of similarity” between each cell type (Figure S3A). This approach suggested an ability to use gene set scoring to assign identities to organoid cell clusters that correspond to specific segments within HFK. When applied to the organoid-derived populations, this unbiased gene set scoring suggested that while the Ch⁺/Ep⁺-org population at isolation exhibited a DN / connecting segment identity, this shifted with culture in UE medium such that the resulting Ch⁺/Ep⁺-UE cells showed the strongest congruence to ureteric tip (Figure 2F). Two-way differential expression analysis between mCherry⁺ epithelium before and after culture in UE-promoting conditions (Ch⁺/Ep⁺-org vs Ch⁺/Ep⁺-UE) revealed 573 upregulated and 766 downregulated genes (Figure S3B). Notably, UE-specific markers such as RET, KANK4, MMP7, MOXD1, KCNN4, SLC52A3, SLC9A2 and WNT11 were upregulated in the cultured mCherry⁺ epithelium while there was a corresponding downregulation of DN markers such as TMEM52B, KCNJ1, DEFB1 and CLCNKB (Figure 2G and Table S3). Together, this data suggests that while the mCherry⁺ epithelium within kidney organoids exhibits strongest transcriptional congruence with human fetal DN, this epithelium can undergo a transition towards a transcriptional profile that most closely resembles ureteric tip in response to culture in UE medium.

It has previously been assumed that the UE and nephron epithelia would retain HOX gene expression reflective of their distinct origins. As such, expression of the posterior HOX10/11 clusters would indicate a nephron origin whereas a bona fide UE would not express these more posterior HOX genes to the same extent (Ransick et al., 2019). Upon examination of HOX gene expression within the HFK dataset within all DN and UE subclusters, we observed a distinct suppression of HOXC cluster genes in the UE subcompartments (Figure S3C). Interestingly, suppression of HOXC cluster gene expression was also observed in the Ch⁺/Ep⁺-UE after culture, implying that HOX code can also be modified via the culture conditions and is not fixed by lineage (Figures S3C). This suggests that the DN epithelium within organoids exhibits fate plasticity and is responsive to exogenous cues.

We also assessed the capacity for cell types other than the GATA3⁺ epithelium to form budding UE-like structures (Figure S4A). While UE cultures could not be established from sorted GATA3⁺ stromal and MAFP⁺ podocyte populations (Figure S4B and S4C), the EPCAM⁺/GATA3⁻ fraction could give rise to budding UE structures, albeit at a lower effciency compared to the GATA3⁺ epithelial fraction (Figure S4B). GATA3 expression was also induced in these structures, as evidenced by expression of the mCherry reporter, after approximately 1 week of culture in UE medium (Figure S4D). The EPCAM⁺/GATA3⁻ population encompasses both GATA3⁻ regions of the DN (N. Distal_EN, N.LOH) together with the proximal tubule. To specifically determine whether proximal tubule cells can adopt a UE fate, we FACS-purified the YFP⁺ fraction from kidney organoids generated from iPSCs harbouring the YFP reporter inserted into the proximal tubule marker, HNF4A (Vanslambrouck et al., 2019a). Although the YFP⁺ sorted fraction could give rise to small epithelial colonies when plated at high density (20–50 K cells per well) (Figure S4B and S4E), GATA3 expression was not detectable even after 3 weeks of culture in UE medium (Figure S4F). Together, these results suggest that the capacity to transition towards a UE identity is most likely restricted to cells within the connecting segment, DN and loop of Henle nephron segments.

Induced UE cultures can be passaged, cryopreserved and established from iPSCs with diverse genetic backgrounds

Having developed a capacity to induce and expand a relatively homogenous population of human UE from kidney organoids, we next evaluated the feasibility of re-establishing budding structures following the dissociation, cryopreservation and subsequent thawing of an existing culture. In these experiments, mCherry⁺ epithelium FACS-purified from kidney organoids was expanded in UE medium for 2 weeks prior to enzymatic dissociation into a single cell suspension (Figure S5A). The cells were cryopreserved in 10% DMSO using a controlled-rate freezer and then transferred to liquid nitrogen to facilitate long term storage. Frozen vials could be retrieved from liquid nitrogen, thawed and transferred back into UE medium with minimal loss of cell viability (data not shown). Moreover, thawed cultures maintained GATA3-driven mCherry reporter expression and were able to reform the complex, 3-dimensional budding structures that were observed prior to dissociation and cryopreservation (Figure S5A). The integrity of the reconstituted structures was also confirmed by whole mount immunofluorescence (Figure S5B). Bulk RNAseq transcriptional profiling was performed on UE cultures that had undergone: 1) dissociation and replating (passaged); 2) cryopreservation, thawing and replating (freeze/thaw); or 3) extended culture (3 weeks), to examine if these processes affected cell identity. Using the HFK gene set scoring analysis described above, we demonstrate that a stable ureteric tip identity for all cultures was maintained (Figure S5C).

Because the UE cultures established previously were generated from GATA3⁺ epithelium that had been FACS-purified directly from kidney organoids, we next explored the possibility of establishing human UE cultures from existing organoids without this purification step. In these experiments, kidney organoids derived from MAFB^mTagBFP2:GATA3^mCherry iPSCs were dissociated and plated at varying densities (2–50 K cells per well) in UE medium (Figure 3A and Figure S5D). After 1–2 weeks, we observed the appearance of budding mCherry⁺ structures that exhibited a morphology comparable to sorted UE cultures (Figure 3B). A comparable number of budding mCherry⁺ structures was observed when established with or without a FACS-purification step (Figure S5D). While unsorted UE cultures initially contained cells that did not express the mCherry reporter, presumably carried over from the initial organoid dissociation step, after a single round of dissociation and replating, >99% of cells exhibited high levels of mCherry expression in the passaged cultures (Figure 3C).

Figure 3. Induced ureteric epithelial cultures can be established in the absence of fluorescent activated cell sorting.

(A) Overview of the strategy used to establish UE cultures from kidney organoids in the absence of a FACS-purification step. (B) Budding mCherry⁺ structures established from MAFB^mTagBFP2:GATA3^mCherry kidney organoids in the absence of FACS-purification. Scale bars, 100 μm. (C) Flow cytometry analysis of an unsorted UE culture established from MAFB^mTagBFP2:GATA3^mCherry kidney organoids that initially contained <10% mCherry⁺ cells. The UE culture was passaged after 14 days and analysed after an additional 14 days. (D) Immunofluorescence analysis of kidney organoids derived from HNF4A^YFP iPSCs. Scale bars, 100 μm. Left image is stitched. (E) Immunofluorescence analysis of UE cultures established from HNF4A^YFP kidney organoids, in the absence of a FACS-purification step. Scale bars, 100 μm. (F) Brightfield and fluorescent images of UE cultures established from WNT9B^mCerulean iPSCs. Scale bars, 100 μm. (G) Flow cytometry analysis of WNT9B^mCerulean UE cultures.

Since unsorted UE cultures can be derived from kidney organoids that contained <10% mCherry⁺ cells, this suggests that our UE culture conditions selectively favours propagation of the mCherry⁺ epithelium. These findings indicate that it may be possible to establish UE cultures from other human pluripotent stem cells. To test this hypothesis, we attempted to establish UE cultures from additional iPSC lines that do not carry DN-specific reporters. Using our modified differentiation protocol we generated kidney organoids from HNF4A^YFP iPSCs, which were dissociated at day 25 and seeded at at low density (5–10 K cells per well) in UE culture medium. This resulted in the appearance of budding epithelial structures arising 1–2 weeks after plating, which expressed common UE markers such as KRT8, PAX2, SOX9, GATA3, and RET as determined by whole mount immunofluorescence (Figure 3E). We have also established UE cultures from several additional iPSC lines, including one that harbours the mCerulean fluorescent reporter under the control of the endogenous WNT9B locus (Figure S5E–G). Although a classic and specific marker of UE, WNT9B expression is not observed in kidney organoids generated using standard differentiation protocols. However, reporter gene expression was observed in UE cultures established from WNT9B^mCerulean iPSCs (Figure 3F) and was detected in >35% of cells as determined by flow cytometry (Figure 3G). We have also established UE cultures from iPSCs that do not contain any fluorescent reporters, including the female fetal fibroblast line CRL-1502 (Takasato et al., 2015) (Figure S5H and S5I).

UE cultures exhibit a ureteric tip identity

To interrogate the cellular composition of the UE cultures further, we performed single cell transcriptional profiling of budding structures generated from two independent iPSC lines (MAFB^mTagBFP2:GATA3^mCherry and SIX2^CreGAPDH^Dual) (Vanslambrouck et al., 2019a) in the absence of a FACS-purification step. Data from both samples was integrated to generate a single dataset of 12,359 cells, with tight congruence between cellular identity from both samples (Figure 4A). We then utilised the annotated HFK dataset to perform a “label transfer” within the Seurat package (Stuart et al., 2019). Here, the labelling from the reference (HFK) dataset is transferred onto a query (UE culture) dataset. This enables a cellular identity to be assigned to each and every cell in our UE culture dataset, based upon the cellular compartment that the cell most closely aligns with in the HFK dataset (Figure 4B). In this context, the highest proportion of cells (6042, 48.9%) were identified as ureteric tip, while 26.4% (3257) of cells appeared to retain an identity more akin to early DN (Figure 4B and 4C). Nonetheless, label transfer identified the vast majority of cells (8664, 70.1%) as having a UE subcluster identity. The label transfer process identifies the most similar reference segment for each individual cell based on its maximum similarity score. Upon close analysis, the majority of cells identified as U.Tip, N.DST, U.Med_Inner and S.Mesangial exhibited a similarity score >50% (Figure 4C). To further validate the label transfer calls, feature plots for key genes were also examined across the dataset. This revealed that RET and WNT11 were broadly expressed, supportive of a ureteric tip identity (Figure 4D). WNT9B was also expressed in a subset of cells at low levels (Figure 4D) and is consistent with the reporter gene expression detected in UE cultures established from WNT9B^mCerulean iPSCs (Figure 3F). A similar pattern of WNT9B expression was also observed within the UE cluster of the HFK dataset. While very few AQP2-expressing cells were detected in induced UE cultures, the uroplakin gene, UPK2, was expressed in a distinct subset of cells predicted to represent inner medullary UE. This gene is specifically expressed in ureteric epithelial cells of the human fetal kidney (Figure 1E) although uroplakin gene expression is also seen in ureter and urothelium.

Figure 4. Single cell transcriptional profiling of induced ureteric epithelial cultures reveals a prominently ureteric tip identity.

(A) UMAP overlay of UE cultures established from two iPSC lines. The structures isolated for scRNAseq analysis are shown (inset). (B) UMAP plot with individual cells coded according to the “label transfer” analysis, which assigns an identity to each cell in the UE culture dataset. This is based upon the cellular compartment that the cell most closely aligns with in the HFK dataset. (C) Distribution of the maximum score from the label transfer of HFK annotation onto the single cell UE culture datasets. (D) Gene expression profile of ureteric tip (RET, WNT11), ureteric stalk (WNT9B, WNT7B, AQP2, UPK2), and markers in UE cultures.

The ureteric tip gives rise to all derivative components of the collecting duct epithelium. To functionally validate the identity of the cells present within our UE cultures, we used a reaggregation assay previously developed to assess cellular identity of exogenous cell types (Davies et al., 2012; Hendry et al., 2013; Vanslambrouck et al., 2019b) including UE (Xia et al., 2013). Dissociated E13.5 embryonic kidney from a Hoxb7-GFP transgenic mouse strain (Srinivas et al., 1999) was reaggregated with dissociated 14 day mCherry⁺ UE cultures at a ratio of 9:1 and analysed by immunofluorescence 6 days later (Figure 5A). Using an anti-RFP (mCherry) antibody to definitively identify human cells and an anti-GFP antibody to mark the mouse collecting duct, we observed consistent and specific integration of human UE cells within the collecting duct compartment of the organoid co-cultures (Figure 5B). Converesly, we did not observe integration of human mCherry⁺ cells within CDH1⁺/Hoxb7⁻ DN structures.

Figure 5. Kidney recombination assay confirms UE identity.

(A) Outline of the embryonic mouse recombination assay. (B) Immunofluorescence analysis demonstrating three examples of the integration of human mCherry⁺ cells (red) to endogenous Hoxb7⁺/GFP⁺ (green) mouse collecting duct structures in an embryonic mouse recombination assay. Cells did not contribute to CDH1⁺/ Hoxb7⁻ DN structures (arrow heads). Insets show CHD1 staining only. Scale bars, 50 μm.

UE cultures show functional properties of nephron induction and fusion.

In vivo, the ureteric tip induces the nephron progenitor population to commit to nephron formation with the resulting early nephrons fusing to the tip to make a patent epithelium (Carroll et al., 2005; Georgas et al., 2009). As UE cultures displayed a ureteric tip-like identity, we evaluated the capacity for UE cultures to induce human iPSC-derived nephron progenitor (NP) commitment (Figure 6A, Figure S6). UE / NP “hybrid” organoids were established in the absence of a CHIR (WNT) pulse, which is typically required for the commencement of effective nephrogenesis in kidney organoids (Takasato et al., 2015) (Figure 6B, Figure S6C). NP cultures were generated using either SIX2^EGFP (marks nephron progenitors, Figure 6) or MAFB^mTAGBFP (marks podocytes; Figure S6) reporter iPSC lines (Vanslambrouck et al, 2019a). These were differentiated for 5 days in the presence of CHIR followed by 5 days of FGF9, which favours specification of NPs (Howden et al., 2019; Vanslambrouck et al., 2019a). NP cultures were dissociated at day 10 and combined with dissociated UE cultures generated using either MAFB^mTagBFP2:GATA3^mCherry (Figure 6) or GATA3^mCherry alone (Figure S6) reporter iPSC lines respectively. UE / NP hybrid organoids were cultured in medium without growth factors for an additional 18 days, at which point the formation of both patterning and segmenting nephrons and UE-derived mCherry⁺ epithelial structures could clearly be observed, indicating the induction of nephrons by the UE (Figure 6C, Figure S6B). Analysis of UE / NP hybrid organoids by whole mount immunofluorescence revealed nephrons containing MAFB⁺ podocytes and EPCAM⁺/GATA3⁺/mCherry⁻ DN tubules (derived from NP culture), some of which formed continuous connections to GATA3⁺/mCherry⁺ epithelial structures (derived from UE culture), illustrating fusion of nephrons with the UE (Figure 6D). When UE / NP hybrids were generated between SIX2^EGFP NP and MAFB^mTagBFP2:GATA3^mCherry UE, podocytes showed no mTagBFP fluorescence (Figure 6C), however when generated between MAFB^mTagBFP NP and GATA3^mCherry UE, there was clear evidence of podocyte/proximal nephron-specific mTagBFP2 fluorescence (Figure S6B), showing that nephrons formed specifically from the NP. Collectively, these findings suggest that UE cultures can induce human iPSC-derived NPs to form segmented nephrons in the absence of an exogenous WNT signal and that these nephrons can fuse to the ureteric tip epithelium.

Figure 6. UE cultures show functional properties of nephron induction and fusion.

(A) Outline of the strategy to evaluate capacity of UE cultures to induce and fuse with nephrons. (B) Brightfield images showing impaired nephrogenesis in kidney organoids from SIX2^EGFP iPSCs in the absence of a CHIR pulse. Scale bars, 200 μm. (C) Brightfield and fluorescent images of a “hybrid” kidney organoid generated from the aggregation of human UE derived from the MAFB^mTAGBFP:GATA3^mCherry dual reporter line and NPs derived from the SIX2^EGFP reporter line co-cultured in the absence of a CHIR pulse. (D) Immunofluorescence analysis of the same UE/NP hybrid organoid showing segmented nephrons (derived from NP culture) connected to mCherry⁺ epithelial structures (derived from UE culture). Scale bars, 100 μm. Top image is stitched.

Induced UE cultures can transition to a ureteric stalk identity

Although single cell profiling of induced UE cultures suggested a predominantly ureteric tip identity, these cells are considered as the progenitors that give rise to stalk cells, which further differentiate into the more mature cell types within the collecting duct. This differentiation process is associated with declining RET expression and reduced cell proliferation (Shakya et al., 2005; Watanabe and Costantini, 2004). Yuri et al previously demonstrated a remarkable plasticity between cultured mouse ureteric tip and stalk cells in vitro, with a substantial upregulation of ureteric stalk markers, such as Wnt7b, in the absence of Wnt signalling (Yuri et al., 2017). We also examined whether we could shift UE cultures toward a more ureteric stalk-like identity by removing both WNT and GDNF signalling from the culture medium. In these experiments, 14 day UE cultures derived from MAFB^mTagBFP2:GATA3^mCherry kidney organoids were passaged and CHIR / GDNF was withdrawn from the medium (4 days post-passage) and cultured for an additional 10 days. While UE cultures maintained high levels of GATA3-driven reporter expression in the absence of CHIR / GDNF (Figure S7A and Figure S7B), we observed a marked reduction in the formation of budding structures and cell proliferation. Transcriptional profiling by bulk RNAseq revealed a shift away from ureteric tip towards a more cortical UE / medullary UE like phenotype, although this was observed in only two of the three replicates (Figure 7A). Moreover, while WNT7B and CALB expression was upregulated substantially in UE cultured in the absence of CHIR and GDNF, AQP2 expression remained virtually undetectable (Figure 7B), suggesting a lack of more mature cells of the collecting duct, such as the AQP2⁺ PCs. PCs express the vasopressin receptor (AVPR2) and respond to vasopressin by relocating AQP2 to the apical luminal membrane of the cell (Grassmeyer et al., 2017; Katsura et al., 1995). PCs should also express the ENaC sodium channel and upregulate this gene in response to aldosterone signalling via the NR3C2 nuclear hormone receptor (Pearce et al., 2015). Furthermore, PC identity is regulated via the transcription factor ELF5 (Grassmeyer et al., 2017). While expression of these genes is not exclusive to collecting duct, expression of ELF5, AVPR2 and NR3C2 was clearly detected in cultures grown in the absence of GDNF and CHIR (Figure S6C). In light of these findings and the recent report that aldosterone and vasopressin improves the maturation of iPSC-derived collecting duct cells (Uchimura et al., 2019), we tested the effect of aldosterone and vasopressin (1–100 nM) addition to the culture medium. This led to an induction of AQP2 expression, with the highest response in cells treated with the lowest dose of vasopressin (1 nM) (Figure S7D). Upregulation of the ureteric stalk marker WNT7B and a corresponding downregulation of the tip marker WNT11 was also observed in cultures grown in the presence of aldosterone and vasopressin as determined by both qRT-PCR (Figure S7F) and bulk RNAseq analyses (Figure 7A). Moreover, when compared to the HFK dataset using the gene set scoring method described previously, a more robust shift towards a cortical UE / medullary UE phenotype was observed in all three replicates (Figure 7A).

Figure 7. UE cultures can be transitioned towards a ureteric stalk identity and model autosomal recessive polycystic kidney disease.

(A) Heatmap comparing expression of ureteric tip-specific and ureteric stalk specific-specific genes in standard UE (tip) cultures and cultures grown in the absence of CHIR (C) and GDNF (G) with and without the addition of aldosterone (A) and vasopressin (V). (B) Comparison of the transcriptional profiles of standard UE (tip) cultures and cultures grown in the absence of CHIR (C) and GDNF (G) with and without the addition of aldosterone (A) and vasopressin (V) to the HFK dataset using the “gene set scoring” approach. (C) Strategy used to knock-out PKHD1 in GATA3^mCherry iPSCs using CRISPR/Cas9. (D) Overview of the strategy to establish UE stalk cultures from PKHD1^null and PKHD1^wt iPSCs. (E) Representative brightfield and fluorescent images of PKHD1^null and PKHD1^wt stalk cultures. Images were taken after 11 days in stalk medium. Scale bars, 100 μm. Quantification of cyst area (F) and cyst perimeter (G) in PKHD1^null and PKHD1^wt stalk cultures. Statistical analysis was performed using Welsch’s t test (unpaired data with variable distributions); ****p<0.0001. n=52 for PKHD1^null, n=63 for PKHD1 ^wt.

PKHD1^null UE stalk cultures recapitulate cystogenesis

A capacity to generate and maintain a relatively uniform culture of primary collecting duct epithelium poses the question of whether we can accurately model diseases of the collecting duct. ARPKD is a fibrocystic disease of the kidneys and liver. Approximately 95% of cases are caused by homozygous or compound heterozygous mutations in the PKHD1 gene (Bergmann, 2017). Within the kidney, the renal collecting ducts in patients with ARPKD display cystic expansion which obliterates adjacent functioning kidney tissue. To investigate whether we could model ARPKD in vitro, we used the CRIPSR/Cas9 gene-editing system to elicit disruption of the PKHD1 locus in GATA3^mCherry iPSCs. This involved the use of a pair of crRNAs that bind within exon 2 and exon 3 of PKHD1 (Figure 7C and Figure S7E) to facilitate the homozygous deletion of the intervening sequence and the generation of a premature stop codon approximately within exon 4 of the PKHD1 coding region (Figure S7F and S7G). GATA3^mCherry:PKHD1^null iPSCs were differentiated into kidney organoids alongside GATA3^mCherry (PKHD1^wt) iPSCs. Organoids from both iPSC lines showed similar morphology and composition as determined by whole mount immunofluorescence (Figure S7H). The mCherry⁺ epithelium from PKHD1^null and PKHD1^wt organoids was purified by FACS, cultured in UE-promoting conditions and then passaged after 2 weeks before transitioning to stalk medium (FGF2, RA, Y-27632, aldosterone, vasopressin) for an additional 10–14 days (Figure 7D). PKHD1^null and PKHD1^wt UE cultures were indistinguishable when grown in complete UE medium (FGF2, RA, Y-27632, CHIR, GDNF) (Figure S7I). However, when transitioned to stalk medium, the spontaneous formation of large dilated structures was apparent within the PKHD1^null ureteric stalk cultures. These cyst-like structures were consistently observed in PKHD1^null but not PKHD1^wt ureretric stalk cultures (Figure 7E). Moreover, quantitative analyses revealed a significant increase in both cyst area (Figure 7F) and cyst perimeter (Figure 7G) in PKHD1^null cultures compared to PKHD1^wt cultures. Whilst these cystic structures arose spontaneously in PKHD1^null stalk cultures, this phenotype could also be exacerbated following overnight exposure to forskolin (Figure S7J and S7K), which acts to further increase intracellular cAMP. Collectively, these findings validate the robustness of our strategy for establishing UE cultures and their utility for modelling diseases of the collecting duct, such as ARPKD.

DISCUSSION

Recent advances in the directed differentiation of pluripotent stem cells to kidney cell types have significantly advanced our ability to study human kidney development and disease. The formation of UE consisting of proliferative tip and maturing stalk regions is a major target for the field. However, the transcriptional overlap between the DN / connecting segment and the UE has represented a considerable challenge and can result in misleading interpretations. Based on a careful dissection of gene expression within the UE and DN within the developing HFK, we redefine the GATA3⁺ epithelium present in previously described kidney organoids as DN. However, the subsequent culture of this epithelial population in the presence of a WNT agonist (CHIR), retinoic acid, GDNF and FGF2 shifts the cellular identity of this population to UE. This manifests as reduced expression of DN genes and an upregulation of ureteric tip genes, including RET, WNT9B, KCNN4 and WNT11. This induced UE population was able to be cultured long term and retain identity even after cryopreservation and thawing, facilitating the substantial expansion of UE cultures. Single cell transcriptional profiling of induced UE identified one main epithelial cluster which, when directly compared with HFK, most tightly correlated with ureteric tip. A smaller subset of cells showing initiation of UPK2 expression was also detected, suggesting the formation of a watertight stalk within the 3D culture. When combined with mouse embryonic kidney, induced human UE cells specifically integrated into the mouse Hoxb7⁺ epithelium. More critically, induced UE was able to both induce nephron formation and fuse with these newly formed nephrons, both critical functionalities of a bona fide ureteric tip. In response to withdrawal of WNT and GDNF signalling and the addition of vasopressin and aldosterone, UE cultures could be transitioned away from a RET⁺ ureteric tip towards cortical and medullary UE identities, facilitating the accurate modelling of cyst formation in the context of PKHD1 loss-of-function mutations.

The central dogma that the UE and nephron epithelia arise from distinct progenitor populations during kidney organogenesis has driven the hypothesis that these populations should be derived separately from human pluripotent stem cells using distinct differentiation protocols (Taguchi et al., 2014; Taguchi and Nishinakamura, 2017). Several groups have previously reported the selective specification of UE-like cells directly from mouse or human pluripotent stem cells (Mae et al., 2020; Mae et al., 2013; Taguchi and Nishinakamura, 2017; Tan et al., 2020; Xia et al., 2013). Here we report an alternative approach, whereby UE is instead induced from a plastic DN population in response to exogenous growth factor signalling. This transition in cell fate is also accompanied by an erasure of the pre-existing HOX expression code. Although there is little evidence to suggest that this level of plasticity exists between DN and UE during normal embryogenesis, plasticity within the UE and DN segments of the postnatal kidney has in fact been reported (Assmus et al., 2020; Takito et al., 1996; Trepiccione et al., 2016), whereby a transition between PC and IC fate occurs in response to drug treatment (Trepiccione et al., 2013) and renal challenge (Iervolino et al., 2020).

Marked fate plasticity has also recently been uncovered following the directed differentiation of human pluripotent stem cells towards other lineages, such as lung epithelium. NKX2–1⁺ progenitors can be shifted between proximal and distal lung epithelial identities via changes to exogenous growth factors (McCauley et al., 2017). Cellular plasticity can also result in reversion of progenitors to non-lung epithelium cell fates, although this declines over time as cells mature, allowing the propagation of PSC-derived lung cells with more stable phenotypes (Hurley et al., 2020). Whether the plasticity between DN and UE reported in our study is limited in time to a developmental window is yet to be determined. However we clearly demonstrate the maintenance of a stable UE phenotype across passage and freeze/thaw together with a continued capacity to shift from ureteric tip to ureteric stalk.

Although a number of distinct protocols have now been described which report the generation of UE from mouse or human pluripotent stem cells (Mae et al., 2013; Taguchi and Nishinakamura, 2017; Tan et al., 2020; Xia et al., 2013), cellular identity is mostly inferred from qRT-PCR and/or immunofluorescence for markers that are often observed in both UE and DN. In this study, we perform an unbiased evaluation of cellular identity based on global and single cell expression profiling. We adopt an unbiased gene set prediction approach, based upon the transcriptional analysis of HFK, to provide a greater degree of certainty around the identity of the cell types generated in vitro. A similar approach was recently described to thoroughly characterise the podocyte population in kidney organoids (Tran et al., 2019). Such studies will continue to provide a better framework for the reanalysis and refinement of existing protocols.

The development of robust methods for the specification of cell lineages from pluripotent stem cells without the need to enrich for progenitors by FACS or other means is also advantageous. The inherent plasticity of progenitors can lead to drift back towards heterogeneous mixes of target and non-target cell types even after enrichment, as observed in the specification of lung (Hurley et al., 2020; McCauley et al., 2018), and in approaches for enriching of pluripotent fractions (Hough et al., 2009). The ability to generate and selectively expand UE in the absence of FACS enrichment represents a considerable advantage for both cellular manufacturing and disease modelling applications. We demonstrate that induced UE cultures can be established from numerous independent iPSC lines with diverse genetic backgrounds and without the need for FACS enrichment.

A capacity to robustly expand relatively homogenous populations of UE also enables the development of more accurate in vitro models of disease states, such as ARPKD. Primarily caused by mutations in PKHD1, ARPKD is a severe kidney disorder that presents very early in life (sometimes in utero) as cyst formation specifically within the ureteric epithelial compartment. A recent study reported a patient-iPSC derived model of ARPKD using a differentiation protocol similar to that described by Takasato et al (2015) in which differential tubular cyst swelling was observed in ARPKD patient-specific kidney organoids compared to isogenic (gene-corrected) control organoids (Low et al., 2019). Although the kidney organoids also lacked a bona fide UE, cyst swelling was reportedly localised to the proximal tubule following exposure to forskolin (Low et al., 2019). Whilst transient and self-resolving proximal tubule dilation has been reported in knockout mouse models and one report in human fetal kidney (prior to the discovery of PKHD1) (Nakanishi et al., 2000) the definitive disease process in vivo is invariably restricted to the collecting duct. Here, we show spontaneous cyst formation in PKHD1 null ureteric stalk cultures in the absence of forskolin addition, as evidenced by the enlarged dilated epithelial structures in mutant cultures compared to wildtype isogenic controls, thereby successfully modelling an ARPKD phenotype.

In summary, this study has used a comprehensive analysis of the developing human DN and UE to develop a more stringent approach with which to assess the successful generation of UE in vitro. Using this knowledge, we describe the selective propagation of a UE population after isolation as DN epithelium from a standard kidney organoid culture. This suggests that the cellular properties of the UE can be adopted by DN elements in vitro, highlighting the potential to engineer this region of the kidney from DN. It also provides an accurate model of ARPKD cyst formation in vitro within the appropriate epithelial context.

Limitations of Study

This protocol for generating UE requires the induction of UE from a DN population first established in the context of a kidney organoid. As such, this protocol requires an additional 2–3 weeks compared to previously described protocols where UE is derived directly from iPSCs (Mae et al., 2020; Taguchi and Nishinakamura, 2017). However, it does not require cell sorting and the added convenience of stable cell identity, even after serial passage, cryopreservation and thawing, facilitates the generation of expansive stocks.

While transcriptional profiling and functional assays presented here validate the identity of the UE cultures generated in this study, as yet we have not performed an epigenetic analysis to further corroborate a transition from DN to UE identity in vitro. Indeed, further characterization of stalk cultures is also warranted to gain a better understanding of the cellular composition of these cultures, particularly the relative proportion of PCs and ICs. The generation of reporter lines that specifically mark PCs and/or ICs would aid in the optimization of conditions that favour the specification of each of these key collecting duct cell types. There are also many biological questions around how the PKHD1 mutation leads to cyst formation. Importantly, using the model developed in this study, we can now begin to address these questions, hopefully leading to a greater understanding of ARPKD and the development of novel treatments.

STAR METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Melissa H. Little (melissa.little@mcri.edu.au).

Materials Availability

The cell lines generated in this study are available upon request.

Data and Code Availability

The datasets generated during this study are available at Gene Expression Omnibus, under GEO: GSE161255. Code related to analyses in this study is available at https://github.com/KidneyRegeneration/HowdenWilson2020

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell lines

All iPSC lines were maintained and expanded at 37°C, 5% CO₂ and 5% O₂ in Essential 8 medium (Thermo Fisher Scientific) on Matrigel-coated (Corning) plates with daily media changes and passaged every 3–4 days with EDTA in 1X PBS as previously described (Chen et al., 2011). The genomic integrity of iPSCs were confirmed by molecular karyotyping using Infinium CoreExome-24 v1.1 SNP arrays (Illumina), performed by Victorian Clinical Genetics Services (Melbourne, Australia).

Human samples

Use of human tissue was reviewed and approved by The University of Michigan Institutional Review Board. De-identified human fetal kidney tissue was obtained from the University of Washington Laboratory of Developmental Biology. Tissue was shipped overnight in Belzer-UW Cold Storage Solution (Thermo Fisher Scientific), with cold packs, as previously published (Menon et al., 2018; Miller et al., 2020).

METHOD DETAILS

Human fetal kidney single cell RNA sequencing

Human fetal tissue was dissociated as previously described (Miller et al., 2020). Briefly, tissue was mechanically minced into small fragments in a petri dish filled with ice-cold 1X HBSS (with Mg2+, Ca2+). The tissue was then transferred to a 15 ml conical tube and dissociation enzymes and reagents from the Neural Tissue Dissociation Kit (Miltenyi) were added. All incubation steps were carried out in a refrigerated centrifuge pre-chilled to 10°C and tubes and pipette tips used to handle cell suspensions were pre-washed with 1% BSA in HBSS to prevent adhesion of cells to the plastic. Tissue was treated for 15 min at 10°C with Mix 1. Mix 2 was added to the digestion, and tissue was incubated for 10 min increments at 10°C until digestion was complete. After each 10 min incubation, tissue was agitated using a P1000, and tissue digestion was visually assessed under a stereo microscope. This process continued until the tissue was fully digested. Cells were filtered through a 70 μm filter coated with 1% BSA in 1X HBSS, spun down at 500 g for 5 min at 10°C and resuspended in 500 μl 1X HBSS (with Mg2+, Ca2+). 1 ml Red Blood Cell Lysis buffer (Roche) was then added to the tube and the cell mixture was placed on a rocker for 15 min at 4°C. Cells were spun down (500 g for 5 min at 10°C), and washed twice with 2 ml of HBSS + 1% BSA followed by centrifugation. Cells were counted using a haemocytometer, then spun down and resuspended (if necessary) to reach a concentration of 1000 cells/μl and kept on ice. Single cell droplets were immediately prepared on the 10x Chromium according to manufacturer instructions at the University of Michigan The Advanced Genomics Core, with a target of capturing 5,000–10,000 cells. Single cell libraries were prepared using the Chromium Next GEM Single Cell 3’ Library Construction Kit v3.1 according to manufacturer instructions. ScRNAseq data was deposited at EMBL-EBI ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) under the accession number E-MTAB-9083.

Kidney organoid production

The day prior to differentiation, cells were dissociated with TrypLE (Thermo Fisher Scientific), counted using a haemocytometer and seeded onto Laminin521-coated 6-well plates at a density of 5 ×10⁴ cells per well in Essential 8 medium. Intermediate mesoderm induction was performed by culturing iPSCs in TeSR-E6 medium (Stem Cell Technologies) containing 7 μM CHIR99021 (R&D Systems) for 3 days. On day 3, cells were switched to TeSR-E6 medium supplemented with 600 ng/mL FGF2 (PeproTech) and 1 μg/mL Heparin (Sigma Aldrich). On day 7, cells were dissociated with TrypLE, diluted 5-fold with TeSR-E6 medium, transferred to a 15 ml conical tube and centrifuged for 5 minutes at 300 g to pellet cells. The supernatant was discarded, cells were resuspended in residual medium and transferred directly into a syringe for bioprinting. Syringes containing the cell paste were loaded onto a NovoGen MMX bioprinter, primed to ensure cell material was flowing, and 10 × 10⁵ aliquots (cells per organoid) were deposited on 0.4 μm Transwell polyester membranes in 6-well plates (Corning). Following bioprinting, organoids were cultured for 1 hour in the presence of 7 μM CHIR99021 in TeSR-E6 medium in the basolateral compartment and subsequently cultured until Day 12 in TeSR-E6 medium supplemented with 600 ng/mL FGF2 and 1 μg/ml Heparin. From Day 12 to Day 25, organoids were grown in TeSR-E6 medium supplemented with 2 μM all-trans retinoic acid (ATRA).

UE culture

Day 20–25 organoids were dissociated with a 1:1 TrypLE/Accutase solution (~0.2 ml per organoid) at 37°C for 15–25 min, with occasional agitation (flicking) until large clumps were no longer clearly visible. The cell suspension was then centrifuged (3 min, 300 g) and the dissociation solution removed. Cells from organoids derived from iPSCs harbouring a GATA3 promoter driven mCherry reporter were resuspended in HBBS supplemented with 2% FBS and the mCherry⁺ fraction was isolated by fluorescent activated cell sorting (FACS) as described below. The purified mCherry⁺ cells were centrifuged and resuspended in UE medium (TesR-E6 supplemented with 200 ng/ml FGF2, 3 μM CHIR99021, 2 μM ATRA, 10 μM Y-27632 and 100 ng/ml GDNF), counted using a haemocytometer and transferred to 24-well Transwell inserts (5–10 × 10⁴ cells per well). In the absence of a FACS-purification step, cells from individual kidney organoid were resuspended in UE medium (2–5 × 10⁵ cells per ml) and 2–5 × 10⁴ cells plated per well of a 24-well Transwell plate. Cells were mounted in a 1:1 Matrigel:UE medium solution (placed on top of the Transwell) and 0.5 ml UE medium was placed in the basolateral subcompartment. Media was changed every other day. For passaging, UE cultures were dissociated by adding 0.5 ml of a 1:1 TrypLE/Accutase above and below the Transwell insert followed by incubation at 37°C for 15–25 min with occasional agitation by gentle pipetting using a P1000 pipette. The cell suspension was then transferred to a 1.5 ml tube and centrifuged (3 min, 300 g), resuspended in fresh UE medium (2–5 × 10⁵ cells per ml) and 2–5 × 10⁴ cells plated per well of a new 24-well Transwell plate. For transition to ureteric stalk, UE medium was replaced with stalk medium (TesR-E6 supplemented with 200 ng/ml FGF2, 2 μM ATRA, 10 μM Y-27632, 2 μM aldosterone, 1 nM vasopressin) four days post-passage, with medium changes every 2 days.

Flow cytometry and fluorescent activated cell sorting

Kidney organoids were dissociated as described above and resuspended in 1 ml of HBBS supplemented with 2% FBS. The cells were passed through a 40 μM FACS tube cell strainer (Falcon). Flow cytometry was performed using a LSRFortessa Celll Analyzer (BD Biosciences). FACS was performed using a FACSAria Fusion (BD Biosciences). Data acquisition and analysis was performed using FACsDiva (BD) and FlowLogic software (Inivai). Gating was performed on live cells based on forward and side scatter analysis.

Whole Mount Immunofluorescence

Organoid and UE cultures were fixed with ice cold 2% paraformaldehyde (Sigma Aldrich) for 20 min followed by 15 min washing in three changes of phosphate-buffered saline (PBS). For immunofluorescence, blocking and antibody staining incubations were performed on a rocking platform for 3 hours at room temperature or overnight at 4°C, respectively. Blocking solution consisted of 10% donkey serum with 0.3% Triton-X-100 (TX-100; Sigma Aldrich) in PBS. Antibodies were diluted in 0.3% TX-100/PBS. Primary antibodies were detected with Alexa Fluor-conjugated fluorescent secondary antibodies (Invitrogen), diluted 1:500. Organoids and UE cultures were washed in at least 3 changes of PBS for a minimum of 1 hour following primary and secondary antibody incubations. Imaging was performed in glass-bottomed dishes (MatTek) with glycerol-submersion using either the Zeiss LSM 780 or Andor Dragonfly Spinning Disk confocal microscopes. Image stitching was performed in FIJI using the Grid/Collection stitching function.

Single cell transcriptional profiling and data analysis

UE cultures were dissociated as described above and passed through a 40 μM FACS tube cell strainer. Following centrifugation at 300 g for 3 minutes, the supernatant was discarded and cells resuspended in 50 μl TeSR-E6 medium. Viability and cell number were assessed and samples were run across separate runs on a Chromium Chip Kit (10x Genomics). Libraries were prepared using Chromium Single Cell Library kit V3 (10x Genomics) and sequenced on an Illumina Novaseq with 100 bp paired-end reads. Cell Ranger (v1.3.1) was used to process and aggregate raw data from each of the samples returning a count matrix. Quality control and analysis was performed in R using the Seurat package (v3.1.4) (Stuart et al., 2019). Cells with less than 1500 genes, more than 100,000 UMIs expressed, less than 3% or more than 25% reads assigned to mitochondrial genes were filtered out. Data normalisation, scaling, variable gene identification and cell cycle phase identity regression were performed using the implementation of SCTransform within Seurat. Genes with less than two counts across the whole dataset were also filtered out. Integration of the two datasets was performed using the anchor based integration method with Seurat. The final integrated dataset had 12359 cells and 23578 identified genes. Cell type identification was performed by utilising the label transfer in Seurat using the GATA3⁺ populations from the human fetal kidney dataset as the reference.

Bulk cell transcriptional profiling and data analysis

For the analysis of cells populations within organoid, day 22 organoids were dissociated as described above and resuspended in HBBS + 2% FBS (50–100 μl per organoid). Cells were stained for 30 min with Alexa Fluor 488 anti-EpCAM antibody (Biolegend) followed by three washes with HBBS + 2% FBS and resuspension in 2 ml HBBS + 2%. Cell populations were sorted based on mCherry reporter expression and EpCAM-488. Following collection, the cells were pellet by centrifugation (3 min, 300 g) and RNA extracted using the Isolate II RNA Mini/Micro Kit (Bioline). Libraries were prepared using the Illumina TruSeq Stranded Total RNA library Prep Kit and sequenced on the Illumina NextSeq 500 Sequencing System (Illumina) with 75 bp single-end reads to a depth of ~20 M reads. Reads were trimmed for quality using Trimmomatic (v0.35) to remove bases from the ends of reads with lower than 25 quality and exclude reads less than 30 bp and Illumina adaptors. Trimmed reads were then mapped to the hg38 reference genome using STAR aligner (v2.5.2a) (Dobin et al., 2013) in two pass single end mode and gene level counts calculated using featureCounts (v1.5.0-p3) (Liao et al., 2014) and the GENCODE_V20 annotation. Analysis was performed using R. Genes with less than one count per million in 3 or more samples were excluded from analysis. Counts were normalised using TMM (Robinson and Oshlack, 2010). Differential expression analysis was performed using voom transformed counts (Law et al., 2014) and limma (Ritchie et al., 2015). Testing relative to a threshold (TREAT) (McCarthy and Smyth, 2009) analysis from limma was also performed, assessing for a log2-fold change greater than 1 (2-fold) when calculating differential expression.

Gene set scoring method

To characterise the transcriptional profile of the bulk profiling we leveraged the human fetal kidney dataset to generate lists of genes that were specifically enriched within each annotated cell type. This was performed using the FindMarkers function in Seurat to iteratively identify the differentially expressed genes for each cluster compared to every other cluster. This list was then filtered for genes with an average log-fold change greater than 1 and adjusted p-value less than 0.005. For each cluster, the genes were counted for the number of comparisons they were identified in. Genes that were identified as differentially expressed in a given cluster compared to at least 26 of the other 29 clusters were defined as specific and included in the “gene set” for that cluster. The AddModuleScore function from Seurat was then modified to operate on a voom object instead of a Seurat object and was used to generate scores for the lists of genes generated from the human fetal kidney dataset. Higher scores indicate higher expression of the genes as a group, and were used as a metric to measure the similarity of transcriptional identity to each annotated cell type within the bona fide human fetal kidney.

RNA purification and quantitative RT-PCR (qRT-PCR)

Total RNA was isolated from cells using the Isolate II RNA Mini/Micro Kit (Bioline) with DNase I treatment. 0.5–1 ug total RNA of each sample was reverse transcribed using the SensiFAST^™ cDNA Synthesis Kit (Bioline). Quantitative real-time PCR was carried out using the QuantStudio^™ 5 Real-Time PCR System (Applied Biosystems) and GoTaq® qPCR Master Mix (Promega). mRNA expression levels were quantified by the DCt method and normalised to GAPDH. All of the sample analyses were carried out in triplicates.

Embryonic mouse kidney recombination assay

Recombination assays were performed as described previously (Davies et al., 2012), with modifications to format and culture conditions (Vanslambrouck et al., 2019b), exogenous cell proportion (10% UE cells) and recombination culture media (10 μM Y27632, included for UE survival). Recombinations were harvested after 6 days of culture for fixation and immunofluorescence.

Generation of UE / NP hybrid organoids

NP cultures were differentiated from SIX2^EGFP or MAFB^mTagBFP2 iPSCs as previously described (Howden et al., 2019). Briefly, cells were dissociated with TrypLE (Thermo Fisher Scientific) the day prior to differentiation, counted using a haemocytometer and seeded onto Laminin521-coated 6-well plates at a density of 5 ×10⁴ cells per well in Essential 8 medium. Cells were then differentiated in TeSR-E6 medium (Stem Cell Technologies) containing 7 μM CHIR99021 (R&D Systems) for 5 days, with medium changes every other day. On day 5, cells were switched to TeSR-E6 medium supplemented with 200 ng/mL FGF9 (R&D Systems) and 1 μg/mL Heparin (Sigma Aldrich). On day 10, cells were dissociated with TrypLE, diluted 5-fold with TeSR-E6 medium, transferred to a 15 ml conical tube and centrifuged for 5 minutes at 300 g to pellet cells. Cells were resuspended in TeSR-E6 medium, counted using a haemocytometer and 2 ×10⁵ cells were placed into a 1.5 ml tube containing 2 ×10⁴ UE cells. UE was prepared by first removing Matrigel with 10 mg/ml dispase solution for 15 min at 37°C. The cells were then transferred to a 1.5 ml tube, pelleted and dissociated further by resuspending in 0.5 ml of 1:1 TrypLE / Accutase solution. After removal of the dissociation solution, cells were resuspended in TeSR-E6 medium supplemented with 10 uM Y-27632 and 2% FBS. The NP / UE mixture was aggregated by centrifugation and the pellet was transferred to a Transwell filter. The organoids were in TeSR-E6 medium supplemented with 10 uM Y-27632 and 2% FBS for 2 days and then switched to TeSR-E6 medium for an additional 16 days.

Generation of WNT9B mCerulean3-tagged iPSCs

BJFF.6 iPSCs was generated from human BJ (ATCC ID: CRL-2522) by the Genome Engineering and iPSC Center (GEiC) at Washington University (St. Louis) using the CytoTune-iPS 2.0 Sendai reprogramming kit (ThermoFisher) following the manufacturer’s recommended protocol. This line is described on the RBK website (https://www.rebuildingakidney.org/chaise/record/#2/Cell_Line:Parental_Cell_Line/RID=Q-2D6W). The WNT9B C-terminal mCerulean3 tagged cell line (WNT9B^mCerulean) was generated by GEiC using the BJFF.6 iPSC line. Approximately 1 to 1.5 × 10⁶ hPSCs were washed in DPBS and resuspended in P3 primary buffer (Lonza) with 1 μg of a WNT9B-specific gRNA (5’-AGACTGGCTTGCTGGGCAGT-3’) expression plasmid (cloned into MLM3636), 1.5 μg of the Cas9 expression vector, p3s-Cas9HC, and 1.5 ug of hWNT9B-mCerulean3 donor plasmid (GeneArt, ThermoFisher). The cells were subsequently electroporated with a 4D-Nucleofector (Lonza) using the CA-137 program. Following nucleofection, cells were then single-cell sorted as previously described (Chen and Pruett-Miller, 2018). Single cell clones were expanded and screened by PCR using primer sets that flank the 5’ (WNT9B_5’F and mCeruleanR) and 3’ (mCeruleanF and WNT9B_3’R) recombination junctions. The WNT9B-mCerulean3 tagged region of successfully targeted clones was also verified by Sanger sequencing.

Generation of PKHD1 knock-out iPSCs

A homozygous deletion was introduced into GATA3^mCherry iPCS using the Alt-R CRISPR-Cas9 system (Integrated DNA Technologies) using two crRNAs that bind within exon 2 (GAGTATTGAAGTACTACTTT) and exon 4 (TGATCCACGTTCCCCCTGCA) of the PKHD1 coding region. RNP complexes were formed according to manufacturer’s recommendations (Integrated DNA Technologies) and transfected into cells using the Neon Transfection System (Thermo Fisher Scientific) as follows. GATA3^mCherry iPSCs were harvested with TrypLE (Thermo Fisher) 2 days after passaging and resuspended in Buffer R at a final concentration of 1 × 10⁷ cells/ml. Electroporation was performed in a 100 μl tip using 1100 V, 30 ms, 1 pulse for human iPSCs. Electroporated human iPSCs were plated on 6-well Matrigel-coated plates containing Essential 8 medium with 5 μM Y-27632 (Tocris). Individual iPSC colonies were isolated and expanded and genomic DNA was isolated using the DNeasy Blood & Tissue Kit (QIAGEN) in accordance with the manufacturer’s protocol. Clones were screened by PCR analysis, performed on gRNA using ODNs that flank the crRNA target sites (PKHD1_AscF and PKHD1_SscR) and GoTaq Green Master Mix (Promega). Clones that yielded amplicons carrying the intended (~1.7 kb) deletion were expanded for differentiation. Amplicons were also sequence verified by Sanger sequencing (performed by the Australian Genome Research Facility).

Cyst quantification

Wild type and PKHD1-knockout ureteric stalk cultures were imaged blinded with a Dragonfly spinning disk confocal microscope 14 days after transition to stalk media. Fifty cysts were imaged per condition. Images were processed in ImageJ v1.52p. Using the Polygon selections tool, outlines were drawn manually around all cysts that did not cross the border of the image. Area and perimeter were output using the Analyze>Measure function. All cysts were measured for each image until over 50 cysts had been measured (n=52 for knockout and n=63 for wildtype). Graphs and statistical analyses were performed in GraphPad Prism v7.04. Graphs are presented with all data points in grey and superimposed bars representing mean and standard error of the mean. Significance of the difference between means was calculated with Welsch’s t test for unpaired data with variable distributions.

Supplementary Material

3

Supplementary Movie 1. Timelapse imaging of Cherry+ ureteric epithelial culture showing bud formation with time. Cultures were imaged after 13 days culture in UE medium, Related to Figure 2.

4

Supplementary Movie 2. Timelapse imaging of Cherry+ ureteric epithelial culture showing bud formation with time. Cultures were imaged after 13 days culture in UE medium, Related to Figure 2.

5

Table S1. Genes used to identify broad cell lineages in human fetal kidney, Related to Figure 1.

6

Table S2. Genes used to identify distinct cell types in human fetal kidney, Related to Figure 1.

7

Table S3. Top differentially expressed genes (logfold ranked output) for the following sample comparisons: Ch+/Ep+_org versus Ch+_org; Ch+/Ep+_org versus Ep+_org; Ch+/Ep+_org versus Ch+/Ep+_UE, Related to Figure 2.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse anti-EpCAM (Alexa Fluor 488 conjugated)	Biolegend	Cat#324210
Mouse monoclonal anti-GATA3	Thermo Fisher Scientific	Cat#MA1-028
Goat polyclonal anti-GATA3	R&D Systems	Cat#AF2605
Rabbit polyclonal anti-SL312A1	Proteintech	Cat#18970-1-AP
Mouse monoclonal anti-KRT8	Abcam AB115959	Cat#AB115959
Goat polyclonal anti-SOX9	R&D systems	Cat#AF3075
Rabbit polyclonal anti-PAX2	Zymed Laboratories Inc.	Cat#71-6000
Rabbit monoclonal anti-HNF4A	Life technologies	Cat#MA1-199
Mouse monoclonal anti-CALB1	Sigma Aldrich	Cat#C9848
Rat polyclonal anti-aPKC	Santa Cruz	Cat#sc-216
Biological Samples
Human fetal kidney	Human fetal kidney samples from 96 day male and 108 day female (gestational age) donors were obtained from the University of Washington Laboratory of Developmental Biology with oversight from The University of Michigan Institutional Review Board.	N/A
Chemicals, Peptides, and Recombinant Proteins
Y-27632	Tocris	Cat#1254
CHIR99021	R&D Systems	Cat#4423
Heparin	Sigma Aldrich	Cat#H4784
Human recombinant FGF9	R&D Systems	Cat#273-F9-025
Human recombinant FGF2	PeproTech	Cat#100-18B
Matrigel (growth factor reduced)	Corning	Cat#354277
Human recombinant GDNF	R&D Systems	Cat#212-GD
All-trans retinoic acid	Sigma Aldrich	R2625
Aldosterone	Sigma Aldrich	Cat#A9477
Arg-Vasopressin	Sigma Aldrich	Cat#V9879
Laminin521	BioLamina	Cat#77003
Critical Commercial Assays
SensiFAST™ cDNA Synthesis Kit	Bioline	Cat# BIO-65054
Isolate II RNA Mini Kit	Bioline	Cat#BIO-52073
Isolate II RNA Micro Kit	Bioline	Cat#BIO-52075
GoTaq® qPCR Master Mix	Promega	Cat#A6001
DNeasy Blood & Tissue Kit	QIAGEN	Cat#69506
GoTaq Green Master Mix	Promega	Cat#M7123
Deposited Data
Raw HFK single cell dataset	Holloway et al., 2020	EMBL-EBI ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) accession # E-MTAB-9083
Raw single cell RNA-Seq Ureteric Epithelial cultures	This study	GEO (https://www.ncbi.nlm.nih.gov/geo) accession # GSE161255
Analysed single cell RNA-Seq data	This study	GEO (https://www.ncbi.nlm.nih.gov/geo) accession # GSE161255
Raw and analysed bulk RNA-Seq data	This study	GEO (https://www.ncbi.nlm.nih.gov/geo) accession # GSE161255
Experimental Models: Cell Lines
Human iPSC line: MAFBmTagBFP2:GATA3mCherry	Vanslambrouck et al., 2019	N/A
Human iPSC line: MAFBmTagBFP2	Vanslambrouck et al., 2019	N/A
Human iPSC line: SIX2EGFP	Vanslambrouck et al., 2019	N/A
Human iPSC line: HNF4AYFP	Vanslambrouck et al., 2019	N/A
Human iPSC line: GAPDHDual	Howden et al., 2019	N/A
Human iPSC line: GATA3mCherry	Vanslambrouck et al., 2019	N/A
Human iPSC line: 1502	Takasato et al., 2015	N/A
Human iPSC line: GATA3mCherry:PKHD1null	This study	N/A
Human iPSC line: WNT9BmCerulean	This study	N/A
Human iPSC line: BJFF.6 iPSCs	RBK consortium	N/A
Oligonucleotides
PKHD1_AscF: AAAGCTATGCTGCTCCAATC	Integrated DNA Technologies	N/A
PKHD1_AscR: CAATCTAATGATTATGACAAG	Integrated DNA Technologies	N/A
PKHD1_CscR: CAGACAGGTATACCTGGTCC	Integrated DNA Technologies	N/A
WNT9B_5’F: AGAGACAGCTTTCCTGTACGC	Integrated DNA Technologies	N/A
WNT9B_3’R: GGACCCCTGCAAGGTCTTCAT	Integrated DNA Technologies	N/A
mCeruleanF: CTGGAGTACAACGCCATCC	Integrated DNA Technologies	N/A
mCeruleanR: AGGGGTCTTGTAGTTGCCGTCG	Integrated DNA Technologies	N/A
GAPDH_qRT_F: CTCTCTGCTCCTCCTGTTCGA	Integrated DNA Technologies	N/A
GAPDH_qRT_R: TGAGCGATGTGGCTCGGCT	Integrated DNA Technologies	N/A
AQP2_qRT_F: ACGTCTCCGTTCTCCGAGCC	Integrated DNA Technologies	N/A
AQP2_qRT_F: AGCCGTCGTGCTGTTGCTGAG	Integrated DNA Technologies	N/A
WNT11_qRT_F: TATCCGGCCTGTGAAGGACT	Integrated DNA Technologies	N/A
WNT11_qRT_F: GTCTTGTTGCACTGCCTGTC	Integrated DNA Technologies	N/A
WNT7B_qRT_F: GGCAAGAGCTCCGAGTAGG	Integrated DNA Technologies	N/A
WNT7B_qRT_R: GAGAAGTCGATGCCGTAACG	Integrated DNA Technologies	N/A
Recombinant DNA
p3s-Cas9HC	A gift from Jin-Soo Kim	Addgene #43945
MLM3636 (human gRNA expression vector)	A gift from Keith Joung	Addgene #43860
Software and Algorithms
SnapGene v4.1.2	GSL Biotech	https://www.snapgene.com/
FlowLogic v700.0A	Inivai Technologies	https://www.inivai.com/flowlogic
FACSDiva	BD Biosciences	http://www.bdbiosciences.com/
ZEN v.10.0.19	Zeiss	https://www.zeiss.com/
Seurat v3.1.4	Stuart, Butler et al. 2019	N/A
Other
Essential 8	Thermo Fisher Scientific	Cat#A1517001
TESR-E6	Stem Cell Technologies	Cat#05946
HBSS	Thermo Fisher Scientific	Cat#14025076
Fetal bovine serum	Hyclone	Cat#SH30084.03
Accutase	Stem Cell Technologies	Cat#07922
TrypLE Select	Thermo Fisher Scientific	Cat#12563029
Alt-R S.p. Cas9 Nuclease V3	Integrated DNA Technologies	Cat#1081058
Alt-R CRISPR-Cas9 crRNA	Integrated DNA Technologies	See methods for details related to crRNA sequence
Alt-R CRISPR-Cas9 tracrRNA	Integrated DNA Technologies	Cat#1072532
Neon Transfection Kit	Thermo Fisher Scientific	Cat#MPK10096
Cell Nucleofector Kit 1	Lonza	Cat#VPH-5012
Red Blood Cell Lysis buffer	Roche	Cat#1181438900
Belzer-UW Cold Storage Solution	Thermo Fisher Scientific	Cat#NC0952695
Neural Tissue Dissociation Kit	Miltenyi	Cat#130-092-628

Highlights.

• Accurate prediction of cell identity within human kidney organoids

• Evidence of distal nephron-specific epithelial plasticity

• Ureteric epithelium induced from distal nephron can be cultured and expanded in vitro

• Modelling of autosomal recessive polycystic kidney disease