Simple 3D culture of dissociated kidney mesenchyme mimics nephron progenitor niche and facilitates nephrogenesis Wnt-independently

Kidney mesenchyme (KM) and nephron progenitors (NPs) depend on WNT activity, and their culture in vitro requires extensive repertoire of recombinant proteins and chemicals. Here we established a robust, simple culture of mouse KM using a combination of 3D Matrigel and growth media supplemented with Fibroblast Growth Factor 2 (FGF2) and Src inhibitor PP2. This allows dissociated KM to spontaneously self-organize into spheres. To reassess the requirement of WNT activity in KM self-organization and NPs maintenance, cells were cultured with short pulse of high-dose GSK3β inhibitor BIO, on a constant low-dose or without BIO. Robust proliferation at 48 hours and differentiation at 1 week were observed in cultures with high BIO pulse. Importantly, dissociated KM cultured without BIO, similarly to that exposed to constant low dose of BIO, maintained NPs up to one week and spontaneously differentiated into nephron tubules at 3 weeks of culture. Our results show that KM is maintained and induced to differentiate in a simple culture system. They also imply that GSK3β/WNT-independent pathways contribute to the maintenance and induction of mouse KM. The robust and easy 3D culture enables further characterization of NPs, and may facilitate disease modeling when applied to human cells.

Organotypic cultures, such as the transfilter system and explants of the whole kidney anlagen, have profoundly deepened our knowledge of the developmental mechanisms of kidney development 10 . However, it has been shown that when the KM is cultured in isolation, it stops proliferating, does not differentiate and decays in a few days 10,11 . First attempts towards the long-term KM in vitro culture were undertaken by dissociating and reaggregating whole embryonic kidneys where UB cells were present 12 . A variety of factors required for the maintenance, proliferation, and differentiation of the KM has been discovered [13][14][15][16] . Recently, the maintenance and propagation of purified nephrogenic progenitor cells was achieved 17,18 . Although efficient in maintaining the purified SIX2 + NPs, these cultures exclude stromal niche cells. In addition, the maintenance protocols are tedious, depend on complex procedures and require multiple synthetic agents and growth factors.
The majority of renal cell cultures rely on 2D monolayers [19][20][21][22] that do not accurately model the 3D architecture of the tissue. The 3D architecture and cell-cell contacts are essential for propagation SIX2 + NPs in vitro both in isolation and when pluripotent stem cells are differentiated into kidney organoids 17,23 .
Until recently, the induction of an isolated kidney mesenchyme for differentiation was based on its recombination with the UB or heterologous inducing tissues such as embryonic spinal cord 21,24,25 . The GSK3β inhibitor 6-bromoindirubin-3-oxime (BIO) has been shown to induce differentiation of isolated rat and mouse mesenchyme via the canonical WNT signaling pathway 9 . Chemical induction with GSK3β inhibitors has been utilized not only for kidney explants but also for generating organoids from hiPSC that show nephron marker patterns typical for that of nephrogenesis in vivo 26 .
Here we describe a robust, simple culture of mouse KM using a combination of 3D Matrigel and growth media supplemented with FGF2 and PP2. We show that this method facilitates dissociated KM to self-organize into spheres where also renal stromal cells are present. We further demonstrate that NPs and stromal niche propagate for a week, and thereafter NPs undergo spontaneous MET to form nephron tubules without additional WNT activation. Our results imply that GSK3β/WNT-independent pathways complement regulation of self-renewal and differentiation decisions in mouse KM.

Results
Self-organization and formation of metanephric niche. To characterize NPs self-organization in vitro, KM from E11.5 mouse embryos were microsurgically dissected from the UB and dissociated into single cells ( Fig. 1A-C). Previous studies have shown that FGF family proteins are essential for the KM maintenance [27][28][29] . Therefore, basic culture medium (BM) was supplemented with 50 ng/ml of FGF2. We then plated 30,000 dissociated KM cells on top of 3D Matrigel per well of 96-plate in 200 µl either BM or BM + FGF2 (Fig. 1D, E respectively). Time-lapse imaging revealed that single cell suspensions self-organized into spheroids within 17 hours of culture (Supplemental video 1). To reassess the role of WNT in the self-organization and propagation of the NPs, the dissociated KM cell culture was pulsed with 10 µM of BIO for 1 hour (BIO pulse), thereafter the medium was replaced with fresh BM + FGF2 without BIO (Fig. 1F, Supplemental video 2). Bright field imaging after 72 hours of culture showed that dissociated KM cultured in BM-only did not form spheres and decayed with time ( Fig. 2A), whereas in the BM + FGF2 and BM + FGF2 + BIO conditions dissociated KM self-organized into cell clusters comparably (Fig. 2B,C).
To better understand the composition and self-organization of KM spheres we examined the cell identities by immunofluorescence at 72 hours of culture and compared these to E11.5 whole kidney reaggregation cultured in classical Trowell-type culture 12 . Cross-sections of the KM spheres were immunostained with different cell type markers to identify whether both stromal and nephrogenic lineages are present. This revealed that the spheres were composed of MEIS1/2 + stroma, SIX2 + and PAX2 + NPs progenitors ( Fig. 3A-F). The BM + FGF2 cultured KM spheres comprised of MEIS1/2 labeled cells (Fig. 3A, C) neighboring SIX2 + progenitors (Fig. 3B,C), whereas in spheres cultured in BM + FGF2 + BIO pulse PAX2 + NPs formed pretubular aggregate-like condensates, which were surrounded by MEIS1/2 positive stroma cells. Quantification of immunofluorescence images demonstrated, that KM spheres cultured in BM + FGF2 for 72 hours contained 48.56% ± 6.34% of SIX2 + cells, whereas BIO pulse significantly reduced the SIX2 + progenitors (4.31% ± 2.13%, p < 0.05, n = 6, Supplemental Fig. 1). Analysis of niche self-organization in the whole kidney dissociation and reaggregation culture revealed that E11.5 reaggregated kidneys were decreased in size at 72 hours (Supplemental Fig. 2) and were negative for NP marker SIX2 and nephron epithelial markers CDH1 and lotus tetragonolobus lectin (LTL) (Supplemental Fig. 3E-H). These immunostainings revealed that in our dissociated KM culture method, stroma and NPs spontaneously self-organize into stroma-NP niche with a capacity to recapitulate structures such as pretubular aggregates.

FGF2 promotes maintenance of kidney mesenchyme spheres while BIO induces their expansion.
To assess the survival of KM spheres in different conditions we re-dissociated and incubated the spheres with Sytox, a live cell impermeant agent allowing analysis of live cell proportions by flow cytometry (Fig. 2D-G). This revealed, in accordance with previous findings, that dissociated KM cultured in BM + FGF2 showed almost three times higher cell survival than those cultured in BM only (57.4% ± 20.9% and 20.1% ± 8.8% respectively, n = 4, Fig. 2D, E). A four-fold increase in cell survival was observed when dissociated KM cells were cultured with BM + FGF2 + BIO pulse (79.8% ± 14.4%, n = 4, Fig. 2B).
We then characterized proliferation in KM spheres for up to three weeks by EdU incorporation assay and Click-it fluorescent labeling of cells in S-phase of cell cycle (Supplemental Fig. 4). At 48 hours of culture, proliferating cell ratio of the KM spheres cultured in BM + FGF2 was 0.20 ± 0.05 while the same ratio was 0.38 ± 0.06 in spheres supplemented with BIO pulse (Supplemental Fig. 4A, B). This difference in proliferating cell ratios was abolished after one week of culture and remained similar also at three weeks of culture (Supplemental Fig. 4B). The deceleration of proliferation in KM spheres cultured with BIO pulse coincided with a rapid switch from progenitor to differentiation state in response to strong BIO pulse, as evident from immunofluorescence staining of one week cultured KM spheres (Supplemental Fig. 5).

Kidney mesenchyme spheres WNT-independently undergo tubulogenesis. Several studies
have reported that constant basic WNT activity, induced by continual presence of low dose of BIO or CHIR (another potent WNT-activator) facilitates NPs maintenance 17,30 . Since strong BIO pulse failed to maintain NP state and proliferation in long-term, we set to investigate the effect of low, constant BIO-dose on our 3D dissociated KM cultures. For this, the BIO concentration was titrated to identify appropriate concentration. After one week of culture, 50 nM BIO concentration maintained SIX2 + progenitors without much of the positivity for HPA (Supplemental Fig. 6) indicating maintenance of the progenitor status. No differences were observed by bright field microscopy between dissociated KM cultured in BM + FGF2 and BM + FGF2 + constant BIO in self-organization or KM sphere size (see Fig. 2H, one week time point). This concentration of BIO was chosen for following studies addressing progenitor status (SIX2 + ) and tubulogenesis as a readout of differentiation (lectin Helix pomatia agglutinin (HPA), a basolateral marker of nephron epithelium).
We next analyzed the long-term maintenance of SIX2 + NPs in KM spheres cultured in BM + FGF2 and BM + FGF2 + constant BIO. Whole mount immunofluorescence staining showed that at one week of culture, SIX2 + NPs were maintained in both BM + FGF2 and BM + FGF2 + constant BIO cultured KM spheres (Fig. 4A, left panel). Quantification of the whole mount immunofluorescence images, using custom image analysis pipelines on the modular workflow system ANIMA 31 , showed no significant difference in ratios of SIX2 + and Pax2 + -to-total cell counts between BM + FGF2 and BM + FGF2 + constant BIO (SIX2 + 15.32% ± 2.52%, 18.70% ± 3.78% n = 3, p = 0.25, and PAX2 + 10.32% ± 4.81%, 16.68% ± 6.28%, n = 3, p = 0.26, respectively, Fig. 4A'). Importantly, these data demonstrate a significantly longer maintenance of cultured NP cells than previously reported in such a simple culture system 22 . To assess the nephrogenic potential of NP cells cultured for one week in BM + FGF2 and BM + FGF2 + constant BIO we next exposed them to WNT activation, which revealed equal epithelization capacity comparable to that seen with naïve, freshly isolated MM (Supplemental Fig. 8). To test the long-term survival of NP cells in KM spheres we next analyzed SIX2 expression in spheres cultured for three weeks. Very few clusters of SIX2 + NP cells were detected in spheres cultured for three weeks regardless whether they were supplemented with FGF2 alone or together with constant BIO (4.46% ± 2.56% and 4.80% ± 2.56%, respectively, n = 3, p = 0.93, Fig. 4B, B' , and Supplemental Fig. 10). However PAX2 + -to-total cell counts between BM + FGF2 and BM + FGF2 + constant BIO were much higher and with later culture condition causing significantly higher PAX2 expression (31.25 ± 15.33%, and 48.73% ± 17.50% respectively, n = 4, p < 0.05). At this stage, the sphere cultures appeared also not suitable for recombination experiments, which failed due to poor quality of the cells. Of note, epithelial nephron tubules were detected in KM spheres cultured in both conditions for three weeks indicating loss of NP cells by spontaneous differentiation (Supplementary Figs 9, 10).
We finally characterized nephrogenesis in the KM spheres cultured for one and three weeks. Staining with PAX2, a protein upregulated in nephron precursors undergoing epithelialization 32  www.nature.com/scientificreports www.nature.com/scientificreports/ aggregates, reminiscent of pretubular aggregates and lumen containing nephron tubules in spheres cultured with BM + FGF2 for one week (Fig. 4A, right panel). In BM + FGF2 + constant BIO conditions PAX2 + cells were predominantly found in lumen forming tubules (Fig. 4A, right panel arrowheads). This suggests that the MET www.nature.com/scientificreports www.nature.com/scientificreports/ was initiated, and to some extent also completed at one week of KM spheres culture. At three weeks, however, we observed a robust PAX2 signal with varying intensities in KM spheres cultured in BM + FGF2 (Fig. 4B, left  panel). To ascertain the presence of the tubules, we performed 3D reconstruction of PAX2 immunofluorescence within KM spheres cultured for one and three weeks. At one week of culture, we again confirmed the presence of PAX2 + in the early cell aggregate of spheres cultured in BM + FGF2 and larger aggregates in spheres cultured in BM + FGF2 + constant BIO (Fig. 4C, upper panel). At three weeks a network of PAX2 labelled complex tubular structures were observed in both BM + FGF2 and BM + FGF2 + constant BIO cultured KM spheres (Fig. 4C). We have further confirmed that these tubules are CDH1 positive (Supplemental Fig. 10), however lack of segment specific markers such as LTL in CHD1 + tubules suggests that the nephron epithelium is fairly immature. Taken together, the panel of immunofluorescence staining and their 3D reconstructions indicate that KM spheres spontaneously differentiate into tubular epithelium even without activation of WNT signaling.

Discussion
The organization of an in vivo NP niche involves presence of all three tissue layers contributing to normal kidney organogenesis, namely KM itself, stroma and UB epithelium. In vitro, the self-organization and propagation of kidney nephron progenitors require activation of a complex mixture of different cascades aiming to mimic signals normally secreted by UB and stroma 17,30 . Simple systems, which allow manipulation of the cells, have utilized dissociation and re-aggregation together with the UB-derived inductive signals such as GSK3 inhibition, UB itself or spinal cord expressing several Wnts 1,12,20,22,33 . Until recently, in vitro expansion of nephron progenitors was not possible and current protocols now in use are not only expensive but also tedious 17,34 .
Various in vivo and in vitro studies have indicated that FGF signaling is indispensable for survival and maintenance of metanephric mesenchyme 18,22,[27][28][29] . Our 3D cell culture model confirms this. Contrary to previous studies, here we present a simple 3D culture method to aggregate and propagate dissociated KM into spheres resembling NP niche and maintaining nephrogenic potential up to one week. This new method utilizes the 3D Matrigel in combination with FGF2 and PP2 to generate conditions allowing self-organization into nephron progenitors and stromal cells where both lineages interact with each other and support their maintenance. As  www.nature.com/scientificreports www.nature.com/scientificreports/ indicated by previous studies, a clear lineage separation of the nephron and stromal progenitor is established early in renal differentiation [35][36][37] . Stromal compartment in cooperation with ureteric epithelium is also shown to be crucial for NP self-renewal and differentiation [38][39][40] . In the light of a growing interest towards generating kidney organoids from human cells 41,42 , our culture method presents more simple approach by providing an opportunity to investigate the cell aggregation 43 , self-organization, cross-talk between stroma and NPs in the conditions where signals from ureteric epithelium or other inductive tissues are not needed.
Likely, the most interesting finding with our newly developed dissociated KM culture is the observation that NPs within KM spheroids, cultured without exogenous WNT-activation, undergo spontaneous epithelization. This exciting finding underlines the robustness of the nephron induction and tubulogenesis programs, implying that similarly to rat kidneys 44 , signaling pathways other than just GSK3β/WNT-activation supplement regulation of self-renewal and differentiation decisions in mouse KM. Indeed, Schmidt-Ott and colleagues previously identified conserved UB tip expressed gene signature in mouse and rat kidneys, and demonstrated that e.g. secreted cytokine receptor, cytokine receptor-like factor 1 (CLF-1) in complex with its ligand induces tubulogenesis in isolated mouse MM 45 . Moreover, several studies also support the involvement of FGF induced signaling in differentiation of nephrons 43,46,47 . Thus, the robust and easy 3D culture method presented here provides a tool to study niche-NPs interaction and can be utilized to generate new knowledge about NPs characteristics and nephron differentiation.

Material and Methods
Tissue culture. Mouse embryonic kidneys were microdissected from E11.5 NMRI embryos as described 48 .
The kidneys were dissected in PBS, and subsequently treated with collagenase type IV for 15 min at 37 °C. Thereafter, the kidney mesenchyme (KM) was separated from the ureteric epithelium (UB) by microdissection in the DMEM supplemented with 10% fetal calf serum. The KM was incubated with TrypLE Express and dispersed into single cells in Basic Media (BM) (DMEM/F12, 10% fetal calf serum (FCS), and 10 µM PP2 (to reduce neuronal outgrowths and augment sphere aggregation) or BM with 50 ng/ml FGF2 9  Embryonic kidney dissociation and reaggregation. Embryonic kidneys were dissociated and reaggregated as described in (Unbekandt and Davies, 2010) 12 with following modifications: E11.5 embryonic kidneys were isolated as described 48 , the isolated kidneys were dissociated into single cells by pipetting after incubation of 5 min in TrypLE at 37 °C. The digestion was blocked by washing in 10 parts of DMEM supplemented in 10% FCS. Cell were counted and at least 80,000 cells were used for reaggregation by centrifugation at 500 × g for 3 min. Pellets were transferred on nucleopore carbon filter and cultured on Trowell-type system at air liquid interface in DMEM supplemented with 10% FCS.
Tissue recombination. After five days of culture, the KM spheres were recovered from Matrigel and transferred on top of a piece of E11.5 mouse spinal cord ( Supplementary Fig. 7). The recombinant tissue was cultured in Trowell type tissue culture for 72 hours, followed by fixation with 4% PFA for 1 hour at room temperature and subjected to immunofluorescence staining.
Immunofluorescence. The KM spheres were fixed for 20 min at RT with 4% PFA (unless otherwise stated), washed with PBS, pelleted in 1.5 ml low protein binding tubes, embedded in Richard-Allan Scientific HistoGel (for paraffin sections) or OCT embedding compound (for cryosections), and subsequently cut at 5 µm section in microtome or cryostat. The sections were blocked with secondary antibody host specific 10% serum for 1 hour at RT. Samples were incubated with primary antibodies overnight at +4 °C, followed washes in PBS and 1 hour incubation with secondary antibodies. The samples were then counterstained with 10 µg/ml of Hoechst or DAPI, and mounted for imaging.
Whole mount immunofluorescence. The KM spheres were fixed for 30 min at RT with 4% PFA, washed with PBS, pelleted in 1.5 ml low protein binding tubes and pellets were transferred on microscopic slide. The pellets were permeabilized for 20 min with PBS Triton x-100 0.5%, blocked for 1 hour with secondary antibody host specific 10% serum at RT. Samples were rinsed with PBS and incubated with primary antibodies at +4 °C o.n., followed by wash and incubation with secondary antibody o.n. at +4 °C. After wash samples were counterstained with DAPI mounted and imaged. www.nature.com/scientificreports www.nature.com/scientificreports/ Proliferation assay. At the indicated time points the proliferating cells were labeled with 10 µM of 5-ethynyl-2′-deoxyuridine (EdU) for 2 hours and detected by Click-IT EdU Imaging Kit (AlexaFluor-594 conjugated, Life Technologies) according to the standard protocol. Subsequently, the samples were co-immunostained as described above.
Flow cytometric analysis of cell viability. After 72 hours in culture the KM spheres were harvested from Matrigel, transferred to low protein binding tubes, and collected by centrifugation and subsequently dissociated into single cells by TrypLE Express. Cells were stained with Sytox Red Dead Cell Stain (ThermoScientific) according manufacturer's instruction, and analyzed with Accuri Flow Cytometer (BD Biosciences) at the HiLIFE Biomedicum Flow Cytometry core facility.
Imaging and image analysis. The samples were imaged by Zeiss AxioImager Z1 or Leica SP8 confocal microscope. Bright field images were taken by Nikon Eclipse Ti-E. The proliferation and protein expression ratios (labeled cells to-total cell count) were quantified using custom developed pipelines on the modular workflow system ANIMA 31 . In short, images were pre-processed, segmented for cell nuclei by Graphcut algorithm using Farsight library 49 , and mapped for mask, which was used to measure intensity of label channel. The intensity values and their standard deviations were clustered by k-mean algorithm to determine negative staining threshold and signal intensity classes. KM spheres 3D structure was reconstructed using Imaris software (Bitplane) from convoluted (Huygens Professional software, SVI) immunofluorescence and DAPI counterstained-stained confocal microscopy images. Imaging, image deconvolution and 3D rendering has been performed at HiLIFE Biomedicum imaging core facility.

Data analysis and statistical methods. For all figures the independent experiments (different mouse litters
with no less than 8 embryos) were performed at least three times. The proliferation ratio was calculated from at least three independent experiments, each of which contained pooled KM sample from at least eight embryos. To determine the proliferation ratio and cell survival percentage no less than 10,000 cell counts were used. For SIX2 and PAX2-to-total cell count positive cell quantification, images form at least 3 independent experiments were used and calculated as described in image analysis section above. The independent two-sample t-test (significance p < 0.05) analysis was performed on the means of the ratios (calculated with OriginPro statistical analysis software).
Ethical approval. All methods were carried out in accordance with relevant guidelines and regulations. All animal husbandry and experimental procedures were approved by EU legislation and Finnish Animal Care and Use Committee.
Significance statement. The kidney mesenchyme (KM), also known as metanephric mesenchyme, contains both nephron progenitors (NPs) and stromal cells. In vitro maintenance and propagation of dissociated KM and uninduced NPs is not trivial, and current culture protocols rely on complex procedures. This study demonstrates that a combination of three dimensional (3D) cell culture and Fibroblast Growth Factor 2 (FGF2) expands NPs that self-organize into spheres resembling the niche. They also form epithelial tubes without inductive tissues or exogenous source for WNT activation. This robust and easy 3D culture method provides a tool to study niche-NPs interaction and can be utilized to generate new knowledge about NPs characteristics.

Data Availability
The datasets and quantification pipelines generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.