Mosaic analysis of cell rearrangements during ureteric bud branching in dissociated/reaggregated kidney cultures and in vivo

Abstract

Background

Cell rearrangements mediated by GDNF/Ret signaling underlie the formation of the ureteric bud (UB) tip domain during kidney development. Whether FGF signaling also influences these rearrangements is unknown. Chimeric embryos are a powerful tool for examining the genetic controls of cellular behaviors, but generating chimeras by traditional methods is expensive and laborious. Dissociated fetal kidney cells can reorganize to form complex structures including branching UB tubules, providing an easier method to generate renal chimeras.

Results

Cell behaviors in normal or chimeric kidney cultures were investigated using time-lapse imaging. In Spry1^−/− ↔ wild-type chimeras, cells lacking Spry1 (a negative regulator of Ret and FGF receptor signaling) preferentially occupied the UB tips, as previously observed in traditional chimeras, thus validating this experimental system. In Fgfr2^UB−/− ↔ wild-type chimeras, the wild-type cells preferentially occupied the tips. Independent evidence for a role of Fgfr2 in UB tip formation was obtained using “Mosaic mutant Analysis with Spatial and Temporal control of Recombination” (MASTR).

Conclusions

Dissociation and reaggregation of fetal kidney cells of different genotypes, with suitable fluorescent markers, provides an efficient way to analyze cell behaviors in chimeric cultures. FGF/Fgfr2 signaling promotes UB cell rearrangements that form the tip domain, similarly to GDNF/Ret signaling.

Introduction

The renal collecting duct system is formed by the extensive branching and growth of the ureteric bud (UB), an epithelial tube that emerges from the nephric duct (ND) at E10.5 in the developing mouse (Al-Awqati and Goldberg, 1998; Costantini, 2006; Short et al., 2014). The collecting duct system connects to the nephrons and conveys the filtrate from the nephron to the ureter and bladder. The formation, growth and branching of the UB are stimulated by several growth factors expressed by the surrounding metanephric mesenchyme cells, including GDNF, FGF10, and others (reviewed by Bates, 2011a; Song et al., 2011; Costantini, 2012; Davis et al., 2014). GDNF signals through the receptor tyrosine kinase Ret (Takahashi, 2001), which is expressed only by ureteric bud tip cells (Pachnis et al., 1993), while FGF10 in the kidney signals primarily through FGF receptor 2 (Fgfr2), which is expressed throughout the tips and trunks of the UB epithelium (Ohuchi et al., 2000; Zhao et al., 2004; Eswarakumar et al., 2005; Bates, 2011b). Knockout of Ret, Gdnf or the GDNF co-receptor Gfra1 result in a high frequency of renal agenesis, due to failure of the UB to emerge from the nephric duct (reviewed by Costantini and Shakya, 2006; Davis et al., 2014); in contrast, specific deletion of Fgfr2 in the UB epithelium (abbreviated Fgfr2^UB−/−) (Zhao et al., 2004), or global deletion of Fgf10 (Ohuchi et al., 2000), rarely cause renal agenesis but usually cause renal hypoplasia, due to reduced UB branching within the developing kidney. Gdnf and Fgf10 appear to have synergistic effects, as simultaneous deletion of Gdnf and Fgf10 leads to fully penetrant renal agenesis (Michos et al., 2010).

In studying the role of Ret signaling during ureteric bud formation, the use of chimeric embryos has proven to be a powerful tool for examining the cell-autonomous effects of genes in the Ret signaling pathway on nephric duct cell behaviors. Ret^−/− ↔ wild-type chimeras were generated, in which the mutant and wild-type ND and UB cells were labeled with different fluorescent proteins to permit them to be distinguished during live-imaging. These studies showed that wild-type nephric duct cells preferentially moved to the site where the UB was forming, thus contributing to the tip of the primary ureteric bud, while the Ret−/− cells failed to undergo these movements and were thus excluded from the primary bud tip (Shakya et al., 2005; Chi et al., 2009b). In Spry1^−/− ↔ wild-type chimeras, in contrast, the nephric duct cells lacking Spry1 (a negative regulator of signaling by Ret and other receptor tyrosine kinases, Basson et al., 2005) preferentially moved to form the primary ureteric bud tip, while the wild-type cells were largely excluded from this domain (Chi et al., 2009b). As Spry1 expression normally decreases Ret signaling, mutant Spry1^−/− cells have higher levels of signaling than wild-type cells. This study, as well as the examination of other chimeric combinations (Chi et al., 2009b; Kuure et al., 2010), led to a model in which the subset of nephric duct cells with the highest level of Ret signaling will preferentially give rise to the primary UB tip domain (Costantini, 2012). More recent studies, in which genetic mosaics for Ret, or for the downstream transcription factor Etv4 (Lu et al., 2009), were generated using Mosaic Analysis with Double Markers (MADM) (Zong et al., 2005) have shown that similar, Ret signaling-dependent cell movements also take place during ureteric bud branching within the developing kidney (Riccio et al., 2016).

However, generating chimeric embryos by traditional methods is expensive and laborious, requiring: (1) the generation of embryonic stem (ES) cell lines from embryos of the desired mutant genotypes; (2) micro-injection of the ES cells into (or aggregating them with) wild type pre-implantation embryos; and (3) surgical implantation of the manipulated embryos into pseudopregnant foster mothers. MADM uses genetic methods to generate mosaic embryos, and is thus technically simpler, but currently can be performed only for genes on four of the 20 mouse chromosomes (Zong et al., 2005; Hippenmeyer et al., 2010; Tasic et al., 2012; Hippenmeyer et al., 2013).

Here, we use two newer methods to generate chimeric or mosaic kidneys, and apply them to study the effects of Spry1 and Fgfr2 on cellular behaviors during ureteric bud branching. It was recently shown that when mouse fetal kidney cells are dissociated to single cells, and the cells are then allowed to reaggregate, they can self-organize to form complex renal structures containing branched ureteric bud tubules as well as nephrons (Lusis et al., 2010; Unbekandt and Davies, 2010). Among the many potential applications of this system (Ganeva et al., 2011; Xinaris et al., 2012) is the ability to easily generate chimeric reaggregates by mixing cells from dissociated kidneys of two different genotypes. A similar approach (but using siRNA-treated wild-type kidney cells, mixed with untreated wild-type kidney cells) was used to demonstrate a cell-autonomous role for the transcription factor Wt1 during nephrogenesis (Unbekandt and Davies, 2010). In this study, we first use the renal dissociation/reaggregation system, with kidneys expressing fluorescent proteins in different cell lineages, to investigate the cellular and morphogenetic events during the re-formation of the ureteric bud tubules from dissociated cells. We then test if this system can be used to study cellular events during UB branching, and specifically, the formation and maintenance of the tip domain, by analyzing Spry1^−/− ↔ wild-type dissociation/reaggregation chimeras and comparing them to the chimeras previously generated by traditional methods (Shakya et al., 2005; Chi et al., 2009b). After validating the use of this system to study tip domain formation, we apply it to address a new question: whether Fgfr2 signaling has an effect similar to Ret signaling on UB cell behaviors. Finally, to independently test the importance of Fgfr2 in UB tip formation, we apply a new method for the generation of mosaics by genetic means: Mosaic mutant Analysis with Spatial and Temporal control of Recombination (MASTR) (Lao et al., 2012). The combined results of these two methods reveal a novel role for Fgfr2 signaling during ureteric bud branching morphogenesis.

Results and Discussion

Early events in the formation of renal organoid structures from dissociated/reaggregated kidney cells

While the ability of dissociated/reaggregated kidney cells to form renal organoid structures with branched ureteric bud tubules joined to multiple nephrons has been demonstrated (Lusis et al., 2010; Unbekandt and Davies, 2010), the cellular events by which these structures re-form from dissociated kidney cells have not been analyzed in detail. We therefore used mouse fetal kidneys expressing fluorescent proteins in the ureteric bud or metanephric mesenchyme cells, together with time-lapse imaging, to visualize the reaggregation of kidney cells and the early morphogenesis of multicellular renal structures in such cultures.

To focus on ureteric bud reaggregation, we used kidneys carrying Hoxb7-GFP (Srinivas et al., 1999) or Hoxb7-mVenus (Chi et al., 2009a) transgenes that express fluorescent proteins specifically in UB cells. Kidneys from E12.5 fetuses were dissociated to single cells, reaggregated by centrifugation and cultured on Transwell filters, essentially as described (Unbekandt and Davies, 2010), while being photographed every 20–30 minutes (see Experimental Procedures for details). This time-lapse imaging revealed that, over the first 24 hours of culture, individual UB cells re-aggregated to form large multicellular structures. Many of the individual UB cells first extended and retracted long cellular processes that established contact with other UB cells, allowing multiple cells to join each other. Cells within these aggregates continued to send out similar protrusions, allowing the aggregates to merge with each other and thus to form larger structures (Fig. 1 and Movie 1). The protrusive activity of dissociated UB cells resembles cells at the caudal tip of the nephric duct during ductal elongation, a region which has not yet formed an epithelial tube. These protrusive cells are thought to be involved in the caudal migration of the duct tip (Chia et al., 2011; Weiss et al., 2014). This suggests that dissociated UB cells may transiently revert to an earlier developmental state in this culture system, leading to the reformation of epithelial tubules.

Fig. 1. Ureteric bud cell dynamics during the reaggregation of dissociated fetal kidney cells.

Hoxb7-GFP E12.5 kidneys were dissociated to single cells, which were then allowed to reaggregate while being monitored by time-lapse epifluorescence microscopy to allow ureteric bud cellular behaviors to be visualized (Movie 1 in Supplementary materials). Although mesenchymal cells were also present in the cultures, they were not labeled and therefore are not visible in fluorescence images. A–E show images between 1 and 28 hours of culture, while G–R show more frequent images of the region indicated by the white box in B. Many of the UB cells send out long, thin processes (some indicated by yellow arrows) that contact other UB cells, allowing the cells to reaggregate. In G–R, most of the UB cells in G (those circumscribed by yellow dotted line) re-join by 25 hrs to form a large aggregate (white arrow in R). Scale bar, 100 μm; G–R are 2x magnified.

To visualize the behavior of the metanephric mesenchyme cells during the early stages of reaggregation, we used kidneys in which the mesenchyme cell lineage was labeled with a red fluorescent protein (Six2^Cre/+; Rosa26^Tomato/+) and the UB cells with a green fluorescent protein (Hoxb7-mVenus) (Movie 2 and Fig. 2). Initially, Six2-lineage mesenchyme cells were widely dispersed (Fig. 2A), but soon after multicellular UB aggregates were formed, the mesenchyme cells began to migrate to surround the UB aggregates (Movie 2 and Fig. 2B–D), and by ~24 hours most of the Six2-lineage mesenchyme cells formed tight clusters around the UB structures (Fig. 2E). These clusters resemble the “caps” of nephrogenic mesenchyme that closely surround the UB tips during normal kidney development. The mechanism by which the cap cells closely associate with the UB tips is not known, and these results suggest that UB cells might secrete chemo-attractive signals that allow the mesenchyme cells to cluster around the UB tips. The dissociation/reaggregation system may be a useful tool to further investigate the mechanism of cap cell attraction by the UB.

Fig. 2. Condensation of Six2-lineage metanephric mesenchyme cells to form “caps” around ureteric bud cell aggregates.

A–D show selected frames (at the indicated times) from a 24-hour time lapse movie (see Movie 2) of a culture of dissociated/reaggregated cells from E12.5 kidneys. Ureteric bud cells are labeled green by Hoxb7-mVenus and Six2-lineage mesenchyme cells are labeled red by Six2^Cre-driven recombination of Rosa26^Tomato. The mesenchyme cells are at first broadly dispersed (1–6 hrs), but after the UB cells form aggregates, the mesenchyme cells gradually move to closely surround most of the UB aggregates. Scale bar, 100μm.

Branched ureteric bud tubules can be formed by two different processes: “pseudo-branching” and bona fide branching

It has been shown that highly branched ureteric bud tubules can form in dissociated/reaggregated kidneys (Unbekandt and Davies, 2010; Ganeva et al., 2011), but the temporal process of branching in this system has not been well-characterized. We therefore followed the time-course of branching using Hoxb7-GFP or Hoxb7-mVenus kidneys. The first branched structures appeared in many cultures within the first ~24 hours (Fig. 3, Movies 3 and 4). However, these did not form by the usual process that occurs during normal kidney development, in which the tips of a pre-existing UB tubule bifurcate to form two new branches. Instead, they formed during the initial reaggregation of the dissociated UB cells. While some of the initial UB cell aggregates were spheroid or oblong in shape, other aggregates immediately took the shape of branched tubules during their initial formation. We call this process “pseudo-branching” to distinguish it from the bona fide branching that occurs later on. The proportion of larger, pseudo-branched structures, compared to smaller spheroid or oblong UB aggregates, varied from culture to culture, for reasons that were not clear. In extreme cases (e.g., Fig. 3F–O, Movie 4), many of the UB cells in a culture formed one, or a few, large, highly branched aggregates, while in other cases the initial UB aggregates were mostly smaller and spheroid (e.g., Fig. 2).

Fig. 3. Branched ureteric bud tubules can be generated by “pseudo-branching”: the direct reaggregation of dissociated ureteric bud cells.

Hoxb7-GFP E12.5 kidneys were dissociated to single cells and allowed to reaggregate for the indicated times (hrs:min). Ureteric bud cells, and the structures formed by their aggregation can be seen due to Hoxb7-GFP fluorescence, while mesenchymal cells were not labeled. A–E shows individual frames from a time-lapse movie of one culture which formed both small, round aggregates and larger, tubular structures, some of them branched (arrowheads in E) (see Movie 3). F–O shows individual frames from another movie of a culture which formed a highly branched ureteric bud plexus (see Movie 4). Initial pseudo-branching is usually followed by bona fide branching (see Fig. 4). Scale bars, 100 μm.

At later stages of culture (typically after the first 24–48 hours), many of the dissociated/reaggregated UB structures began to branch and elongate in a manner that more closely resembled normal, in vivo branching morphogenesis (Fig. 4), as has been reported previously (Unbekandt and Davies, 2010). This bona fide branching is carried out both by large, pseudo-branched structures (Fig. 4A′–E′), and by smaller aggregates (Fig. 4A″–E″). Movie 5 and Fig. 4F–J show additional examples of bona fide branching, which were visualized by time-lapse microscopy.

Fig. 4. Time-lapse analysis of reaggregated renal cultures expressing ureteric bud-specific Hoxb7-mVenus reveals extensive, bona fide branching after ~ 2 days of culture.

A–E show a Hoxb7:mVenus reaggregated culture generated from E12.5 kidney cells, and photographed once per day. A′–E′ and A″– E″ are enlarged images of the regions marked by purple and orange dotted boxes in A–E. Many of the UB spheroids elongate from 1 day to 2 days (brackets in A and B). While the branched structure visible after one day (A, upper left) apparently arose by “pseudo-branching”, beginning on day two, bona fide branching of many of the UB tubules (purple dotted boxes) and UB spheroids (orange dotted boxes) was observed. The UB structure in A′–E′ undergoes a second round of branching between day 4 and day 6. F–J show selected time points from a time-lapse movie of a different culture, photographed every 30 min over 4 days, and also revealing bona fide branching (see Movie 5). Some of the branching tubules are marked by solid lines. Scale bars, 200 μm.

Reaggregated ureteric bud tubules re-establish a tip-specific gene expression domain characteristic of normal kidneys

A hallmark of in vivo kidney development is the presence of a specialized domain at the tips of the branching ureteric bud, which serves as a progenitor cell population for UB growth (Michael and Davies, 2004; Shakya et al., 2005; Short et al., 2014; Riccio et al., 2016). This tip domain expresses many genes important for normal UB growth and branching, including Ret (Pachnis et al., 1993) and Wnt11 (Majumdar et al., 2003), which are not expressed in the UB trunks. Since we wished to use the dissociation/reaggregation system to study genes involved in the formation of the UB tip domain, it was important to determine if the normal tip-specific gene expression patterns were regenerated in dissociated/reaggregated UB tubules.

For this purpose, we used two mouse alleles that express fluorescent proteins under the control of tip-specific genes: the knock-in reporter allele Ret^GFP (Jain et al., 2006) and the transgene Wnt11-TagRFP/Cre/ERT2 (McMahon, 2012). In kidneys that developed in vivo, these two reporters are expressed predominantly at the tips of the UB (Fig. 5A, C). The kidneys also carried a marker of the entire UB epithelium: in kidneys carrying Ret^GFP the entire UB was labeled “red” by the recombination of the Rosa26^Tomato reporter allele driven by Hoxb7-Cre, while in kidneys carrying Wnt11-TagRFP/Cre/ERT2 the entire UB was labeled “green” by Hoxb7-mVenus (Fig. 5A, C). After dissociation of E12.5 kidneys, reaggregation, and several days of culture, most of the long, branched UB tubules displayed tip-specific expression of Ret^GFP (Fig. 5B) or Wnt11-TagRFP/Cre/ERT2 (Fig. 5D). This revealed that the UB tip domain is reestablished in dissociated/reaggregated cultures, and therefore that this experimental system can be used to study cellular events involved in tip domain formation.

Fig. 5. Re-formed ureteric bud tubules in reaggregated renal cultures reestablish tip-specific domains of gene expression similar to those in fetal kidneys.

A, E12.5 kidneys (before cell dissociation) in which the entire ureteric bud epithelium is labeled by expression of the red fluorescent protein Tomato (Hoxb7-Cre;Rosa26^Tomato) while the ureteric bud tips are labeled green by Ret^GFP. B, a culture of reaggregated kidney cells after 4d of culture. Most of the tips are green, indicating expression of Ret^GFP, while most of the trunks lack Ret^GFP expression. C, E12.5 kidneys (before cell dissociation) in which the entire ureteric bud epithelium is labeled by Hoxb7-mVenus (green) while the ureteric bud tips are labeled by Wnt11^RFP-IRES-Cre (red). D, a culture of reaggregated kidney cells after 6d of culture. Wnt11^RFP-IRES-Cre (red) is expressed specifically at the tips of reaggregated ureteric buds. Scale bars, 100 μm.

The behavior of Spry1^−/− UB cells in chimeric Spry1^−/− ↔ wild-type reaggregated kidney cultures recapitulates their behavior in traditional in vivo chimeras

Because of the difficulty in generating chimeras by traditional methods, and their utility in studying cell behaviors during organogenesis, we wished to determine if ureteric bud cells in chimeric renal organoids, formed by dissociation/reaggregation, behaved similarly to their in vivo behaviors. While such chimeric cultures have been used to study the cell-autonomous role of the Wt1 gene in the nephron cell lineage (Unbekandt and Davies, 2010), they have not previously been used to study epithelial cell behaviors during UB branching morphogenesis. We chose Spry1 knockout kidneys (Basson et al., 2005; Chi et al., 2009b) to test the utility of this system, rather than Ret knockouts, since Ret^−/− embryos at E12.5 have either no kidneys or very small kidneys with a minimal number of ureteric bud cells (Schuchardt et al., 1994; Schuchardt et al., 1996).

Spry1^−/− E12.5 kidneys carrying a green UB marker (Hoxb7-mVenus) and wild-type (Spry1^+/+) kidneys carrying a red UB marker (Hoxb7-Cre; R26R^Tomato) were dissociated to single cells. Equal numbers of cells from the mutant and wild-type (WT) kidneys were mixed, reaggregated and cultured. As a control, WT kidneys carrying Hoxb7-mVenus were dissociated to single cells and reaggregated with an equal number of WT cells from dissociated Hoxb7-Cre; R26R^Tomato kidneys. Fig. 6A–C shows three representative examples of branching ureteric bud tubules that formed in the Spry1^−/− ↔ WT chimeric cultures, and Fig. 6D shows a representative example of a similar structure in a WT ↔ WT chimeric culture. In the Spry1^−/− ↔ WT chimeric UB tubules, after one day of culture, both the mutant (green) and WT (red) UB cells remained widely dispersed with no apparent pattern. However, after two or three days of culture, the green Spry1^−/− UB cells appeared to be preferentially localized to the tips (Fig. 6A–C, yellow arrowheads). In contrast, in the control WT ↔ WT chimeric ureteric buds, the green and red cells remained dispersed in an apparently random pattern throughout the UB tips and trunks, even after three days of culture.

Fig. 6. In chimeric Spry1^−/− ↔ wild-type reaggregated kidney cultures, Spry1^−/− ureteric bud cells undergo rearrangements to preferentially occupy the tips.

E12.5 kidneys with green (Hoxb7-mVenus) or red (Hoxb7-Cre; Rosa26^Tomato) ureteric buds were dissociated, the cells were reaggregated in a 1:1 ratio and cultured. In A–C and E the kidneys with green UB cells were Spry1^−/− and those with red UB cells were wild-type for Spry1 (WT); in D, all kidneys were WT for Spry1. A–C, three examples of branching UB tubules (indicated by gray outlines) in Spry1^−/− ↔ WT chimeras; after 1 day of culture, both the green Spry1^−/− and red wild-type UB cells are scattered throughout the UB tubules, but by days 2 and 3, the Spry1^−/− cells are enriched at the tips (arrowheads) and depleted in the trunks, while the WT cells show no such redistribution. D, in a branching UB tubule in a WT↔WT control chimeric culture, both red and green cells remain widely dispersed throughout the UB after 3 days. E, Time-lapse analysis of the rearrangement of Spry1^−/− and WT UB cells. A chimeric Spry1^−/− ↔ WT culture (generated with a 1:5 ratio of mutant:WT cells) was imaged every 30 minutes for the first ~2 days (see Movie 6), and then once per day. In the reaggregated UB structure shown (outlined in gray) green Spry1^−/− UB cells are widely dispersed for the first ~20 hours, but between 20 and 32 hours most mutant cells move to four clusters at the edges of the expanding UB tubule (arrowheads at 32 hrs), and remain at the tips of four new branches as they emerge (arrowheads at 6 days), while the central trunk region becomes enriched in red wild-type UB cells. Scale bars, 100 microns.

To quantify the distribution of Spry1^−/− vs. WT cells, we selected 30 random tips and 30 random trunk areas and measured the mean intensity of the green and red fluorescence as an estimate of the percentage of mutant vs. WT cells in each area (see Fig. 7A and Experimental Procedures). This analysis confirmed that Spry1^−/− cells were significantly enriched in the chimeric UB tips, which contained an average of 59% ± 17% mutant cells, vs. the trunks which contained an average of 45% ± 16% mutant cells (p=0.0023, tips vs. trunks, t-test). Fig. 7B shows a histogram of the number of tips or trunk areas that contained different percentages of mutant (green) cells. In contrast, in control chimeras (Fig. 7C), there was no significant difference between the percentage of green cells in the tips (average 50% ± 9%) vs. the trunks (average 54% ± 8%; p=0.12, tips vs. trunks, t-test).

Fig. 7. Contributions of Spry1^−/− or Fgfr2^−/− cells to ureteric bud tips and trunks in chimeric renal organoids.

A, examples of data analysis (see Experimental Procedures for details). For each chimera type (WT↔Spry1^−/−, WT↔WT or WT↔Fgfr2^UB−/−) after 3 days of culture, the red and green fluorescence was measured in randomly selected tip and trunk areas (30 each; e.g., yellow circles numbered 1 and 2 in photograph), as an estimate of the percentage of “green” vs. “red” cells. B–D, histograms of the number of tip or trunk areas displaying different percentages of green cells. For example, tip area #1 in A with 72% green cells is included in the “60–80%” category in B, while trunk area #2 in A with 44% green cells is included in the “40–60%” category. B, in WT↔Spry1^−/− chimeras (e.g., Fig. 6), tips (black bars) contained significantly higher percentages of green (Spry1^−/−) cells than trunks (gray bars). C, in control WT↔WT chimeras (e.g., Figs. 6 and 8), the percentage of green cells was not significantly different in tips vs. trunks. D, in WT↔Fgfr2^UB−/− chimeras (e.g., Fig. 8), the tips contained significantly lower percentages of green (Fgfr2^−/−) cells than trunks.

To better visualize the cellular events by which the Spry1^−/− cells became preferentially localized at the UB tips, we generated time-lapse movies of the early stages of a Spry1^−/− ↔ WT chimeric culture. Movie 6 shows the development of a branched chimeric tubule from a spheroid aggregate during the first ~43 hours of culture, and Fig. 6E shows selected frames from the movie, as well as still images taken once a day thereafter. After 8 hours of culture, the green Spry1^−/− cells remained dispersed throughout the aggregate, but by 32 hours most of the Spry1^−/− cells had moved outward to form four small clusters (yellow arrowheads), and on subsequent days each of these four clusters of mutant cells remained at the tip of a new branch as it extended from the central aggregate.

The behavior of Spry1^−/− cells in these dissociated/reaggregated kidney cultures is very similar to their behavior previously observed in traditional chimeras, in which Spry1^−/− cells in chimeric nephric ducts preferentially moved to the site of ureteric bud outgrowth and clustered at the tips of the primary ureteric buds (Chi et al., 2009b). Therefore, this analysis demonstrates that chimeras generated by dissociation/reaggregation are a useful alternative to in vivo chimeras for studying the genes that influence cell rearrangements during ureteric bud branching morphogenesis.

In Fgfr2^UB−/− ↔ wild-type chimeric reaggregated kidney cultures, the cells lacking Fgfr2 are deficient in their ability to contribute to the ureteric bud tips

FGF signaling through Fgfr2 is required for normal ureteric bud branching (Ohuchi et al., 2000; Zhao et al., 2004), and has been shown to cooperate with GDNF/Ret signaling to promote ureteric bud outgrowth and branching morphogenesis (Michos et al., 2010). Given the role of Ret signaling in nephric duct cell rearrangements that form the primary UB tip domain (Shakya et al., 2005; Chi et al., 2009b; Kuure et al., 2010), and in continued cell movements at the UB tips during renal branching (Riccio et al., 2016), we wished to test whether Fgfr2 signaling also played a role in these cell rearrangements. One reason for suggesting that the role of Fgfr2 might be different is that Fgfr2 is expressed throughout the UB epithelium (Cancilla et al., 2001; Zhao et al., 2004) (www.gudmap.org), unlike Ret, which is expressed only at the UB tips (Pachnis et al., 1993). We had previously attempted to use traditional chimeric embryos to address this question, by injecting Fgfr2^−/− embryonic stem cells (marked by Hoxb7-mVenus) into wild type host blastocysts. However, the vast majority of kidneys at E12.5 displayed no visible contribution from the Fgfr2^−/− ES cells (O. Michos and F.C., unpublished data). We suspected that the Fgfr2^−/− cells were out-competed by wild-type cells at an earlier stage of embryogenesis, in the mesodermal cell lineage leading to nephric duct and ureteric bud. Therefore, we used the renal dissociation/reaggregation system to examine the behavior of Fgfr2^−/− UB cells in chimeric cultures.

We first compared the development of non-chimeric, dissociation/reaggregation cultures from Fgfr2^UB−/− E12.5 kidneys with those generated from WT kidneys. This showed that in the mutant cultures there was typically much less elongation and branching of the reaggregated UBs (Fig. 8A, B), consistent with the ureteric bud defects observed in intact Fgfr2^UB−/− kidneys (Zhao et al., 2004). We then generated chimeras by dissociating Fgfr2^UB−/− and WT kidneys, each expressing a different fluorescent protein marker in UB cells, and reaggregating equal mixtures of the two cell populations. In the Fgfr2^UB−/− ↔ WT chimeric cultures (Fig. 8C), both mutant and WT ureteric bud cells were widely dispersed in the forming tubules after one day of culture. However, by three days, the red WT cells appeared to be preferentially clustered at the tips of the ureteric bud tubules (Fig. 8C, yellow arrowheads), and to be depleted in the trunks, which were composed largely of the green mutant cells. In contrast, in the WT ↔ WT control chimeras, both the red and green UB cells remained widely distributed throughout the UB reaggregates (Fig. 8D). The distribution of Fgfr2^−/− vs. WT ureteric bud cells was estimated in 30 randomly selected tips and 30 randomly selected trunk regions (using the same method illustrated for Spry1^−/− ↔ WT chimeras Fig. 7A). By this analysis, an average of 44% ± 20% of the UB tip cells were Fgfr2^−/−, while, in contrast, 58% ± 13% the UB trunk cells were Fgfr2^−/−, a significant difference (p=0.0024, tips vs. trunks, t-test). Fig. 7D shows a histogram of the number of tips or trunk areas that contained different percentages of Fgfr2^−/− mutant (green) cells, and illustrates the contrasting behavior of Spry1^−/− and Fgfr2^−/− cells in this assay.

Fig. 8. In chimeric Fgfr2^UB−/− ↔ WT reaggregated kidney cultures, WT ureteric bud cells undergo rearrangements to preferentially occupy the tips.

A–B, cultures of reaggregated cells from dissociated Fgfr2^UB−/− kidneys (A) or wild-type (WT) kidneys (B) after 4 days of culture; the mutant cultures exhibited reduced growth and branching. C–D, E12.5 kidneys with “green” or “red” ureteric buds were dissociated, and the cells were reaggregated in a 1:1 ratio and cultured. In C, the donor kidneys with green UB cells were Fgfr2^UB−/− (Hoxb7-mVenus/+; Hoxb7-CreGFP/+; Fgfr2^flox/null) and those with red UB cells (Hoxb7-Cre/+; Rosa26^{Tomato/Tomato}) were WT for Fgfr2. Arrowheads indicate tips where wild type (red) cells become enriched at 3 days. In D, both the donor kidneys with green UB cells (Hoxb7-mVenus/+) and those with red UB cells (Hoxb7-Cre/+; Rosa26^{Tomato/Tomato}) were WT for Fgfr2. Note the apparently random distribution of red and green cells. The ratios of red and green cells in tips vs. trunks are quantitated in Fig. 7. Scale bars, 100 μm.

Previous chimeric studies showed that nephric duct cells with higher levels of Ret signaling preferentially moved to form the primary UB tip domain, while cells with lower levels (such as Ret^−/− cells) failed to undergo such movements and were thus absent or underrepresented in the tip of the primary ureteric bud (Shakya et al., 2005; Chi et al., 2009b; Kuure et al., 2010). The current in vitro assay suggests that cells lacking Fgfr2, but wild type for Ret, also have a cell-autonomous deficiency (when competing with WT cells) in the ability to undergo cell movements that generate new UB tips during branching. This is consistent with the idea that Ret and Fgfr2 stimulate similar downstream signaling pathways in nephric duct cells and UB cells (Michos et al., 2010; Song et al., 2011). Interestingly, in branching epithelia of the developing fly, FGF signaling has a somewhat similar role in the selection of tip cells (Cabernard and Affolter, 2005; Ghabrial and Krasnow, 2006).

Mosaic mutant analysis with spatial and temporal control of recombination (MASTR) confirms a deficiency in the ability of Fgfr2^−/− cells to contribute to ureteric bud tips in vivo

As an independent test of the role of Fgfr2 in the contribution of ureteric bud cells to the tip domain, we used MASTR (Lao et al., 2012) to delete Fgfr2 in a small fraction of kidney cells, and then locate the mutant UB cells after several days of kidney development in vivo. To this end, we generated embryos containing the following alleles (Fig. 9A): Rosa26^FlpoERT2, expressing tamoxifen-inducible FlpoERT2 recombinase in all embryonic cells; Rosa26^MASTR, expressing a GFPcre fusion protein, but only in cells in which a transcriptional stop cassette flanked by frt (Flp target) sequences has been deleted by FlpoERT2; and either Fgfr2^flox/− (in “mutant” embryos) or Fgfr2^+/− (in control embryos). The activation of FlpoERT2 by binding to tamoxifen results in recombination of the Rosa26^MASTR allele in a small and random subset of cells, which then express GFPcre. The GFPcre makes it possible to visualize the cells in which recombination occurred, while the strong Cre activity of GFPcre will delete the Fgfr2^flox allele in the same cells (Lao et al., 2012), rendering them Fgfr2^−/−.

Fig. 9. MASTR (mosaic mutant analysis with spatial and temporal control of recombination) of the contributions of Fgfr2^−/− or Fgfr2^+/− cells to ureteric bud tips in Fgfr2^+/− kidneys in vivo.

A, Recombination of Fgfr2^flox allele to generate an Fgfr2⁻ (null) allele, and simultaneous GFP labeling of the recombinant cells, by MASTR. In mice carrying one Rosa26^FlpoER allele and one Rosa26^MASTR allele, tamoxifen injection activates FlpoER recombinase, leading to deletion of the frt-flanked “stop” cassette in Rosa26^MASTR, in a random subset of cells; this leads to expression of a GFPcre fusion protein, which labels the cells with GFP and causes deletion of the Fgfr2^flox allele. B, experimental protocol. Fgfr2^flox/− (experimental) and Fgfr2^+/− (control) embryos were exposed to tamoxifen at E10.5 (4 mg injected into the mother i.p.) and the kidneys were analyzed at E13.5. Kidneys were stained with anti-calbindin to label the entire UB epithelium (green), and with anti-GFP to label the recombinant cells (red). Stacks of optical sections at 3 μm intervals were generated by confocal microscopy. C–D, examples of optical sections used to count recombinant cells. In C–E, the recombinant cells were Fgfr2^+/−, while in F–H they were Fgfr2^−/−. The numbers of recombinant cells in all UB tips and trunks throughout each kidney (N=3 experimental and 3 control kidneys) were counted by examining every optical section (yellow dotted lines indicate the boundaries between tips and trunks). Red cells outside of the calbindin-labeled UB (i.e., mesenchymal cells) were not counted. I, the average percentage of recombinant cells in tips/(tips+trunks), in control vs. experimental kidneys. Error bars indicate S.D. Scale bar, 100 μm.

As diagrammed in Fig. 9B, pregnant females, whose embryos carried these alleles, were injected with tamoxifen at E10.5; this should cause recombination to occur between approximately E11.0 and E11.5, in a subset of cells throughout the forming ureteric bud (as well as in the surrounding mesenchymal and stromal cells). The kidneys were analyzed at E13.5, after the UB had undergone several rounds of branching. Both mutant and control mosaic kidneys appeared grossly normal, with similar numbers of UB branches, consistent with a low frequency of recombination of the Fgfr2^flox allele. Because the fluorescence of GFPcre was too weak to visualize clearly, we used whole-mount immunohistochemistry with anti-GFP to identify the cells in which GFPcre was expressed (and which are therefore presumed to have deleted the Fgfr2^flox allele) (Lao et al., 2012). Because recombination occurs in a fraction of all cell types, including the surrounding mesenchyme and stroma, as well as the UB, we could not visualize the GFP+ UB cells clearly in whole mount preparations. Therefore, we used confocal microscopy to examine every “optical section” throughout the kidneys, in which the entire UB was labeled with anti-Calbindin (Fig. 9B).

For each kidney (N=3 mutants and 3 controls), we counted all the GFP+ cells in the ureteric bud epithelium, and classified them according to their location in UB tips or trunks; examples of the optical sections used to score the location of GFP+ UB cells are shown in Fig. 9C–H. In control kidneys, where both the GFP+ and the GFP− cells were Fgfr2^+/−, 52% ± 4% of all GFP+ ureteric bud cells were found in the tips. In the experimental kidneys, where the GFP+ cells were Fgfr2^−/− and the GFP− cells were Fgfr2^flox/−, only 38% ± 5% of the GFP+ ureteric bud cells were found in the tips, a significant reduction (p=0.035, t-test) (Fig. 9I).

Thus, the MASTR experiments lead to a similar conclusion as the Fgfr2^UB−/− ↔ WT dissociation/reaggregation chimera studies: that Fgfr2^−/− UB cells have a cell-autonomous deficiency in their ability to contribute to and/or remain in the tips during early branching morphogenesis. The results of the two methods cannot be compared quantitatively, as the methods of scoring were different, by necessity. In the MASTR studies, where GFP+ recombinant cells were rare, we used confocal microscopy to score every GFP+ cell in the ureteric buds of experimental and control kidneys. In contrast, in the chimera studies, there were too many mutant cells to individually count, and the shapes and tip/trunk proportions of the UB structures were variable, and different than those in normal kidneys; therefore, we estimated and compared the proportion of mutant vs. wild-type cells in randomly selected tip and trunk regions. However, the qualitatively similar findings in the two experimental systems help to validate the use of dissociation/reaggregation cultures for studies of this type.

The deficiency in the contribution of Fgfr2^−/− UB cells to ureteric bud tips, in both experimental systems, was much less severe than previously observed in Ret^−/− ↔ WT traditional chimeric embryos, where very few if any Ret^−/− cells were observed in the UB tips (Shakya et al., 2005; Chi et al., 2009b). This may be due to a lesser role for Fgfr2 compared to Ret in the cell rearrangements that form the UB tip domain, or it may be due to differences in the experimental methods. Side-by-side comparisons of Ret and Fgfr2 mutant cells in the same experimental system would be needed to perform a direct comparison. Nevertheless, the combination of MASTR and dissociation/reaggregation chimeras provides clear evidence that Fgfr2 signaling plays a role in this critical developmental process.

MASTR is a powerful technique that can be used to generate genetic mosaics for any mouse gene for which a floxed allele is available (Lao et al., 2012). While the mutant cells could be identified in fixed tissues by staining with anti-GFP, an improved version of the MASTR allele, encoding a more strongly fluorescent protein, would be useful as it would permit time-lapse imaging of the behaviors of mutant cells in live kidney cultures.

Experimental Procedures

Animals

All experiments using mice were approved by the Columbia University Medical Center Animal Care and Use Committee. Mouse strains were on a 129X1/SvJ, 129S6/SvEvTac, C57BL/6J or mixed backgrounds, and included the following alleles:

• Fgfr2^flox (Mouse Genome Informatics: Fgfr2tm1Dor) (Yu et al., 2003)

• Fgfr2⁻ (Mouse Genome Informatics: Fgfr2tm1.1Dor) (Yu et al., 2003)

• Hoxb7-creGFP (Mouse Genome Informatics: Tg(Hoxb7-cre)5526Cmb) (Zhao et al., 2004)

• Hoxb7-cre (Mouse Genome Informatics: Tg(Hoxb7-cre)13Amc) (Yu et al., 2002)

• Hoxb7-GFP (Mouse Genome Informatics: Tg(Hoxb7-EGFP)33Cos) (Srinivas et al., 1999)

• Hoxb7-mVenus (Mouse Genome Informatics: Tg(Hoxb7-Venus*)17Cos) (Chi et al., 2009a)

• Ret^GFP (Mouse Genome Informatics: Rettm13.1Jmi) (Jain et al., 2006)

• Rosa26^Tomato (Mouse Genome Informatics: Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J) (Madisen et al., 2010)

• Rosa26^FlpoERT2 (Mouse Genome Informatics: Gt(ROSA)26Sortm3(CAG-FLPo/ERT2)Alj)

• Rosa26^MASTR (Mouse Genome Informatics: Gt(ROSA)26Sortm2(EGFP/cre)Alj)

• Six2-EGFP/cre (Mouse Genome Informatics: Tg(Six2-EGFP/cre)1Amc/J) (Kobayashi et al., 2008)

• Spry1⁻ (Mouse Genome Informatics: Spry1tm1.1Jdli) (Basson et al., 2005)

• Wnt11-TagRFP/cre/ERT2 (Mouse Genome Informatics: Tg(Wnt11-TagRFP/cre/ERT2)28Amc/J) (McMahon, 2012)

Dissociation of embryonic kidneys and culture of reaggregated kidney cell pellets

Dissociation and reaggregation of kidney cells were performed essentially as described (Unbekandt and Davies, 2010). Kidneys were obtained either from freshly-dissected E12.5 mouse embryos, or from embryos that had been stored for 1–3 days on ice while genotyping was performed, as described (Davies, 2006). Kidneys were dissected in CO₂-Independent medium (Life Technologies), then placed in 0.025% Trypsin/EDTA (Life Technologies) for 5 min at 37°C. Organs were stabilized in 200–500μl kidney culture medium (KCM, consisting of Dulbecco’s modified Eagle’s minimal essential medium, 10% fetal bovine serum, 1% penicillin/streptomycin) for 10 min at 37°C, then dissociated by trituration (using a standard plastic pipette tip attached to a Gilson 20 or 200μl PipetMan) and filtered through a 40μm cell strainer (BD Falcon, Oxford, UK) by gravity to remove undissociated cells. The cells were counted using a hemocytometer, one wild-type E12.5 kidney typically yielding about 40,000 cells.

For reaggregation cultures, 80,000–200,000 dissociated cells in KCM were centrifuged in 1.5ml Eppendorf tubes at 1600g for 8min. A Wiretrol glass pipette (Drummond Scientific) was used to carefully remove the pellet from the wall of the Eppendorf tube. The pellet was placed on a 0.4μm Transwell Clear filter (Costar) and cultured on the filter at the air–medium interface, in a humidified incubator at 37°C and 5% CO2. The initial culture medium was KCM with a ROCK inhibitor, either 10μmol/l Y27632 (Sigma), or 1.25 μmol/l Glycyl-H1152 dihydrochloride (Tocris, Bristol, UK); either of these inhibitors is required to prevent apoptosis of the kidney cells (Unbekandt and Davies, 2010). After 24 hrs, the well was washed 2x with PBS to remove the Y27632 or Glycyl-H1152 (as their continued presence interferes with nephrogenesis) and fresh KCM was added. The KCM was replaced every 2 days thereafter.

Time-lapse imaging

Kidney cell reaggregates were cultured in an incubation chamber at 37°C, 100% humidity, and 5% CO₂, and images were acquired every 20 min, 30 min or 1 hr for 2–4 days using a Zeiss AxioObserver Z1 microscope with Zen software.

Quantitative analysis of chimeric cultures

Three-day old chimeric cultures were used for analysis. Color images were converted to grayscale so that red and green areas were no longer discernable, and scored by a third party (unfamiliar with the identity of the cultures, and therefore unbiased), who catalogued the UB tips with a number. Thirty tips for analysis were selected using an online random integer sequence generator (https://www.random.org). Thirty UB trunk regions were also randomly selected by the same method. The same individual performed this selection for the Spry1, Fgfr2 and control chimeric cultures, to achieve consistency in the analysis. For each of the tip and trunk regions, now using the original color images, the mean intensity of the red and green channels within a 40-micron wide circle was measured using ImageJ (Schneider et al., 2012), and used as an estimate of the percentage of green vs. red cells in that region, as illustrated in Fig. 7A. This measurement was performed in the same way for the Spry1↔WT, Fgfr2↔WT and WT↔WT chimeric cultures.

Whole-mount Immunofluorescence staining for analysis of MASTR kidneys

E13.5 kidneys were fixed in 4% PFA in PBS overnight at 4°C, then washed in PBS (3 × 10 min). Samples were then incubated overnight at 4°C in 10% NDS in TSP before incubation in the following primary antibodies: rabbit anti-GFP (Invitrogen, 1:400) and goat anti-calbindin D28K (Santa Cruz, 1:200). Primary antibody was diluted in TSP and 10% NDS and incubation was for 4 days at 4°C followed by TSP washes (3 × 40 min, room temp). Secondary antibodies (Cy3 donkey anti-rabbit, Cy2 donkey anti-goat, respectively; Jackson ImmunoResearch) were diluted in TSP and 10% NDS and samples were incubated for 3 days at 4°C followed by TSP washes (3 × 40 min, room temp).

Confocal Microscopy and Statistics

After staining with anti-GFP and anti-calbindin, kidneys were methanol-dehydrated and cleared with BABB (Benzyl alcohol: Benzyl benzoate; 1:1, Fisher Scientific). Optical sections at 3-micron intervals through the whole kidneys were obtained by confocal microscopy using a Nikon Ti Eclipse inverted microscope. All recombinant (GFP+) cells within the ureteric trees were then counted, and the percentages of recombinant cells in the tips vs. the trunks were then calculated for experimental (Rosa26^{FlpoERT2/MASTR};Fgfr2^flox/−) and control (Rosa26^{FlpoERT2/MASTR};Fgfr2^+/−) samples. The averaged percentages for three kidneys of each genotype were compared with Student’s t-test.

Supplementary Material

Supp Movie S1. Movie 1.

This movie corresponds to Figure 1. Ureteric bud cell dynamics during the reaggregation of dissociated fetal kidney cells.

Supp Movie S2. Movie 2.

This movie corresponds to Figure 2A–E. Condensation of Six2-lineage metanephric mesenchyme cells to form “caps” around ureteric bud cell aggregates.

Supp Movie S3. Movie 3.

This movie corresponds to Figure 3A–E. Branched ureteric bud tubules can be generated by “pseudo-branching”: the direct reaggregation of dissociated ureteric bud cells.

Supp Movie S4. Movie 4.

This movie corresponds to Figure 3F–O. Branched ureteric bud tubules can be generated by “pseudo-branching”: the direct reaggregation of dissociated ureteric bud cells.

Supp Movie S5. Movie 5.

This movie corresponds to Figure 4F–J. Time-lapse analysis of reaggregated renal cultures expressing ureteric bud-specific Hoxb7-mVenus reveals extensive, bona fide branching after ~ 2 days of culture.

Supp Movie S6. Movie 6.

This movie corresponds to Figure 6E. In chimeric Spry1^−/− ↔ wild-type reaggregated kidney cultures, Spry1^−/− ureteric bud cells undergo rearrangements to preferentially occupy the tips. The images on the left are the merged red (WT)/green (Spry1^−/−) channels, and the images on the right are the green (Spry1^−/−) channel alone.

Bullet points.

• Branched ureteric bud tubules re-form in a two-step process: pseudo-branching, followed by bona fide branching.

• Ureteric buds reestablish tip-specific domains of Ret and Wnt11 expression.

• In Spry1^−/− ↔ wild-type chimeric cultures, Spry1^−/− cells preferentially occupy the ureteric bud tips.

• In Fgfr2^UB−/− ↔ wild-type chimeras, the wild-type cells preferentially occupy the tips.

• Mosaic mutant Analysis with Spatial and Temporal control of Recombination (MASTR) further supports a role for Fgfr2 in ureteric bud cell rearrangements to form the tip domain.