Postnatal prolongation of mammalian nephrogenesis by excess fetal GDNF

The mammalian kidney is a non-regenerating blood filtrating organ with essential detoxification and electrolyte balancing functions. The nephrons executing these functions are born during the fetal period, as the adult kidney shows only limited repair capacity or compensatory glomerular enlargement (hypertrophy) (Chawla and Kimmel, 2012; Rinkevich et al., 2014; Romagnani et al., 2013). Congenital low nephron mass is a widely accepted risk factor for hypertension, proteinuria, glomerulosclerosis and end-stage renal disease (Bertram et al., 2011; Hoy et al., 2008) and thus greatly influences renal health throughout life.

The nephrogenic program ends between gestation weeks 35-37 in humans and postnatal day (P) 3 in mouse (Hartman et al., 2007; Hinchliffe et al., 1991; Ryan et al., 2018), but the mechanisms contributing to the cessation are poorly understood. It has been postulated that stage-specific changes in growth patterns (Cebrián et al., 2004) and cell intrinsic age-sensing mechanisms (Chen et al., 2015a) are involved. Molecularly, bone morphogenetic protein (BMP)-induced SMAD signaling (Brown et al., 2015), hamartin (Tsc1) (Volovelsky et al., 2018) and microRNA regulation (Yermalovich et al., 2019) participate in determination of the nephrogenic program duration. However, plasticity of the nephrogenic program has not been fully explored.

Nephrons differentiate from a nephron progenitor (NP) pool, also known as metanephric or cap mesenchyme. NPs expressing a unique repertoire of markers including SIX2, glial cell line derived neurotrophic factor (GDNF) and CITED1, are present only in the embryonic kidney where they surround each ureteric bud (UB) tip from its establishment (Boyle et al., 2008; Self et al., 2006). The nephrogenesis process is synchronized with the UB branching, which simultaneously maintains self-renewal in the undifferentiated NP population and induces a subset of NPs to undergo gradual mesenchyme-to-epithelium transition via well-defined precursor stages (pretubular aggregates, PA; renal vesicles, RV; comma-shaped bodies, CSB; S-shaped bodies, SSB), which eventually differentiate into functional nephrons (Kobayashi et al., 2008; Lindström et al., 2018).

The balance in self-renewal versus differentiation within the NP population depends on signals mediating the reciprocal inductive tissue interactions taking place between the UB and the neighboring mesenchyme. Metanephric mesenchyme and NP-derived signals, such as GDNF and fibroblast growth factors (FGFs), regulate UB morphogenesis, through which the embryonic kidney grows in size, achieves its shape and establishes the collecting duct system (Costantini and Kopan, 2010). GDNF-activated RET signaling in the UB tips regulates a transcriptional profile responsible for the control of collecting duct progenitors and UB morphogenesis (Kurtzeborn et al., 2018; Li et al., 2019; Lu et al., 2009; Ola et al., 2011; Riccio et al., 2016). GDNF-regulated transcripts also comprise secreted proteins, such as WNT11 and CRLF1, with demonstrated potential to regulate the nephrogenic program (O'Brien et al., 2018; Schmidt-Ott et al., 2005). Other UB-derived nephrogenesis regulators are WNT9b and FGF-ligands influencing NP maintenance and differentiation (Barak et al., 2012; Carroll et al., 2005; Kiefer et al., 2012).

We have previously reported that genetic disruption of the Gdnf 3′ untranslated region causes a 3- to 6-fold increased expression of endogenous GDNF and results in renal hypoplasia due to the UB branching defect (Kumar et al., 2015). Unlike newborn lethality in the Gdnf knockouts, the Gdnf^hyper/hyper mice survive up to 3 weeks and have revealed the essential role of GDNF in collecting duct progenitor self-renewal (Costantini, 2012; Kumar et al., 2015; Li et al., 2019). Here, we demonstrate that excess GDNF, despite its negative effect on overall kidney growth, can expand NP lifespan, thus further strengthening the newly recognized plasticity of mammalian kidneys.

The spatial arrangement of SIX2-positive NPs in the embryonic and early postnatal kidneys follows a stereotypical, well-described pattern, in which the initial multi-cell-layered tissue thins over time without much affecting the overall niche organization (Short et al., 2014). NPs in Gdnf^hyper/hyper kidneys were distributed around abnormally wide UB as a thinner layer (Fig. 1A-F; Fig. S1A,B). Quantification of NPs at E11.5 revealed an initial increase, which was transiently normalized at E12.5 but severely decreased at E14.5 in Gdnf^hyper/hyper kidneys (Fig. 1E-G; Fig. S1C-F). Mitosis analysis demonstrated that both endogenously increased and exogenously supplemented GDNF causes significant reduction in pHH3+ NPC (Fig. 1H; Fig. S1G). Collectively, the data show that excess GDNF, which primarily functions in the UB, where its receptors are expressed (Pachnis et al., 1993), induces a rapid drop-off in NP cell proliferation during early kidney development.

It is known that premature or accelerated NP differentiation can cause NP cell decline in their niche (Kobayashi et al., 2008; O'Brien, 2018). To assess this, nephrogenesis was examined by staining with LEF1 and cyclin D1, which label the earliest differentiation-committed cells and nephrons undergoing full epithelialization, respectively (Lindström et al., 2015). This showed a general decline in early nephron precursor numbers and indicated reduced nephron differentiation in embryonic Gdnf^hyper/hyper kidneys (Fig. 2). Glomerular density assessment further supported this and demonstrated lower overall densities in hypodysplastic Gdnf^hyper/hyper kidneys than in wild-type (WT) kidneys (Fig. S1G). Interestingly, analysis of the cortical glomeruli within the total measured area, and especially cortical-to-total glomeruli ratios, revealed slightly higher than WT ratios in Gdnf^hyper/hyper (Table 1; 17% in Gdnf^hyper/hyper, 12% in WT). This indicates that despite the early deceleration of NP self-renewal and differentiation, the kidneys facing excess GDNF during organogenesis have ongoing postnatal nephrogenesis with a minor increase in the latest born nephrons, which are most cortically located (Rumballe et al., 2011; Short et al., 2014).

Many mouse models with deficiency in kidney growth, either due to UB morphogenesis and/or nephron differentiation, die immediately after birth (Kuure and Sariola, 2020). Despite the significant hypoplasia, abnormalities in UB morphology and severely disrupted embryonic nephrogenesis (Fig. 2), Gdnf^hyper/hyper mice survive up to 3 weeks after birth (Kumar et al., 2015; Li et al., 2019), providing a possibility to examine postnatal nephrogenesis.

As has been established (O'Brien, 2018), our analysis of the NP populations at P1-P10 revealed dramatic loss in the WT kidneys at P3, in which SIX2-positivity was mainly detected in the differentiating nephron precursors, and only 10% of the analyzed UB tips (niches) were capped with nephron progenitor cells (NPCs) (Fig. 3A, n=2 kidneys). However, SIX2-positive NPs were detected in 59% of Gdnf^hyper/hyper niches at P3 (Fig. 3B, n=2 kidneys). Similarly, PAX2-positive NPs were detected in NP niches of Gdnf^hyper/hyper kidneys at P3, but the PAX2-positive cells localized exclusively in the differentiating nephron precursors in WT kidneys (Fig. 3C,D).

Nephrogenesis in postnatal Gdnf^hyper/hyper kidneys demonstrated significant improvement from that in embryonic kidneys, as exemplified by similar nephron precursor patterns in WT and Gdnf^hyper/hyper kidneys (Fig. 3E-H; compare with Fig. 2). Finally, we found that ∼18% of the NP niches sustained SIX2-positive NPCs at P4, and committed NPCs were present in Gdnf^hyper/hyper kidneys until P6, whereas committed NPCs were detected in WT kidneys only until P4 (Fig. 4; Fig. S1H). This demonstrates that the NP lifespan in vivo is not pre-fixed, and suggests that modulation of growth factor levels, specifically GDNF, contributes to the timing of the final nephron differentiation wave.

We hypothesized that persistent NPCs in postnatal Gdnf^hyper/hyper kidneys could either result from general compensatory mechanisms provoked by UB branching defects leading to renal hypoplasia (Kumar et al., 2015; Li et al., 2019) or alternatively derive from the GDNF-specific effects on the NP population. To discriminate between the options, postnatal NPs were analyzed in other models of renal hypoplasia. Fgf9;20-deficient kidneys show similar thinning of embryonic NP cell layers (Barak et al., 2012) as detected in Gdnf^hyper/hyper kidneys, but we failed to detect SIX2- and/or PAX2-positive cells in their postnatal NP niches (Fig. S2). To further elaborate on this, we next analyzed Hoxb7Cre; Mek1^fl/fl;Mek2^+/− kidneys (Ihermann-Hella et al., 2014) in which, similar to Gdnf^hyper/hyper mice, renal hypoplasia is caused by impaired UB branching. This model also lacked signs of prolonged nephrogenesis, as detected by the presence of SIX2- and/or PAX2-positive cells in their postnatal NP niches (Fig. S3), supporting the view that renal hypoplasia as such is not enough to maintain postnatal nephrogenesis.

Published data also demonstrates that renal hypoplasia alone is unable to support postnatal NP maintenance (Kuure and Sariola, 2020), which further suggests an active role for GDNF. However, we cannot completely exclude the possibility that excess growth factors other than GDNF would have the same effect.

Next, the differentiation potential of sustained NPs in the postnatal Gdnf^hyper/hyper kidneys was examined. Ki67 analysis revealed comparable postnatal proliferation patterns in WT and Gdnf^hyper/hyper kidneys until P4 (Fig. 5A,B). From P5 onwards, the Ki67-positive cells decreased in the cortical differentiation zone of WT kidneys (Fig. 5C,E,G). Such a drop in cortical proliferation was not detected in Gdnf^hyper/hyper kidneys, which showed actively cycling cells in nephron precursor-like structures at P7 (Fig. 5D,F,H) but no longer at P12 (Fig. S4A,B).

Postnatal nephron differentiation analysis demonstrated an excessive quantity of nephron precursors in Gdnf^hyper/hyper kidneys until P7, whereas these were detected in WT kidneys only up to P4 (Fig. 6; Fig. S4C-F). Such persistent postnatal nephrogenesis, however, failed to increase glomerular density from that in late embryonic stages (Fig. S5A-C; 6.4±2.1 versus 10.2±1.2, respectively; mean±s.e.m.). The cortical glomeruli proportion in Gdnf^hyper/hyper kidneys at P7 (59%) is 1.5× higher than in WT kidneys (39%) (Table 1) supporting the view that the sustained postnatal NP pool in Gdnf^hyper/hyper kidneys induces new nephrons after the normal nephrogenesis cessation.

Gdnf^hyper/hyper mice with prolonged nephrogenesis in severely hypodysplastic kidneys show no signs of neoplasia or tumorigenesis during their short lifespan (Kumar et al., 2015; Turconi et al., 2020). Gdnf^wt/hyper kidneys show milder embryonic decline of NPCs than Gdnf^hyper/hyper kidneys. These mice live a normal lifespan and do not show tumors in kidneys and other organs (Turconi et al., 2020) (Fig. S6A-D, n=62). The twofold GDNF increase during embryogenesis, however, fails to maintain the NP population in postnatal Gdnf^wt/hyper kidneys (Fig. S6E,F), suggesting a tight threshold-limit for the ability of GDNF to modulate the nephrogenic program.

Gdnf expression is lost from WT kidneys by P2 (www.gudmap.org), whereas its mRNA and protein are still present in Gdnf^hyper/hyper kidneys as late as P7 (Fig. 7A-D). The conservative GDNF receptor complex is expressed only in the epithelial UB tips (Costantini, 2012; Golden et al., 1999). Owing to the receptor localization in the tissue adjacent to the NPC population, we suspected that the effects of increased endogenous GDNF on nephrogenesis must be mediated through UB-derived, secreted, GDNF-dependent molecule(s) (http://www.signalpeptide.de/index.php) (Lu et al., 2009; Ola et al., 2011; Schmidt-Ott et al., 2005) (Table 2), or other molecules, the expression of which could change due to the oversized UB.

The expression analysis of UB-derived, secreted GDNF-targets revealed significant upregulation of Crlf1, Srgn and Wnt11 in Gdnf^hyper/hyper kidneys at E14.5 (Fig. 7E). Also, Wnt9b expression is significantly increased (Fig. 7E). Further analysis of the WNT/β-catenin signaling targets (Clevers and Nusse, 2012) in E14.5 control and Gdnf^hyper/hyper kidneys demonstrated significant upregulation of canonical target Axin2 (P=0.003371), whereas NP-expressed WNT targets (Karner et al., 2011) showed a trend of downregulation, likely reflecting the reduced overall number of NPCs (Fig. S7A). Together, these indicate that not only augmented GDNF-induced signaling but also increased Wnt9- and Wnt11-induced signaling with known functions in NPC regulation participate in mediating reciprocal communication from mutant UB to the adjacent NP pool at E14.5 (Karner et al., 2011; Kiefer et al., 2012; Nagy et al., 2016; O'Brien et al., 2018).

Molecular characterization of P5 Gdnf^hyper/hyper kidneys showed persistent upregulation of Crlf1, Etv4, Etv5, Cited1 and Uncx4.1 (also known as Uncx), but also revealed increased expression of potential niche factor Lama1 (Rayagiri et al., 2018), nephron differentiation inducer Wnt4 (Stark et al., 1994) and WNT signaling agonist R-spondin 1 (Rspo1) (Fig. 7F; Fig. S7B). Such a transcriptional profile suggests an augmented UB-derived effect on NPs, which appear to undergo a differentiation wave based on the decreased Fgf9 and Fgf20 expression and shifted localization of SIX2+ cells (Figs 4C,D and 7F).

We functionally tested whether the most significantly increased Crfl1, Wnt11 and Wnt9b expressions possibly affect the nephrogenic program in mouse. CRLF1, in complex with its physiological ligand cardiotrophin-like cytokine, induces differentiation in isolated rat kidney mesenchyme cultures (Schmidt-Ott et al., 2005). Characterization of Crfl1-knockout kidneys revealed comparable kidney size and nephrogenesis with WT kidneys (Fig. S6C-F), suggesting dispensable CRLF1 functions in vivo.

The contribution of increased canonical WNT signaling to the phenotypes caused by excess GDNF was tested next. IWR1 inhibits tankyrases 1 and 2, which are necessary for normal kidney development (Chen et al., 2009; Huang et al., 2009; Karner et al., 2010). IWR1 treatment alleviated UB tip numbers and morphology in the presence of excess GDNF (Fig. S8). Finally, we asked whether significantly increased Wnt11 expression in Gdnf^hyper/hyper kidneys contributes to the prolonged nephrogenic program in Gdnf^hyper/hyper mice by crossing Wnt11 knockout-allele to the Gdnf^hyper background. This failed to generate double homozygote Gdnf^hyper/hyper;Wnt11^−/− offspring (Table 3; n=95; P=0.005 Gdnf^wt//hyper;Wnt11^+/− breeding). These results indicate embryonic lethality for Gdnf^hyper/hyper;Wnt11^−/− mutants and forced us to focus the analysis on Gdnf^hyper/hyper;Wnt11^+/− kidneys.

Removal of one Wnt11 allele in Gdnf^hyper/hyper background slightly improved UB morphology and reduced the kidney size further from that in Gdnf^hyper/hyper kidneys, likely reflecting reduced formation of collecting duct cysts upon Wnt11 dosage decrease (Fig. S9A-C). However, SIX2-positive NP population was enhanced (29.8 NPC/tip versus 40.7 NPC/tip at E14.5 and 20.3 NPC/tip versus 30.9 at P0) and nephrogenesis was improved in the cortical differentiation zone of Gdnf^hyper/hyper;Wnt1^+/− kidneys (Fig. 8; Fig. S9D-F; compare also with Fig. 1F). Thus, these results suggest that increased GDNF levels augment several bud-derived GDNF/Ret targets, which, not alone but in combination with each other and together with increased canonical WNT signaling, are regulating NPCs in embryonic kidneys.

Nephron count is defined during fetal life and can vary greatly between individuals. NP expansion and lifespan greatly influences the final nephron endowment, and therefore identification of regulatory mechanisms that control NP self-renewal is important.

Here, we show that the neurotropic factor GDNF, known to initiate renal development and regulate UB morphogenesis (Costantini, 2010; Kurtzeborn et al., 2018; Li et al., 2019), is, despite its negative effects on kidney growth, capable of prolonging the nephrogenic program beyond its normal cessation. Our results demonstrate that excess GDNF has a bi-phasic effect on NP self-renewal and maintenance. Based on our analysis, we propose a model of how excess GDNF in Gdnf^hyper/hyper kidneys functions through mechanisms that involve changes in cell cycle progression and capacity to sense growth factor levels.

Initially in embryonic kidneys, high GDNF levels cause a rapid decline in NP numbers per niche, which is typical for many mouse models with primary deficiency in UB (Carroll and Das, 2013; Chen et al., 2015b; Costantini and Kopan, 2010; Liu et al., 2018). The premature NP decline seen in these previously published models is typically caused by their untimely differentiation, which is not the case in Gdnf^hyper/hyper kidneys. Instead, excess GDNF is capable of maintaining active cell cycle in the postnatal kidney cortex significantly longer than is seen in WT kidneys. Proliferation occurs in NP cells, which are maintained longer than in WT kidneys, and they keep producing new nephrons. In the future, it will be interesting to see whether it will be possible to administer excess GDNF in a manner favorable for continued postnatal nephrogenesis without severely affecting UB morphogenesis to distort the overall organ growth.

Our data demonstrate that the early decline in embryonic NPs of Gdnf^hyper/hyper kidneys is due to decreased cell proliferation rather than increased premature differentiation. This cellular mechanism is supported by molecular changes detected in known GDNF targets (Crlf1, Wnt11 and Srgn) (Lu et al., 2009; Ola et al., 2011) and in important regulators of NP cell maintenance signaling (Wnt9b, Wnt11 and their transcriptional targets) (Karner et al., 2011; Kiefer et al., 2012; O'Brien et al., 2018). Previous examination of longevity and resilience of NP cells by ablating the NP population with diphtheria toxin A, or by disturbing MAPK/ERK activation, which both resulted in rapid decline in NP cells without a positive effect on their lifespan (Cebrián et al., 2014; Ihermann-Hella et al., 2018), together with our results suggest that NP cell proliferation inherently plays a major role in defining their total lifetime. In agreement, NP cells in late embryonic kidneys show significantly reduced proliferation rates and accelerated differentiation, whereas the transplantation of old NP cells into earlier-stage kidneys prolongs their lifespan (Chen et al., 2015a).

Based on our results, we propose that GDNF contributes to the mechanism through which NP cells sense their developmental age and ultimately their lifespan. In a normal kidney, GDNF expression is high at the initiation of kidney morphogenesis and during the active growth phase, whereas a clear decline in GDNF levels occurs in late embryonic kidneys, resulting in loss of expression by early postnatal stage (Figs 7 and 9). GDNF levels in early Gdnf^hyper/hyper kidneys are abnormally high, which appears to be inhibitory for NP cell proliferation, likely due to the detected cellular and molecular changes within the mutant UB such as an altered WNT pathway (Fig. 7E,F; Fig. S7A,B), which has known effects on NPC biology (Karner et al., 2011; Kiefer et al., 2012; Nagy et al., 2016; O'Brien et al., 2018). Similar to WT kidneys, Gdnf mRNA and GDNF protein levels decrease in postnatal Gdnf^hyper/hyper kidneys in which Gdnf is manipulated at the post-transcriptional level allowing endogenous temporal regulation of the expression (Kumar et al., 2015). Likely, this diminishes GDNF expression to the level that is permissive for NP cell self-renewal and differentiation, which are detected as a sustained nephrogenic program in postnatal kidneys (Fig. 9). Such a newly recognized growth factor-dependent plasticity in renal differentiation bears hope for its use in regenerative purposes in the future.

The first clues for the possibility of lengthening the nephron differentiation period came from experiments inhibiting BMP/SMAD1/5 signaling, which led to slightly bigger kidneys with more nephrons than in vehicle-treated kidneys (Brown et al., 2015). Decreased dosage of Tsc1, encoding an mTOR inhibitor, maintains NP cells for an additional day, without reported effects on embryonic NP cell numbers (Volovelsky et al., 2018). Prolonged nephrogenesis was observed also in mice with overexpression of Lin28, Lin28b or inactivated Let-7 (Let-7a1; Let-7d; Let-7f1) (Urbach et al., 2014; Yermalovich et al., 2019), which also either show direct tumorigenesis or activation of the Igf2 or H19h loci associated with pediatric kidney cancer Wilms' tumor (Chen et al., 2018; Ogawa et al., 1993). As a note, no changes in pSMAD1/5 localization or levels were detected in Gdnf^hyper/hyper kidneys, which are hypodysplastic and compatible with life for three weeks.

The present study focuses on the characterization of GDNF function in nephrogenesis, which has been previously examined by diminishing endogenous Gdnf dosage and results in reduced nephron endowment (Cullen-McEwen et al., 2001, 2003). The canonical receptor complex for GDNF is formed by the RET tyrosine kinase and its co-receptor GFRa1, which are exclusively co-expressed only in the UB epithelium (Costantini, 2012; Golden et al., 1999; Pachnis et al., 1993). GDNF is a key niche factor for UB tip cells (Chi et al., 2009; Riccio et al., 2016; Shakya et al., 2005). We have previously shown that excess GDNF expands collecting duct progenitor pool at the expense of ureteric trunk elongation, resulting in expanded tip and short trunk phenotype due to defects in progenitor motility and shortened cell cycle length (Li et al., 2019). During the embryonic stages the combined effect of abnormally high GDNF, anomalous UB tip morphology-induced biophysical changes and associated molecular alterations in GDNF targets and the WNT pathway appear to jointly push NP cells towards diminished proliferation. We show here that the expanded UB tip morphology ameliorates over the time but persists at mild forms up to P4 in postnatal Gdnf^hyper/hyper kidneys (Fig. 5). UB tip cells are molecularly still present in P5 Gdnf^hyper/hyper kidneys, which express high levels of GDNF-regulated tip markers Crlf1, Etv4, Etv5 and Lama1, NPC-expressed canonical WNT targets Cited1 and Uncx4.1, as well as WNT agonist R-spondin1 (Fig. 7E,F; Fig. S7A,B) (Karner et al., 2011; Lu et al., 2009; Vidal et al., 2020). Such molecular changes in the postnatal UB likely thus provide a favorable environment for sustained NP self-renewal beyond P2/P3.

We identified UB-derived secreted CRLF1 and WNT11 as GDNF targets that, together with augmented canonical WNT signaling, additionally boosted by increased Wnt9b expression, mediate its effects on NP regulation. Our data also indicate that the function of CRLF1 is redundant with other cytokines as its inactivation does not affect renal differentiation. Both WNT9b and WNT11 regulate many aspects of NPCs and nephrogenesis (Carroll et al., 2005; Karner et al., 2011; Majumdar et al., 2003; Nagy et al., 2016; O'Brien et al., 2018). Upregulation of UB-expressed Wnt9b, which is known to regulate many aspects of nephrogenesis, in Gdnf^hyper/hyper kidneys may mimic its forced expression in NPs, which disrupts their maintenance versus differentiation equilibrium (Kiefer et al., 2012), and thus mechanistically contributes to imbalanced NP self-renewal in Gdnf^hyper/hyper nephrogenesis. This is supported by improved tip morphology upon canonical WNT inhibition by IWR1 in whole kidneys facing excess GDNF.

Our results show that IWR1 inhibition itself has a major effect on NPCs, which are dispersed and loosely adhered to UB tips. Canonical WNT inhibition in NPCs themselves likely contributes to the failure to augment their phenotype in the presence of excess GDNF. This is supported by our genetic experiments in which lowering UB-expressed Wnt11 expression in the Gdnf^hyper/hyper background improved NP maintenance and differentiation, but failed to fully rescue the renal hypoplasia caused by the excess GDNF. This may at least partially be because of the failure to inactivate both Wnt11 alleles in the Gdnf^hyper/hyper background due to the embryonic lethality of Wnt11^−/− (Nagy et al., 2016). Thus, one remaining Wnt11 allele may even augment the GDNF cascade as suggested earlier (Majumdar et al., 2003) and thus in part prohibit full rescue of nephrogenesis.

Our current study cannot rule out the possibility that some effects of GDNF on NPs are due to non-canonical GDNF signaling, as its potential alternative signaling receptors are expressed by the NPs (Brodbeck et al., 2004; Enomoto et al., 2004; Keefe Davis et al., 2013; Paratcha et al., 2003). However, both deletion of NCAM or expression of Gfra1 in UB only (by placing it under the control of the Ret promoter) maintain normal renal phenotype (Cremer et al., 1994; Heuckeroth et al., 1999; Keefe Davis et al., 2013; Rossi et al., 1999; Sariola and Saarma, 2003). This suggests that, at least on their own, neither NCAM nor GFRa1, are essential for the nephrogenic program.

In summary, our experiments suggest that GDNF, which has been, and currently is, in clinical trials for Parkinson's disease (Kirkeby and Barker, 2019), may bear potential to serve as a prospective means to improve nephron endowment in the future. Additional experimentation is needed to identify possible means to excessively activate GDNF signaling without harmful effects on kidney growth.

Animal care and research protocols were performed in accordance with the Code of Ethical Conduct for Animal Experimentation as well as European Union directives (Directive 2010/EU/63), and were approved by the National Animal Experiment Board of Finland.

Mice were housed in individually ventilated cages with food and water ad libitum. Optimal humidification and heating were provided constantly at the certified Laboratory Animal Centre of the University of Helsinki.

The generation and genotyping of the Gdnf^hyper (Kumar et al., 2015) and Fgf9;20 (Barak et al., 2012) mouse models has been described previously. Gdnf^hyper;Wnt11⁻ pups are from intercrosses of C57BL6 Wnt11⁻ (Nagy et al., 2016) and Gdnf^hyper lines, and were genotyped as previously described (Kumar et al., 2015; Nagy et al., 2016). All the mouse lines used in the studies were maintained on a 129Ola/ICR/C57bl6 triple mixed background.

The generation of Crlf knockout mice has been described previously (Alexander et al., 1999). PCR genotyping was performed using a Neo-specific primer set (5′-AGAGGCTATTCGGCTATGACTG-3′ and 5′-CCTGATCGACAAGACCGGCTTC-3′) together with gene-specific primer sets for Crlf1 (5′-GCCTAATAGGTGCTGGGTGA-3′ and 5′-GACCCTATCTGCGTTTTCCA-3′).

Embryos were staged according to the criteria of Theiler (1989) as described previously (Kuure et al., 2005) and collected after cervical dislocation of pregnant females at deep anesthesia with CO₂. Late embryonic and postnatal pups were sacrificed via decapitation. Tissue was dissected in Dulbecco's medium supplemented with 0.2% bovine serum albumin (BSA) followed by fixation with 4% paraformaldehyde (PFA) or organ culture.

Kidneys isolated from E11.5 and E12.5 mouse embryos were placed at the air-liquid interface supported by Transwell^® polyester membrane system (Costar) and cultured at 37°C in a humidified 5% CO₂ atmosphere with kidney culture medium [F12:DMEM (1:1)+Glutamax (Gibco)] supplemented with 10% fetal bovine serum and penicillin-streptomycin (Ihermann-Hella and Kuure, 2019). For the in vitro culture with 100 ng/ml exogenous GDNF (ProSpec Ltd) (Sainio et al., 1997) aimed at NPC quantification, each urogenital block containing metanephros, the definitive kidney rudiment, was halved along the linea mediana ventralis. One-half of each sample was cultured with exogenous GDNF while the other half served as the control. WNT inhibition experiments were first tested with 50-100 ng/ml GDNF and 50-100 µM IWR1 (I0131-25, Sigma-Aldrich). Based on these dose dependence experiments the optimal concentrations were 50 ng/ml GDNF and 75 µM IWR1.

For studies investigating the marker of interest on paraffin sections, kidneys were dissected from the embryos or pups of the indicated stages, followed by tissue processing and sectioning as previously described (Li et al., 2019). A minimum of five sections from each kidney and three to five different kidneys per genotype were used in each analysis. The exact sample numbers are provided in the figure legends.

Hematoxylin and Eosin (H&E) and immunohistochemistry performed on paraffin sections were completed as previously described (Li et al., 2019). In brief, the dissected tissue was fixed overnight with 4% PFA (pH 7.4), followed by dehydration and paraffinization with an automatic tissue processor (Leica ASP 200). Sections of 5 µm thickness were prepared and dewaxed in Xylene followed by rehydration before being stained with Harris H&E or undergoing heat-induced antigen retrieval in antigen retrieving buffer (10 mM Citrate; pH 6.0). For immunostaining, 3% BSA was used for blocking (1 h at room temperature) followed by 30 min 0.5% hydrogen peroxide treatment for chromogenic staining. Sections were then incubated with primary antibodies overnight at +4°C followed by species-specific secondary antibody incubation for 2 h at room temperature. Chromogenic detection was implemented with EnVision Detection System-HRP (DAB) kit (Dako).

E11.5 and E12.5 kidneys subjected to whole-mount immunofluorescence staining were fixed and permeabilized with 100% ice-cold methanol for 10 min before overnight incubation with primary antibodies. Incubation with the corresponding species-specific secondary antibodies was implemented on the second day following a vigorous wash with PBST (0.1% Tween 20 in PBS). Details about the primary and secondary antibodies used in the present study are listed in Table S1.

In situ hybridization was conducted as previously described (Ihermann-Hella et al., 2014). In brief, an anti-sense RNA probe against Gdnf exons was transcribed and hybridized on thick sections derived from P4 kidneys (10-15 sections/kidney, n=3 kidneys/genotype) embedded in 4% low melting agarose (NuSieve GTG, Lonza). BM Purple was used for colorimetric reaction.

GDNF protein levels in P7.5 kidneys were measured via ELISA as reported previously (Kumar et al., 2015). GDNF Emax^® Immunoassay (Promega) was used with acid treatment when carrying out the ELISA.

In detail, for each genotype, three kidneys were lysed immediately after being dissected. After measuring the protein concentration with DC protein Assay (Bio-Rad), 25 µg of total protein was loaded into the well on ELISA plate. Each sample was analyzed in duplicate.

Immunostaining and in situ hybridization on sections were imaged using either a Zeiss AxioImager equipped with the HXP 120V fluorescence light source, a Hamamatsu Orca Flash 4.0 LT B&W camera and Zeiss AxioCam 105 color camera, or with a Leica DM5000B equipped with Leica EL 6000 metal halide light source and Hamamatsu Orca-Flash4.0 V2 sCMOS camera. Whole-mount immunofluorescence staining was imaged using a Zeiss AxioImager with the Zeiss ApoTome application. Whole-mount imaging for intact kidney size measurements was performed using a Leica MZFLIII equipped with a Leica DFC7000T camera.

Quantification of the glomerular density was calculated from the sequential sections of E18.5 and P7 kidneys. The kidneys were sectioned through, and sections for glomerular quantification were picked every 125 µm to avoid iteration count of glomeruli. The number of glomeruli in each section were counted manually and the area of each kidney was measured with Fiji ImageJ software (V1.51) through outlining the contour manually. To calculate the glomerular densities, the total number of glomeruli and number of glomeruli in the cortex were counted separately. The average total kidney area and overall and cortical glomeruli numbers were then calculated for each sample. The average glomerular density was calculated by dividing the average glomerular number by the average measured area. The average of cortical glomeruli within the entire average area was calculated by dividing the average cortical glomerular number by the average measured total area. Only glomeruli with intact shape were recorded.

Quantification of the total SIX2-positive NPs at E11.5 (n=4 for WT, n=5 for Gdnf^hyper/hyper) and E12.5 (n=10 kidneys/treatment, three independent experiments) were based on the whole-mount immunofluorescence staining. The images were captured with the application of Zeiss ApoTome and then processed with Imaris Software (Bitplane) using spot tracking. All images were processed with identical analysis protocols. In order to normalize the variation between the litters, the mean quantity of SIX2-positive NPs in the WT kidneys within each litter was set to 100%, and the quantity of SIX2-positive NPs in the Gdnf^hyper/hyper kidneys was compared with the mean quantity in the WT kidneys within the litter.

The mitotic index of SIX2-positive NPs in E12.5 kidneys (n=10 kidneys/treatment, three independent experiments) cultured with exogenous GDNF were calculated in a similar way to normalize the variation between litters. The mean proportion of proliferating SIX2-positive NPs in the WT kidneys and kidneys cultured without exogenous GNDF within each litter was set to 100% and the proportion of proliferating SIX2-positive NPs in the presence of excess GDNF (Gdnf^hyper/hyper kidneys and exogenous application) from the same litter was plotted in the figure after comparing with the WT littermates. The mitotic index of SIX2-positive NPs in E11.5 kidneys cultured with exogenous GDNF were plotted as the original data/value.

The calculation of the average SIX2-positive NPs per nephrogenic niche (E14.5 and P0) and nephrogenic niches with progenitors (P3-P6) was conducted by immunofluorescence staining in paraffin sections. The number of SIX2-positive NPs was counted with Fiji ImageJ software (V1.51) on each section selected for the analysis using particle analysis and the number of tips was counted manually on the corresponding sections. The number of analyzed samples at E14.5 was three kidneys for WT and Gdnf^hyper/hyper kidneys and two kidneys for Gdnf^hyper/hyper;Wnt11^+/−. The number of tips analyzed at E14.5 was 207 in WT, 431 in Gdnf^hyper/hyper and 107 for Gdnf^hyper/hyper;Wnt11^+/− kidneys. The number of analyzed samples at P0 was three kidneys per genotype, and number of tips analyzed, which is presented in the corresponding figures, varies from 10 to 315, being lowest in WT P6 kidneys, in which the tips are rarely present.

UB tip measurements were carried out using the LAS X program of the Leica microscope operating system after 48 h culture in the presence of specified chemicals and visualization of the tips with E-cadherin antibody (Table S1). The tips were measured from six kidneys in each condition of two independent set of experiments in which the number of tips measured was: 101 for GDNF, 131 for IWR1 and 115 for 75 µM IWR1+50 ng/ml GDNF.

The kidney size measurements in experiments assessing the effect of Wnt11 deletion in Gdnf^hyper/hyper background were conducted with Fiji ImageJ software (V1.51) on whole-mount images through manual tracing of the contour of the kidney.

All values are presented as mean±s.e.m. except the main effects analysis of Gdnf and Wnt11 on kidney size, which are shown as mean± s.d. Statistical significance was set at P<0.05 for all the analysis. Independent-samples two-tailed t-test was used for pairwise comparisons with SPSS Statistics software (IBM; Version 25) in the study unless otherwise noted. The data obtained via in vitro tissue culture with exogenous GDNF was analyzed with the paired-sample t-test using SPSS Statistics software. The impact of genetically modified Gdnf and Wnt11 expression on kidney size was analyzed via two-way ANOVA followed by main effects test with SPSS Statistics software. The statistical significance for the deviation of gene distribution from Mendelian ratio was performed using χ² test also with SPSS Statistics software. The data obtained from ELISA, qRT-PCR and WNT inhibition experiments were analyzed with one-way ANOVA followed by Bonferroni post-hoc test using SPSS Statistics software.

Quantitative RT-PCR (qRT-PCR) experiments were conducted as previously reported (Kumar et al., 2015). In more detail, 150-500 ng of total RNA from each sample (n=3 kidneys/genotype at given stage) was treated with RNase-free DNase I (Thermo Fisher Scientific). The reverse transcription reaction was performed with random hexamer primers and RevertAid Reverse Transcriptase (Thermo Fisher Scientific) immediately after inactivation of DNase I with 5 mM EDTA at 65°C. Complementary DNA (cDNA) was diluted 10× and stored at −20°C until qRT-PCR.

For qRT-PCR, LightCycler 480 SYBR Green I Master (Roche) or Bio-Rad C1000 Touch Thermal Cycler upgraded to CFX384 System (Bio-Rad), supplied with SYBR Green I Master (Roche) and 250 pmol primers was loaded into the well of 384-well plates with a total volume of 10 µl. Negative control was always included in every reaction via setting conditions with minus-reverse transcription or water. A combination of Pgk1, Hprt and Gapdh was used as the reference genes in all the qRT-PCR experiments, except the one with P7.5 kidneys, in which mouse Actb was reference gene. Primer sequences can be found in Table S2.

Data obtained from qRT-PCR were analyzed as described previously (Kumar et al., 2015). Normalization was performed according to the geometric mean of the reference genes. All samples were analyzed in a duplicate manner. Results for a biological repeat were discarded when the Cq value for one or more of the replicates was 40 or 0, or when the Cq difference between replicates was >1.

We thank the personnel of Biomedicum Imaging and Light Microscopy Units at Helsinki Institute of Life Sciences (HiLIFE) for their valuable help in image analysis and quantification related techniques. The Wnt11 mice were kindly provided by Professor Seppo Vainio (University of Oulu, Finland) and sent to the Laboratory Animal Centre of HiLIFE, University of Helsinki where all animals were housed by the GM-unit of HiLIFE, University of Helsinki.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.197475