p53 enables metabolic fitness and self-renewal of nephron progenitor cells

Congenital or acquired nephron deficit results in hypertension and chronic kidney disease (Brenner et al., 1988; Hoy et al., 2008), a major health problem worldwide. Nephron endowment at birth is determined, in part, by the availability of the nephron progenitor cells (NPCs) and their efficient differentiation into nephrons. Nephron abundance varies amongst individuals and populations, with a demonstrated influence of genetics and maternal nutritional status on nephron number in humans (Benz and Amann, 2010; Hoy et al., 2006; Luyckx and Brenner, 2010; Wlodek et al., 2008). Self-renewal of NPCs ensures a supply of cells for nephrogenesis until the cessation of nephrogenesis at postnatal day 4 in mice and at 35 weeks of gestation in humans (Hartman et al., 2007; Rumballe et al., 2011). The nephron progenitor niche includes the cap mesenchyme (CM), in which the NPCs reside, and provides extrinsic cues, such as nutrients and morphogens that drive cell intrinsic signaling and metabolic pathways, to promote the availability of NPCs for subsequent nephrogenesis. Renewal of the NPC pool is dependent upon ureteric bud (UB) secreted Wnt9b-β-catenin signaling to Six2⁺/Cited1⁺ NPCs, as well as FGF, EGF and BMP signaling pathways, with their genetic inactivation resulting in defects in NPC renewal and differentiation (Brown et al., 2011, 2013; Karner et al., 2011).

Fate mapping studies with Cited1-CreER^T2 and Six2-GFPCreERT (Six2^GCE) mice have demonstrated that the Cited1⁺/Six2⁺ subcompartment marks the definitive self-renewing nephron progenitor domain that will differentiate to the inducible Cited1⁻/Six2⁺ and Wnt4⁺/β-catenin⁺/Lef1⁺ cells to form nascent nephrons (Boyle et al., 2008; Kobayashi et al., 2008; Mugford et al., 2009). Expression of Six2 is essential to maintain the progenitor state (Self et al., 2006). Loss of self-renewal, proliferative or survival capacity in the Cited1⁺/Six2⁺ CM will result in a diminished cap and a smaller NPC pool.

In its canonical role as a tumor suppressor, p53 (Tp53 – Mouse Genome Informatics) plays a key role in cell fate regulation by transcriptionally regulating genes that control cell cycle arrest, DNA repair, apoptosis or senescence, thus limiting the propagation of cells with damaged genomes (Amariglio et al., 1997; Asker et al., 1999; Aylon and Oren, 2007; Vousden and Prives, 2009). p53 also regulates genes in metabolic pathways such as oxidative respiration and glycolysis for energy generation and glucose homeostasis, genes in cell adhesion and migration via Rho signaling pathways, genes regulating polarity of cell division, and autophagy (Armata et al., 2010; Balaburski et al., 2010; Buchakjian and Kornbluth, 2010; Cicalese et al., 2009; Gadea et al., 2007; Olovnikov et al., 2009; Tasdemir et al., 2008). Recent studies in hematopoietic, mammary and neuronal stem cells link p53 with the regulation of self-renewal potential (Cicalese et al., 2009; Liu et al., 2009; Meletis et al., 2006). Although data from some tissue lineages indicates that p53 restricts self-renewal capacity and the size of the stem and/or progenitor pool, data from mouse embryonic stem cells suggest that p53 serves as a positive regulator of self-renewal, by maintaining strict genome integrity quality-control that is essential in proliferative self-renewing progenitor populations (Lee et al., 2010; Schoppy et al., 2010; Xu, 2005). Therefore, the requirement of p53 in the renewal or differentiation of stem cells and lineage-committed progenitors is clearly cell type and tissue dependent.

Integrative analysis of differential gene expression data from p53-null embryonic kidneys with p53 ChIP-Seq data has identified nearly 10% of the possible p53 target genes as enriched in the CM and nascent nephrons, indicating a substantial involvement of p53-mediated transcription in nephrogenesis (Li et al., 2013). To directly assess the contribution of p53 to NPC renewal and differentiation, we conditionally deleted p53 from the Six2⁺ CM. Six2Cre⁺;p53^fl/fl mice have hypoplastic kidneys and a nephron deficit (Saifudeen et al., 2012). Here, we show that the Six2(p53-null) CM exhibits a diminished NPC pool size and marked disorganization of the mesenchymal cells around the ureteric tip. The Cited1⁺ domain is completely lost by the time of birth. Further, adult mutant animals exhibit high blood pressure. RNA-Seq analysis of wild-type and mutant embryonic CM cells revealed that p53 is critically involved in regulation of cellular energy metabolism and cell adhesion pathways. These novel physiological functions of p53 on progenitor cell renewal, metabolism and adhesion have hitherto not been reported in a developing organ system.

To determine the functional significance of p53 in the CM, we conditionally deleted p53 from the Six2⁺ mesenchyme by crossing Six2-GFPCre [Six2^GC (Kobayashi et al., 2008; Park et al., 2007)] to p53^loxP/loxP mice to generate Six2^GC+;p53^loxP/loxP mice (hereafter referred to as Six2Cre⁺;p53^fl/fl). The breeding strategy used produced mutant and wild-type littermates of genotype Six2^GC+;p53^loxP/loxP or Six2^GC−;p53^loxP/loxP (Six2Cre⁻;p53^fl/fl), respectively. Crosses of Six2^GC+;p53^+/+mice were performed to enable fluorescence activated cell sorting (FACS)-mediated isolation of p53-wild-type Six2(GFP⁺) cells. Loss of p53 expression in the CM was confirmed by using real-time (RT)-PCR analysis (Fig. 1A) and subsequent RNA-Seq (below) (supplementary material Fig. S1).

The Six2Cre⁺;p53^fl/fl kidneys are hypoplastic as early as E13.5, with sparse, disorganized CM and UB-branching defects (Fig. 1B). This can also be visualized by using GFP fluorescence in Six2Cre⁺;p53^+/+ and Six2Cre⁺;p53^fl/fl kidneys (Fig. 2A,B). Histological examination after Hematoxylin and Eosin staining draws attention to the absence of a clearly defined nephrogenic zone and a marked deficiency of differentiating nephrons in postnatal day 0 (P0) kidney sections (Fig. 1C), as well as an increase in interstitial stroma. Of note are the glomeruli in PAS-stained P0 kidneys that display proteinaceous and/or hyaline material in Bowman's space, indicative of glomerular dysfunction (Fig. 1D). Immunofluorescent staining against the CM markers amphiphysin and Six2 accentuates the substantially diminished and less compact CM at E14.5-15.5 (Fig. 2C-F). Amphiphysin is a cell surface marker of the CM. The functional impact of a decrease of amphiphysin in a subset of Six2Cre⁺;p53^fl/fl GFP⁺ cells is not known at this time. GFP is expressed as a fusion protein with Cre and localizes to the nucleus and cytoplasm (Park et al., 2007). The diminished Cited1 domain is completely lost by P0 in Six2Cre⁺;p53^fl/fl kidneys (Fig. 2G-J). Despite the loss of the Cited1⁺ domain, Six2⁺ cells persist at P0, when a markedly smaller Six2⁺ CM is observed (Fig. 2K,L). In the absence of increased apoptosis (details below), an alternative explanation for the presence of a smaller (but not absent) Cited1⁻/Six2⁺ population could be diminished self-renewal and premature exit of the Cited1⁺ self-renewing cells to the transit, non-self-renewing pool (Cited1⁻/Six2⁺).

Flow cytometry analyses showed a significant (30%) decrease in Six2⁺ cell number from mutant kidneys (n=3, P<0.05) (Fig. 2M). As stated above, this decrease is not due to increased apoptosis, as the apoptotic index (staining of PARP1) in the E13.5 CM was not different between Six2Cre⁺;p53^fl/fl and Six2Cre⁻;p53^fl/fl littermate kidneys (Fig. 2N). Paradoxically, p53 loss results in a significantly lower proliferative index of Six2 cells (Fig. 2O,P). Unlike in the wild-type CM, the S-phase marker BrdU was present predominantly in laterally located Six2⁺ cells in Six2Cre⁺;p53^fl/fl kidneys, as opposed to the cells ventral to the ureteric tip (Fig. 2O). Cell cycle analysis showed that more Six2Cre⁺;p53^fl/fl cells were in the G1 phase, with a significantly smaller fraction in the S, G2 or M phases in comparison with the Six2Cre⁺;p53^+/+ cells (Fig. 2Q, Table 1). Of note, the branching defect in the mutant kidneys affects niche number and, thereby, impacts the total Six2⁺ cell number in these kidneys and, alongside the compromised proliferative capacity, is expected to contribute to the decreased count of mutant cells.

In contrast to the compact aggregates of cap cells around the ureteric tip in wild-type kidneys, cells of the Six2Cre⁺;p53^fl/fl CM are dispersed and loosely organized around the ureteric tip (Fig. 3B,D). Immunostaining of neural cell adhesion molecule (NCAM1, referred to here as NCAM), which decorates the intercellular domains of CM and nascent nephrons, is greatly decreased in the mutant CM (Fig. 3C,D). NCAM is required for cell-cell and cell-matrix interactions during development and differentiation (Shin et al., 2002). The transcription factor Pax2 is expressed in both mesenchymal and UB lineages (Dressler et al., 1990); here, Pax2 expression was decreased in CM but not UB of Six2Cre⁺;p53^fl/fl kidneys (Fig. 3C,D), as described previously (Saifudeen et al., 2012). Another intriguing feature of the Six2Cre⁺;p53^fl/fl phenotype is the loss of demarcation between the CM and surrounding Meis1⁺ cortical stroma, resulting in gaps between adjacent cap cells and between cap and UB cells, as well as the presence of stromal cells in close apposition to the UB tip (Fig. 3F, Fig. 2D), instead of a smooth uninterrupted CM adjacent to the UB (Fig. 3E).

The Six2Cre⁺;p53^fl/fl kidneys exhibited a noticeable deficit in renal vesicles upon immunostaining for Lhx1⁺ at E13.5 (Fig. 4A,B). Consequently, fewer LTA⁺ proximal tubules and WT1⁺ glomeruli were evident in P0 mutant kidneys (Fig. 4C,D). To determine whether nephron deficit in mutant kidneys was simply due to a decrease in NPC number or due to a possible defect in nephrogenesis, we induced canonical Wnt pathway-activated mesenchyme to epithelial transition (MET) in isolated Six2⁺ NPCs. FACS-isolated Six2(GFP⁺) cells from Six2Cre⁺;p53^+/+ and Six2Cre⁺;p53^fl/fl kidneys were aggregated and cultured for a total of 48 h in medium supplemented with glycogen synthase kinase-3 inhibitor 6-bromoindirubin-3′-oxime (BIO) for the first 24 h followed by medium with vehicle (DMSO) for the subsequent 24 h, as described previously (Park et al., 2012). NPC that have been induced to differentiate through transient activation of the Wnt/β-catenin pathway express E-cadherin, indicative of MET. Six2Cre⁺;p53^+/+ cells, upon induction, expressed E-cadherin and formed multicellular structures, as expected (Fig. 4G,I,K). By contrast, Six2Cre⁺;p53^fl/fl cells showed weak E-cadherin expression and failed to form complex structures (Fig. 4H,J,L), supporting the notion that defective MET is a contributor to the nephron deficit. Altogether, our data demonstrate that the nephron deficit in Six2Cre⁺;p53^fl/fl kidneys is due to at least two developmental defects: (a) impaired renewal of progenitor cells, and (b) a differentiation defect that is downstream of Wnt signaling which disrupts proper MET.

Therefore, next, we examined the expression of genes that are the earliest markers of nephrogenesis and are essential for MET. In situ hybridization analyses showed a clear decrease in Fgf8 mRNA in nascent nephrons in mutant kidneys (Fig. 4M,N). Decreased Fgf8 expression was also observed by using RT-PCR analyses of RNA from mutant kidneys (data not shown). Surprisingly, however, expression of the tubulogenic factors Pax8 and Wnt4 did not show a marked change in expression (supplementary material Fig. S2). Expression of Fgf8 and Wnt4 has been shown to be concordant in the induced mesenchyme, and these factors function downstream of the inductive signal from Wnt9b (Carroll et al., 2005; Grieshammer et al., 2005; Perantoni et al., 2005). Wnt4 and Lhx1 are not expressed in Fgf8-null kidneys, and advanced epithelial structures do not form. Fgf8 hypomorphs express Wnt4 and Lhx1 and progress through nephrogenesis (Grieshammer et al., 2005). Despite being Fgf8 hypomorphs and expressing Wnt4 at normal levels, epithelialization occurs inefficiently in Six2Cre⁺;p53^fl/fl kidneys, indicating additional defects that impair differentiation of the Six2Cre⁺;p53^fl/fl CM. Indeed, addition of Fgf8-soaked beads to E13.5 Six2Cre⁺;p53^fl/fl kidneys did not facilitate even a partial rescue of nephrogenesis (supplementary material Fig. S3). Thus, whether defective MET in the mutant NPCs is a result of an impaired ability to be induced or of impaired epithelialization remains to be determined.

Reduced nephron number from birth is strongly associated with the development of adult-onset diseases, such as hypertension (Brenner et al., 1988). Because both germline and conditional p53^−/− mice exhibit impaired kidney development, renal hypoplasia and nephron deficit (Saifudeen et al., 2009, 2012; this study), we sought to determine the physiological impact of p53 loss on blood pressure. Continuous ambulatory blood pressure was monitored for 6 or 7 days in 2-month-old wild-type or mutant mice. The average time of tumor onset in p53^−/− mice is ∼3-4 months of age (Donehower, 1996). Younger mice could not be used because the small size of the mice increased the risk of death after implanting telemetry transmitters. Daytime blood pressure was significantly higher in p53^−/− mice compared with that of wild-type mice (mean arterial pressure: 109.4±2.3 versus 95.5±1.1 mmHg, P<0.05, n=6 per genotype) (Fig. 5A), as it was in Six2Cre⁺;p53^fl/fl mice (mean arterial pressure: 110.0±0.97 versus 100.8±0.68 mmHg, P<0.005, n=3 per genotype) (Fig. 5B). These data implicate developmental defects in p53-deficient kidneys as an underlying cause of hypertension early in life.

To obtain a global transcription profile of Six2Cre⁺;p53^+/+ and Six2Cre⁺;p53^fl/fl cells at E15.5, we utilized the Illumina TruSeq platform for paired-end RNA sequencing. Nearly 200 million total reads were obtained for each wild type and mutant (supplementary material Table S1). The expression of UB-specific genes (Ret and Wnt9b) was not detected, validating the enrichment of CM cells over UB lineage cells after FACS. Additionally, undetectable levels of reads from p53 exon 2 to exon 10 confirmed efficient deletion of this loxP-flanked fragment by Cre recombinase (supplementary material Fig. S1). RNA-Seq data was analyzed to obtain differential expression using GeneSpring, and confirmed by using EBSeq and RSEM.

Six2 expression does not decrease in mutant CM (Table 2), confirmed by using RT-PCR (Fig. 6A), as well as immunostaining (Fig. 2F,L). A decrease in Cited1 mRNA (−1.3 fold), as detected using RNA-Seq, in Six2Cre⁺;p53^fl/fl cells (Table 2) confirmed the decrease observed by using immunostaining (Fig. 2H). Decreased expression of other known CM genes (GDNF, Sall1, Bmp7, Fgf20, Mdm2, Meox1) (Table 2 and GEO accession GSE56253) was not detected, implicating a previously unidentified p53-mediated mechanism that is independent of these regulators in the maintenance of nephron progenitor cell number.

The top most downregulated genes included phosphoenol pyruvate carboxykinase 1 (Pck1, −14.3 fold), cytochrome P450 2D26 (Cyp2d26, −13.0 fold), meprin1a (Mep1a, −12.5 fold), aldolase b (Aldob, −12.0 fold), Rec8 (−9.3 fold) and Tmem174 (−9.0 fold). The most upregulated genes included Fau (4.3 fold), Trim25 (2.2 fold) and H2aZ (H2afz – Mouse Genome Informatics; 1.9 fold). Fau is a fusion protein of an N-terminal ubiquitin-like protein Fubi fused to the ribosomal protein S30 at the C-terminal (Rossman et al., 2003). Trim25 belongs to the tripartite motif (TRIM) family and is an E3 ubiquitin ligase as well as a newly identified RNA-binding protein in mouse embryonic stem cells, suggesting a role in stem cell physiology (Kwon et al., 2013). H2AZ is a highly conserved histone variant that is essential for embryonic development in mice, and it might participate in telomere maintenance and regulate chromatin dynamics (Morillo-Huesca et al., 2010; Morrison and Shen, 2009).

Differentially expressed genes (supplementary material Table S2) were subjected to Gene Ontology (GO) analysis using the Ingenuity Pathway Analysis (IPA) software. Based on gene expression changes between wild-type and Six2Cre⁺;p53^fl/fl kidneys, IPA analysis predicted a significant alteration of function in six categories – development and morphology (39%), adhesion and migration (23%), diseases and abnormalities (16%), cell survival (14%), metabolism (6%) and ion transport (2%) (Fig. 6B). Enrichment analysis showed the highest enrichment of differentially expressed genes for (a) metabolic disorders; (b) cellular movement, growth and proliferation; and (c) carbohydrate and lipid metabolism (Table 3). A complete list of genes expressed in E15.5 Six2⁺ CM are posted on GEO under accession GSE56253.

The expression of several genes from the differential expression list were validated by using Nanostring nCounter digital gene expression analysis (Kulkarni, 2011). As shown by using nCounter quantification of mRNA expression in Fig. 7, a good correlation in expression trends between the RNA-Seq and nCounter data was obtained. The magnitude of fold change is higher with RNA-Seq, possibly from sensitivity and data normalization differences that exist between the two techniques.

The mRNA profile of Six2Cre⁺;p53^fl/fl cells revealed dysregulation of genes involved in glucose metabolism, the pentose phosphate pathway and the electron transport chain (ETC) (Table 4). The expression of several genes involved in ATP production, namely, components of the ETC or genes required to maintain a proton gradient for ATP production during oxidative respiration, was dysregulated. These are among the most downregulated genes in RNA-Seq data (Table 4). The decreased expression was confirmed by using quantitative nCounter expression analysis (Fig. 7), although the decrease was not of the same magnitude as that found by using RNA-Seq. Whether these genes are direct p53 targets in the CM remains to be determined. None of the tricarboxylic acid cycle (TCA) cycle enzymes showed expression changes.

Because of the indicated repression of energy pathways in Six2Cre⁺;p53^fl/fl cells from the gene expression data, we compared ATP levels between isolated E15.5 Six2Cre⁻;p53^fl/fl and Six2Cre⁺;p53^fl/fl cells. Mutant cells exhibited a significant, nearly 2-fold decrease in ATP levels than the wild-type cells (P<0.0001) (Fig. 8A). Because this could reflect mitochondrial dysfunction in Six2Cre⁺;p53^fl/fl cells, and because mitochondria are the major source of reactive oxygen species (ROS) (Holmström and Finkel, 2014), we measured ROS levels in these cells. ROS levels were consistently lower by ∼30% in mutant cells (Fig. 8B). In summary, these data clearly indicate energy metabolism differences between wild-type and p53-deficient Six2⁺ cells and provide strong evidence of metabolic dysregulation of the p53-null CM.

Expression of cell cycle inhibitors p21 (Cdkn1a) and p57 (Cdkn1c) was decreased and, thus, cannot account for the proliferation defect of the Six2Cre⁺;p53^fl/fl CM (supplementary material Table S3). Decreased proliferation might result from the unavailability of materials that are required for new cell synthesis, such as nucleotides, lipids and ATP. The downregulation of genes regulating amino acid metabolism, such as ornithine decarboxylase (Odc1, −1.6 fold) and arginine decarboxylase (Adc; Azin2 – Mouse Genome Informatics; −1.6 fold), and contributing to polyamine synthesis was also observed. Polyamines are maintained within a very narrow range because a decrease in their concentrations inhibits cell proliferation, whereas excess appears to be toxic (Landau et al., 2012). Because decreased polyamine synthesis is usually observed in aging cells (Minois et al., 2011), we tested whether the Six2Cre⁺;p53^fl/fl cells were senescent. We found neither increased expression of cell cycle inhibitor p21 nor positive staining for senescence-associated β-galactosidase (SA-βgal) (supplementary material Fig. S4), indicating that the Six2Cre⁺;p53^fl/fl cells are not in replicative senescence.

The ability of p53 to restrict cell proliferation is associated with its well-characterized function as a tumor suppressor. Contrary to the prototypical roles of p53 in cancer and embryonic stem cells, our data point to a previously unknown role for p53 in enabling the renewal of multipotent progenitors in the nephrogenic niche. Mechanistically, our findings suggest that p53 is required to maintain energy metabolism and cell-matrix interactions, and consequently niche integrity. We propose that the absence of p53 triggers a ‘metabolic crisis’, resulting in loss of niche architecture, depletion of the self-renewing Cited1⁺ population and nephrogenesis defects that result in disease early in adulthood. Nephron progenitor loss in this conditional p53-deletion model is independent of apoptosis, cell senescence or impaired expression of known regulators of NPC maintenance and survival (Six2, Fgf20, Bmp7, Sall1, WT1 and Mdm2) (Barak et al., 2012; Basta et al., 2014; Blank et al., 2009; Hilliard et al., 2014; Kobayashi et al., 2008; Self et al., 2006).

In accord with the loosely organized CM in Six2Cre⁺;p53^fl/fl kidneys, RNA-Seq showed a high enrichment of altered gene expression in cell adhesion and migration pathways (Fig. 7B). These include genes encoding collagens, which contribute to the extracellular matrix, as well as genes involved in cytoskeletal remodeling (Table 5). Although the functions of many of the products of these genes in the NPC are unknown, disruption of biophysical and biomechanical cues from the extracellular matrix to cells within the niche can influence renewal and differentiation programs (Cox and Erler, 2011).

The Six2Cre⁺;p53^fl/fl kidneys clearly harbor fewer generations of nascent nephrons. However, unlike Six2, Fgf9, Fgf20 or Sall1 mutant kidneys (Barak et al., 2012; Basta et al., 2014; Self et al., 2006), the loss of Cited1⁺/Six2⁺ progenitors in Six2Cre⁺;p53^fl/fl kidneys is not a result of precocious differentiation. No ectopic renal vesicles were visualized to suggest the smaller cap is due to advanced differentiation of progenitors. Indeed, this was not anticipated because Six2 expression is robust. The overall decrease in nephron number is multifactorial: impaired progenitor cell renewal, secondary UB branching defects and loss of the compact niche architecture, which disrupts the CM microenvironment and the signaling pathways between the CM and UB lineages and between the CM and stroma lineages (Fig. 9A) that are crucial for NPC renewal and differentiation (Das et al., 2013; Fetting et al., 2014). A simple decrease in proliferation would also result in fewer Six2⁺ cells, but not in the progressive loss and premature absence of Cited1⁺ cells observed here. The progressive loss of Cited1⁺ cells suggests a decreased capacity for self-renewal because, as per our current knowledge, Cited1 marks the self-renewing population. Proliferation of Six2⁺ cells (see BrdU staining, Fig. 2O) explains why Six2⁺ cells are not entirely lost by P0.

One of the most interesting findings of this study is that the top dysregulated genes belong to energy metabolism pathways. Enzyme deficiencies in metabolic pathways such as glycolysis, the TCA cycle and the ETC result in embryonic lethality (Kane and Buckley, 1977; Rahman et al., 2014). These pathways support the production of intermediates that act as electron carriers, and they play an essential role in anabolic pathways, such as fatty acid and nucleotide biosynthesis, which are required components for cells undergoing proliferation. The balance of biomass synthesis versus energy production through glycolysis and oxidative respiration, respectively, is crucial for cell proliferation. Downregulation of TCA cycle genes blocks mitosis in Caenorhabditis elegans embryos (Rahman et al., 2014), whereas mitochondrial depolarization in HCT116 cells arrests cells in G1/S, as does reduction of ATP levels in Drosophila cells (Mandal et al., 2005; Mitra et al., 2009).

p53 counteracts aberrant proliferation via the inhibition of glycolysis and the stimulation of oxidative phosphorylation through direct regulation of TIGAR (also known as fructose-2,6-bisphosphatase) and Sco2, respectively (Bensaad et al., 2006; Matoba et al., 2006). p53 loss in cancer results in increased glucose metabolism by means of glycolysis over oxidative phosphorylation even under aerobic conditions, and is termed the Warburg effect (Vander Heiden et al., 2009). Unlike in cancer cells, nephron progenitors lacking p53 showed no alterations in expression of either TIGAR or Sco2. Rather, our data show transcriptional repression of mitochondrial respiration and glucose metabolism pathways. A recent study describes the ability of p53 to adaptively regulate mitochondrial respiration in Drosophila cells and to maintain the supercompetitive status of the Myc⁺ cells by enhancing the metabolic flux (de la Cova et al., 2014). Interestingly, p53 loss in these cells reduces their viability and results in impaired metabolism, as it does in Six2Cre⁺;p53^fl/fl cells, suggesting a cell context-dependent regulation of cellular processes. p53-mediated stress-induced competition has also been described in the mammalian hematopoietic system using a mosaic mouse model with graded levels of p53 activity (Bondar and Medzhitov, 2010). Here, competition did not involve apoptosis, rather, the outcompeted cell did not contribute to clonal expansion by dropping out of the proliferative pool. In this system, the outcompeted ‘loser’ cells undergo a permanent growth arrest and are refractory to growth factors. Interestingly, genes that control interactions of the cell with its environment, such as cell adhesion and migration, are implicated in competitive interactions. p53 knock down in hematopoietic stem cells (HSCs) demonstrates that p53 positively regulates self-renewal independently of apoptosis (Milyavsky et al., 2010). Self-renewal in the HSCs is dependent upon p53-mediated genome stability and the maintenance of DNA damage response. Although we did not observe an increase in foci of phosphorylated γH2Ax in Six2Cre⁺;p53^fl/fl CM (data not shown), an indicator of DNA damage, p53 might play a similar protective role in the nephron progenitors by providing metabolic stability to the NPCs under stress.

ATP is necessary to fuel cell proliferation, migration and adhesion. Energy depletion resulting from decreased ATP production provides a mechanistic link to the Six2Cre⁺;p53fl/fl CM phenotype, as does a decrease in ROS, an essential regulator of redox-dependent signaling that is crucial in neural and intestinal stem cell self-renewal, hematopoietic progenitor maturation in Drosophila and genomic stability (Holmström and Finkel, 2014). We propose that a niche-defined change in energy metabolism pathway usage would drive the Cited1⁺/Six2⁺ self-renewing cells to Cited1⁻/Six2⁺ transit cells (Fig. 9B, purple energy sign). Untimely alterations in energy metabolism pathways (Fig. 9B, yellow energy sign) could impair self-renewal and promote an en masse transition of Cited1⁺/Six2⁺ cells to non-renewing Cited1⁻/Six2⁺ cells, gradually resulting in the complete loss of Cited1⁺ cells but maintenance of Six2⁺ cells (expectedly fewer) (Fig. 9B). Indeed, the significant decrease in the expression of glycolysis pathway genes that has been reported in nephron progenitors at the end of nephrogenesis emphasizes the involvement of metabolic regulation in cell fate decisions (Brunskill et al., 2011), and whether p53 is involved in these temporal metabolic and cell fate decisions remains to be determined.

The energy metabolism model posits how the progenitor pool self-renews while simultaneously contributing to the differentiation-competent pool in the nephrogenic niche. p53 might play an adaptive role in the self-renewing Cited1⁺ progenitors, assimilating multiple external and internal cues into a homeostatic metabolic response in order to maintain a robust cell that is capable of self-renewing. Differences in cellular response might determine the continued participation of the Cited1⁺/Six2⁺ cell in renewal or its exit to the transit Cited1⁻/Six2⁺ population to undergo induction and differentiate. In this scenario, a p53-null cell would be refractory to external stimuli (positional, nutrient, mitogenic) and unable to modulate an internal metabolic response in order to fit the needs of the cells. Although this model points to a cell-intrinsic defect in the proliferation of Six2Cre⁺;p53^fl/fl cells, the defect might, in part, be cell extrinsic – for example, limited growth factor availability at the cell surface due to adhesion defects and inadequate or inappropriate cell-cell interactions (Fig. 9A). Premature entry of less fit cells into the non-renewing pool would result in a smaller pool of progenitors being available for nephrogenesis, leading to nephron deficit. However, this outcome protects the viability of the organism and could be more homeostatic than death of the unfit cells that would lead to either an extreme dearth or an absence of nephrons. Whether adhesion defects precede or follow the altered metabolism status of the mutant CM remains to be determined.

In summary, the present study demonstrates that p53 is required for the metabolic fitness of nephron progenitors and explains the apparent paradox of why p53 loss impeded rather than enhanced cell proliferation. p53 fulfills a physiological function in development and is required to maintain progenitor cell homeostasis. Further mechanistic studies to define the roles of p53, energy metabolism and cell adhesion and migration in self-renewal versus differentiation are ongoing.

The Six2^GC transgenic mice express a GFP-fused Cre recombinase under the control of a Six2 BAC transgene [a gift from A. McMahon (Kobayashi et al., 2008; Park et al., 2007)]. p53^loxP/+ mice were obtained from Jackson Laboratories. Six2^GC;p53^loxP/loxP mice were crossed to p53^loxP/loxP mice and embryonic kidneys harvested at desired time points. All animal protocols utilized in this study were approved by and in strict adherence to guidelines established by the Tulane University Institutional Animal Care and Use Committee.

GFP⁺ kidneys were processed enzymatically to single cell suspensions for GFP-mediated FACS of p53-deleted Six2 cells (Six2Cre⁺;p53^fl/fl). Wild-type GFP⁺ cells were similarly sorted from Six2Cre⁺;p53^+/+ mice. Gating of GFP⁺ cells was performed based on control GFP⁻ cells. For quantification, CM cells were counted from pooled E15.5 kidneys from at least three litters of Six2Cre⁺;p53^+/+ and Six2Cre⁺;p53^fl/fl embryos.

Treatments using 500 nM and 2 µM BIO were performed as described previously (Park et al., 2012). Briefly, 1×10⁵ FACS-isolated Six2Cre⁺;p53^fl/fl or Six2Cre⁺;p53^+/+ cells were pelleted and placed on a Transwell filter. DMEM/F12 medium with fetal bovine serum (FBS) was supplemented with BIO or DMSO. After 24 h, BIO medium was replaced with complete medium containing DMSO, for another 24 h. At this time, cells were fixed in 4% paraformaldehyde for 10 min and stained with an antibody against E-cadherin (1:200, 610181, BD Biosciences). Treatments using BIO were performed on three separate pools of FACS-isolated cells from three litters.

Manual magnetic-activated cell sorting (MACS) was used to sort Six2⁺ cells from E15.5 kidneys. Depletion of stromal, endothelial and erythroid cells was performed as previously described (Brown et al., 2013). Six2⁺ cells were isolated and plated in nephron progenitor expansion medium (NPEM) (A.B. and L.O., unpublished formulation) for the measurement of ATP levels.

Cell cycle analysis (n=3, each genotype) was performed on pools of 100,000 cells from embryos from multiple litters. FACS-isolated GFP⁺ Six2Cre⁺;p53^fl/fl or Six2Cre⁺;p53^+/+ cells were fixed in ice-cold ethanol and stained with propidium iodide and treated with RNase A. The stained cells were loaded into the BD FACSArray bioanalyzer for DNA content-based cell cycle analysis.

Kidneys at E15.5 were fixed in 10% formalin and processed for paraffin embedding and sectioning. Immunostaining was performed on 5 µm sections as previously described (Saifudeen et al., 2009). See supplementary material Methods for antibody information and image capture details. Whole-mount staining was performed on E12.5-13.5 kidneys, as previously described (Saifudeen et al., 2009). Staining for SA-βgal was performed on frozen tissue sections as per the manufacturer's protocol (BioVision). Hematoxylin and Eosin staining was performed on 5 µm paraffin-embedded tissue sections, as described previously (Cardiff et al., 2014). Periodic acid-Schiff (PAS) staining was performed as described previously (Saifudeen et al., 2009).

Section in situ hybridization was performed on paraffin-embedded kidney sections as described previously (Hilliard et al., 2014). The following cDNA probes were used: Pax8 (from A. McMahon), Wnt4 (from A. McMahon) and Fgf8 (from Gail Martin and John Rubenstein).

PBS-washed Affigel Blue beads were soaked in recombinant Fgf8 (1 μg/ml, R&D Systems) for 30 min at 37°C, washed in PBS and placed on kidneys at E13.5 for 72 h. At this time, kidneys were fixed in 2% paraformaldehyde and stained with an antibody against Lhx1. Bovine serum albumin (BSA)-soaked beads were used as control.

RNA was purified from Six2⁺ FACS-isolated cells from Six2Cre⁺;p53^+/+ and Six2Cre⁺;p53^fl/fl kidneys at E15.5. Poly(A) RNA was selected for 100 bp/read paired-end Illumina TruSeq RNA-Seq on the HiSeq 2000 system. Sequence aligner Novoalign v2.08.02 was used to align the raw reads to mouse genome assembly GRCm38/mm10. The quality of raw reads was tested by using FastQC (Babraham Bioinformatics). Before analyzing differential gene expression between wild-type and mutant cells using the GeneSpring 12.5 NGS module, the following parameters were applied to the reads – ‘Benjamini Hochberg FDR Multiple Testing’ corrected P-value (Audic Cleverie Test) <0.05, and FPKM cutoff ≥3 on wild-type samples. FPKM reflects the normalized enrichment of a gene (Garber et al., 2011). TPM represents Transcript Per Million, which measures the abundance of gene transcripts (isoforms) according to RNA-Seq readout (Wagner et al., 2012). Fold change (knockout divided by wild type) cutoff was set to 1.5 for Ingenuity Pathway Analysis. Differential expression data was processed using GeneSpring, EBSeq or RSEM. Predicted genes and pseudo-genes were excluded to improve the accuracy of gene identification by IPA. Processed datasets are available in Gene Expression Omnibus (GEO accession number GSE56253). See supplementary material Methods for quality codes.

Direct, digital detection of mRNA was performed using the nCounter Analysis System (Nanostring Technologies) as per the manufacturer's protocol (Geiss et al., 2008; Kulkarni, 2011). mRNA was pooled from wild-type or mutant kidneys from multiple litters, and at least three independent pools per genotype were processed for expression analysis. RNA was loaded at 100 ng per sample, hybridized with a custom gene expression codeset containing unique pairs of 35-50 bp reporter probes and biotin-labeled capture probes, in addition to positive and negative controls. Probe target sequences are listed in supplementary material Table S4. Three housekeeping genes were used for normalization: B2m, Tbp and Pgk1. Data analysis was performed using nSolver Analysis Software v2. All samples passed quality control; no sample was flagged. All samples were normalized to have the same geometric mean of counts of an nCounter-provided positive-control probe, which was spiked into all reactions. See supplementary material Methods for hybridization and processing details.

RT-PCR was performed on RNA from FACS-isolated Six2Cre⁺;p53^+/+ and Six2Cre⁺;p53^fl/fl cells at E15.5 to determine the expression of the genes encoding p53, Cited1, Six2, Meox and Osr1. All primer probes for the TaqMan gene expression assay were purchased from Applied Biosystems. For p53 expression, primer-probes spanning exon 9 to exon 10 (Mm01337166-mH) were used. Reactions were prepared using the TaqMan RNA-to-CT 1-Step Kit and amplified using Stratagene Mx3000P and Mx3005P instruments. Expression was normalized against endogenous Gapdh levels. Experimental and biological triplicates were applied.

ATPlite (Perkin Elmer) luminescent assay was performed as per the manufacturer's instructions. This assay is based on light produced by conversion of Luciferin by Luciferase to Oxyluciferin in an ATP-dependent reaction. The emitted light is proportional to ATP concentration in the cell lysate. MACS-isolated Six2Cre⁺;p53^fl/fl or Six2Cre⁻;p53^fl/fl cells (2.5×10⁴/well) were seeded onto an ATP-free microplate. At 18-20 h after seeding, cells were lysed in the well, and the luminescence was measured after the addition of substrate. Values were normalized to cell number. Multiple experiments were performed with cells isolated from different litters (n=3).

5.0×10⁴ FACS-isolated Six2Cre⁺;p53^fl/fl or Six2Cre⁺;p53^+/+ cells were seeded per Transwell membrane. At 24 h after plating, Cyto-ROX reagent (Invitrogen) with Cy5 label was added to the medium. After a 30 min uptake period, cells were washed twice in PBS, trypsinized and counted by using flow cytometry. GFP⁺ cells were counted and factored for Cy5 intensity.

Telemetry transmitters were implanted into the carotid arteries of 2-month-old p53^−/− (n=6) and p53^+/+ (n=6) male mice on a C57BL/6 genetic background, as described previously (Li et al., 2012). After a 2-week recovery period post-surgery, blood pressure was recorded continuously for 7 days. Six2Cre⁺;p53^fl/fl and wild-type mice (n=3 per genotype) were monitored in a similar manner.

Our sincere appreciation to Dr Samir El-Dahr, for his insightful discussions and critical reading of the manuscript. We thank Dr Hua Lu for helpful discussions. We also thank Dr Erik Flemington at the Tulane Cancer Center and the COBRE Cancer Crusaders Next Generation Sequence Analysis Core for his assistance with bioinformatics [supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Number P20GM103518]. We thank Mylinh Bernardi at the Genomic and RNA Profiling Core at Baylor College of Medicine for her assistance with RNA-Seq and Nanostring nCounter gene expression analyses. Our thanks also to the Tulane Hypertension and Renal Centers of Excellence Molecular and Imaging Core and Mouse Phenotyping Core, and Dr Sylvia Hilliard for her assistance with the Fgf8 bead experiments.