Hamartin regulates cessation of mouse nephrogenesis independently of Mtor

Significance

Mammals form the final numbers of nephrons, the functional units of their kidneys, before or within a few days after birth. This initial endowment declines throughout life without the ability to replace nephrons lost to injury. Therefore, humans with low nephron numbers (such as premature infants) have higher rates of kidney disease in adulthood. Here we report that partial reduction in hamartin activity within mouse renal progenitors can lead to a significant increase in nephron numbers in newborns. Nephrogenesis was prolonged by at least a day and correlated with higher engraftment rates in the developing niche in a stem cell engraftment assay. We provide evidence that this phenotype is independent of the inhibitory effect of hamartin on the Mtor complex.

Abstract

Nephrogenesis concludes by the 36th week of gestation in humans and by the third day of postnatal life in mice. Extending the nephrogenic period may reduce the onset of adult renal and cardiovascular disease associated with low nephron numbers. We conditionally deleted either Mtor or Tsc1 (coding for hamartin, an inhibitor of Mtor) in renal progenitor cells. Loss of one Mtor allele caused a reduction in nephron numbers; complete deletion led to severe paucity of glomeruli in the kidney resulting in early death after birth. By contrast, loss of one Tsc1 allele from renal progenitors resulted in a 25% increase in nephron endowment with no adverse effects. Increased progenitor engraftment rates ex vivo relative to controls correlated with prolonged nephrogenesis through the fourth postnatal day. Complete loss of both Tsc1 alleles in renal progenitors led to a lethal tubular lesion. The hamartin phenotypes are not dependent on the inhibitory effect of TSC on the Mtor complex but are dependent on Raptor.

Mammalian nephrogenesis progresses through a series of developmental stages during mid to late gestation culminating with the metanephros, which persists as the definitive kidney in the adult. The kidneys contain multiple nephrons, consisting of glomerular capillary tufts surrounded by parietal epithelial capsules and a connecting network of tubules that modify glomerular filtrate composition to aid in fluid and salt homeostasis and perform different metabolic functions, including vitamin D metabolism. The associated juxtaglomerular apparatus contributes to hormonal regulation of blood pressure. Nephrogenesis ends by the 36th week of gestation in humans (1, 2) and by the third day of life in mice (2, 3). The number of nephrons in individual kidneys may vary 10-fold, from ∼200,000 to over 2.5 million. Not all factors contributing to this variation are known, but prematurity, intrauterine growth retardation, and maternal starvation during gestation are strongly associated with low nephron endowment (4). Low nephron numbers contribute to significantly increased risk of chronic kidney disease (CKD), hypertension, and end-stage renal disease (5–8). As the postnatal kidney can repair but not replace nephrons, intervention-free survival depends on the initial nephron number. The burden of renal and cardiovascular disease is immense: More than eight million Americans are affected by CKD, and over 100,000 begin dialysis each year while awaiting renal transplantation, mainly secondary to obesity, diabetes, and hypertension. Thus, a greater understanding of the basic mechanisms regulating nephron endowment and any indication of a path to increasing nephron numbers are of marked scientific and clinical relevance.

Nephron progenitor cells (NPCs) in the metanephric mesenchyme (MM) coalesce into “cups” around ureteric bud (UB) tips. Reciprocal interactions between the UB and NPCs induce differentiation of two nephron primordia, one at each “arm pit” of the new branch (9, 10), via a mesenchymal–epithelial transition in response to UB-derived Wnt9b (11, 12). The NPCs in turn produce GDNF and other signals that induce and maintain UB branching (13). Six2, a transcription factor, maintains the NPC population; Six2 deletion in the NPC leads to the loss of nephrons due to premature differentiation (14–16). Differentiating NPCs extinguish Six2 and create nephrons through formation of an epithelial renal vesicle, which grows into a comma-shaped and then into an S-shaped body. Using an ex vivo transplantation assay, we have demonstrated that postnatal day 0 (P0) progenitors preferentially exit the stem cell niche and differentiate into nephrons relative to young NPCs. Importantly, P0 NPCs can remain in the niche if surrounded by E12.5 Fgf20⁺ progenitors (17) or can self-renew in culture nearly indefinitely (18–20), indicating tunable plasticity. NPC populations are heterogeneous, shifting with embryonic age toward a signature rich with ribosomal components, PolyA-binding proteins, and other transcripts suggestive of an increase in Mtor (mammalian target of rapamycin) activity (17). Mtor, a serine/threonine kinase, serves as a central regulator of cell metabolism, growth, proliferation, autophagy, and survival when in complex with Raptor (forming Mtor complex 1, Mtorc1) or Rictor (forming Mtorc2). Hamartin (coded by the Tuberous sclerosis 1 gene, Tsc1), a protein widely expressed in normal tissues, complexes with tuberin (Tsc2) (21, 22) and inhibits MtorC1. These findings motivated us to investigate the role of Mtor and Tsc1 in tuning NPC plasticity.

Herein we report that conditional deletion of Mtor in NPC profoundly disrupted nephrogenesis, and hemizygous deletion led to a significant reduction in nephron endowment. By contrast, conditional deletion of Tsc1 led to a lethal proximal tubular lesion. Surprisingly, hemizygous deletion of Tsc1 was associated with significantly (25%) increased nephron numbers due to the delayed cessation of nephrogenesis. Accordingly, NPCs hemizygous for Tsc1 displayed increased stem cell niche engraftment relative to age-matched wild-type cells, which could reflect a slower rate of aging. Finally, epistasis experiments demonstrated that the effects of Tsc1 deletion were not dependent on its interaction with the Mtor pathway. This suggests that a hamartin target is one variable affecting nephron endowment, representing a potential therapeutic approach to augment nephrogenesis.

Results

Mtor Activity Is Necessary for Normal Development of the Kidney.

To study the role of the Mtor pathway in nephron development, we deleted Mtor in NPCs using Six2^TGC+/tg (14) and the Mtor^f/f alleles (23). Reciprocally, we deleted Tsc1 using a similar strategy. Notably, the Six2^TGC+/tg allele is not benign: Fewer nephrons form in hemizygotes relative to wild type (see below). Since a large fraction of gravid dams carrying a litter with Tsc1 hemizygote kidneys delivered prematurely (before the Six2^TGC+/tg controls), we assigned age to all mice in this study as postcoitum days (PCD) based on the plug date.

One-quarter of the pups born to a Six2^TGC+/tg; Mtor^+/f male mated with an Mtor^f/f female died within 2 d and displayed severe generalized edema (Fig. 1A) not seen in other mice lacking nephrons (e.g., Notch mutants) (24). Genotyping confirmed complete loss of Mtor in these pups; no Mtor-null mice survived longer than 2 d. The kidneys of Mtor mutants were pale and dysplastic (Fig. 1B), with tubules in the parenchyma, paucity of glomeruli, and multiple cysts noted in the collecting ducts (Fig. 1 C and D). These results demonstrate that Mtor activity is essential for proper kidney development.

Fig. 1.

Mtor-deficient kidneys are severely impaired. (A) PCD19.5 mice whose kidneys lack Mtor appear edematous. (B) Mtor-deficient kidneys are small and pale. (C) H&E staining of a kidney cross-section (4× objective). (D) Magnified view with a 40× objective of the region shown in the dashed red box in C. Note multiple cysts and an individual glomerulus.

Although complete loss of Mtor led to dysplastic and dysfunctional kidneys, we next explored the hypothesis that Mtor signaling might still affect nephron endowment in hemizygotes. Loss of one Mtor allele led to a 24% decrease in the kidney/body weight ratio at PCD19.5 compared with Six2^TGC+/tg mice (0.47 ± 0.03 vs. 0.624 ± 0.02, P < 0.005), and to a 42% decrease in the ratio compared with wild-type mice (0.47 ± 0.03 vs. 0.9 ± 0.03, P < 0.005) (Fig. 2A). Accordingly, nephron counts were significantly (P < 0.005) lower in Six2^TGC+/tg; Mtor^+/f kidneys (1,140 ± 139) compared with both controls: Six2^TGC+/tg (2,850 ± 3) and wild type (5,100 ± 377.5) (Fig. 2B), reflective of impaired nephrogenesis. As Six2^TGC+/tg itself affects nephrogenesis, Six2^TGC+/tg mice serve as controls in this study.

Fig. 2.

Kidney size and nephron number are sensitive to Mtor dose. The weight (A) and nephron count (B and C) in controls (1), Six2^TGC+/tg (2), and Six2^TGC+/tg Mtor^+/f (3) kidneys of PCD19.5 pups (A and B) and adults (C). In B and C, each data point depicts the sum of nephrons in both kidneys from one individual. *P < 0.05.

We then assessed the contribution of postnatal nephrogenesis in Mtor hemizygotes. Six2^TGC+/tg; Mtor^+/f P30 kidneys had 22.5% fewer nephrons than Six2^TGC+/tg controls (13,382 ± 389 vs. 17,250 ± 194, P < 0.0001) and 59% fewer nephrons than wild-type kidneys (13,382 ± 194 vs. 27,967 ± 298, P < 0.0001) (Fig. 2C). Notably, at P200 Six2^TGC+/tg animals developed albuminuria (Fig. S1A). Despite reduced nephron numbers, glomerular function, assessed in aggregate by serum creatinine (Fig. S1B), blood urea nitrogen (BUN) levels (Fig. S1C), and albuminuria, was not further compromised in Six2^TGC+/tg; Mtor^+/f ∼ P200 mice relative to Six2TGC^+/tg controls.

Tsc1 Loss Is Lethal Due to Severe Renal Tubular Lesion.

While Mtor loss may compromise progenitor survival due to metabolic insufficiency, it is possible that elevated Mtor activity would accelerate the cessation of nephrogenesis. To explore this, we first deleted Tsc1 in NPCs. Although Six2^TGC+/tg Tsc1^f/f pups (henceforth, “Tsc1-null kidneys”) were born at the expected Mendelian frequencies and had a normal appearance at birth, they failed to survive longer than 48 h, consistent with renal failure. Tsc1-null kidneys produced urine throughout gestation, as demonstrated by amniotic fluid volumes similar to those in hemizygous littermates and the presence of urine in the bladder when pups were killed at PCD19.5 (Fig. S2A). Accordingly, and in contrast to Mtor-null kidneys, no glomerulopathy was identified in histological sections of Tsc1-null kidneys.

Strikingly, Tsc1-null kidneys displayed severe proximal tubular lesions: H&E staining showed that most cortical proximal tubules in Tsc1-null kidneys lacked a patent lumen compared with wild-type tubules, and tubular epithelial cells frequently exhibited a random orientation of the nuclei in contrast to the orderly basilar location in wild-type tubules (Fig. 3). To quantify the degree of tubular lesion, we counted the fraction of LTL⁺ tubules with a patent lumen in Tsc1-null and wild-type kidneys. All control LTL⁺ tubules had a patent lumen; ∼90% of LTL⁺ tissues in the Tsc1-null kidneys lacked a lumen (Fig. S2 B and C). Tubular lesions were detected as early as PCD15.5 (Fig. 3). To further characterize the phenotype, we performed periodic acid Schiff (PAS) staining on Tsc1-null and control embryonic kidneys to examine the brush border in cortical proximal tubule cells (Fig. 4A). The PAS stain generally showed no well-developed brush border along the apical surface and a lack of intracytoplasmic resorption droplets, both findings divergent from wild-type stains and indicative of loss of tubular cell polarity. Ultrastructural analysis confirmed the lack of a brush border in swollen proximal tubule cells with large eccentric nuclei. The normal number and the distribution of mitochondria implied that some forms of active transport may still exist. Glomerular morphology was unaffected by Tsc1 loss, consistent with the production of urine. The lesion was limited to the cortical proximal tubules; although it did not obstruct urine flow, it was severe enough to compromise survival (Fig. 4B). While we cannot rule out the possibility that loss of Tsc1 in other Six2-expressing tissues (e.g., cranial nerves and dorsal root ganglia) compromised the animals, at birth they were viable, motile, and feeding normally. We postulate that the mechanism most likely to impose rapid demise may be impaired tubular function driving the accumulation of electrolytes in the blood and leading to fatal arrhythmia.

Fig. 3.

Severe tubular lesions in kidneys with complete loss of Tsc1. H&E staining of kidneys from different embryonic stages (PCD15.5–PCD19.5) and from wild-type (1), Tsc1-hemizygous (2), and Tsc1-null (3) kidneys. Note the complete occlusion of tubular lumen by large cells that lost cellular polarity as shown by lack of brush borders and atypical nuclear location. Note also that the glomeruli appear normal in all three genotypes. Images were acquired with a 20× objective.

Fig. 4.

Complete Tsc1 loss in the kidneys led to the lack of a tubular brush border. (A) PAS staining of the cortical proximal tubules from different embryonic stages and wild type (1), Tsc1-hemizygous (2), and Tsc1-null (3) kidneys. Images were acquired with a 20× objective. (B) Transmission electron microscopy of proximal tubules from Six2^TGC+/tg Tsc1^f/f PCD19.5 mice. Note lack of a brush border (bb) in Tsc1-deficient kidney. Large tubular cells have enlarged nuclei (nuc) and high numbers of mitochondria (mit).

Tsc1-Hemizygous Kidneys Have More Nephrons than Controls in Two Independent Models.

We next studied the effect of removing one Tsc1 allele on nephron numbers. We confirmed that hamartin levels were reduced by Western blot analysis of purified NPCs (Fig. S3). Surprisingly, the kidney/body weight ratio of the Tsc1 hemizygotes (1.5% ± 0.05) was indistinguishable from that of wild-type mice (1.5% ± 0.05, P = 0.49) and was 22% higher than that of Six2^TGC+/tg mice in adulthood (1.22% ± 0.04, P < 0.05) (Fig. 5A). Furthermore, deletion of one copy of Tsc1 in kidney progenitors restored nephron numbers at P30 to the wild-type levels (Fig. 5B). The average nephron number in Six2^TGC+/tg mice was 17,180 ± 261, significantly lower than that in wild-type mice (25,769 ± 1,756, P < 0.001); by contrast, the average nephron number in Tsc1-hemizygous kidneys (26,640 ± 514, P = 0.66) was indistinguishable from wild-type kidneys and was significantly higher than in Six2^TGC+/tg kidneys (P < 0.0001). Overall, the loss of one Tsc1 allele in NPCs increases kidney size and nephron numbers in the Six2-Cre background without any apparent impact on viability or fecundity.

Fig. 5.

Tsc1 deletion increased kidney size and nephron number in an Mtor-independent manner. (A and B) The weight (A) and nephron count (B) in P30 wild-type (1) and Six2^TGC+/tg (2) controls and Six2^TGC+/tg Tsc1^+/f-hemizygous (3) kidneys. (C) Reduction in Mtor in Six2^TGC+/tg Tsc1^+/f Mtor^+/f (4) did not impact nephron count relative to Six2^TGC+/tg Tsc1^+/f (3). In B and C, each data point shown is the sum of nephrons in both kidneys of one individual. (D) Nephron counts in wild-type (1), Fgf20-Cre (5), and Fgf20-Cre Tsc1^+/f (6) kidneys. Glomeruli were counted from a single kidney removed from pups aged between P7 and P11. All litters were born at the same time as controls. *P < 0.05.

The loss of a Tsc1 allele may act only to correct the uncharacterized defect in Six2^+/TGC (a Bac insertion transgene). To address this possibility we used Fgf20^+/Cre mice (25), a knockin line increasing progenitor exit by a fraction due to reduced Fgf20/Fgf9 signaling (17, 18). Again, we observed a significant increase in nephron numbers in Fgf20^+/Cre; Tsc1^+/f mice, which became statistically indistinguishable from wild type (Fig. 5D). This line serves as an independent control confirming that the reduction in hamartin did not act solely by countering the defect introduced by the Six2^TGC transgene.

Increased Perdurance of Tsc1-Hemizygous Progenitors in the Niche.

Hamartin could act by increasing niche numbers, by increasing progenitor proliferation, by prolonging nephrogenesis, or by other mechanisms. Blinded observers quantified niches and progenitors by tomography-based counts on littermates with the genotypes wild type, Six2^TGC+/tg, and Six2^TGC+/tg; Tsc1^+/f on PCD15.5 and PCD19.5. Cell number per niche was indistinguishable in the three groups at both ages. Although they trended upward at PCD19.5 in Six2^TGC+/tg; Tsc1^+/f mice, niche numbers were statistically indistinguishable in Six2^TGC+/tg and Six2^TGC+/tg; Tsc1^+/f mice and were higher in wild-type mice on PCD15.5 and PCD19.5 (Fig. 6).

Fig. 6.

Hemizygous loss of Tsc1 is not associated with an increase in niche number or cells per niche in early and late gestation. (A–C) OPT-based niche (A and B) and cell (C) counts derived from confocal images of Six2 antibody-stained kidneys from wild-type (1) and Six2^TGC+/tg (2) controls and Six2^TGC+/tg Tsc1^+/f-hemizygous (3) kidneys at PCD15.5 (A) and PCD19.5 (B). Cell numbers per niche (C) were counted at PCD15.5 (1a, 2a, and 3a) and PCD19.5 (1b, 2b, and 3b). (D) The Tsc1-mediated increase in nephron number requires Raptor. Shown are nephron counts in wild-type controls (1), Six2^TGC+/tg (4), Six2^TGC+/tg Rap^+/f (5), and Six2^TGC+/tg; Rap^+/f; Tsc1^+/f (6) kidneys. Glomeruli were counted from a single kidney removed from pups aged between P7 and P11. All litters were born at the same time as controls. *P < 0.05; *** P < 0.0001.

To determine when NPCs were exhausted (typically by P3 in mice), we stained PCD22.5–PCD25.5 kidneys (P3–P6 in Six2^TGC+/tg mice) from the genotypes mentioned above for Six2 (MM) and cytokeratin 8 (UB). For all genotypes, Six2⁺ NPCs were identified on PCD22.5 but were absent on PCD24.5 (P5 in controls). At PCD23.5, controls (four wild type and two Six2^TGC+/tg) lacked Six2⁺ NPCs, while many niches in all seven Tsc1 hemizygotes contained Six2⁺ progenitors (Fig. 7A and Fig. S4). As an additional control, age-matched wild-type (Tsc1^+/f) littermates exposed to the same uterine environment as Six2^TGC+/tg; Tsc1^+/f pups were examined; no Six2⁺ niches were observed. Finally, Six2⁺ progenitors, and thus nephrogenesis, persisted an additional ∼24 h in the Fgf20^+/Cre; Tsc1^+/f kidney as well (Fig. S4C). Importantly, since none of the wild-type littermates had Six2⁺ progenitors at that age (Fig. 7B), the gain in nephrons is the result of the increased lifespan of NPCs hemizygous for Tsc1.

Fig. 7.

Tsc1 loss in renal progenitors leads to prolonged nephrogenesis and improved engraftment. (A) Magnified view of the nephron progenitor cells (arrow) in the periphery identified with anti-Six2 antibody in PCD23.5 (P4.5) Six2^TGC+/tg Tsc1^+/f kidneys. (B) Magnified view of kidney’s periphery from Tsc1^f/f (wild-type) littermates in which nephrogenesis has ended (the arrow points to background staining). Purple: cytokeratin 8; green: Six2. (Scale bars represent 50 µM.) (C and D) Niche engraftment success presented as the percentage of Six2^TGC+/tg; Tsc1^+/f Rosa^+/tdTomato and Six2^TGC+/tg; CAG-eCFP^+/tg cells in the total number of cells detected at the end of 4-d culture. (C) Comparison between Tsc1^+/f and Tsc1^+/− cells of different ages. (D) Comparison between Tsc1^+/f and Tsc1^+/− cells of the same age.

Tsc1-Hemizygotous Renal Progenitor Cells Have Higher Engraftment Rate.

The above observation suggests a possible shift in the rate of NPC aging. To address this, we investigated how progenitor engraftment was affected by reduced hamartin activity. Deletion of Tsc1 increases the activity of Mtor, which may be driving the “aged” signature in our single-cell RNA-sequencing (RNA-seq) analysis (17). If elevated Mtor activity caused poor engraftment of older progenitors, we anticipated that Tsc1-deficient progenitors would engraft poorly relative to same-age wild-type cells. Alternatively, if Mtor activity was not impeding engraftment, but hamartin activity was, we expected to see improved engraftment.

We used an ex vivo transplantation assay to differentiate between these hypotheses. The engraftment potential of NPCs declines with time, as older NPC tend to exit the niche and differentiate in larger numbers than younger cells. To determine the performance of PCD18.5 Tsc1^+/f progenitors engrafted relative to PCD12.5 progenitors, we FACS-purified PCD18.5 Six2^TGC+/tg; Tsc1^+/f; Rosa^+/tdTomato old progenitors (red+green, due to GFP expressed from the Six2^TGC+/tg allele) and PCD12.5 Six2^TGC+/tg; CAG-eCFP^+/tg young progenitors (blue+green). These cells were injected in a 1:1 ratio into the cap mesenchyme (CM) of a PCD12.5 recipient kidney explant. After 4 d the injected kidneys were fixed and stained for markers of MM and UB (Six2 and cytokeratin 8, respectively); we then quantified the number of red and blue Six2⁺ cells in z stacks acquired by confocal imaging (17). To normalize the data across experiments, we calculated the percentage of red and blue cells out of the total injected cells remaining in the niche at day 4. IMARIS software enabled analysis of the engraftment rate; the young cells engrafted better than old cells (86 ± 3% vs. 13 ± 3%, P < 0.0001, 11 sites injected on three dates) (Fig. 7C). Therefore, the reduction in Tsc1 dosage did not alter the negative effect of aging.

We next tested if Tsc1 loss impacted relative engraftment by comparing same-age cells of different genotypes. We FACS-purified PCD18.5 Six2^TGC+/tg; Tsc1^+/f; Rosa^+/tdTomato (red+green, Tsc1-hemizygote) and Six2^TGC+/tg; CAG-eCFP^+/tg (blue+green, control) progenitors and mixed them at a 1:1 ratio. When populations of PCD18.5 progenitors were injected into the CM of a PCD12.5 host kidney, Tsc1-hemizygote progenitors had a significant engraftment advantage compared with the controls (85 ± 5% vs. 15 ± 4%, P < 0.0001, 10 sites injected on five dates) (Fig. 7D). Therefore, although Tsc1 loss did not reverse the overall effect of aging on engraftment, decreased hamartin in NPCs does improve the cell engraftment rate.

The Effect of Tsc1 on Nephron Number and Proximal Tubules Is Agnostic to Mtor Dose but Requires both Alleles of Raptor.

The increase in nephron numbers could result exclusively from elevated Mtor activity, from reduced hamartin function(s) unrelated to Mtor signaling, or both. To differentiate between these possibilities, we examined nephron numbers in compound hemizygous (Six2^TGC+/tg; Mtor^+/f; Tsc1^+/f) mice. Removal of one Mtor allele resulted in reduced nephron numbers (Fig. 2). If this reflected enhanced inhibition of Mtor complex(es) due to an increased Tsc/Mtor ratio, we would expect nephron numbers in Six2^TGC+/tg; Mtor^+/f; Tsc1^+/f kidneys to return to the levels seen in Six2^TGC+/tg kidneys. Instead we found that Six2^TGC+/tg; Mtor^+/f; Tsc1^+/f and Tsc1-hemizygous kidneys were indistinguishable: At P30, the average nephron number in Six2^TGC+/tg; Tsc1^+/f mice (21,355 ± 1,819) is similar to the average number in Six2^TGC+/tg; Mtor^+/f Tsc1^+/f mice (19,325 ± 1,822, P = 0.46). Both were significantly higher than in Six2^TGC+/tg control mice (13,838 ± 755, P = 0.01) (Fig. 5C). We concluded that Mtor-independent hamartin activity limits nephron number and that the loss of one Tsc1 allele alleviated this restriction.

To test whether the tubular lesions of Tsc1-null kidneys can be reversed by decreasing Mtor activity in progenitors, we mated Six2^TGC+/tg; Mtor^+/f; Tsc1^+/f males with Tsc1^f/f females; 12.5% of the pups had the relevant genotype (Six2^TGC+/tg; Tsc1^f/f; Mtor^+/f), all of which died within the first 2 d of life regardless of their Mtor genotype. The fraction of tubules displaying lesions under light microscopy in Six2^TGC+/tg; Tsc1^f/f; Mtor^+/f kidneys was the same as noted in Tsc1-null kidneys (Fig. S2D). Thus, the frequency and severity of the tubular lesions caused by complete deletion of Tsc1 are not modified by the Mtor dose.

Finally, we asked whether the scaffold protein Raptor was involved in the Tsc1 phenotypes. Whereas Six2TGC^+/tg; Mtor^+/f kidneys contained significantly fewer nephrons than Six2TGC^+/tg kidneys (Fig. 2), nephron numbers in Six2TGC^+/tg; Rap^+/f (n = 4) kidneys were statistically indistinguishable from Six2TGC^+/tg (n = 5) kidneys (Fig. 6D). Although Raptor and the Mtor kinase form MtorC1, Raptor may be in excess such that the loss of free Raptor had no impact whereas the loss of Mtor did. If Raptor activities were mediated via MtorC1, we expected that reducing hamartin levels would elevate nephron number in Rap hemizygotes as it did in Mtor hemizygotes. Surprisingly, loss of Tsc1 in Six2TGC^+/tg; Tsc1^+/f; Rap^+/f kidneys (n = 5) had no impact on nephron numbers (Fig. 6D). These series of genetic experiments suggest a complex mechanism by which hamartin regulates nephrogenesis duration in mice, proposed in Fig. S6.

Discussion

Here we describe a role for hamartin in regulating NPC exhaustion during kidney development. Complete deletion of either Mtor or Tsc1 in mice NPCs led to postnatal lethality but apparently by different mechanisms. Mtor-deficient NPC failed to develop a functional kidney, reminiscent of the loss of Mtor in other stem cell compartments (26–31). The lethal phenotype reflects the vital role of the kidney in neonates: In utero, the filtering function of the kidney is replaced by the maternal placenta.

Complete deletion of Tsc1 spared the glomerulus but disrupted cortical proximal tubule morphology with loss of the brush border and minimal luminal space. The proximal tubules act in resorption and homeostasis of glucose, ions, and water; the loss of Tsc1 must have rendered them sufficiently impaired to cause lethality. This phenotype is reminiscent of one described previously (32) in which a later, mosaic deletion of Tsc1 in proximal tubules, glomeruli, and ducts resulted in severe, MtorC1-dependent polycystic kidney. Our results reflect an earlier, fully penetrant effect of hamartin that is Mtor independent.

Importantly, hemizygous deletion of Tsc1 increased nephron numbers in Six2^TGC+/tg Tsc1^+/f kidneys, restoring them to the level seen in the wild type. While environmental factors such as a high-protein diet (33) can increase nephron endowment in rats, few genetic manipulations can cause the same effect (34). We found that the number of nephron progenitors and niches were the same in Tsc1^+/− and controls at PCD15.5 and PCD19.5. However, Tsc1^+/− kidneys displayed prolonged nephrogenesis, with Six2⁺ NPCs detected at PCD23.5 in a fraction of the niches, while there was no detectable Six2 staining past P25.5/PCD22.5 in control mice. Extended nephrogenesis by even a fraction of a day can account for the observed increase in nephron number.

Extended nephrogenesis in Tsc1 hemizygotes was consistent with the effect of losing one Tsc1 allele on the behavior of individual NPCs. Our explant system permits simultaneous interrogation of genetically defined cells engrafted in genetically defined environments. Importantly, we found that older NPCs lacking one Tsc1 allele have significantly higher engraftment rates than wild-type NPCs of the same age, as is consistent with a prolonged nephrogenic period and, presumably, a slower rate of aging. Note that not all NPCs have to display these properties for a gain of 25% in nephron numbers. Because hamartin loss impacted both Six2^TGC+/tg and Fgf20^+/Cre NPCs, its impact implies a more general mechanism that compensates indirectly for the deficiency in these models.

In humans, tuberous sclerosis disease is caused by loss of function of TSC1 or TSC2. Kidney involvement manifests with angiomyolipoma (AML) and cysts, which contribute to bleeding and CKD, respectively. It is not known whether newborns with TSC1 loss have more nephrons at birth, as any gain may be obscured by the secondary disease complications. It is possible that an increased nephron number offers a physiological advantage, as cysts promote nephron loss. In AML, in addition to germline loss of TSC1 or 2, somatic loss of heterozygosity is frequently identified (35, 36). We found the complete deletion of Tsc1 in NPCs leads to severe and lethal tubular lesions, and others have demonstrated severe cysts in Tsc1-null kidneys (32). Therefore, the loss of heterozygosity may lead to similar, mTOR-independent tubular lesions and eventually, in an mTORC1-dependent manner, to cystogenesis.

Inhibition of mTOR by rapamycin revolutionized tuberous sclerosis management and was particularly helpful in treating hamartomas. However, we have shown that viability was not restored and that the severity of the tubular lesions in Tsc1-null kidneys was not ameliorated by Mtor deletion. In addition, it did not modify nephron numbers in Six2^TGC+/tg Tsc^+/− kidneys. It should be noted that Mtor deletion could affect MtorC1, MtorC2, and other Mtor-containing complexes (37). Since reduced hamartin levels increase nephron number in Mtor, Fgf20, and Six2^TGC hemizygotes but not in Raptor hemizygotes, we conclude that hamartin activities are Mtor independent but Raptor dependent. This is reminiscent of recent reports that free Raptor has mTOR-independent activity (38), which hamartin may be able to modulate (Fig. S6).

Several questions remain to be addressed by future investigation. The molecular mechanisms behind our observed Tsc1^+/− NPC niche retention and higher engraftment rates remain unknown, as are the hamartin targets involved. Hamartin and tuberin form a complex that negatively regulates the Rheb (Ras homolog enriched in brain) GTPase, the main regulator of mTORC1. Hamartin also interacts with other pathways that affect nephrogenesis: For example, it can modulate TGF-β/Smad2/3 signaling independently from tuberin (39). Smad signaling has an important role in kidney development via regulation of NPC exhaustion (20). However, we could not detect changes in the level of phospho-Smad in Tsc1 hemizygotes (Fig. S5). Our genetic analysis suggested that hamartin impacts nephron number independently of Mtor or the dosage-sensitive Fgf20 but that its activity does require Raptor. We speculate that hamartin may participate in balancing the impact of free Raptor on nephrogenesis. When hamartin levels are reduced, a net positive effect is gained (Fig. S6). The pathogenesis of Tsc1^−/−-associated tubular lesions remains to be fully characterized, and, last, the implications for human health need to be explored. At present, the field is focused on modulating Mtor; our study suggests Mtor-independent targets should also be considered.

In conclusion, we find that a reduction in hamartin has a net positive effect on nephron number by enabling progenitor expansion, acting independently of the mTor complex but requiring Raptor. This suggests distinct targets and offers a therapeutic path toward increasing nephron numbers in premature newborns. Over the long run, augmenting nephron numbers could also help address the high rates of adult renal and cardiovascular disease. These findings may also provide a model that will enhance our understanding of the initial steps in the process leading to AML and cyst formation in tuberous sclerosis patients.

Materials and Methods

Animals.

All mice were maintained in the Cincinnati Children’s Hospital Medical Center (CCHMC) animal facility according to the animal care regulations. The Animal Studies Committee of CCHMC approved the experimental protocols (IACUC2016-0022/0032). The following lines were used: Tg(Six2-EGFP/cre)1^Amc (herein Six2Cre^tg/+) (14), Rosa^tom/tom (40), CAG-eCFP^tg/tg (herein ECFP^tg/tg), Mtor^f/f (23), Fgf20^+/Cre (25), Tsc1^f/f (41), and Rptor^tm1.1Dmsa (herein Rap^f/f) (42). For NPCs in niche engraftment we crossed males of the genotypes Six2 Cre^tg/+ Rosa^tom/tom and Six2 Cre^tg/+ CAG-eCFP^tg/tg, Six2 Cre^tg/+ Mtor^f/+ or Six2 Cre^tg/+ Tsc1^f/+ with Mtor^f/f, Tsc1^f/f or CD1 females.

Nephron Count.

Nephron count was performed as described in ref. 18 and in Supporting Information. Two individuals, blinded to the genotypes of kidneys being scored, performed the counts independently, and results were averaged.

Histology.

Embryonic and adult kidneys were dissected in ice-cold PBS and fixed overnight in fresh 4% paraformaldehyde in PBS. Kidneys were embedded in paraffin. For overall morphology, tissue was stained in hematoxylin for 4 min and eosin for 1 min. Automated PAS staining was performed by the pathology core of CCHMC. Electron microscopy methods are provided in Supporting Information.

FACS Sorting of Renal Progenitor Cells.

Progenitors were sorted as described by Chen et al. (17). Detailed protocol can be found in Supporting Information.

Kidney Organ Culture.

Mouse metanephric organ cultures were performed as detailed in Supporting Information.

Transplantation of Renal Progenitor Cells.

The injection procedure is detailed elsewhere (17). Six recipient PCD12.5 CD1 kidneys were incubated in two Transwell chambers (Falcon) for 1–2 h before transplantation. The 1:1 (red/blue) cell mixture was injected in one or two niches in each kidney. Injected kidneys were then cultured for 4 d. Twenty-two niches were injected in this study.

Immunostaining, Confocal Imaging, and Image Analysis.

Detailed protocols and a list of the antibodies we used can be found in Supporting Information.

BUN Measurement.

BUN measurement was carried out by the Cincinnati Veterinary Laboratory (Cincinnati, OH).

Niche and Progenitor Cell Count.

Whole-mount and confocal microscopy and optical projection tomography (OPT) were carried out according to published protocols (43). Cell counts per niche (confocal microscopy) and niche counts (OPT) were performed as reported (43).

Statistical Analysis.

Two-tailed unpaired t tests were performed for the analysis of kidney size, nephron counts, and BUN levels. The percentages of red or blue cells out of the total number of cells coinjected in multiple sites on different days were combined to derive the average percentage and SD. A two-tailed Student t test was used to calculate the P value. Error estimates of all pooled data were calculated as SEM. The percentage of single cells vs. groups was calculated separately for cells of each color in individual injected niches. The data are presented using the GraphPad Prism version 7.