MicroRNAs are critical regulators of tuberous sclerosis complex and mTORC1 activity in the size control of the Xenopus kidney

Significance

The kidney is an essential organ to remove metabolic waste products and retain essential nutrients. To successfully execute these processes, the kidney and its functional unit, the nephron, must be correctly proportioned. Surprisingly, little is known about the processes governing kidney size. Here, we use the amphibian pronephric kidney as a paradigm to study the molecular mechanisms regulating size control. We show for the first time, to our knowledge, that growth factors belonging to the Insulin growth factor family are critically important in establishing kidney size. Moreover, we demonstrate that the strength of the signal is dependent on the presence of small noncoding RNAs termed microRNAs. These provide robustness to size control and ascertain that the kidney develops properly.

Abstract

MicroRNAs (miRNAs) are major posttranscriptional regulators of a wide variety of biological processes. However, redundancy among most miRNAs has made it difficult to identify their in vivo functions. We previously demonstrated that global inhibition of miRNA biogenesis in Xenopus resulted in a dramatically smaller pronephric kidney. This suggested that microRNAs play a pivotal role in organ size control. Here we now provide a detailed mechanistic explanation for this phenotype. We identified that the activation of the mechanistic target of rapamycin complex 1 (mTORC1) by Insulin and insulin-like growth factor (Igf) 2 is an important regulator in kidney growth, which in turn is modulated by microRNAs. Molecular analyses demonstrate that microRNAs set a threshold for mTORC1 signaling by down-regulating one of its core negative regulators, tuberous sclerosis 1 (Tsc1). Most importantly, this rheostat can be reprogrammed experimentally. Whereas knockdown of miRNAs causes growth arrest, concomitant knockdown of Tsc1 restores mTORC1 activity and proximal tubular size. Together, these data establish a previously unidentified in vivo paradigm for the importance of posttranscriptional regulation in organ size control.

The kidney is essential for excreting metabolic waste products and regulating electrolyte and water homeostasis in the body (1–3). In vertebrates, three successively more complex kidney structures, the pro-, meso-, and metanephros, have evolved. The metanephros is the adult kidney present in higher vertebrates such as humans and mice, whereas the mesonephros is the adult kidney of fish and amphibians. The pronephros is the simplest and earliest kidney form. Even though only a rudiment in mammals, its development is a prerequisite for the formation of the meso- and metanephric kidney (4, 5). Despite the different complexities of the three kidney types, their functional unit, the nephron, is organized very similarly. Many transcription factors, structural proteins, and signaling pathways have been shown to pattern the pro- and metanephros in an evolutionarily conserved way (6, 7).

The efficiency of the kidney is directly linked to the total number as well as the overall size of the individual nephrons. Size control of the kidney is established during embryonic development and even small disturbances can have a substantial impact on kidney function. Defective organ size control does not only affect children, but also carries health risks for adults (8). Studies in humans have demonstrated that individuals with smaller kidneys are more prone to suffer from acute kidney injury resulting in an increased number of patients undergoing dialysis or kidney transplants (9, 10). However, very little information is available on the molecular pathways regulating the size of the kidney.

In our previous study on the role of microRNAs (miRNAs) during Xenopus development, we observed that inhibiting miRNA biogenesis causes a reduction in the size of the pronephric kidney (11). miRNAs are small noncoding RNA molecules that regulate gene expression at the posttranscriptional level by binding to the 3′ UTR of target mRNAs (12). Here we now demonstrate that one of the major organ size control pathways, mechanistic target of rapamycin complex 1 (mTORC1) signaling, is directly regulated by miRNAs and that this crosstalk is pivotal in organ size control. We provide mechanistic data showing how miRNAs impinge on mTORC1 signaling and function as a rheostat, guaranteeing a properly sized kidney.

Results

The Size of the Xenopus Proximal Tubules Is Tightly Controlled.

We have previously demonstrated that the loss of miRNAs in early Xenopus development resulted in a range of pronephric kidney phenotypes (11). We further determined that part of the kidney phenotype was due to the loss of miR-30 miRNA family expression and the ensuing up-regulation of Lhx1, a key transcription factor in kidney development. However, miR-30 miRNAs could not explain one particular aspect of the phenotype observed upon inhibition of miRNA biogenesis, i.e., a smaller pronephric kidney. Thus, to more accurately assess the effect of miRNAs on kidney organ size control in vivo, we developed a method to accurately determine the size of individual nephron segments (in particular focusing on proximal tubules) and their number of mitotic cells (Fig. 1 A–A′′) (13). Using this approach, we observed a 2.4-fold increase in proximal tubular cell numbers between stages 37 and 42 in uninjected control embryos. However, Dicer antisense morpholino oligomer (xDicer-MO)-injected embryos only exhibited a modest 1.4-fold increase (Fig. 1B). This phenotype could not be attributed to a delay in pronephros formation, as it was still evident at stage 45 (Fig. S1 A and B), or to an increase in apoptosis, as xDicer-MO injected only exhibited 2 of 1,684 proximal tubular cells that were TUNEL-positive at stage 42 (compared with 0 in 1,762 in uninjected embryos, Fig. 1 D and D′). Instead, the percentage of mitotic events [i.e., phosphohistone H3 (pH3)-positive cells] was greatly reduced at stage 37 in Dicer-MO–injected embryos (Fig. 1C), suggesting that miRNAs regulate cell division.

Fig. 1.

Growth characteristics of the Xenopus proximal tubules. (A–A′′) Representative immunofluorescence analysis of proximal tubules at stage 42 marked by 3G8 (green); mitotic cells were labeled by pH3 (red), and nuclei were visualized by DAPI (blue). Asterisks indicate dividing cells in the proximal tubules. (B and C) Graphs depicting the total number of proximal tubular cells (B) and the percentage of pH3-positive cells (C) in uninjected controls and xDicer-MO–injected embryos at stages 37 and 42. The number of embryos analyzed is indicated in the bars. *P < 0.005; ns, not statistical significant change. (D and D′) TUNEL staining (red) of stage 42 embryos; images were counterstained with DAPI (blue) and ECL (green) to visualize nuclei and proximal tubules, respectively. Inset in D shows a DNase I-treated sample as positive control. (E) Total number of cells present in the proximal and distal tubule of a single pronephros during stages 34–46 [equivalent of 45–106 h postfertilization (hpf)]. Proximal tubular growth follows a sigmoidal trend (blue line, with an R² = 0.9274), whereas the distal tubular one is linear (black line, with an R² = 0.6914). Inset shows scheme subdividing proximal tubular expansion into three distinct phases, baseline (B, stages 34–37), growth (G, stages 38–42), and stationary (S, stages 43–46). The number of embryos analyzed is presented in Dataset S1. (F) Scatterplot depicting total number of cells and those in mitosis; each individual embryo used for the time course analysis in E from stages 34–46 is represented by a blue diamond. Note that the number of mitotic cells does not directly correlate with the overall cell numbers, but instead follows a polynomial trend (red line, with an R² = 0.8009). Black line indicates an extrapolation of a direct correlation considering only proximal tubules of 500 cells or fewer (with an R² = 0.7604). (G–U) Immunofluorescence staining of sectioned proximal tubules at different developmental stages using antibodies recognizing pErk1/2[T202/Y204], pAkt[Ser473], and pS6[Ser235/236]. Tubular structures are outlined by yellow dotted lines. Inset in I shows pErk1/2 staining in the somites as control for antibody activity. Insets in P and U show immunofluorescence analysis of total Akt and ribosomal S6, respectively. All error bars correspond to the SD.

In the frog Rana temporaria, growth of the pronephric kidney is characterized by a dramatic expansion phase (14). However, in Xenopus, the growth dynamics of proximal tubular expansion and the factors that regulate it are poorly described. To better understand the contributions of miRNAs, we performed an extensive characterization of normal tubule growth by analyzing kidneys from the onset of tubulogenesis (stage 34) until late tadpole stage (stage 46) (Fig. 1E and Fig. S1 C and D). To our surprise, proximal tubules displayed a very distinct growth pattern. Whereas proximal tubular cells did not extensively divide between stages 34 and 36, they expanded dramatically thereafter. After stage 43, proliferation slowed down and stopped once the final pronephric kidney size was attained. The accelerated proliferation was proximal tubular specific, because the distal tubular segment of the pronephros exhibited a rather linear growth pattern. This suggested that proximal tubule development occurs in a distinct pattern consisting of a baseline, growth, and stationary phase (Fig. 1E, Inset). To better understand the growth over time, we mapped the number of proximal tubular cells and their mitotic subpopulation for each individual counted pronephros in a scatterplot (Fig. 1F). Curve fitting analyses did not suggest a linear correlation between overall cell numbers and cells in mitosis. Instead, their relation was best modeled using a polynomial formula (red line in Fig. 1F). Whereas in smaller kidneys the size and the number of cells in mitosis directly correlated, larger kidneys exhibited an unproportionally low number of pH3-positive cells.

MAPK and Akt signaling are well-established proliferation regulators (15, 16). Immunofluorescence analyses using phospho-Erk1/2 and phospho-Akt as readouts for these pathways demonstrated that Akt, but not Erk1/2, was active in the pronephric kidney (Fig. 1 G–P). More importantly, Akt activity matched the growth curve; Akt phosphorylation was initially at baseline levels, increased at the beginning of the growth phase, and ceased upon reaching the stationary phase (Fig. 1 L–P). This effect was not due to differences in total protein levels, because Akt levels remained constant throughout kidney development (Fig. 1P, Inset and Fig. S2). Akt phosphorylation is a key event in regulating mTORC1 signaling (17–19), an important pathway in organ size control (20, 21). Whole-mount in situ hybridizations confirmed that mTORC1 signaling components are expressed in the pronephric kidney (Fig. S3). Moreover, activation of a downstream target of mTORC1 signaling, phosphorylation of the small ribosomal protein S6 (pS6) (22), followed the same time kinetics as that of Akt (Fig. 1 Q–U and Fig. S2).

mTORC1 Signaling Regulates Proximal Tubule Growth.

To functionally test whether mTORC1 signaling is involved in the growth control of the proximal tubules, we decided to block the pathway using three pharmacological inhibitors, LY294002, rapamycin, or Torin2 (Fig. 2 A–J). LY294002 is a PI3 kinase inhibitor, interfering with the upstream activation of mTORC1 by Insulin/Insulin-like growth factor (Igf) signaling, but will also interfere with other signaling pathways downstream of PI3 kinase (23); rapamycin and Torin2 directly inhibit mTOR, the core subunit of mTORC1, which functions as a serine/threonine protein kinase (24, 25). Embryos treated with these inhibitors from stage 34 until stage 42 (Fig. 2B) exhibited significantly reduced proximal tubules and a very low percentage of pH3-positive cells (Fig. 2 K–O and Fig. S4 A and B). Importantly, drugs did not interfere with the overall size of the Xenopus embryos, muscle formation, or the patterning of the pronephros (Fig. 2 P–R and Fig. S4 C–F′′).

Fig. 2.

mTORC1 controls proximal tubular expansion. (A) Schematic of the mTORC1 signaling pathway indicating the three pharmacological inhibitors (rapamycin, Torin2, and LY294002) and their targets. (B) Schematic of the experimental layout. (C–J) pAkt and pS6 immunofluorescence analysis of stage 42 embryos either untreated or treated with 4 μM rapamycin, 1.2 nM Torin2, or 2 μM LY294002 from stage 34 onward. Tubular structures are outlined by yellow dotted lines. (K–O) 3G8 whole-mount immunohistochemistry and quantification of proximal tubular cells in untreated controls and embryos with inhibited mTORC1 signaling at stage 42. (P–R) Comparison of embryo length and skeletal muscle formation using the monoclonal antibody 12/101 in the presence or absence of 4 μM rapamycin at stage 42. Number of embryos analyzed and the SD are indicated in the individual bars. *P < 0.005.

Based on the broad expression of mTORC1 signaling components (Fig. S3) it is likely that other organ systems are equally dependent on this signaling pathway. Thus, to complement the drug screen and to rule out off-target effects, we next wondered if we could identify the endogenous signal activating mTORC1. mTORC1 signaling integrates a range of external cues (26, 27). The fact that the growth of the proximal tubules could be inhibited by LY294002 (Fig. 2 N and O) prompted us to investigate Insulin and Igf signaling, an evolutionarily conserved pathway regulating overall body and organ size (20, 21, 28–30). Whereas Insulin and Igf receptors and the downstream signaling intermediate Irs1 were expressed throughout the embryo, two ligands, Insulin and Igf2 mRNAs, were specifically expressed in the proximal tubules of the kidney at stage 38 (Fig. 3 A–B′′ and Fig. S5). Moreover, their expression paralleled the stages, when we observed proximal tubular expansion. Insulin and Igf2 were detected as early as stage 34 and persisted at least until stage 42 (Fig. 3 A–B′′ and Fig. S5 I–L). Antisense MO-mediated knockdown of both Insulin and Igf2 (Ins/Igf2-MOs, Fig. S6) interfered with ligand-mediated autophosphorylation of the Insulin/Igf receptor at stage 40 and the phosphorylation of the downstream targets, Akt and S6, in the pronephric tubules at stage 42 (Fig. 3 C–H′). Morphologically, the knockdown resulted in a lower number of total as well as pH3-positive proximal tubular cells (Fig. 3 I–L). In fact, the kidneys of Ins/Igf2-MO–injected embryos resembled early stage pronephroi that did not initiate the accelerated growth phase and only a baseline level of growth persisted (red circles in Fig. 3L). We believe that this growth is likely downstream of other pathways such as Wnt and/or Fgf8 signaling (31, 32).

Fig. 3.

Insulin/Igf2 activate mTORC1 in the proximal tubules. (A–B′′) Whole-mount in situ hybridizations and paraplast sections thereof showing expression of Insulin or Igf2 in the proximal tubules of stage 38 Xenopus embryos. (C–D′) Phospho-Insulin/IGF receptor (pInsR/IgfR) immunofluorescence analysis of uninjected controls and embryos injected with Ins/Igf2-MOs at stage 40. C and D show pInsR/IgfR staining only; C′ and D′ show merged images with DAPI (blue) and ECL (green) to visualize nuclei and proximal tubules, respectively. Note the apical activation of the Insulin/IGF receptor present in control embryos, which is lost in tubules lacking Insulin and Igf2. (E–H′) pAkt and pS6 immunofluorescence analysis of uninjected controls and Ins/Igf2-MO–injected embryos at stage 42. E, F, G, and H show pAkt or pS6 staining only; E′, F′, G′, H′ show merged images with DAPI (blue) and 3G8 (green) to visualize nuclei and proximal tubules, respectively. Tubular structures in C, D, E, F, G, and H are outlined by yellow dotted lines. (I–K) 3G8 whole-mount immunohistochemistry and total proximal tubular cell counts of control embryos and Insulin/Igf2 morphants at stage 42. Number of embryos analyzed and the SD is indicated in the individual bars. *P < 0.005. (L) Stage 42 cell counts from uninjected (blue) and Ins/Igf2-MO–injected embryos (red) superimposed on the scatterplot of Fig. 1F (gray) demonstrate that impaired growth is caused by decreased numbers of mitotic cells. The fact that the pronephroi of the Insulin/Igf2 morphants shift to the right indicates that proliferation is not completely abolished but reduced to a baseline level.

miRNAs Act as a Rheostat in mTORC1 Signaling.

Based on the discovery of Insulin and Igf2 as crucial growth factors, we wondered whether there is a crosstalk between mTORC1 signaling and miRNAs, which could explain the reduced proximal tubule size upon loss of miRNAs (Fig. 1B). Immunofluorescence analysis of Dicer morphants revealed dramatically reduced pAkt and pS6 staining in the pronephric kidney (Fig. 4 A–D′). This suggested that mTORC1 signaling is impaired in the absence of mature miRNAs. To further substantiate this finding and to demonstrate that this effect was not a unique feature of the Xenopus pronephros, we used Lewis lung carcinoma-porcine kidney 1 (LLC-PK1) cells, a proximal tubular epithelial cell line, which has been shown to respond to Insulin (33–35). Extending on this observation, we could demonstrate that these cells responded to either Insulin or Igf2 in a time- and dose-dependent manner using phosphorylation of the ribosomal protein S6 as readout for mTORC1 activity (Fig. S7 A–D). In agreement with the Xenopus data, this effect was dependent on miRNAs and lipofecting an sDicer-MO blunted the response (Fig. 4 E–G′). More importantly, miRNAs generated a threshold for Igf2 signaling (Fig. 4 H and I). Whereas 5 μM sDicer-MO completely inhibited any response to Igf2, cells lipofected with 4 μM of the MO responded to 0.1 nM, but not 0.01 nM Igf2; at an even lower sDicer-MO concentration (3 μM) 0.01 nM Igf2 was now sufficient to trigger mTORC1 signaling. This effect could also be confirmed in another proximal tubular cell line, human HK-2 cells (Fig. S7 E–J).

Fig. 4.

miRNAs regulate mTORC1 pathway activity. (A–D′) pAkt and pS6 immunofluorescence analysis of uninjected controls and xDicer-MO–injected embryos at stage 42. A, B, C, and D show pAkt or pS6 staining only; A′, B′, C′, and D′ show merged images with DAPI (blue) and ECL (green) to visualize nuclei and proximal tubules, respectively. (E–G′) pS6 immunofluorescence comparing LLC-PK1 proximal tubular cells lipofected with scr-MO or sDicer-MO in the presence or absence of 0.1 μM Insulin or 0.01 nM Igf2. (H and I) Western blot analysis of LLC-PK1 cells lipofected with decreasing amounts of scr-MO or sDicer-MO in the presence of 0.01 nM or 0.1 nM recombinant Igf2 protein. α-Actin serves as a loading control.

Tuberous Sclerosis 1 Is a Primary miRNA Target.

To demonstrate that miRNAs directly modulate mTORC1 signaling we performed an extensive in silico analysis. By combining miRNA binding site prediction algorithms with miRNA array data, we identified several mTORC1 signaling intermediates that could be targeted by miRNAs in the kidney (Fig. 5 A and B). Because the proximal tubular xDicer-MO phenotype is characterized by reduced Insulin/Igf2 signaling (Fig. 4 A–D′), we focused on the two negative regulators of the mTORC1 pathway, the phosphatase and tensin homolog (Pten) and tuberous sclerosis (Tsc) 1; loss of miRNAs would cause increased translation or delayed mRNA degradation of these two genes and thus inhibit mTORC1 signaling (Fig. 5 L and L′). This hypothesis was tested in Xenopus by a combined knockdown with Dicer (Fig. 5C). The Pten-MO did not rescue the effects of xDicer-MO, even though the MO has been shown to be effective in Xenopus (36). Conversely, coinjection of the xTsc1-sMO restored proximal tubular expansion; the ratio of proximal tubular cells in xTsc1-sMO/xDicer-MO vs. xDicer-MO was statistically different from that of xTsc1-sMO vs. uninjected embryos (1.50 vs. 1.34 with a P-value of <0.05). Reactivation of mTORC1 signaling was confirmed by pS6 and pAkt immunofluorescence (Fig. 5 D–G and Fig. S7 K–N). Furthermore, Tsc1 protein was up-regulated in xDicer-MO–injected embryos (Fig. 5 H and H′). Whereas there is little detectable Tsc1 in the pronephric kidney of uninjected control embryos, the levels were clearly increased in the absence of miRNAs. This regulatory mode was also observed in LLC-PK1 cells, where concomitant knockdown of Dicer and Tsc1 reactivated mTORC1 signaling (Fig. 5I). Finally, the activity of miRNAs on Tsc1 mRNA was confirmed by luciferase assays. A 1-kb fragment of the Tsc1 3′ UTR, which contains the majority of miRNA binding sites, was sufficient to repress 50% of the reporter activity (Fig. 5 J and K and Fig. S7O). This repressive effect was abolished by interfering with miRNA biogenesis, which restored luciferase activity to control levels. To confirm that Tsc1 is targeted by miRNAs, we used miRNA mimics, double-stranded RNA molecules that imitate endogenous mature miRNA molecules. Both miR-19a/b and miR-130a/b have two predicted, overlapping binding sites in the Tsc1 3′ UTR (Fig. 5J). By adding miR-19b and miR-130b mimics individually or in combination to LLC-PK1 cells lipofected with sDicer-MO, the activity of the luciferase reporter was reduced (Fig. 5K and Fig. S7O). Finally, site-directed mutagenesis proved that this effect was dependent on the two predicted binding sites (Fig. S7O).

Fig. 5.

Tsc1 is a target of miRNA activity in proximal tubular cells in Xenopus and LLC-PK1 cells. (A) Flowchart of the in silico analysis identifying in vivo relevant miRNA/mRNA target pairs. (B) Table of conserved and mouse kidney-specific miRNAs potentially regulating mTORC1 pathway components. (C) Total number of proximal tubular cells of uninjected controls and MO-injected embryos at stage 42. Number of embryos analyzed and the SD is indicated in the individual bars. *P < 0.005; ns, a not statistical significant change. (D–G) pS6 immunofluorescence comparing the proximal tubules of Dicer, Tsc1, and Tsc1/Dicer double morphants. (H and H′) Tsc1 immunofluorescence of uninjected controls and xDicer-MO–injected embryos. (I) Western blot analysis of LLC-PK1 cells lipofected with sDicer-MO, sTsc1-sMO, and a combination of both MOs using pS6 and α-actin antibodies. (J) Schematic of the Luciferase reporter (pmirGLO-Tsc1) containing the most proximal portion of the Tsc1 3′ UTR. Black bars indicate the predicted miRNA-binding sites. The miR-19a/b (dark blue) and miR-130a/b (light blue) were identified in our in silico analysis and were selected for further analysis because of their dual binding sites. (K) Luciferase assays of LLC-PK1 cells transfected with the empty vector, pmirGLO-Tsc1 alone or in combination with lipofected sDicer-MO, miR-19b, or miR-130b miRNA mimics. SD is indicated in the individual bars. *P < 0.005. (L and L′) Model of mTORC1 signaling, in which the presence (L) or absence of miRNAs (L′) modulates the abundance of Tsc1.

Discussion

Together, these data support an in vivo model (Fig. 5 L and L′), where miRNAs restrict the activity of mTORC1 signaling to regulate the size of the pronephric kidney. They tightly control the expression levels of the negative regulator Tsc1 and thereby set a threshold for the upstream ligands, Insulin and Igf2. This assures that only high ligand concentrations can activate downstream signals and permit proximal tubular growth. Interestingly, we recently observed that perturbing endo- and exocytosis also results in smaller pronephric kidneys (37). Based on the work presented here, we now believe that this phenotype was caused by impaired secretion of Insulin and Igf2.

Our data do not rule out that other mTORC1 signaling components are regulated by miRNAs. In fact, the in silico analysis revealed that 9 of 13 genes in this pathway contain potential miRNA binding sites (Fig. 5B). Moreover, other groups have reported that Insulin and Igf receptors, as well as Pten, are targeted by miRNAs in vivo (38–40). Because Tsc1 is part of the tuberous sclerosis complex, the most proximal negative regulator of the pathway (26, 27), its up-regulation in the absence of miRNAs has most likely obscured any changes in the upstream signaling components. In fact, we believe that—as proposed for other systems (41–43)—the concerted regulation of multiple mTORC1 signaling components by miRNAs infuse robustness. This effect of Insulin/Igf signaling is probably not kidney specific, but is a more general principle (28–30). It will be regulated by the repertoire of miRNAs expressed in a given tissue. For example, mTORC1 signaling can be curbed by miRNAs like miR-19a/b or miR-130a/b in the epithelial cells of the kidney (Fig. 5K), but by a completely different set of miRNAs in mesenchymal cells. The plethora of miRNAs assures the tight regulation of mTORC1 activity in multiple tissues and cell types.

One intriguing aspect of this study is that the overall size of the pronephric proximal tubules in Xenopus always plateaus out at approximately the same number of cells (Fig. 1E). Even under conditions that show an increase in mitosis (e.g., in Tsc1 morphants) this number is not surpassed. Instead, cell division slows down earlier to maintain the final organ size. This mechanism seems to be evolutionarily conserved as suggested in earlier studies in R. temporaria (14). The underlying molecular mechanism is unknown. Based on the dramatic effects of miRNAs on Tsc1 expression levels, it is tempting to speculate that miRNA levels are responsible for the transition from the growth to the stationary phase. Such a concept will have major implications not only for organ development, but also for tissue repair and disease. Maintaining the correct number of cells in an organ is of the utmost importance and disruption of this rheostat can contribute to disease such as cancer or polycystic kidney disease.

Methods

Embryo Manipulations.

All animal experiments performed in this study were approved by the Cleveland Clinic Foundation Institutional Animal Care and Use Committee. Xenopus embryos were obtained by in vitro fertilization and maintained in 0.1× modified Barth medium (44) and staged according to Nieuwkoop and Faber (45). Antisense morpholino oligomers were obtained from GeneTools. The sequences of the antisense MOs used in this study were: 5′-TGC AGG GCT TTC ATA AAT CCA GTG A-3′ (xDicer-MO), 5′-GCA GAC ACT GCA TCC ATA GAG CCA T-3′ (Ins-MO), 5′-AAA TGT GTT GGC AGC TCC TCT GGG A-3′ (Igf2-MO), 5′-AAC ACC AAG ATA CCT GCT TGG GTC-3′ (xTsc1-sMO), and 5′-CGA ACT CCT TGA TGA TGG CGG TCA T-3′ (Pten-MO). Antisense morpholino oligomers were diluted to a concentration of 1 mM. The xDicer-MO and Pten-MO have been previously published (11, 36). The Ins-MO and Igf2-MO were tested using GFP fusion protein constructs (Fig. S6) following a strategy previously described (46). For all of the injections, a total of 4 nL of morpholino oligomer solution was injected into one of the two blastomeres at the two-cell stage. In the case of Insulin and Igf2, both MOs were always combined at a 1:1 ratio, which is referred to as Ins/Igf2-MOs. To identify the injected side, MOs were combined with 10 mg/mL fluorescein-labeled Dextran (D1820; Invitrogen). For the drug experiments with rapamycin, Torin2, and LY294002, embryos were treated from stage 34 onward and replaced every 24 h (Fig. 2B).

Whole-Mount in Situ Hybridization.

The procedure for the in situ hybridizations and the analysis by paraplast sectioning has been previously described (47, 48). To generate antisense probes, plasmids were linearized and transcribed as follows: pSK-β1-Na/K-ATPase–EcoRI/T7 (48), pCS2-Igf1–EcoRI/T7 (49), pCS2-Igf2–EcoRI/T7 (49), pCS2-Igf3–EcoRI/T7 (49), pCS2-Igf-Receptor–EcoRI/T7 (49), pCS2-Insulin–EcoRI/T7 (49), pCMV-SPORT6-Insulin-Receptor–SalI/T7 (clone ID: 7011987), pCMV-SPORT6-Insulin-related-Receptor–SmaI/T7 (clone ID: 6639745), pCMV-SPORT6-Irs1–EcoRV/T7 (clone ID: 4407025), pSK-mTOR –EcoRI/T7, pSK-Ncc–EcoRI/T7 (48), pSK-Nkcc2–SmaI/T7 (48), pSK-Raptor–BamHI/T7 (clone ID: XL072d22), pSK-Rictor–EcoRI/T7 (clone ID: XL021i17), pCMV-SPORT6-Sglt1-K–SalI/T7 (50), pSK-Tsc1–EcoRI/T7 (clone ID: XL062p12), and pSK-Tsc2–XhoI/T3 (clone ID: XL091f02).

Xenopus Immunofluorescence Analysis.

The determination of the total number of proximal tubular cells and their mitotic subpopulation was performed as previously described (13). All of the numbers correspond to a single pronephros. To determine the cell numbers in the distal tubule, the 4A6 antibody was used instead of 3G8 (51). Because 4A6 can only be detected from stage 38 onward, the growth curve lacks the first four time points. Moreover, to limit the counts to the distal tubule only, the convergence of the pronephros toward the midline was used as a morphological indicator. The protocol was slightly modified for the analysis of MO-injected embryos. In this case, the coinjected fluorescein-labeled dextran was visualized using an anti-fluorescein-horseradish peroxidase-coupled antibody (1:1,000; Roche) and the ImmPACT DAB kit (Vector Laboratories). It is noteworthy that we only selected MO-injected embryos, which showed clear one-sided labeling, for subsequent analysis. This stringent criterion eliminated ∼50% of the embryos, but was necessary to accurately assess kidney growth.

TUNEL staining was performed using the DeadEnd Fluorometric TUNEL system (Promega) and an Alexa-647–coupled anti-fluorescein antibody (Jackson Immunoresearch). Immunofluorescence analyses were performed on Dent’s fixed embryos using the following antibodies: 12/101 (Developmental Studies Hybridoma Bank), pAkt(Ser473) (4060; Cell Signaling Technology), phospho-Erk1/2(Thr202/Tyr204) (E7028; Sigma), phospho-S6 ribosomal protein (Ser235/236) (2211; Cell Signaling Technology), Akt 1/2/3 (sc-8312; Santa Cruz Biotechnology), S6 ribosomal protein (2317; Cell Signaling Technology), phospho-Insulin receptor (Tyr1150/1151)/IGF-I receptor (Tyr1135/1136) (3024; Cell Signaling Technology), and Tsc1 (sc-12082; Santa Cruz Biotechnology). Primary antibodies were visualized using Alexa-647–coupled secondary antibodies. Proximal tubules were visualized using either 3G8 immunostaining (13) or Erythrina cristagalli lectin (ECL; Vector Laboratories) and nuclei were counterstained with DAPI. To perform the studies as quantitative as possible, all immunofluorescence analyses comparing different treatments or time points were performed in parallel and images were taken at precisely identical settings.

Cell Culture Experiments.

LLC-PK1 and HK2 cells were maintained following standard culture conditions. For lipofection, an 80% confluent 10 cm dish was trypsinized, resuspended in 10 mL media, and 800 μL seeded into a six-well plate. After 10 h, the medium was replaced with glucose, glutamine, and FBS-free DMEM containing 5 μL/mL Endoporter (Gene Tools) and the indicated amounts of sDicer-MO (5′-GTT CTC ATT AGT ACC TGA TAT TTT C-3′), sTsc1-sMO (5′-CCC CAT ATT TGC TTG CTG GGC CAT T-3′), and scrambled MO (scr-MO, 5′-CTG GCA CAG CCA TGG TCG CAT ACG A-3′). Endoporter was included in all of the wells to avoid effects due to changes in the overall lipid composition of the cells. After 12 h, the media was replaced with 1× PBS to starve cells for an additional 24 h. Cells were stimulated with media (DMEM containing 1% glutamine, 1% penicillin/streptomycin, and 5% (vol/vol) FBS diluted in a ratio of 1:1 in 1× PBS) in the presence of the indicated amounts of bovine Insulin (I6634; Sigma-Aldrich) or recombinant human Igf2 (407245; EMD Millipore). Cells were washed briefly in 1× PBS and lysed in 200 μL Laemmli sample buffer containing 2-mercaptoethanol. Lysates were sonicated for 20 min, cleared by centrifugation, and processed for Western blot analysis using the antibodies against phospho-S6 ribosomal protein (Ser235/236) (2211; Cell Signaling Technology) and the α-actin (JLA20; Developmental Studies Hybridoma Bank).

For immunofluorescence analyses, cells were processed identically as described above. The only difference was that cells were seeded on glass cover dishes. They were fixed with 4% (vol/vol) PFA/1× PBS for 15 min at room temperature, washed with tris-buffered saline containing tween-20 at room temperature for 1 hr, and processed for immunofluorescence analyses using the anti-pS6 antibody and an Alexa-647–coupled secondary antibody. Slides were embedded in Prolong Gold with DAPI (Invitrogen) and visualized using a Leica DM5500B fluorescent microscope.

For the Dual-Luciferase assays, cells were lipofected with MOs as described above. After 24 h, cells were transfected with either 1 μg of pmirGLO, pmirGLO-Tsc1 [containing the most proximal part of the 3′ UTR (2700–4000 bp)] or pmirGLO-Tsc1-mut (in which the miR-19/130 miRNA binding sites have been mutated) in the presence or absence of 10 μL of the miR-19a and miR-130a miRNA mimics (Sigma-Aldrich) using the X-treme GENE HP DNA transfection reagent (Roche). Luciferase activity was determined after 24 h using the Dual Luciferase assay kit (Promega).

In Silico Analysis.

The 3′ UTRs of human and mouse insulin, igf2, insulin-receptor, igf2-receptor, irs-1, akt1, akt2, akt3, Pten, PraS50, Tsc1, Tsc2, mTOR, Tti1, deptor, rictor, S6 Kinase β1, and ribosomal protein S6 were screened for potential miRNA binding sites using two prediction algorithms, Target Scan (www.targetscan.org) and DIANA microT (http://diana.cslab.ece.ntua.gr/microT), using the default parameters. miRNAs with sites both in the human and the mouse genes were then assayed for their presence in the kidney using data from a miRNA expression array comparing E14.5, P1, P7, and P21 mouse kidneys (Miltenyi Biotec). Only those miRNAs that showed a differential expression profile and had multiple targets, were included in the final table shown in Fig. 5B.

Statistical Analysis.

All experiments were repeated at least three times using independent fertilizations from different females (in the case of Xenopus experiments) or independent transfections/lipofections (in the case of the cell culture studies). Error bars show SD. Data were analyzed by Student t test. An asterisk indicates a P value of <0.005 in all of the figures. Growth curves were generated by nonlinear regression using Prism 4 software.