Transforming Growth Factor β1-induced Apoptosis in Podocytes via the Extracellular Signal-regulated Kinase-Mammalian Target of Rapamycin Complex 1-NADPH Oxidase 4 Axis*

TGF-β is a pleiotropic cytokine that accumulates during kidney injuries, resulting in various renal diseases. We have reported previously that TGF-β1 induces the selective up-regulation of mitochondrial Nox4, playing critical roles in podocyte apoptosis. Here we investigated the regulatory mechanism of Nox4 up-regulation by mTORC1 activation on TGF-β1-induced apoptosis in immortalized podocytes. TGF-β1 treatment markedly increased the phosphorylation of mammalian target of rapamycin (mTOR) and its downstream targets p70S6K and 4EBP1. Blocking TGF-β receptor I with SB431542 completely blunted the phosphorylation of mTOR, p70S6K, and 4EBP1. Transient adenoviral overexpression of mTOR-WT and constitutively active mTORΔ augmented TGF-β1-treated Nox4 expression, reactive oxygen species (ROS) generation, and apoptosis, whereas mTOR kinase-dead suppressed the above changes. In addition, knockdown of mTOR mimicked the effect of mTOR-KD. Inhibition of mTORC1 by low-dose rapamycin or knockdown of p70S6K protected podocytes through attenuation of Nox4 expression and subsequent oxidative stress-induced apoptosis by TGF-β1. Pharmacological inhibition of the MEK-ERK cascade, but not the PI3K-Akt-TSC2 pathway, abolished TGF-β1-induced mTOR activation. Inhibition of either ERK1/2 or mTORC1 did not reduce the TGF-β1-stimulated increase in Nox4 mRNA level but significantly inhibited total Nox4 expression, ROS generation, and apoptosis induced by TGF-β1. Moreover, double knockdown of Smad2 and 3 or only Smad4 completely suppressed TGF-β1-induced ERK1/2-mTORactivation. Our data suggest that TGF-β1 increases translation of Nox4 through the Smad-ERK1/2-mTORC1 axis, which is independent of transcriptional regulation. Activation of this pathway plays a crucial role in ROS generation and mitochondrial dysfunction, leading to podocyte apoptosis. Therefore, inhibition of the ERK1/2-mTORC1 pathway could be a potential therapeutic and preventive target in proteinuric and chronic kidney diseases.

Podocyte damage is one of the key determinants of the majority of diabetic and non-diabetic glomerular diseases, leading to chronic kidney diseases and end-stage renal disease (1). Leakage of plasma proteins in the urine indicates the onset of renal diseases that are associated with podocyte injury (2)(3)(4). Recent evidence has shown that mammalian target of rapamycin (mTOR) 3 signaling activation is an important mediator of diabetic nephropathy in mice and humans (5,6). Increased mTOR activity has been reported in different glomerular diseases, and mTORC1 inhibition by rapamycin and everolimus shows beneficial effects against diabetic nephropathy, focal segmental glomerulosclerosis, minimal change disease, membranous nephropathy, etc. (7)(8)(9)(10). But contrasting evidence also indicates that mTORC1 inhibition by the immunosuppressive drug rapamycin leads proteinuria and podocyte apoptosis and develops primary focal segmental glomerulosclerosis in renal transplant recipients (11). Therefore, it might be important to investigate the detailed molecular mechanisms to answer this puzzle and identify therapeutic targets. mTOR is an atypical serine/threonine kinase belonging to the phosphatidylinositol 3-kinase-related kinase and is conserved from yeast to humans (12,13). It forms two functionally different multiprotein complexes: mTOR complex 1 (mTORC1), sensitive to rapamycin, and mTOR complex 2 (mTORC2), rapamycin-insensitive (14). Functional mTORC1 consists of mTOR, mLST8, and raptor, whereas mTORC2 includes mTOR, mLST8, mSIN1, and rictor. These signaling pathways regulate major physiological events, including ribosome biogenesis and protein translation (15), lipid homeostasis (16), and mitochondrial biogenesis (17). However, it is also implicated in various disease progressions, such as type 2 diabetes and obesity (14). mTORC1 activation results in phosphoryl-ation of ribosomal protein S6 via p70S6K and eukaryotic translation initiation factor 4E binding protein (4EBP), which potentiates the initiation of protein translation (18,19). However, mTORC2 regulates cell survival through phosphorylation of Akt (Ser 473 ) and cytoskeleton organization through phosphorylation of PKC␣ (20,21). Tuberous sclerosis 1 (TSC1)/ hamartin forms a heterodimer with tuberous sclerosis 2 (TSC2)/tuberin to negatively regulate mTORC1 activity. Many growth factors, including insulin, insulin-like growth factor, and TGF-␤ and nutrients such as amino acids exert their cellular input through modulation of mTOR complexes (22,23). Akt, ERK1/2, and AMP-activated protein kinase are well known intermediate signaling molecules that transmit extracellular signals to mTOR (15,24).
TGF-␤ is a multifunctional cytokine involved in many cellular processes, including migration, growth inhibition, invasion, extracellular matrix remodeling, epithelial-mesenchymal transition, and fibrosis (25). Increased activity of TGF-␤ plays critical roles in the progression of renal diseases in animal models and human renal diseases where podocyte loss is highly evident (26,27). Furthermore, accumulation of TGF-␤ and its pathogenic roles in primary focal segmental glomerulosclerosis, progressive glomerulosclerosis, and diabetic nephropathy have been reported (26,28). In addition, adenovirus-mediated TGF-␤1 gene transfer to kidney glomeruli leads to proteinuria in rats (29). It has been reported recently that podocyte-specific overexpression of TGF-␤ induces segmental glomerulosclerosis with dysfunction in endothelial cells and podocyte depletion (30). We have also published previously that TGF-␤1 induces podocyte apoptosis through transcriptional up-regulation of NADPH oxidase 4 (Nox4) by activation of the Smad pathway (31). There is also evidence that high glucose-mediated inactivation of AMP-activated protein kinase activates mTOR in a TSC2-dependent manner, which accounts for Nox4-mediated podocyte depletion in diabetic mice (32). However, the possibility of mTOR activation by TGF-␤1 in podocyte apoptosis has not been addressed so far. Moreover, the molecular mechanism of Nox4 regulation by the TGF-␤1-mTOR axis has not yet been explored in detail.
In this study, we observed that Nox4 is up-regulated by TGF-␤1 via rapamycin-sensitive mTORC1 activation because Nox4-induced oxidative stress and mitochondrial dysfunctions are ameliorated with rapamycin treatment or silencing of mTOR or its downstream target p70S6K. ERK1/2 plays an important role in TGF-␤1-induced mTOR activation and translational regulation of Nox4 protein. Finally, Smad signaling regulates TGF-␤1-mediated ERK1/2 phosphorylation in podocytes. Therefore, we suggest that inhibition of the ERK1/ 2-mTOR pathway could provide effective protection against podocyte depletion in glomerular kidney diseases.

Experimental Procedures
Cell Culture and Drugs-Immortalized mouse podocytes were obtained from the laboratory of Prof. Peter Mundel (Harvard Medical School) and cultured as described previously (31). Cells were grown on collagen-coated 100-mm dishes using lowglucose Dulbecco's modified Eagles medium supplemented with 5% FBS, 100 units/ml penicillin, and 100 g/ml strepto-mycin. Cell proliferation was achieved at 33°C in the presence of 20 units/ml mouse recombinant interferon ␥, and, for differentiation, these cells were thermoshifted to 37°C in the absence of interferon␥ for 10ϳ14 days. Before experiments, cells were changed with media containing 0.2% FBS for 24 h.
Quantitative Real-time PCR-To isolate total RNA, podocytes were grown on 100-mm dishes. Cells were washed twice with PBS before trypsinization. RNA was isolated according to the RNeasy kit (catalog no. 74134, Qiagen GmbH, Hilden, Germany). cDNA was synthesized from 0.5-1 g of RNA with a reverse transcription kit (Applied Bioscience, Foster City, CA) using oligo(dT) in a reaction volume of 20 -40 l according to instructions of the company. cDNA was subjected to real-time PCR to evaluate mRNA expression of Nox4 with the help of a sequence-specific primer (forward, 5Ј-CCACAGACCTGGA-TTTGGAT-3Ј; reverse, 5Ј-TGGTGACAGGTTTGTTGCTC-3Ј). The ␤-actin (forward, 5Ј-AAGAGCTATGAGCTGCC-TGA-3Ј; reverse, 5Ј-CACAGGATTCCATACCCAAG-3Ј) primer was considered a reference control. Experiments were performed in triplicate in a real-time PCR machine (7900HT, Applied Bioscience) using SYBR Green PCR Master Mix (catalog no. 204143, Qiagen GmbH). Data analysis was performed according to the ⌬⌬C T method (33).
Measurement of total Nox Activity-Total NADPH oxidase activity from podocyte lysate was measured as described previously (31). Podocytes were differentiated on 100-mm culture dishes for 13 days. Cells were kept on ice for 15 min, washed three times with ice-cold PBS, and collected from the dishes by scrapers in PBS containing protease inhibitor. Cells were cen-trifuged at 2000 rpm for 3 min at 4°C. The supernatant was discarded, and cell pellets were dissolved in lysis buffer (20 mM KH 2 PO 4 , 1 mM EGTA, and protease inhibitor mixture (pH 7.0)). Using a Dounce homogenizer, cell suspensions were homogenized with 100 strokes, and the homogenates were used freshly to measure total Nox activity. 50 g of homogenates was added to 50 mM phosphate buffer (pH 7.0) containing 1 mM EGTA, 150 mM sucrose, 5 M lucigenin, and 100 M NADPH. Photoparticles generated during the reaction were detected by a luminometer (Synergy 2, BioTek Instruments, Winooski, VT) for 10 min continuously and expressed as relative light units. Superoxide generation was expressed as relative light units per minute per microgram of protein.
TUNEL Assay-Podocytes were grown on 12-mm coverglasses coated with rat collagen and were differentiated for 10 days at 37°C without interferon ␥. Cells were exchanged with 0.2% FBS-containing medium 24 h before addition of TGF-␤1 and inhibitors. In the case of siRNA treatment, 0.2% FBS-containing medium was replaced 24 h after siRNA treatment. TUNEL was performed according to the instructions of the manufacturer using an ApopTag fluorescein in situ apoptosis detection kit (catalog no. S7110, Millipore). Cells were counterstained with DAPI. Signals from the nuclei of apoptotic cells merged with DAPI (blue) and fluorescein (green) were considered TUNEL-positive.
siRNA Transfection-Cells were grown on 6-well cell culture dishes and allowed to differentiate for 10 days. The culture medium was changed to antibiotic-free medium 1 day before siRNA transfection. Transfection was carried out using Dhar-maFECT 1 siRNA transfection reagent (catalog no. T-2001-03, Thermo Fisher Scientific) according to the directions of the manufacturer. Opti-MEM-reduced serum medium was used to prepare the siRNA-transfection reagent complex. The medium was changed 6 h after the addition of transfection reagent. Protein was isolated 72 h after siRNA treatment. SiGENOME SMARTpool siRNA duplexes for Smad2, Smad3, Smad4, and p70S6K were purchased from Dharmacon (Thermo Fisher Scientific Inc.). TSC2 siRNA was purchased from Bioneer Inc. (Daejeon, Korea). siRNAs for mTOR and non-targeting sequences (siControl) were synthesized from Bioneer Inc. The sequences for mTOR, TSC2, and non-targeting siRNA were 5Ј-CAUUCGCAUUCAGUCCAUA-3Ј, 5Ј-CAUCAUUGAAC-GACUACUU-3Ј, and 5Ј-CCUACGCCACCAAUUUCGU-3Ј, respectively. The sequence for mTOR siRNA was obtained from a published article where the efficiency of siRNA had already been tested (35).
Adenoviral Infection-Adenoviral mTOR constructs of wildtype (mTOR-WT), constitutively active (mTOR⌬), kinasedead (mTOR-KD), and control virus expressing firefly luciferase (Fluc) were gifts from Prof. C. Rhodes (University of Chicago, Chicago, IL). mTOR⌬ was generated by deleting amino acid sequences 2430 -2450 with 5-fold higher activity compared with wild-type mTOR assayed under in vitro conditions. mTOR-KD carried an Asp 2338 -to-Ala mutation rendering catalytic inactivation (36). Each mTOR construct was attached to an N-terminal AU1 tag, facilitating identification of exogenous mTOR expression. All four adenoviruses were amplified in HEK293 cells and purified by an ibiPure TM Adeno

ERK1/2-mTORC1-Nox4 Axis in TGF-␤1-induced Podocyte Apoptosis
adenovirus purification kit (catalog no. 60681) according to the instructions of the manufacturer. For all experiments, infections with a multiplicity of infection of 100 were used to infect podocytes as described previously (36).
Statistical Analysis-Experimental data are expressed as mean Ϯ S.E. All data shown here represent three or more independent experiments where n indicates the number of independent experiments. Statistical comparisons were made by Student's t test or one-way analysis if not stated otherwise, and p Ͻ 0.05 was considered to be significant.

TGF-␤1 Phosphorylates and Activates the mTOR Signaling
Cascade in Podocytes-Immortalized podocyte cell line expressing the differentiation marker synaptopodin at 37°C in the absence of interferon ␥ has been checked previously (31). Under similar culture conditions, podocytes were treated with TGF-␤1 (5 ng/ml) in a time-dependent manner to check the activation of mTOR signaling. Phosphorylation of mTOR at Ser 2448 was evident 5 min after TGF-␤1 treatment and reached its peak at 6 h. At the same time, p70S6K and 4EBP1, which are the downstream targets of mTORC1, were phosphorylated at Thr 421 /Ser 424 and Thr 37/46 , respectively, reached a maximum level at 6 h, and were reduced slightly at the 24-h time point. No statistically significant changes in expression of t-mTOR and t-p70S6K protein level were observed during this time interval. Additionally, Akt, which is a well known upstream regulator of the mTOR pathway, was also checked. Phosphorylation of Akt (Thr 308 and Ser 473 ) was increased by TGF-␤1, with a maximum peak at 6 h (Fig. 1, A and B). The TGF-␤ receptor I (TGF-␤RI)-specific inhibitor SB431542 (10 M) completely blocked the activation of mTOR signaling. TGF-␤1-mediated phosphorylations of mTOR, p70S6K, and 4EBP1 was prevented completely in the presence of SB431542. Inhibition of p-Smad2 and p-Smad3 by SB431542 was used as a positive control for TGF-␤ signaling activation in this experiment (Fig. 1, C-E).
mTOR Regulates Nox4 Protein Synthesis in Podocytes-We have described previously that TGF-␤1-induced Nox4 is responsible for TGF-␤1-mediated podocyte apoptosis (31). Different growth factors are known to regulate mTOR activity to increase protein synthesis in a cell type-dependent manner. Therefore, the role of mTOR on TGF-␤1-mediated Nox4 protein synthesis was evaluated here in podocytes. To confirm the role of mTOR in TGF-␤1-mediated oxidative stress by Nox4, mTOR was knocked down using siRNA (100 nM). Application of simTOR significantly reduced the mTOR protein level, as validated by Western blotting (Fig. 2A). Treatment with sim-TOR suppressed TGF-␤1-induced Nox4 protein up-regulation and completely recovered TGF-␤1-stimulated ROS generation in podocytes (Fig. 2, B and C). The protein translation inhibitor cycloheximide was applied to confirm the role of translational regulation as one of the possible mechanisms for increased Nox4 protein synthesis by TGF-␤1 application. TGF-␤1-mediated Nox4 protein synthesis was abolished completely by cycloheximide treatment (Fig. 2D). Using an adenoviral delivery system, we further evaluated the effect of mTOR on Nox4 protein expression through loss-of-function and gain-of-function mutations in mTOR. Overexpression of mTOR constructs was confirmed because t-mTOR expression was increased substan-FIGURE 1. TGF-␤1 activates mTOR signaling in podocytes. A, activation of mTOR was checked by TGF-␤1 treatment in podocytes. Fully differentiated cells were treated with 0.2% FBS for 24 h before treatment with 5 ng/ml TGF-␤1 at different time points. p-mTOR (Ser 2448 ), t-mTOR, p-p70S6K (Thr 421 /Ser 424 ), t-p70S6K, p-4EBP1 (Thr 37/46 ), p-Akt (Thr 308 and Ser 473 ), and t-Akt and ␤-actin were detected by Western blot analysis. B, densitometry analysis of p-mTOR, p-p70S6K, and p-4EBP1 from Fig. 1A (n ϭ 3). C, inhibition of TGF-␤1 and mTOR signaling using SB431542 (SB) was assessed. SB431542 (10 M) was applied 1 h before addition of TGF-␤1, and protein was isolated after 6 h of TGF-␤1 stimulation. Western blots for p-mTOR (Ser 2448 ), t-mTOR, p-p70S6K (Thr 421 /Ser 424 ), t-p70S6K, p-4EBP1 (Thr 37/46 ), p-Smad2, t-Smad2, p-Smad3, t-Smad3, and ␤-actin were performed. For Western blot analysis, 15 g of protein was loaded for each sample except ␤-actin (5 g). Experiments were performed three times or more. D and E, quantitative analyses of p-mTOR and p-p70S6K from C. Values are mean Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01. DECEMBER 25, 2015 • VOLUME 290 • NUMBER 52 tially and AU1 expression was clearly detected in all groups except the control group (Fluc). p70S6K phosphorylation was increased at the basal level as well as by TGF-␤1 treatment in podocytes expressing mTOR-WT and mTOR⌬, whereas p-p70S6K was reduced in the mTOR-KD group in comparison with the control (Fig. 2E). Overexpression of mTOR-WT increased TGF-␤1-stimulated Nox4 expression, ROS generation, and TUNEL-positive apoptotic nuclei compared with the control vector. Expression of mTOR⌬ further augmented the Nox4 level, ROS generation, and podocyte apoptosis. However, mTOR-KD completely abrogated the above changes under a TGF-␤1 stimulus (Fig. 2, F-H).

ERK1/2 Regulates TGF-␤1-induced mTOR Activation in a TSC2-independent Manner in Podocytes-Previous literature
has shown that phosphorylation of ERK1/2 regulates mTOR activation in different cell types (24,37,38). To elucidate the role of ERK1/2 in mTOR activation by TGF-␤1, ERK1/2 activation was inhibited by U0126 and PD184352, and 1 h of preincubation with these inhibitors significantly reduced mTOR and p70S6K phosphorylation by TGF-␤1 (Fig. 5, A-C). Next we wanted to identify the mechanism of mTOR phosphorylation by ERK1/2. PMA-induced TSC2 phosphorylation at position Ser 664 via ERK has been described by Ma et al. (38). Stimulation with TGF-␤1 significantly phosphorylated TSC2 (Ser 664 ) at 6 h, and this phosphorylation was inhibited completely by U0126 and PD184352. As a positive control, HeLa cells were treated with nocodazole (10 M) for 1 h to stimulate phosphorylation of TSC2 at Ser 664 (Fig. 5D). Another potential target of mTOR activation by TGF-␤1 is protein kinase B or Akt. Akt (Thr 308 ) phosphorylation was clearly detected after 1 h of TGF-␤1 treatment, and the downstream target TSC2 was phosphorylated simultaneously (Ser 939 and Thr 1462 ) by TGF-␤1. These phosphorylations were clearly inhibited by the PI3K inhibitor wortmannin in a dose-dependent manner (Fig. 5E). However, wortmannin did not show any significant inhibition of p70S6K phosphorylation mediated by TGF-␤1 stimulation in podocytes (Fig. 5F). To confirm the role of ERK1/2-p-TSC2 (Ser 664 ) in the activation of mTOR by TGF-␤1, TSC2 was knocked down by specific siRNA oligos (100 nM). The TSC2 protein level was reduced significantly by siRNA-mediated TSC2 knockdown. However, TGF-␤1-induced phosphorylation of p70S6K was not changed by silencing of TSC2 in podocytes. When U0126 and rapamycin were added to siTSC2-treated groups, phosphorylation of p70S6K by TGF-␤1 was diminished com-pletely. Moreover, insulin-induced phosphorylation of p70S6K was also not inhibited by TSC2 knockdown in podocytes (Fig.  5G).

Discussion
Here we report on TGF-␤1-stimulated Nox4 up-regulation through ERK1/2 and mTOR signaling, which accelerates oxidative stress-induced mitochondrial dysfunction and apoptosis in immortalized podocytes. Inhibition of either mTORC1 by low-dose rapamycin and p70S6K siRNA treatment or of ERK1/2 by U0126 and PD184352 protected podocytes from activation of caspase-3 and apoptosis. Previously, we have reported that TGF-␤1 selectively up-regulates the transcription of Nox4 mRNA by Smad2/3, resulting in an elevated mitochondrial Nox4 protein level, oxidative stress, mitochondrial dysfunction, and apoptosis in podocytes (31). This study identifies a distinct molecular mechanism of translational regulation of Nox4 protein that is considered another crucial step in TGF-␤1  DECEMBER 25, 2015 • VOLUME 290 • NUMBER 52 signaling to augment oxidative stress in podocytes. Activation of ERK1/2-mTORC1 signaling by TGF-␤1 occurs in a Smaddependent pathway, leading to an increased Nox4 protein level. We also show that mTOR activation is independent of ERK1/2 or Akt-targeted TSC2 phosphorylation. Taken together, we suggest that TGF-␤1-induced transcriptional regulation of Nox4 mRNA by the Smad-dependent pathway converges with translational regulation by the Smad-ERK1/2-mTORC1 pathway, accounting for oxidative stress-induced apoptosis in cultured podocytes.

ERK1/2-mTORC1-Nox4 Axis in TGF-␤1-induced Podocyte Apoptosis
Although the main function of mTORC1 is in growth control, recent findings suggest its overactivity in the progression of different glomerular diseases in human and animal models. In contrast, inhibition of mTORC1 also leads to pathogenic alterations such as podocyte apoptosis and proteinuria (11). Several pieces of evidence suggest that TGF-␤1 can stimulate mTORC1 and mTORC2 activation in a cell type-dependent manner (39 -42). We observed rapid mTOR phosphorylation in podocytes after 5 min of TGF-␤1 treatment. Within the same time window, Smad2, Smad3 (31), and mTOR signaling are activated when podocytes are chal-lenged with exogenous TGF-␤1 (Fig. 1A). This observation led us to identify a possible cross-talk between Smad signaling and mTOR activation in podocytes. Upon binding of TGF-␤1 to TGF-␤RII on the extracellular surface, it transmits a signal by phosphorylating TGF-␤RI, which activates either Smad-dependent or -independent signaling. TGF-␤RI kinase activity is a prerequisite for phosphorylation of Smad2 and 3, which can be inhibited completely by SB431542, a potent inhibitor of TGF-␤RI. However, Smadindependent activation of p38 MAPK in mesangial cells does not rely on TGF-␤RI kinase activity (43,44). In our results, phosphorylation of mTOR and p70S6K was inhibited completely by SB431542, implying that TGF-␤RI kinase activity is a mandatory to initiate mTORC1 (Fig. 1, C-E). It has been observed previously that TGF-␤1 stimulation induces apoptosis in undifferentiated podocytes cultured at 33°C (27). We also observed strong phosphorylation of p70S6K by TGF-␤1 treatment to undifferentiated podocytes and inhibition of caspase-3 activation by rapamycin, indicating conservation of the mTORC1-mediated apoptotic signal even in undifferentiated podocytes (data not shown). . Smad regulates phosphorylation of ERK1/2 and mTOR activation in podocytes. A, podocytes were transfected with a mixture of siSmad2 (50 nM) and siSmad3 (50 nM), and expression levels of p-ERK1/2, t-ERK1/2, p-p70S6K, t-p70S6K, t-Smad2, and t-Smad3 were checked by Western blot analysis after TGF-␤1 treatment for 6 h. ␤-actin was used as a loading control. B and C, densitometry analyses of p-ERK1/2 and p-p70S6K from A, respectively (n ϭ 3). D, cells were treated with 50 nM of siSmad4 and TGF-␤1 for 6 h. Western blot analysis was performed for p-ERK1/2, t-ERK1/2, p-mTOR, t-mTOR, p-p70S6K, t-p70S6K, Smad4, and ␤-actin. E and F, densitometry analyses of p-ERK1/2 and p-p70S6K from D, respectively (n ϭ 3). G, Western blot analysis showing that siRNA-mediated individual knockdown of Smad2 or Smad3 (100 nM each) did not recover TGF-␤1-induced p-ERK1/2 and p-p70S6K phosphorylation (n ϭ 3). H and I, densitometry analyses of p-ERK1/2 and p-p70S6K from G, respectively. All values are mean Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; n.s., not significant.
The roles of Nox4 in the effacement and apoptosis of podocytes have been studied in an experimental diabetic mice model and in a puromycin aminonucleoside-induced rat model of focal segmental glomerulosclerosis (32,(45)(46)(47). We also have reported, in a previous publication, that inhibition of Nox4 by the pharmacological inhibitor diphenyleneiodonium or RNA interference ameliorates TGF-␤1-mediated oxidative stress, mitochondrial dysfunction, and apoptosis in podocytes (31).
Here we observed that change in mTOR protein expression modulates the Nox4 level in podocytes. Gain-of-function (using mTOR-WT and mTOR⌬) and loss-of-function (using mTOR⌬ and simTOR) studies indicate that mTOR kinase activity is crucial for Nox4 protein expression, determining the ROS level and apoptosis (Fig. 2, B, C, and E-H). Additionally, pharmacological inhibition of mTORC1 by rapamycin suppressed TGF-␤1-stimulated Nox4 protein up-regulation and total Nox activity, dissipated ROS generation, and recovered from loss of mitochondrial membrane potential, caspase-3 activation, and apoptosis (Fig. 3, A and C-G). Moreover, silencing of p70S6K showed a reduction in Nox4 and ROS generation to the level similar to rapamycin treatment, confirming the role of mTORC1 in the regulation of Nox4 in podocytes (Fig. 3, H-K). However, the up-regulated Nox4 mRNA level was not changed at all when rapamycin was applied (Fig. 3B). This implies that mTOR may up-regulate translation mechanism or increase protein stability by inhibiting the proteasomal degradation of Nox4 under TGF-␤1 treatment in podocytes. To confirm this finding, cycloheximide, a protein translation inhibitor, was employed, and, indeed, pretreatment with cycloheximide blocked up-regulation of Nox4 protein, confirming that mTOR activation by TGF-␤1 regulates the translation of the Nox4 protein level in podocytes (Fig. 2D).
Evidence shows that activation of the ERK MAPK pathway induces pathogenic changes in podocytes. Gain-of-function mutation in the transient receptor potential C6 (TRPC6) channel facilitates a Ca 2ϩ influx that activates ERK signaling (48). Yu et al. (49) have shown that TGF-␤1-induced Smad3-ERK1/2-NF-B activation and Fyn-dependent TRPC6 phosphorylation lead to podocyte apoptosis. Interestingly, we observed that TGF-␤1 mediates biphasic phosphorylation of ERK1/2 in podocytes (Fig. 4A). Rapid phosphorylation of ERK1/2 at 5 min may be due to Smad-independent direct phosphorylation of ShcA by TGF-␤RI, which associates with Grb2 and Sos, leading to Ras-Raf-MEK-ERK pathway activation (50). The later peak at 6 h might result from a Smad-dependent increase in mRNA transcription and translation of ERK1/2 (51). However, we did not further investigate this mechanism in this study. We observedthatNox4proteinup-regulationdependsonERKphosphorylation because U0126 and PD184352 completely inhibited the increase in Nox4 protein level and total Nox activity in podocytes (Fig. 4, E-G). Similar to rapamycin treatment, ERK inhibition did not show any reduction in Nox4 mRNA level increased by TGF-␤1 (Fig. 4F), implying regulation of the Nox4 translation process by the ERK pathway. Moreover, ERK inhibition also partially recovered loss of the mitochondrial membrane potential, reduced ROS generation, and prevented caspase-3 activation and apoptosis (Fig. 4, H-K). These results clearly suggest that podocyte apoptosis is mediated via ERK1/2 activation and that a probable cross-talk exists between ERK and mTOR signaling in podocytes.
We considered three possible intermediate pathways that might regulate TGF-␤1-induced mTOR activation. These are the Akt, ERK, and Smad pathways, and they were explored to find the potential upstream regulator of mTOR in podocytes. Lamouille et al. (41) observed that translational regulation of epithelial-mesenchymal transition genes is mediated by TGF-␤-induced mTOR activation through PI3K and Akt. Akt is considered a prosurvival pathway, and TGF-␤1 is known to activate Akt through CD2AP in podocytes (52). Similar to the observation by Abe et al. (53), we detected TSC2 phosphorylation at Thr 1462 . In addition, p-TSC2 at Ser 939 by TGF-␤1 was also detected (Fig. 5E). Phosphorylation of these two positions on TSC2 by TGF-␤1 indicates the possible involvement of Akt in mTOR activation (15). Surprisingly, inhibition of PI3K-Akt signaling using wortmannin did not significantly suppress mTORC1 activation, as assessed by phosphorylation of p70S6K (Fig. 5F). A similar result was obtained when Akt was blocked directly by triciribine, an Akt inhibitor (data not shown). The significance of in vitro phosphorylation of TSC2 at Ser 939 and Thr 1462 by TGF-␤1 was hardly conceivable to us until now. However, these data clearly indicate that the PI3K-Akt pathway is not involved in TGF-␤1-mediated mTOR activation in podocytes. The role of Akt in mTOR phosphorylation in vivo may also be excluded by the observation that there is no increase but, rather, a decrease in p-Akt in the isolated glomeruli from db/db mice compared with db/ϩ mice (54). Notably, TGF-␤1 expression is increased in the kidney cortex of db/db mice (55).
There is evidence of Smad pathways regulating ERK1/2 and mTOR signaling. Phosphorylation of Smad3 stimulates mTOR activation to increase the synthesis of collagen in mesangial cells (42). In another study, sustained activation of Smad3 by TGF-␤1 suppresses deptor expression to activate mTORC1 and mTORC2 in a similar cell type (39). In renal fibroblasts, Smad3-Nox4-induced ROS generation phosphorylates ERK1/2, leading to myofibroblastic transformation (56). In podocytes, Smad3 phosphorylation has been reported previously to regulate ERK1/2 activation (49). However, their study has limitations because no inhibitory evidence of Smad3 for ERK1/2 activation is presented. Combined silencing of Smad2 and 3 or only Smad4 completely suppressed TGF-␤1-stimulated ERK1/2, mTOR, and p70S6K phosphorylation, indicating that Smad activation is upstream of the ERK-mTOR cascade (Fig. 6,  A-F). Surprisingly, knockdown of either Smad2 or Smad3 did not prevent TGF-␤1-induced ERK1/2 and p70S6K phosphorylation (Fig. 6, G-I). This observation shows the redundancy of Smad2 and 3 in ERK1/2-mTOR activation and differs from the transcriptional regulation of Nox4 mRNA (31). Therefore, Smad signaling simultaneously initiates two independent pathways. One pathway regulates transcription in the nucleus and increases the concentration of Nox4 mRNA, whereas the other one activates translation machineries in the cytoplasm to accelerate Nox4 protein synthesis in podocytes. Our data also indicate that ERK1/2 activation by TGF-␤1 varies between podocytes and mesangial cells and, therefore, should be considered carefully when developing new therapeutic strategies. Activation of mTOR by ERK1/2 plays a critical role in cancer progression, and it occurs through phosphorylation of TSC2 at Ser 664 . Inactivation of TSC2 by ERK1/2-mediated phosphorylation accelerates mTOR activation (37,38). ERK1/2 can also phosphorylate raptor through phosphorylation of the p90 ribosomal S6 kinase and, thereby, phosphorylates mTOR (57). Direct phosphorylation of raptor by ERK1/2 is also known to activate mTOR by phosphorylation (58). We identified that ERK1/2 is a potential upstream regulator of TGF-␤1-induced mTOR activation in podocytes because U0126 and PD184352 clearly inhibited mTOR and p70S6K phosphorylation (Fig. 5,  A-C). We also detected Ser 664 phosphorylation on TSC2 after TGF-␤1 treatment, which was completely inhibited by U0126 (Fig. 5D). Surprisingly, siRNA-mediated knockdown of TSC2 did not suppress TGF-␤1-stimulated p70S6K phosphorylation, indicating that TSC2 phosphorylation does not play any role in TGF-␤1-ERK1/2-mediated mTOR activation (Fig. 5G). However, when the ERK1/2 inhibitor U0126 was applied to siTSC2treated groups, p-p70S6K disappeared completely. These observations clearly point out that TSC2-independent activation of mTOR occurs through ERK1/2 in podocytes. Studies have also shown that TSC2 is not a critical target of Akt during Drosophila development (59) and that Akt can activate mTOR by regulating ATP and AMP-activated protein kinase activity (60). Therefore, the role of TSC2 phosphorylation by ERK1/2 in mTOR activation is questionable in podocytes. The detailed mechanism of mTOR activation remains to be elucidated, and we are working in that direction at present. However, in this manuscript, we clearly demonstrate the role of ERK1/2 in TGF-␤1-mediated mTOR activation, which occurs in a TSC2-independent pathway in cultured podocytes. It should also be noted that insulin-induced mTOR activation was not prevented by TSC2 knockdown, suggesting the existence of some other signaling routes transmitting a physiologic and pathophysiologic response across the cell membrane. A report has shown that insulin can bypass TSC2 phosphorylation in Akt-dependent mTOR activation through PRAS40 (proline-rich Akt/PKB substrate, 40 kDa) (61). A recent study has also shown that regulation of TSC2 by TGF-␤1 is cell type-specific. Reduction of TSC2 in renal tubular cells mediates EMT and tubulointerstitial fibrosis by TGF-␤1 (62). Therefore, although TSC2-mediated mTOR activation has been documented as a mediator of diabetic nephropathy (6, 32), TSC2-independented mTOR activation may participate in diabetes and/or other different types of nephritic diseases.
In summary, using a differentiated podocyte cell line, we demonstrated the role of the ERK1/2-mTORC1 signaling axis in TGF-␤-induced podocyte apoptosis. Activation of Nox4 by mTOR produces oxidative stress-induced mitochondrial dysfunction and apoptosis in podocytes. The TGF-␤1-induced ERK1/2-mTORC1 pathway accelerates the translation of Nox4 protein without participating in the transcription of Nox4 mRNA, which is primarily mediated by Smad2 and 3 (Fig. 7). Moreover, this mechanism of disease progression may not be restricted to Nox4 regulation and may be applicable to other TGF-␤ target genes in podocytes and other cell types. We also propose the beneficial effect of rapamycin against proteinuric kidney diseases caused by elevated TGF-␤1. Moreover, the prosurvival effect of the ERK pathway inhibitor could be another promising therapy against podocyte depletion, especially PD184352, which may need more attention because it is used frequently in cancer research and is already being tested in clinical trials (63).
Author Contributions-K. S. P. and R. D. conceived and coordinated the study and wrote the paper. R. D. prepared most of the figures. S. X., T. T. N., X. Q., S. K. C., and S. J. K. performed some experiments and provided technical assistance. E. Y. L. and S. K. C. contributed to data analysis and discussions. All authors reviewed the results and approved the final version of the manuscript. . The proposed mechanism of TGF-␤1-stimulated podocyte apoptosis. This schematic illustrates the mechanism of Nox4 up-regulation and oxidative stress-induced apoptosis by TGF-␤1 in podocytes. We have reported previously that activation of the transcription factors Smad2 and Smad3 increases the Nox4 mRNA level (dotted lines). Here we show that simultaneous Smad-dependent activation of the ERK1/2-mTORC1 axis augments translation of the elevated Nox4 mRNA pool to elevate Nox4 protein to a pathogenic level, resulting in oxidative stress and podocyte apoptosis (solid lines). A detailed explanation of this figure is provided under "Discussion." DPI, diphenyleneiodonium; NAC, N-acetylcysteine.