mTOR Regulation of N-Myc Downstream Regulated 1 (NDRG1) Phosphorylation in Clear Cell Renal Cell Carcinoma

The mechanistic target of rapamycin (mTOR) kinase is a component of two signaling complexes that are known as mTOR complex 1 (mTORC1) and mTORC2. We sought to identify mTOR-phosphorylated proteins that are differently expressed in clinically resected clear cell renal cell carcinoma (ccRCC) relative to pair-matched normal renal tissue. Using a proteomic array, we found N-Myc Downstream Regulated 1 (NDRG1) showed the greatest increase (3.3-fold) in phosphorylation (on Thr346) in ccRCC. This was associated with an increase in total NDRG1. RICTOR is a required subunit in mTORC2, and its knockdown decreased total and phospho-NDRG1 (Thr346) but not NDRG1 mRNA. The dual mTORC1/2 inhibitor, Torin 2, significantly reduced (by ~100%) phospho-NDRG1 (Thr346). Rapamycin is a selective mTORC1 inhibitor that had no effect on the levels of total NDRG1 or phospho-NDRG1 (Thr346). The reduction in phospho-NDRG1 (Thr346) due to the inhibition of mTORC2 corresponded with a decrease in the percentage of live cells, which was correlated with an increase in apoptosis. Rapamycin had no effect on ccRCC cell viability. Collectively, these data show that mTORC2 mediates the phosphorylation of NDRG1 (Thr346) in ccRCC. We hypothesize that RICTOR and mTORC2-mediated phosphorylation of NDRG1 (Thr346) promotes the viability of ccRCC cells.

To this end, we used a cancer-specific proteomic array to simultaneously detect approximately 440 cancer-associated proteins and phospho-proteins in ccRCC tumors and their pair-matched normal renal tissue controls (N = 22). This array contains probes for proteins that are phosphorylated by mTORC1 and mTORC2 [33]. The results showed that the level of mTORC1 signaling in ccRCC was not different from its levels in pair-matched normal renal tissue, given that the phosphorylation of primary mTORC1 targets (phospho-p70 S6 Kinase (Thr389), phospho-S6 (S235/236), and phospho-4E-BP1 (Thr37/46)) was not significantly increased in ccRCC samples relative to normal renal tissue. The phosphorylation of NDRG1 (on Thr346) showed the greatest statistically significant increase in ccRCC compared with noncancerous renal tissue. Further study showed the phosphorylation of NDRG1 (Thr346) in ccRCC cells was maintained by RICTOR and mTOR in complex 2. Treatments that inhibit RICTOR and mTORC2 reduce phospho-NDRG1 (Thr346) and cancer cell viability. Collectively, these data show RICTOR and mTOR in complex 2 maintain the phosphorylation of NDRG1 (Thr346) in ccRCC. We hypothesize that phospho-NDRG1 (Thr346) promotes the survival of cancer cells, and thus its targeting could potentially be efficacious for ccRCC.

Differentially Expressed Proteins and Phospho-Proteins in ccRCC
To characterize changes in signaling at the level of proteins and phospho-proteins in ccRCC, we carried out protein array studies using Reverse Phase Protein Array (RPPA) technology to measure the expression of approximately 440 cancer-associated proteins and phospho-proteins. Twenty-two clinically resected ccRCC samples and their 22 pair-matched normal renal tissue controls were compared. Kidney samples were obtained from treatment naïve male patients. Men are more prone (two-fold) to ccRCC, and the vast majority of the participants were male, which is why we chose to study 22 male samples in this study [5][6][7]. The median age at diagnosis was 57 years. The Body Mass Index (BMI) was greater than 30 for all patients, except for one patient who had a BMI of 29. Based on Fuhrman nuclear grading, three samples were grade I, 11 were grade II, four were grade III, and four were grade IV tumors. We identified 39 unphosphorylated and eight phosphorylated proteins, of which 20 were upregulated and 27 were downregulated by at least 50% in ccRCC relative to pair-matched normal renal tissue (Table 1). We identified statisticallys ignificant changes in protein expression that correspond with changes in cellular metabolism. Mitochondrial proteins (Mitochondrial Encoded Cytochrome C Oxidase, Succinate Dehydrogenase Complex Iron-Sulfur Subunit B, Cytochrome C Oxidase Subunit 4l1, and Succinate Dehydrogenase Complex Flavoprotein Subunit A) were significantly (p < 0.05) reduced, and proteins involved in glycolysis (Hexokinase 2, Pyruvate Kinase, and Lactate Dehydrogenase) were significantly (p < 0.05) increased in ccRCC compared with non-cancerous renal tissue (Table 1). NDRG1 showed the greatest fold increase (3.3-fold) in phosphorylation (on Thr346) in ccRCC relative to normal renal tissue ( Table 1). Considering that the phosphorylation of primary mTORC1 downstream targets was not different in ccRCC samples relative to pair-matched normal renal tissue, we focused on the regulation of NDRG1 phosphorylation on Thr346 in ccRCC cancer cells.

Total NDRG1 Protein and Phospho-NDRG1 (Thr346) Are Increased in ccRCC
Western blot experiments were performed to validate the protein array data. The results showed the level of total NDRG1 protein and phosphorylated NDRG1 (Thr346) was significantly increased in renal tumors (T) compared to pair-matched normal (N) renal tissue ( Figure 1). Notably, these ccRCC and pair-matched normal renal tissue samples (N = 7) were different from those used in the initial protein array study. These data show that the increase in total NDRG1 and phospho-NDRG1 (Thr346) occurs repeatably in ccRCC ( Figure 1).

Reducing RICTOR Causes a Reduction in Total NDRG1 Protein and Phospho-NDRG1 (Thr346) in ccRCC Cells
In glioblastoma, mTORC2 regulates the phosphorylation of NDRG1 (on Thr346) [34]. To test whether NDRG1 is a mTORC2 target in ccRCC, we sought to selectively disrupt the activity of mTORC2 without perturbing the activity of mTORC1. To this end, we transiently transfected 786-0 ccRCC cells with short interfering RNA (siRNA) corresponding to RICTOR or non-targeting (NT) siRNA. The 786-0 cell line was chosen because it is defective in VHL expression and thus is a good cell line model of ccRCC, given that VHL is inactivated in approximately 80% of ccRCC cases [20][21][22]35]. The 786-0 cell line also expressed high levels of high phospho-NDRG1 (Thr346) and total NDRG1 (Figures 2-4), which is consistent with the clinical samples ( Figure 1). RICTOR is an obligatory subunit of mTORC2 that is not a component of mTORC1 [12,15]. We assessed two different RICTOR siRNAs (Ri #1 and Ri #2) that were designed to target different regions of RICTOR mRNA. Using this approach, we reduced RICTOR protein by 40% (Ri #1) and 60% (Ri #2) compared with control NT siRNA ( Figure 2B). The decrease in RICTOR was associated with a statistically significant reduction in total NDRG1 protein (by~50%) and phosphorylated NDRG1 (on Thr346) (between~30 and 50%, for Ri #1 and Ri #2, respectively) relative to cells transfected with NT siRNA ( Figure 2B). The interaction between mTORC1 and mTORC2 is dynamic [36], such that inhibition of mTORC2 can cause an increase in the activity of mTORC1, as shown by the statistically significant increase in the phosphorylation of S6 (S235/236), which is a readout of mTORC1 activity, in cells transfected with Ri # 2 relative to control cells transfected with NT-siRNA ( Figure 2B). The increase in phospho-S6 (S235/236) in response to Ri #2 was not associated with an increase in total S6 protein ( Figure 2B). The reduction in total NDRG1 protein in response to RICTOR knockdown was not associated with a decrease in the levels of NDRG1 mRNA ( Figure 2C). These data suggest that RICTOR facilitates the phosphorylation of NDRG1 (Thr346) in part by promoting NDRG1 protein stability.

Total NDRG1 Protein and Phospho-NDRG1 (Thr346) Are Increased in ccRCC
Western blot experiments were performed to validate the protein array data. The results showed the level of total NDRG1 protein and phosphorylated NDRG1 (Thr346) was significantly increased in renal tumors (T) compared to pair-matched normal (N) renal tissue ( Figure 1). Notably, these ccRCC and pair-matched normal renal tissue samples (N = 7) were different from those used in the initial protein array study. These data show that the increase in total NDRG1 and phospho-NDRG1 (Thr346) occurs repeatably in ccRCC ( Figure 1). Quantification of (B) NDRG1 and C phosphorylated NDRG1 (Thr346) western blot data. Differences between ccRCC and normal tissue protein expression (N = 7) were analyzed using paired t-tests.

Reducing RICTOR Causes a Reduction in Total NDRG1 Protein and Phospho-NDRG1 (Thr346) in ccRCC Cells
In glioblastoma, mTORC2 regulates the phosphorylation of NDRG1 (on Thr346) [34]. To test whether NDRG1 is a mTORC2 target in ccRCC, we sought to selectively disrupt the activity of mTORC2 without perturbing the activity of mTORC1. To this end, we transiently transfected 786-0 ccRCC cells with short interfering RNA (siRNA) corresponding were analyzed using paired t-tests.

Concurrent Inhibition of mTOR in Complex 1 and Complex 2 Decreases the Levels of Phospho-NDRG1 (Thr346) in ccRCC Cells
To further investigate the regulation of NDRG1 by mTORC2, we treated 786-0 cells with increasing concentrations (2, 10, 50, 250, and 1000 nM) of Torin 2 for 24 h. Torin 2 is a recent mTOR inhibitor that concurrently inhibits mTOR in complex 1 and complex 2 [37]. At 2 nM, Torin 2 statistically significantly reduced the phosphorylation of NDRG1 at Thr346 by~60% ( Figure 3B). Torin 2 at higher concentrations induced greater reductions in phospho-NDRG1 (Thr346) (by~85% at 10 nM, >90% at 50 and 250 nM, and complete suppression at 1000 nM) ( Figure 3B). Torin 2 stimulated reductions in phospho-NDRG1 (Thr346) were not associated with reductions in total NDRG1 ( Figure 3B). As shown by the western blot, Torin 2 induced a shift in the total NDRG1 band towards higher mobility, which is consistent with a decrease in the phosphorylation of NDRG1 ( Figure 3A). As anticipated, because Torin 2 also inhibits mTOR in complex 1, the phosphorylation of the mTORC1 target, S6 (on S235/236), was significantly reduced in cells treated with Torin 2 ( Figure 3). Maximal suppression of NDRG1 phosphorylation (Thr346) was correlated with a statistically significant increase in NDRG1 mRNA in response to 250 and 1000 nM Torin 2 treatments ( Figure 3C). tion of S6 (S235/236), which is a readout of mTORC1 activity, in cells transfected with Ri # 2 relative to control cells transfected with NT-siRNA ( Figure 2B). The increase in phospho-S6 (S235/236) in response to Ri #2 was not associated with an increase in total S6 protein ( Figure 2B). The reduction in total NDRG1 protein in response to RICTOR knockdown was not associated with a decrease in the levels of NDRG1 mRNA ( Figure 2C). These data suggest that RICTOR facilitates the phosphorylation of NDRG1 (Thr346) in part by promoting NDRG1 protein stability.  We also assessed the effect of rapamycin on the phosphorylation of NDRG1 (on Thr346). Rapamycin is a potent inhibitor of mTORC1 that can partially inhibit mTORC2 after long-term exposure in some cell lines [36]. Rapamycin was applied to 786-0 cells at increasing concentrations (2, 10, 50, 250, and 100 nM) for 24 h. Western blot data showed rapamycin had no impact on the levels of total NDRG1 protein, phospho-NDRG1 (Thr346), or NDRG1 mRNA ( Figure 4A-C). Rapamycin reduced the phosphorylation of the mTORC1 target phospho-S6 (S235/236) ( Figure 4B). Given that Torin 2 is a potent inhibitor of both mTORC1 and mTORC2 that completely ablates the phosphorylation of NDRG1 (Thr346) (Figure 3), and that phospho-NDRG1 (Thr346) is resistant to rapamycin at all concentrations tested (Figure 4), these data indicate that mTOR in complex 2, but not complex 1, mediates the phosphorylation of NDRG1 (on Thr346) in ccRCC cells.

mTOR and RICTOR Promote the Viability of ccRCC Cells
Because mTOR promotes cell growth and survival, we hypothesized that mTOR is essential for ccRCC cell viability. We questioned if this mTOR effect is mediated through complex 1 or complex 2. We also probed whether the requirement for mTOR is higher in cancer cells than in non-cancerous cells. To test this hypothesis, 786-0 and HK2 cells were treated with rapamycin or Torin 2 at increasing (2, 10, 50, 250, and 1000 nM) concentrations for 3 days. The effect on cell viability was assessed by calcein fluorescence. The results showed that 786-0 and non-cancerous HK2 renal epithelial cells were resistant to the mTORC1 inhibitor rapamycin ( Figure 5A). There was no change in the percentage of live cells in response to rapamycin at all tested concentrations ( Figure 5A). Conversely, both cell lines were equally sensitive to Torin 2. In both cell lines, cell viability was significantly reduced by 35% at 10 nM, by 50% at 50 nM, and by 75% at 250 and 1000 nM in response to Torin 2 ( Figure 5B).
in phospho-NDRG1 (Thr346) (by ~85% at 10 nM, >90% at 50 and 250 nM, and complete suppression at 1000 nM) ( Figure 3B). Torin 2 stimulated reductions in phospho-NDRG1 (Thr346) were not associated with reductions in total NDRG1 ( Figure 3B). As shown by the western blot, Torin 2 induced a shift in the total NDRG1 band towards higher mobility, which is consistent with a decrease in the phosphorylation of NDRG1 ( Figure 3A). As anticipated, because Torin 2 also inhibits mTOR in complex 1, the phosphorylation of the mTORC1 target, S6 (on S235/236), was significantly reduced in cells treated with Torin 2 (Figure 3). Maximal suppression of NDRG1 phosphorylation (Thr346) was correlated with a statistically significant increase in NDRG1 mRNA in response to 250 and 1000 nM Torin 2 treatments ( Figure 3C). We also assessed the effect of rapamycin on the phosphorylation of NDRG1 (on Thr346). Rapamycin is a potent inhibitor of mTORC1 that can partially inhibit mTORC2 after long-term exposure in some cell lines [36]. Rapamycin was applied to 786-0 cells at increasing concentrations (2, 10, 50, 250, and 100 nM) for 24 h. Western blot data showed rapamycin had no impact on the levels of total NDRG1 protein, phospho-NDRG1 (Thr346), or NDRG1 mRNA ( Figure 4A-C). Rapamycin reduced the phosphorylation of the mTORC1 target phospho-S6 (S235/236) ( Figure 4B). Given that Torin 2 is a potent inhibitor of both mTORC1 and mTORC2 that completely ablates the phosphorylation of Given that concurrent inhibition of mTORC1 and mTORC2 by Torin 2 non-selectively reduced the viability of HK2 and 786-0 cells, we tested whether selective inhibition of mTORC2 would reduce the viability of 786-0 cells without reducing the viability of HK2 cells. To this end, cells were transfected with RICTOR siRNAs (Ri #1 and Ri #2). As noted, RICTOR is a required subunit of mTORC2 [15,19]. The results showed that RICTOR siRNAs significantly reduced (by~60%) the levels of RICTOR protein in HK2 cells ( Figure 5C). This, however, did not affect the viability of HK2 cells ( Figure 5D). We previously showed that the RICTOR siRNAs (Ri #1 and Ri #2) significantly reduced (by~50%) the RICTOR protein in 786-0 cells (in Figure 2). This was associated with a significant reduction in 786-0 cell viability compared with 786-0 cells transfected with non-targeting control siRNA ( Figure 5D).

Torin 2 Stimulates Apoptosis in 786-0 Cells
We asked whether the decrease in 786-0 cell viability in response to Torin 2 was correlated with an increase in apoptosis. The accumulation of cleaved PARP is a well-established readout of apoptosis [38]. This can be assessed by western blotting with an anti-cleaved PARP antibody. The cells (786-0) were treated with increasing (2, 10, 50, 250, and 1000 nM) concentrations of rapamycin or Torin 2 for 24 h. The results showed a statistically significant increase in the accumulation of cleaved-PARP in response to increasing concentrations of Torin 2 ( Figure 6B). The increase in cleaved PARP was correlated with a significant decrease in the levels of full-length PARP in response to Torin 2 ( Figure 6B). Etoposide is known to induce apoptosis in cancer cells [39], and it increased the levels of cleaved PARP in 786-0 cells as a positive control ( Figure 6A). The levels of cleaved-PARP were below the level of detection in 786-0 cells treated with rapamycin ( Figure 6A,C). The inability of rapamycin to induce apoptosis was anticipated, given that rapamycin did not reduce the percentage of viable 786-0 cells ( Figure 5A).  (Figure 3), and that phospho-NDRG1 (Thr346) is resistant to rapamycin at all concentrations tested (Figure 4), these data indicate that mTOR in complex 2, but not complex 1, mediates the phosphorylation of NDRG1 (on Thr346) in ccRCC cells. . Statistical analyses were performed using one-way ANOVA followed by Dunnett's multiple comparison tests.

mTOR and RICTOR Promote the Viability of ccRCC Cells
Because mTOR promotes cell growth and survival, we hypothesized that mTOR is essential for ccRCC cell viability. We questioned if this mTOR effect is mediated through complex 1 or complex 2. We also probed whether the requirement for mTOR is higher in cancer cells than in non-cancerous cells. To test this hypothesis, 786-0 and HK2 cells were treated with rapamycin or Torin 2 at increasing (2, 10, 50, 250, and 1000 nM) concentrations for 3 days. The effect on cell viability was assessed by calcein fluorescence. The results showed that 786-0 and non-cancerous HK2 renal epithelial cells were resistant to the mTORC1 inhibitor rapamycin ( Figure 5A). There was no change in the percentage of live cells in response to rapamycin at all tested concentrations ( Figure 5A). Conversely, both cell lines were equally sensitive to Torin 2. In both cell lines, cell viability was significantly reduced by 35% at 10 nM, by 50% at 50 nM, and by 75% at 250 and 1000 nM in response to Torin 2 ( Figure 5B).
Given that concurrent inhibition of mTORC1 and mTORC2 by Torin 2 non-selectively reduced the viability of HK2 and 786-0 cells, we tested whether selective inhibition of mTORC2 would reduce the viability of 786-0 cells without reducing the viability of HK2 cells. To this end, cells were transfected with RICTOR siRNAs (Ri #1 and Ri #2). As noted, RICTOR is a required subunit of mTORC2 [15,19]. The results showed that RICTOR siR-NAs significantly reduced (by ~60%) the levels of RICTOR protein in HK2 cells ( Figure  5C). This, however, did not affect the viability of HK2 cells ( Figure 5D). We previously showed that the RICTOR siRNAs (Ri #1 and Ri #2) significantly reduced (by ~50%) the RICTOR protein in 786-0 cells (in Figure 2). This was associated with a significant reduction in 786-0 cell viability compared with 786-0 cells transfected with non-targeting control siRNA ( Figure 5D).  Collectively, these data show that a reduction in phospho-NDRG1 (Thr346) that is greater than 90% is correlated with an increase in the apoptosis of 786-0 cells. The levels of phospho-NDRG1 (Thr346) dropped below 90% when Torin 2 was applied at 50, 250, and 1000 nM concentrations, which also stimulated the accumulation of cleaved-PARP ( Figures 3B and 6B) in 786-0 cells. Torin 2, when applied at 10 nM, induced an 85% reduction in phospho-NDRG1 (Thr346), which was correlated with a 35% reduction in cell viability without a correlating increase in apoptosis (Figures 3B and 6B). This suggests that the residual phospho-NDRG1 (Th346) in cells treated with 10 nM Torin 2 might contribute to the prevention of apoptosis. We, therefore, hypothesize that a greater than 90% reduction in phospho-NDRG1 (Thr346) is the threshold concentration to stimulate apoptosis in 786-0 cells.

Discussion
In this study, we characterized the regulation of NDRG1 phosphorylation on Thr 346 in ccRCC. The premise for this line of investigation stemmed from protein array results showing that this phosphorylation mark showed the greatest fold increase in ccRCC relative to pair-matched non-cancerous renal tissue (Table 1). We show that RICTOR and mTOR in complex 2 promote the levels of NDRG1 protein and phosphorylation of NDRG1 (Thr346) (Figures 2 and 3). Reducing RICTOR and dual inhibition of mTORC1 and mTORC2, but not selective inhibition of mTORC1, significantly reduced the levels of phospho-NDRG1 (Th346) in 786-0 cells (Figures 2-4). This was associated with a decrease in cancer cell viability that was correlated with an increase in apoptosis when phospho-NDRG1 (Thr346) was reduced by greater than 90% (Figures 5 and 6). Given that the loss of phospho-NDRG1 (Thr346) was correlated with an increase in cancer cell apoptosis in response to Torin 2, we hypothesize that phospho-NDRG1 (Thr346) promotes the survival of ccRCC cancer cells.
The anti-apoptotic effect of NDRG1 confers cancer cell resistance to hypoxia and anticancer drugs [34,[40][41][42]. NDRG1 protein expression is robustly induced in cancer cells in response to hypoxia [34,[40][41][42]. The induction of NDRG1 in response to hypoxia in cancer cells inhibits apoptosis and thus promotes cancer cell survival in tumors that have high Collectively, these data show that a reduction in phospho-NDRG1 (Thr346) that is greater than 90% is correlated with an increase in the apoptosis of 786-0 cells. The levels of phospho-NDRG1 (Thr346) dropped below 90% when Torin 2 was applied at 50, 250, and 1000 nM concentrations, which also stimulated the accumulation of cleaved-PARP ( Figures  3B and 6B) in 786-0 cells. Torin 2, when applied at 10 nM, induced an 85% reduction in phospho-NDRG1 (Thr346), which was correlated with a 35% reduction in cell viability without a correlating increase in apoptosis (Figures 3B and 6B). This suggests that the residual phospho-NDRG1 (Th346) in cells treated with 10 nM Torin 2 might contribute to the prevention of apoptosis. We, therefore, hypothesize that a greater than 90% reduction in phospho-NDRG1 (Thr346) is the threshold concentration to stimulate apoptosis in 786-0 cells.

Discussion
In this study, we characterized the regulation of NDRG1 phosphorylation on Thr 346 in ccRCC. The premise for this line of investigation stemmed from protein array results showing that this phosphorylation mark showed the greatest fold increase in ccRCC relative to pair-matched non-cancerous renal tissue (Table 1). We show that RICTOR and mTOR in complex 2 promote the levels of NDRG1 protein and phosphorylation of NDRG1 (Thr346) (Figures 2 and 3). Reducing RICTOR and dual inhibition of mTORC1 and mTORC2, but not selective inhibition of mTORC1, significantly reduced the levels of phospho-NDRG1 (Th346) in 786-0 cells (Figures 2-4). This was associated with a decrease in cancer cell viability that was correlated with an increase in apoptosis when phospho-NDRG1 (Thr346) was reduced by greater than 90% (Figures 5 and 6). Given that the loss of phospho-NDRG1 (Thr346) was correlated with an increase in cancer cell apoptosis in response to Torin 2, we hypothesize that phospho-NDRG1 (Thr346) promotes the survival of ccRCC cancer cells.
The anti-apoptotic effect of NDRG1 confers cancer cell resistance to hypoxia and anti-cancer drugs [34,[40][41][42]. NDRG1 protein expression is robustly induced in cancer cells in response to hypoxia [34,[40][41][42]. The induction of NDRG1 in response to hypoxia in cancer cells inhibits apoptosis and thus promotes cancer cell survival in tumors that have high levels of hypoxia [41]. In glioblastoma, alkylating anticancer drugs are less effective in cancer cells that express high levels of NDRG1 [34]. In this context, NDRG1 provides cancer cell resistance to apoptosis by stabilizing DNA repair proteins that counter the apoptotic-inducing activity of alkylating drugs in glioblastoma [34]. In hepatocellular carcinoma, siRNA-mediated reduction of NDRG1 sensitized cancer cells to doxorubicininduced apoptosis. Regarding a potential mechanism, phospho-NDRG1 (Thr346) binds, stabilizes, and increases the activity of proteins that have functional roles in DNA repair and apoptosis pathways [34,42]. The binding of NDRG1 to these regulatory proteins requires NDRG1 to be phosphorylated at Thr346 [34]. Further, the phosphorylation of NDRG1 (Thr346) promotes NDRG1 activity [43]. In line with these studies, we show that a decrease in phospho-NDRG1 (Thr346) is correlated with an increase in apoptosis in ccRCC cells treated with increasing concentrations of Torin 2. We hypothesize this is mediated by phosphorylated NDRG1 (Thr346), which promotes the activity/and or stability of proteins in ccRCC cells that promote cell viability.
NDRG1 was previously shown to be increased in ccRCC compared with paired normal renal tissue [44,45]. We now show that phospho-NDRG1 (Thr346) is increased and that this phosphorylation is maintained by mTORC2 and RICTOR (Figures 2-4). NDRG1 was identified to inhibit the proliferation of ccRCC cells in vitro and their growth as tumors in mice [44]. We show that the decrease in NDRG1 activity due to inhibition of its phosphorylation is correlated with an increase in apoptosis in ccRCC cells treated with Torin 2 (Figures 3 and 6). When viewed together, these studies show that NDRG1 has at least two roles in ccRCC: it inhibits cancer cell proliferation, and when phosphorylated, it inhibits apoptosis. This suggests that ccRCC tumors that have high levels of NDRG1 might grow slower and be more resistant to apoptosis.
Our results show that Torin 2 stimulates NDRG1 mRNA expression. As mentioned, hypoxia in the tumor microenvironment stimulates NDRG1 expression in cancer cells [34,[40][41][42]. This, in turn, promotes cancer cell survival [41]. Radiotherapy and alkylating drugs induce NDRG1 mRNA and protein in glioblastoma [34]. This, in turn, mediates tumor resistance to cancer therapy [34]. Di-2-pyridylketone thiosemicarbazones, which are iron chelators, stimulate an increase in NDRG1 transcription and protein expression [46,47]. This effect, in turn, has a tumor-suppressive effect on cancer cells [46,47]. The NDRG1 response to Torin 2 is different from other responses, given that increases in mRNA are not associated with an increase in total NDGR1 protein ( Figure 3). This is likely due to at least two factors. First, total NDRG1 is the sum of unphosphorylated and phosphorylated NDRG1. Torin 2 induces a profound decrease in phospho-NDRG1, and this reduces a large percentage of total NDRG1 ( Figure 3). Second, increases in NDRG1 mRNA must be translated into NDRG1 protein, and Torin 2, by inhibiting mTORC1, might interfere with the translation of NDRG1 mRNA into NDRG1 protein. The significant induction of NDRG1 mRNA occurred in response to high (250 and 1000 nM) concentrations of Torin 2 ( Figure 3). It is possible that the loss of phospho-NDRG1 (Thr346) at these concentrations induces an increase in NDRG1 mRNA as a compensatory mechanism to restore the loss of activated NDRG1. The induction of NDRG1 mRNA could also be a response to apoptosis in cells treated with 250 and 1000 nM Torin 2 ( Figure 6).
Our results suggest that RICTOR has two roles regarding the regulation of NDRG1. First, it promotes NDRG1 protein expression without changing the levels of NDRG1 mRNA ( Figure 2). This suggests RICTOR promotes the stability of the NDRG1 protein. Second, RICTOR facilitates the phosphorylation of NDRG1 (Thr346) as a required component of mTORC2. This is in line with publications by other investigators that have shown RICTOR has dependent and independent mTORC2 roles [48]. We hypothesize that RICTOR could stimulate NDRG1 protein stability by associating with NDRG1 in a complex that does not contain mTOR. This could explain why reducing RICTOR, but not inhibition of mTOR reduced the levels of total NDRG1 protein (Figures 2 and 3).
In conclusion, we set out to identify mTOR-phosphorylated proteins that are differentially expressed in renal tumors compared with normal renal tissue. The impetus for this question was based on genomic studies showing that approximately 20% of ccRCC cases had a mutation in at least one gene associated with the mTOR pathway [21]. The second reason to study mTOR signaling in ccRCC is that mTORC1 inhibitors are FDA-approved for the treatment of advanced ccRCC. Our proteomic array data shows the mTORC1 pathway is not statistically significantly increased in the 22 renal tumor samples we assayed relative to normal renal tissue. It is possible that a larger sample size might have detected increased mTORC1 signaling in a set of tumors. We also could not evaluate a correlation between the grade of the tumor and mTORC1 due to the sample size. We, however, did find a recurrent statistically significant increase in total NDRG1 and phospho-NDRG1 (T346) in the renal tumors that we analyzed (by western blot and protein array) ( Table 1 and Figure 1). We also established a correlation between decreases in phospho-NDRG1 (Thr346) and increases in cancer cell apoptosis in cells treated with Torin 2. We propose that new therapies that selectively target the kinase that phosphorylates NDRG1 (Thr346) in renal cancer cells might be efficacious in ccRCC.

Tissue Samples
De-identified frozen patient renal tissue samples were obtained from the biorepository bank at the Edwards Comprehensive Cancer Center at Cabell Huntington Hospital, Huntington, WV. Pathology established that renal tumors were clear-cell renal carcinoma and that pair-matched normal renal tissue was non-cancerous. De-identified samples were obtained with patient consent. Renal tissue samples were collected and immediately taken to the surgical pathology department for identification by a pathologist. Malignant and normal renal tissue were isolated using sterile techniques, with blade changes made between cutting tumor and non-tumor tissue. Isolated tissues were immediately flash-frozen in liquid nitrogen and stored in the liquid nitrogen vapor phase.

Reverse-Phase Protein Array
Frozen renal tissue samples were shipped on dry ice to the Reverse Phase Protein Array (RPPA) Core Facility at the University of Texas MD Anderson Cancer Center. Sample lysis and protein extraction in RPPA lysis buffer were carried out by the RPPA Core. Briefly, tissue extract was serially diluted two-fold for 5 dilutions (undiluted, 1:2, 1:4, 1:8; 1:16) and arrayed on nitrocellulose-coated slides in an 11 × 11 format to produce sample spots. Sample spots were then probed with antibodies by a tyramide-based signal amplification approach and visualized by a DAB colorimetric reaction to produce stained slides. Stained slides were scanned on a Huron TissueScope scanner to produce 16-bit tiff images. Sample spots in tiff images were identified and their densities were quantified by the Array-Pro Analyzer. Relative protein levels for each sample were determined by interpolating each dilution curve produced from the densities of the 5-dilution sample spots using a standard curve (SuperCurve) for each slide (antibody). SuperCurve is constructed by a script in R written by Bioinformatics. All relative protein level data points were normalized for protein loading and transformed to linear values.

Live-Dead Assays
Cells were seeded in quadruplicate in 96-well plates at 5000 cells/well and treated the following day with vehicle, rapamycin, or Torin 2, and the % of viable cells was determined by calcein-AM (2 µM) fluorescence (ex/em 495 nM/515 nM) and calculated relative to the vehicle group. The cells (786-0) were transfected with siRNAs (100 nM) and plated in 96-well plates at 5000 cells/well. HK2 cells were plated in 96-well plates and, the following day, transfected with siRNAs (100 nM). At 96 h post-transfection, the % of viable cells was determined by calcein-AM fluorescence and calculated relative to the control siRNA group. Calcein-AM was purchased from Thermo Fisher Scientific (L3224). Fluorescence was measured by a 96-well plate reader.

Statistical Methods
For RPPA (N = 22), log2 fold changes were determined by taking the log2 expression in the tumor minus the log2 expression in the normal for each patient and then taking the mean of those. Paired t-tests were conducted for each protein and phospho-protein, and then Benjamini-Hochberg was used to calculate adjusted p-values (false discovery rates) to account for multiple hypothesis testing. Differences between tumor and normal tissue protein expression from quantified western blot data were analyzed using paired t-tests using GraphPad Prism 9. Statistical analyses were performed using a two-tailed t-test or one-way ANOVA, followed by Dunnett's multiple comparisons test for the remaining in vitro studies. In vitro data are represented as the means with SEM of three independent experiments. Cell viability data are indicated as the means with SD of four replicates.