SGK1 (serum- and glucocorticoid-induced protein kinase 1) is a member of the AGC (protein kinase A/protein kinase G/protein kinase C) family of protein kinases and is activated by agonists including growth factors. SGK1 regulates diverse effects of extracellular agonists by phosphorylating regulatory proteins that control cellular processes such as ion transport and growth. Like other AGC family kinases, activation of SGK1 is triggered by phosphorylation of a threonine residue within the T-loop of the kinase domain and a serine residue lying within the C-terminal hydrophobic motif (Ser422 in SGK1). PDK1 (phosphoinositide-dependent kinase 1) phosphorylates the T-loop of SGK1. The identity of the hydrophobic motif kinase is unclear. Recent work has established that mTORC1 [mTOR (mammalian target of rapamycin) complex 1] phosphorylates the hydrophobic motif of S6K (S6 kinase), whereas mTORC2 (mTOR complex 2) phosphorylates the hydrophobic motif of Akt (also known as protein kinase B). In the present study we demonstrate that SGK1 hydrophobic motif phosphorylation and activity is ablated in knockout fibroblasts possessing mTORC1 activity, but lacking the mTORC2 subunits rictor (rapamycin-insensitive companion of mTOR), Sin1 (stress-activated-protein-kinase-interacting protein 1) or mLST8 (mammalian lethal with SEC13 protein 8). Furthermore, phosphorylation of NDRG1 (N-myc downstream regulated gene 1), a physiological substrate of SGK1, was also abolished in rictor-, Sin1- or mLST8-deficient fibroblasts. mTORC2 immunoprecipitated from wild-type, but not from mLST8- or rictor-knockout cells, phosphorylated SGK1 at Ser422. Consistent with mTORC1 not regulating SGK1, immunoprecipitated mTORC1 failed to phosphorylate SGK1 at Ser422, under conditions which it phosphorylated the hydrophobic motif of S6K. Moreover, rapamycin treatment of HEK (human embryonic kidney)-293, MCF-7 or HeLa cells suppressed phosphorylation of S6K, without affecting SGK1 phosphorylation or activation. The findings of the present study indicate that mTORC2, but not mTORC1, plays a vital role in controlling the hydrophobic motif phosphorylation and activity of SGK1. Our findings may explain why in previous studies phosphorylation of substrates, such as FOXO (forkhead box O), that could be regulated by SGK, are reduced in mTORC2-deficient cells. The results of the present study indicate that NDRG1 phosphorylation represents an excellent biomarker for mTORC2 activity.
The SGK (serum- and glucocorticoid-induced protein kinase) isoforms are members of the AGC (protein kinase A/protein kinase G/protein kinase C) family kinases and their activity is stimulated by growth factors and other agonists [1,2]. There are three isoforms of SGK (SGK1, SGK2 and SGK3) that are widely expressed and are reported to possess distinct, as well as overlapping, roles to other AGC kinase family members such as Akt (also known as protein kinase B) [1,2]. One of the best characterized processes controlled by SGK involves its ability to stimulate sodium transport into epithelial cells by enhancing the stability and expression of the ENaC (epithelial sodium channel) (reviewed in ). This is achieved by SGK phosphorylating the NEDD4-2 (neural-precursor-cell-expressed developmentally down-regulated 4-2) ubiquitin E3 ligase, promoting its interaction with 14-3-3 proteins, thereby preventing it from binding to ENaC and targeting it for degradation [4,5].
Insulin and growth factors stimulate activation of SGK1 as well as other AGC kinases, such as Akt, S6K (S6 kinase), RSK (ribosomal S6K) and PKC (protein kinase C) isoforms by enhancing the phosphorylation of these enzymes at their T-loop kinase domain residue (Thr256 in SGK1), as well as at a C-terminal non-catalytic residue, termed the hydrophobic motif (Ser422 in SGK1). Previous studies have established that the activation of SGK1 and S6K is dependent on the activation of PI3K (phosphoinositide 3-kinase) and the production of the second messenger PtdIns(3,4,5)P3 [6,7]. This induces phosphorylation of SGK1 and S6K at its hydrophobic motif, promoting the interaction with PDK1 (phosphoinositide-dependent kinase 1) [8,9]. PDK1 next activates SGK1 and S6K by phosphorylating the T-loop residues of these enzymes [8,9]. Consistent with this model, SGK1 and S6K activity is suppressed by inhibiting PI3K or by preventing PDK1 from interacting with the phosphorylated hydrophobic motif of SGK1 or S6K [10,11]. Activation and phosphorylation of Akt depends on PtdIns(3,4,5)P3 interacting with a PH (pleckstrin homology) domain on Akt that is not found on SGK1 or S6K. This induces a conformational change in Akt that enables PDK1 to phosphorylate the T-loop Thr308 residue [12–15]. PDK1 also contains a PH domain that binds with high affinity to PtdIns(3,4,5)P3 and other phosphoinositides which co-localizes PDK1 and Akt at the plasma membrane [15,16]. Binding of PDK1 to PtdIns(3,4,5)P3 is important for the activation of Akt, but not SGK1, as a knock-in mutation that prevented PDK1 binding to PtdIns(3,4,5)P3, suppressed the activation of Akt, but not SGK .
Recent work has established that complexes of the mTOR (mammalian target of rapamycin) protein kinase, termed mTORC1 (mTOR complex 1) and mTORC2 (mTOR complex 2), play a vital role in mediating the hydrophobic motif phosphorylation of Akt and S6K, as well as certain isoforms of PKC [18,19]. mTORC1 phosphorylates the hydrophobic motif residue of S6K (Thr389), whereas mTORC2 phosphorylates the hydrophobic motif of Akt (Ser473). mTORC1 consists of mTOR, raptor (regulatory associated protein of mTOR) and mLST8 (mammalian lethal with SEC13 protein 8; previously known as GβL). mTORC1 is activated by growth factors via a PI3K-regulated pathway, involving Akt-mediated phosphorylation of TSC2 (tuberous sclerosis complex 2) protein and PRAS40 (proline-rich Akt substrate of 40 kDa), which leads to the activation of the Rheb GTPase (reviewed in ). The activity of mTORC1 is also stimulated by nutrients such as amino acids via a distinct pathway involving the Rag GTPases binding to raptor [21,22]. mTORC1, and hence activation and phosphorylation of S6K, is also acutely inhibited by the macrolide rapamycin [23–25]. mTORC2 consists of mTOR, rictor (rapamycin-insensitive companion of mTOR; also known as mAVO3), Sin1 (stress-activated-protein-kinase-interacting protein 1; also known as mSin1 or MIP1), mLST8 [26–31] and protor (protein observed with rictor) [32–34]. Unlike mTORC1, mTORC2 is insensitive to acute rapamycin treatment, although prolonged incubation disrupts mTORC2 assembly in certain cells . mTORC2 is also activated by PI3K through an unknown mechanism, but, unlike mTORC1, its activity is not regulated by amino acids [26,27,29,36]. In the present study we explore whether mTOR complexes may play a role in regulating SGK1. Our results indicate that the hydrophobic motif phosphorylation, and hence activity of SGK1, is regulated by mTORC2.
MATERIALS AND METHODS
Protein G–Sepharose, glutathione–Sepharose and [α-32P]ATP were purchased from Amersham Biosciences. Pre-cast SDS polyacrylamide Bis-Tris gels and Lipofectamine™ 2000 were from Invitrogen. Tween 20 and dimethyl pimelimidate were from Sigma, and CHAPS and rapamycin were from Calbiochem. PI-103 was synthesized by Dr Natalia Shpiro at the University of Dundee. The wild-type control and mLST8-knockout MEFs (mouse embryonic fibroblasts) have been described previously  and were provided by Dr David Sabatini (Whitehead Institute for Biomedical Research, Cambridge, MA, U.S.A.). The wild-type control and Sin1-knockout MEFs have been described previously  and were provided by Dr Bing Su (Yale University School of Medicine, New Haven, CT, U.S.A.). The wild-type control and rictor-knockout MEFs have been described previously  and were provided by Dr Mark Magnuson (Vanderbilt University School of Medicine, Nashville, TN, U.S.A.). The muscle extracts from wild-type and SGK1-knockout mice have been described previously [17,38] and were provided by Dr Krishna M. Boini and Dr Florian Lang (Department of Physiology, University of Tübingen, Tübingen, Germany).
The following antibodies were raised in sheep and affinity-purified on the appropriate antigen: anti-mLST8 (S837B, 3rd bleed) was raised against the human full-length mLST8 protein expressed in Escherichia coli (used for immunoblotting); anti-mTOR [S683B, 2nd bleed; residues 2–20 of human mTOR LGTGPAAATTAATTSSNVS, used for immunoblotting in HEK (human embryonic kidney)-293 cells and immunoprecipitation]; anti-protor-1 (S020C, 3rd bleed) was raised against the human full-length protor-1 protein expressed in E. coli (used for immunoblotting); anti-raptor (S682B, 3rd bleed; residues 1–20 of human raptor MESEMLQSPLLGLGEEDEAD, used for immunoblotting and immunoprecipitation); anti-rictor (S654B, 3rd bleed; residues 6–20 of human rictor RGRSLKNLR-VRGRND, used for immunoblotting in HEK-293 cells and immunoprecipitation); anti-rictor (S274C, 1st bleed; residues 6–20 of mouse rictor RGRSLKNLRIRGRND, used for immunoblotting); anti-Sin1 (S8C, 1st bleed) was raised against the human full-length Sin1 protein expressed in E. coli (used for immunoblotting); and anti-SGK1 phosphorylated at Thr256 (S987, 1st bleed; residues 251–262 of human SGK1 NSTTSTpFCGTPE, used for immunoblotting). An anti-NDRG1 (N-myc downstream regulated gene 1) antibody (S276B, 2nd bleed) was made in sheep using recombinant GST (glutathione transferase)-fusion of full-length NDRG1 (used for immunoblotting). An antibody that recognizes NDRG1 phosphorylated at Thr346, Thr356 and Thr366 (S911B, 2nd bleed; termed pNDRG1 3xThr-P) was raised against the nonapeptide RSRSHpTSEG, whose sequence is common to all three sites (used for immunoblotting). Anti-Akt1 (S695B, 3rd bleed; residues 466–480 of human Akt1 RPHFPQFSYSASGTA, used for immunoblotting); anti-S6K (S417B, 2nd bleed; residues 25–44 of human S6K1 AGVFDIDLDQPEDAGSEDEL, used for immunoblotting); and anti-S6K2 (S469A, 3rd bleed; residues 476–495 of human S6K2, RPPSGTKKSKRGRGRPGR, used for immunoblotting) were also used. Anti-GST (S902A, 1st bleed) antibody was raised against the GST tag expressed from pGex4T (used for immunoblotting). An anti-mTOR antibody used for immunoblotting of mouse mTOR in MEFs was purchased from Santa Cruz Biotechnology (catalogue number sc-1549). For phospho-immunoblotting of the hydrophobic motif of SGK1(Ser422), we employed the Thr389 S6K antibody (catalogue number 9205) from Cell Signaling Technology which we previously demonstrated cross-reacted with the phosphorylated Ser422 of SGK1 . For the immunoblotting of endogenous SGK1 in Figure 6, we employed the phospho-SGK1 Ser422 antibody from Santa Cruz Biotechnology (catalogue number sc-16745-R) and the anti-SGK1 antibody from Upstate (catalogue number 07-315). The phospho-Akt Ser473 (catalogue number 9271), phospho-S6K1 Thr389 (catalogue number 9234), phospho-RSK Ser235 (catalogue number 4856) and total RSK (catalogue number 2217) antibodies, used for immunoblotting, were also purchased from Cell Signaling Technology. Secondary antibodies coupled to HRP (horseradish peroxidase) used for immunoblotting were obtained from Thermo Scientific.
Tissue culture, immunoblotting, restriction enzyme digests, DNA ligations and other recombinant DNA procedures were performed using standard protocols. DNA constructs used for transfection were purified from E. coli DH5α using the Qiagen plasmid Mega or Maxi kit according to the manufacturer's protocol. All DNA constructs were verified by DNA sequencing, which was performed by the Sequencing Service, School of Life Sciences, University of Dundee, Dundee, Scotland, U.K., using DYEnamic ET terminator chemistry (Amersham Biosciences) on Applied Biosystems automated DNA sequencers. MEFs were cultured with additional non-essential amino acids and 1% sodium pyruvate solution.
The following buffers were used: Tris lysis buffer [50 mM Tris/HCl (pH 7.5), 1 mM EGTA, 1 mM EDTA, 0.3% CHAPS, 1 mM sodium orthovanadate, 10 mM sodium-β-glycerophosphate, 50 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.27 M sucrose, 0.15 M NaCl, 0.1% 2-mercaptoethanol, 1 mM benzamidine and 0.1 mM PMSF]; buffer A [50 mM Tris/HCl (pH 7.5), 0.1 mM EGTA and 0.1% 2-mercaptoethanol]; Hepes lysis buffer [40 mM Hepes (pH 7.5), 120 mM NaCl, 1 mM EDTA, 0.3% CHAPS, 10 mM sodium pyrophosphate, 10 mM sodium-β-glycerophosphate, 50 mM sodium fluoride, 0.5 mM sodium orthovanadate, 1 mM benzamidine and 0.1 mM PMSF]; Hepes kinase buffer [25 mM Hepes (pH 7.5) and 50 mM KCl]; TBS (Tris-buffered saline)-Tween buffer [50 mM Tris/HCl (pH 7.5), 0.15 M NaCl and 0.1% Tween 20]; and sample buffer [50 mM Tris/HCl (pH 6.8), 6.5% (v/v) glycerol, 1% (w/v) SDS and 1% (v/v) 2-mercaptoethanol].
HEK-293 cells, HeLa cells, MCF-7 cells or MEFs were cultured and treated as described in the legends to the Figures. Following treatment, cells were rinsed twice with ice-cold PBS and then lysed using Tris lysis buffer. Whole-cell lysates were centrifuged (18000 g at 4 °C for 20 min), supernatants were removed and stored in aliquots at −80 °C until required.
Plasmids and transfection
A full-length cDNA encoding human SGK1 from an infant brain library was obtained from the IMAGE (Integrated Molecular Analysis of Genomes and their Expression) Consortium (clone ID42669). DNA constructs encoding full-length SGK1 or the truncated form of SGK1 (ΔNSGK1) lacking the N-terminal 60 residues, as well as the point mutations in the SGK1 gene, have been described previously . For transfection studies, typically ten 10-cm-diameter dishes of HEK-293 or HeLa cells were cultured and each dish was transfected with 5–10 μg of the indicated plasmids using the polyethylenimine method . MCF-7 cells were transfected with 5 μg of the indicated plasmids using Lipofectamine™ 2000 as described by the manufacturer. MEFs were transfected with 10 μg of the indicated plasmids using the MEF Nucleofector® Kits 1 and 2 as described by the manufacturer (Amaxa). Electroporation was performed using the A-23 nucleofector program (Amaxa).
Immunoprecipitation of endogenous mTOR complexes for immunoblot analysis
HEK-293 cells or MEFs were lysed in Hepes lysis buffer. Lysate (1–4 mg) was pre-cleared by incubating with 5–20 μl of Protein G–Sepharose conjugated to pre-immune IgG. The lysate extracts were then incubated with 5–20 μl of Protein G–Sepharose conjugated to 5–20 μg of the indicated antibodies or pre-immune IgG. All antibodies were covalently conjugated to Protein G–Sepharose using dimethyl pimelimidate. Immunoprecipitations were carried out for 1 h at 4 °C on a vibrating platform. The immunoprecipitates were washed four times with Hepes lysis buffer, followed by two washes with Hepes kinase buffer. The immunoprecipitates were resuspended in 30 μl of sample buffer (not containing 2-mercaptoethanol), filtered through a 0.22 μm Spin-X filter, and 2-mercaptoethanol to a concentration of 1% (v/v) was added. Samples were subjected to electrophoresis and immunoblot analysis as described below.
GST-pulldown of transfected SGK1 for immunoblot analysis
At 36 h post-transfection, HEK-293 cells, HeLa cells, MCF-7 cells or MEFs were lysed in Tris lysis buffer. Lysate (0.5–1 mg) was affinity-purified on glutathione–Sepharose. Incubations were carried out for 1 h at 4 °C on a vibrating platform. The resultant precipitates were then washed twice with Tris lysis buffer containing 0.5 M NaCl, followed by two washes with buffer A. The immunoprecipitates were resuspended in 20 μl of sample buffer, filtered through a 0.22 μm Spin-X filter, and samples were subjected to electrophoresis and immunoblot analysis as described below. Amounts of GST–SGK1 were assessed as described below using LI-COR scanning methodology.
mTOR complexes kinase assays
HEK-293 cells or MEFs were freshly lysed in Hepes lysis buffer. Lysate (1–4 mg) was pre-cleared by incubating with 5–20 μl of Protein G–Sepharose conjugated to pre-immune IgG. The lysate extracts were then incubated with 5–20 μl of Protein G–Sepharose conjugated to 5–20 μg of either mTOR, rictor, raptor or pre-immune IgG. All antibodies were covalently conjugated to Protein G–Sepharose. Immunoprecipitations were carried out for 1 h at 4 °C on a vibrating platform. The immunoprecipitates were washed four times with Hepes lysis buffer, followed by two washes with Hepes kinase buffer. For raptor immunoprecipitates used for phosphorylating S6K1 and SGK1 (Figures 4E and 4F), the buffer for the initial two wash steps included 0.5 M NaCl to ensure that we obtained optimal kinase activity . GST–ΔNSGK1 (4 μg of total protein, of which 35% is GST–SGK1) was isolated from serum-deprived HEK-293 cells incubated with PI-103 (1 μM for 1 h). GST–S6K1 (0.5 μg) was purified from serum-deprived HEK-293 cells incubated with rapamycin (0.1 μM for 1 h). mTOR reactions were initiated by adding 0.1 mM ATP and 10 mM Mg2+ in the presence or absence of GST–ΔNSGK1 (4 μg of total protein, of which 35% is GST–SGK1) or GST–S6K1 (0.5 μg of total protein). Reactions were carried out for 60 min at 30 °C on a vibrating platform and stopped by the addition of SDS sample buffer. Reactions were then filtered through a 0.22 μm Spin-X filter and samples were subjected to electrophoresis and immunoblot analysis.
SGK1 kinase assay
HEK-293 cells or MEFs were lysed in Tris lysis buffer. Lysate (50–200 μg) was incubated with 5–20 μl of glutathione–Sepharose for 1 h at 4 °C on a vibrating platform. SGK1 activity was assayed exactly as described previously  using the Crosstide peptide (GRPRTSSFAEG) at 30 μM. Incorporation of [32P]-phosphate into the peptide substrate was determined by applying the reaction mixture on to P81 phosphocellulose paper and scintillation counting after washing the papers in phosphoric acid. One unit of activity was defined as that which catalysed the incorporation of 1 nmol of [32P]-phosphate into the substrate. For the activity assay of purified SGK1 after phosphorylation with mTORC2 and/or PDK1 (Figure 3C), GST–ΔNSGK1 was phosphorylated with mTORC2 (as described above) in the presence or absence of purified GST–PDK1 (0.1 μg), which was added during the last 10 min of the reaction. Independent aliquots (1 μg) of the resultant phosphorylated GST–ΔNSGK1 were then assayed for SGK1 activity using the Crosstide assay described above.
Quantifying the amount of GST–SGK1
The amount of GST–SGK1 isolated from MEF cells, as well as HEK-293 cells, was quantified by undertaking quantitative Coomassie Blue staining of SDS gels in which GST–SGK1 precipitates were run side-by-side with known amounts of BSA markers. The gels were analysed using a LI-COR Odyssey IR detection system following the manufacturer's guidelines. The band intensity was quantified using LI-COR software.
Total cell lysate (20 μg) or immunoprecipitated samples were heated at 95 °C for 5 min in sample buffer, and subjected to PAGE and electrotransferred on to nitrocellulose membranes. Membranes were blocked for 1 h in TBS-Tween buffer containing 10% (w/v) dried skimmed milk powder. The membranes were probed with the indicated antibodies in TBS-Tween containing 5% (w/v) dried skimmed milk powder or 5% (w/v) BSA for 16 h at 4 °C. Detection was performed using HRP-conjugated secondary antibodies and ECL (enhanced chemiluminescence) reagent.
Indication that mTORC2 regulates SGK1 hydrophobic motif phosphorylation
To explore whether mTORC2 regulates SGK1, we first expressed a form of SGK1 lacking the N-terminal PEST degradation motif [6,42] in wild-type and rictor-deficient MEFs that lack mTORC2 activity . Cells were cultured in the presence of serum and SGK1 activity, as well as hydrophobic motif Ser422 phosphorylation, was analysed. In wild-type cells, SGK1 was active (∼2 units/mg GST–SGK1) and significantly phosphorylated at its hydrophobic motif. Mutation of Ser422 to alanine abolished recognition of SGK1 by the phospho-Ser422 antibody as well as SGK1 activity. In contrast, in rictor-knockout MEFs, SGK1 was expressed at a similar level to wild-type cells, but was inactive and not detectably phosphorylated at its hydrophobic motif (Figure 1A). In agreement with the rictor-deficient cells lacking mTORC2 activity, Akt was not phosphorylated at its hydrophobic motif (Ser473). However, the rictor-knockout cells still possessed mTORC1 function, as S6K was still phosphorylated at its hydrophobic motif (Thr389) (Figure 1A).
mTORC2-deficient MEFs lack SGK1 activity and hydrophobic motif phosphorylation
Activation of SGK1 is triggered by the interaction of PDK1 with SGK1 following the phosphorylation of Ser422 in the hydrophobic motif. An SGK1 mutant in which the hydrophobic motif Ser422 residue is mutated to aspartate, in order to mimic phosphorylation, is active when expressed in cells owing to its ability to constitutively interact with PDK1 [6,10,43]. It would therefore be expected that expression of the SGK1[S422D] mutant in rictor-knockout cells should bypass the requirement for TORC2 in activating this enzyme. Consistent with this, we found that SGK1[S422D], in contrast with wild-type SGK1, was significantly active when expressed in rictor-deficient cells (Figure 1B).
Full-length SGK1, although expressed at lower levels than SGK1 lacking the PEST motif, was also phosphorylated at Ser422 in wild-type, but not in rictor-deficient, MEFs (Figure 1C). We also found that SGK1 was not detectably phosphorylated at Ser422 in MEFs that lack the other critical mTORC2 subunits, namely mLST8 (Figure 1D) or Sin1 (Figure 1E). Similar to rictor-deficient MEFs and consistent with previous studies [28–31], mLST8- and Sin1-knockout cells lacked Akt Ser473 phosphorylation, but still displayed S6K Thr389 phosphorylation, confirming a lack of mTORC2, but not mTORC1, activity in these cells (Figures 1D and 1E).
Evidence that mTORC2 regulates activity of endogenous SGK1
To investigate whether mTORC2 controlled SGK1 activity in vivo, we examined the phosphorylation of NDRG1, a previously characterized substrate for SGK1 . SGK1 phosphorylates NDRG1 at three residues, Thr346, Thr356 and Thr366, that lie within a repeated decapeptide sequence . To ensure that this approach represented a reliable readout for SGK1 activity, we first studied the phosphorylation of NDRG1 in skeletal muscle derived from wild-type and SGK1-knockout, mice and found that NDRG1 was phosphorylated in extracts derived from wild-type, but not SGK1-knockout, mice (Figure 2). We then examined NDRG1 phosphorylation in MEFs deficient in rictor, mLST8 or Sin1, and observed that NDRG1 was markedly phosphorylated in all of the wild-type control MEFs, but not in any of the mTORC2-subunit-deficient cells (Figure 2).
Dependence of NDRG1 phosphorylation on mTORC2
mTORC2 phosphorylates the hydrophobic motif of SGK1 in vitro
We next isolated endogenous mTORC2 complex from HEK-293 cells by immunoprecipitating rictor and investigated whether it was capable of phosphorylating the hydrophobic motif of SGK1. The rictor immunoprecipitates contained the known mTORC2 complex subunits (mTOR, rictor, mLST8, Sin-1 and protor-1), but not the specific mTORC1 component raptor (Figure 3A). In the presence of MgATP and recombinant SGK1 (isolated from serum-starved HEK-293 cells incubated with a PI3K inhibitor), immunoprecipitated mTORC2 induced phosphorylation of Ser422 (Figure 3B). In parallel experiments, a control immunoprecipitation undertaken with a pre-immune antibody, failed to phosphorylate SGK1 at Ser422 (Figure 3B). mTORC2 did not phosphorylate SGK1 at its T-loop Thr256 residue, whereas PDK1 phosphorylated SGK1 at its T-loop, but not the hydrophobic motif (Figure 3C). Phosphorylation of SGK1 with mTORC2 in the absence of PDK1 did not stimulate SGK1 activity significantly (Figure 3C), consistent with the evidence that phosphorylation of Thr256 is required to trigger activation of SGK1 . Incubation of SGK1 with PDK1 in the absence of mTORC2 induced substantial activation of SGK1 to an activity of ∼3 units/mg GST–SGK1 (Figure 3C), similar to the specific activity of wild-type SGK1 expressed in MEFs (Figure 1A). In the presence of both mTORC2 and PDK1, SGK1 was phosphorylated at both Thr256 and Ser422 and its activity was further stimulated to 5–6 units/mg GST–SGK1 (Figure 3C).
Immunoprecipitated mTORC2 phosphorylates SGK1 in vitro
To demonstrate that an intact mTORC2 complex is required to catalyse hydrophobic motif phosphorylation of SGK1, we immunoprecipitated rictor from either wild-type or mLST8-knockout MEFs and tested how this affected phosphorylation of SGK1 at Ser422. Consistent with previous work , we found that the lack of mLST8 abolished the interaction of rictor with mTOR, without affecting the ability of rictor to interact with Sin1 and protor (Figure 3D). Rictor immunoprecipitates derived from mLST8-deficient cells failed to phosphorylate SGK1 under conditions which rictor immunoprecipitated from wild-type control MEFs phosphorylated SGK1 at Ser422 (Figure 3E).
Immunoprecipitated mTORC1 phosphorylates the hydrophobic motif of S6K, but not SGK1
To investigate whether the mTORC1 complex was capable of phosphorylating SGK1 at Ser422, we immunoprecipitated mTOR from rictor-knockout MEFs that lack mTORC2, but still possess a functional mTORC1 complex, as emphasized by the observations that S6K is still phosphorylated in these cells (Figure 1). Consistent with previous studies [31,37], immunoprecipitates of mTOR derived from rictor-deficient MEFs were associated with the mTORC1 components raptor and mLST8, but not with the mTORC2 subunits rictor, Sin1 and protor-1 (Figure 4A). mTOR immunoprecipitated from rictor-deficient MEFs phosphorylated S6K at Thr389in vitro to the same extent as mTOR isolated from wild-type cells, in agreement with mTORC1 mediating this reaction (Figure 4B). However, in parallel experiments, mTOR immunoprecipitated from rictor-knockout MEFs failed to phosphorylate SGK1 at Ser422 under conditions which mTOR immunoprecipitated from wild-type MEFs phosphorylated SGK1 (Figure 4C). We also isolated mTORC1 by immunoprecipitating raptor from HEK-293 cells and demonstrated that it was associated with mTOR and mLST8, but not with rictor, Sin-1 or protor-1 (Figure 4D). Consistent with the conclusion that mTORC1 is unable to phosphorylate SGK1 at Ser422, raptor immunoprecipitates failed to phosphorylate SGK1 at Ser422, under conditions which they phosphorylated S6K at Thr389 (Figures 4E and 4F). In parallel experiments, rictor immunoprecipitates phosphorylated SGK1 at Ser422, but did not phosphorylate the hydrophobic motif of S6K (Figures 4E and 4F).
Immunoprecipitated mTORC1 phosphorylates S6K1, but not SGK1, in vitro
Phosphorylation of the hydrophobic motif of SGK1 is not inhibited by the mTORC1 inhibitor rapamycin
To study whether mTORC1 has any role in regulating hydrophobic motif phosphorylation of SGK1 in vivo, we investigated how the activation and phosphorylation of SGK1 was affected by short-term treatment of HEK-293 cells with the mTORC1 inhibitor rapamycin, conditions that do not affect mTORC2 . HEK-293 cells expressing full-length SGK1 were cultured in serum and treated in the presence or absence of rapamycin (100 nM) for 30 min. As expected, rapamycin abolished mTORC1-regulated phosphorylation of S6K at Thr389, as well as phosphorylation of the S6 protein substrate at Ser235, without affecting mTORC2-mediated phosphorylation of Akt at Ser473 (Figure 5). However, rapamycin did not significantly inhibit SGK1 activity, its phosphorylation at Ser422 or phosphorylation of its NDRG1 substrate (Figure 5). As expected, treatment of cells with the PI3K inhibitor PI-103, which also inhibits mTORC1 and mTORC2 , suppressed SGK1 activity and phosphorylation of Ser422 (Figure 5). Consistent with this, PI-103 also inhibited the phosphorylation of NDRG1, Akt, S6K and the S6 protein.
The mTORC1 inhibitor rapamycin does not suppress hydrophobic motif phosphorylation or activation of SGK1 in three different cell lines
A recent study by Hong et al. , reported that in MCF-7 and HeLa cells, hydrophobic motif phosphorylation of SGK1 is controlled by mTORC1 and thus is inhibited by rapamycin. To analyse this further, we expressed full-length SGK1 in MCF-7 cells (Figure 5B) as well as in HeLa cells (Figure 5C) and tested whether phosphorylation of the hydrophobic motif of SGK1 or NDRG1 was inhibited by rapamycin. In contrast with the findings of Hong et al. , we observed that phosphorylation of SGK1 at Ser422 or phosphorylation of NDRG1 was not affected by rapamycin under conditions which this drug inhibited hydrophobic motif phosphorylation of S6K at Thr389. As expected, phosphorylation of SGK1 at Ser422 in MCF-7 and HeLa cells was abolished following treatment with PI-103.
Hong et al.  utilized an anti-phospho-Ser422-SGK1 antibody (catalogue number sc-16745-R; Santa Cruz Biotechnology) to analyse the phosphorylation of endogenous SGK1. We have purchased this antibody and immunoblotted MCF-7, HeLa and HEK-293 cell extracts utilized in the studies shown in Figure 5. The Santa Cruz Biotechnology antibody recognizes a significant non-specific band that migrates at a similar position to GST–SGK1 (∼85 kDa), making it hard to assess hydrophobic motif phosphorylation of overexpressed GST–SGK1 (Figure 6). However, the Santa Cruz Biotechnology phospho-Ser422-SGK1 antibody also strongly recognizes a phosphorylated protein of ∼70 kDa whose phosphorylation is inhibited by both rapamycin and PI-103 (Figure 6). In our opinion, this protein is likely to comprise endogenous S6K1 and/or S6K2, rather than SGK1, as this signal co-migrates with the band recognized by antibodies against S6K1/S6K2 protein. Immunoblotting with an antibody that recognizes endogenous SGK1 demonstrates that it migrates at the expected ∼48 kDa position (Figure 6). Following a long exposure of the immunoblot, the Santa Cruz Biotechnology phospho-Ser422-SGK1 antibody weakly recognizes bands of ∼45–55 kDa that are likely to represent endogenous SGK1 and/or other isoforms/splice variants of SGK (Figure 6). Importantly, phosphorylation of proteins attributed to endogenous SGK isoforms was inhibited by PI-103, but not by rapamycin (Figure 6).
Analysis of endogenous phosphorylation of SGK
The results of the present study demonstrate that mTORC2 phosphorylates the hydrophobic motif of SGK1. This is based on the finding that in MEFs lacking the critical mTORC2 subunits, SGK1 is not phosphorylated at its hydrophobic motif and is thus inactive (Figure 1). Moreover, NDRG1 is not phosphorylated at the residues targeted by SGK1 in mTORC2-deficient cells (Figure 2). MEFs lacking active mTORC2 still possessed functional mTORC1 activity as emphasized by S6K being phosphorylated in these cells. However, despite possessing mTORC1 activity, SGK1 was not detectably phosphorylated at its hydrophobic motif, suggesting that mTORC1 does not contribute to the phosphorylation of SGK1, at least in MEFs. This conclusion is also consistent with our present observations (Figures 5 and 6), as well as previous reports [6,7,47], that acute rapamycin treatment of HEK-293, HeLa or MCF-7 cells does not inhibit serum-induced SGK1 activity or phosphorylation, under conditions which it inhibited phosphorylation and activation of S6K. To further demonstrate that mTORC2 controls SGK1, we found that immunoprecipitated mTORC2 can phosphorylate SGK1 at Ser422in vitro and that this phosphorylation is dependent upon the presence of the rictor and mLST8 subunits of mTORC2 (Figure 3). Although isolated endogenous mTORC1 phosphorylated the hydrophobic motif of S6K, in a parallel reaction it did not phosphorylate the hydrophobic motif of SGK1 (Figure 4).
We propose a model in Figure 7 in which growth factors and other agonists that stimulate PI3K lead to the activation of mTORC2. Activated mTORC2 phosphorylates the hydrophobic motif of SGK1 triggering its interaction with PDK1, via the PIF-motif substrate-docking site. PDK1 then phosphorylates the T-loop of SGK1 resulting in its activation. This model accounts for the sensitivity of SGK1 activation to PI3K inhibitors [6,7], but not rapamycin. It is also explains why prior hydrophobic motif phosphorylation and integrity of the PIF-pocket of PDK1 is essential for SGK1 activation and phosphorylation [10,11]. It also accounts for why the binding of PtdIns(3,4,5)P3 to PDK1 is not required for the activation of SGK1 in vivo , as phosphoinositide binding is not required for PDK1 to interact with the hydrophobic motif of SGK1 and phosphorylate its T-loop residue . One study reported that the WNK1 [with no lysine (K) 1] kinase, which is activated by osmotic shock and regulates salt uptake into cells , activated SGK1 by controlling the hydrophobic phosphorylation of SGK1 . How this fits into the model of SGK1 activation shown in Figure 7 is unclear, unless WNK1 plays a role in controlling the activity of mTORC2.
Mechanism of SGK1 activation
Hong et al.  have recently reported, in disparity to the observations made in the present study, that the phosphorylation of the hydrophobic motif of SGK1 was regulated by mTORC1. Key evidence to support this conclusion was based on the finding that phosphorylation of the hydrophobic motif of SGK1 was sensitive to the mTORC1 inhibitor rapamycin in WM35, MCF-7 and HeLa cells . However, we have been unable to reproduce these findings and observe that phosphorylation of the hydrophobic motif of SGK1 is not inhibited by treatment of HEK-293, MCF-7 or HeLa cells with rapamycin, under conditions where phosphorylation of the hydrophobic motif of S6K1 is inhibited (Figure 5). Other groups have also reported that SGK1 activation in HEK-293 or HeLa cells is not inhibited by rapamycin [6,7,47]. We have found that the Santa Cruz Biotechnology phospho-Ser422-SGK1 antibody used in the Hong et al.  study recognizes phosphorylated endogenous S6K1 and/or S6K2 (∼70 kDa) on immunoblot analysis of cell extracts much more strongly than it recognizes endogenous phosphorylated SGK1 (∼48 kDa) (Figure 6). Long exposure of immunoblots was required to detect hydrophobic motif phosphorylation of endogenous SGK in MCF-7 or HeLa cells that was inhibited by PI-103, but not by rapamycin (Figure 6). To avoid confusion when employing phospho-Ser422-SGK1 antibodies, caution is required to discriminate between the lower abundance phosphorylation of SGK1 migrating at ∼48 kDa compared with the much stronger recognition of S6K1/S6K2 migrating at ∼70 kDa. There seems to be significant antigenic similarity between the phosphorylated hydrophobic motifs of SGK1 and S6K1, as antibodies raised against the phosphorylated hydrophobic motif of S6K1 also recognize the phosphorylated hydrophobic motif of SGK1 .
A key implication of the present findings is that mTORC2-deficient cells investigated in the present study are devoid of SGK1 activity, and therefore the phosphorylation of cellular substrates of SGK1 should be markedly reduced. In contrast with SGK1, Akt is still activated to a significant extent in mTORC2-deficient cells [30,31,37], as it is phosphorylated at Thr308 by PDK1 in a reaction that is not dependent upon mTORC2. In this regard, phosphorylation of well-characterized Akt substrates, such as GSK3 (glycogen synthase kinase 3) and TSC2, was not observed to be significantly impaired following the loss of critical mTORC2 subunits in Sin1-  or mLST8-  deficient MEFs. This is probably due to partial activation of Akt being capable of inducing near-normal phosphorylation of its substrates. In contrast with TSC2 and GSK3, the phosphorylation of FOXO (forkhead box O) 1 (Thr24) and FOXO3a (Thr32) was markedly reduced in cells lacking mTORC2 activity [30,31]. This observation was interpreted in one study to imply that Akt phosphorylated at only Thr308 may possess a different substrate specificity to Akt phosphorylated at both Thr308 and Ser473 . However, an alternative explanation suggested by the present study, is that the lack of SGK activity in mTORC2-deficient cells accounts for inhibition of FOXO phosphorylation. Consistent with this, previous reports have demonstrated that SGK isoforms are capable of efficiently phosphorylating FOXO [50,51]. Moreover, employing an antibody that recognizes substrates phosphorylated at an Akt/SGK phosphorylation motif, Jacinto et al.  reported that an non-identified protein of 48 kDa was heavily phosphorylated in wild-type, but not Sin1-knockout, MEFs. It is tempting to speculate that the protein visualized in this experiment was in fact NDRG1, that possesses a molecular mass of ∼48 kDa. Undertaking a screen to identify proteins whose phosphorylation is markedly suppressed in mTORC2-deficient cells would be a good approach to identify SGK substrates. Identification of novel SGK substrates would expand our understanding of the roles of these poorly characterized kinases. There is also much on-going research to develop drugs that suppress the activity of mTORC2 for the treatment of cancer . The present study suggests that specific mTORC2 inhibitors will be more effective in suppressing phosphorylation of SGK1 substrates than Akt substrates.
In conclusion, we have identified SGK1 as a novel substrate for the mTORC2 complex. We demonstrate that mTORC2 phosphorylates SGK1 at Ser422 and that this is required for its SGK1 activation. Our results indicate that the SGK-phosphorylated form of NDRG1 represents an excellent biomarker of mTORC2 activity and that mTORC2-deficient cells could be utilized to identify substrates of SGK.
We thank Dr Krishna M. Boini and Dr Florian Lang (Department of Physiology, University of Tübingen, Tübingen, Germany), Dr Mark Magnuson (Vanderbilt University School of Medicine, Nashville, TN, U.S.A.), Dr David Sabatini (Whitehead Institute for Biomedical Research, Cambridge, MA, U.S.A.) and Dr Bing Su (Yale University School of Medicine, New Haven, CT, U.S.A.) for the provision of reagents. We are also grateful to the Sequencing Service (School of Life Sciences, University of Dundee, Dundee, Scotland, U.K.) for DNA sequencing, the Post Genomics and Molecular Interactions Centre for Mass Spectrometry facilities (School of Life Sciences, University of Dundee, Dundee, Scotland, U.K.) and the protein production and antibody purification teams [DSTT (Division of Signal Transduction Therapy), University of Dundee, Dundee, Scotland, U.K.] co-ordinated by Hilary McLauchlan and James Hastie for the expression and purification of antibodies.
protein kinase A/protein kinase G/protein kinase C
epithelial sodium channel
forkhead box O
glycogen synthase kinase 3
human embryonic kidney
mouse embryonic fibroblast
mammalian lethal with SEC13 protein 8
mammalian target of rapamycin
N-myc downstream regulated gene 1
phosphoinositide-dependent kinase 1
protein kinase C
regulatory associated protein of mTOR
rapamycin-insensitive companion of mTOR
protein observed with rictor
ribosomal S6 kinase
serum- and glucocorticoid-induced protein kinase
stress-activated-protein-kinase-interacting protein 1
tuberous sclerosis complex 2
with no lysine (K) 1
This work was supported by the Medical Research Council and AstraZeneca (a grant to J. M. G. M.).