Biochemical Journal

Research article

mTORC2 targets AGC kinases through Sin1-dependent recruitment

Angus J. M. Cameron, Mark D. Linch, Adrian T. Saurin, Cristina Escribano, Peter J. Parker

Abstract

The protein kinase TOR (target of rapamycin) is a key regulator of cell growth and metabolism with significant clinical relevance. In mammals, TOR signals through two distinct multi-protein complexes, mTORC1 and mTORC2 (mammalian TOR complex 1 and 2 respectively), the subunits of which appear to define the operational pathways. Rapamycin selectively targets mTORC1 function, and the emergence of specific ATP-competitive kinase inhibitors has enabled assessment of dual mTORC1 and mTORC2 blockade. Little is known, however, of the molecular action of mTORC2 components or the relative importance of targeting this pathway. In the present study, we have identified the mTORC2 subunit Sin1 as a direct binding partner of the PKC (protein kinase C) ϵ kinase domain and map the interaction to the central highly conserved region of Sin1. Exploiting the conformational dependence for PKC phosphorylation, we demonstrate that mTORC2 is essential for acute priming of PKC. Inducible expression of Sin1 mutants, lacking the PKC-interaction domain, displaces endogenous Sin1 from mTORC2 and disrupts PKC phosphorylation. PKB (protein kinase B)/Akt phosphorylation is also suppressed by these Sin1 mutants, but not the mTORC1 substrate p70S6K (S6 kinase), providing evidence that Sin1 serves as a selectivity adaptor for the recruitment of mTORC2 targets. This inducible selective mTORC2 intervention is used to demonstrate a key role for mTORC2 in cell proliferation in three-dimensional culture.

  • cancer
  • mammalian target of rapamycin (mTOR)
  • nucleotide pocket
  • protein kinase B (PKB)
  • protein kinase C (PKC)
  • Sin1

INTRODUCTION

Over recent years, the TOR (target of rapamycin) pathway has emerged as one of the key regulators of cell growth, providing a nexus for the integration of growth factor signalling with the availability of energy and nutrients [1]. Among the key targets through which TOR exerts its effects on cell growth, proliferation and survival are members of the AGC family of kinases [2]. This large group of kinases, which includes PKC (protein kinase C), PKB (protein kinase B) and p70S6K (S6 kinase) share a well-conserved mechanism for regulating the activity of their kinase domains. This typically involves phosphorylation of the kinase activation loop by PDK1 (phosphoinositide-dependent kinase 1) and phosphorylation of two conserved motifs within kinase domain C-terminal tails [3]. These C-terminal phosphorylation sites, termed the turn motif and hydrophobic motif, are commonly targeted by TOR and, once phosphorylated, serve to lock the kinase domain in a stable activity-competent conformation [46].

TOR is a serine/threonine protein kinase initially identified and characterized in yeast as the target of the anti-proliferative bacterial metabolite rapamycin [7]. In Saccharomyces cerevisiae, there are two TOR genes (TOR1 and TOR2), which contribute to two functionally distinct protein TOR complexes: TORC1 (TOR complex 1; TOR1 or TOR2) and TORC2 (TOR complex 2; TOR2) [8]. Rapamycin acts through binding to FKBP12 (FK506-binding protein 12) which, as a complex, binds to and inhibits the TOR kinase activity of TORC1, but not TORC2. In mammals, these complexes are conserved as mTORC1 (mammalian TORC1) and mTORC2 (mammalian TORC2), which are also differentially sensitive to rapamycin. In mammals, both mTORCs share the single catalytic subunit mTOR (mammalian TOR) and the adaptor protein mLST8 in addition to the complex specific subunits raptor (regulatory associated protein of mTOR), for mTORC1, or rictor (rapamycin-insensitive companion of mTOR), protor (protein observed with rictor) and Sin1 for mTORC2 [1,913]. mTORC1 and mTORC2 have been shown to target distinct members of the AGC family. mTORC1, for example, phosphorylates and activates p70S6K to regulate protein synthesis, whereas mTORC2 has been shown to target the hydrophobic and turn motifs of both PKB and PKC [1419]. Despite an explosion in our understanding of the signalling pathways regulated by TOR, little is known about the mechanism of action or how the components of the distinct TORCs target specific downstream effectors [12].

In contrast with PKB where phosphorylation is acutely regulated in response to PI3K (phosphoinositide 3-kinase) activation, PKC isoforms are generally constitutively phosphorylated on the three conserved kinase domain-priming sites. To circumvent the limitations imposed by constitutive PKC priming, we have exploited the finding that stable PKC phosphorylation is contingent on occupation of the nucleotide-binding pocket. Nucleotide-binding mutants, such as the kinase-dead PKCϵ K437M and PKCα K368M are virtually unprimed, but allow for acute regulation of priming allosterically through the introduction of ATP-competitive inhibitors of PKC (described in detail in [20]). Using this system, we demonstrate an essential role for mTORC2 in the acute priming of PKC. Furthermore, we have identified Sin1 as a direct binding partner of PKC and mapped the site of interaction to the conserved central region of Sin1. We have used this information to derive dominant interfering forms of Sin1, which block PKC phosphorylation through mTORC2. This approach provides a unique route to mTORC2 intervention in vivo without disrupting mTORC2 assembly or mTORC1-mediated signalling. Exploiting these insights, we demonstrate that expression of an inducible interfering Sin1 mutant protein, specifically restricts phosphorylation of mTORC2 target kinases, and inhibits cell growth in suspension cultures. Thus mTORC2 can play a critical role in growth control.

EXPERIMENTAL

Reagents

Reagents were purchased from Sigma unless specified otherwise. Ku0063794 was from Tocris Bioscience. Anti-GFP (green fluorescent protein) (3E1 and 4E12/8) and anti-Myc (9E10) mAbs (monoclonal antibodies) were prepared by the Cancer Research UK antibody facility. Rabbit anti-PKCϵ (pThr710) (PPA218), anti-PKCϵ (pThr566) (PPA204) and anti-PKCα (pThr497) (PPA164) are as described previously [21]. Rabbit anti-PKCϵ (pSer729) and anti-PKCα (pSer657) were from Millipore. Rabbit anti-p70S6K (total and pThr389), anti-PKB (pSer473) and mouse mAb against total PKB (clone 40D4) were from Cell Signaling Technology. HRP (horseradish peroxidase)-linked anti-rabbit and anti-mouse antibodies were from GE Healthcare. Protein G–Dynabeads™, Protein A/G-conjugated HRP and Alexa Fluor® 680-conjugated goat anti-rabbit antibodies were from Invitrogen. IRDye 800-conjugated goat anti-mouse secondary antibody was from LI-COR Biosciences. Polyclonal antibodies against Sin1, rictor and mTOR were from Bethyl Laboratories.

Plasmids

Previously described GFP–PKCϵ and GFP–PKCϵ K437M [22] were subcloned into pcDNA5/FRT/TO. Cloning of PKCϵ into EGFP (enhanced GFP)–C1 [23] and generation of kinase-dead GFP–PKCα K368M in pEGFP-C2 are as described previously [20]. HA (haemagglutinin)–PKB and HA–ΔPH PKB (kinase domain) in pCMV5 vector were a gift from Jongsun Park and Brian Hemmings (Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland). GFP–PKC mutants were generated using the QuikChange® system (Stratagene). Full-length Sin1 was a gift from Gillian Bushell (Griffith University, Brisbane, QLD, Australia). Untagged Sin1 was subcloned into the EcoRI site of pcDNA3.1 or a modified version of pcDNA5.0 in which EGFP had been previously inserted into the KpnI site to generate N-terminally tagged Sin1. For C-terminally Myc-tagged Sin constructs, Sin1.1–Myc was obtained from Addgene courtesy of the David Sabatini laboratory. Sin1.1–Myc was subcloned into pcDNA5.0. Truncations and deletions of Sin1 were generated using the QuikChange® system.

Cell culture and treatments

HEK (human embryonic kidney)-293 cells were cultured in DMEM (Dulbecco's modified Eagle's medium), 10% fetal bovine serum, penicillin (50 units/ml) and streptomycin (0.05 mg/ml) under 10% CO2. Tetracycline-inducible HEK-293 cell lines were generated using the FRT T-Rex™ system (Invitrogen) according to the manufacturer's instructions. GFP–PKCϵ or Sin1 expression was induced with 100 ng/ml tetracycline. HEK-293 cells were transiently transfected using Lipofectamine™ LTX (Invitrogen). For transient transfection of tetracycline-inducible cell lines, GFP–PKCϵ K437M and GFP–PKCα K368M constructs were expressed from pEGFP-C1 and -C2 vectors respectively to avoid suppression of expression in the absence of tetracycline. For siRNA (short interfering RNA) transfections, cells were transfected using 10 nM Dharmacon siRNA pools (four duplexes for each pool) using HiPerfect (Qiagen). The specific siRNAs used were: mTOR siGenome SMART pool (#1, 5′-GAGAAGAAAUGGAAGAAAU-3′; #2, 5′-CCAAAGUGCUGCAGUACUA-3′; #3, 5′-GAGCAUGCCGUCAAUAAUA-3′; #4, 5′-GGUCUGAACUGAAUGAAGA-3′), rictor siGenome SMART pool (#1, 5′-UCAACGAGCUCACAUAUGA-3′; #2, 5′-GUACGAAGACUACUUUAUU-3′; #3, 5′-UGACCGAUCUGGACCCAUA-3′; #4, 5′-GUACUUGGGC-UCAUAGCUA-3′), raptor siGenome SMART pool (#1, 5′-GCCCGUCGAUCUUCGUCUA-3′; #2, 5′-UGGAGAAGCGU-GUCAGAUA-3′; #3, 5′-AGAAGGGCAUUACGAGAUU-3′; #4, 5′-GAAACCAUCGGUGCAAAUU-3′), Sin1 siGenome SMART pool (#1, 5′-AGACUCAGGGCUAUGUAUA-3′; #2, 5′-GAUU-AGAACGACUCCGAAA-3′; #3, 5′-CAUUCUAGCUCAUA-UUCGA-3′; #4, 5′-GAUAUUACCUCAAGUUGGG-3′), and Sin1 ON-TARGETplus (#1, 5′-GGUCUGACAUCCAAAGA-GU-3′; #2, 5′-GUGGACAACACAAAGGUUA-3′; #3, 5′-AA-UUAAAUGCUGCUCAUGGA-3′; #4, 5′-GUACUUUGGCCC-UGGUUGA-3′).

Immunoprecipitation and immunoblotting

Two lysis buffers were used in immunoprecipitation studies. For mapping of the Sin1–PKC/PKB interaction, cells were lysed in 0.5% Nonidet P40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris/HCl (pH 7.5) and 120 mM NaCl supplemented with Complete™ protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail I (Sigma). For immunoprecipitation of endogenous mTORC2, 0.3% CHAPS was used as a detergent to maintain complex integrity [9,24]; cells were lysed in 0.3% CHAPS, 40 mM Hepes (pH 7.4), 120 mM NaCl, 1 mM EDTA, 50 mM NaF, 10 mM pyrophosphate, 10 mM 2-glycerophosphate and 2 mM sodium orthovanadate supplemented with Complete™ protease inhibitor cocktail. Cell lysates were centrifuged at 14000 g for 5 min to remove insoluble material and incubated with 1–5 μg of antibody conjugated to Protein G–Dynabeads™ for 1 h. Immunoprecipitations were washed four times with 1 ml of lysis buffer.

For immunoblotting, lysates or immunoprecipitates were resolved by SDS/PAGE and transferred on to PVDF membranes. Immunoblots were blocked in TBST [Tris-buffered saline (20 mM Tris/HCl, pH 7.5, and 125 mM NaCl) containing 0.1% Tween 20] with 3% BSA and probed with primary antibodies as indicated. Phosphoprotein immunoblotting was carried out with relevant dephosphopeptide (1 μg/ml) to block immunodetection of dephosphorylated proteins. Following incubation with appropriate secondary antibodies, bands were visualized by ECL (enhanced chemiluminescence) (GE Healthcare) or using the Odyssey infrared imaging system (LI-COR Biosciences). For immunoblotting of Sin1 in immunoprecipitations, Protein A-conjugated HRP was used as a secondary reagent in place of HRP-conjugated antirabbit antibodies to prevent cross-reaction with the immunoprecipitating antibody, which otherwise masks the signal.

For quantification of specific protein phosphorylations in Sin1-inducible cell lines using the Odyssey infrared imaging system, stimulations were carried out in 24-well format in triplicate. Western blots were simultaneously probed for phospho-PKC (using Alexa Fluor® 680-conjugated anti-rabbit secondary antibodies) and GFP (using IRDye 800-conjugated anti-mouse secondary antibodies) to control for protein loading (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/439/bj4390287add.htm). Data are calculated as phospho-PKC/GFP and are normalized such that the average value for Bim I-stimulated control cells is equal to 1. Likewise for PKB, phospho-PKB (Ser473) was detected with rabbit polyclonal antibody while simultaneously probing for total PKB with mAb (Clone 40D4) using the same secondary antibodies as for PKC above. PKB data are calculated as phospho-PKB/total PKB. The data are presented as the means±S.D. and are representative of a minimum of three separate experiments. Typically, a single measurement was made for PKCϵ K437M and PKCα K368M phosphorylation levels in unstimulated cells (where signal is invariably very low) and consequently no error bars are given for these data points.

Proliferation assays

Cell proliferation/viability was assessed in 96-well format using the Cell Titer Glo assay system according to the manufacturer's instructions (Promega). For tetracycline-inducible cell lines, cells were pre-treated with 100 ng/ml tetracycline or vehicle (ethanol) for 48 h before plating 103 cells/well. The three-dimensional culture of HEK-293/FRT cells in Matrigel™ (BD Biosciences) was performed as described previously [25,26]. In brief, cells in exponential-phase growth were trypsinized and resuspended in standard medium supplemented with 2% low-growth factor Matrigel™ (BD Biosciences) at 2 ×104 cells/ml. Each well of an eight-well chamber slide (BD Biosciences) was pre-coated with 30 μl of 100% Matrigel™ to which 400 μl of the cell suspension was added. Medium, 2% Matrigel™ with or without tetracycline, was changed on alternate days for 5 days at which time four phase images were taken per well. The cross-sectional area of the resulting HEK-293/FRT spheroids were estimated from the images using ImageJ software (NIH, version 1.40g) and data were analysed for statistical significance using two-way ANOVA.

RESULTS

Sin1 is a direct binding partner of PKCϵ

We have identified Sin1, a component of mTORC2, as a direct interaction partner with the kinase domain of PKCϵ in a yeast two-hybrid screen. Using the PKCϵ kinase domain as bait, a partial clone of Sin1 isoform 1.2 comprising residues 148 to the C-terminus was isolated (see the schematic diagram of Sin1 in Figure 2). This, along with pharmacological data and published studies identifying PKC as a potential mTORC2 substrate, prompted us to investigate the role for this interaction in the recruitment of PKC to mTORC2. An interaction between Sin1 and both PKCϵ and the well-established mTORC2 target PKB was confirmed in co-expression studies (Figure 1). Immunoprecipitation of GFP–PKCϵ and kinase-dead GFP–PKCϵ K437M, both as full-length proteins or as kinase domains alone, co-purified untagged co-expressed Sin1 (Figure 1A). Reciprocally, immunoprecipitation of N-terminally tagged GFP–Sin1 co-purified both full-length Myc–PKCϵ and Myc–PKCϵ kinase domain (Figure 1B). In either configuration, GFP itself failed to co-purify either Sin1 or Myc–PKCϵ. We assessed the interaction of Sin1 with a well-characterized mTORC2 target, PKBα. GFP–Sin1 also co-purified HA-tagged PKBα (HA–PKBα) or the PKBα kinase domain alone (Figure 1C). Taken together, these data identify Sin1 as a direct interacting partner of mTORC2 target kinases of the PKC and PKB families.

Figure 1 Sin1 interacts with PKC and PKB

(A) HEK-293 cells were transiently transfected for 24 h with GFP, GFP-tagged PKCϵ or GFP–PKCϵ K437M as full-length proteins or as the kinase domains alone (Kin) in the presence of untagged Sin1. GFP immunoprecipitates (left-hand panels) and cell lysates (right-hand panels) were analysed for Sin1 and GFP by Western blotting. (B and C) Cells were co-transfected with GFP or GFP–Sin1 with either Myc–PKCϵ (B) or HA–PKB (C) constructs as full-length or kinase domains. Lysates and GFP immunoprecipitates were examined by Western blotting for GFP and either PKCϵ or PKB. IP, immunoprecipitation.

PKC interacts with the CRIM domain of Sin1

Sin1 proteins contain a number of distinct domains and are expressed as a variety of splice variants likely to perform functionally distinct roles (Figure 2A). For clarity, we refer to distinct Sin1 isoforms using the nomenclature described by Frias et al. [24]. The first 192 amino acids of Sin1 contains no recognizable protein domains, but have been proposed to play a role in assembly into mTORC2. Adjacent to the putative mTORC2-interacting region lies a highly conserved region found in all Sin1 orthologues termed the CRIM (conserved region in middle) domain which is significantly also the only region of Sin1 present in all of the splice variants so far described [27]. The C-terminal half of Sin1 comprises an RBD (Ras-binding domain) and a PH (pleckstrin homology) domain which are variously disrupted or missing from distinct Sin1 splice variants [28] (Figure 2A).

Figure 2 PKCϵ interacts with the CRIM domain of Sin1

(A) Schematic diagram of the Sin1 domain structure, splice variants and interaction mapping constructs. There are five described variants of Sin1 (Sin1.1–Sin1.5) all of which contain a central conserved region (CRIM) of unknown function. Sin1.5 utilizes a distinct splice site and stop codon, ending with residues -ACD as indicated. The construct labelled Y2H represents the fragment of Sin1.2 isolated in a yeast two-hybrid screen using the PKCϵ kinase domain as bait. GFP–Sin1 truncation and deletion mutants were constructed along naturally occurring splice boundaries. (B) Cells were transfected with GFP or GFP–Sin1 mapping constructs along with Myc–PKCϵ K437M. GFP immunoprecipitates (IP) and cell lysates were examined for GFP and PKCϵ. An indication of which Sin1 constructs efficiently co-purify PKCϵ is provided to the right on the schematic diagram (+/−). Sin FL, full-length Sin1.

To gain insight into the relationship between Sin1 and PKC, we sought to map the site of interaction and designed a series of N-terminally GFP-tagged Sin1 truncations informed by the naturally occurring splice variant boundaries (Figure 2A). As N-terminally tagged Sin1 proteins appear not to be incorporated into mTORC2 [24], this series of experiments was intended to examine specifically the interaction between Sin1 and PKC in isolation. HEK-293 cells were co-transfected with GFP–Sin1 constructs and Myc–PKCϵ K437M. GFP–Sin1 was immunoprecipitated via the GFP tag, and Western blots were analysed for associated PKCϵ (Figure 2B). The kinase-dead non-phosphorylated PKCϵ K437M mutant was used to probe for association due to its enhanced ability to bind Sin1 when compared with the WT (wild-type) kinase. Neither the N-terminal 192 amino acids comprising the mTORC-binding domain, nor the C-terminal putative PH domain were found to interact strongly with PKCϵ, although both regions did co-purify more PKC than did GFP alone. The most robust association was found between full-length Sin1 or the Sin1 central region alone comprising amino acids 193–403. Significantly, GFP-tagged Sin1 comprising the full sequence of the mSin1.5 splice variant (residues 1–321) was also found to interact robustly with PKCϵ. This isoform comprises the putative mTORC2-interaction domain, which does not efficiently bind PKC, in tandem with the highly conserved intact CRIM domain. Intriguingly, this isoform, when incorporated into mTORC2, has been shown to be sufficient to mediate in vitro Ser473 phosphorylation of PKB in contrast with mSin1.4 [24].

Phosphorylation of PKC is dependent on PDK1 and mTOR activity

To determine whether the interaction of Sin1 with PKCϵ is related to mTORC2 action as an upstream kinase for PKCϵ, we exploited the acute inducible phosphorylation of an ATP-binding-defective mutant of PKCϵ that becomes phosphorylated following inhibitor occupation of the ATP-binding pocket [20].

Cells were transfected with Dharmacon smart pools for 48 h before tetracycline induction of GFP–PKCϵ K437M expression for a further 24 h. Cells were then exposed to Bim I treatment for various times, and phosphorylation of all three priming sites was monitored by Western blot analysis. siRNA directed at the mTORC2 components Sin1, rictor and mTOR significantly reduced the ability of Bim I to induce phosphorylation compared with control siRNA (Figure 3A). Conversely, siRNA against the mTORC1 subunit raptor was unable to suppress PKC phosphorylation.

Figure 3 PKCϵ K437M phosphorylation requires mTORC2

(A) Expression of mTORC proteins was suppressed using siRNA for 48 h before induction of GFP–PKCϵ K437M expression with tetracycline for 24 h. Cells were treated with Bim I for the times indicated and phosphorylation of priming sites was assessed. Numbers beneath the blots indicate quantification of Ser729 phosphorylation relative to total GFP–PKCϵ expression. (B) Suppression of rictor, raptor and mTOR expression was assessed by Western blotting. (C) Sin1 siRNA decreases the amount of immunoreactive Sin1 detectable in rictor immunoprecipitates and induces a band shift in rictor. Assessment of Sin1 suppression in total cell lysates was hampered by non-specific immunoreactive bands. (D) Multiple distinct Sin1 siRNAs and PDK1 siRNA inhibit priming of PKCϵ K437M. Expression of Sin1 and PDK1 were suppressed using siRNA for 48 h before induction of GFP–PKCϵ K437M expression with tetracycline for 24 h. Cells were treated with Bim I for 20 min and phosphorylation of the activation loop (Thr566), turn motif (Thr710) and hydrophobic motif (Ser729) priming sites of PKCϵ were assessed by Western blotting. Sin1-directed siRNAs are: Pool 1, SiGenome SMART pool; Pool 2, ON-TARGETplus; 1&2, oligonucleotides 1 and 2 from ON-TARGETplus; 3&4, oligonucleotides 3 and 4 from ON-TARGETplus (see the Experimental section). The histogram beneath the blots represent quantification of PKCϵ K437M pSer729 from multiple experiments and show the means±S.D. (n=3). (E) GFP–PKCϵ K437M-expressing cells were treated with the indicated concentrations of PP242 (PP) or Ku0063794 (Ku) before stimulation of PKC priming phosphorylations with Bim I (2 μM) for 15 min. IP, immunoprecipitation; Veh, vehicle control.

Depletion of rictor, mTOR and raptor was demonstrated by Western blotting (Figure 3B). The depletion of Sin1 following siRNA proved problematic as the polyclonal antibody used to detect Sin1 cross-reacts with an unknown protein of a similar molecular mass confounding assessment of Sin1 loss. However, consistent with knockdown of Sin1, Sin1-directed siRNA was found to deplete the amount of Sin1 detectable in rictor immunoprecipitates (Figure 3C). Furthermore, close examination of rictor blots demonstrates a small, but significant, band shift in the apparent size of rictor in response to Sin1 siRNA, but not control siRNA, consistent with a functional disturbance of mTORC2. To verify that the suppression of PKC phosphorylation with Sin1 siRNA is indeed a consequence of Sin1 depletion, we used a separate distinct siRNA pool along with two pairs of siRNAs from this pool (Figure 3D). All Sin1-directed siRNAs tested suppressed PKC phosphorylation induced by Bim I stimulation, as did siRNA directed against the well-established PKC activation loop kinase PDK1 (Figure 3D). Finally, the specific mTOR catalytic inhibitors PP242 and Ku0063794 were found to potently block the acute Bim I-induced phosphorylation of PKCϵ K437M at all three priming sites (Figure 3E). The co-operative nature of PKC-priming phosphorylations [4,29,30] means that inhibition of any priming kinase results in loss of phosphorylation at all three sites (for a full description of this behaviour, see [20]). In contrast, mTOR inhibition has little effect on WT PKCϵ priming under the same conditions, indicative of the relative stability of the constitutive phosphorylations. Taken together, our data strongly support a role for mTORC2 and its catalytic activity in the acute priming of PKCϵ K437M.

Truncated Sin1 disrupts mTORC2 targeting of PKC and PKB

As PKC interacts with a region of Sin1 distinct from that interacting with mTORC2, we speculated that Sin1 species lacking the PKC-interaction domain, but capable of being recruited to mTORC2 might act as dominant-negatives with respect to PKC recruitment and priming, by competing with endogenous Sin1 for mTOR association (Figure 4A). To test this hypothesis, we generated stable cell lines expressing tetracycline-inducible C-terminally Myc-tagged Sin1 constructs. Expression of Sin1 constructs was assessed by Western blotting for the Myc epitope tag (Figure 4B).

Figure 4 Truncated Sin1 (Sin Δ193–522) inhibits mTORC2 targeting of PKC and PKB but not p70S6K

(A) Schematic diagram indicating disruption of the PKC-binding domain, but not the mTORC2-interaction domain, in Sin Δ193–522. (B) Myc-tagged full-length Sin1 (Sin FL) and Sin Δ193–522 are expressed in response to tetracycline induction in HEK-293/FRT stable cell lines and expression is enhanced by MG132 treatment. Molecular masses are indicated in kDa. (C) Tetracycline induction of Sin Δ193–522 expression, but not full-length Sin1 (Sin FL) blocks Bim I induced phosphorylation of PKCϵ K437M. Sin constructs were induced with tetracycline where indicated for 48 h before transient transfection with GFP–PKCϵ K437M. Cells were stimulated with Bim I for 10 min and immunoblotted for PKC hydrophobic site phosphorylation. Blots were quantified using the Odyssey system which allows simultaneous blotting for phospho-PKCϵ (pSer729) and protein loading (GFP) (example blots are provided in Supplementary Figure S1A at http://www.BiochemJ.org/bj/439/bj4390287add.htm). For a detailed description of Odyssey quantification, see the Experimental section. The effect of Sin Δ193–522 expression on the phosphorylation status of endogenous mTORC targets was assessed: (D and F) PKB pSer473 and (E and F) p70S6K pThr389. The effect of rapamycin (Rap; 100 nM) or mTOR inhibitor PP242 (1 μM) pre-treatment on PKB and p70S6K was also assessed. Data are quantified from Western blots as the ratio of phosphorylation/total protein and are the means±S.D. for at least three separate experiments. ctrl, control; FL, full-length; Rapa, rapamycin; tet, tetracycline; Veh, vehicle control.

Consistent with the proposed mechanism whereby Sin1 interaction recruits PKC, tetracycline-induced expression of a truncated mutant Sin1 comprising the mTORC2-interaction domain alone (Sin Δ193–522, denoting deletion of residues 193–522 of full-length Sin1.1) dramatically inhibited the ability of Bim I to induce phosphorylation of either PKCϵ K437M (Figure 4C and Supplementary Figure S1A) or PKCα K368M (Figure 5C, and see below) on their respective hydrophobic motif phosphorylation sites. Induction of Sin Δ193–522 also significantly suppressed the phosphorylation of endogenous PKB on Ser473 (Figures 4D and 4E), but had no effect on phosphorylation of the mTORC1 target p70S6K on Thr389 (Figures 4D and 4F). In contrast, induction of full-length Sin1 had no effect on the phosphorylation of any of these mTOR targets (Figures 4D–4F). Rapamycin effectively suppressed p70S6K, but not PKB, phosphorylation, whereas an mTOR kinase inhibitor (PP242) suppressed both of these pathways. These data suggest that PKC and PKB are targeted by mTORC2 through a conserved mechanism involving Sin1 recruitment.

Figure 5 The CRIM domain of Sin1 is essential for mTORC2 to target PKC

(A) Schematic representation of Sin1-deletion mutants. (B) Tetracycline-induced expression of Sin1 mutants was assessed by Western blotting for the Myc tag. The histogram represents expression quantified from Western blots as the Myc tag expression normalized to tubulin and are the mean±S.D. results for three separate experiments. The effect of Sin1 mutant expression on Bim I-induced phosphorylation of PKCα K368M (C) and basal constitutive PKB Ser473 phosphorylation (D) were analysed by Western blotting and quantified as described in the Experimental section. Data are means±S.D. (n=3) and are representative of at least three separate experiments. ctrl, control; FL, full-length; tet, tetracycline; Veh, vehicle control.

The CRIM domain of Sin1 is essential for mTORC2 to target PKC

In contrast with other systems for disrupting mTORC2 function, such as gene knockouts and RNA interference, the inducible Sin1 system described in the present paper can readily be exploited to investigate the functional importance of distinct regions of Sin1. Thus, to determine more accurately the region of Sin1 necessary for supporting phosphorylation of PKC, we examined the effects of inducibly expressing various Sin1-deletion mutants on mTORC2 function (Figure 5). The deletions tested were informed by existing splice boundaries (Figures 2A and 5A).

Despite the FRT T-Rex™ system being designed to generate equal expression of tetracycline-induced constructs from a single locus, significant differences were observed in steady-state protein levels of the various Sin1 mutants, which probably reflects protein stability (Figure 5B). Four informative constructs, however, were expressed to comparable levels: full-length Sin1 and Sin Δ193–522 (both examined above), along with Sin Δ0–192 (which lacks the putative mTOR-interaction domain) and Sin Δ236–249. This final mutant (Sin Δ236–249) lacks a region of the CRIM domain comprising only 14 amino acids (EDDGEVDTDFPPLD), identified for selective deletion due to the high degree of homology observed for this region between all known Sin1 orthologues [27]. As PKCα K368M consistently proved to be a more sensitive read-out for disruption of mTORC2 signalling compared with PKCϵ K437M, this was used as a primary assay for comparing Sin1 mutants (Figure 5C and Supplementary Figure S1B). Deletion of the highly conserved peptide within the CRIM domain was sufficient to disrupt mTORC2 targeting PKCα K368M for phosphorylation, whereas expression of Sin Δ0–192, which is not predicted to incorporate into mTORC2, does not. Of the remaining four less well expressed mutants, two which disrupt the CRIM domain also act as dominant-negatives towards PKCα K368M phosphorylation, whereas a deletion which disrupts both the RBD and the PH domain (Sin Δ321–403), or a truncation which just disrupts the PH domain (Sin Δ404–522), had little effect (Figure 5C).

The ability of all the Sin1 mutants to interfere with steady-state endogenous PKB phosphorylation was also assessed, demonstrating a pattern of behaviour similar to that of PKCα K368M; CRIM-disrupting mutants cause a decrease in basal PKB Ser473 phosphorylation (Figure 5D). In HEK-293/FRT cells, PKB is constitutively highly phosphorylated even under serum-starved conditions making assessment of acute PKB-directed mTORC2 activity problematic. The observed reduction in steady-state levels thus reflects a shift in the equilibrium between kinase and phosphatase action and consequently may indicate a considerably greater decrease in cellular mTORC2 activity, as is suggested by the greater effect of mutant Sin1 expression on acutely induced PKCα K368M phosphorylation. Significantly for PKB phosphorylation, the two additional mutants that disrupt the PH domain (Sin Δ321–403 and Sin Δ404–522) caused a small, but significant, decrease in steady-state PKB phosphorylation (P<0.01, n=6), whereas deletion of the CRIM region along with the C-terminal RBD and PH domains (Sin Δ193–522) was considerably more effective at suppressing PKB phosphorylation than was single domain disruptions (Figure 5D). These data perhaps suggest a role for multiple domains of Sin1 in the targeting of endogenous PKB. Notably, transient transfection with Sin1 mutants did not efficiently or consistently interfere with mTORC2 signalling when compared with the FRT T-Rex™ inducible system exploited in the present study. This perhaps reveals advantages in reproducibility and penetration afforded by chronic inducible expression from a stable genomic integration site over commonly employed transient transfection techniques.

Sin1 mutants that interfere with PKC phosphorylation displace endogenous Sin1 from mTORC2

To examine the mechanism whereby mutant forms of Sin1 interfere with PKC and PKB phosphorylation, we sought to examine the composition of endogenous mTORC2 in the inducible cell lines before and after tetracycline induction. Tetracycline treatment decreased the levels of endogenous Sin1 detectable in rictor immunoprecipitates in all of the cell lines studied, with the exception of the parental HEK-293/FRT cell line and the Sin Δ0–192 mutant (Figures 6A and 6B). Conversely, little effect was observed on levels of endogenous mTOR associated with rictor. Incorporation of full-length Myc-tagged Sin1 and Sin Δ236–249 into mTORC2 can be directly observed in rictor immunoprecipitates where the diffuse endogenous Sin1 band is replaced by a single (ectopic) Sin1-immunoreactive band (Figure 2B) of molecular mass comparable with that of full-length Sin1. The relatively low intensity of this ectopic Sin1 band compared with endogenous Sin1 is likely to be a result of the addition of the Myc tag directly in tandem with the epitope recognized by the anti-Sin1 polyclonal antibody. Unfortunately, it was not possible to detect the Myc-tagged proteins directly in the rictor immunoprecipitates because of low signal and high background problems with the anti-Myc (mAb 9E10) immunoblots.

Figure 6 Sin1 mutants displace endogenous Sin1 from mTORC2

(A) The HEK-293/FRT parental (FRT) and Sin1 mutant cell lines were treated with vehicle or tetracycline for 48 h and transiently transfected overnight before lysis and immunoprecipitation with anti-rictor and anti-Myc (mAb 9E10) antibodies or control anti-rabbit and anti-mouse IgG respectively. Immunoprecipitates were probed for the presence of Sin, rictor, mTOR and Myc (rabbit polyclonal antibody) as indicated. Ectopic Myc-tagged full-length Sin1 and Sin Δ236–249 in rictor immunoprecipitates are indicated by *. Myc-tagged constructs in the immunoprecipitations are indicated by >. Expression of Myc constructs and endogenous mTOR and rictor were also assessed in cell lysates. (B) The amount of Sin1 associated with rictor was quantified and data are expressed as the ratio of Sin/rictor signal of tetracycline-treated cells relative to vehicle-treated controls. Data are means±S.D. (n=3) and are representative of three separate experiments. (C) Sin1, rictor or control rabbit IgG immunoprecipitates were probed for the presence of Sin1, rictor and mTOR as indicated. The anti-Sin1 polyclonal antibody was found to non-specifically immunoprecipitate and detect an unknown 60 kDa protein in addition to Sin1 (between arrows labelled Sin1). IP, immunoprecipitation; Tet, tetracycline.

To overcome this problem, we immunoprecipitated the Myc-tagged Sin constructs from each of the cell lines and immunoblotted for the presence of endogenous rictor and mTOR. All Myc-tagged Sin1 constructs were found to co-purify both mTOR and rictor, with the exception of Sin Δ0–192. Expression and purification of Myc-tagged constructs following tetracycline induction was confirmed by immunoblotting using a rabbit polyclonal antibody directed against the Myc tag. This, together with the inability of Sin Δ0–192 to displace endogenous Sin1 from mTORC2, confirms this region as the mTORC2-interaction domain. These data support a model where expression of Sin1 mutants bearing the mTORC2-interaction domain can replace endogenous Sin1 in mTORC2. Such a model is supported further by the loss of both rictor and mTOR from Sin immunoprecipitates following induction of Sin Δ193–522 (Figure 6C). This mutant lacks the Sin1 polyclonal epitope, indicating that mTORC2 no longer contains endogenous Sin1 following tetracycline induction. Close examination of Sin1 immunoprecipitates also reveals loss of endogenous Sin1 bands following mutant expression, suggesting that displaced Sin1 may be degraded when not in a complex with rictor and mTOR (Figure 6C). The additional Sin1-immunoreactive band in Sin1 immunoprecipitates, which does not co-purify with rictor, is a non-specific protein recognized by the polyclonal antibody (see above). To assess the extent of endogenous Sin1 displacement, the relative amount of Sin1 co-precipitated with rictor pre- and post-tetracycline induction was quantified (Figure 6B). Interestingly, all mutants bearing the mTORC2-interaction domain displaced more than 60% of endogenous Sin1, despite significant variation in total mutant expression levels (Figure 5B).

In summary, we have mapped the interaction of PKC with Sin1 to the CRIM domain. Integration of multiple CRIM domain mutants into mTORC2 mutants interferes with the targeting of PKC for phosphorylation. Extensive attempts were made to co-purify PKC in complex with intact mTORC2 and the various Sin1 mutant complexes to directly assess the effect of various mutants on target recruitment. In line with previous studies [17,19], we were unable to purify endogenous PKC or PKB with mTORC2. In some experiments, the non-phosphorylated PKC K437M mutant was recovered with intact mTORC2 in a manner that was inhibited by expression of truncated Sin Δ193–522 (Supplementary Figure S1C). However, this complex appears to be unstable and is easily lost during immunoprecipitation wash steps. These experiments were consequently difficult to reproduce sufficiently to accurately assess the effect of selective CRIM disruption on recruitment to intact mTORC2. Taken together, however, our data strongly support a mechanism where interaction between PKC and the CRIM domain is critical for productive recruitment to mTOR.

mTORC2 positively regulates cell proliferation

The selective inhibitory effect on mTORC2 afforded by truncated Sin1 induction provides a unique opportunity for examining the effect of sustained suppression of mTORC2 activity in an isogenic context. To probe for a role in proliferation and cell survival, the effects of prolonged expression of truncated Sin1 Δ193–522 were examined under various conditions of cell culture. The Sin Δ193–522 mutant was selected as it was consistently the most efficient at uncoupling mTORC2 signalling to PKC and PKB. Under normal adherent two-dimensional cell culture growth conditions, suppression of mTORC2 with truncated Sin1 had only a modest effect on cell viability when compared with cells expressing full-length Sin1 (Figures 7A and 7B). In contrast, colony growth in Matrigel™ suspension cultures was significantly inhibited by expression of truncated, but not full-length, Sin1 (Figures 7C and 7D). These data reveal an important role for mTORC2 in non-adherent transformed cell growth.

Figure 7 Expression of dominant-negative Sin1 inhibits cell proliferation

(A) Full-length Sin (Sin FL) cells or (B) Sin Δ193–522 cells were seeded into 96-well tissue culture plates and treated with vehicle or tetracycline (100 nM). Cell viability was assessed using Cell Titer Glo (Promega) at 3, 5 and 7 days. Data are means±S.D. (n=4). (C) Sin Δ193–522 decreased spheroid growth in Matrigel™ at 5 days. The mean cross-sectional area was measured in at least 150 spheroids per condition in each of the four duplicate experiments performed. The population of spheroids was non-Gaussian, so error bars indicate the range. Non-parametric statistical analysis demonstrated a highly significant difference (***P<0.0001). (D) Representative phase images of HEK-293/FRT spheroids after 5 days of growth in Matrigel™ with and without tetracycline-induced expression of Sin Δ193–522. Tet/tet, tetracycline; Veh, vehicle control.

DISCUSSION

The mTOR pathway has emerged as an important clinical target particularly in cancer therapy [31]. Compounds acting to block mTOR fall into two distinct categories on the basis of their specificity and mechanism of action. First, rapamycin and its analogues selectively block activity of mTORC1 indirectly through an interaction with the mTOR-binding protein FKBP12. Alternatively, both arms of the mTOR pathway can be targeted with more classical ATP-pocket-directed inhibitors of the mTOR kinase itself. Although rapamycin analogues have been in clinical use for some time, showing efficacy against some tumour types, increasing focus is falling on the development of mTOR kinase inhibitors as they block both mTORCs [3235]. The discovery that mTORC2 plays a key role in regulating the activity of both PKC and PKB signalling has stimulated speculation concerning the potential benefits of blocking both mTORC1 and mTORC2 activity directly in the treatment of cancer. The insights derived from the present study have not only provided an understanding of mTORC2 action, but also enabled selective intervention providing direct evidence to support the targeting of mTORC2 in blocking the growth of this transformed cell model. It will be interesting to transfer this system to distinct cell types and in vivo tumour models to examine the effects of suppressing mTORC2 in more clinically relevant settings.

Studying PKC-priming phosphorylation has historically been hampered for two key reasons. First, unlike PKB, the priming sites of PKC are generally constitutively occupied, with acute regulation of activity being primarily driven allosterically through regulatory domain interactions. Secondly, once the priming sites are occupied, they are very stable and thus, under normal in vitro growth conditions, inhibition of potential upstream kinase pathways is generally not sufficient to impinge on their steady-state occupancy. Although prolonged exposure with mTOR inhibitors such as PI-103 [19], or indeed genetic disruption of mTORC2 [36], does result in reduced endogenous PKC phosphorylation, such experiments are limited by their time scales and hampered by the potential for secondary effects. In contrast, PKB phosphorylation is much more accessible to study due to the acute and reversible regulation of both the Thr308 and Ser473 sites in response to growth factors. This property is also reflected in the ability of our Sin1-dominant constructs to reduce endogenous PKB phosphorylation on Ser473. Here a reduction in mTORC2 activity is reflected in a reduction in mTORC2 target phosphorylation.

Our recent description of the dependence of PKC-priming phosphorylations on the conformation of the nucleotide-binding pocket has provided a route to the dissection of upstream PKC kinases. Here, PKC priming of ATP-binding mutants (PKCϵ K437M and PKCα K368M) can be acutely regulated by ATP-competitive PKC inhibitors [20,37]. As this system is directly allosteric in mechanism, stimulation of priming does not depend on intermediate signalling steps as is the case for example with PKB. Using this inhibitor-inducible PKC phosphorylation as a sensitive readout for mTORC2 activity, we have been able to probe the function of the mTORC2 subunit Sin1. Our data suggest that the CRIM domain of Sin1, conserved among all known Sin orthologues, acts to recruit target kinases to mTORC2. Expression of mutants that lack this region displace endogenous Sin1 from mTORC2 and render the complex impotent. An interaction between Sin1 and PKB has been demonstrated previously by co-immunoprecipitation and a role as a scaffold has been proposed [17]. Additionally, two studies have provided evidence that the mTORC2 subunit rictor interacts with PKC isoforms [19,38]. Our identification of Sin1 as a direct interactor with PKC and that disruption of this interaction blocks phosphorylation of PKC and PKB by mTORC2 despite retention of rictor in that complex provides the first evidence that Sin1-driven recruitment is of key functional importance.

Recent studies have provided evidence that mTORC2 can both associate with and be regulated by ribosomes [39,40]. mTORC2 has been proposed to phosphorylate nascent PKB on its turn motif (Thr450) to prevent its ubiquitination and degradation and a similar mechanism is proposed to occur for PKCα which is degraded in the absence of priming phosphorylations [40]. Association with the ribosome has also been shown to regulate Ser473 PKB phosphorylation in response to insulin and oncogenic PI3K signalling [39]. The present study demonstrates that, whereas phosphorylation of nascent PKC in principle might occur co-translationally, phosphorylation of the PKC C-terminal sites can also occur post-translationally. As kinase-dead PKCϵ K437M and PKCα K368M localize at the plasma membrane, it seems likely that mTORC2 can act on PKC in this compartment, as is the case for phosphorylation of PKB on its hydrophobic motif in response to growth factor stimulation. With this in mind, it is perhaps a little surprising that disruption of the Sin1 PH domain, known to bind the plasma membrane [28], had little affect on the ability of mTORC2 to phosphorylate PKC in our model system (Figure 3). Furthermore, as the inhibitor-induced phosphorylation of PKC acts directly though an allosteric mechanism, it implies that constitutive mTORC2 activity in this compartment is not rate-limiting. In addition, culture of cells under conditions of nutrient or growth factor starvation has little impact on the ability of mTORC2 to target PKC in this model, implying that it is PKC conformation rather than modified mTORC2 activity which determines the degree of priming site occupancy under these conditions. A role for ribosome-associated mTORC2 in the acute or biosynthetic pathway of PKC phosphorylation remains to be elucidated.

No compounds are currently available showing selectivity for mTORC2. Elucidating the roles specific to mTORC2 has thus relied on genetic ablation of mTORC2-specific components such as rictor and Sin1 either in the form of mouse knockouts or through suppression of expression with RNA interference technologies. The dominant-negative Sin1 constructs described in the present paper thus provide a unique system for assessing the potential relevance of targeting mTORC2 activity while leaving the mTORC1 pathway intact. As an additional benefit, introduction of mutated Sin1 constructs does not physically disrupt mTORC2, but rather uncouples mTORC2 from some or all of its downstream targets. In contrast, deletion of Sin1 disrupts the interaction between rictor and mTOR with unknown consequences [17]. By maintaining dominant constructs under tetracycline-inducible control, we are able to suppress mTORC2 activity chronically without the well-documented off-target issues associated with RNA-suppression technologies. Our studies using this approach reveal a role for mTORC2 in mediating cell proliferation in suspension cultures with obvious implications for cancer therapy.

AUTHOR CONTRIBUTION

Angus Cameron conceived and executed the bulk of the study and co-wrote the paper. Mark Linch performed the three-dimensional growth assays. Adrian Saurin identified Sin1 in a yeast two-hybrid screen. Cristina Escribano performed initial phosphorylation experiments. Peter Parker conceived the study and co-wrote the paper.

FUNDING

We would like to acknowledge Cancer Research UK for financial support.

Abbreviations: CRIM, conserved region in middle; EGFP, enhanced green fluorescent protein; FKBP12, FK506-binding protein 12; GFP, green fluorescent protein; HA, haemagglutinin; HEK, human embryonic kidney; HRP, horseradish peroxidase; mAb, monoclonal antibody; mTOR, mammalian target of rapamycin; mTORC, mTOR complex; PDK1, phosphoinositide-dependent kinase 1; PH, pleckstrin homology; PI3K, phosphoinositide 3-kinase; PKB, protein kinase B; PKC, protein kinase C; raptor, regulatory associated protein of mTOR; RBD, Ras-binding domain; rictor, rapamycin-insensitive companion of mTOR; S6K, S6 kinase; siRNA, short interfering RNA; TOR, target of rapamycin; TORC, TOR complex; WT, wild-type

References

View Abstract