Research ArticleCell Biology

ER Stress Inhibits mTORC2 and Akt Signaling Through GSK-3β–Mediated Phosphorylation of Rictor

See allHide authors and affiliations

Science Signaling  22 Feb 2011:
Vol. 4, Issue 161, pp. ra10
DOI: 10.1126/scisignal.2001731

Abstract

In response to environmental cues, cells coordinate a balance between anabolic and catabolic pathways. In eukaryotes, growth factors promote anabolic processes and stimulate cell growth, proliferation, and survival through activation of the phosphoinositide 3-kinase (PI3K)–Akt pathway. Akt-mediated phosphorylation of glycogen synthase kinase–3β (GSK-3β) inhibits its enzymatic activity, thereby stimulating glycogen synthesis. We show that GSK-3β itself inhibits Akt by controlling the mammalian target of rapamycin complex 2 (mTORC2), a key activating kinase for Akt. We found that during cellular stress, GSK-3β phosphorylated the mTORC2 component rictor at serine-1235, a modification that interfered with the binding of Akt to mTORC2. The inhibitory effect of GSK-3β on mTORC2-Akt signaling and cell proliferation was eliminated by blocking phosphorylation of rictor at serine-1235. Thus, in response to cellular stress, GSK-3β restrains mTORC2-Akt signaling by specifically phosphorylating rictor, thereby balancing the activities of GSK-3β and Akt, two opposing players in glucose metabolism.

Introduction

Growth factor signaling coordinates various cellular processes, including proliferation, metabolism, and survival (1). Akt is a critical downstream effector of the growth factor–dependent phosphoinositide 3-kinase (PI3K) pathway. How growth factors regulate Akt signaling has been studied intensively, and it has become clear that growth factor signaling depends on the cellular environment and is sensitive to multiple stress pathways (2). For example, the anabolic effects of growth factors (such as insulin) are opposed by the extent of endoplasmic reticulum (ER) stress, which causes activation of catabolic reactions associated with autophagy (3). ER stress coupled to the unfolded protein response is a frequently occurring pathophysiological state associated with human cancers as well as with metabolic and neuronal disorders (4, 5). Presently, how cellular stress conditions suppress growth factor signaling is not well understood.

After activation of PI3K, which generates phosphatidylinositol 3,4,5-trisphosphate (PIP3), Akt translocates to the plasma membrane by binding to PIP3 through its N-terminal pleckstrin homology (PH) domain (6). In addition to the Akt PH domain, ubiquitination of Akt also facilitates its membrane translocation (7). At the plasma membrane, Akt is phosphorylated at two sites; these phosphorylation events are required for its full activation. The Thr308 site resides in the activation loop and is phosphorylated by the phosphoinositide-dependent kinase 1 (PDK1) (8, 9), and the Ser473 “hydrophobic motif” site is located at the C terminus of Akt (6). mTORC2 (mammalian target of rapamycin complex 2) is a major kinase that phosphorylates the Ser473 site in Akt (10), and the essential mTORC2 component, rictor, is required for phosphorylation of Ser473 in cells (1114). Under conditions that induce DNA damage, the DNA protein kinase (DNA-PK) also phosphorylates Ser473 in Akt (15).

mTOR is a protein kinase that functions as a central component of an essential, conserved growth pathway. Biochemical studies reveal that mTOR and its interacting proteins mLST8 and DEPTOR [DEP (disheveled, Egl-10, pleckstrin; protein) domain–containing mTOR-interacting protein] exist in at least two distinct complexes. The binding of raptor to mTOR defines the first nutrient-sensitive complex, mTORC1, which regulates protein synthesis by phosphorylating its substrates S6 kinase 1 (S6K1) and eIF4E-binding protein 1 (4E-BP1). The second mTOR complex, mTORC2, is assembled by binding of rictor and Sin1 [stress-activated protein kinase (SAPK)–interacting protein] to mTOR. The mTORC2 kinase complex prefers to phosphorylate members of the AGC (protein kinase A, G, and C) family on the hydrophobic motif site located at the C terminus (16). This feature is characteristic of mTORC2 substrates, such as Akt (10) and SGK (serum- and glucocorticoid-induced protein kinase) (17) and its downstream effector PKCα (protein kinase Cα) (12, 18). The regulation of mTORC2 remains poorly characterized. Previously, the Raf-like Ras binding and PH domains of Sin1 have been proposed to regulate Sin1 activity (19). However, the precise mechanism or mechanisms by which Sin1 regulates mTORC2 have yet to be determined. Sequence analysis of the 1708–amino acid human rictor polypeptide does not reveal homology to known functional domains in other proteins (18). In previous studies, several rictor phosphorylation sites have been identified, and the Thr1135 site attracted attention as a growth factor– and S6K1-regulated site (2023). However, the functional importance of the Thr1135 site remains unknown because it does not appear to affect the kinase activity of mTORC2 (23).

Here, we aimed to characterize the regulation of mTORC2-Akt signaling by ER stress. Our work identified the mechanism by which ER stress induces phosphorylation of rictor at a specific site to prevent phosphorylation and activation of Akt by mTORC2.

Results

Inhibition of Akt signaling by ER stress is associated with phosphorylation of Ser1235 in rictor

Deregulation of Akt signaling by various stress conditions is common to the pathogenesis of human diseases such as diabetes and cancer (25). To address whether mTORC2 signaling is sensitive to cellular stress, we analyzed the phosphorylation of Akt at Ser473, the kinase activity of mTORC2, and the phosphorylation of rictor after osmotic stress. We studied the well-defined mTORC2 substrate Akt (10) because its regulation is sensitive to growth factor and stress response signaling. Another mTORC2 substrate, SGK, was not addressed in this study because of its low abundance (17). Acute osmotic stress induced by incubating cells with high-osmolarity sorbitol inhibited mTORC2-Akt signaling, as shown by a decrease in mTORC2 kinase activity and in phosphorylation of Akt at its regulatory Ser473 site (Fig. 1A). High osmolarity as induced by sorbitol also decreased S6K1-dependent phosphorylation of Thr1135 in rictor. In contrast, phosphorylation of rictor at Ser1235 (20) was increased after osmotic stress, as detected in cell lysates or in rictor immunoprecipitates (Fig. 1A). Our phosphospecific antibody did not detect a form of rictor with a mutation at the phosphorylation site (fig. S1A), thus validating its specificity in detecting rictor phosphorylated at Ser1235. Osmotic adsorption of water from cells is associated with induction of ER stress (24). Tunicamycin and thapsigargin are two well-characterized compounds that cause ER stress. To extend our cellular data, we induced ER stress in mice by injection of tunicamycin or thapsigargin (25). ER stress was detected in liver extracts, as assessed by phosphorylation of the ER stress marker PERK (protein kinase RNA–like endoplasmic reticulum kinase) at Thr980 (Fig. 1B). Tunicamycin or thapsigargin treatment also reduced phosphorylation of Akt at Ser473, while increasing that of rictor at Ser1235 (Fig. 1B). Thus, these results suggest that Ser1235 in rictor is a previously unknown ER stress–inducible phosphorylation site.

Fig. 1

Inhibition of mTORC2 signaling by ER stress is linked to rictor phosphorylation. (A) Inhibition of mTORC2 activity by osmotic stress correlates with rictor Ser1235 phosphorylation. HEK 293T cells were incubated with 0.5 M sorbitol for 30 min to induce osmotic stress. Rictor immunoprecipitates (IP) were used for in vitro kinase assays with full-length wild-type (WT) Akt1 as a substrate. Immunoblotting was used to detect phosphorylation of Akt at Ser473 and rictor at Ser1235 and Thr1135, as well as the amounts of Akt, rictor, Sin1, or α-tubulin in the kinase assays, immunoprecipitates, and cell lysates. (B) Regulation of rictor Ser1235 phosphorylation by ER stress in mice. Thapsigargin (1 μg/g in PBS) or tunicamycin (2 μg/g in 150 mM glucose) was injected intraperitoneally into 4-month-old mice for 24 hours to induce ER stress. Tissue extracts prepared from liver were subjected to SDS-PAGE and were immunoblotted to detect the phosphorylation of PERK at Thr980, phosphorylation of Akt at Ser473, phosphorylation of rictor at Ser1235, and the total abundance of rictor, PERK, and Akt. A blot representative of n = 3 independent experiments is shown for each panel; the data are means ± SD for triplicate measurements shown in table S1.

GSK-3 inhibition prevents ER stress–dependent phosphorylation of rictor at Ser1235 and promotes mTORC2 kinase activity

Alignment of rictor orthologs indicates that the Ser1235 site is conserved in vertebrates (fig. S1B). Examination of the sequence surrounding Ser1235 in rictor reveals two serine residues preceding Ser1235 that is followed by proline and serine residues (SS-S1235-PS). Glycogen synthase kinase–3 (GSK-3) is one kinase that is activated by ER stress (26). It phosphorylates Ser394 in BCL3 (B cell lymphoma 3–encoded protein) in a sequence (SS-S394-PS) resembling that surrounding Ser1235 in rictor (27). Therefore, we considered the possibility that stress-inducible phosphorylation of Ser1235 in rictor might also be regulated by GSK-3. Treating cells with thapsigargin or tunicamycin induced phosphorylation of PERK and Ser1235 in rictor (Fig. 2A). Moreover, these treatments also activated GSK-3, as indicated by both the induction of tyrosine phosphorylation of GSK-3 (28, 29) and the dephosphorylation of the inhibitory Ser9 site, which is phosphorylated by Akt (Fig. 2A). Additionally, ER stress–induced GSK-3 activation correlated with increased phosphorylation of a GSK-3 substrate, glycogen synthase (GS), at Ser641 (Fig. 2A). Treatment of stressed cells with the GSK-3 inhibitor 6-bromoindirubin-3′-oxime (BIO) (30) did not reduce PERK phosphorylation, but did block the induction of phosphorylation of Ser1235 in rictor, suggesting that GSK-3 might regulate this phosphorylation event. We also observed that ER stress–induced Akt dephosphorylation was dependent on GSK-3 activity, indicating an inverse correlation between rictor Ser1235 and Akt Ser473 phosphorylation. Conversely, phosphorylation of rictor at Thr1135 was slightly increased by BIO treatment (fig. S2A), which is consistent with the known effects of Akt on mTORC1 activation and phosphorylation of S6K1 at Thr389 (31, 32).

Fig. 2

ER stress–dependent phosphorylation of rictor at Ser1235 and mTORC2 signaling requires GSK-3 activity. (A) GSK-3 mediates ER stress–induced phosphorylation of rictor at Ser1235. MDA-MB-435 cells were treated with 2 μM thapsigargin (TG) or tunicamycin (TM) (2 μg/ml) for 6 hours, and then with 5 μM of the GSK-3 kinase inhibitor BIO 3 hours before cell harvest. Immunoblotting was used to detect phosphorylation of rictor at Ser1235, phosphorylation of Akt at Ser473, phosphorylation of PERK at Thr980, phosphorylation of glycogen synthase (GS) at Ser641, phosphorylation of GSK-3β at Tyr216 or Ser9, and the total abundance of rictor, Akt, GS, GSK-3β, or α-tubulin in cell lysates. CTL, control. (B) GSK-3 inhibitors induce Akt activation. MDA-MB-435 cells were incubated with 5 μM BIO or 10 μM SB-216763 for 5 hours under low- or high-serum (1 or 10% FBS, respectively) conditions. (C) GSK-3 inhibition induces mTORC2 kinase activity. MDA-MB-435 cells were incubated with BIO or SB-216763 as in (B) under low-serum conditions. Rictor immunoprecipitates were used for in vitro kinase assays with full-length WT Akt1 as a substrate. In (B) and (C) the samples were analyzed as in (A). A blot representative of n = 3 independent experiments is shown for each panel; the data are means ± SD for triplicate measurements shown in table S1.

That ER stress–induced GSK-3 activation leads to phosphorylation of rictor at Ser1235 and dephosphorylation of Akt implies rictor Ser1235 might mediate the inhibitory effects of GSK-3 on mTORC2-Akt signaling. We determined whether GSK-3 inhibited Akt activity in cells grown in culture medium containing low (1%) or high (10%) serum. Two distinct GSK-3 inhibitors, BIO or SB-216763 (33), induced phosphorylation of Akt at Ser473 and Thr308 in both low- and high-serum conditions (Fig. 2B). Thus, GSK-3 inhibition is sufficient to stimulate Akt signaling in multiple cell contexts and is consistent with previous findings that BIO and SB-216763 promote Akt-mediated phenotypes, such as increased cell proliferation and glucose uptake (34, 35).

We wondered whether increases in Akt phosphorylation resulting from GSK-3 inhibition might be linked to alterations in mTORC2 activity because both GSK-3 inhibitors caused dephosphorylation of rictor at Ser1235. To test this notion, we isolated mTORC2 complexes from cells treated with BIO, SB-216763, or vehicle and performed mTORC2 kinase assays with Akt as the substrate. We detected robust phosphorylation of Akt by mTORC2 when mTORC2 was purified from cells treated with either GSK-3 inhibitor (Fig. 2C). BIO treatment had the most potent effect on mTORC2, increasing phosphorylation of Akt by mTORC2 at least twofold, and SB-216763 treatment stimulated phosphorylation of Akt by mTORC2 by ~80% (fig. S2B). These results suggest that GSK-3 inhibition increases phosphorylation of Akt by mTORC2; furthermore, this effect was consistently accompanied by dephosphorylation of rictor at Ser1235 (Fig. 2C).

We wondered whether the regulation of TORC2 and Akt by GSK-3 might be conserved in the invertebrate fly Drosophila melanogaster. Analysis of rictor homolog amino acid sequences showed that the first 1000 amino acids within the human polypeptide of 1708 amino acids are conserved across eukaryotes (18). The Drosophila rictor sequence shows poor alignment with the human rictor sequence following this conserved sequence; however, a small stretch of 19 amino acids within the sequence surrounding the human rictor Ser1235 site (from amino acids 1209 to 1238) does align with its Drosophila ortholog (from amino acids 1209 to 1237) with a potential phosphorylation site at Ser1233 (fig. S3A). To test whether the effects of GSK-3 on TORC2 and Akt signaling are conserved, we inhibited GSK-3 in Drosophila cells. Under low-serum conditions, phosphorylation of Drosophila Akt (dAkt) at the TORC2-dependent Ser505 site (10) was increased after incubation with BIO (fig. S3B). Only one gene in Drosophila encodes GSK-3, which we knocked down with double-stranded RNA (dsRNA)–mediated RNA interference (RNAi) (10). Similar to the effects of BIO, GSK-3 knockdown induced robust phosphorylation of Akt at Ser505 (fig. S3C), suggesting that GSK-3 inhibits Akt activity in Drosophila cells. Moreover, phosphorylation of dAkt at Ser505 was inhibited by ER stress, and this inhibition required GSK-3 (fig. S3D).

Phosphorylation of rictor at Ser1235 requires GSK-3β

Our findings thus far indicated that ER stress–mediated rictor phosphorylation and mTORC2 activity depended on GSK-3 activity. In mammals, GSK-3 has two homologous but functionally distinct isoforms, GSK-3α and GSK-3β. It is unlikely that BIO and SB-216763 are isoform-specific inhibitors because they target the adenosine 5′-triphosphate (ATP) binding pocket, which is structurally similar in GSK-3α and GSK-3β (33). To determine whether both GSK-3 isoforms are required for phosphorylation of rictor at Ser1235, we depleted each isoform from cells, using two distinct short hairpin RNAs (shRNAs) specifically targeting GSK-3α or GSK-3β (10) (Fig. 3A). We observed that knockdown of GSK-3β, but not GSK-3α, inhibited phosphorylation of rictor at Ser1235. To further address the role for GSK-3β in regulating phosphorylation of rictor and Akt, we used GSK-3β–null mouse embryonic fibroblasts (MEFs). In wild-type MEFs, switching from high- to low-serum conditions induced phosphorylation of rictor at Ser1235, and this increased phosphorylation was associated with decreased phosphorylation of Akt at Ser473. In contrast, switching GSK-3β–null MEFs from high to low serum did not induce phosphorylation of rictor at Ser1235, and mTORC2-dependent Akt phosphorylation remained detectable in these cells (Fig. 3B). Next, we assessed ER stress responses in both wild-type and GSK-3β–null cells. Although wild-type cells had an ER stress response similar to that of human cancer cells (Fig. 2A), induction of ER stress in the GSK-3β–null cells did not trigger phosphorylation of rictor at Ser1235 or perturb Akt signaling (Fig. 3C). These data indicate a role for GSK-3β in regulating ER stress–dependent phosphorylation of rictor and Akt. Several groups have found that the two GSK-3 isoforms have nonoverlapping functions (3638). Our study thus expands on this work by demonstrating that GSK-3β, and not GSK-3α, mediates phosphorylation of rictor at Ser1235 and inhibits mTORC2-Akt signaling during ER stress.

Fig. 3

GSK-3β is required for ER stress–induced phosphorylation of rictor at Ser1235 and inhibition of Akt signaling. (A) Phosphorylation of rictor at Ser1235 in cells requires GSK-3β. shRNAs targeting luciferase, GSK-3α, or GSK-3β were lentivirally transduced into MDA-MB-435 cells. Immunoblotting was used to analyze the phosphorylation of rictor at Ser1235 and the total amounts of rictor, GSK-3α, or GSK-3β in cells with stable decreases in GSK-3α or GSK-3β abundance. (B) Phosphorylation of rictor at Ser1235 was not detected and phosphorylation of Akt at Ser473 phosphorylation was increased in GSK-3β–null MEFs. GSK-3β+/+ and GSK-3β−/− MEFs were incubated with 10 or 1% FBS–containing medium for 16 hours. Immunoblotting was used to detect the total abundance of the indicated proteins in cell lysates. (C) GSK-3β–null MEFs show resistance to ER stress–dependent Akt inactivation. GSK-3β+/+ and GSK-3β−/− MEFs were treated with tunicamycin (2 μg/ml) for 6 hours. Immunoblotting was used to detect the abundance of the indicated proteins in cell lysates. A blot representative of n = 3 independent experiments is shown for (B) and (C); the data are means ± SD for triplicate measurements shown in table S1.

GSK-3β phosphorylates rictor at Ser1235 and decreases phosphorylation of Akt by mTORC2

Because GSK-3β–dependent phosphorylation of rictor leads to decreased phosphorylation of Akt by mTORC2, we hypothesized that manipulation of GSK-3β abundance might also alter mTORC2 function. Indeed, overexpression of wild-type or constitutively active GSK-3β was sufficient to inhibit Akt phosphorylation by mTORC2 and was associated with increased phosphorylation of rictor at Ser1235 (Fig. 4A). Because rictor Ser1235 phosphorylation in cells depends on the presence of GSK-3β, we tested whether GSK-3β might phosphorylate rictor in vitro. To address this question, we performed in vitro mTORC2 kinase assays in the presence or absence of active GSK-3β. We found that Akt phosphorylation by mTORC2 was decreased in the presence of GSK-3β, and concomitantly, Ser1235 in rictor was robustly phosphorylated (Fig. 4B). Moreover, these effects were sensitive to GSK-3 inhibition, suggesting that GSK-3, and not a contaminating kinase, was responsible for the decreased phosphorylation of Akt. Thus, GSK-3β inhibits phosphorylation of Akt by mTORC2 in a manner that inversely correlates with rictor Ser1235 phosphorylation.

Fig. 4

GSK-3β phosphorylates rictor at Ser1235 and inhibits phosphorylation of Akt by mTORC2. (A) GSK-3β overexpression enhances phosphorylation of rictor at Ser1235 and inhibits phosphorylation of Akt by mTORC2. Rictor immunoprecipitates were prepared from lysates of COS-7 cells transfected with WT or mutant GSK-3β and used for in vitro kinase assays with full-length WT Akt1 as a substrate. Immunoblotting was used to detect the phosphorylation of Akt at Ser473, phosphorylation of rictor at Ser1235, and the amounts of indicated proteins in the kinase assay, immunoprecipitates, and cell lysates. (B) GSK-3β phosphorylates rictor at Ser1235 and inhibits phosphorylation of Akt by mTORC2. GSK-3β was immunopurified from HEK 293T cells transfected with GSK-3β; cells were treated with tunicamycin (2 μg/ml) for 3 hours before harvest. Rictor immunoprecipitates were prepared from MDA-MB-435 cell lysates. In vitro GSK-3β kinase assays were performed on rictor immunoprecipitates. As a control, GSK-3β was preincubated with 10 μM BIO before the kinase reaction. In vitro mTORC2 kinase assays were subsequently performed on rictor immunoprecipitates. Immunoblotting was used to detect the phosphorylation of Akt at Ser473, phosphorylation of rictor at Ser1235, and the amounts of indicated proteins in the kinase assay and immunoprecipitates. A blot representative of n = 3 independent experiments is shown for each panel; the data are means ± SD for triplicate measurements shown in table S1.

GSK-3β–dependent phosphorylation of rictor at Ser1235 interferes with substrate binding to mTORC2

We hypothesized that if Ser1235 in rictor is a major GSK-3β regulatory site in the mTORC2 complex, mutation of this site should affect GSK-3β–mediated regulation of mTORC2 phosphorylation of Akt. Upon expressing either wild-type rictor or its nonphosphorylatable Ser1235→Ala (S1235A) phospho-mutant in rictor-null MEFs at equivalent abundances, we observed that mTORC2-dependent phosphorylation of Akt was higher in cells expressing the rictor phospho-S1235A mutant compared to those expressing wild-type rictor (Fig. 5A). The coexpression of GSK-3β with wild-type rictor caused reduced phosphorylation of Akt at Ser473 (Fig. 5A). In contrast, mTORC2-dependent Akt phosphorylation was not affected by coexpression of the S1235A rictor mutant and GSK-3β (Fig. 5A). Therefore, mutation of the Ser1235 site in rictor is sufficient to prevent the inhibitory effect of GSK-3β on mTORC2 signaling.

Fig. 5

Phosphorylation of rictor at Ser1235 inhibits mTORC2 kinase activity by interfering with mTORC2 binding to Akt. (A) GSK-3β inhibits Akt phosphorylation by phosphorylation of rictor at Ser1235. Rictor-null MEFs grown in 10% serum were transfected with cDNAs encoding WT rictor or the S1235A mutant and with increasing amounts of GSK-3β cDNA. Immunoblotting was used to detect the amounts of the indicated proteins in cell lysates. (B) Stable expression of the rictor S1235A phospho-mutant shows high basal phosphorylation of Akt that is insensitive to ER stress. Rictor-null MEFs constitutively expressing WT rictor or the phospho-mutants were treated with tunicamycin (0.3 μg/ml) for 8 hours. Samples were analyzed as in (A). (C) Increased phosphorylation of Akt is observed with mTORC2 complexes with the rictor S1235A mutant. Rictor immunoprecipitates from rictor-null MEFs stably expressing WT rictor or the phospho-mutants and grown in 10 or 1% serum for 16 hours were used for in vitro kinase assays. The phosphorylation of Akt at Ser473 and the amounts of the indicated proteins were analyzed as in (A). (D) ER stress inhibits mTORC2 substrate binding. MDA-MB-435 cells were treated with tunicamycin or sorbitol (Sorb.) to induce stress. Rictor immunoprecipitates were used in in vitro substrate binding assays with WT GST-Akt as substrate and analyzed by immunoblotting as in (A). (E) Phosphorylation of rictor at Ser1235 regulates mTORC2 substrate binding. Rictor immunoprecipitates from rictor-null MEFs stably expressing WT rictor or the phospho-mutants were used for in vitro substrate binding assays with WT GST-Akt as substrate and analyzed as in (A). A blot representative of n = 3 independent experiments is shown for each panel; the data are means ± SD for triplicate measurements shown in table S1.

Similarly, we detected high basal phosphorylation of Akt at Ser473 in cells stably expressing the rictor S1235A mutant compared to those expressing wild-type rictor or the Ser1235→Asp (S1235D) mutant (Fig. 5B). GSK-3 inhibition by BIO in cells expressing the rictor S1235A mutant did not further increase Akt phosphorylation, suggesting that GSK-3β–dependent stimulation of mTORC2 can be saturated in these cells (fig. S2C). We also examined ER stress responses in the MEFs reconstituted with wild-type or phospho-mutant rictor. In each cell line, tunicamycin treatment equivalently induced ER stress and activated GSK-3β (Fig. 5B). However, Akt phosphorylation was insensitive to ER stress in cells expressing the rictor S1235A phospho-mutant (Fig. 5B), thus suggesting that the Ser1235 site in rictor is critical for controlling the ER stress–induced effects of GSK-3β on mTORC2 and Akt. Furthermore, cells expressing the rictor S1235D mutant (which mimics constitutive phosphorylation of Ser1235) showed sensitivity to ER stress similar to that detected in the cells expressing wild-type rictor. It suggests that ER stress exerts its inhibitory effect on Akt signaling not only by inhibiting mTORC2-dependent phosphorylation but also by interfering with growth factor–dependent PI3K signaling upstream of this kinase complex.

Several possible mechanisms might be responsible for the effects of the phosphorylation status of Ser1235 in rictor on mTORC2 function. Decreased phosphorylation of Ser1235 in rictor might increase the abundance of mTORC2; alternatively, it might alter its ability to phosphorylate Akt. To determine which hypothesis was correct, we characterized the composition and activity of mTORC2 complexes containing either wild-type rictor or its phospho-mutant forms. We detected similar amounts of mTORC2 complexes in rictor immunoprecipitates from MEFs expressing wild-type, S1235A, or S1235D-rictor (Fig. 5C), indicating that neither phospho-mutant alters the abundance of the mTORC2 complex. In contrast, phosphorylation of Akt by mTORC2 containing the rictor S1235A mutant was increased compared with that by mTORC2 complexes containing wild-type rictor or its S1235D mutant, regardless of whether the complexes were isolated from cells grown in high or low serum (Fig. 5C, upper panel). These results suggest that phosphorylation of Ser1235 in rictor is critical to the ability of mTORC2 to phosphorylate Akt.

How one phosphorylation site on rictor can alter the activity of a large kinase complex of ~550 kD is an interesting question. Rictor does not have a kinase domain, and, in conjunction with its interacting partner mSin1, it determines the substrate specificity of mTORC2. Thus, the rictor Ser1235 phosphorylation site might act to regulate substrate binding to mTORC2. To address this possibility, we developed an mTORC2 substrate binding assay in which purified kinase complexes were incubated with glutathione S-transferase (GST)–Akt protein without ATP, nonbound substrate was washed away with lysis buffer, and the amount of Akt bound to mTORC2 was determined. First, we assessed Akt binding to mTORC2 complexes immunoprecipitated from actively growing or tunicamycin-treated (ER-stressed) cells. We found that binding of the recombinant GST-Akt protein to mTORC2 complexes purified from ER-stressed cells was lower compared to control (Fig. 5D, left panel). A similar decrease in GST-Akt binding to mTORC2 was detected in cells exposed to osmotic stress induced by sorbitol (Fig. 5D, right panel). Thus, ER stress inhibits substrate binding to mTORC2. We further proposed that if GSK-3β mediates ER stress–induced inhibition of mTORC2, GSK-3β–dependent phosphorylation of rictor might be engaged in regulating substrate binding to mTORC2. Next, we assessed the effect of the rictor phospho-mutants on mTORC2 substrate binding. We immunopurified similar amounts of mTORC2 from rictor-null cells constitutively expressing wild-type rictor or the phospho-mutants. Measurement of GST-Akt bound to the purified complexes showed that mTORC2 containing wild-type rictor bound more recombinant substrate compared to control (Fig. 5E, left panel). Compared to mTORC2 complexes containing wild-type rictor, complexes containing the rictor S1235A phospho-mutant bound more recombinant substrate, whereas those containing the rictor S1235D mutant bound less, comparable to nonspecific controls. Thus, preventing phosphorylation of rictor at Ser1235 facilitates the ability of mTORC2 to phosphorylate Akt by enhancing substrate binding, whereas mimicking constitutive phosphorylation of this site hinders substrate binding, leading to inhibition of mTORC2 activity (Fig. 5C). In this mTORC2 substrate binding assay, the interaction of the substrate with mTORC2 was ATP-dependent because this interaction was not detectable if ATP was present in the reaction buffer (Fig. 5E, right panel). Thus, if ATP is not present in the kinase buffer, the GST-Akt substrate is not efficiently released from mTORC2, and under such conditions, abundance of the bound GST-Akt protein reflects the substrate-binding capacity of the mTORC2 complex.

GSK-3β–dependent phosphorylation of rictor at Ser1235 regulates cell proliferation and tumor growth

Our results thus far indicated that preventing GSK-3β–mediated phosphorylation of rictor at Ser1235 increases the ability of mTORC2 to phosphorylate Akt at Ser473. Because Akt regulates cell proliferation, we reasoned that the rictor S1235A mutant might more potently promote cell proliferation compared to wild-type rictor or the S1235D mutant. To test this idea, we analyzed the proliferation rate of rictor-null MEFs stably expressing similar amounts of wild-type rictor or its phospho-mutants (fig. S5A). First, the proliferation rate of rictor-null MEFs was increased by restoring expression of wild-type rictor (2.27 times compared to unreconstituted rictor-null MEFs). Expression of the rictor S1235A mutant had a greater effect on cell proliferation compared to that of wild-type rictor (2.67 times relative to unreconstituted rictor-null MEFs and 30% greater than MEFs reconstituted with wild-type rictor), whereas cells expressing the rictor S1235D mutant did not proliferate to the same degree as those expressing either the wild type or the S1235A mutant (1.83 times relative to unreconstituted rictor-null MEFs and 50% less than MEFs reconstituted with wild-type rictor) as shown in Fig. 6A. These results are consistent with the effects of these mutants on mTORC2 signaling (Fig. 5). That MEFs reconstituted with the rictor S1235A mutant proliferated twice as fast as those reconstituted with the S1235D mutant suggests that regulation of this site is critical to the effects of GSK-3β on cell proliferation. We obtained similar differences in the proliferation rates of MEFs reconstituted with wild-type or phospho-mutant rictor grown in high serum, indicating that GSK-3β–dependent mTORC2 regulation occurs in at least two different cellular contexts (fig. S4). To extend these findings to an in vivo model, we assessed the effects of phosphorylation of rictor at Ser1235 on subcutaneous tumor growth. To establish tumors that would engraft in their hosts, we transformed MEFs reconstituted with wild-type or mutant rictor by overexpressing Ras (fig. S5B). Mice injected with the rictor S1235A mutant–expressing MEFs had larger tumors compared to those injected with cells expressing wild-type rictor (Fig. 6, B and C). In contrast, mice injected with cells expressing the rictor S1235D mutant developed smaller tumors than those injected with either wild-type or S1235A mutant rictor–containing cells (Fig. 6, B and C). These results suggest that GSK-3β–dependent regulation of mTORC2 is important in the control of cell proliferation and tumor growth.

Fig. 6

Phosphorylation of rictor at Ser1235 inhibits cell proliferation and tumor growth. (A) WT rictor or the phospho-mutants were reintroduced into rictor-null MEFs with a lentiviral expression system. Cell proliferation measurements were performed by counting cells 48 hours after incubation in low-serum conditions. The ratio of proliferation rate was graphed with GraphPad Prism 5 software. *P < 0.001 for all pairwise comparisons; one-way ANOVA, post hoc intergroup comparisons with Holm-Sidat test. (B) MEFs described in Fig. 5C were transformed by overexpression of the oncogenic form of H-Ras and injected into 6-week-old immunodeficient nude mice (n = 5 for each group; 5 × 106 cells per mouse). Tumor size was measured after 15 days, the tumor sizes were calculated, and the volumes were presented by histogram. *One-way ANOVA, post hoc intergroup comparisons with Holm-Sidat test; the P values are indicated. (C) Mice injected with MEFs expressing each form of rictor were killed and representative images of the excised tumors are shown.

Discussion

Despite the involvement of deregulation of Akt signaling and ER stress in the pathogenesis of several human diseases, the mechanism (or mechanisms) by which ER stress suppresses Akt signaling is not well characterized (4, 5). We found that ER stress triggers phosphorylation of the mTORC2 component rictor at Ser1235 by GSK-3β, thereby inhibiting mTORC2 kinase activity by hindering substrate binding. Thus, one specific phosphorylation site on rictor determines the substrate-binding capacity of a multiprotein complex of ~550 kD. Phosphorylation-dependent inhibition of substrate binding is not a common mechanism in the regulation of kinase activity. Rictor and Sin1 determine the substrate specificity of mTORC2 and both proteins appear to be involved in the recognition of Akt as a substrate. Eliminating ATP from the kinase-substrate milieu results in stabilization of Akt binding to mTORC2. Other rictor phosphorylation sites near to Ser1235, namely, Thr1135 and Ser1177 (23), do not regulate mTORC2 activity. Therefore, we propose that the Ser1235 site in rictor resides in close proximity to the Akt substrate binding pocket of mTORC2 and that phosphorylation of this site interferes with Akt binding to mTORC2.

How ER stress regulates the activity of GSK-3β remains poorly characterized. As a conserved and essential multitasking kinase, GSK-3β regulates cellular metabolism, proliferation, and survival (33, 39). Our study indicates that GSK-3β, and not GSK-3α, is a major kinase that phosphorylates the Ser1235 site in rictor to control mTORC2-Akt signaling. Moreover, we find that GSK-3β–mediated regulation of mTORC2 is most likely conserved not only in vertebrates but also in the invertebrate fly D. melanogaster (fig. S3). Many GSK-3 substrates require priming by phosphorylation at neighboring sites to subsequently facilitate binding to and phosphorylation by GSK-3. However, several GSK-3 substrates are phosphorylated without this priming event, such as BCL3, whose GSK-3 phosphorylation site resembles the rictor Ser1235 site. Phosphorylation of some GSK-3 substrates triggers their ubiquitination by the E3 ubiquitin ligase β-TrCP1 (β-transducin repeats containing protein 1) and subsequent proteasomal degradation (40). It will be interesting to determine whether GSK-3β–mediated phosphorylation of rictor might also target rictor for degradation.

We have identified a molecular mechanism to explain how the ER stress–regulated kinase, GSK-3β, restrains the anabolic effects of the mTORC2-Akt pathway. GSK-3 was first identified as an Akt substrate (41). Akt-mediated phosphorylation of GSK-3 inhibits GSK-3 activity toward GS, ultimately promoting the conversion of glucose to glycogen. Here, we found that GSK-3β itself inhibits Akt by controlling a key upstream activating kinase, mTORC2. The Ser1235 site in rictor is a key point in the regulation of the ER stress–induced effects of GSK-3β on mTORC2-Akt signaling. Our finding is consistent with the previous work describing the suppression by ER stress of insulin-dependent PI3K-Akt signaling through activation of GSK-3β (42). In addition, our work also agrees with mouse genetic studies that show correction of diabetes in mice deficient in GSK-3β, thus implying that GSK-3β interferes with insulin signaling (43).

GSK-3β is an increasingly promising target for the treatment of various human diseases, including metabolic disorders (33, 39). Akt is a critical regulator of glucose homeostasis (44, 45). Our study is consistent with the idea that GSK-3β and Akt have opposing effects on glucose metabolism and expand our understanding of the mechanisms by which GSK-3 inhibitors might regulate glucose metabolism. Activation of GS by inhibition of GSK-3 mimics insulin action by promoting the conversion of glucose to glycogen and is associated with the lowering of blood glucose concentrations (33). We propose that the insulin-mimetic effects of the GSK-3 inhibitors are also carried out by activation of the Akt signaling linked to the elimination of GSK-3β–mediated inhibition of mTORC2.

Materials and Methods

Materials

Reagents were obtained from the following sources: Dulbecco’s modified Eagle’s medium (DMEM)/F12 from Life Technologies; fetal bovine serum (FBS) from HyClone; complete protease inhibitor cocktail from Roche; protein G–Sepharose from Pierce; thapsigargin, tunicamycin, BIO, and SB-216763 from Sigma; DreamFect transfection reagent from OZ Biosciences; antibodies to V5 tag from Invitrogen; antibodies to phospho-Thr1135 and phospho-Ser1235 rictor, mTOR, raptor, phospho-Ser473 and phospho-Thr308 Akt, Akt, phospho-Ser641 GS, GS, phospho-Ser9 and phospho-Tyr216 GSK-3β, GSK-3β, GSK-3α, phospho-Thr980 PERK, and PERK from Cell Signaling Technologies; antibodies to rictor and tubulin and horseradish peroxidase (HRP)–labeled anti-rabbit, anti-mouse, and anti-goat secondary antibodies from Santa Cruz Biotechnology. Lentiviral shRNAs targeting human GSK-3α and GSK-3β were obtained from Sigma (MISSION shRNA). The targeting sequences of the oligonucleotides are as follows: GSK-3β sense strand, GCTGAGCTGTTACTAGGACAA (NM_002093.2-974s1c1); GSK-3β sense strand, CCCAAACTACACAGAATTTAA (NM_002093.2-1087s1c1); GSK-3α sense strand, CGGACATCAAAGTGATTGGCA (NM_019884.1-473s1c1); GSK-3α sense strand, CCTCTCTTCAACTTCAGTGCT (NM_019884.1-1375s1c1). pcDNA3–hemagglutinin (HA)–GSK-3β–wild type and pBabe-puro–GSK-3β (S9A) were obtained from Addgene.

Cell lines and culture

MEFs, HeLa, MDA-MB-435, and COS-7 cells were obtained from the American Type Culture Collection and cultured in DMEM/F12 with 10% FBS and penicillin/streptomycin in 5% CO2 at 37°C. Cell lines were cultured at a density that allowed cell division throughout the course of the experiment. MEFs were transfected by DreamFect following the manufacturer’s protocol.

Mutagenesis of rictor

The rictor complementary DNA (cDNA) from the pRK5 plasmid was resubcloned to pcDNA4 by means of the Xpress tag with the pcDNA4/HisMax TOPO TA Expression Kit from Invitrogen. The large size of the plasmid (11 kb) is incompatible with the polymerase chain reaction (PCR)–based mutagenesis. To decrease the size of the plasmid to 6 kb, we subcloned the fragment of rictor containing the sequences of the phospho-mutants into pBluescript by Bgl II and Not I digestion. The primers for mutagenesis were designed based on the QuikChange Primer Design Program (http://www.stratagene.com). The pBluescript plasmid containing the rictor fragment was mutagenized with the QuikChange II XL Mutagenesis Kit (Stratagene). After validation of mutations by sequencing, the mutated rictor fragments were reinserted into the Xp-rictor pcDNA4 plasmid by Bgl II and Not I digestion.

Immunoprecipitations and kinase assays

For immunoprecipitation experiments, the lysis buffer contained 0.3% CHAPS instead of 1% Triton to preserve the integrity of the mTOR complexes. Two micrograms of rictor antibody was added to the cleared cellular lysates (1 mg of protein content in 700 μl) and incubated with rotation at 4°C for 90 min. After 1 hour of incubation with 40 μl of 50% slurry of protein G–agarose, immunoprecipitates captured by protein G–agarose were washed four times with CHAPS-containing lysis buffer and once with rictor-mTOR kinase buffer [25 mM Hepes (pH 7.5), 100 mM potassium acetate, and 2 mM MgCl2]. For in vitro mTORC2 kinase reactions, immunoprecipitates were incubated in a final volume of 15 μl at 37°C for 20 min in rictor-mTOR kinase buffer containing 500 ng of inactive Akt1-GST and 500 μM ATP. The reaction was stopped by the addition of 200 μl of ice-cold enzyme dilution buffer [20 mM Mops (pH 7.0), 1 mM EDTA, 0.3% CHAPS, 5% glycerol, 0.1% 2-mercaptoethanol, and bovine serum albumin (BSA; 1 mg/ml)]. After a quick spin, the supernatant was removed from the protein G–agarose, and a 15-μl portion was analyzed by immunoblotting for phospho-Ser473 Akt and total Akt. The pelleted protein G–agarose beads were also analyzed by immunoblotting to determine the abundance of rictor, Sin1, and mTOR in the immunoprecipitates. For in vitro GSK-3β kinase reactions, immunoprecipitates were prepared from MDA-MB-435 cell lysates with rictor antibody. GSK-3β was immunoprecipitated from lysates of human embryonic kidney (HEK) 293T cells expressing GSK-3β that were treated with tunicamycin (2 μg/ml) for 3 hours before cell harvest. In vitro GSK-3β kinase assays were performed at 37°C for 30 min in a final volume of 25 μl of GSK-3β kinase buffer containing 4 mM Mops (pH 7.2), 2.5 mM β-glycerophosphate, 1 mM EGTA, 0.4 mM EDTA, 4 mM MgCl2, 0.05 mM dithiothreitol (DTT), 40 μM BSA, and 500 μM ATP. Immunoprecipitates captured by protein G–agarose were washed with rictor-mTOR kinase buffer for subsequent in vitro mTORC2 kinase reactions as indicated above.

Cell lysis and immunoblotting

Cells were rinsed with ice-cold phosphate-buffered saline (PBS) before lysis in buffer containing 40 mM Hepes (pH 7.5), 120 mM NaCl, 1 mM EDTA, 10 mM sodium pyrophosphate, 10 mM sodium glycerophosphate, 50 mM NaF, 1% Triton X-100, and protease inhibitor cocktail (Roche). Scraped lysates were incubated for 20 min at 4°C to complete lysis. The soluble fractions of cell lysates were isolated by centrifugation at 13,000 rpm at 4°C for 12 min. Samples of the cellular lysates containing an equal amount of proteins were resolved by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride (PVDF) membrane. Proteins were then visualized by immunoblotting and detected with enhanced chemiluminescence (ECL) with the Immobilon Western kit (Millipore).

Retroviral production and infection

Retroviral vectors were propagated in and purified from XL-10 Gold bacterial cells and cotransfected together with the Δ VPR and VSVG plasmids into actively growing cells as previously described (18). One day before transfection, HEK 293T cells (1.2 × 106) were plated on 6-cm dishes in 3 ml of DMEM supplemented with 10% FBS. For production of retroviruses, HEK 293T cells were transfected by the calcium phosphate method with 3 μg of transfer vector pMSCV, 0.6 μg of envelope-coding plasmid VSVG, and 2.4 μg of Gag-pol–expressing plasmid. Retroviruses were harvested 48 hours after transfection and centrifuged at 3000g at 4°C for 15 min to eliminate any remaining HEK 293T cells. One day before infection, cells to be infected were seeded in six-well plates. Viral supernatant was added at a ratio of 1:1 to the culture medium in the presence of polybrene (8 μg/ml), and the cells were centrifuged at 1800 rpm for 45 min to increase the infection rate. Cells were incubated with retroviruses for 24 hours. A second infection was performed the next day following the same protocol. After an additional 24 hours of recovery in normal medium, infected cells were passaged and selected with puromycin (2 μg/ml for 2 days).

Preparation of liver tissue extracts for immunoblotting

Powdered livers were homogenized on ice with a Teflon glass homogenizer in 1 ml of ice-cold buffer [20 mM tris-HCl (pH 7.4), 20 mM NaCl, 1 mM EDTA, 20 mM β-glycerophosphate, 5 mM EGTA, 1 mM DTT, and 1 mM phenylmethylsulfonyl fluoride (PMSF)] containing 0.1% Tween 20 and were centrifuged at 1000g for 20 min at 4°C to pellet insoluble material. The protein concentration of the supernatant was determined. Liver extract samples (containing 50 to 100 μg of total protein) were subjected to SDS-PAGE in 7.5% polyacrylamide gels followed by Western blot analysis with indicated antibodies.

Tumor allografts

MEFs constitutively expressing wild-type, S1235A, or S1235D-rictor were transformed by H-Ras overexpression. Protein extracts from wild-type–, S1235A-, or S1235D-rictor–expressing MEFs indicated equal protein amounts of rictor and H-Ras by Western blotting analysis. The MEFs (5 × 106 cells per mouse) were injected subcutaneously into the upper flank region of 6-week-old immunodeficient nude mice (n = 5 for each group). Tumor size was measured after 15 days and the tumor volume was determined with the standard formula L × W2 × 0.5, where L is the longest length and W is the shortest length in millimeters. The differences in the tumor volume from mice injected by wild-type–, S1235A-, or S1235D-rictor–expressing MEFs were compared by one-way analysis of variance (ANOVA). Nude mice were killed and the tumors were excised.

In vitro substrate binding assay

For the in vitro substrate binding assay, rictor immunoprecipitates were incubated in a final volume of 15 μl at 37°C for 20 min in rictor-mTOR kinase buffer containing 100 ng of Akt1-GST with or without ATP. The immunoprecipitates were then washed four times with CHAPS-containing lysis buffer. The pelleted protein G–agarose beads were analyzed by immunoblotting to determine the protein abundance of Akt1-GST, rictor, Sin1, and mTOR in the immunoprecipitates.

Drosophila RNAi and analysis

RNAi against Drosophila GSK-3 was performed as previously described (10). dsRNA targeting Drosophila GSK-3 was synthesized by in vitro transcription (IVT) in 20-μl reactions with a MEGAscript T7 kit (Ambion). DNA templates for IVT were generated by PCR from total Drosophila S2 cellular genomic DNA. The primers (which incorporated a 5′ and 3′ T7 promoter) for dGSK-3 dsRNA synthesis were as follows: dGSK-3 forward primer, 5′-TAATACGACTCACTATAGGCCGTTGACGAGTTTGTGTGT; dGSK-3 reverse primer, 5′-TAATACGACTCACTATAGGAAACTCGGCGACTGTTTGTT. The underlined region indicates the T7 promoter sequence. dsRNA products were purified with an Invitrogen PureLink PCR Purification Kit. Final dsRNA concentrations were measured on a NanoDrop spectrophotometer. Drosophila S2 cells were prepared for dsRNA addition by seeding 1 × 106 cells in 2 ml of Drosophila Schneider’s medium to each well in six-well culture plates. dsRNAs were transfected with FuGENE 6 (Roche). Briefly, 6 μl of FuGENE was added to 94 μl of Drosophila SFM (Invitrogen) followed by the addition of 4 μg of dGSK-3 dsRNA. Four micrograms of green fluorescent protein (GFP) dsRNA was used as control. Tubes were gently mixed and incubated for 20 min at room temperature. FuGENE-dsRNA complexes were then administered to cells by adding the entire mixture dropwise around wells and then swirling to ensure even dispersal. After 6 hours of dsRNA addition, the medium was changed to avoid potential negative effects of FuGENE on cell viability. Additional FuGENE-dsRNA complexes were added to wells on each of the following 3 days. After 4 days of incubation to allow turnover of the target mRNAs, cell lysates were subjected to immunoblotting for the amounts of phospho- and total dAkt, GSK-3, and tubulin.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/4/161/ra10/DC1

Fig. S1. Validation of the phospho-rictor (Ser1235) antibody and alignment of rictor orthologs.

Fig. S2. Effect of GSK-3 inhibition on the mTORC2 complex.

Fig. S3. GSK-3–dependent regulation of Akt is conserved in the invertebrate fly D. melanogaster.

Fig. S4. The rate of cell proliferation is dependent on phosphorylation of rictor at Ser1235.

Fig. S5. Analysis of rictor and Ras expression in MEFs.

Table S1. Quantification of the indicated immunoblotting signals.

References and Notes

  1. Acknowledgments: We thank D. M. Sabatini and M. A. Magnuson for providing rictor-null MEFs, J. Woodgett for providing GSK-3β–null MEFs, and X. Lin and B. D. Manning for reagents. We thank Cell Signaling Technology Inc. (Danvers, MA) for developing and providing the phospho-Ser1235 rictor antibodies and S.-C. J. Yeung for help with the statistical analysis. We gratefully acknowledge D. M. Sabatini for the critical reading and editing of our manuscript. Funding: This work was supported by the MD Anderson Trust Fellow Fund and NIH grant CA133522 (D.D.S.). T.R.P. was supported by the Ludwig Cancer Center Fellowship and the American Diabetes Association. Author contributions: C.-H.C., T.R.P., and D.D.S. conceived the project. C.-H.C. and D.D.S. designed the experiments and analyzed the data. C.-H.C. performed most of the experiments. T.S. set up in vitro kinase studies. R.A. and A.K.B. performed the rictor mutagenesis and established the stable cell lines. S.-W.L., J.W., and H.-K.L. performed in vivo studies. T.R.P. and D.D.S. wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article