Eukaryotic Elongation Factor 2 Kinase Activity Is Controlled by Multiple Inputs from Oncogenic Signaling

Abstract

Eukaryotic elongation factor 2 kinase (eEF2K), an atypical calmodulin-dependent protein kinase, phosphorylates and inhibits eEF2, slowing down translation elongation. eEF2K contains an N-terminal catalytic domain, a C-terminal α-helical region and a linker containing several regulatory phosphorylation sites. eEF2K is expressed at high levels in certain cancers, where it may act to help cell survival, e.g., during nutrient starvation. However, it is a negative regulator of protein synthesis and thus cell growth, suggesting that cancer cells may possess mechanisms to inhibit eEF2K under good growth conditions, to allow protein synthesis to proceed. We show here that the mTORC1 pathway and the oncogenic Ras/Raf/MEK/extracellular signal-regulated kinase (ERK) pathway cooperate to restrict eEF2K activity. We identify multiple sites in eEF2K whose phosphorylation is regulated by mTORC1 and/or ERK, including new ones in the linker region. We demonstrate that certain sites are phosphorylated directly by mTOR or ERK. Our data reveal that glycogen synthase kinase 3 signaling also regulates eEF2 phosphorylation. In addition, we show that phosphorylation sites remote from the N-terminal calmodulin-binding motif regulate the phosphorylation of N-terminal sites that control CaM binding. Mutations in the former sites, which occur in cancer cells, cause the activation of eEF2K. eEF2K is thus regulated by a network of oncogenic signaling pathways.

INTRODUCTION

Eukaryotic elongation factor 2 (eEF2) mediates the translocation of the ribosome (between successive codons) during the elongation phase of mRNA translation, which is where most of the energy and amino acids required for protein synthesis is used. The activity of eEF2 is regulated by its phosphorylation (1), which impairs its interaction with the ribosome, inhibiting eEF2 and thus translation elongation (2–4).

eEF2 is phosphorylated by a highly specific protein kinase, eEF2 kinase (eEF2K [5–7]). eEF2K belongs to the small group of so-called α-kinases, which are quite distinct from the main eukaryotic protein kinase superfamily (8) and remain poorly understood, especially in terms of their regulation. eEF2K is the only calcium/calmodulin (Ca/CaM)-activated α-kinase (6, 7, 9) (see also Fig. 1). eEF2K activity controlled by multiple signaling pathways, in particular through the mammalian target of rapamycin complex 1 (mTORC1), which promotes the phosphorylation of eEF2K at inhibitory sites (Fig. 1). mTORC1 signaling acts to inhibit eEF2K and thus promote dephosphorylation and activation of eEF2 (10). mTORC1 signaling is stimulated by amino acids and anabolic, mitogenic, or hypertrophic stimuli, e.g., insulin (11), thereby providing a mechanism by which these agents can activate translation elongation. Some, but not all, the functions of mTORC1 are inhibited by rapamycin (12, 13); in contrast, compounds that directly inhibit the catalytic activity of mTOR, such as AZD8055 (14), block all known functions of mTOR.

FIG 1.

Schematic depiction of structural organization of eEF2K. The upper illustration shows the overall layout of eEF2K, including phosphorylation sites controlled by mTOR that were known prior to the present study; all three inputs impair eEF2K activity. The lower part shows a model for the interplay between different domains of eEF2K in its function. Residue numbering corresponds to human eEF2K. Solid arrows show known kinase inputs; the dashed arrow indicates a phosphorylation event where the kinase involved was unknown. The illustration is not to scale.

Ribosomal protein S6 kinase (S6K), which is activated by mTORC1, provides one link between mTORC1 and the control of eEF2K by phosphorylating eEF2K at Ser366 (15), the phosphorylation of which decreases the sensitivity of eEF2K to activation by Ca/CaM. mTORC1 signaling also promotes the phosphorylation of eEF2K at Ser78, adjacent to the CaM-binding site (16). This event interferes with CaM binding and thus inhibits eEF2K activity. mTORC1 can also regulate the phosphorylation of Ser359, another inhibitory site (17, 18). However, it is not known how mTORC1 signaling controls the phosphorylation of Ser78, whether mTORC1 affects additional phosphorylation sites in eEF2K, whether mTORC1 directly phosphorylates eEF2K, or how sites within the linker region affect the function of eEF2K.

Ser366 is also phosphorylated by p90^RSKs (15), which are activated by the mitogen-activated protein (MAP) kinase pathway (19), downstream of the proto-oncogenes Ras and Raf. However, no instance has thus far been reported where eEF2 and eEF2K are controlled specifically via this mechanism in response to cellular stimuli. Even where MEK/extracellular signal-regulated kinase (ERK) does affect eEF2 phosphorylation (e.g., in cardiomyocytes), this is mediated through mTORC1 (20), likely through the ability of MEK/ERK to activate mTORC1 signaling (21). eEF2K can also be regulated by p38 mitogen-activated protein (MAP) kinases or their downstream effectors (17, 22) and by the cell cycle regulator cdc2 (18).

Recent work suggests eEF2K is cytoprotective in cancer cells during nutrient starvation (23). eEF2K expression is high in certain cancers (23–25). It is therefore important to gain further insights into the mechanisms by which signaling pathways, especially those that are activated in cancer cells, control eEF2K activity. Both mTORC1 and MEK/ERK signaling are often activated by oncogenic mutations in their upstream regulators (26).

The molecular architecture of eEF2K and the mechanisms underlying its control remain poorly understood. The α-kinase domain lies toward its N terminus, and N-terminal to that is the CaM-binding site (27–29) (Fig. 1). C-terminal to the kinase domain is a “linker” region that contains several regulatory phosphorylation sites (Fig. 1) but is predicted to be disordered. The C-terminal part of eEF2K contains several predicted α-helical SEL1-like repeats, which can mediate protein-protein interactions (30). Its extreme C terminus is required for ability to phosphorylate eEF2 but not a peptide substrate, MH1 (29). Furthermore, although an N-terminal fragment containing residues 1 to 402 has kinase activity (e.g., undergoes autophosphorylation), it only phosphorylates eEF2 or MH1 in the presence of the C-terminal 478-725 fragment (29). This suggests that the C- and N-terminal regions come closer together to create the conformation required to phosphorylate substrates in trans, perhaps because the C-terminal region helps recruit them to the catalytic domain (Fig. 1). Several other phosphorylation sites have been identified in eEF2K, some of which modulate its activity, but it is unclear how they work together to do this.

We provide here several key new insights into the regulation of eEF2K. In particular, our findings reveal additional regulatory inputs into eEF2K from several signaling pathways and demonstrate interplay between different domains of eEF2K in its regulation.

MATERIALS AND METHODS

Materials.

Rat hepatoma H4IIE cells were from the American Type Culture Collection (catalog number CRL-1548). Human KB cells and mouse glycogen synthase kinase 3α(S21A) [GSK-3α(S21A)]/GSK3β(S9A) knock-in mouse embryonic fibroblasts (MEFs) were kindly provided by D. Alessi (MRC Protein Phosphorylation Unit, Dundee, United Kingdom). Signaling inhibitors were obtained as follows: AZD6244, Selleck; AZD8055, Selleck; PP242, Sigma; and rapamycin, Calbiochem.

All antibodies were obtained from Cell Signaling Technology, except those for eEF2K phosphorylated at Ser359, Ser377, or Ser396, which were kindly provided by the Division of Signal Transduction Therapy [DSTT], College of Life Sciences, University of Dundee, United Kingdom, and additional antibodies for new phosphorylation sites in eEF2K, which were generated by Eurogentec.

Recombinant protein kinases were obtained as follows: ERK2, Cdc2/cyclin B, and p38 MAP kinase were from Invitrogen; glutathione S-transferase (GST)–eEF2K was expressed in Escherichia coli and purified as described previously (29).

Cell culture, transfection, treatment, and lysis.

KB, MEF, rat hepatoma, and HEK293 cells were grown in Dulbecco modified Eagle medium supplemented with 10% (vol/vol) fetal calf serum, 100 U/ml penicillin, and 0.1 mg/ml streptomycin. When the cells reached the appropriate degree of confluence (80% for KB cells, MEFs, and HEK293 cells or 50% for rat hepatoma cells), they were removed from serum and maintained in serum-free medium for 18 h (KB, MEF, and HEK293 cells) or 29 h (rat hepatoma cells) prior to the start of the experiment treatments.

VB6 and BICR6 cells were maintained in MEM-Alpha medium (Gibco; catalog no. 11900-016) supplemented with 10% fetal calf serum, 0.18 μM adenine, 5 μg/ml insulin, 0.5 μg/ml hydrocortisone, 10 ng/ml epidermal growth factor, and 1× Glutamax-I (Gibco).

HEK293 or KB cells were transfected using calcium phosphate or Fugene HD (Promega), respectively. Cells were pretreated with the signaling inhibitors for 40 min before exposure to insulin. Lysis buffer was composed of 50 mM Tris-HCl (pH 7.4), 50 mM β-glycerophosphate, 0.2 mM EDTA, 0.2 mM EGTA, 50 mM NaCl, 1% (vol/vol) Triton X-100, and 1 mM sodium orthovanadate plus 1× protease inhibitors (cocktail from Roche) and 15 mM β-mercaptoethanol.

Gel electrophoresis and Western blotting.

The protein concentrations were determined as described previously (31). Cell lysates were heated at 95°C for 5 min in sample buffer (62.5 mM Tris-HCl, 7% (wt/vol) sodium dodecyl sulfate (SDS), 20% (wt/vol) sucrose, and 0.01% (wt/vol) bromophenol blue) and subjected to polyacrylamide gel electrophoresis (PAGE) and electrophoretic transfer to nitrocellulose membranes. Membranes were then blocked in phosphate-buffered saline (PBS)–Tween containing 5% (wt/vol) skimmed milk powder for 30 min at room temperature. The membranes were probed with the indicated primary antibody overnight at 4°C. After incubation with fluorescently tagged secondary antibody, signals were scanned using a Li-Cor Odyssey imaging system.

Molecular biology and mutagenesis.

Point mutations were created by PCR mutagenesis, using the QuikChange system (Stratagene). Vectors were always resequenced prior to use. eEF2K and its mutants were expressed as GST fusion proteins in E. coli as recently described (32).

Immunoprecipitation (IP) of eEF2K.

Portions (2 μg) of anti-eEF2K or anti-Flag antibody were prebound to 10 μl of protein G-beads at 4°C overnight. The beads were washed thrice with PBS, 1 mg (protein) of cell lysate was added, and the mixture was tumbled at 4°C for 90 min. Next, the beads were finally washed thrice with lysis buffer, and bound proteins were subjected to SDS-PAGE.

CaM binding to eEF2K.

HEK293 cells were lysed in buffer comprising 50 mM HEPES (pH 7.4), 50 mM β-glycerophosphate, 50 mM NaCl, 1 mM CaCl₂, 0.3% (wt/vol) CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate}, and 1 mM sodium orthovanadate plus 1× protease inhibitors (cocktail from Roche) and 15 mM β-mercaptoethanol.

Protein kinase assays.

eEF2K was assayed as described earlier (16), with the following modifications. HEK293 cells overexpressing eEF2K were harvested in high Ca²⁺ buffer (50 mM HEPES [pH 7.4], 50 mM NaCl, 50 mM β-glycerophosphate, 0.3% [wt/vol] CHAPS, and 1 mM CaCl₂) in order to maintain the eEF2 kinase-CaM interaction. Next, 10 ng of cell lysate was used per reaction in a total volume of 40 μl containing 20 mM 3-(N-morpholino)propanesulfonic acid (MOPS; pH 7.0), 5 mM MgCl₂, 0.4 mM EGTA, 2 mM EDTA, 1 mM dithiothreitol, 2% (vol/vol) glycerol, and 0.67 mM CaCl₂ in the presence of purified eEF2 (2 μg) and [γ-³²P]ATP. The activity was assayed at 30°C for 10 min. The reaction was stopped by the addition of SDS-PAGE sample buffer, and the incorporation of ³²P into eEF2 was determined by SDS-PAGE, followed by staining with Coomassie brilliant blue. Radioactivity was detected either by autoradiography or by using a phosphorimager (Typhoon; GE Healthcare).

For mTOR assays against eEF2K (or, as a positive control, 4E-BP1), the method described earlier (33) was used, except that MnCl₂ was omitted and the reaction time was 1 h.

Assays for eEF2K phosphorylation were performed using 0.5 μg of purified wild-type (WT) GST-eEF2K or the indicated mutant proteins, expressed in E. coli, in kinase buffer (final concentrations: 30 mM Tris [pH 7.4], 10 mM MgCl₂, 50 mM KCl, and 0.1 mM EGTA) and with ATP (1 mM). Reaction mixtures containing the potential priming kinase were performed for 25 min (at 30°C), at which time GSK3 was added for 10 min (in some cases together with radioactive ATP [see the figure legends]). Reaction mixtures contained the following amounts of the relevant kinases: Ck1δ (0.04 mg/ml; DSTT), CK2 (0.05 mg/ml; Invitrogen; PV3624), ERK2 (0.05 mg/ml; Invitrogen, PV3313), Cdk5/p25 (0.06 mg/ml; Invitrogen; PV4676), or p38 MAP kinase α (0.1 mg/ml; DSTT). GSK3 (0.12 mg/ml; DSTT) was added at the indicated time points, in some cases, together with [γ-³²P]ATP (where indicated). Assays were continued for a further 10 min, after which the products were analyzed by SDS-PAGE, followed by phosphorimage analysis (radioactive reactions) or Western blotting (nonradioactive assays).

Reproducibility of data.

All experiments were performed on at least three independent occasions and the data reflect consistent findings from these multiple replicates.

RESULTS

Signaling through mTORC1 and MEK maintains active eEF2 in cancer cells.

It has previously been shown that some cancers (e.g., glioma) express high levels of eEF2K and that it helps protect such cells against nutrient deprivation (23–25). However, since eEF2K is a negative regulator of protein synthesis, a process which is required for cell growth and proliferation, we surmised that cancer cells should possess mechanisms to restrain eEF2K activity and thus maintain eEF2 in its active dephosphorylated state.

To examine this possibility, we surveyed a range of cancer cell lines for their levels of eEF2K expression and eEF2 phosphorylation (Fig. 2A). Several, especially the oral carcinoma cell lines VB6 and BICR6, show these characteristics, as do the KB cells we used in earlier studies (e.g., see references 16 and 17). VB6 and BICR6 cells also show relatively high levels of phosphorylated (active ERK), suggesting this pathway may contribute to maintaining low levels of eEF2 phosphorylation (Fig. 2B). To study this, we used signaling inhibitors, including AZD6244, a potent and specific inhibitor of MEK (34), and rapamycin, an allosteric inhibitor of mTORC1 that inhibits many of its effects. When used alone, AZD6244 caused a modest increase in phosphorylated eEF2 levels, similar to that caused by rapamycin, indicating this pathway also contributes to keeping eEF2 dephosphorylated. When used together, AZD6244 and rapamycin caused a strong increase in phosphorylated eEF2 in both cell lines (Fig. 2B). Thus, these two pathways cooperate to maintain low phospho-eEF2 levels. Interestingly, in both cell lines, the mTOR kinase inhibitor AZD8055 (14) had a stronger effect on phospho-eEF2 than rapamycin, indicating that rapamycin-insensitive outputs from mTOR also contribute to control of eEF2/eEF2K in these cells (Fig. 2C). It was therefore important to identify additional links between these pathways and the control of eEF2K. It is evident from the data in Fig. 2B that certain signaling inhibitors cause an increase in the mobility of eEF2K on SDS-PAGE; reasons for this are explored later in the present study.

FIG 2.

Regulation of eEF2 phosphorylation in cancer cell lines. (A) Samples of lysate from the indicated cell lines, grown in medium containing serum, were analyzed by SDS-PAGE/Western blotting with the indicated antibodies. (B and C) VB6 or BICR6 cells were maintained in growth medium containing 10% serum in the presence or absence of AZD6244 (A6), rapamycin (R), and/or AZD8055 (A8) for 2 h. Cells were then lysed, and lysates were analyzed by immunoblotting with the indicated antibodies.

Inputs from mTOR to the control of eEF2K and the phosphorylation of eEF2.

In previous studies (e.g., see reference 16), we showed that in KB cells insulin consistently induces the rapid dephosphorylation of eEF2 (Fig. 3) and a slower increase in phospho-S6. This effect was not blocked by rapamycin or the S6K1 inhibitor PF4708671 (35), although it was attenuated by these compounds especially at later times (45 min). At this time point they also impaired the insulin-induced phosphorylation of Ser366 in eEF2K, the site that is phosphorylated by S6K1 (15) (Fig. 3). It should be noted that although PF4708671 primarily inhibits S6K1 (35), rather than the S6K2 isoform which is reported to be the main kinase for S6 in some cell types (36), it does block S6 phosphorylation in KB cells, suggesting S6K1 may be the main kinase responsible here. These data indicate that, although mTORC1/S6K1 signaling contributes to the control of eEF2 phosphorylation and eEF2K in response to insulin, other regulatory inputs are also involved. Mobility shifts are again evident, linked to responses to insulin (retarded) or treatment with rapamycin (faster) but not the S6K inhibitor.

FIG 3.

Role of mTORC1 signaling in the regulation of the phosphorylation of eEF2 and eEF2K. Serum-starved KB cells were preincubated for 40 min with 100 nM rapamycin (R) or 10 μM PF4708671 (P4) and then stimulated with 100 nM insulin for indicated times.

GSK3 provides an input to the control of eEF2K.

The above data indicate that additional signaling connections impinge on the regulation of eEF2 phosphorylation and probably eEF2K. GSK3, which is inactivated by signaling through phosphatidylinositol 3-kinase and protein kinase B, regulates several proteins involved in anabolic processes, including the TSC2 protein, which regulates mTORC1 signaling (37). We therefore tested whether GSK3 played a role in controlling eEF2K, using mouse embryonic fibroblasts (MEFs) in which knock-in mutations have been made in the genes for GSK3α and -β to change the regulatory phosphorylation sites at Ser21 and Ser9, respectively, to alanines {GSK3(α[S21A]/β[S9A]) MEFs (38)}; this abrogates the regulation of GSK3 by insulin. As judged from the phosphorylation of S6, activation of mTORC1 signaling by insulin was not affected by the mutations in GSK3 (Fig. 4A). In wild-type MEFs, insulin induced the dephosphorylation of eEF2 in a rapid and sustained manner (Fig. 4A). In contrast, the rapid phase of eEF2 dephosphorylation was lost in the GSK3(α[S21A]/β[S9A]) MEFs (10- and 20-min time points; Fig. 4A and B). This indicates that, in response to insulin, inactivation of GSK3 through Ser9/21 contributes to the regulation of eEF2 phosphorylation and thus, presumably, to the control of eEF2K. The simplest explanation for this would be that GSK3 itself phosphorylated eEF2K.

FIG 4.

GSK3-KI cells. (A) Serum-starved wild-type or GSK-3α(S21A)/GSK3β(S9A) knock-in MEFs were stimulated with 100 nM insulin for the indicated times. Cells were lysed, and equal amounts of protein were subjected to analysis by Western blotting with the indicated antibodies. (B) Quantitation of data from multiple experiments as in panel A; data are given as means ± the standard errors of the mean (SEM; control cells without insulin = 1; n = 3). (C) Sequences around potential GSK3 sites in human eEF2K are shown.

Phosphorylation sites for GSK3 that are regulated via its phosphorylation are generally followed by a “priming” phosphoserine or phosphothreonine (39) at position +4, which allows phosphorylation by GSK3. The sequence of eEF2K contains several possible pairs of GSK3/priming sites, e.g., Ser27/31, Ser70/74, Ser392/396, and Ser470/474, as well as Ser74/78 and Ser474/478, none of which has previously been studied (Fig. 4C and Table 1), apart from Ser78, an insulin-stimulated site (16). Phosphoproteomic analysis of eEF2K expressed in HEK293 cells revealed that Ser74 and Ser470 are phosphorylated; Ser27 and Ser392 may not have been detected in this analysis since they lie in negatively charged or very large tryptic peptides. Ser396 has been reported previously (22).

TABLE 1.

Sequences flanking selected phosphorylation sites in human and rat eEF2K

^aPotential GSK3 sites are shaded, and the corresponding possible priming sites are indicated in boldface. In the case of Ser74 and Ser474, the residue indicated in this way is also a potential GSK3 site, the priming site (Ser78/478) is indicated by double underlining. Ser359 is also included (single underlining), although it is not a potential site for GSK3.

To test whether specific residues were substrates for GSK3, eEF2K or mutants in which a site of interest was converted to alanine were incubated with GSK3 and [γ-³²P]ATP, in some cases after preincubation with a potential priming kinase (predicted on the basis of the sequence C-terminal to the potential priming site). For example, based on the presence of several C-terminal Asp/Glu residues, Ser31 appeared likely to be phosphorylated by CK2 (40). GSK3 phosphorylated eEF2K only weakly without preincubation with a possible priming kinase (Fig. 5A); incubation with CK2, p38 MAP kinase α, Erk2, or CDK5 (depending on the sites under study) substantially enhanced subsequent phosphorylation by GSK3. This was decreased by mutation of the probable GSK3 sites, often quite strongly (S392A, S70A, S470A) (Fig. 5A). This suggested that S70, S392, and S470 might be targets for GSK3 in vivo. The data are less clear for S27. We have not studied Ser74, since basal phosphorylation of its potential priming site (Ser78) is low and increased by insulin (16), which inactivates GSK3 or Ser474, because it is unclear what kinase might phosphorylate Ser478.

FIG 5.

Phosphorylation of recombinant eEF2K in vitro by GSK3 and other kinases. (A) Recombinant GST-eEF2K or point mutants were incubated with [γ-³²P]ATP and the indicated potential priming kinases (in the presence of nonradioactive ATP) for 25 min and, where shown, GSK3 and [γ-³²P]ATP were then added for a further 10 min. The top part shows a phosphorimage of the gel; the bottom part shows the Coomassie blue-stained gel. (B) Phospho-specific antisera for the indicated sites in eEF2K were tested by dot blotting by applying the indicated dilutions of a 10-mg/ml solution of the corresponding phospho- and nonphosphopeptides to the nitrocellulose membrane. The primary antibody was used in the presence of the non-phospho-peptide at 0.5 μg/ml. (C) GST-eEF2K or mutants were incubated with the indicated kinases in the presence of nonradioactive ATP. Products were analyzed by SDS-PAGE/Western blotting with the indicated phospho-specific antisera (and anti-eEF2K as a “loading control”).

To allow us to investigate the phosphorylation of these potential GSK3 sites, we generated phospho-specific antisera for each of them; all four antisera were fully phospho-specific (i.e., did not react with the corresponding nonphosphorylated peptide; Fig. 5B). Ser27 was basally phosphorylated and was not altered by insulin treatment or the GSK3 inhibitor CT99021 (41) (Fig. 6). Phosphorylation of Ser70 and Ser470 was low or undetectable in serum-starved cells (Fig. 7A) and insulin increased phosphorylation of both (Fig. 7A). This is not consistent with any of these sites being phosphorylated by GSK3 in cells. In contrast, Ser392 was basally phosphorylated, and this was decreased by the GSK3 inhibitor CT99021, indicating that it is indeed basally phosphorylated by GSK3 in cells (Fig. 7C). However, phosphorylation at Ser392 was actually increased by insulin (Fig. 7B and C), indicating that it is not primarily controlled by GSK3 in vivo, and cannot explain the input from GSK3 to the control of eEF2K. Additional work, beyond the scope of the present study, is needed to explore how GSK3 regulates eEF2K. It should be noted that the potential priming site for Ser392, Ser396, was reported to be a constitutively phosphorylated (22). Although Ser396 can be phosphorylated by p38 MAP kinases in vitro, they do not appear to be responsible for its phosphorylation in vivo (22). The control of this site is studied further below. Our data identify four new in vivo phosphorylation sites in eEF2K, three of which (not Ser70) are conserved in rat and mouse eEF2K (Table 1).

FIG 6.

Regulation of phosphorylation of Ser27 in eEF2K. Serum-starved KB cells were preincubated for 40 min with 3 μM CT99021 (CT) or treated with insulin for the indicated times. The cells were lysed, and eEF2K was immunoprecipitated from equal amounts of lysate protein prior to analysis.

FIG 7.

Control of specific phosphorylation sites in eEF2K. (A) Serum-starved KB cells were treated with 100 nM insulin for the times indicated. The cells were lysed, and eEF2K was immunoprecipitated from equal amounts of lysate protein. The levels of the indicated phosphorylated and total proteins were determined by immunoblotting. (B) Same as in panel A, but cell lysates were analyzed. (C) Serum-starved KB cells were preincubated for 40 min with 5 or 10 μM CT99021 (CT) or treated with insulin for the indicated times. In panels A to C, cells were lysed, and equal amounts of protein were subjected to analysis by Western blotting with the indicated antibodies. (D) Sequence alignment showing conservation of S377 and adjacent residues: Hs, Homo sapiens; Mm, Mus musculus; Bt, Bos taurus; Ac, Anolis carolinensis; Tg, Taeniopygia guttata (zebra finch); Dr, Danio rerio (zebra fish). (E) Same as in panel A, but eEF2K was immunoprecipitated from equal amounts of lysate protein. Values below each lane show quantitation of the signal for P-S377, corrected for the eEF2K signal (means ± the SEM, n = 4). (F) Same as above, but the cells were preincubated for 40 min with AZD6244 (A6) or PF3644022 (P3) at either 5 or 10 μM; eEF2K was then immunoprecipitated from equal amounts of lysate protein prior to analysis.

Insulin also rapidly increased the phosphorylation of Ser377 (Fig. 7B), the other previously known site in the linker region (22). This site and its local context are strongly conserved in vertebrate eEF2K proteins (Fig. 7D). Its insulin-induced phosphorylation was not affected by rapamycin or AZD8055 but was decreased by the MEK inhibitor AZD6244 (Fig. 7E and F) and by the MK2 inhibitor PF3644022 (Fig. 7F). Insulin does not activate p38 MAP kinase as indicated using an antibody that recognizes phosphorylated (activated) p38 MAP kinase isoforms (Fig. 7F). These data suggest insulin might stimulate the phosphorylation of Ser377 at least in part via MEK/ERK and MK2.

Role of specific signaling pathways in the control of the novel phosphorylation sites in eEF2K.

We next used the phospho-specific antisera for the insulin-regulated sites Ser70, Ser392, and Ser470 in human eEF2K to study which signaling pathways control their phosphorylation. Assessment of the phosphorylation of S6K1 (at Thr389) and ERK provide positive controls, respectively, for the efficacy of the mTORC1/mTOR and MEK inhibitors (Fig. 8A). In KB cells, the insulin-induced increase in phosphorylation of Ser70 was impaired by the MEK inhibitor AZD6244 and completely blocked by rapamycin (Fig. 8A), indicating it that is mainly regulated via mTORC1. Given that AZD6244 also slightly decreases the activation (phosphorylation) of S6K at Thr389, it seems likely that AZD6244 exerts its effect on this site in part via mTORC1 (since MEK/ERK signaling can contribute to activation of mTORC1 [see, for example, reference 42]). For Ser392, AZD6244 had little effect, whereas rapamycin substantially inhibited its phosphorylation (Fig. 8A). The mTOR kinase inhibitor AZD8055 (14), which blocks all functions of mTORC1 and mTORC2, inhibited the phosphorylation of Ser392 more strongly than rapamycin, and AZD6244/AZD8055 together completely blocked Ser392 phosphorylation (Fig. 8A). Insulin caused only a modest increase in phosphorylation at Ser470, which was partially phosphorylated in starved cells (Fig. 8A). This increase was not affected by AZD6244, only partially inhibited by rapamycin and strongly blocked by AZD8055 (Fig. 8A). Interestingly, insulin also increased the phosphorylation of eEF2K at Ser359, a site that can be phosphorylated by p38 MAP kinase δ (17) and by cdc2/cyclin B (18). Its phosphorylation was partially inhibited by AZD6244 and blocked by either rapamycin or AZD8055 (Fig. 8A). Insulin therefore induces the phosphorylation of all four sites via mTORC1/MEK signaling. The increased mobility of eEF2K caused by inhibition of mTOR is investigated further in Fig. 10.

FIG 8.

Signaling events in the control of specific phosphorylation sites in eEF2K. (A) Serum-starved KB cells were preincubated for 40 min with AZD6244 (A6, 5 μM) and/or 100 nM rapamycin (R, 100 nM) and then stimulated with 100 nM insulin as indicated. (B) Serum-starved KB cells were preincubated for 40 min with PF4708671 (P4; 10 μM), rapamycin (R; 100 nM), or AZD8055 (A8; 1 μM) and then stimulated with 100 nM insulin as indicated. The cells were lysed, and equal amounts of protein were subjected to analysis by Western blotting with the indicated antibodies. In panel A, the data for pSer359 is a reprobe of the blot for pS70; these antibodies were raised in different species, and the reprobe used secondary antisera with a different fluorescent tag.

FIG 10.

mTORC1 signaling regulates Ser392/Ser396 in eEF2K. (A) FLAG-eEF2K was expressed in HEK293 cells, which were treated with AZD8055 (A8; 1 μM for the times shown). Equal amounts of the resulting cell lysates were analyzed by Western blotting with the indicated antibodies. (B) Same as in panel A, except that FLAG-eEF2K was immunoprecipitated from the cell lysates, and, where indicated, treated with λ-phosphatase (30 min). Samples were analyzed in nonadjacent lanes of the same gel by Western blotting with anti-FLAG. (C) FLAG-eEF2K was expressed in HEK293 cells, together with Rheb or empty vector (E.V.), where indicated. Cells were treated with AZD8055 (A8; 1 μM for 40 min) or transferred to D-PBS (90 min). Samples of lysate were analyzed as in panel A. (D) WT FLAG-eEF2K or the S392A/S396A mutant were expressed in HEK293 cells, which were treated with AZD8055 (A8; 1 μM for 40 min) where shown. Samples were subjected to FLAG immunoprecipitation (IP) prior to analysis. (E) HEK293 cells were transfected with vectors encoding the indicated proteins. After 40 h, cells were treated with insulin (100 nM) for 45 min or 1 μM AZD8055 for 40 min before lysis. Samples were analyzed by Western blotting with the indicated antibodies. (F) HEK293 cells were transfected with the indicated vectors for FLAG-eEF2K and mutants. Where shown, cells were treated with AZD8055 (1 μM for 40 min). Samples of cell lysate were processed for eEF2K assay (top two sections) or Western blot (middle two sections) using the indicated antibodies. The bottom section (graph) shows the eEF2K activity data ± the SEM (n = 3). (G) KB cells were transfected with WT-eEF2K or S392A/S396A (SS/AA), S392D/S396D (SS/DD), or S392E/S396E (SS/EE) FLAG-eEF2K. After 24 h, the cells were starved of serum for 16 h and then stimulated with insulin (100 nM) for 45 min. Samples were subjected to FLAG IP prior to analysis. The arrows shown in several panels denote the two main species of eEF2K which were resolved by SDS-PAGE.

These data indicate that a protein kinase(s) that is activated by mTORC1 may be involved in the insulin-stimulated phosphorylation of eEF2K at Ser70, -359, -392, and -470, although we cannot exclude the possibility that mTORC1 (also) regulates the corresponding protein phosphatases. S6 kinases are the only well-characterized mTORC1-activated protein kinases. We therefore examined the effect of PF4708671 (35) on the insulin-induced phosphorylation of these sites in eEF2K. Quantitative analysis of data from multiple experiments showed that PF4708671 had no effect on the ability of insulin to induce the phosphorylation of Ser70, -359, or -392 (Fig. 8B); in any case, their local sequence does not conform to the S6K consensus. It did strongly inhibit the phosphorylation of S6K substrate, S6, at S240/244, confirming its efficiency. The modest residual phosphorylation of S6 at these sites may reflect phosphorylation by S6K2, which is less efficiently inhibited by PF4708671 (35). These data imply that neither S6 kinases nor a (putative) kinase activated by them act on these sites in eEF2K. PF4708671 also failed to affect the phosphorylation of Ser396, although AZD8055 did cause a small decrease, especially in control cells.

To assess whether these sites are also regulated in another cell-type from a different species, we used rat H4IIE hepatoma cells where eEF2 phosphorylation can be strongly regulated (43). Because the sequences around Ser359 and Ser392 in human eEF2K are very similar or identical (respectively) to those around the equivalent sites in rat eEF2K (Table 1; Ser358/Ser391 in rat eEF2K; numbering differs by −1 from the corresponding residues in human eEF2K), we were also able to study the control of their phosphorylation by insulin in rat hepatoma cells. (We cannot study the other sites because the antibodies do not cross-react due to sequence differences [Table 1].) Hepatoma cells were treated with insulin in the presence or absence of AZD6244 and/or rapamycin. A (weak) signal was detected for Ser395 even in serum-starved cells (Fig. 9A), and this was not increased by insulin or affected by these signaling inhibitors, at least over the time periods studied here.

FIG 9.

Regulation of eEF2K in rat hepatoma cells. (A) Serum-starved hepatoma cells were treated with insulin (100 nM) for the indicated times. Equal amounts of the resulting cell lysates were analyzed by Western blotting with the indicated antibodies (the positions of sites in rat eEF2K and their equivalents in the human sequence are indicated). Where indicated, cells were preincubated with AZD6244 (A6; 5 μM) and/or rapamycin (R; 100 nM) for 40 min. (B) Same as in panel A, but the cells were pretreated, as shown, with roscovitine (10 μM for 40 min). (C) Recombinant GST-eEF2K was incubated with the indicated kinases and radioactive [γ-³²P]ATP for the times shown. After resolving the sample by SDS-PAGE, the gel was stained with Coomassie blue (lower part) and dried, and the radioactivity was visualized by using a phosphorimager (upper part). (D) Same as in panel C, but GST-eEF2K(S359A) was analyzed alongside wild-type eEF2K, and only ERK was used.

While the phosphorylation of Ser358 in eEF2K was almost undetectable in serum-starved cells, it was strongly increased by insulin (Fig. 9A). AZD6244 and rapamycin each partially inhibited this increase, at both times tested and, in combination, they blocked it almost completely. AZD6244 did not interfere with the phosphorylation of S6 at Ser240/244, confirming that inhibition of MEK signaling does not impair mTORC1 signaling and the S6 kinases in this setting. These data are important because they show that the effect of AZD6244 on phosphorylation of Ser358 is not due to effects on an input mediated via mTORC1. Thus, insulin induces Ser358 phosphorylation via a combination of MEK and mTORC1 signaling. An earlier study showed IGF-1 regulated this site via mTORC1 signaling in KB cells (17). The phosphorylation of Ser391 behaved similarly to that of Ser358 (Fig. 9A).

Cdc2/cyclin B can also phosphorylate Ser359 in human eEF2K (18) but appears not to be involved here since the cdc2 inhibitor roscovitine did not affect the ability of insulin to elicit phosphorylation of Ser358 in hepatoma cells (Fig. 9B).

ERK directly phosphorylates eEF2K.

In the light of these data, it was important to determine how MEK signaling regulates the phosphorylation of Ser358/359 (in rat/human eEF2K); this residue is followed by a proline, so it could be a direct substrate for ERK. ERK phosphorylated eEF2K to a similar level to that seen with cdc2/cyclin B, which efficiently phosphorylates eEF2K (18) (Fig. 9C), a finding consistent with earlier data (17).

To identify the site(s) in eEF2K which is phosphorylated by ERK, recombinant eEF2K, expressed in E. coli, was radiolabeled using ERK and subjected to tryptic digestion and analysis by reversed-phase high-pressure liquid chromatography. Several phosphorylated peptides were identified (data not shown), including ones phosphorylated on Ser18, Ser74, Thr254, Ser359, or Ser396, all of which are followed by a proline, consistent with the specificity of ERK. We therefore created five mutants of eEF2K in which each of these sites was individually mutated to an alanine residue. eEF2K(S18A), eEF2K(S74A), and eEF2K(S254A) were phosphorylated to similar extents as WT eEF2K, indicating that these are, at most, very minor sites of phosphorylation by ERK (data not shown). In contrast, eEF2K[(S359A) showed markedly decreased phosphorylation by ERK; at early times, phosphorylation was decreased by >60%, indicating that the principal site phosphorylated rapidly by ERK is Ser359 (Fig. 9D), although other sites do undergo phosphorylation, albeit probably more slowly. We therefore tested an eEF2K(S359A/S396A) double mutant; this still showed similar residual phosphorylation to eEF2K(S359A), indicating that Ser396 is only a minor phosphorylation site here (data not shown). Thus, ERK mainly phosphorylates eEF2K on Ser359, although it can weakly phosphorylate additional sites in vitro. Importantly, phosphorylation of eEF2K on Ser359 strongly inhibits its activity (17).

Ser391 (like Ser392 in human eEF2K) does not lie in an ERK or mTORC1 consensus, and it remains to be established how signaling through mTORC1 and MEK controls its phosphorylation.

mTORC1 regulates Ser392/Ser396 in the linker region.

To explore the reason for the alterations in the mobility of endogenous eEF2K caused by signaling inhibitors, we studied the behavior of FLAG-tagged eEF2K expressed in HEK293 cells (use of ectopic eEF2K allows us to create and test specific mutants). We noted that AZD8055 caused a marked time-dependent shift in the migration of eEF2K on SDS-PAGE to a faster-migrating species (Fig. 10A; also Fig. 8A and B). To assess whether this reflected dephosphorylation of eEF2K, we incubated eEF2K from control or AZD8055-treated cells with λ phosphatase. This caused the upper band to shift to the position of the lower one (Fig. 10B), but it had no effect on the mobility of the faster-moving species from AZD8055-treated cells. Thus, dephosphorylation appeared the probable explanation for the AZD8055-induced mobility shift. The rather slow nature of this effect suggests that mTOR's involvement here might be indirect, perhaps through transcriptional induction of a phosphatase or repression of the actual kinase.

Since AZD8055 inhibits both mTORC1 and mTORC2, the shift could result from inhibition of signaling through either of them. In contrast, it is generally accepted that starving cells of amino acids inhibits mTORC1 but not mTORC2 (44). Shifting cells to medium without amino acids (Dulbecco's phosphate-buffered saline [D-PBS]) caused a similar shift in the mobility of FLAG-eEF2K that was prevented by overexpressing Rheb (Fig. 10C), which opposes the effect of starvation on mTORC1 signaling (see phosphorylation of RpS6 in Fig. 10C). This indicates that the dephosphorylation event(s) causing the mobility shift reflects the inhibition of mTORC1.

We noted that Ser392 and Ser396 were phosphorylated in the upper band but not the lower one (Fig. 10A), suggesting these might be the sites whose dephosphorylation caused the shift. We therefore created eEF2K(S392A/S396A). This mutant protein migrated mainly as a faster-moving species similar to that seen for wild-type eEF2K after treatment with AZD8055, and this compound did not cause any further shift in its migration. We also checked several other mutants, including eEF2K(S27A/S31A) and eEF2K(S70A/S74A). All behaved similarly to wild-type eEF2K (data not shown). These observations indicate that the dephosphorylation of Ser392 and Ser396 contributes to the mobility shift caused by AZD8055 or inhibition of mTORC1 signaling by amino acid starvation. Expression of Rheb promoted the phosphorylation of Ser392 and Ser396 in amino acid-starved cells but did not prevent the effect of AZD8055. This indicates that these sites are regulated by mTORC1 (Fig. 10C). We also created the potentially phosphomimetic mutants eEF2K(S392D/S396D) and eEF2K(S392E/S396E). Both these mutant proteins migrated more slowly than eEF2K(S392A/S396A) and AZD8055 did not cause them to shift to the faster-migrating species (Fig. 10E), supporting the conclusion that dephosphorylation of Ser392/396 in eEF2K is the principal cause of the AZD8055-induced mobility shift.

We noted that the phosphorylation of Ser70 and Ser78 was reduced in eEF2K(S392A/S396A) (Fig. 10D), indicating that phosphorylation of Ser392/Ser396 affects phosphorylation at the N-terminal sites, at least one of which, Ser78, strongly affects CaM binding (16). Consistent with this, eEF2K(S392A/S396A) showed higher basal binding to CaM (Fig. 10D). Furthermore, AZD8055 markedly increased the binding of WT eEF2K to CaM but had no effect for the S392A/S396A mutant (Fig. 10D). In line with the enhanced binding to CaM, eEF2K(S392A/S396A) showed higher basal activity than the wild-type enzyme, similar to that of wild-type eEF2K from cells treated with AZD8055 (Fig. 10F).

We also evaluated the ability of insulin to induce the phosphorylation of Ser70 and Ser78 in the phosphomimetic mutants. In KB cells, the insulin-stimulated phosphorylation of both was enhanced when Ser392/6 were mutated to phosphomimetic acidic residues, especially in the case of eEF2K(S392E/S396E) (Fig. 10G).

Ser392/Ser396 affect the accessibility of Ser70/Ser78 for phosphorylation.

Given that the C-terminal part of eEF2K may recruit substrates to the catalytic domain (Fig. 1), it was possible that phosphorylation of Ser392/396 altered eEF2K's conformation in such a way that it affected the accessibility of Ser70 and Ser78 to the as-yet-unknown kinases that act on them. It could therefore be informative to study this using C-terminally truncated versions of eEF2K; however, this was not feasible, since they were unstable in mammalian cells.

As an alternative, we therefore expressed WT eEF2K or eEF2K(S392E/S396E) in E. coli and tested their ability to undergo phosphorylation at Ser70 by GSK3 in vitro in the presence or absence of the relevant priming kinase, Cdk5 (Fig. 5). (We adopted this approach since the physiological kinase acting on Ser70 is not known). GSK3 phosphorylated Ser70 much more rapidly in eEF2K(S392E/S396E) than for WT eEF2K (Fig. 11A). With the caveat that these kinases do not seem to be physiologically relevant ones, the data indicate that modification of Ser392/396 renders the N-terminal sites in eEF2K more accessible to phosphorylation. Importantly, this indicates interplay between the control of phosphorylation sites in the N-terminal region of eEF2K and those in the linker region.

FIG 11.

mTOR phosphorylates eEF2K in vitro. (A) GST-eEF2K or GST-eEF2K(S392E/S396E) were expressed in E. coli, purified, and then incubated with cdk5 for 20 min before adding GSK3 all in the presence of nonradioactive ATP. Samples were collected 5, 10, 15, and 20 min after the addition of GSK3 and analyzed by SDS-PAGE and Western blotting with the indicated antibodies. (B) In vitro phosphorylation of bacterially expressed recombinant GST-4E-BP1/GST-eEF2K by mTOR immunoprecipitates from HEK293 cells (using [γ-³²P]ATP) in the presence or absence of AZD8055 (1 μM; incubation time, 20 min). Radioactivity was detected by phosphorimager analysis. The lower section shows the Coomassie blue staining of the same gel. (C) Same as for panel B, but using nonradioactive ATP. The reaction time was 1 h, and phosphorylation was detected by immunoblotting with the indicated phospho-specific antibodies. (D) Schematic representation of suggested model for control of eEF2K via multisite phosphorylation.

mTORC1 phosphorylates Ser78 and Ser396 in eEF2K.

To test whether mTOR directly phosphorylates eEF2K, recombinant GST-eEF2K was incubated with mTOR immunoprecipitated from cells and [γ-³²P]ATP. Clear phosphorylation of GST-eEF2K and a known mTOR substrate, 4E-BP1, was observed (Fig. 11B), while GST itself was not phosphorylated (data not shown). Phosphorylation of both eEF2K and 4E-BP1 was strongly inhibited by AZD8055, showing that mTOR was mainly responsible.

To test which site(s) in eEF2K is phosphorylated by mTOR, we conducted similar assays with nonradioactive ATP, analyzing the products with our phospho-specific antisera. Clear phosphorylation of Ser396 was observed, which was eliminated by AZD8055 (Fig. 11C), indicating it is catalyzed by mTOR. This is consistent with our finding that phosphorylation of Ser396 decreases in response to amino acid starvation or AZD8055 treatment. Ser78 was also phosphorylated by mTOR in an AZD8055-sensitive manner, but we either saw no signal for the other sites tested (Ser70/Ser366/Ser392/Ser470) or phosphorylation was insensitive to AZD8055, suggesting a contaminating kinase was responsible (Ser359). Thus, mTOR can phosphorylate Ser78 and Ser396 in eEF2K. However, while Ser396 is constitutively phosphorylated in cells in medium containing amino acids (likely due to basal mTORC1 activity) and is insensitive to rapamycin, Ser78 is regulated by insulin in a rapamycin-sensitive manner (16). Interestingly, Ser396 and Ser392 affect the phosphorylation of Ser78 in vivo (Fig. 10G).

The mutations at Ser366 and Ser396 which occur in cancer cells enhance eEF2K activity.

Inspection of the COSMIC (Catalogue of Somatic Mutations in Cancer) database (http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/) revealed a number of mutations in the EEF2K gene that occur in solid tumors, including two at regulatory phosphorylation sites controlled by mTORC1 signaling, S366F (reported in esophageal or urinary carcinomas) and S396F (malignant melanoma). We also created more conservative Ser-to-Ala mutations at these sites. The mutants and wild-type eEF2K were expressed in HEK293 cells. Both Ser-to-Phe mutants and the S366A mutant bound more CaM (Fig. 12A). The Phe variants also displayed markedly higher activity than WT eEF2K (Fig. 12B). Levels of phosphorylation of the N-terminal sites Ser70 and especially Ser78 were markedly decreased in these mutants (Fig. 12A); this likely explains their enhanced levels of CaM binding and activity. These mutations may also override the inputs to these sites from mTORC1 signaling which is activated in many cancers. These data also further confirm the ability of alterations in the linker region to influence events at the N terminus, presumably through conformational changes, and thus to affect eEF2K activity.

FIG 12.

The S366F and S396F mutations in eEF2K activate it. WT FLAG-tagged eEF2K or the indicated mutants were expressed in HEK293 cells. Where indicated, cells were treated with AZD8055 (AZD; 1 μM for 40 min). (A) After lysis and FLAG IP, samples were analyzed by Western blotting with the indicated antibodies. (B) eEF2K activity was analyzed using eEF2 as the substrate. Top section shows a phosphorimage of the Coomassie blue-stained stained, radioactive gel. The lower section shows Western blots of the corresponding cell lysates. Also shown is quantitation of the activity data from three independent experiments ± the SEM, where the activity of eEF2K from untreated cells is equal to 1.

Overall, the finding that these mutations activate eEF2K are consistent with the emerging concept that eEF2K is cytoprotective in tumor cells or tumors (see, for example, reference 23) and that it is overexpressed or shows increased activity in certain solid tumors (23, 45). Our data show that, in addition to enhanced expression levels, activating mutations in eEF2K also occur in human tumors.

DISCUSSION

Our data identify additional inputs from oncogenic signaling pathways (ERK, mTORC1) to the regulation of eEF2K, which likely serve to restrain eEF2K activity in cancer cells, to allow eEF2 to remain active and mediate the high rates of protein synthesis required for cancer cell growth and proliferation. These inputs include phosphorylation events under the control of the MEK/ERK and mTORC1 signaling pathways, which are activated in a high proportion of tumors, and from GSK3. Under conditions of nutrient starvation, where eEF2K helps protect cancer cells (see, for example, reference 23), eEF2K will be activated by the AMP-activated protein kinase, an important energy sensor (46), and due to impaired mTORC1 signaling which occurs under such conditions. Table 2 summarizes the available information on the control of the phosphorylation sites in eEF2K examined here, highlighting the new data provided in the present study.

TABLE 2.

Summary of phosphorylation sites in eEF2K studied here

Site (in human eEF2K)	Effect of insulin	Signaling pathway	Kinase/comments	Source and/or reference
Ser70	Increase	mTORC1	Not known	First reported here
Ser78	Increase	mTORC1	mTOR	Site and effect of insulin reported by Browne and Proud (6)
Ser359	Increase	MEK/ERK mTORC1	Probably ERK; not known	Site reported by Knebel et al. (17); role of ERK first reported here
Ser366	(Increase)	MEK/ERK and mTORC1	p90RSK and S6Ks	Reported by Wang et al. (15)
Ser377	Increase	MEK/ERK/MK2	MK2?	Site reported by Knebel et al. (22); effect of insulin first reported here
Ser392	Increase	mTORC1 ERK?	Basally GSK3; not known for insulin effect	First reported here
Ser396	No change	mTORC1	mTOR	Site reported by Knebel et al. (22); role of mTOR first reported here
Ser470	Increase	??	Not known	First reported here

These signaling links probably also serve to couple the relevant pathways to the control of translation elongation in response to physiological signals such as nutrients or growth factors. Previous studies using a range of stimuli in several different cell types have shown that agonist-induced dephosphorylation of eEF2 is mediated through the mTORC1 signaling pathway, as evidenced by the observation that in all case the dephosphorylation of eEF2 was blocked by rapamycin (10, 15, 21, 43, 47). Consistent with this, earlier work identified three phosphorylation sites in eEF2K which are regulated in an mTORC1-dependent manner and phosphorylation of which causes the inactivation of eEF2K. These are Ser78 (16), Ser359 (17, 18), and Ser366 (15).

The present studies identify several new mTORC1-regulated phosphorylation sites, show that Ser78 and Ser396 in eEF2K are direct substrates for mTOR and reveal that the phosphorylation of some sites is regulated by the phosphorylation status of other sites remote from them in the sequence of eEF2K, which likely reflects conformational rearrangements caused by the phosphorylation of the linker sites, allowing better access for the kinases that phosphorylate Ser70 and Ser74 (i.e., in the latter case, for mTOR) (Fig. 11D). We made strenuous attempts to use FRET to examine conformational changes in (tagged) eEF2K in living cells but were unable to observe energy transfer. It is clearly an important to gain structural insights into the interactions and interplay between these domains, as well as the linker and extreme N-terminal, CaM-binding regions of eEF2K. Our findings also demonstrate that eEF2K can be controlled by mTORC1-independent mechanisms. For example, we show that ERK phosphorylates eEF2K at a site (Ser359) that profoundly inactivates eEF2K (22). This may be important in the control of protein synthesis by ERK signaling, which is often activated in cancers. The currently known inputs to the regulation of eEF2K are summarized in Fig. 13. The mechanisms by which mTORC1 signaling controls the phosphorylation sites that are not direct substrates for mTOR may include effects via the regulation of the relevant phosphatases.

FIG 13.

Schematic summary of signaling links to eEF2K. This includes both previously known links (Fig. 1) and those first identified in the present study. Solid lines denote direct phosphorylation events, and dashed lines indicate indirect ones.

In hepatoma cells, phosphorylation of Ser358 in eEF2K, an event which strongly inhibits eEF2K activity, is under the combined control of MEK- and mTORC1-dependent signaling. Indeed, in these cells, insulin-induced dephosphorylation of eEF2 can be blocked by inhibiting MEK/ERK and mTORC1 signaling, indicating that these two pathways independently regulate eEF2 phosphorylation, presumably via the control of eEF2K through its phosphorylation at some or all of the sites mentioned above. The input from MEK is likely mediated by ERK, which directly phosphorylates Ser359 in eEF2K. The link from mTORC1 signaling to the regulation of this site remains unclear; although p38 MAPKδ and cdc2/cyclin B can also phosphorylate this site (17, 18), the former is not regulated by mTORC1, while the data for roscovitine rule out cdc2/cyclin B. An unknown, mTORC1-regulated kinase may be responsible.

Of the additional mTORC1-regulated phosphorylation sites in eEF2K which we have identified (Ser70, Ser392, Ser396, and Ser470), only Ser396 has been reported previously (22) and was not known to be controlled via mTORC1 (Fig. 13). Importantly, we show that phosphorylation of Ser392 and Ser396 modulates the phosphorylation of Ser70 and Ser78 (a key regulator of CaM binding to eEF2K), pointing to long-range interactions (e.g., conformational changes) between the central linker region of eEF2K and its N terminus, such that phosphorylation of Ser392/396 facilitates phosphorylation of Ser70 and Ser78, thereby controlling binding of eEF2K to CaM. This complements our earlier observations that cooperation between the C-terminal “SEL1-repeat” region of eEF2K and the N-terminal catalytic domain is essential for efficient phosphorylation of substrates by eEF2K (29).

Our data show that eEF2K is phosphorylated by mTOR, providing a direct input from this pathway to the control of eEF2K. The sequence contexts at these two sites are consistent with the specificity of mTOR (48) (Ser78 is followed by an aromatic residue [Phe] and Ser396 by proline). However, unlike other direct substrates for mTORC1 (49), eEF2K does not contain an obvious TOR-signaling (TOS) motif.

Our data also show for the first time that regulation of GSK3 provides an input to the control of eEF2K in response to insulin, although further work is needed to define the molecular mechanisms involved. Ser392 is thus far the only residue which appears to be phosphorylated by GSK3 in cells and, since insulin increases phosphorylation of this site, GSK3 is not the major insulin-regulated input affecting it. Therefore, Ser392 cannot account for the regulation of eEF2K by GSK3.

eEF2K is subject to an array of regulatory inputs, suggesting its proper control is important in diverse settings, and consistent with its key roles, e.g., in cell survival (23) and learning/memory (50).