Stimulation of the AMP-activated protein kinase leads to activation of eukaryotic elongation factor 2 kinase and to its phosphorylation at a novel site, serine 398

: Protein synthesis consumes a high proportion of the metabolic energy of mammalian cells, and most of this is due to peptide-chain elongation. An important regulator of energy supply and demand in eukaryotic cells is the AMP-activated protein kinase (AMPK). The rate of peptide-chain elongation can be modulated through the phosphorylation of eukaryotic elongation factor (eEF) 2, which inhibits its activity and is catalysed by a specific calcium/calmodulin-dependent protein kinase termed eEF2 kinase. Here we show that AMPK directly phosphorylates eEF2 kinase and we identify the major site of phosphorylation as Ser398 in a regulatory domain of eEF2 kinase. AMPK also phosphorylates two other sites (Ser78, Ser366) in eEF2 kinase in vitro. We develop appropriate phosphospecific antisera and show that phosphorylation of Ser398 in eEF2 kinase is enhanced in intact cells under a range of conditions that activate AMPK and increase the phosphorylation of eEF2. Ser78 and Ser366 do not appear to be phosphorylated by AMPK within cells. Although cardiomyocytes appear to contain a distinct isoform of eEF2 kinase, it contains a site corresponding to Ser398 that is phosphorylated by AMPK in vitro . Stimuli that activate AMPK and increase eEF2 phosphorylation within cells increase the activity of eEF2 kinase. Thus, AMPK and eEF2 kinase may provide a key link between cellular energy status and the inhibition of protein synthesis, a major consumer of metabolic energy. Abbreviations: 4E-BP1, eukaryotic initiation factor 4E-binding protein 1; ACC, acetyl-CoA carboxylase; AICAR, 5-amino-4-imidazolecarboxamide riboside; AMPK, AMP-activated protein kinase; ARVC, adult rat ventricular cardiomyocytes; CaM, calmodulin; CCCP, carbonyl cyanide 3-chlorophenylhydrazone; 2-DOG, 2-deoxy-D-glucose; eEF, eukaryotic elongation factor; GST, glutathione S -transferase; mTOR, the mammalian target of rapamycin; S6K1, the 70 kDa ribosomal protein S6 kinase, isoform 1; TFP, trifluoperazine.

9 immersion in 75 mM phosphoric acid. After washing four times with 75 mM phosphoric acid and then once with acetone, the papers were dried and analysed by Čerenkov counting. One unit (U) of kinase activity is the amount of enzyme that phosphorylates 1 nmol of peptide in 1 min.

Phosphorylation of eEF2 kinase by AMPK and protein chemical analyses
The AMPK used was an extensively-purified preparation (5U/ml) from rat liver, purified through five steps as far as the gel filtration column, and was kindly provided by Professor D. G.
Hardie, University of Dundee (22). This material was then further immunopurified as indicated using antibodies raised against the α1 and α2 catalytic subunits of AMPK coupled to protein G Sepharose and used to phosphorylate GST-eEF2 kinase as described below.
Phosphorylation of bacterially expressed GST-eEF2 kinase (1µg fusion protein per reaction) was carried out using 5mU of immunopurified AMPK in AMPK assay buffer plus 200µM ATP (1000cpm/pmol [γ− 32 P]ATP), where indicated, 200µM AMP (where indicated), and 15nM MgCl 2 . For eEF2 kinase autophosphorylation reactions the immunopurified AMPK and AMP were omitted from the reaction and 0.5 µg/ml calmodulin plus 1mM CaCl 2 added to the reaction mixture. Incubations were at 30 o C for 15 min and were stopped by addition of SDS-PAGE loading buffer and heating to 96 o C for five min. Incorporation of phosphate into GST-eEF2 kinase was determined following electrophoresis of samples on a SDS-PAGE gel by autoradiography or following western transfer to Immobilon-P membrane by immunoblotting using phosphorylated residue specific antibodies.
To map the sites on eEF2 kinase phosphorylated by AMPK GST-eEF2 kinase was incubated with immunopurified AMPK as described above except that the reaction was (Milligen) as described previously (23). Phosphoamino acid analysis was used to confirm the type of residue phosphorylated in each tryptic peptide.
The thin-layer cellulose plates were allowed to dry and then subjected to autoradiography.

Expression and mutagenesis of GST-eEF2 kinase
A modified cDNA encoding human eEF2 kinase with an N-terminal FLAG tag cloned between the BamHI sites of the vector pGEX-4T was kindly provided by Dr Maria Deak of the Division of Signal Transduction Therapy, University of Dundee. For expression of GST-eEF2 kinase in E.coli BL21 cells were transformed with this vector and induced with isopropyl β-thiogalactoside (100 µM) for 3 h at 26°C. Cells were lysed (by sonication) into 50mM Tris pH 7.5, 150mM NaCl, 0.03% Brij-35, 1mM EDTA, 0.1mM EGTA, 5% (v/v) glycerol, 0.1% (v/v) βmercaptoethanol and proteinase inhibitors (1 mM each of phenylmethylsulfonyl fluoride, leupeptin, benzamidine and pepstatin), and extracts were prepared. GST-eEF2 kinase was isolated by chromatography on glutathione-Sepharose and eluted using 100 mM glutathione.
For expression as a maltose binding protein fusion, human eEF2 kinase was subcloned into the modified pMal-HA vector (kindly provided by Dr Jane Leitch, Division of Signal Transduction Therapy, University of Dundee) using the BamH1 restriction sites. Expression and purification were as described above except that chromatography was on amylose resin and elution with 20 mM maltose.
Mutagenesis of the Ser78 phosphorylation site in human eEF2 kinase to alanine was performed by PCR using 'QuikChange' (Stratagene). The forward primer was 5'-CGGCAAACGCCTTCCACTTCAAGGAAGCC -3' and the reverse primer was 5'-

AMPK phosphorylates eEF2 kinase
We and others have shown previously that conditions that activate the AMP-activated protein kinase (AMPK) lead to increased phosphorylation of eEF2 and activate eEF2 kinase (11)(12)(13). We therefore studied whether eEF2 kinase was a substrate for AMPK in vitro. Recombinant human eEF2 kinase was made as a GST-fusion protein in E. coli. It was then incubated with a highly purified preparation of AMPK and [γ-32 P]ATP as described in the Experimental procedures. As shown in Fig. 1A, this resulted in the incorporation of radiolabel into eEF2 kinase which was markedly stimulated by addition of AMP to the reaction, confirming that phosphorylation is indeed due to AMPK. As an alternative, we used immunopurified AMPK, which also catalysed efficient incorporation of radiolabel into GST-eEF2 kinase. However, no such incorporation was seen when we tested the pellet from a mock immunopurification in which pre-immune serum rather than anti-AMPK was used (Fig. 1A). To identify the site(s) phosphorylated by AMPK, eEF2 kinase was radiolabeled by AMPK and then subjected to tryptic digestion. The resulting peptides were resolved by reverse-phase HPLC: two main peaks of radioactive material were observed (Fig. 1B). Further analysis by MALDI mass spectrometry revealed that they corresponded to the peptides containing residues 69 -82 (peak 1) or 364 -406 of eEF2 kinase (peak 2; Fig. 1C,D).
Solid-phase Edman degradation revealed that the peptide in peak 1 was phosphorylated at position 10, indicating that it is YSSSGSPANSpFHFK, where the phosphorylated residue is Ser78 of eEF2 kinase (Fig. 1C). Peak 2 contains a very large peptide and, to facilitate further analysis, it was subdigested with Asp-N prior to Edman sequencing. This sample showed release of label at three cycles -1, 3 and 8 (Fig. 1D). Inspection of the sequences of the predicted products of Asp-N digestion shows that the release at cycle 3 could be due to Ser366 or Thr389.
However, since phosphoamino acid analysis revealed only P-Ser and no P-Thr (data not shown), the Thr389 can be ruled out. The only possible explanation for release at cycle 8 is that it corresponds to Ser398 in the most C-terminal of the Asp-N fragments derived from the original tryptic peptide. The radioactivity at cycle 1 is likely due to leaching from the matrix of some of this peptide that is coupled via the side chain of its N-terminal Asp residue rather than the Cterminal α-carboxy group. Each Edman sequencing run proceeded through sufficient cycles to go beyond the end of the peptide: thus, there can be no further sites of phosphorylation within these peptides. These data indicate that, in vitro, AMPK phosphorylates three sites, Ser78, Ser366 and Ser398, in eEF2 kinase. Other data suggest that Ser398 is the major site phosphorylated by AMPK (see below).
Ser366 and Ser398 lie C-terminal to the catalytic domain of eEF2 kinase (Fig. 1E) in a region that contains several known sites of phosphorylation (12,17,18). Ser398 has not previously been identified as a phosphorylation site in eEF2 kinase. Ser78 is also a novel site: it lies Nterminal to the catalytic domain, immediately adjacent to the CaM binding site (26,27) (Fig. 1E).
To confirm the identity of the three major sites of in vitro AMPK phosphorylation in eEF2 kinase and for use as tools for investigating the regulation of the in vivo phosphorylation of these sites we generated phosphospecific antisera for each of the two novel sites (Ser78, Ser398) by immunising rabbits with appropriate phosphopeptides. Antibodies were purified as described in the Experimental procedures. Fig. 2A-C shows data for their characterisation. When GST-eEF2 kinase was incubated with AMPK and cold ATP and subsequently analysed by SDS-PAGE and western blotting, clear signals were seen using the anti-(P)Ser78 and anti-(P)Ser398 antisera, as well as for a phosphospecific antiserum for Ser366 that has previously been developed by us (10) ( Fig. 2A-C). No signal was observed for the anti-(P)Ser78 and (P)Ser398 antisera when mutants of eEF2 kinase were used in which these sites had been individually mutated to alanines, thus demonstrating the specificity of the phosphospecific antisera for their intended epitopes.
eEF2 kinase undergoes extensive autophosphorylation in the presence of its activators, Ca-ions and CaM (28,29). The data in Fig. 2A also show that neither Ser78 nor Ser366 is a site of autophosphorylation, since neither becomes phosphorylated when eEF2 kinase is incubated with Ca 2+ -ions and CaM in the absence of AMPK.
Although the AMPK used here was obtained by specific immunoprecipitation from an already highly purified preparation of enzyme, it was theoretically possible that coimmunoprecipitating kinases might contribute to the observed phosphorylation of eEF2 kinase.
To confirm that these three sites are indeed phosphorylated by AMPK rather than any possible co-immunoprecipitating kinases, we performed phosphorylation reactions using the immunopurified AMPK in absence or presence of AMP. As shown in Fig. 2D, AMP very markedly increased the phosphorylation of all three sites. This strongly suggests that their phosphorylation is indeed catalysed by AMPK.
Serine 398 is the major site of AMPK phosphorylation in eEF2 kinase.
In order to determine which sites in eEF2 kinase were preferentially phosphorylated by AMPK we again radiolabeled eEF2 kinase in vitro using AMPK and subjected the labeled protein to digestion successively with trypsin and the proteinase Asp-N. The resulting peptides were then analysed by two-dimensional mapping followed by autoradiography to locate radiolabeled species. As shown in Fig. 3A, three radiolabeled peptides were observed. These comprise two minor phosphopeptides (labeled 'a' and 'b' in Fig. 3A) and one major species ('c' in Fig. 3A). To identify which of these are due to phosphorylation at Ser398, we performed a similar analysis using a mutant of eEF2 kinase in which Ser398 is mutated to alanine. As shown in Fig. 3A, the major species ('c') seen in the map from wildtype eEF2 kinase is entirely absent from the map derived from the Ser398Ala mutant strongly indicating that Ser398 is the major site in eEF2 kinase that is phosphorylated by AMPK in vitro. The phosphopeptides 'a' and 'b' were identified in a similar manner as the Ser366 and Ser78 phosphopeptides respectively (GJB, unpublished data). Ser398 and adjacent amino acids are completely conserved in the known sequences of eEF2 kinases from mammals.
eEF2 kinase undergoes extensive autophosphorylation, at least in vitro (see, e.g., (28,29)), and this can affect its activity, e.g., render it partially independent of Ca 2+ /CaM (29). It was therefore important to establish whether Ser398 was a site of autophosphorylation. We therefore incubated GST-eEF2 kinase with either AMPK or Ca 2+ -ions and CaM in the presence of cold ATP. Although a clear signal was seen using the anti-Ser398 phosphospecific antibody when eEF2 kinase was pretreated with AMPK, none was seen under autophosphorylation conditions ( Fig. 3B) showing that Ser398 is not a site for autophosphorylation in eEF2 kinase.
We have previously shown that ATP depletion or treatment of cardiomyocytes with an agent that activates AMPK (AICAR) leads to increased phosphorylation of eEF2 (12). There is some evidence that cardiac myocytes may possess a distinct form of eEF2 kinase (12) although this enzyme has neither been purified nor cloned. To examine whether cardiac eEF2 kinase contained a phosphorylation site corresponding to Ser398, we subjected a cardiac extract to chromatography on FPLC (Mono Q Column) as described earlier (12), in order to partially purify eEF2 kinase. Fractions were monitored for eEF2 kinase activity. The material in the peak fraction was subjected to immunoprecipitation with anti-eEF2 kinase antibodies and the immunoprecipitate was incubated with AMPK and unlabeled ATP. The reaction products were analysed by SDS-PAGE followed by immunoblot analysis using the anti-Ser398 phosphospecific antibody. As shown in Fig. 3C, following treatment of this material with AMPK and cold ATP, a signal was detected with the anti-Ser398 antiserum, indicating that eEF2 kinase from cardiac tissue contains a site corresponding to Ser398 that is phosphorylated by AMPK. A reaction using GST-human eEF2 kinase was performed as a positive control. To be certain that this represents eEF2 kinase from cardiomyocytes, rather than other cell types present in the heart, we immunoprecipitated eEF2 kinase from isolated cardiomyocytes and again treated this with AMPK, followed by analysis by SDS-PAGE and western blotting. As shown in Fig. 3D, a clear signal was again observed for the Ser398 phosphospecific antibody, confirming that a site corresponding to Ser398 is a substrate for AMPK in eEF2 kinase from cardiomyocytes.

Ser78 and Ser398 are both phosphorylated in intact cells
Ser366 has previously been shown to be a substrate for S6K1 and p90 RSK (10) and phosphorylation here inhibits the activity of eEF2 kinase. It was therefore surprising to find that AMPK phosphorylates this site, given that conditions that activate AMPK increase, rather than decrease, the phosphorylation of eEF2. It was therefore important to study whether Ser78, Ser366 and Ser398 are phosphorylated within cells in a manner consistent with their being physiological targets of AMPK i.e. under conditions where AMPK is activated. To test this, eEF2 kinase was immunoprecipitated either from lysates of serum-fed KB cells or from lysates of cells that had been pretreated with relatively low concentrations of 2-DOG, which depletes cellular ATP leading to a rise in AMP levels. We measured the activity of AMPK using a peptide substrate to confirm that 2-DOG treatment does indeed stimulate it in KB cells (Fig. 4A).
To study the phosphorylation of eEF2 kinase, it was also immunoprecipitated from the cell lysates, and then analysed by SDS-PAGE and western blotting. In serum-fed cells, a clear signal was seen with the anti-(P)Ser78 antibody indicating that Ser78 is indeed phosphorylated within cells (Fig. 4B). However, under these conditions no signal was seen with the anti-(P)Ser398 antiserum. Treatment of cells with 2-DOG led to increased phosphorylation of Ser398 Immunoblotting of samples of cell lysate with the phosphospecific (anti-[P]Thr56) antibody for eEF2 itself revealed that 2-deoxyglucose increased the phosphorylation of eEF2 itself (Fig. 4C), under conditions where AMPK is activated as shown by the increased phosphorylation of ACC at Ser79 (Fig. 4C). This is consistent with our earlier findings for the effects of ATP-depletion on eEF2 phosphorylation in other cell types (11,12).
The above data suggest that Ser398 in eEF2 kinase may be a direct target for AMPK within cells, but strongly imply that Ser78 and Ser366 are not. It was thus possible that Ser78 and Ser366 are only phosphorylated by AMPK in vitro and not physiologically, perhaps because phosphorylation at these sites by AMPK is relatively inefficient compared to that of Ser398.
We also studied whether 2-DOG treatment affected the phosphorylation of other sites that have recently been identified in eEF2 kinase, i.e., Ser359, Ser377 and Ser396, for which phosphospecific antisera are also available (17,18). In the case of Ser359, 2-DOG treatment resulted in a decrease in phosphorylation. For Ser377, a marked increase was observed, while for Ser396 no change was seen (Fig. 4B). Ser377 has previously been reported to be phosphorylated by MAP kinase-activated protein kinase 2 (18). This enzyme lies downstream of the SB203580sensitive p38 MAP kinase α/β pathway. We therefore tested whether SB203580 affected the 2-DOG glucose-induced phosphorylation of this site. As shown in Fig. 4D, SB203580 completely eliminated the ability of 2-DOG to increase the phosphorylation of this site, suggesting its phosphorylation is mediated through the p38 MAP kinase α/β pathway, not via AMPK.
Phosphorylation of Ser398 in eEF2 kinase and the AMPK substrate ACC was not affected by SB203580, indicating that this drug does not interfere with the activity or activation of AMPK.
Furthermore, in vitro experiments in which eEF2 kinase was incubated with AMPK and samples were then analysed by SDS-PAGE/western blotting using anti-(P)Ser377 antisera showed that this site cannot be phosphorylated by AMPK in vitro. Thus the increased phosphorylation of Ser377 seen in response to 2-DOG is mediated via activation of the stress-regulated p38 MAP kinase pathway rather than through AMPK.

Ser398 is phosphorylated in cells under conditions where AMPK is activated and eEF2
phosphorylation increases.
Since the data described above suggest that Ser78 and Ser366 are very unlikely to be intracellular targets for AMPK, the remainder of this study focuses on Ser398, the main site that is phosphorylated by AMPK in vitro and one whose phosphorylation does increase in response to AMPK activation induced by 2-DOG.
However, it was conceivable that the effects of 2-DOG on the phosphorylation of eEF2 kinase might be due to consequences of ATP depletion other than the activation of AMPK. For example, severe ATP depletion, induced by very high concentrations of 2-DOG, has been reported to impair mTOR signaling (16), which in turn is known to modulate the phosphorylation state of eEF2 (reviewed in (7)). To assess whether this was the case in KB cells and at the low concentrations of 2-DOG used here, we studied the regulation of S6K1, a well-known target for mTOR signaling. This analysis makes use of the facts that activation of S6K1 involves its phosphorylation at multiple sites and that this causes a marked retardation of its mobility on SDS-PAGE. In untreated cells, S6K1 migrated as a ladder of bands that differ in their states of phosphorylation indicating that the enzyme is at least partially activated under this condition To study this further, we also tested the effects of AICAR, which following entry to the cell, is converted into a compound that specifically activates AMPK (30,31). In contrast to the situation for 2-DOG, upon treatment of cells with AICAR, there was no shift in the pattern of bands observed for S6K1 (Fig. 5A) indicating that activation of AMPK per se does not interfere with mTOR signaling. Nonetheless, treatment of KB cells with AICAR increased the level of phosphorylation of eEF2 at Thr56, as reported earlier by us for primary rat hepatocytes, Chinese hamster ovary (CHO)-K1 cells and cardiac myocytes (11,12) (Fig. 5B).
Interestingly, AICAR, 2-DOG and rapamycin raised the phosphorylation of eEF2 to similar extents ( Fig. 5A; middle section), suggesting that effects other than inhibition of mTOR signaling may underlie the increase in eEF2 phosphorylation elicited by 2-DOG and in particular by AICAR. All these treatments increased the phosphorylation of ACC, with 2-DOG having the most marked effect (Fig. 5A: bottom section).
However there was a robust increase in ACC phosphorylation with 2-DOG and a slight activation with AICAR. Given that all three treatments, 2-DOG, AICAR and rapamycin, have similar effects on the phosphorylation of eEF2 (Fig. 5A: middle section), while the first two have little or no effect on mTOR signaling, it is clear that 2-DOG, and in particular AICAR, increase eEF2 phosphorylation through events that are independent of the mTOR pathway.
A further, and potent, way of activating AMPK, is to treat cells with the mitochondrial uncoupler, CCCP. This induces a marked increase in the phosphorylation of eEF2 (Fig. 5B).
CCCP treatment also induced increased the phosphorylation of eEF2 kinase to a similar extent to that observed with AICAR (Fig. 5C). Treatment of KB cells with AICAR or 2-DOG increased the phosphorylation of Ser398 in eEF2 kinase (Fig. 5D). However, rapamycin did not induce an increase in the phosphorylation of this site (data not shown), indicating that it is not a consequence of inhibition of mTOR signaling.
The data obtained using AICAR provide several important pieces of information. Firstly, they give strong evidence that Ser398 in eEF2 kinase is a target for AMPK within living cells.
Secondly, they show that the increase in phosphorylation of eEF2 kinase at Ser398 is unlikely to be a secondary consequence of inhibition of mTOR signaling. Thirdly, they also show that the increase in eEF2 phosphorylation cannot be attributed to impairment of mTOR signaling (which can under conditions have this effect), and they thus imply that it is likely to be a direct consequence of activation of AMPK and perhaps of the phosphorylation of eEF2 kinase at Ser398. Taken together, our data are consistent with the conclusion that phosphorylation of Ser398 is a consequence of the activation of AMPK, probably because this site is a direct substrate for this enzyme.
Several other phosphorylation sites have been identified within eEF2 kinase. These include Ser366 and Ser78, which are also targets for AMPK in vitro (see above). It was of interest to ascertain whether agents that affect eEF2 kinase activity and activate AMPK, affected their phosphorylation. As shown in Fig. 4B, the phosphorylation of eEF2 kinase at Ser78 or Ser366 decreased following treatment of KB cells with 2-DOG. Although treatment of cells with AICAR, 2-DOG or CCCP caused a marked increase in the phosphorylation of eEF2 itself, none of these treatments affected the phosphorylation of Ser366 in eEF2 kinase, as revealed by western blot analysis of immunoprecipitates from KB cells pretreated with these agents (Fig. 5E).

Conditions that stimulate AMPK lead to increased activity of eEF2 kinase
The observation that ATP depletion or AICAR leads to increased phosphorylation of eEF2 suggested that such conditions were also likely to activate eEF2 kinase. The activity of eEF2 kinase is normally completely dependent on Ca 2+ and calmodulin (28,32,33). There are therefore, in principle, several ways in which such activation might come about. It could involve, for example, an increase in the maximal activity of eEF2 kinase (i.e., at saturating Ca 2+ /CaM) or an increase in the affinity of eEF2 kinase for CaM.
To study the possible effects of AICAR or 2-DOG treatment on the activity of eEF2 kinase, the enzyme was immunoprecipitated from lysates of KB cells or ARVC that had been subjected to various treatments and harvested in the presence of Ca 2+ to ensure that any tightly bound CaM remained associated with the eEF2 kinase. Kinase assays were then performed in the presence of a low concentration of Ca 2+ (1 µM) and in the presence or absence of added CaM.
Under the latter conditions, eEF2 kinase activity is dependent upon the CaM that remains bound to eEF2 kinase during the IP. To verify the specificity and efficiency of the immunoprecipitation, we also performed immunoprecipitations using pre-immune IgG as well as anti-eEF2 kinase, and measured eEF2 kinase activity in the pellets, and immunoblotted the supernatant for residual eEF2 kinase. As shown in Fig. 6A, the anti-eEF2 kinase antiserum completely immunoprecipitated the eEF2 kinase from the cell lysate, as judged either by immunoblot or kinase assay. In the case of the pre-immune IgG, both kinase protein and activity remained in the supernatant. Thus, no kinase activity against eEF2 is immunoprecipitated non-specifically. To further verify that the activity against eEF2 shows the characteristics expected for eEF2 kinase, we checked its dependence on Ca 2+ and CaM. In anti-eEF2 kinase immunoprecipitates from ARVC, kinase activity against eEF2 was completely dependent upon added Ca 2+ /CaM ( Fig. 6B; left side). In lysates from KB cells, a trace of activity was still seen in the absence of added Ca 2+ /CaM but this was eliminated by addition of the CaM antagonist trifluoperazine (TFP; Fig.   6B, right side) suggesting the trace basal activity was due to endogenous CaM or contaminating Ca 2+ . Thus, the kinase activity against eEF2 behaves exactly as expected for the known eEF2 kinase indicating that there is unlikely to be any contribution from other (unknown) enzymes.
In assays in which additional CaM was added, no effect of pretreatment of the KB cells with AICAR or 2-DOG on eEF2 kinase activity was evident (Fig. 6C). In contrast, rapamycin did elicit a marked increase in eEF2 kinase activity, when measured in this way, consistent with our earlier conclusion that AMPK activation and inhibition of mTOR signaling affect eEF2 kinase activity in distinct ways. However, when the assays of eEF2 kinase from KB cells were performed without added CaM, it was clear that pretreatment of KB cells with AICAR or 2-DOG caused a significant and reproducible increase in eEF2 kinase activity (Fig. 6D). Data from multiple experiments are analysed in Fig. 6E. It is clear that treatment of KB cells with any of several treatments that activate AMPK -AICAR, 2-DOG or CCCP -activates eEF2 kinase. This likely explains their common ability to increase the phosphorylation of eEF2. It is also evident from these data that rapamycin also elicits a marked activation of eEF2 kinase as measured is this way. The observation that the effects of agents that activate AMPK are only seen when measured without addition of CaM to the assay could be interpreted to indicate that such agents, and perhaps the phosphorylation of eEF2 kinase at Ser398, increase the binding of eEF2 kinase to CaM, so that more of it co-immunoprecipitates with eEF2 kinase and is therefore able to activate eEF2 kinase in the assay.
In the case of ARVC, a different picture emerged, in that an increase in eEF2 kinase activity was seen even under conditions where CaM was added to the assay (Fig. 6F), while in the absence of CaM a similar pattern was observed although the overall activity was, as expected, lower (data not shown). This could suggest that agents that activate AMPK lead to increased maximal activity of eEF2 kinase in cardiomyocytes. This accords with our earlier conclusions (12) that cardiomyocytes may contain a different isoform of eEF2 kinase and that its regulation, in particular the modulation of its control by Ca 2+ -ions, differs from that of the well-characterised isoform found in many other cell types.

Discussion
The present data are of considerable potential importance for understanding the links between cellular energy metabolism and the regulation of protein synthesis. Given that protein synthesis uses a high proportion of cellular energy, it is perhaps surprising that so little work has previously been devoted to understanding how cells might co-ordinate the rate of protein synthesis with the availability of metabolic energy. Our data thus provide new insights into the poorly understood links between energy status and the rate of protein synthesis. It has been previously noted that low oxygen per se, i.e., in the absence of ATP depletion, does not affect protein synthesis rates, while depletion of ATP does lead to a fall in the rate of translation (34).
This fits well with our findings which suggest that AMPK is a key link between cellular status and the regulation of protein synthesis: AMPK would be activated under the latter condition, but not under the former.
In this study, we identify a new target of AMPK, i.e., eEF2 kinase. We show that AMPK can phosphorylate three sites in eEF2 kinase in vitro. Of these, Ser398 appears to be more efficiently phosphorylated than either Ser78 or Ser366. Using an antibody that we have developed for eEF2 kinase phosphorylated at Ser398, we show that its phosphorylation is enhanced in response to a range of treatments designed to activate AMPK. These include treatment of cells with 2-DOG, with mitochondrial poisons such as CCCP or with the specific AMPK activator, AICAR. Phosphorylation of Ser78 or Ser366 does not increase under such decreases, and in some cases even falls (e.g., Ser78 in cells treated with 2-DOG). Thus, it is likely that Ser398 is a physiological target for AMPK, while Ser78 and Ser366 are almost certainly not.
Our data show that Ser398 is the main site in eEF2 kinase whose phosphorylation is increased under conditions of ATP depletion and the major site in eEF2 kinase that is phosphorylated by AMPK in vitro. Phosphorylation at Ser398 in eEF2 kinase is increased by a range of conditions that activate AMPK and correlates well with phosphorylation of the substrate, eEF2 itself. These conditions all result in activation of eEF2 kinase as measured against eEF2 in vitro suggesting that phosphorylation of eEF2 kinase at Ser398 leads to an increase in its activity. The fact that these effects are smaller than those of rapamycin likely reflects the facts that rapamycin also affects the phosphorylation of other sites in eEF2 kinase, e.g., Ser359 and Ser366 (10,17) which results in additional activation of eEF2 kinase. Rapamycin can decrease the phosphorylation of both sites, and since phosphorylation of either exerts an inhibitory effect on eEF2 kinase activity, this likely contributes to the relatively larger increase in eEF2 kinase activity seen following rapamycin treatment.
As phosphorylation of Ser366 serves to decrease the activity of eEF2 kinase (10), it would make little sense for its phosphorylation to be enhanced under conditions of ATP depletion where eEF2 phosphorylation actually increases. Very recent work in this laboratory suggests that phosphorylation at Ser78 may also decrease the activity of eEF2 kinase and a similar argument would thus also apply to this site (GJB, unpublished findings).
2-DOG treatment also increased the phosphorylation of eEF2 kinase at Ser377, and, as expected, this was mediated through the SB203580-sensitive p38 MAP kinase pathway (18).
Since phosphorylation here is not thought to affect the activity of eEF2 kinase, this modification is unlikely to contribute to its activation following 2-DOG treatment.
Recent work from Rider's group shows that in anoxic cardiomyocytes AMPK activation was associated with an increase in eEF2 phosphorylation and an inhibition of protein synthesis (35). These authors identified Ser78 in eEF2 kinase as a potential site for phosphorylation by AMPK in vitro. Our data corroborate this finding but reveal that, although Ser78 is phosphorylated in vivo, its phosphorylation does not change in a manner consistent with it being a site for phosphorylation by AMPK in living cells.
In principle, one could test the effect of phosphorylation of eEF2 kinase at Ser398 by incubating purified eEF2 kinase with AMPK in vitro and then measuring its activity or ability to bind CaM. However, this possible approach is made complicated by the fact that AMPK also phosphorylates at least two other sites in eEF2 kinase, Ser78/366, and by the fact that phosphorylation of at least, possibly both, of these sites also affects eEF2 kinase activity ( (10) and unpublished data).
29 translation elongation rather than initiation. This does seem logical, as far more energy is consumed during elongation that during initiation (for the assembly of a typical protein, >99% of the energy is used in elongation). Furthermore, inhibition of elongation will retard the rate of progress of ribosomes along the mRNA, thus saving energy, but without causing polysomes to dissociate (in fact, there would likely accumulate). This would therefore allow translation to resume fully and quickly once cellular energy levels had recovered and eEF2 has been dephosphorylated, without any need to reassemble polysomes. This regulatory mechanism may be of considerable importance in matching the rate of translation elongation to fluctuations in cellular energy status.