The phosphorylation of eukaryotic initiation factor eIF4E in response to phorbol esters, cell stresses, and cytokines is mediated by distinct MAP kinase pathways.

Initiation factor eIF4E binds to the 5'-cap of eukaryotic mRNAs and plays a key role in the mechanism and regulation of translation. It may be regulated through its own phosphorylation and through inhibitory binding proteins (4E-BPs), which modulate its availability for initiation complex assembly. eIF4E phosphorylation is enhanced by phorbol esters. We show, using specific inhibitors, that this involves both the p38 mitogen-activated protein (MAP) kinase and Erk signaling pathways. Cell stresses such as arsenite and anisomycin and the cytokines tumor necrosis factor-alpha and interleukin-1beta also cause increased phosphorylation of eIF4E, which is abolished by the specific p38 MAP kinase inhibitor, SB203580. These changes in eIF4E phosphorylation parallel the activity of the eIF4E kinase, Mnk1. However other stresses such as heat shock, sorbitol, and H2O2, which also stimulate p38 MAP kinase and increase Mnk1 activity, do not increase phosphorylation of eIF4E. The latter stresses increase the binding of eIF4E to 4E-BP1, and we show that this blocks the phosphorylation of eIF4E by Mnk1 in vitro, which may explain the absence of an increase in eIF4E phosphorylation under these conditions.

tion but can be unwound by eIF4A (1,2). Binding of eIF4E to eIF4G can be blocked by regulator proteins termed eIF4Ebinding proteins (4E-BPs), 1 which interact with the same region of eIF4E that binds eIF4G (4). The best characterized of them is 4E-BP1 (also called PHAS-I, (4)). It is regulated by phosphorylation at multiple sites, in response, e.g. to insulin, which causes its dissociation from eIF4E (4). eIF4E is a phosphoprotein, and its phosphorylation is generally enhanced by agents that activate translation (reviewed in Refs. 1 and 2). Phosphorylation of eIF4E increases its affinity for the cap and for mRNA and may also favor its entry into initiation complexes (1,2). Both effects may be important in the activation of translation under conditions that increase eIF4E phosphorylation. The phosphorylation site in eIF4E is Ser 209 (5,6), although the identity of the protein kinase responsible for its phosphorylation in vivo is less clear. We have recently shown that insulin-induced eIF4E phosphorylation requires the MAP kinase signaling pathway (also termed the Erk, extracellular signal-regulated kinase pathway, the term used here) (7). However, eIF4E is not a substrate for the Erks, and we have shown that it is phosphorylated instead at Ser 209 by a novel Erk-activated protein kinase, MAP kinase signal-integrating kinase-1 (Mnk1) (8,9). Mnk1 is also phosphorylated and activated by an additional enzyme related to Erk, p38 MAP kinase, which lies on a distinct signaling pathway activated by cell stresses and cytokines (8 -10).
Here we show that the increased phosphorylation of eIF4E brought about by the phorbol ester tetradecanoylphorbol 13acetate (TPA), which activates members of the protein kinase C (PKC) family, requires the Erk and p38 MAP kinase pathways. Furthermore, we show that eIF4E phosphorylation is enhanced by agents, such as certain stresses and cytokines, that activate p38 MAP kinase and that this is blocked by a specific inhibitor of this enzyme. Our data support the identity of Mnk1 as a physiologically important eIF4E kinase. Certain stresses that activate p38 MAP kinase do not increase eIF4E phosphorylation. This is likely to be due to the increased association of eIF4E with 4E-BP1 that these conditions bring about, because 4E-BP1 inhibits phosphorylation of eIF4E by Mnk1.

MATERIALS AND METHODS
Chemicals and Biochemicals-Unless otherwise stated, chemicals were obtained as described previously (7,11). Anti-(P)Erk was from New England BioLabs. Anti-human eIF4E was raised against a syn-thetic peptide corresponding to residues 5-23 of the protein. Antibodies to eIF4G were a kind gift from S. J. Morley (Sussex). The anti-4E-BP1 antibody was raised against a peptide corresponding to residues 101-113 of human 4E-BP1. Recombinant Mnk1-GST and 4E-BP1 were prepared as described previously (8,12). The expression vector for 4E-BP1 was a generous gift from R. M. Denton (Bristol).
Cell Culture, Treatment, and Extraction-Human embryonic kidney 293 and Chinese hamster ovary (CHO.K1) cells were grown as described previously (8,13). Human umbilical vein endothelial cells (HUVECs) were grown in modified MCDB131 medium (Clonetics), supplemented with 2% (v/v) fetal calf serum, 10 ng/ml epidermal growth factor, 1 g/ml hydrocortisone, 50 g/ml gentamicin, 50 ng/ml amphotericin B, and 12 g/ml bovine brain extract. Where used, inhibitors were added 1 h prior to treatment of the cells. Cells were extracted in our standard buffer, which contains a mixture of protein phosphatase and proteinase inhibitors (13). The transfection protocol for 293 cells was described previously (8).
Assessment of Protein Kinase Activities-Mnk1 was assayed using eIF4E as substrate as described by Waskiewicz et al. (8). p38 MAP kinase activity was assessed by measuring the activity of the downstream kinase MAP kinase-activated kinase-2 (MAPKAP-K2 (10)) using recombinant hsp25 as substrate (14).
Isolation and Analysis of eIF4E-eIF4E was isolated from cell extracts by affinity chromatography on m 7 GTP-Sepharose as described previously (11). Its phosphorylation state was assessed by isoelectric focusing/immunoblotting as described previously (11). Its association with 4E-BP1 and with components of eIF4F was analyzed by Western blotting using antisera to eIF4G and 4E-BP1 (15,16).

RESULTS AND DISCUSSION
TPA-induced Phosphorylation of eIF4E Requires the Erk and p38 MAP Kinase Pathways-In several cell types, phorbol esters that activate PKC enhance eIF4E phosphorylation (11,(17)(18)(19). Exposure of 293 cells to TPA increased the level of eIF4E phosphorylation from almost zero to about 30% (Fig.  1A). To test whether, as for insulin (7), this effect required the Erk pathway, we used the compound PD098059, a specific inhibitor of MEK activation (20). As expected, PD098059 reduced the ability of TPA to enhance eIF4E phosphorylation (Fig. 1A) and also completely inhibited the activation of Erk by TPA in 293 cells (Fig. 1B, lanes 1-4). However, the inhibition of eIF4E phosphorylation was incomplete indicating that other signaling pathways were involved. We therefore tested the effect of a specific inhibitor of p38 MAP kinase SB203580 (21) on eIF4E phosphorylation. It has been shown not to interfere with stress-, cytokine-, or growth factor-induced activation of other signaling pathways such as Erk, JNK, or p70 S6 kinase (21)(22)(23)(24). SB203580 partially blocked TPA-induced eIF4E phosphorylation, and its use together with PD098059 completely abolished TPA-induced eIF4E phosphorylation (Fig. 1A). As shown in Fig. 1B (lanes 5-8), TPA activates MAPKAP-K2, which is activated by p38 MAP kinase (10), in 293 cells. These data suggest that TPA acts through both Erk and p38 MAP kinase to increase eIF4E phosphorylation in 293 cells. Waskiewicz et al. (8) previously showed that Mnk1 could be activated in vitro by either Erk or p38 MAP kinase. To assess Mnk1 activity in 293 cells subjected to these treatments, we transfected 293 cells with a vector encoding wild-type Mnk1 fused to GST (8) and subjected the transfected cells to treatment with TPA in the absence or the presence of kinase inhibitors. Cells were extracted, and Mnk1-GST was isolated and assayed. SB203580 or PD098059 only partially prevented the activation of Mnk1 by TPA, but use of both completely abolished it (Fig. 1C). The changes in Mnk1 activity parallel the alterations in eIF4E phosphorylation observed under these conditions, entirely consistent with a key role for Mnk1 in mediating eIF4E phosphorylation (8).
TPA also increases eIF4E phosphorylation in CHO.K1 cells, which involves the so-called conventional isoforms of PKC (11). Analysis of the roles of signaling pathways in the phosphorylation of eIF4E in these cells is more complex than for 293 cells due to the high basal level of eIF4E phosphorylation (Fig. 1D). This was partially decreased either by PD098059 or SB203580, although both were required to completely suppress it (Fig.  1D). Each compound alone also decreased the level of eIF4E phosphorylation in TPA-treated CHO.K1 cells, but both were again required to abolish eIF4E phosphorylation completely (Fig. 1D). Taken together, the data reinforce the conclusion that both the Erk and p38 MAP kinase cascades are involved in mediating changes in eIF4E phosphorylation. Our data imply that the phorbol ester-induced phosphorylation of eIF4E is not, as has previously been suggested (1), directly mediated by PKC. A more likely route by which TPA increases eIF4E phosphorylation is through activation of the Erk/p38 MAP kinase cascades, via PKC, leading to the activation of Mnk1, which itself directly phosphorylates eIF4E (8). This is consistent with all the published data on phorbol ester-induced eIF4E phosphorylation (11,(17)(18)(19).
Arsenite Induces the Phosphorylation of eIF4E, Which Is Blocked by SB203580 -The above data prompted us to ask whether other treatments that activate p38 MAP kinase affect eIF4E phosphorylation. Arsenite potently activates the p38 MAP kinase pathway in 293 cells ( Fig. 2A) and also markedly increased eIF4E phosphorylation (Fig. 2B, lanes 1-5). The p38 MAP kinase inhibitor SB203580 (21) blocked both this and the arsenite-induced activation of MAPKAP-K2 (Fig. 2, A and B). Arsenite did not activate Erk in 293 cells (data not shown), and the MEK inhibitor PD098059 did not affect arsenite-induced eIF4E phosphorylation (Fig. 2B). Thus, arsenite-induced eIF4E phosphorylation appears to be mediated by the p38 MAP ki-nase pathway (8). Changes in Mnk1 activity again parallel those in eIF4E phosphorylation (Fig. 2C). The ability of arsenite to increase eIF4E phosphorylation is not restricted to 293 cells because the same effect was also observed in CHO.K1 cells (Fig. 2B, lanes 6 -9), and again, SB203580 blocked both arsenite-induced eIF4E phosphorylation and p38 MAP kinase activation ( Fig. 2B and data not shown). Anisomycin activates Mnk1 in 293 cells, and this activation was blocked by SB203580 ( Ref. 8 and Fig. 2C). Anisomycin increased eIF4E phosphorylation in 293 cells (to about 80%, data not shown), and this effect was also completely blocked by SB203580. Anisomycin also increases eIF4E phosphorylation in NIH 3T3 cells, and this increase is blocked by SB203580 (25).
The finding that arsenite stimulates eIF4E phosphorylation is surprising given that arsenite potently inhibits protein synthesis (26), whereas eIF4E phosphorylation is normally associated with its activation. It is likely that arsenite inhibits other steps in translation, and, indeed, we have shown that it increases phosphorylation of the ␣-subunit of eIF2, which is well known to lead to inhibition of peptide chain initiation (27).
Why do some stresses increase eIF4E phosphorylation, whereas others cause a decrease, even though they also activate Mnk1? To try to explain this apparent paradox, we analyzed the association of eIF4E with its regulator 4E-BP1; we have previously shown that in CHO.K1 cells, most stresses increase binding of 4E-BP1 to eIF4E (16). This effect is also seen in 293 cells (Fig. 3A). (The exception here (as in CHO cells) is arsenite, which does not cause increased binding of 4E-BP1 to eIF4E. This is probably because it can activate the rapamycin-sensitive signaling pathway (28), which leads to the phosphorylation of 4E-BP1 and its dissociation from eIF4E (4).) These findings raised the possibility that the association of 4E-BP1 with eIF4E might impair phosphorylation of the latter by Mnk1. To test this, we examined the effect of 4E-BP1 on the ability of Mnk1 to phosphorylate eIF4E in vitro. The data (Fig.  3B) clearly show that 4E-BP1 substantially inhibits the phosphorylation of eIF4E by Mnk1. The highest amount of 4E-BP1 used represents saturation of the eIF4E with 4E-BP1 as indicated by the fact that addition of further 4E-BP1 resulted in (i) it not being retained on m 7 GTP-Sepharose, i.e. not being associated with eIF4E, and (ii) phosphorylation of the excess 4E-BP1 by the Erk present in the activated Mnk1, with only free 4E-BP1 (and not the 4E-BP1/eIF4E complex) being a substrate for Erk (12) (data not shown). SB203580 had no effect on the association of eIF4E with 4E-BP1, either under stress or control conditions (16). 2 4E-BP1 did not affect the phosphorylation of another substrate, the cAMP-response element binding protein, by Mnk1 (data not shown). This suggests that inhibition of eIF4E phosphorylation by 4E-BP1 reflects the inability of Mnk1 to phosphorylate eIF4E in the eIF4E/4E-BP1 complex , determined by PhosphorImager analysis of the dried gel. B, assessment of eIF4E phosphorylation. Lanes 1-5, 293 cells. Cells were pretreated with or without SB203580 and/or PD098059 for 60 min and then exposed, where indicated, to sodium arsenite (0.1 mM) for 20 min, as indicated. Lanes 6 -9, CHO.K1 cells. Cells were treated as for lanes 1-5 (except that PD098059 was not used here). Data are presented as Fig. 1A. C, activity of Mnk1-GST in 293 cells. Stimuli and inhibitors were used under the conditions described above. The data were obtained and are presented as in Fig. 1C. unstim, unstimulated; SB, SB203580; HS, heat shock; SORB, sorbitol; ARS, arsenite; ANISO, anisomycin. D, assessment of the level of phosphorylation of eIF4E. Lanes 1-5, 293 cells; lanes 6 -10, CHO.K1 cells. All stress conditions are as same as those in A: con, control; ars, arsenite; HO, hydrogen peroxide; HS, heat shock; sor, sorbitol. and is likely to be specific for this substrate.
The inhibition of Mnk1-catalyzed eIF4E phosphorylation by 4E-BP1 provides an explanation for the differing effects of stresses on eIF4E phosphorylation, and, in particular, for the ability of heat shock to reduce eIF4E phosphorylation (reviewed in Ref. 29). Some studies have shown that rapamycin reduces the level of eIF4E phosphorylation (25,30). Our data suggest that this may be due to increased association of eIF4E with 4E-BP1 caused by rapamycin (due to dephosphorylation of 4E-BP1 (4)) and resulting inhibition of Mnk1-catalyzed eIF4E phosphorylation. 4E-BP1 also blocks the phosphorylation of eIF4E by PKC (31). The binding site for 4E-BP1 in eIF4E has recently been identified (32). Because it is some distance from Ser 209 in the three-dimensional structure of the protein, it seems unlikely that 4E-BP1 actually occludes the phosphorylation site. The effect of 4E-BP1 may instead reflect interference with the interaction between these kinases and other regions of the eIF4E protein required for kinase/substrate binding.
The increased binding of 4E-BP1 to eIF4E was accompanied by a decrease in the binding of eIF4G to eIF4E (Fig. 3C), as expected from the mutually competitive nature of their interactions (33). Consistent with its lack of effect on 4E-BP1 binding, arsenite also had no effect on the association of eIF4E with eIF4G.
Regulation of eIF4E Phosphorylation by Cytokines That Activate p38 MAP Kinase-It was important to ascertain whether treatment of cells with physiological activators of p38 MAP kinase, such as cytokines, also altered the phosphorylation of eIF4E. Tumor necrosis factor-␣ (TNF␣) is a physiological regulator of endothelial cell function (34), and in HUVECs it markedly activates the p38 MAP kinase pathway without any apparent effect on Erk activity (data not shown and Ref. 34). We therefore studied its effect on the phosphorylation of eIF4E. TNF␣ increased the phosphorylation of eIF4E (Fig. 4A), and this increase was prevented by SB203580, which blocked activation of the p38 MAP kinase pathway and hence of MAPKAP-K2. These data show for the first time that cytokines increase phosphorylation of eIF4E through the p38 MAP kinase pathway. In CHO.K1 cells, another cytokine, interleukin-1␤ (IL-1␤) activates p38 MAP kinase, although less markedly than TNF␣ does in HUVECs (data not shown). IL-1␤ (5 ng/ml) increased the phosphorylation of eIF4E in CHO.K1 cells (Fig. 4B). This effect, like that of TNF␣ in HUVECs, was prevented by SB203580 (data not shown).
Conclusions-The ability of activators of p38 MAP kinase to increase eIF4E phosphorylation was seen in three different cell types, human embryonic kidney (293) cells, CHO cells, and HUVECs and in response to stresses and cytokines. In all cases, SB203580 blocked the phosphorylation of eIF4E. The data for TNF␣ in HUVECs are of particular note given that the TNF␣-stimulated induction of the cell adhesion molecule V-CAM is mediated by the p38 MAP kinase pathway (34) and involves post-transcriptional effects that might be related to changes in eIF4E phosphorylation. Both the stress stimuli (arsenite and anisomycin) and TNF␣ also activate the JNK pathway. However, the ability of SB203580 (which does not affect JNK activity in the cells used here) 2 to block eIF4E phosphorylation indicates that the JNK pathway is not involved in modulating eIF4E phosphorylation.
We have previously shown that insulin-induced phosphorylation of eIF4E requires the Erk pathway (7). Taken together, our findings show that eIF4E phosphorylation can be mediated by two distinct signaling pathways, the Erk and p38 MAP kinase pathways, depending on the stimulus, consistent with the established regulatory properties of the eIF4E kinase Mnk1, which is a target for activation by both (8,9). Changes in eIF4E phosphorylation largely mirror alterations in Mnk1 activity, consistent with a physiological role for this kinase in eIF4E phosphorylation. The only exceptions are stress conditions that increase binding of 4E-BP1 to eIF4E. In almost all cases, such conditions activate Mnk1 but decrease eIF4E phosphorylation. Cells were subjected to stress conditions as in Fig. 2A and then extracted and analyzed for 4E-BP1 and eIF 4E. The figure shows a Western blot developed using ECL. con, control; ars, arsenite; HO, hydrogen peroxide; sor, sorbitol; HS, heat shock. B, effect of 4E-BP1 on the phosphorylation of eIF4E by Mnk1. Lanes 1-4, eIF4E was incubated with Mnk1-GST that had previously been activated by incubation with Erk1 and ATP/Mg, followed by washing of the Mnk1-GST bound to glutathione-Sepharose with LiCl (0.5 M) to remove as much Erk1 as possible (8). Reactions for the phosphorylation of eIF4E by Mnk1-GST were performed in the absence (lane 1) or the presence (lanes 2-4) of 4E-BP1 and radiolabeled ATP. Lanes 2-4 contain increasing amounts of 4E-BP1 (in the ratio 1:3:7). Numbers below each lane indicate the relative labeling of eIF4E as determined by PhosphorImager analysis. C, dissociation of eIF4F complexes in stressed cells (see Fig. 2A), assessed by analyzing samples isolated as described under "Materials and Methods" (m 7 GTP-Sepharose-bound material) on an 8% polyacrylamide gel followed by blotting with anti-eIF4G.

FIG. 4. Cytokines increase eIF4E phosphorylation.
A, HUVECs were pretreated with or without SB203580 (25 M) for 60 min before exposure to TNF␣ (5 ng/ml) for 10 min, as indicated, and extracts were analyzed for eIF4E phosphorylation. Data are presented as in Fig. 1A. B, CHO.K1 cells were treated for 20 min with or without IL-1␤ (indicated concentration). Cells were then extracted, and extracts were analyzed for eIF4E phosphorylation. Data are presented as Fig. 1A.