Research ArticlePhysiology

MNK2 Inhibits eIF4G Activation Through a Pathway Involving Serine-Arginine–Rich Protein Kinase in Skeletal Muscle

See allHide authors and affiliations

Science Signaling  14 Feb 2012:
Vol. 5, Issue 211, pp. ra14
DOI: 10.1126/scisignal.2002466


Skeletal muscle mass is regulated by activity, metabolism, and the availability of nutrients. During muscle atrophy, MNK2 expression increases. We found that MNK2 (mitogen-activated protein kinase–interacting kinase 2), but not MNK1, inhibited proteins involved in promoting protein synthesis, including eukaryotic translation initiation factor 4G (eIF4G) and mammalian target of rapamycin (mTOR). Phosphorylation at serine 1108 (Ser1108) of eIF4G, which is associated with enhanced protein translation, is promoted by insulin-like growth factor 1 and inhibited by rapamycin or starvation, suggesting that phosphorylation of this residue is regulated by mTOR. In cultured myotubes, small interfering RNA (siRNA) knockdown of MNK2 increased eIF4G Ser1108 phosphorylation and overcame rapamycin’s inhibitory effect on this phosphorylation event. Phosphorylation of Ser1108 in eIF4G, in gastrocnemius muscle, was increased in mice lacking MNK2, but not those lacking MNK1, and this increased phosphorylation was maintained in MNK2-null animals under atrophy conditions and upon starvation. Conversely, overexpression of MNK2 decreased eIF4G Ser1108 phosphorylation. An siRNA screen revealed that serine-arginine–rich protein kinases linked increased MNK2 activity to decreased eIF4G phosphorylation. In addition, we found that MNK2 interacted with mTOR and inhibited phosphorylation of the mTOR target, the ribosomal kinase p70S6K (70-kD ribosomal protein S6 kinase), through a mechanism independent of the kinase activity of MNK2. These data indicate that MNK2 plays a unique role, not shared by its closest paralog MNK1, in limiting protein translation through its negative effect on eIF4G Ser1108 phosphorylation and p70S6K activation.


Skeletal muscle mass is maintained by the delicate balance between protein synthesis and protein degradation. Under conditions that cause atrophy, the E3 ubiquitin ligase MuRF1 mediates proteasomal degradation of myosin heavy chain (1) and other components of the thick filament (2). A second muscle-specific E3, MAFbx (also known as Atrogin-1) inhibits protein translation by ubiquitinating the initiation factor eIF3F (eukaryotic translation initiation factor 3F) (3, 4). On the anabolic side, IGF1 (insulin-like growth factor 1) is critical for protein synthesis in skeletal muscle. IGF1 initiates signaling through the IGF1 receptor that promotes muscle hypertrophy by stimulating phosphoinositide 3-kinase (PI3K), leading to the activation of the protein kinase AKT. AKT inactivates the negative regulator of protein synthesis GSK-3β (glycogen synthase kinase 3β) and activates mTOR (mammalian target of rapamycin), a kinase that is the key positive regulator of protein synthesis (5). mTOR is found in two complexes: TORC1, which is inhibited by rapamycin and promotes protein synthesis, and TORC2, which is not inhibited by rapamycin and functions in cell metabolism and cytoskeletal organization. In addition to mTOR, mTORC1 includes the proteins Raptor, GβL, and Pras40 (6). Through TORC1, mTOR phosphorylates positive and negative regulators of protein synthesis, such as p70S6K (70-kD ribosomal protein S6 kinase) and the translational inhibitor protein 4E-BP (eIF4E-binding protein), which enhances the activity and availability of protein translation components, such as ribosomal protein S6 (RPS6) and eIF4E (712). This pathway functions in many cells and promotes skeletal muscle mass (13).

The assembling of the translation initiation complex is a regulated process (14), the first step of which is the formation of the eIF4F complex (including eIF4G, eIF4E, and eIF4A). At the center of the eIF4F complex is eIF4G, a modular protein and the docking platform for several translation initiation factors and regulatory proteins, including mitogen-activated protein kinase–interacting kinase 1 and 2 (MNK1/2) (1517). Formation of the eIF4F complex and the subsequent binding of 5′ RNA cap are dependent on the amount of free eIF4E and the extent of eIF4G phosphorylation. The availability of free eIF4E is in turn directly correlated to the amount of 4E-BP1 phosphorylation; when in complex with nonphosphorylated 4E-BP1, eIF4E is unable to bind eIF4G (18). The phosphorylation state of 4E-BP1 is tightly regulated by mTOR activity. When mTOR is inactive, such as during nutrient-poor conditions, 4E-BP1 is not phosphorylated and sequesters eIF4E. Conversely, under nutrient-rich conditions, mTOR phosphorylates 4E-BP1, causing its dissociation from eIF4E, and facilitates eIF4F formation (19).

eIF4G is a phosphoprotein with more than 30 identified serine or threonine phosphorylation sites (2024). Among them, phosphorylation at Ser1108, Ser1148, and Ser1192 is stimulated by serum and inhibited by rapamycin treatment in human embryonic kidney (HEK) 293 cells (25). In particular, Ser1108 [found in the sequence -SSL(p)SRER-(XX)] has been characterized as correlating with active translation (26). In addition to serum, anabolic agents such as insulin and IGF1 increase Ser1108 phosphorylation in a rapamycin-sensitive manner, indicating crosstalk between this phosphorylation event and the IGF1 to mTOR pathway (27, 28). Furthermore, enhanced protein synthesis in rat skeletal muscle and heart after feeding is associated with Ser1108 phosphorylation and assembly of the active eIF4E-eIF4G complex (29, 30). In contrast, Ser1108 phosphorylation is diminished under conditions that negatively affect protein synthesis, such as starvation and sepsis (31, 32). These results suggested that eIF4G phosphorylation at Ser1108 is a molecular marker associated with enhanced protein translation. However, the molecular mechanism, particularly the kinase and phosphatase regulating this phosphorylation event, has yet to be characterized. eIF4G interacts with both eIF4E and MNKs and facilitates the phosphorylation of eIF4E at Ser209 by MNKs (14, 33). MNKs bind to the C-terminal part of eIF4G, near Ser1108. Although one might predict that MNKs may directly mediate this phosphorylation event or promote it to have a positive effect on translation, we identified a pathway involving MNK2 and SRPK (serine-arginine–rich protein kinase) that reduced phosphorylation of eIF4G at Ser1108 in skeletal muscle under atrophy conditions, indicating that MNK2 is a negative regulator of translation.


Regulation of eIF4E and eIF4G phosphorylation by MNKs

Phosphorylation of eIF4G at Ser1108 is regulated by nutrient availability (29). We monitored the change in phosphorylation of this site in myotubes formed by the muscle cell line C2C12 over the course of an hour, after myotubes were switched into amino acid–deficient medium, Dulbecco’s phosphate-buffered saline (DPBS) (Fig. 1A). Replacing differentiation medium with DPBS resulted in a time-dependent depletion of Ser1108 phosphorylation, with near-complete dephosphorylation by 60 min. Phosphorylation of eIF4E at Ser209 appeared unchanged or possibly slightly increased (Fig. 1A).

Fig. 1

MNK2 negatively regulates eIF4G phosphorylation at Ser1108. (A) C2C12 cells were transfected with the respective siRNA at day 1 of differentiation. At day 3 of differentiation, when myotubes were fully formed, medium was changed into DPBS for the indicated times. Cell lysates were analyzed for eIF4E and eIF4G phosphorylation status by Western blotting. An antibody recognizing p-MNK (Thr197/202) was used for MNK2 detection. A parallel set of samples (three biological replicates) were analyzed for MNK1 and MNK2 expression at day 3 of differentiation by qPCR; averages with SD are shown in the graph. (B) C2C12 cells were transduced with the adenovirus expressing the respective MNK. The phosphorylation status of eIF4E and that of eIF4G were analyzed at day 3 of differentiation. Ad-MNK2-AA, MNK2 with T197A/T202A mutations. These experiments have been repeated three times for (A) and four times for (B). Quantification of (A) is shown in fig. S1.

Because MNKs bind to eIF4G, we studied the effect of knocking down MNK1 or MNK2 with small interfering RNA (siRNA) on the status of eIF4G phosphorylation (Fig. 1A). We observed a 62% knockdown of transcripts for MNK1 and an 80% knockdown of transcripts for MNK2. The changes in phospho-eIF4E Ser209 upon knockdown of either MNK1 or MNK2 were not statistically significant throughout the course of hard starvation (Fig. 1A and fig. S1). Surprisingly, phospho-eIF4G Ser1108 increased in abundance specifically upon MNK2 and not MNK1 knockdown. This effect of MNK2 knockdown was even more pronounced under the nutrient-deprived condition, where Ser1108 phosphorylation was barely detectable in the control siRNA or MNK1 knockdown cells, yet was readily detectable in the MNK2 knockdown cells. Although eIF4G phosphorylation in the MNK2 knockdown cells did decrease in response to nutrient deprivation, phosphorylation was maintained at 30 and 60 min after starvation (Fig. 1A and fig. S1). These data indicated that MNK2 negatively regulates eIF4G phosphorylation. However, we could not completely rule out that MNK1 played a similar role because of the less than optimal knockdown of MNK1.

To investigate the role of each MNK on the regulation of eIF4G phosphorylation, we overexpressed MNK1 and MNK2 separately in C2C12 cells (Fig. 1B). Overexpression of either MNK1 or MNK2 resulted in enhanced eIF4E phosphorylation, indicating that both kinases were active when overexpressed. Consistent with the knockdown data, overexpression of MNK2, but not MNK1, abrogated eIF4G Ser1108 phosphorylation. Overexpression of the catalytically inactive MNK2-AA (T197A/T202A), where the two threonine residues at the activation segment were changed to alanine, did not alter eIF4G Ser1108 phosphorylation. This result suggested that MNK2, but not MNK1, through its kinase activity negatively regulates eIF4G phosphorylation.

To determine whether the C2C12 tissue culture data could be recapitulated in vivo, we analyzed eIF4G phosphorylation in the gastrocnemius muscle of MNK1, MNK2, and MNK1/2 knockout mice and compared it to that in wild-type control animals (Fig. 2). Basal eIF4G phosphorylation at Ser1108 was higher in MNK2 and MNK1/2 knockout mice and lower in MNK1 knockout mice compared to that of wild-type mice (Fig. 2). Ser209 phosphorylation of eIF4E was nearly undetectable in the MNK1/2 knockout mice, consistent with previous reports indicating that these are the principal kinases phosphorylating this site (34). These in vivo genetics data confirmed the results from the C2C12 cells, indicating a negative role for MNK2, but not MNK1, on the regulation of eIF4G phosphorylation at Ser1108. This effect of MNK2 was opposite of what one would expect for a kinase, suggesting that MNK2 is acting indirectly, either as a negative regulator of a kinase required to phosphorylate Ser1108 on eIF4G or as a positive regulator of a phosphatase.

Fig. 2

Phosphorylation status of eIF4E and eIF4G in gastrocnemius from MNK knockout mice. Gastrocnemius extracts (25 μg of protein) prepared from wild-type (WT) (n = 4), MNK1 knockout (KO) (n = 3), MNK2 KO (n = 4), and MNK1/2 KO (n = 4) mice were analyzed by Western blotting for eIF4E and eIF4G phosphorylation and MNK1 and MNK2. An antibody recognizing p-MNK (Thr197/202) was used for MNK2 detection.

mTOR and eIF4G phosphorylation

IGF1, a growth factor that is produced and released into the bloodstream upon anabolic exercise and that promotes muscle hypertrophy, enhances eIF4G phosphorylation at Ser1108 (28). Inhibition of the IGF1 downstream target mTOR complex 1 (TORC1) with rapamycin blocks this phosphorylation (25, 28). To understand the relationship between MNK2 and IGF1 pathways on Ser1108 phosphorylation, we investigated the effect of MNK2 overexpression on IGF1-induced Ser1108 phosphorylation in C2C12 myotubes. Although IGF1 induced eIF4G phosphorylation, this induction was mostly blocked by MNK2 overexpression (Fig. 3A). This effect of MNK2 was not due to general inhibition of the IGF1 pathway because AKT phosphorylation induced by IGF1 was not affected. A catalytically inactive mutant of MNK2 where the active-site lysine has been changed to a methionine (MNK2-K/M; K113M) was substantially less effective (Fig. 3A and fig. S2).

Fig. 3

Effect of MNK2 on the changes in eIF4G phosphorylation by IGF1 and mTOR signaling. (A) C2C12 cells were transduced with the indicated MNK-expressing adenovirus at day 1 of differentiation. At day 3 of differentiation, when myotubes were fully formed, cells were starved for 4 hours with serum-free DMEM and then either untreated or treated with IGF1 (10 nM) for 1 hour. Cell lysates were analyzed for eIF4G, AKT, and p70S6K phosphorylation by Western blotting. M2, adeno-MNK2; K/M, adeno-MNK2 with ATP binding site K-to-M mutation. (B) C2C12 cells were transfected with the respective siRNA at day 1 of differentiation. At day 3 of differentiation, cells were either untreated or treated with rapamycin (50 nM) or AZD8055 (50 nM) for 2 hours. Cell lysates were analyzed for eIF4E, eIF4G, AKT, and p70S6K phosphorylation by Western blotting. (C) C2C12 cells were transduced with the indicated MNK-expressing adenovirus at day 1 of differentiation. At day 3 of differentiation, when myotubes were fully formed, cells were starved for 4 hours with serum-free DMEM and then either untreated or treated with IGF1 (10 nM) for 1 hour. Cell lysates were analyzed for eIF4G, eIF4E, AKT, RPS6, and p70S6K phosphorylation by Western blotting. M1, adeno-MNK1; M2, adeno-MNK2; AA, adeno-MNK2 with T(197/202)-to-A mutations. These experiments have been repeated four times for (A) and (B) and three times for (C). Quantification of (A) is shown in fig. S2.

In a complementary experiment, the effect of MNK knockdown on rapamycin-mediated inhibition of eIF4G phosphorylation was assessed (Fig. 3B). Rapamycin-mediated inhibition of Ser1108 phosphorylation was blocked by siMNK2 but not siMNK1, indicating a role for MNK2 downstream of mTOR and upstream of eIF4G. This effect of siMNK2 appeared specific for rapamycin’s inhibitory action on eIF4G phosphorylation because rapamycin-mediated inhibition of p70S6K phosphorylation at Thr389 was not restored by MNK2 knockdown (Fig. 3B). Furthermore, we tested the effect of MNK2 knockdown on eIF4G phosphorylation in the presence of the mTOR ATP (adenosine 5′-triphosphate) binding site inhibitor AZD8055, which inhibits both TORC1 and TORC2 signaling (35). Exposure of C2C12 myotubes to AZD8055, but not to rapamycin, inhibited AKT phosphorylation at Ser473, which is a TORC2-mediated event. AZD8055 also inhibited both eIF4G and p70S6K phosphorylation. However, like rapamycin, only eIF4G phosphorylation was recovered upon MNK2 knockdown.

An interaction between MNK2 and TORC1

MNK2 or MNK2-KM overexpression reduced basal p70S6K phosphorylation at Thr389 through a mechanism independent of the kinase activity of MNK2 (Fig. 3A and fig. S2). The reduction in p70S6K Thr389 phosphorylation was specific for MNK2 and was not observed in cells overexpressing MNK1 (Fig. 3C). In addition, p70S6K phosphorylation at Ser371, another rapamycin-sensitive phosphorylation site in p70S6K (36), was also reduced upon MNK2 overexpression. A similar effect was observed with another catalytically inactive mutant of MNK2, MNK2-AA. Conditions that reduced p70S6K phosphorylation also reduced the phosphorylation of RPS6 at Ser240/244, which is a target of p70S6K (37). The similarities between MNK2 overexpression and rapamycin on eIF4G Ser1108 phosphorylation and p70S6K activation prompted us to investigate the effect of MNK2 overexpression on protein synthesis in differentiated C2C12 cells. We found that both basal and IGF1-induced protein synthesis were decreased by MNK2 overexpression. Basal conditions showed a 19% reduction compared to control cells overexpressing green fluorescent protein (GFP); IGF1-induced conditions showed a 7.5% reduction (fig. S3). This basal decrease is similar in magnitude to the reported effect of rapamycin on protein synthesis (28).

Because MNK2 inhibited the phosphorylation of a TORC1 substrate, we investigated whether MNK2 interacted with mTOR (Fig. 4). We found that Flag-tagged MNK2, but not MNK1, coimmunoprecipitated not only mTOR, but also Raptor and GβL, suggesting that MNK2 interacts with TORC1. Rictor, which is a component of TORC2, was not coprecipitated. However, this interaction did not require the kinase activity of MNK2 because the MNK2-AA mutant coimmunoprecipitated with TORC1 components (Fig. 4A), consistent with the ability of this catalytically inactive mutant to inhibit p70S6K phosphorylation (Figs. 3C and 4B). To identify regions of MNK2 responsible for the interaction with TORC1, we performed domain swapping between MNK1 and MNK2, selecting divergent regions between the two paralogs. We found that replacing both N-terminal and C-terminal regions of MNK1 with those from MNK2 created a chimera (MNK1-NC2) that coimmunoprecipitated with TORC1, whereas replacing both the N-terminal and the C-terminal regions of MNK2 with those of MNK1 (MNK2-NC1) eliminated the interaction with TORC1 (Fig. 4A). The MNK1-NC2 chimera also gained some ability to inhibit eIF4G phosphorylation (Fig. 4B). Although the MNK2-NC1 chimera failed to exhibit the MNK2 function of interacting with TORC1, it exhibited the MNK2 function of inhibiting eIF4G phosphorylation. The ability of MNK2 to inhibit p70S6K phosphorylation appeared dependent on its ability to interact with TORC1 because MNK1-NC2, but not MNK2-NC1, inhibited p70S6K phosphorylation.

Fig. 4

Interaction of MNK2 with TORC1 contributes to its ability to inhibit p70S6K phosphorylation, but not eIF4G phosphorylation. (A) C2C12 cells were transduced with the indicated MNK-expressing adenovirus at day 1 of differentiation. Cell lysates were prepared at day 3 of differentiation. Immunoprecipitation was performed with anti-Flag antibody. (B) The same lysates were analyzed for eIF4G and p70S6K phosphorylation by Western blotting. MNK1, adeno-MNK1; MNK2, adeno-MNK2; AA, adeno-MNK2 with T(197/202)-to-A mutations; MNK1-NC2 and MNK2-NC1, chimeras between MNK1 and MNK2. These experiments have been repeated four times for (A) and three times for (B).

Of the several known proteins that constitute the mTORC1 complex, Pras40 is the only component that did not coimmunoprecipitate with MNK2. Because Pras40 is known to bind Raptor (38), we asked whether Pras40 and MNK2 compete for Raptor binding.

Pras40 coimmunoprecipitated with Raptor when the immunoprecipitation was performed with antibody recognizing Raptor; however, the amount of Pras40 that coimmunoprecipitated with Raptor was lower in MNK2-overexpressing cells, under either basal or serum-free conditions (fig. S4). This result indicated that MNK2 and Pras40 compete for binding to Raptor. In the presence of IGF1, a condition that causes the phosphorylation at Thr246 and the release of Pras40 from TORC1 (38), only minimal amounts of Pras40 were pulled down together with Raptor in all conditions (fig. S4). Under our assay conditions, we observed only a minimal increase in Pras40 Thr246 phosphorylation.

Identification of the putative kinases responsible for eIF4G phosphorylation at Ser1108

To identify the kinases responsible for Ser1108 phosphorylation, we performed a screen of eIF4G Ser1108 phosphorylation in C2C12 myoblasts with a panel of siRNAs directed against mouse kinases (see Materials and Methods). The validity of the screen was indicated by the detection of mTOR as a strong hit in this screen (table S1). Among the other kinases identified, SRPK1 was a high-scoring candidate. The SRPK family is composed of three members, SRPK1, 2, and 3. We therefore tested the effect of knocking down individual members, as well as all three members combined, on Ser1108 phosphorylation in myotubes (Fig. 5A). Knockdown of individual members of the SRPK family [efficiency of knockdown as assessed by quantitative transcript analysis by qPCR (quantitative real-time polymerase chain reaction): SRPK1, 77%; SRPK2, 34%; and SRPK3, 87%] did not cause a statistically significant change in eIF4G phosphorylation. However, a statistically significant decrease occurred when all three were simultaneously knocked down, suggesting that all three members could contribute to eIF4G phosphorylation (Fig. 5A). In addition, knocking down all three members of SRPK significantly diminished the positive effect of MNK2 knockdown on eIF4G phosphorylation (Fig. 5A). The triple-knockdown experiment did not completely abolish eIF4G phosphorylation at Ser1108, which could be the result of either residual SRPK activity or activity of other kinases contributing to its phosphorylation. The abundance of SRPK3 increases during differentiation of myoblasts into myotubes (39), which may have contributed to the inability to demonstrate a significant decrease in Ser1108 phosphorylation upon knocking down of individual SRPKs in myotubes. Because the primary screen was performed in myoblasts, this may explain why knocking down of SRPK1 alone was sufficient to show diminished Ser1108 phosphorylation in the screen.

Fig. 5

SRPK family kinases are putative kinases responsible for eIF4G phosphorylation at Ser1108. (A) C2C12 cells were transfected with the respective siRNA at day 1 of differentiation. At day 3 of differentiation, cell lysates were analyzed for eIF4G phosphorylation status by Western blotting. The bar graph shows the ratio of p-eIF4G to total eIF4G from three independent experiments. Error bars indicate the SD. P values were determined by unpaired t test. #P < 0.05 versus siAllStar; ##P < 0.01 versus siMNK2. (B) C2C12 cells were transduced with adenovirus carrying the indicated expression construct at day 1 of differentiation. At day 3 of differentiation, cell lysates were prepared after cells were starved for 4 hours with serum-free DMEM and analyzed by Western blotting with the indicated antibodies. (C) C2C12 cells were transduced with adenovirus carrying the indicated expression construct at day 1 of differentiation. At day 3 of differentiation, rapamycin was added at 40 nM for 1 hour, and lysates were analyzed by Western blot with the indicated antibodies. Experiments in (B) and (C) have been repeated three times. Quantification of (C) is shown in fig. S6.

To further confirm the involvement of SRPKs and their relationship with MNK2, we overexpressed SRPK1 in the absence or presence of MNK2 overexpression in C2C12 myotubes. SRPK1 overexpression enhanced basal eIF4G phosphorylation at Ser1108. However, this effect was completely abolished by overexpressing the wild-type MNK2, but not the catalytically inactive MNK2-AA, suggesting that SRPK1 is downstream of MNK2 (Fig. 5B). When tagged forms of either MNK2 or SRPK1 were overexpressed in C2C12 cells, eIF4G coimmunoprecipitated with either MNK2 or SRPK1 (fig. S5), indicating that eIF4G interacts directly or indirectly with both MNK2 and SRPK1.

Consistent with a role in regulation of protein synthesis, we found that rapamycin affected the ability of overexpressed SRPK1 to promote eIF4G phosphorylation (Fig. 5C and fig. S6). SRPK1 overexpression increased eIF4G phosphorylation in rapamycin-treated cells compared to that in control cells treated with rapamycin, but this phosphorylation was still less than that observed in cells not exposed to rapamycin. Thus, SRPK1 may be regulated downstream of mTOR. In summary, these data implicate SRPKs in regulation of protein synthesis and suggest that MNK2 may target SRPKs directly or indirectly to reduce eIF4G phosphorylation at Ser1108 and that SRPKs are required for the increase in eIF4G phosphorylation observed upon MNK2 knockdown.

MNK expression and eIF4G phosphorylation in atrophy and starvation conditions

To investigate the role of MNK2 under conditions that cause atrophy or nutrient depletion in vivo, we compared muscles of wild-type and MNK knockout mice from a dexamethasone-induced atrophy model, a denervation-induced atrophy model, and in response to starvation. Dexamethasone treatment of mice induced the expression of MNK2 and reduced the expression of MNK1 in gastrocnemius muscle (Fig. 6A). Furthermore, we observed substantial decreases in eIF4G Ser1108 phosphorylation in the gastrocnemius of dexamethasone-treated wild-type mice even though total eIF4G did not change (Fig. 6, B and C). In muscles from the MNK2 knockout mice, eIF4G Ser1108 phosphorylation was increased compared to that in wild-type mouse muscles even when the mice were treated with dexamethasone (Fig. 6, B and C). We obtained a similar result with denervation-induced atrophy, where MNK2 expression was also induced (fig. S7). Upon starvation of mice for 16 hours, wild-type mice showed an almost complete elimination of eIF4G Ser1108 phosphorylation in gastrocnemius, as well as a partial decrease in total eIF4G, even though in wild-type animals MNK2 expression did not change in response to the 16-hour starvation (Fig. 7). However, increases in eIF4G Ser1108 phosphorylation were preserved in muscles of the MNK2 knockout animals even after fasting. This result demonstrates that MNK2 is necessary for the down-regulation of eIF4G Ser1108 phosphorylation during starvation. These in vivo data provided additional support for a role of MNK2 in regulating eIF4G phosphorylation. However, in both atrophy models, we failed to observe protection from muscle loss in the MNK2 knockout mice (fig. S8), most likely reflecting the multifactor nature of these models.

Fig. 6

MNK1 and MNK2 expression and eIF4G phosphorylation status in gastrocnemius of a dexamethasone-induced atrophy model. (A) MNK1 and MNK2 expression in gastrocnemius at the indicated days of dexamethasone (Dex) or vehicle (Veh) treatment in the WT mice were analyzed by qPCR (n = 5). (B) The phosphorylation status of eIF4E and eIF4G in WT and MNK2 KO gastrocnemius at day 21 of dexamethasone treatment was analyzed by Western blotting (n = 3 to 4). (C) Bar graph presentation of p-eIF4G–to–total eIF4G ratio of samples in (B). Error bars indicate the SD. P values were determined by unpaired t test. #P < 0.05; ##P < 0.0001 versus WT Veh; ###P < 0.0001 versus WT Dex.

Fig. 7

Effect of fasting on eIF4G Ser1108 phosphorylation in WT and MNK2 knockout mouse gastrocnemius. Mice were either fed ad libitum or left without food for 16 hours. Gastrocnemius extracts (25 μg of protein) prepared from WT and MNK2 KO mice were analyzed by Western blotting for eIF4E and eIF4G phosphorylation. The accompanying bar graphs show the ratio of p-eIF4G to total eIF4G (left) and MNK1 and MNK2 mRNA expression (right). Error bars indicate the SD. P values were determined by unpaired t test. #P < 0.001; ##P < 0.0001 versus WT Fed; ###P < 0.0001 versus WT Fasted.


The eukaryotic translation initiation factor eIF4G is a large, 225-kD phosphoprotein and the central docking platform of the eIF4F complex (17). Phosphorylation of eIF4G regulates its function; notably, phosphorylation at Ser1108 has been associated with enhanced eIF4E-eIF4G complex formation in skeletal muscle upon meal feeding, determined by leucine perfusion of the hindlimb (30, 40, 41). Here, we have investigated the signaling pathway that regulates eIF4G phosphorylation at Ser1108 and in the process discovered a previously unidentified signaling pathway activated upon muscle atrophy conditions and starvation.

We focused our study on the role of the MNKs on this phosphorylation event because MNKs bind to the C-terminal portion of eIF4G, in proximity to Ser1108 (15). Our results showed that MNK2, but not MNK1, negatively regulated Ser1108 phosphorylation. In C2C12 myotubes, MNK2 knockdown enhanced and MNK2 overexpression suppressed Ser1108 phosphorylation. This was somewhat of a surprise, because previous data demonstrated that MNKs mediate an increase in eIF4E phosphorylation, which might have suggested a positive role for these kinases in translation signaling (34). However, MNK2 expression increases during certain settings of atrophy, consistent with the idea that it helps mediate the required shutdown of translation under such conditions. The most unequivocal evidence for a negative role of MNK2 in regulating eIF4G phosphorylation came from analysis of the Ser1108 phosphorylation status in MNK knockout mice. Ser1108 phosphorylation was markedly enhanced in the skeletal muscle of MNK2 knockout and MNK1/2 double-knockout mice, but not in MNK1 knockout mice. Furthermore, in muscle atrophy settings in which increased MNK2 expression was observed, there was also a decrease in eIF4G phosphorylation in the wild-type mice, whereas eIF4G phosphorylation remained increased in the MNK2 knockout mice. A 16-hour starvation of wild-type mice did not affect MNK2 expression but was sufficient to reduce eIF4G Ser1108 phosphorylation. This would be desired in settings where amino acids are not available. However, in the absence of MNK2, this down-regulation was largely prevented. These in vivo results complement the biochemical data and establish a requirement for MNK2 in negatively regulating eIF4G Ser1108 phosphorylation. Nevertheless, this negative effect of MNK2 on Ser1108 phosphorylation clearly indicated that MNK2 is not the kinase responsible for the phosphorylation of eIF4G at Ser1108.

mTOR has been recognized as a key molecule in integrating various growth signals, from amino acid stimulation of protein synthesis to activation of translation by growth factors such as IGF1 (6, 42, 43). Phosphorylation of eIF4G at Ser1108 is regulated downstream of mTOR because rapamycin treatment inhibits serum- and IGF1-mediated stimulation of Ser1108 phosphorylation in HEK293 and C2C12 cells (25, 28). This result would suggest that mTOR, through TORC1, could potentially be the kinase responsible for the phosphorylation of eIF4G at Ser1108. However, we found that MNK2 overexpression suppressed both basal and IGF1-induced Ser1108 phosphorylation without affecting IGF1-induced AKT phosphorylation. Whereas rapamycin inhibited the phosphorylation of both eIF4G and p70S6K, MNK2 knockdown only reversed the effect of rapamycin on eIF4G and not on p70S6K phosphorylation. It is also unlikely that mTOR phosphorylates eIF4G through TORC2 because inhibition of eIF4G phosphorylation by the mTOR ATP binding site inhibitor AZD8055 was also reversed by MNK2 knockdown. These results indicate that although a distinct interplay exists between MNK2 and mTOR, mTOR is not responsible for mediating MNK2’s ability to inhibit Ser1108 phosphorylation.

By screening a panel of mouse kinase–directed siRNAs, we identified the SRPK family of kinases as the putative enzymes responsible for Ser1108 phosphorylation. Knocking down all three members of the SRPK family diminished not only the basal but also MNK2 knockdown–induced Ser1108 phosphorylation. Overexpression of SRPK1, the only one of the three members tested, enhanced basal eIF4G phosphorylation and MNK2 overexpression blocked this phosphorylation event. We also found that rapamycin suppressed the increase in eIF4G phosphorylation in SRPK1-overexpressing cells. Together, these data indicate that SRPK is negatively regulated by MNK2 and positively regulated by TORC1. The net balance of these opposing effects apparently determines the status of eIF4G phosphorylation. In addition, the MNK2 knockdown can reverse the negative effect of TORC1 inhibition, indicating that SRPK retains higher basal activity in the absence of MNK2. SRPKs were initially identified as kinases specifically phosphorylating serine residues within regions rich in serine-arginine dipeptide motifs of splicing factors (44, 45). Even though amino acid residues surrounding Ser1108 of eIF4G, RRVVGRSSLS(p)RERG, do not have the S-R dipeptide motif of the typical SRPK substrates, they are rich in serine and arginine; thus, eIF4G may be a substrate of SRPK. The possibility that MNK2, through SRPK, participates in the regulation of mRNA splicing and other pathways is intriguing and warrants further investigation.

MNK1 and MNK2 are the only kinases that phosphorylate eIF4E at Ser209. However, the physiological role of that phosphorylation event in the context of protein translation is not clear (46, 47). Although both MNK1 and MNK2 are expressed in skeletal muscle, only MNK2 was induced in two animal models of atrophy, denervation-induced atrophy and dexamethasone-induced atrophy. Here, we have identified a unique role of MNK2, not shared by MNK1, in the negative regulation of eIF4G Ser1108 phosphorylation. In addition, we showed that MNK2, but not MNK1, physically interacts with TORC1. Our coimmunoprecipitation data suggest that not only MNK2 but also SRPK1 physically interacts with eIF4G. Therefore, we have uncovered a protein interaction network that could explain how MNK2 negatively regulates eIF4G Ser1108 phosphorylation through SRPK and TORC1 (fig. S9).

Finally, even though eIF4G phosphorylation at Ser1108 has been associated with enhanced protein translation, there are many other phosphorylation sites in eIF4G that could or do regulate eIF4G activity. For example, the phosphorylation event catalyzed by Pak2 at Ser896 inhibits translation (48), and the phosphorylation at Ser1186 by PKCα (protein kinase C α) regulates the binding of MNK1 to eIF4G (49). In addition, activities of other translation initiation factors, such as eIF4B, and ribosome components, such as RPS6, are regulated by phosphorylation (43). It remains to be elucidated how these other phosphorylation events in combination with eIF4G Ser1108 phosphorylation affect the assembly of eIF4F and higher-level protein translation complexes.

Materials and Methods

Cell culture

C2C12 cells were obtained from the American Type Culture Collection and maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (HyClone), 1× GlutaMAX, Hepes, and penicillin and streptomycin (Invitrogen). For differentiation, C2C12 cells were grown to 100% confluency and were induced to differentiate by switching into differentiation medium [DMEM supplemented with 2% horse serum, 1× GlutaMAX, Hepes, and penicillin and streptomycin (Invitrogen)] and incubated at 37°C at 7.5% CO2–92.5% air. HEK293 cells were maintained in DMEM (Invitrogen) supplemented with 10% heat-inactivated FBS (Invitrogen), 1× GlutaMAX, and penicillin and streptomycin (Invitrogen).

Overexpression and gene knockdown with siRNA

Target protein overexpression in C2C12 myotubes was performed by adenoviral-mediated transduction. C2C12 cells at day 1 of differentiation were transduced with the respective adenovirus at 1 × 109 particles per milliliter in fresh differentiation medium. Cell lysates were prepared 48 hours later for Western blot analysis or coimmunoprecipitations. For siRNA knockdown experiments, siRNAs against target genes from Qiagen were used. C2C12 cells were transfected on day 1 of differentiation with DharmaFECT 1 according to the manufacturer’s protocol. RNA and protein lysates were prepared 48 hours after transfection.

Cell lysate preparation

Lysates used for Western blot analysis were prepared by lysing cells in PBS (without calcium chloride and magnesium chloride) containing 1% Nonidet P40 and protease and phosphatase inhibitors (Roche Applied Science) and incubating for 30 min at 4°C. Then, clear lysate was collected by centrifugation at 13,000 rpm in a microfuge for 15 min at 4°C.


Cells rinsed twice with ice-cold PBS were lysed in 0.3% CHAPS buffer with no salt (40 mM Hepes, 10 mM pyrophosphate, 2 mM EDTA, 0.3% CHAPS, a tablet of EDTA-free protease inhibitors, and a tablet of phosphatase inhibitors). After cell lysates were rotated for 30 min in the cold, the soluble fractions of cell lysates were isolated by centrifugation at 13,000 rpm in a microfuge for 10 min. For immunoprecipitations, primary antibodies were added to the lysates and incubated with rotation for 2 to 3 hours at 4°C. Protein G agarose beads (20 μl per immunoprecipitation) were added and the incubation continued overnight. On the next day, immunoprecipitates were washed three times each with low-salt wash buffer (125 mM NaCl, 40 mM Hepes, 10 mM pyrophosphate, and 2 mM EDTA). Immunoprecipitated proteins were denatured in sample loading buffer, boiled for 5 min, and analyzed by immunoblotting.

The antibodies recognizing the following proteins or phosphorylated proteins were obtained from Cell Signaling Technology: MNK1, p-MNK (Thr197/202), p-eIF4G (Ser1108), eIF4G, eIF4E, p-RPS6 (Ser240/244), RPS6, p-AKT (Thr308), p-AKT (Ser473), AKT, p-p70S6K (Thr389), p-p70S6K (Ser371), p70S6K, mTOR, Raptor, Rictor, GβL, and Pras40. The antibody recognizing p-eIF4E (Ser209) was obtained from Millipore. Rabbit anti-Flag antibody was from Sigma. Mouse anti-Flag antibody was from Agilent.

MNK knockout mice and dexamethasone- or denervation-induced atrophy model

MNK1, MNK2, and MNK1/2 knockout mice were obtained from RIKEN. Dexamethasone sodium phosphate injection, USP (Baxter Healthcare Corporation) was administered by Alzet mini pump at 2.4 mg/kg per day for 21 days. At the indicated days, the animals were killed by CO2 asphyxiation, and the tibialis anterior, gastrocnemius, and quadriceps muscles were dissected out, weighed, and snap-frozen in liquid nitrogen. For denervation-induced muscle atrophy, mice were anesthetized with 3% isoflurane in oxygen, and, under aseptic conditions, an ~5-mm segment of the sciatic nerve in one leg was resected. Buprenorphine was administered both preemptively and postoperatively for analgesia. Fourteen days after denervation, the animal was euthanized and tissues were dissected out, weighed, and analyzed.

To assess the effects of fasting, we allowed wild-type and MNK2 knockout mice free access to water but not food for 16 hours (6 p.m. to 10 a.m.).

High-content screen to identify potential kinases responsible for eIF4G Ser1108 phosphorylation

An siRNA library was screened in C2C12 myoblasts with high-content image analysis. Cells were seeded at 500 cells per well in 384-well plates, and siRNA was introduced into cells by reverse transfection with DharmaFECT 2. Three days after transfection, cells were serum-starved for 1 hour and stimulated with IGF1 (300 ng/ml) for an additional 2 hours. Cells were fixed with paraformaldehyde and stained with antibody against phospho-eIF4G (Ser1108) followed by the Alexa Fluor 546 secondary antibody. The mean signal intensity of each well was measured with IN Cell Investigator software (GE Healthcare) after imaging in an IN Cell Analyzer 1000 instrument (GE Healthcare). Phospho-eIF4G signal intensity was further normalized with the signal of negative control siRNA (Silencer Negative Control 2, Ambion) and the signal of positive control eIF4G siRNA (SI00992278, Qiagen) by linear scaling to 0 as negative control and −100 as positive control. Four siRNAs (Qiagen) against each gene were screened in duplicate. Two of four siRNAs that resulted in less than −30 of normalized phospho-eIF4G intensity were selected for further confirmation study.


MNK expression constructs were Flag-tagged at the N terminus and subcloned into pcDNA3+. MNK1-NC2 construct was generated by replacing the N-terminal 73 amino acids of MNK1 (end at …QNGK) with the N-terminal 61 amino acids of MNK2 (end at …ITSQ) and the C-terminal 50 amino acids of MNK1 (start at EENE…) with the C-terminal 56 amino acids of MNK2 (start at …DEDL). MNK2-NC1 construct was generated by reciprocal replacement of MNK1 N- and C-terminal fragments into MNK2. Point mutations were introduced with QuikChange site-directed mutagenesis (Stratagene).

Adenoviral constructs were generated in pAd/CMV/V5 DEST vectors with the Gateway cloning system. Recombinant adenoviruses were generated at Welgen Inc. Human myc-DDK–tagged SRPK1 complementary DNA (cDNA) was obtained from Origene.

Transcript analysis

Samples were prepared in biological triplicate. RNA isolation was performed with Tri Reagent (Molecular Research Center). Isolated RNA (1 μg) was reverse-transcribed into cDNA with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Newly synthesized cDNA was quantified with TaqMan (Applied Biosystems). The housekeeping gene TATA-binding protein (TBP) was used as an internal control for the generation of ΔCt values and subsequent assessment of fold differences in gene expression.

Supplementary Materials

Fig. S1. Densitometry quantification of the ratio of phospho/total eIF4G and eIF4E from Fig. 1A.

Fig. S2. Densitometry quantification of the ratio of phospho/total eIF4G and p70S6K from Fig. 3A.

Fig. S3. MNK2 overexpression partially inhibits global protein synthesis.

Fig. S4. MNK2 and Pras40 compete for binding to Raptor.

Fig. S5. MNK2 and SRPK1 interact with eIF4G.

Fig. S6. Densitometry quantification of the ratio of phospho/total eIF4G in Fig. 5.

Fig. S7. Expression of atrophy-associated genes and eIF4G phosphorylation status in gastrocnemius of a denervation-induced atrophy model.

Fig. S8. Muscle mass in dexamethasone-induced and denervation-induced atrophy models.

Fig. S9. Proposed negative role of MNK2 on protein translation through SRPK and TORC1.

Table S1. Summary of primary mouse kinase panel siRNA screen. (Excel file)

References and Notes

Acknowledgments: We thank B. Latario, R. Pulz, P. Rao, B. Clarke, and J. Shi for helpful discussions or comments. We also thank J. Eash for skillful in vivo work and E. Frias and C. Mickanin for their initial work on the siRNA kinase screen. Funding: This work was supported by the Novartis Institutes for BioMedical Research Inc. Author contributions: M.K., S.C., J.C., and X.Q. performed the experiments. C.I. designed and performed in vivo studies. S.Z. and A.C. performed high-content siRNA screen. M.K., S.C., S.-I.H., and D.J.G. designed the experiments and analyzed the data. S.-I.H. and D.J.G. wrote the paper. Competing interests: All authors are employees of Novartis Institutes for Biomedical Research.
View Abstract

Navigate This Article