Phosphorylation of the Eukaryotic Translation Initiation Factor eIF4E Contributes to Its Transformation and mRNA Transport Activities

INTRODUCTION

The eukaryotic translation initiation factor eIF4E is dysregulated in a wide variety of human cancers including acute myeloid leukemia (1) and breast cancers in which it is used as a prognostic indicator (2) . Furthermore, overexpression and dysregulation of eIF4E leads to increased tumor number, invasion, and metastases in mouse models (3) . Transgenic expression of eIF4E leads to a variety of cancers (4) . eIF4E is required for cellular survival, acting in the rate-limiting step of cap-dependent translation initiation in the cytoplasm (5) . In the nucleus, eIF4E forms nuclear bodies and plays a role in the nucleo-cytoplasmic export of a subset of growth-promoting mRNAs including cyclin D1 (6, 7, 8) . In contrast, eIF4E does not alter the mRNA transport of housekeeping genes such as glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and actin. eIF4E must bind the m⁷G cap of the mRNA in both subcellular compartments to function (5 , 7) . In AML, this mRNA-transport activity is dysregulated, and down-regulation of eIF4E leads to return of cyclin D1 mRNA transport to normal (1) . Importantly, mutational studies of eIF4E indicate that the W73A mutant, which is unable to act in translation, promotes cyclin D1 transport as readily as wild-type eIF4E (7 , 8) . Furthermore, this mutant transforms cells as readily as wild-type eIF4E, suggesting that the mRNA transport function of eIF4E is linked to its ability to transform cells (7 , 8) . Clearly regulation of eIF4E in both subcellular compartments is critical for maintaining normal growth control. To date, the only reported post-translational modification to eIF4E is its phosphorylation on S209 (5) . There are a plethora of reports investigating the effects of phosphorylation on the translational activities of eIF4E (9, 10, 11) . In general, it seems that phosphorylation of eIF4E reduces m⁷G cap-binding activity slightly but does not substantially alter cap-dependent translation (9, 10, 11, 12) . No studies have examined the role of phosphorylation on either the ability of eIF4E to transform cells or to modulate its activities in mRNA transport. We examine the effect of phosphorylation on these processes here.

MATERIALS AND METHODS

pMV eIF4E was a kind gift of N. Sonenberg (13) . For Figs. 1 ⇓ , 2 ⇓ , and 4 ⇓ , eIF4E was subcloned into pcDNA3.1 to use the Xpress tag (8) . 1 Confocal microscopy was carried out as described previously (7 , 8) . Antibodies were to eIF4E (Transduction Laboratories, Newington, NH), Xpress tag (Invitrogen, Carlsbad, CA), β-actin (Sigma, St. Louis, MO), or promyelocytic leukemia protein (PML; ref. 7 ). Phospho-specific antibodies to eIF4E were a kind gift of S. Morley and obtained from (Cell Signaling, Beverly, MA). Transformation assays were carried out in NIH3T3 cells as described previously, staining foci with Giemsa (7 , 8) . Experiments were carried out in triplicate. Enhanced chemiluminescence was used to carry out Western analysis. Northern analysis was carried out according to the manufacturer’s instructions (Ambion, Austin, TX). Probes for cyclin D1, GAPDH, tRNA^Lys, and U6snRNA were described previously (8) . CGP57380 was a kind gift of Simon Morley and used at 20 μmol/L for 60 minutes as described previously (14) . One-dimensional vertical isoelectric focusing was done as described previously (15) .

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Phosphorylation of eIF4E enhances its transformation activity. A, NIH3T3 cells were transfected as indicated and transformation was monitored by foci formation assays as described. The number of foci were counted and SDs are shown. Results are from three independent experiments (left panel). Representative plates for each transfection are also shown (right panel). B, Western blot (w.b.) analysis of transfected cells or vector controls. Note that residual phosphorylation observed in the mutant lanes using the phospho-specific eIF4E antibody (eIF4E-P) is attributable to signal from endogenous phosphorylated eIF4E.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Phosphorylation mutants have impaired cyclin D1 mRNA transport. A, NIH3T3 cells were transfected as described and fractionated into nuclear (n) and cytoplasmic (c) fractions. RNA obtained from each fraction (fractionated RNA) was analyzed by Northern blot (N.B.) analysis. GAPDH levels were used to assure equal loading of the samples, whereas tRNA^Lys and U6snRNA were used as fractionation controls for the cytoplasmic and nuclear fractions, respectively. Cyclin D1 mRNA nuclear to cytoplasmic ratios were quantified by densitometry, normalized against GAPDH levels, and the obtained results from three independent experiments are shown on the bar graph as mean ± SD (see the text). B, total cell protein and RNA obtained from the aforementioned cells were analyzed by Western (WB; top panel) and Northern (N.B.; bottom panel) blot analysis, respectively. β-Actin was used as a loading control for Western blot analysis (W.B.; top panel), whereas GAPDH was used as a loading control for Northern blot analysis (NB; bottom panel). Note that the eIF4E antibody recognizes both endogenous and exogenous eIF4E whereas the Xpress antibody only recognizes the exogenous eIF4E. The phospho-specific antibody shows some residual signal in ST209/210AA expressing cells; this is because of the endogenous phosphorylated eIF4E.

RESULTS AND DISCUSSION

We examined the importance of phosphorylation to the ability of eIF4E to transform cells. Previous studies indicate that eIF4E is phosphorylated on S209 and in vitro at T210 when S209 is mutated to alanine (10) . Thus, a double mutant was constructed, ST209/210AA. Transformation was monitored by anchorage-dependent foci formation assays in NIH3T3 cells (Fig. 1A) ⇓ . Clearly, this mutation substantially abrogates the ability of eIF4E to transform cells when compared with wild-type constructs. Similar results were observed for constructs with single site mutations S209A and S209D, in which the aspartic acid should somewhat mimic the negative charge of the phosphate. Note that all mutants are expressed to similar levels and that as expected, the mutants are not phosphorylated (Fig. 1B) ⇓ .

Recent studies indicate that eIF4E phosphorylation does not alter global translation rates, and mice developed normally in the absence of phosphorylated eIF4E (16) . It seems that although phosphorylation may be important for fine-tuning the translation program, it is not critical for translation itself (9) . Thus, we examined the possibility that phosphorylation modulated the nuclear functions of eIF4E. Because it is well established that eIF4E modulates transport of cyclin D1 mRNA, we used cyclin D1 as a model transcript to monitor this activity. Western analysis clearly indicates that cyclin D1 protein levels are substantially reduced in cells expressing the phosphorylation mutants in comparison to the wild-type protein (Fig. 1B) ⇓ . Importantly, actin levels, which are not regulated at the level of eIF4E-dependent mRNA transport (6 , 17) , were not altered.

We extended these studies to determine whether cyclin D1 protein levels were reduced because of a decrease in eIF4E-dependent cyclin D1 mRNA transport (Fig. 2) ⇓ . Appropriately, transfected cells were fractionated into nuclear and cytoplasmic compartments, and the distribution of RNA was monitored by Northern analysis. Resulting bands were scanned, and the intensity of the bands was quantified by ImageQuant Software (Molecular Dynamics, Sunnyvale, CA). Area quantitation report values obtained for cyclin D1 mRNA from each fraction were normalized against corresponding area quantitation report values obtained for GAPDH mRNA. Relative nuclear to cytoplasmic ratios for each sample were calculated by dividing the normalized area quantitation report values for nuclear fraction with the normalized area quantitation report values for cytoplasmic fraction. Experiments were independently repeated four times, and the results were presented on a bar graph as mean value ± SD (Fig. 2) ⇓ . Clearly, there is substantially more cyclin D1 in the cytoplasmic compartment than in the nucleus in eIF4E-overexpressing cells relative to vector controls, consistent with previous reports (6, 7, 8 , 17) . In contrast, cells expressing the ST209/210AA mutant had a higher nuclear to cytoplasmic ratio of cyclin D1 transcripts in comparison with fractions obtained from wild-type eIF4E, indicating that cyclin D1 transcripts are not as readily transported as in the wild-type expressing cells. Importantly, the ST209/210AA eIF4E cells still had a lower nuclear cytoplasmic ratio of cyclin D1 transcripts than the vector controls, indicating that eIF4E phosphorylation is not required for transport but rather enhances it. This is consistent with the observation that this mutant expressed more cyclin D1 protein than vector controls, although substantially less cyclin D1 than observed in wild-type eIF4E cells (Fig. 2B) ⇓ . Importantly, GAPDH mRNA, which is not regulated at the level of eIF4E-dependent mRNA transport (6) , did not have its subcellular distribution altered (Fig. 2) ⇓ . tRNA^Lys and U6snRNA serve as fractionation controls for the nuclear and cytoplasmic compartments, respectively. Furthermore, there is no alteration in total levels of cyclin D1 or GAPDH transcripts in cells expressing either wild-type or mutant eIF4E.

CGP57380, a small molecule inhibitor of the Mnk kinases that exclusively phosphorylates eIF4E, has been described previously (14 , 18) . We used this to examine the effects of phosphorylation on eIF4E-dependent cyclin D1 mRNA transport (Fig. 3) ⇓ . Addition of CGP57380 led to decreased phosphorylation of eIF4E as observed by either isoelectric focusing or an antibody specific to the phosphorylated form of eIF4E (Fig. 3A) ⇓ . Notably, total levels of eIF4E are unchanged (Fig. 3A) ⇓ . Importantly, addition of CGP57380 led to decreased cyclin D1 protein levels without altering GAPDH. To confirm that this occurred at the level of cyclin D1 mRNA transport, untreated or treated cells were fractionated into nuclear and cytoplasmic compartments, and the distribution of cyclin D1 mRNA was monitored (Fig. 3C) ⇓ . As expected, more cyclin D1 mRNA was present in the cytoplasmic fraction relative to the nuclear fraction in untreated cells than in treated cells. The distribution of GAPDH mRNA was unchanged as were the total levels of cyclin D1 and GAPDH mRNA (Fig. 3B and C) ⇓ . Thus, CGP57380 decreased levels of phosphorylated eIF4E, and subsequently down-regulated eIF4E-dependent cyclin D1 mRNA transport.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Inhibition of phosphorylation leads to decreased cyclin D1 mRNA transport in eIF4E-overexpressing cells. A, Western analysis (W.B.) reveals lower levels of cyclin D1 are produced after addition of CGP57380. Both isoelectric focusing (vertical slab gel isoelectric focusing) gels probed for eIF4E or SDS-PAGE gels probed with an antibody specific to eIF4E indicate that addition of CGP57380 leads to a substantial reduction in eIF4E phosphorylation. ∗, phosphorylated form. Note that CGP57380 depletes endogenous phosphorylated eIF4E as well. Actin levels serve as a loading control. B, Northern analysis (N.B.) indicates that total mRNA levels of cyclin D1 and GAPDH do not change upon drug addition. C, fractionation studies indicate that the nuclear to cytoplasmic ratio of cyclin D1 transcripts are lower in cells not treated with CGP57380 indicating cyclin D1 is more efficiently transported in these cells.

One possibility for the reduced activity of unphosphorylated eIF4E is that its subcellular distribution has been altered. We examined this possibility by confocal microscopy (Fig. 4) ⇓ . The distribution of wild-type and ST209/210AA proteins is indistinguishable, with both being found in nuclear bodies and diffusely throughout the cytoplasm (Fig. 4A) ⇓ . Parallel experiments in which cells were fractionated into nuclear and cytoplasmic fractions confirmed that the wild-type and mutant proteins did not have an altered subcellular distribution (data not shown). To assess if the subnuclear distribution of the ST209/210AA mutant was altered, we examined whether a subset of eIF4E nuclear bodies still colocalized with those containing PML (Fig. 4B) ⇓ . Previous studies by multiple groups indicate that a subset of eIF4E bodies colocalize with PML (7 , 19) . Our confocal experiments indicate that a subset of both of the mutant and wild-type proteins colocalize to PML nuclear bodies as expected. This finding strongly suggests that eIF4E phosphorylation does not modulate eIF4E activity by altering its subcellular distribution.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

The subcellular distribution of eIF4E does not apparently depend on its phosphorylation state. Confocal micrographs depict NIH3T3 cells transfected with vector controls, wild-type eIF4E or the ST209/210AA mutant. Exogenous eIF4E is observed using the Xpress antibody (red) and exogenous and endogenous with an eIF4E antibody (green). Additionally, in B, PML is detected using rabbit polyclonal antibody (red; kind gift from Gerd Maul). In both A and B, the overlay is shown in yellow. Confocal micrographs represent a single optical section of ∼300 nm through the cell. Scale bars = 20 μm. DAPI, 4′,6-diamidino-2-phenylindole.

Here we show that phosphorylation of eIF4E, although not necessary nor absolutely required, significantly enhances both the ability of eIF4E to transform cells and eIF4E dependent mRNA transport. Previous studies suggest that phosphorylation of eIF4E is not required for survival of cells where S209A mutants can rescue yeast with eIF4E knocked out (9) , and recent studies indicate the eIF4E phosphorylation is not required for global translation or for normal development in mice (16) . In contrast, phosphorylation is required for normal development of Drosophila (20) , suggesting differences in its importance in vertebrates and invertebrates. Our studies suggest that phosphorylation of eIF4E in the nuclear compartment is not important to its subnuclear distribution, but rather it is likely that it stabilizes interactions with proteins important to the ability of eIF4E to promote mRNA transport. In the nucleus, eIF4E physically associates with a restricted subset of mRNAs. 1 These mRNAs have a common element in their 3′ untranslated region known as an eIF4E sensitivity element (4ESE). 1 In this model for eIF4E-dependent mRNA transport, eIF4E binds the m⁷G cap of the mRNA, and other regulatory proteins bridge the association between the 4ESE and eIF4E. 1 One possibility is that phosphorylation of eIF4E modulates association with these 4ESE-binding proteins and thus modulates the ability of eIF4E to promote transport of these growth-promoting mRNAs and thus to promote growth and transformation. The slight reduction in cap affinity associated with this mutation may allow eIF4E ribonucleoproteins to exchange with other critical ribonucleoproteins associated with transport. Additional elucidation of this regulatory pathway will allow us to understand how eIF4E phosphorylation exerts its affects in the nucleus and how these factors contribute to eIF4E-mediated oncogenic transformation.