The present study is the first report providing evidence for a physiological role of a truncated form of the mRNA cap-binding protein eIF4E1 (eukaryotic initiation factor 4E1). Our initial observation was that eIF4E, which mediates the mRNA cap function by recruiting the eIF4F complex (composed of eIF4E, 4G and 4A), occurs in two forms in porcine endometrial tissue in a strictly temporally restricted fashion. The ubiquitous prototypical 25 kDa form of eIF4E was found in ovariectomized and cyclic animals. A new stable 23 kDa variant, however, is predominant during early pregnancy at the time of implantation. Northern blotting, cDNA sequence analysis, in vitro protease assays and MS showed that the 23 kDa form does not belong to a new class of eIF4E proteins. It represents a proteolytically processed variant of eIF4E1, lacking not more than 21 amino acids at the N-terminus. Steroid replacements indicated that progesterone in combination with 17β-oestradiol induced the formation of the 23 kDa eIF4E. Modified cell-free translation systems mimicking the situation in the endometrium revealed that, besides eIF4E, eIF4G was also truncated, but not eIF4A or PABP [poly(A)-binding protein]. The 23 kDa form of eIF4E reduced the repressive function of 4E-BP1 (eIF4E-binding protein 1) and the truncated eIF4G lacked the PABP-binding site. Thus we suggest that the truncated eIF4E provides an alternative regulation mechanism by an altered dynamic of eIF4E/4E-BP1 binding under conditions where 4E-BP1 is hypophosphorylated. Together with the impaired eIF4G–PABP interaction, the modified translational initiation might particularly regulate protein synthesis during conceptus attachment at the time of implantation.
Protein synthesis is a major determinant of cell growth and differentiation. In the context of conceptus implantation, maternal ovarian steroids and embryonic signals, such as oestrogens [1,2], have been found to influence proliferation and differentiation of the endometrium by modulating those signalling cascades regulating mRNA translation. This view is supported by results showing that signal transduction pathways involving the MAPKs (mitogen-activated protein kinases) and the serine/threonine protein kinase Akt can be activated by 17β-oestradiol [3,4] or P4 (progesterone) . Among other processes, these kinases modulate protein synthesis at the translational level [6–8]. The main targets are translational initiation factors [eIFs (eukaryotic initiation factors)] such as the mRNA-cap-binding protein eIF4E and its repressors 4E-BP1 and 4E-BP2 (eIF4E-binding proteins 1 and 2).
In eukaryotes, more than 95% of proteins are synthesized via cap-dependent translation. In this context, eIF4E plays a key role in the recruitment of mRNAs to ribosomes . The first detected prototypical eIF4E is found in all eukaryotes. It has an apparent molecular mass of approx. 25 kDa and might be the rate-limiting factor in translational initiation [10,11]. This process is influenced by the secondary and tertiary structure of the 5′-UTR (5′-untranslated region) of the mRNA and by the cap structure (m7GpppN) of the mRNA molecules. The functional cap-binding complex eIF4F is composed of the mRNA cap-binding protein eIF4E, eIF4A, an RNA helicase responsible for the unwinding of secondary structures of mRNAs, and eIF4G serving as a scaffold protein for the assembly of the complex . eIF4F mediates the cap function during translation initiation. However, today the most sophisticated model of translational stimulation is the so-called ‘closed loop’ model [12,13]. Simultaneous interactions between the 5′-cap and the 3′-poly(A) sequence of the mRNA are mediated by the protein factors, whereby eIF4G provides binding sites for eIF4E, eIF4A, eIF3 (which bridges eIF4E to the small ribosomal subunit), PABP [poly(A)-binding protein], MNK (MAPK-interacting kinase), the kinase which directly phosphorylates eIF4E, and also for RNA. This complex facilitates ribosome binding and stimulates translation efficiency 8–10-fold .
The biological activity of the cap-binding protein eIF4E can be regulated at different levels. First, eIF4E is phosphorylated at Ser209 in response to a variety of stimuli such as growth factors, cytokines or amino acids . It was shown that eIF4E phosphorylation reduces its affinity to the cap structure. eIF4E released from the cap can probably bind to a different subset of mRNAs and enhance the translation rate .
Secondly, eIF4E-binding proteins can work as repressors of eIF4E function. Specific proteins called 4E-BP1, 4E-BP2 and 4E-BP3 were shown to act as competitors of eIF4E/4G binding. When hypophosphorylated, they form complexes with eIF4E and prevent the formation of the functional eIF4F complex . 4E-BP1 and 4E-BP2 have at least six potential phosphorylation sites, and hierarchical phosphorylation reduces their affinity to eIF4E as well as suppressing their inhibitory effect on the translational initiation [7,18].
Thirdly, subclasses of eIF4E encoded by different genes (eIF4E classes 1–3) or splicing variants have been detected in many organisms from different species [19,20]. Generally, some eIF4E-related proteins may act as translation factors and stimulate either global mRNA recruitment or the recruitment of a subset of mRNAs. Others may exhibit only partial activity and thus act as inhibitors of translation. However, the complexity of the biological function of the different members of the eIF4E family is not fully understood and remains to be investigated in detail, especially in the light of recently detected new eIF4E-binding partners .
Proteolytic processing resulting in stable variants has never been detected for eIF4E in a physiological context, in contrast with other initiation factors such as eIF4G  or PABP . However, structural analyses with recombinant N-terminally truncated mammalian eIF4E forms have been performed [22,23]. The N-terminal region of eIF4E, from amino acids 1 up to Gln40, appeared to form a unique flexible unstructured segment. It was suggested that this segment is dispensable for cap-dependent translation. A truncation of 33 amino acids from the N-terminus yielded a protein binding to the cap structure with the same affinity as the full-length eIF4E . More recently, however, a regulatory impact of the N-terminal segment was suggested . Deletions in this area result in different binding of the repressors 4E-BP1 and 4E-BP2. These alterations depend on the length of the deleted segment and on the interaction state of both molecules [23–25].
In respect to eIF4G binding, deletions of up to 20 amino acids from the N-terminus of eIF4E from yeast did not result in a reduced affinity to eIF4G compared with the wild-type. In contrast, deletions of larger segments (up to 35 amino acids) result in a significantly reduced affinity .
In the present study, we analysed the expression of eIF4E in the porcine endometrium during implantation. At this time, as a consequence of embryonic/maternal cross-talk, structural and functional alterations occur in the endometrial epithelium, allowing attachment of blastocysts and their further development . Surprisingly, we found a naturally occurring proteolytically processed eIF4E variant. The truncated 23 kDa form persists as a stable variant in the porcine endometrium during implantation. Taken together, our results suggest that the proteolytic cleavage of eIF4E represents an alternative regulatory mechanism of translational initiation, which is distinct from the well-known regulation through phosphorylation of eIFs. An altered dynamic of eIF4E/4E-BP1 binding, together with the impaired eIF4G–PABP interaction also observed, might be important in regulating endometrial protein synthesis, particularly during non-invasive implantation in pigs.
Details of the materials and methods used for RNA analysis, Northern blotting and cDNA analysis are available in the Supplementary Experimental section at http://www.BiochemJ.org/bj/432/bj4320353add.htm.
Animals, steroid replacements and unilateral pregnant uterine horn
For all of the animal studies, ethical guidelines were followed and the research has been authorized by German authorities (VI-522a-7221.31-1-036/00).
Mature German Landrace gilts (Sus scrofa domestica; 8.5 months of age, 120–125 kg) were oestrus- and ovulation-synchronized as described previously . Gilts were hysterectomized on C13 [day 13 of the oestrus cycle (n=7)] and on P13 [day 13 of pregnancy (n=10)], P15 (n=15; including five animals used for the model unilateral pregnant uterine horn; see below) and P30 (n=7) [gilts were bred by artificial insemination 24 and 38 h after hCG (human chorionic gonadotropin)]. Additionally, on C10, 24 gilts were OVX (ovariectomized). Steroid replacements with EB (oestradiol benzoate) and P4 were performed as described previously . In further experiments, one uterine horn from each of five pigs was endoscopically closed with a staple line of titanium clips before insemination. This treatment results in pregnancy in one uterine horn only.
Preparation of endometrial tissue extracts
For analysis of initiation factors, 1.5 g of endometrial tissue was used and the homogenate was prepared in the presence of protease inhibitors as described previously , except that the EDTA concentration was 10 mM. Supernatants were used for immunoblots and protein–protein interaction assays. For in vitro protease assays, the tissue was extracted with a buffer containing 150 mM NaCl and 20 mM Tris/HCl (pH 7.2) without any protease inhibitors. The protein concentration was determined as described previously . Granulosa cells, in vitro-matured porcine oocytes and blastocysts were handled as described previously [30,31]. Furthermore, tissues from other organs (muscle, heart, kidney, brain, ovary, spleen and liver) were homogenized as described above.
For the analysis of eIF4F components, PABP, 4E-BP1 and 4E-BP2 by Western blotting, the standard SDS/PAGE and transfer method was used . The blots were developed with the ECL-Plus detection system and were digitalized using the Camilla Camera System (raytest). The absorbance of the bands was measured and evaluated by Aida software (raytest). Binding assays with the different eIF4E forms to m7-GTP–Sepharose was performed as described previously . For the in vitro protease assay, 50 ng of N-terminally GST (glutathione transferase)-tagged full-length eIF4E (51 kDa) served as an external substrate to be cleaved by endometrial lysates (20 μl) collected after OVX and on P15. Incubation was performed for 0, 30, 60 or 120 min at 37 °C. In parallel assays, the samples were treated with 5 mM EDTA or EGTA supplemented with or without 10 mM CaCl2. The reaction was stopped by adding 3×-concentrated SDS/PAGE sample buffer. Samples were analysed by Western blotting and were probed for eIF4E and GST respectively. For semi-preparative eIF4E cleavage and subsequent MS analysis, 1 μg of GST–eIF4E fusion protein was used as an external substrate. To avoid high contamination with endometrial proteins and to achieve maximal cleavage, the cleavage reaction was carried out sequentially by serial addition of OVX or P15 endometrial lysates, diluted 1:20, for 30 min at 37 °C. The addition of diluted lysate was repeated four times. The cleavage reaction was monitored on Coomassie-Blue-stained gels after SDS/PAGE and by Western blotting. For MS analysis, tryptic in-gel digestion was performed and the MALDI–TOF-TOF (matrix-assisted laser-desorption ionization–time-of-flight-time-of-flight) measurement was carried out on the 4800 MALDI-TOF-TOF analyser (Applied Biosystems) as described previously . For the database search, the Mascot search engine (version 2.104; Matrix Science) with a specific NCBI (National Center for Biotechnology Information) sequence database was used.
In vitro translation and protein–protein interaction analysis
For in vitro translation, nuclease-treated RRL (rabbit reticulocyte lysate) from Promega was used to synthesize biotinylated proteins. To evaluate the eIF4E dependency of translation or to mimic the situation in the endometrium, the RRL was modified as follows. In the first case, eIF4E was depleted from the RRL by m7-GTP–Sepharose as described previously  and the RRL was subsequently supplemented either with recombinant eIF4E (0.75 μl/50 μl of assay volume) or unphosphorylated His-tagged 4E-BP1 (0.75 μg/50 μl of assay volume). In the second case, the complete RRL was subjected to limited proteolysis by supplementation with endometrial lysates obtained after OVX or on P15. Titration studies were carried out in advance to achieve optimal cleavage of eIF4E and eIF4G while avoiding exorbitant contamination of the RRLs with endometrial proteins. Lysates containing 1.5 μg of protein were added to 50 μl of RRL. In additional assays, the treated RRLs were supplemented with recombinant 4E-BP1 (1 μg/50 μl of RRL). All treatments of the RRLs were monitored by Western blotting. The in vitro translation reactions were carried out in a final volume of 50 μl with 60% RRL, 140 mM potassium acetate, 0.5 mM magnesium acetate, 10 mM ceratin phosphate, 1 mM amino acid mix and 1 μl of Transcend™ tRNA. The RNA was stabilized by addition of 1 μg of RNasin. The reaction was started by adding 50 ng of capped polyadenylated Xef1 mRNA. Samples were taken at zero time, and after 15 and 30 min. Biotinylated Xef1 translation products were analysed by Western blotting and visualized by streptavidin–HRP (horseradish peroxidase) detection as described previously . Directly after the in vitro translation reaction, the RRLs were analysed by immunoprecipitation with immobilized anti-eIF4E as described previously  and probed for eIF4E, eIF4G, eIF4A, PABP1 and 4E-BP1. To analyse the binding of the eIF4E variants to 4E-BP1, recombinant N-terminally His-tagged full-length 4E-BP1 was bound to Ni-NTA (Ni2+-nitrilotriacetate) magnetic agarose beads according to the manufacturer's instruction. His-tagged 4E-BP1 (15 μl of slurry; 10 μg of His-tagged 4E-BP1) bound to the beads was incubated with endometrial lysates after OVX and on P15 for 2 h at 4 °C under constant agitation. As negative controls, lysates were incubated with the beads and without His-tagged 4E-BP1 under the same conditions. The beads were washed three times with buffer containing 50 mM NaH2PO4 (pH 8.0), 150 mM NaCl, 20 mM imidazole and 0.005% Tween 20. Proteins were eluted from the resin with a buffer of the same composition supplemented with 250 mM imidazole. Bound fractions and precipitated supernatants were analysed by Western blotting and probed for eIF4E and His respectively.
RESULTS AND DISCUSSION
A 23 kDa eIF4E form is preferentially expressed in the endometrium during implantation
A key result of our present study is the detection of a 23 kDa small-size variant of the mRNA-cap-binding protein eIF4E in the porcine endometrium during early pregnancy at the time of implantation.
Western blot analysis for eIF4E (Figure 1A) in the endometrial cytosol of pigs after OVX, on C13 and on P13, P15 and P30 revealed a single band with a molecular mass of 25 kDa in the endometrium after OVX. The same band was predominantly detected on C13. On P13 and P15, the 25 kDa band disappeared almost completely and a band of approx. 23 kDa emerged. Both bands appeared on P30. A similar pattern was obtained when a phospho-specific antibody was used. The abundance of the truncated eIF4E significantly increased on P15 (Figure 1A).
Expression of eIF4E in the porcine endometrium and various other organs and cells, and analysis of eIF4G, eIF4A, PABP, 4E-BP1 and 4E-BP2 on P15
To demonstrate that the eIF4E cleavage is not an artifact of the extract preparation, samples after OVX and on P15 were homogenized either separately as controls or after mixing both tissues. Clearly, extraction of the mixed samples did not reduce the abundance of the prototypical 25 kDa eIF4E nor did additional cleavage products appear (Figure 1B). This indicates that the eIF4E truncation in P15 samples is not an extraction artifact.
Analysis of various other porcine organs (muscle, heart, kidney, brain, ovary, spleen and liver) and cells (oocytes, blastocysts, fibroblasts and granulosa cells) did not reveal a truncated form of eIF4E (Figure 1C). However, traces of smaller proteins were found in the kidney and liver samples. They differ in size compared with the endometrial 23 kDa eIF4E form and could not be detected with the phospho-specific antibody. We conclude that these proteins are not identical with the endometrial 23 kDa eIF4E (see below) and they were not investigated further. Additionally, both the 23 and 25 kDa eIF4E could be precipitated by m7-GTP–Sepharose (Figure 1D).
Further investigations revealed that, in the endometrial lysates on P15, eIF4G was affected by extensive proteolytic cleavage generating discrete fragments (Figure 1E). However, other factors such as eIF4A and PABP were not cleaved. Additionally, this assay revealed a reduced abundance of 4E-BP1 on P15. Moreover, 4E-BP1 appeared in the α-band, which has been shown previously to be the hypophosphorylated form . In contrast, the abundance of 4E-BP2 was not significantly reduced on P15, but a shift to a higher electrophoretic mobility was also observed. We speculate that 4E-BP2 is dephosphorylated on P15 as well. However, the phosphorylation of this protein has not been as studied extensively, thus we cannot exclude that the observed faster migration of 4E-BP2 could even result from truncation. Nevertheless, these different events, namely eIF4G fragmentation, 4E-BP1 dephosphorylation and partial degradation and 4E-BP2 dephosphorylation/truncation, suggest that these modifications are regulated processes specific during early pregnancy.
We envisage that the generation of the truncated eIF4E is a regulated process as well. eIF4E and eIF4G fragments are stably expressed and not an intermediate product arising through general degradation. No signs of apoptosis or general proteolysis were observed in P15 samples. For example, no difference in active caspase 3 after OVX or on P15 was detected. Moreover, cultures of endometrial luminal epithelial cells from P15 remained viable. However, the 23 kDa eIF4E gradually disappeared in these cells and was replaced by the 25 kDa form after at least 24 h (results not shown).
The truncated eIF4E is a proteolytically processed product of the prototypical 25 kDa eIF4E1 protein
Northern blotting and cDNA sequence analysis showed no indication for the occurrence of an alternatively spliced mRNA or an alternative gene encoding the truncated factor (see Supplementary Figures S1 and S2 at http://www.BiochemJ.org/bj/432/bj4320353add.htm). Therefore we analysed whether the 23 kDa eIF4E was generated from the 25 kDa form by proteolytic cleavage (Figure 2). The in vitro protease assay with a GST–eIF4E fusion protein (51 kDa) as an external substrate revealed that the cell lysate after OVX did not generate any cleaved eIF4E, even when Ca2+ was added. In contrast, the P15 lysate generated a distinct fragment of ~23 kDa (Figure 2A). Here, the amount of the recombinant eIF4E decreased with the duration of incubation, while the truncated form increased. This reflects the sum of the 23 kDa form present in the P15 lysate and the newly processed 23 kDa eIF4E generated from the recombinant external substrate. Additionally, the GST blot showed an ~28 kDa fragment which must be composed of the 26 kDa GST tag linked to an approx. 2 kDa N-terminal eIF4E fragment. The results also revealed that the eIF4E cleavage was strongly inhibited by EDTA which can be revoked by applying additional CaCl2, suggesting the action of a Ca2+-dependent protease. In an additional experiment, the more Ca2+-specific chelator EGTA was used to analyse eIF4E cleavage (Figure 2B). The results showed an identical effect as that obtained with EDTA treatment. Quantification of the GST blot is presented in Figure 2(C).
In vitro protease assay with recombinant GST–eIF4E fusion protein as an external substrate in endometrial lysates after OVX and on P15
In general, maximal cleavage was already reached after 30 min of incubation. The truncated eIF4E remained stable for at least 20 h. No further degradation of the truncated eIF4E was observed, even after that long duration of incubation (results not shown).
Lysates from kidney or liver cells, which also had traces of a fast-migrating band (Figure 1C), could not cleave the GST-tagged eIF4E (results not shown). These results support the notion that these proteins are not identical with the endometrial 23 kDa form of eIF4E.
Generation of discrete fragments through proteolytic cleavage has been reported to affect different initiation factors in distinct physiological situations . For instance, eIF4G, eIF3S1 and eIF4B have been shown to be targets of caspase 3 in the pre-apoptotic phase. Additionally, PABP is degraded during apoptosis . In this case, proteolytic cleavage generally reduces the overall net rate of protein synthesis. In contrast, proteolytic processing of eIF4E has not been described to date. Thus we localized the cleavage site in the eIF4E molecule.
The 23 kDa eIF4E form lacks at most the N-proximal 21 amino acids
It was not possible to enrich large amounts of truncated eIF4E from native samples. Therefore, in an alternative in vitro protease assay, we used a larger amount of the GST–eIF4E fusion protein to obtain Coomassie-Blue-stainable fragments for localization of the cleavage site. Edman degradation was not successful, probably through N-terminal blockage of the truncated eIF4E. We thus performed MS/MS (tandem MS) analysis to identify the cleavage site. A Coomassie-Blue-stained gel is shown in Figure 3(A). The external substrate (band 1) which is overlaid by endometrial proteins in the P15 and OVX lane and two specific bands (bands 2 and 3), which were only found in protease assays from P15 lysates, are marked. The identity of these bands was determined by Western blotting (Figure 3B). Band 2 represents GST probably linked to the N-terminal fragment of eIF4E and band 3 was the truncated eIF4E. These three bands were recovered and subjected to MS analysis. The results of the MALDI–TOF-TOF MS are summarized in Figure 3(C). In the GST–eIF4E fusion protein, the GST cloning vector and seven fragments of eIF4E were identified (line 1). Fragments 1–4 comprise entirely the N-terminus, up to amino acid 61. Band 2 was identified as the GST cloning vector. Unfortunately, an N-proximal eIF4E fragment was not found in this assay (line 2). However, the protein sequence of the truncated eIF4E (line 3) was established and was found to begin with amino acid 22 (see Supplementary Figure S1). This result suggests that the truncated eIF4E lacks the N-proximal 21 amino acids and the proteolytic cleavage site is located between the residues at position Lys21 and Thr22. However, Lys21 is targeted by trypsin in the MS analysis. Therefore we cannot exclude the possibility that the eIF4E-specific protease cleaves two to three amino acids further N-terminally. Hence, so far, it is highly speculative to appoint a protease for eIF4E cleavage. For example, it is not clear whether a specific sequence motif or rather tertiary structural elements are responsible to direct cleavage. Nevertheless, the observed Ca2+-dependency and cytosolic localization of the proteolytic action might argue for a tissue-specific calpain-related protein. However, the identification of the eIF4E-specific protease awaits further investigations.
MS analysis of proteolytically cleaved recombinant GST–eIF4E fusion protein
EB, P4 and embryonic signals stimulate the proteolytic processing of eIF4E
Next, we intended to determine whether embryonic signals and/or maternal steroid hormones are involved in the induction of the proteolytic cleavage of eIF4E. However, we failed to establish an endometrial cell culture system in which formation of the truncated eIF4E could be induced. We thus used an animal model for this purpose. The influence of embryonic signals on the expression of the 23 kDa eIF4E was determined in animals where one uterine horn was endoscopically closed before insemination. On P15, the abundance of the 23 kDa eIF4E was high in the endometrium of the embryo-harbouring uterine horn (Figure 4A). The uterine horn without any embryos only had traces of the truncated factor. This observation suggested that embryonic oestrogens stimulate the expression of the eIF4E-processing protease in the P4-primed endometrium. Therefore we performed steroid-replacement experiments. OVX animals were treated with EB, P4 or with both hormones (Figure 4B). This approach revealed a single band with a molecular mass of 25 kDa in the endometrium after OVX and predominantly also after EB or P4 treatment. However, traces of the truncated eIF4E were found in some of the animals treated with EB or P4, probably reflecting the animal-specific hormonal status. In contrast, the consecutive substitution of both steroid compounds resulted in a strong expression of the truncated eIF4E. Furthermore, only lysates from the animals with the double substitution significantly stimulated the cleavage of the 25 kDa eIF4E when incubated for 90 min. These results are supported by an additional experiment. Only in samples from animals having been substituted with both P4 and EB could the formation of the truncated eIF4E be reduced by EDTA. This reduction was abolished when Ca2+ was added (Figure 4C). All of the other lysates did not significantly stimulate any eIF4E processing even if CaCl2 was added.
Analysis of the influence of embryonic signals and steroid replacements on the expression of the truncated eIF4E
In conclusion, our results indicate that, at the time of implantation when the concentration of P4 is high , the generation of the truncated factor is induced by embryonic signals, probably embryonic 17β-oestradiol.
The truncated eIF4E reduces the repressive function of 4E-BP1 in vitro
We suggest that the truncated eIF4E may play an important physiological role at the time of implantation of the conceptus. Therefore we developed a cell-free translation assay which mimicked the situation in P15 samples to examine the potential regulatory impact of the truncated eIF4E in the translation process.
To this end, an in vitro translation assay with nuclease-treated RRL was performed. The abundance of eIF4G, eIF4A, eIF4E and of the recombinant eIF4E and 4E-BP1 in the complete and eIF4E-depleted RRLs is shown in Figure 5(A). The complete RRL did not contain endogenous 4E-BP1 or 4E-BP2. Furthermore, the abundance of eIF4G and eIF4A was not strongly influenced by our eIF4E-depletion method. The in vitro translation assays performed with these lysates showed that the translation of the capped Xef1 mRNA was not stimulated in the eIF4E-depleted and in the 4E-BP1-substituted RRL (Figure 5B). Therefore translational stimulation was strictly eIF4E-dependent in our assay.
Analysis of modified RRL and the in vitro translation assay
Next, we modified the RRL to achieve conditions comparable with the situation in the endometrium. Therefore RRL was treated with endometrial lysates from animals after OVX as a control and on P15 to obtain cleaved eIF4E and eIF4G factors. At zero time, truncated eIF4E or eIF4G were absent or only detected in traces (Figure 5C). Thus the RRL was not heavily contaminated with endometrial factors. As expected, after 60 min of treatment only truncated forms of eIF4E and eIF4G were observed in the P15-treated RRLs (Figure 5C). Interestingly, eIF4G was cleaved into several fragments with molecular masses from approx. 100 to 160 kDa. However, eIF4A and PABP were not affected. OVX treatment did not modify any of these factors and OVX-treated RRL stimulated translation comparable with the control with the complete unmodified RRL (Figure 5D compared with Figure 5B). A weaker stimulation was observed in the treated RRL from P15. In contrast, OVX treatment with 4E-BP1 substitution did not stimulate translation at all. This inhibitory effect of 4E-BP1 was slightly, but noticeably, reduced in the RRL treated with endometrial lysate from P15. The quantification of these results (Figure 5E) shows the eIF4E-dependency of translation (left-hand panel), the inhibitory effect of 4E-BP1 and the partial relief of 4E-BP1 inhibition in the P15-treated RRL (right-hand panel).
We did not, however, observe a complete repression of translation in P15-treated RRL. Only a reduction (~40%) of translation efficiency compared with the OVX-treated samples was observed. This is in accordance with results showing that only RRLs containing endogenous competitor mRNA or which were partially depleted of ribosomes show a cap-poly(A)-mediated synergetic effect, whereas in other cases this effect was only additive .
Truncated eIF4E and eIF4G forms alter the dynamic of eIF4E/4E-BP1 binding and impair eIF4G–PABP interaction
To analyse the formation of the eIF4F complex and PABP binding in the cell-free translation assay, immunoprecipitations with an immobilized anti-eIF4E antibody were performed directly after the in vitro translation reaction. In this experiment, we compared the input and the precipitates with the corresponding supernatant in OVX, P15 and 4E-BP1-substituted RRLs (Figure 6A). This approach confirmed the stable abundance of the truncated eIF4E and eIF4G in the P15-treated RRL. In the OVX-treated control, eIF4F was distinctly formed, as shown by the fact that eIF4G and eIF4A was highly present in the precipitates and only traces remained in the supernatant. Moreover, PABP bound to the complex. As expected, in the 4E-BP1-substituted OVX-treated RRL, the formation of eIF4G and PABP binding was significantly reduced. A different observation was made in the P15-treated RRL. Here, eIF4F was formed, but PABP did not bind to the complex. This indicates that the truncated eIF4G had lost the N-terminally located PABP-binding site. A different behaviour was also observed in the 4E-BP1-substituted P15 assay. No significantly reduced eIF4F formation was observed in comparison with non-substituted P15 RRL. This points to a restricted 4E-BP1 function. Indeed, we observed reduced binding of 4E-BP1 to the truncated eIF4E. Most of the 4E-BP1 molecules remained in the supernatant. Likewise, PABP binding was impaired in this assay.
Analysis of eIF4F formation and PABP binding in the modified RRLs after in vitro translation and analysis of the eIF4E–4E-BP1 interaction
To analyse the altered eIF4E–4E-BP1 interaction in more detail, we performed a protein interaction assay with recombinant His-tagged 4E-BP1 bound to Ni-NTA beads (Figure 6B). This approach showed that, whereas the 25 kDa form of eIF4E binds to 4E-BP1 (43±2.3%), the binding of the truncated factor is remarkably reduced (14.6±4.3%). As a control, the blot was reprobed with an anti-His antibody to ensure equal 4E-BP1 binding to the Ni-NTA resin (Figure 6B, lower panel).
Regarding the regulatory function of eIF4E in translational initiation, it is known that binding of apo-eIF4E to the eIF4G enhances its affinity to the cap [37–39]. Moreover, cap binding enhances the affinity for eIF4G and 4E-BP1 and 4E-BP2 [38,40,41]. eIF4G and its competitors 4E-BP1 and 4E-BP2 bind to the dorsal surface of eIF4E, near the N-terminal region which is located distal to the cap-binding pocket . The N-terminal segment comprises Gln40 and is a unique flexible domain which is strongly conserved between human, mouse, rat and rabbit . Considering the regulatory impact of the N-proximal segment of eIF4E, interaction analysis with recombinant N-terminally truncated eIF4E (Δ33 residues) revealed that the binding dynamic of eIF4E to 4E-BP1 and 4E-BP2 depends on the interaction state . Kinetic-state SPR (surface plasmon resonance) analysis revealed an approx. 2-fold increase in the apparent association rate constant and a ~20-fold decrease in the dissociation rate constant when the truncated eIF4E was compared with the full-length counterpart. In contrast, in a steady-state fluorescence analysis, the association constant was found to decrease by a factor of four. Hence Tomoo et al.  discuss positive and negative contributions of the N-terminal region in 4E-BP1 and 4E-BP2 binding. They describe the N-terminal region of eIF4E as a repressor for the association/dissociation with 4E-BP1 and 4E-BP2. It stabilizes the complex in an equilibrium state and blocks 4E-BP binding in a non-steady kinetic state. However, more recent SPR analysis by this group  revealed that the m7GTP–Δ16 eIF4E binary complex has more or less the same binding affinity for 4E-BP1 and 4E-BP2 compared with the full-length eIF4E. In contrast, the complex with Δ26 eIF4E had a reduced association rate for 4E-BP1 and 4E-BP2, whereas with Δ33 eIF4E the dissociation rate was reduced. These results indicate that deletions between 16 and 26 amino acid residues reduce the affinity of the m7GTP–eIF4E binary complex for 4E-BP1, which is in accordance with our present observations.
Little information is available comparing the interaction of the full-length factor with truncated forms of eIF4E and eIF4G. Studies on yeast eIF4E, which has an ~30% identity with mammalian eIF4E, showed that the N-terminus of eIF4E is required for tight binding to eIF4G and maintaining a long-term eIF4F complex. Deletion of 20 residues did not influence binding, whereas longer deletions (30–35 residues) significantly reduced eIF4E binding to the eIF4G peptide [26,38].
Regarding the scaffold protein eIF4G, binding sites for components of eIF4F and for PABP have been identified to be located in different domains . In the human eIF4G factor, the PABP- and eIF4E-binding sites are located in the N-terminal segment (amino acids 172–200 in PABP; amino acids 557–646 in eIF4E). The eIF4A-binding sites were found in the central and C-terminal regions (amino acids 712–916 and 1241–1356). We detected the N-terminally truncated forms of eIF4G in the P15-treated RRL. Our results additionally showed that, in contrast with OVX, PABP binding to eIF4G was impaired in RRL treated with P15 endometrial lysates. This indicates a decoupling of the PABP from the eIF4E-binding site. That means the PABP site is not present in the N-terminal truncated eIF4G which can be precipitated with the immobilized eIF4E antibody. The estimated apparent size of the stable C-terminal fragments of eIF4G was 100–160 kDa, but the sequence of the N-terminal segment of rabbit eIF4G is not well established. A database search indicates that the N-terminus of the rabbit factor coincides with position 196 of the human eIF4G factor. However, we suggest that, in RRL, the N-terminal region of eIF4G is quite similar to its human counterpart. For instance, eIF4G in RRL was shown to have an apparent molecular mass of ~220 kDa  and our N-terminal eIF4G antibody maps around Gly188, which is located within the PABP-binding site. Additionally, the eIF4G–PABP interaction has been described in unmodified RRL [36,45].
Truncated eIF4E and eIF4G forms provide an alternative regulation mechanism of translational initiation during implantation
It has been shown that either cap or poly(A) stimulate translational initiation in vivo alone or synergistically . Thus the truncated eIF4E and eIF4G described in our present study allow three different scenarios of regulation of translational initiation in P15 samples. These are: (i) reduced inhibitory impact of 4E-BP1 and 4E-BP2; (ii) impaired eIF4G–PABP interaction; and (iii) altered interaction between the truncated eIF4E and eIF4G.
The different interactions of the factors are summarized in Figure 7. Depicted is a highly schematic representation of eIF4E and eIF4G with the presumed cleavage sites marked by arrows (Figure 7A). Although the investigations were performed with recombinant 4E-BP1, the results might also apply for 4E-BP2 due to its similar binding property [25,41]. Formation of the eIF4F complex and PABP binding is dependent on 4E-BP1 abundance after OVX (Figure 7B), whereas such a pronounced effect of 4E-BP1 was not observed during the interaction of the truncated factors on P15. Moreover, PABP1 binding is impaired at this developmental stage. These different interactions have different impacts on translational initiation. On the one hand a general reduction might be suggested by the impaired ‘closed loop’ formation of the mRNA. On the other hand, initiation could be stimulated by a reduced efficacy of the repressor 4E-BP1 even in situations where this protein is hypophosporylated.
Model for eIF4E function in the endometrium of gilts after OVX and on P15
Given the tight spatial restriction of the occurrence of the truncated eIF4E only at the site of implantation, we propose that the truncated factor contributes to establish the non-invasive placentation which is found in pigs. Such an assumption is supported by the fact that, in particular, the truncated form of eIF4E was not found in ongoing pilot experiments (K. Wollenhaupt and W. Tomek, unpublished work) in species with invasive placentation, such as the mouse, and these endometrial lysates did not cleave exogenous substrates. Furthermore, when porcine embryos are transferred to ectopic sites, the trophoblast becomes invasive [47,48]. This results in erosion of the adjacent epithelium. Therefore the truncation of eIF4E and eIF4G might be crucial for the expression of factors favouring non-invasive over invasive implantation. Thus we envisage that the truncated factors differentially regulate a subset of mRNA moieties, even under conditions of persistent hypophosphorylation of 4E-BP1 and 4E-BP2. Mechanistically, these mRNAs might share peculiar properties in their 5′-UTR and quite possibly their activation might not strictly depend on poly(A) [13,49,50].
We are grateful to Hannelore Klückmann, Eva-Maria Kroh, Gesine Krüger, Petra Reckling and Angelika Deike for expert technical assistant.
- C10 etc.
day 10 of the oestrous cycle etc.
eukaryotic initiation factor
matrix-assisted laser-desorption ionization–time-of-flight-time-of-flight
mitogen-activated protein kinase
- P13 etc.
day 13 of pregnancy etc.
rabbit reticulocyte lysate
surface plasmon resonance
Karin Wollenhaupt and Wolfgang Tomek designed and co-ordinated the study, performed the protein analysis and wrote the manuscript. Klaus-Peter Brüssow performed all of the veterinary treatments. Cell separation and culture were performed by Ute Tiemann and Kati Reinke, and the nucleic acid analysis was conducted by Hans-Martin Seyfert. The MS analysis was performed by Dirk Albrecht. All authors read and proofread the final manuscript.
This work was supported by the Deutsche Forschungsgemeinschaft [grant number WO 663/7-1].