Mutants of eukaryotic initiation factor eIF-4E with altered mRNA cap binding specificity reprogram mRNA selection by ribosomes in Saccharomyces cerevisiae.

Recognition of the 5'-end of eukaryotic mRNA by the ribosomal 43 S preinitiation complex involves the eukaryotic translation initiation factor eIF-4E (eIF-4alpha). Deletion mutants of the eIF-4E gene of Saccharomyces cerevisiae (CDC33) encoded proteins with reduced affinity for the 5'-cap. One of these mutant proteins lacked any detectable binding to a cap analogue binding column, yet was still able to support cell growth. More than 17% of the total eIF-4E amino acid sequence could be removed without fully inactivating this factor. At least 30 of the N-terminal amino acids are not essential for function. The minimal functional eIF-4E protein segment therefore comprises at most 176 amino acids. The translation and growth defects of the deletion mutants could be at least partially compensated by increases in eIF-4E synthesis, possibly due to a mass-action effect on mRNA binding. Electroporation of yeast spheroplasts with in vitro synthesized mRNA allowed us to characterize the ability of eIF-4E mutant strains to distinguish between capped and uncapped mRNAs in vivo. Our data show that the cap specificity of eIF-4E determines to what extent the translational apparatus differentiates between capped and uncapped mRNAs and indicate the minimum relative mRNA (cap) binding activity of eIF-4E required for yeast cell viability.


Recognition of the 5-end of eukaryotic mRNA by the ribosomal 43 S preinitiation complex involves the eukaryotic translation initiation factor eIF-4E (eIF-4␣).
Deletion mutants of the eIF-4E gene of Saccharomyces cerevisiae (CDC33) encoded proteins with reduced affinity for the 5-cap. One of these mutant proteins lacked any detectable binding to a cap analogue binding column, yet was still able to support cell growth. More than 17% of the total eIF-4E amino acid sequence could be removed without fully inactivating this factor. At least 30 of the N-terminal amino acids are not essential for function. The minimal functional eIF-4E protein segment therefore comprises at most 176 amino acids. The translation and growth defects of the deletion mutants could be at least partially compensated by increases in eIF-4E synthesis, possibly due to a mass-action effect on mRNA binding. Electroporation of yeast spheroplasts with in vitro synthesized mRNA allowed us to characterize the ability of eIF-4E mutant strains to distinguish between capped and uncapped mRNAs in vivo. Our data show that the cap specificity of eIF-4E determines to what extent the translational apparatus differentiates between capped and uncapped mRNAs and indicate the minimum relative mRNA (cap) binding activity of eIF-4E required for yeast cell viability.
The eukaryotic initiation factor eIF-4E 1 (or eIF-4␣; see Ref. 1) is an essential component of the eukaryotic translational apparatus. eIF-4E constitutes part of the so-called cap-binding complex eIF-4F (eIF-4), which, in higher eukaryotes, also contains eIF-4A and p220 (eIF-4␥) (2)(3)(4). This complex, together with eIF-4B, is thought to mediate binding of the 43 S preinitiation complex to mRNA (5). In the yeast Saccharomyces cerevisiae, eIF-4E has been shown to form a complex with two proteins, p150 (thought to be the homologue of mammalian p220) and p20. The functions of p20 and p150 are as yet unknown (6 -9). However, it is generally assumed that yeast eIF-4F fulfills the same function(s) as its mammalian counterparts.
The sequencing of eIF-4E genes from several species has revealed strong homology within the group of known mammalian polypeptide sequences and less extensive but clearly evident homology between the yeast eIF-4E sequence and the sequences of the counterpart proteins of mammals and wheat. The mammalian and yeast proteins are immunologically distinct (10). Nevertheless, mouse eIF-4E can substitute for its yeast homologue in vivo (11). Thus, there is also at least partial functional homology between eIF-4E from diverse species. At the same time, the functional role(s) of eIF-4E is not yet clear. The two main lines of evidence for this factor's involvement in yeast translation derive from in vitro experiments. First, addition of a cap analogue (12) or a monoclonal antibody against yeast eIF-4E leads to inhibition of translation in a dose-dependent fashion (10). Second, in a cell-free system derived from a yeast strain carrying a temperature-sensitive form of eIF-4E, the translation defect could be complemented by addition of purified wild-type eIF-4E (13). However, the step(s) of in vivo translation in which eIF-4E is involved and its mechanism of action remain uncharacterized. Moreover, other reports have indicated that this protein might be directly or indirectly involved in processes other than translation initiation. For example, the yeast eIF-4E gene has been identified as the locus of a cell cycle mutation (cdc33) that arrests the mitotic cycle at the "start" stage (14). On the other hand, a fraction of the cellular population of eIF-4E in COS cells (15) and yeast (16) localizes to the nucleus.
Much attention has been paid to the fact that eIF-4E can be phosphorylated in the cell. Many apparent correlations between the level of eIF-4E phosphorylation and the rate of cellular protein synthesis have been reported (17). Clearly, any global regulation of protein synthesis mediated via eIF-4E phosphorylation would have to be based on a specific and precisely controlled mechanism, yet this remains an unresolved issue in mammalian cells. In S. cerevisiae, eIF-4E phosphorylation is unlikely to be regulated by the mechanisms proposed for its mammalian counterparts (18).
A common property of all known eIF-4E proteins is recognition of the mRNA cap structure (19). Interestingly, the estimated binding affinity is at least 4 orders of magnitude lower than that of the most tightly binding RNA-binding proteins for their targets (20), yet the significance of this binding activity for the cytoplasmic (and possibly nuclear) functions of eIF-4E remains unclear. A key question here is how the affinity of eIF-4E for the cap relates to the ability of ribosomes to bind and translate mRNA efficiently. This relationship is central to the potential function of eIF-4E as a regulatory factor, whose activity might be modulated by phosphorylation or via its interaction with other proteins. Furthermore, the availability and/or activity of eIF-4E might not influence the translation of all mRNAs equally, yet it remains uncertain whether specific mRNA populations can be differentially activated via this factor. It has been suggested that eIF-4E is "rate-limiting" for translation initiation on at least some eukaryotic mRNAs because of its low abundance in the cell relative to other initiation * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In this paper, we address the above issues by varying both the abundance and the specific activity of eIF-4E in yeast, making use of a series of deletion mutations that progressively reduce its cap binding affinity. A newly developed experimental strategy has allowed us to characterize how altered interactions of these mutant proteins with mRNA influence the translational apparatus in vivo.

MATERIALS AND METHODS
Strains and Media-The yeast strain used as the basis for the construction of haploid eIF-4E deletion strains was 4-2 (a eIF-4E: (11). The diploid strain (GEX1) used for tetrad analysis after transformation with the various eIF-4E expression plasmids was created by mating strains 4-2 and SL988 (␣ met8-1 leu2-1 his3⌬1 trp1 ura3-52) and subsequent elimination of pMDA101. An additional strain, YPM150A (␣ ura3-52 his3), was used for comparative assessment of the amount of eIF-4E present in haploid cells bearing the chromosomal CDC33 gene. Yeast cells were grown in rich medium (2% peptone, 1% yeast extract) containing either 2% glucose (for mutant eIF-4E genes transcribed from the TRP1 promoter) or 2% galactose (for the GPF promoter constructs; see below). For the growth rate determinations and polysomal gradients, a complete medium containing lactate and 2% galactose was used. Yeast transformation as well as plasmid shuffling and tetrad analysis were performed according to standard procedures (26). The Escherichia coli strain used for the expression of the recombinant proteins was TG2 (supE hsd⌬5 thi ⌬(lac-pro AB) ⌬(srl-recA)306::Tn10 (tet r ) FЈ [traD36 proAB ϩ lacI q lacZ⌬M15]). DNA, RNA, and Protein Methods-DNA cloning and sequencing techniques were performed using standard methods (27). Oligodeoxyribonucleotide-directed in vitro mutagenesis of the eIF-4E gene was performed using the system version 2.1 kit from Amersham Corp. Oligodeoxyribonucleotides were synthesized using Applied Biosystems Model 380B and 394 DNA/RNA synthesizers. All new plasmid constructs were checked by means of restriction analysis and DNA sequencing. Proteins were separated on 13.5 or 15% SDS gels according to Laemmli (28). Gels were stained using either Coomassie Brilliant Blue or silver nitrate. Western blots, performed as described by Towbin et al. (29), were treated with polyclonal rabbit antibodies (against eIF-4E and eIF-2␣) diluted 1:2500 (16) or with a monoclonal mouse antibody (against ribosomal protein L15) diluted 1:500. The antibodies against recombinant yeast eIF-4E and eIF-2␣ were obtained as described previously (16,30). The mouse monoclonal antibody against ribosomal protein L15 was kindly given to us by Prof. Juan-Pedro Ballesta (Centro de Biologia Molecular, Madrid, Spain). After washing with 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.05% Tween 20, and 0.5% bovine serum albumin, membranes were treated with the alkaline phosphatase-conjugated goat anti-rabbit (or anti-mouse) secondary antibody and stained using 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium chloride (Promega). pCYTEXP1 (31) was used both for eIF-4E expression in E. coli and to provide single-stranded DNA template for in vitro mutagenesis. Recombinant eIF-4E purification was performed as described previously (16,32). Expression of the wild-type and mutant eIF-4E genes in yeast was achieved using the yeast single-copy vector YCpSupex (33). Two different promoters were employed in this vector: GPF (the inducible GAL1-PGK1 fusion promoter constructed by Oliveira et al. (33)) and the TRP1 promoter (34), which was cloned into the YCp-Supex vector as a 1.6-kilobase pair HindIII-BamHI fragment in place of the GPF promoter. The LUC expression vectors used were YCpLU-CEX1, which has a relatively unstructured 5Ј-UTR, and the derivatives of this plasmid, B3 and X3 (33). The latter two plasmids encode LUC mRNAs whose leaders bear a stem-loop structure positioned either close to the 5Ј-end (B3) or close to the start codon (X3). Luciferase assays and Northern blots were performed as described previously (33).
Preparation of Yeast Cell-free Extracts and Purification of Cap-binding Proteins-Yeast cells were harvested in mid-exponential phase (A 600 ϭ 0.8), washed twice with cold water, and resuspended in buffer A (20 mM Hepes (pH 7.4), 150 mM KCl, 1 mM EDTA, 2 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride). Cells were disrupted by vortexing in the presence of glass beads (0.45-0.5-mm diameter). Cellular debris was removed by centrifugation at 16,000 ϫ g for 20 min. The extracts were either used directly for Western blots or passed through a 7-methyl-GDP-Sepharose column (Pharmacia Biotech Inc.) in order to isolate the cap-binding proteins. The latter were eluted using buffer A containing 0.1 mM 7-methyl-GDP after washing steps using buffer A alone and buffer A containing 0.1 mM GDP, respectively.
Polysomal Gradients-In a procedure adapted from Sagliocco et al. (35), 100-ml yeast cultures (A 600 ϭ 0.5) were harvested for the preparation of extracts. The extracts were loaded onto 12-ml 15-38% sucrose gradients prepared in 50 mM Tris acetate (pH 7.0), 50 mM NH 4 Cl, 12 mM MgCl 2 , and 1 mM dithiothreitol. High salt conditions were achieved by adding NaCl to a final concentration of 0.7 M. The extracts to be loaded onto the high salt gradients were adjusted to 0.8 M NaCl, thus dissociating nontranslating 80 S ribosomes. Gradients were centrifuged for 14 h at 4°C and 60,000 ϫ g. Fractions were collected from these gradients and treated as described previously (36) before being analyzed by Western blotting.
Electroporation and Expression of in Vitro Synthesized mRNA-mRNA encoding luciferase was synthesized using the in vitro transcription vector LUCEX as described previously (37). The resulting run-off transcripts had A 30 tails and were synthesized either capped, by adding the cap analogue m 7 GpppG, or uncapped. Electroporation of spheroplasts was performed essentially as described previously (38), whereby pulses were applied at 800 V, 400 ohms, and 25 microfarads using a Bio-Rad Gene Pulser. The strains used were haploid derivatives of strain 4-2 (11) that were generated by plasmid shuffling (see above). They had a disrupted chromosomal CDC33 gene and were dependent upon expression of either the wild-type CDC33 gene or the ⌬196 mutant on a single-copy plasmid (see above). Extracts for luciferase assays were prepared after a 5-h recovery period.

Construction and Expression of eIF-4E Deletion Mutants-To
initiate structure-function studies of eIF-4E, we constructed a series of N-and C-terminal deletion mutants. In vitro mutagenesis of the gene inserted into pCYTEXP1 was used to create NdeI (CATATG) sites at various positions downstream of the authentic start codon (7,19,30) (Fig. 1). The DNA between the wild-type start codon and the newly created start codon could in each case be eliminated by cleavage with NdeI and religation. The newly created AUG codons functioned as translation initiation codons in E. coli and S. cerevisiae. To shorten the gene at the C terminus, in vitro mutagenesis was used to introduce two adjacent TAG codons at different positions 5Ј of the authentic termination codon (Fig. 1). The above strategy was used to generate prematurely terminated eIF-4E proteins with stops at positions 183, 200, and 206 in the amino acid sequence. The deletion constructs have been named (using the prefix ⌬) according to the positions of the engineered start and stop codons, respectively. Three N-terminal deletion derivatives were combined with ⌬206, yielding ⌬7/206, ⌬19/206, and ⌬30/206. The construction of ⌬196 followed a different strategy. EcoRI cuts both in the eIF-4E gene (at position 582 of the reading frame nucleotide sequence) and in the expression vector downstream of the wild-type reading frame. Religation yielded a gene encoding the wild-type eIF-4E amino acid sequence up to position 196, extended by two codons arising from the vector sequence.
We tested the viability of strains dependent on these eIF-4E mutants for translation. After transformation of the diploid strain GEX1, tetrad analysis was used to assess the viability of haploid cells lacking the wild-type gene but carrying each deletion mutant (Fig. 2). This analysis revealed that all of the deletion mutants except ⌬183 allowed growth (see Tables I and  II). Tetrad analysis with ⌬196 gave very poor spore survival rates when expressed using the TRP1 promoter (see legend to Fig. 2), but since plasmid shuffling was possible (see below), this is not an indicator that ⌬196 normally cannot support growth. In a parallel strategy, the mutant eIF-4E expression plasmids were transformed into strain 4-2, and the resulting transformants were cured of pMDA101. This produced haploid strains that each contained only a mutant form of cdc33. Each of the mutant genes was expressed at low levels using the TRP1 promoter and at high levels using the GPF promoter.
⌬30/206 allowed growth only when expressed from the GPF promoter. Complementation and growth were poor with ⌬196. Western blotting was used to estimate the relative amounts of mutant eIF-4E proteins synthesized using the two promoters (Fig. 3A). Quantitative analysis revealed that expression from FIG. 1. Expression constructs used in this work. The wild-type or mutant eIF-4E genes were expressed using a CEN4-ARS1 plasmid (A). The promoter was either the GPF1 fusion promoter (33) or the TRP1 promoter (18). The start codon of the eIF-4E gene was part of an NdeI site engineered at the beginning of the reading frame (16). Transcription was terminated by the PGK1 terminator. The cleavage sites of the following restriction enzymes are indicated: HindIII (H), BamHI (B), and NdeI (N). B/G indicates the site of a BamHI/BglII fusion. The eIF-4E reading frame was shortened at the N and C termini (B). New start codons were engineered by introducing NdeI sites at various positions downstream of the authentic initiation codon. Premature stops were in most cases introduced as double termination codons (TAG.TAG) upstream of the wild-type stop codon. In ⌬196, this was achieved by cleavage with EcoRI followed by religation, which in this vector results in extension of the reading frame at position 196 by two extra codons (ϩ2). Shown in C is an amino acid sequence comparison of the N-and C-terminal ends of eIF-4E from man and the yeast S. cerevisiae, showing the regions removed in the deletion mutants listed in B.

FIG. 2. Viability of haploid yeast strains dependent on each of the newly constructed mutant forms of eIF-4E for translation.
Shown is the tetrad analysis of the C-terminal mutants. S. cerevisiae GEX1 was transformed using derivatives of the TRP1 promoter version of the vector shown in Fig. 1. After sporulation, the mutants ⌬206 and ⌬200 allowed complementation of the CDC33::LEU2 disruption, yielding mainly 4:0 viable spores. ⌬183, on the other hand, did not allow complementation, yielding only 2:0 viable spores. ⌬196 allowed complementation at only a very low frequency (in 1 out of 24 tetrads), generally also yielding only 2:0 viable spores. It is therefore shown in parentheses here to indicate that its results in tetrad analysis were almost as poor as those of ⌬183. Moreover, ⌬196 supported considerably more rapid growth after plasmid shuffling when expressed from the GPF promoter than when expressed from the TRP1 promoter (see Fig. 3 and Table II).
the GPF promoter yielded a steady-state level of eIF-4E ϳ16fold greater than that supported by the TRP1 promoter (Fig.  3B). Moreover, the steady-state level of eIF-4E was ϳ2-fold lower using the TRP1 promoter plasmid than was observed in a haploid strain (YPM150A195-4C) bearing the chromosomal eIF-4E gene expressed from its own promoter (Fig. 3A). In further experiments, we compared the abundance of eIF-4E in extracts from the strains YPM150A, 4-2 [pMDA101], and 4-2 [YCpSupexTRP1 CDC33] (i.e. the wild-type plasmid described in Fig. 1) (data not shown). The plasmid-borne cdc33 genes supported a steady-state level of eIF-4E that was ϳ50% less than that of the strain with a chromosomal CDC33 gene and also apparently less than that of strain 4-2. This is consistent with an earlier comparison between strains T93C and 4-2 (13).
Cap Binding Behavior of Mutant eIF-4E Proteins-Cell-free extracts were prepared from the yeast mutant strains and used for the isolation of cap complex proteins. The cap binding affinity of eIF-4E was progressively reduced by deletions of increasing length (Fig. 4A), so that this dropped below a detectable level with C-terminal deletions equal to or greater than ⌬196 (Table I). These results indicated that even extreme reductions in cap analogue binding affinity did not prevent eIF-4E from functioning in vivo (Table I). In experiments where eIF-4E is isolated from yeast cells, it remains associated with the other cap complex proteins (p20 and p150) during purification using the cap analogue column. To examine the effects of the deletions on the binding of eIF-4E in the absence of these other proteins, we overexpressed some of the mutants in E. coli and recovered (and renatured) the recombinant initiation factor from the resulting cellular inclusion bodies. We found that the binding behavior of the recombinant mutant proteins was indistinguishable from that of the equivalent mutant proteins from yeast (Fig. 4, A and B). Thus, the data presented in Table  I apply to the behavior of mutant eIF-4E proteins whether they are present in the yeast cap-binding complex or in a free form.
Functional Characterization of eIF-4E Mutants in Vivo-Measurements of the growth rates of the various strains revealed extended doubling times for the mutant genes expressed from the relatively weak TRP1 promoter (Table II). However, use of the GPF promoter resulted in dosage compensation, thus eliminating or reducing the severity of the slow growth phenotype(s). Thus, yeast can grow almost normally even with a mutant eIF-4E protein that is seriously defective in cap binding provided this protein is overproduced in the cell.
In an additional approach to analysis of the eIF-4E mutants, we used electroporation to introduce in vitro transcribed mRNA into yeast spheroplasts (Fig. 5). This method allowed us to compare the translation of capped and uncapped mRNAs in vivo. The results clearly demonstrate that the deletion mutant ⌬196, when expressed at a level comparable to that of eIF-4E in wild-type yeast cells, confers reduced selectivity upon the translational apparatus with respect to capped mRNA. The uncapped mRNA is able to compete more effectively as a translational template in the strain containing the mutated eIF-4E. The luciferase activity encoded by a capped mRNA in a strain containing wild-type eIF-4E was up to 5 times greater than that encoded by an uncapped mRNA, whereby the ratio averaged over the four different amounts of mRNA used (Fig. 5A) in The amount of protein bound to the cap analogue column relative to that obtained with the wild-type protein. The experiments were performed using recombinant proteins from E. coli (F) or both the E. coli-derived proteins and extracts from the appropriate yeast strains (j). Four symbols (e.g. jjjj) ϭ the wild-type binding affinity (100%); three symbols (e.g. FFF) ϭ 75%; two symbols ϭ 50%; one symbol ϭ 25%; Ϫ ϭ not detectable. b Complementation as determined by tetrad analysis and plasmid shuffling experiments.

FIG. 3. Relative levels of expression of eIF-4E from the TRP1 and GPF promoters.
Extracts from transformed haploid yeast strains containing only the indicated form of the eIF-4E gene on the single-copy expression plasmid (Fig. 1) were subjected to SDS-15% polyacrylamide gel electrophoresis and Western blotting. The samples were loaded in pairs, each pair comprising extracts from strains bearing one of the types of eIF-4E reading frames: wild-type (wt) or one of the deletion mutants, under the control of either the GPF1 promoter or the TRP1 promoter (A). Equal amounts of each cell extract within each pair were loaded, permitting direct comparison of the relative amount produced by the respective promoters. The GPF promoter reproducibly supported the maintenance of an amount of eIF-4E greatly exceeding that found in cells using the TRP1 promoter. Serial dilution of extracts from cells expressing eIF-4E from the GPF promoter allowed us to estimate the presence of an ϳ16-fold greater amount of eIF-4E in comparison with the TRP1 promoter. An example of such a comparison is shown in B.
This second Western blot also shows the relationship between the relative level of eIF-4E in cells producing eIF-4E from the chromosomal wild-type promoter (P 4E -4E; in the strain YPM150A) and that maintained by the TRP1 promoter. 16 independent experiments was 4.3. This ratio was never more than 2.0 in a strain containing ⌬196 instead of wild-type eIF-4E, whereby the average for the equivalent set of experiments was 1.9. These results show that eIF-4E has a major influence on the ability of ribosomes in vivo to select mRNA on the basis of the modification status of its 5Ј-end. However, the distinction between capped and uncapped mRNAs is also markedly reduced when wild-type CDC33 is overexpressed (Fig. 5B).
The presence of a defective eIF-4E protein resulted in changes in the profiles obtained from polysomal gradient analysis. Thus, as clearly evident in the case of ⌬196, the relative proportion of ribosomes in the 80 S peak increased markedly (Fig. 6). Salt treatment resulted in the dissociation of most of the 80 S ribosomes into 40 S and 60 S subunits, thus indicating that a large proportion of the 80 S ribosome population comprised nontranslating 80 S "couples" (compare with Ref. 36). Examination of the influence of the promoter used on the relative sizes of the 80 S peaks reveals that there is again an eIF-4E dosage effect. The increased abundance of the ⌬196 eIF-4E mutant synthesized from the GPF promoter compen-sated partially for its reduced specific activity, allowing more ribosomes to be translationally active. A further unusual aspect of the polysomal profiles was the enlarged 40 S/43 S peak, which was particularly exaggerated in extracts from cells containing GPF promoter constructs. One possible explanation of this phenomenon might be the release of non-scanning 40 S subunits that are normally involved in complexes (e.g. posttermination complexes) with the mRNA.
Both wild-type eIF-4E and the mutant derivatives were present in all fractions of the gradient (Fig. 6). This contrasted with the distribution of eIF-2, which was found mainly in the 40 S/43 S fractions, with only low levels in the polysomal fractions. This is consistent with eIF-2 dissociating upon formation of 80 S monosomes and being present in much reduced amounts in the polysomes. The results obtained with the eIF-4E antibody indicate that this factor remains bound to cytoplasmic mRNA at least during translation initiation and elongation. Moreover, the greatly reduced cap binding activity of ⌬196 does not prevent this truncated form of eIF-4E from interacting with mRNA in vivo.
Rate Control on Translation of mRNAs with Structured and Unstructured Leaders-The availability of the expression constructs and mutants described in this paper provided us with a unique opportunity to assess the influence of changes in eIF-4E activity on the translational efficiency of specific mRNAs. It has been proposed that eIF-4E is involved in the "unwinding" of leader structure during the scanning process (2). According to this model, variations in eIF-4E activity are expected to be especially critical for the translation of mRNAs containing stable structure in their 5Ј-UTRs (23). We therefore investigated whether the eIF-4E deletion mutations preferentially attenuate the translation of a reporter mRNA whose leader contains stable inhibitory secondary structure. We compared the translational efficiencies of two luciferase mRNAs bearing a stable stem-loop (see "Materials and Methods" and Ref. 16) with that of an mRNA with a relatively unstructured control leader. The measurements were made in strains carrying either a wild-type or mutant eIF-4E gene expressed at a high (GPF promoter) or low (TRP1 promoter) level (Fig. 7). Northern blot analysis showed that the steady-state luciferase mRNA level was ϳ50% higher when a stem-loop structure was present in the leader (Fig. 8). This means that the relative reduction in translational activity caused by the introduction of stem-loop structures into the 5Ј-UTR is even greater than is immediately obvious from Fig. 7. However, there was no variation in the amount of luciferase mRNA relative to the abundance of cellular PGK1 mRNA associated with the substitution of wild-type eIF-4E by the truncated forms of this factor. Thus, for any one LUC construct, the luciferase activities indicated in Fig. 7 reflect the relative translational efficiencies of the respective mRNAs. According to these results, partial inactivation of eIF-4E does not restrict the translation of the mRNAs with leaders bearing stem-loops more effectively than translation of the control mRNA. DISCUSSION In this work, we have generated and analyzed a series of deletion mutants in CDC33 that encode proteins with reduced affinity for the 5Ј-cap of mRNA. These mutants have proved to be useful tools in investigations of the in vivo function of eIF-4E, which we have studied using a combination of in vivo and in vitro methods.
First, the binding characteristics of eIF-4E have been shown to determine the selectivity of the cellular translational apparatus with respect to capped and uncapped mRNAs in vivo. The competitiveness of uncapped mRNA as a template for translation can be greatly increased as a result of mutations in eIF-4E FIG. 4. Elution of wild-type and mutant eIF-4E proteins from a 7-methyl-GDP-Sepharose column. eIF-4E could be isolated either as part of the yeast cap-binding complex (A) or from E. coli inclusion bodies (B). The silver-stained gel (A) shows the two main fractions (F1 and F2) specifically eluted using 7-methyl-GDP after the column had been washed using buffer A and GDP (see "Materials and Methods"). The amount of protein eluted in this way provides a measure of the relative cap binding affinity of each deletion mutant. Three examples are shown, illustrating the large reductions in affinity caused by the described deletions. ⌬196 (not shown here) was not detectably bound in this assay. Similar elution behavior was seen using inclusion body preparations from E. coli strains expressing the respective mutant eIF-4E genes from pCYTEXP1 (B). The latter two Coomassie Bluestained gels show fractions from all stages of affinity chromatography on 7-methyl-GDP-Sepharose. The combination of the N-terminal deletion ⌬7 with the C-terminal deletion ⌬206 resulted in a further reduction in cap binding affinity over either of the deletions alone. wt, wild-type. that reduce this factor's specificity for the cap. The electroporation method we have used here allows mRNAs to be introduced into the cytoplasm of yeast independently of the normal pathway of (pre-)mRNA synthesis, processing, and transport. Expression of the in vitro synthesized mRNAs reflects primarily the relative rates of their translation. Apart from clearly demonstrating the role of eIF-4E in mediating the interaction between ribosomes and mRNA in vivo, our data show how regulatory changes in eIF-4E binding activity can be expected to influence the global pattern of translation in the cell. For example, potential modulation of the binding activity of eIF-4E due to phosphorylation or via the influence of regulatory proteins would also be expected to change the pattern of mRNA recognition in an analogous fashion. A further striking aspect of these data is that overexpression of eIF-4E (via the GPF promoter) reduces the difference in translation rates between capped and uncapped mRNAs. Saturation of the recognition process is achieved for uncapped mRNA only at higher intracellular activities of eIF-4E. Under normal conditions in the cell, the activity of eIF-4E is set at a level that is optimal for the translation of capped mRNA, but greatly suboptimal for the translation of uncapped mRNA.
Second, while the reductions in cap binding affinity affected the selectivity of ribosomes for capped mRNA, we observed no differential effects related to structure in the mRNA leader. This therefore extends our previous study of the influence of excessive eIF-4E activity in the cell (16) to the region of activity significantly below the normal level. Overall, these data provide no evidence that the activity of eIF-4E is used by the yeast cell to achieve differential regulation of gene expression on the basis of mRNA structure. Moreover, previous investigations yielded no evidence that increases in eIF-4E activity above the wild-type level (however these may be achieved) will either enhance the general translational activity of the cell or differentially increase translation of mRNAs whose translation is restricted by secondary structure in the 5Ј-UTR (Ref. 16 and this paper). Therefore, the yeast translational apparatus does not seem to respond to variations in eIF-4E abundance or activity in the same way as proposed for higher eukaryotic cells (24,25). However, we cannot rule out at this stage that some natural yeast mRNAs respond to changes in eIF-4E activity as a result of properties of the 5Ј-UTR unrelated to the type of secondary structure used here. On the other hand, it should be stressed that the physiological mechanism of the transforming effects caused by overexpression of eIF-4E in higher eukaryotes has yet to be determined. More information about the role of eIF-4E in the control of higher eukaryotic translation could be expected from detailed studies of the effects of progressive reductions in eIF-4E activity in appropriate cell lines. De Benedetti et al. (39) have already demonstrated that the synthesis of antisense RNA directed against eIF-4E mRNA in HeLa cells results in greatly reduced levels of both eIF-4E and p220, strong inhibition of protein synthesis, and ultimately in cell death.
Third, the data provide an indication of the minimal eIF-4E cap affinity and/or activity necessary for cell viability in yeast. The mutant ⌬196 has only a minimal binding preference for capped, as opposed to uncapped, mRNA, and the mutant protein allows only drastically reduced absolute translation rates. Yet cells still survive even with the low level of this protein obtained using the TRP1 promoter. Strikingly, however, the overall rate of translation, but not the ability to differentiate between capped and uncapped mRNAs, is enhanced upon overexpression of the ⌬196 form. One explanation of the dosage compensation of poor binding achieved by increased expression may lie in the mass-action principle. Increasing the cellular concentration of a less tightly binding mutant of eIF-4E may force more of it into an mRNA-bound form.
Fourth, comparison of the amounts of wild-type eIF-4E present in cells in which the gene is transcribed from the wild-type promoter and from the TRP1 promoter (Fig. 3) indicates that in wild-type yeast cells, this factor does not restrict general gene expression by virtue of its limited availability to the translational apparatus, at least under the chosen laboratory growth Typical data are shown, each point representing the average of two parallel measurements. Capped (q) or uncapped (E) polyadenylated mRNA was introduced in various amounts into spheroplasts of a strain with wild-type CDC33 expressed from the relatively weak TRP1 promoter (A). The same transcripts were also introduced into spheroplasts of a strain with the ⌬196 gene expressed from the TRP1 promoter (capped LUC mRNA (f) and uncapped LUC mRNA (Ⅺ)). The second set of results (B) were obtained with spheroplasts derived from strains in which either CDC33 (capped (q) and uncapped (E)) or ⌬196 (capped (f) and uncapped (Ⅺ)) was expressed from the relatively strong GPF fusion promoter. The amounts of mRNA added were all below the level of saturation for the translational apparatus.
conditions. A reduction in the cellular abundance of eIF-4E by a factor of ϳ2 does not limit the growth rate of the host cell (compare Fig. 3 and Table II). It is therefore likely that the cellular concentration of eIF-4E is normally above that which would have a measurably restrictive effect. This observation helps explain why the overproduction of eIF-4E has little effect on cell growth. Only at very high levels is a (negative) effect observed (16), possibly due to the movement of increased amounts of eIF-4E into the nucleus (16), the sequestering of proteins that bind eIF-4E, or other nonspecific effects. However, the fact that eIF-4E levels are not normally restrictive for translation does not rule out that translation can be regulated via modulation of eIF-4E activity. It simply means that greater changes in eIF-4E activity are required in order to achieve a given regulatory effect than would be the case if eIF-4E activity were strongly rate-controlling. Finally, a corollary of the above is that high cap specificity and cap binding activity per se are not necessary for eIF-4E function.
Fifth, large deletions in yeast eIF-4E do not inactivate this protein, and the described mutants help define the minimal functional protein structure. We have shown that up to 30 of the N-terminal amino acids, or up to 17 C-terminal amino acids, can be eliminated without giving rise to a lethal phenotype. In those mutants where amino acids were removed from both ends, deletion of a total of 37 amino acids was tolerated. FIG. 7. Effects of eIF-4E deletions on protein synthesis directed by mRNAs whose translation is restricted by secondary structure in their leaders. The LUC gene was expressed from the GPF promoter using the YCpLUCEX plasmid (33). Three different leaders were used: a control leader and the same leader with a stemloop structure inserted either close to the mRNA 5Ј-end (B3) or close to the start codon (X3). The respective vectors were expressed in a double transformant of strain 4-2 together with an expression plasmid bearing either the wild-type (WT) eIF-4E gene (CDC33) or one of the two deletion mutants ⌬19/206 and ⌬196. The individual values were normalized to the luciferase activity observed using the wild-type eIF-4E gene and the LUC control leader. Plotted here are averages of measurements on at least three independently prepared extracts. The error bars indicate the magnitudes of the respective standard deviations.
This resulted in the generation of far fewer 80 S couples (compare F and D). All strains were grown with galactose as carbon source.
FIG. 6. Polysomal gradient analysis of extracts from yeast cells whose growth is dependent on truncated forms of eIF-4E. The absorbance profile at 260 nm after sucrose gradient centrifugation indicates the sizes and positions of the polysome components. In cells dependent on ⌬196 eIF-4E expressed from the TRP1 promoter (C), the 80 S peak is greatly exaggerated compared with the profile seen with wild-type (WT) eIF-4E (A). Addition of salt leads to dissociation of many of the 80 S ribosomes (compare B and D), indicating that the majority are nontranslating 80 S couples. Western blot analysis was performed on the unusual polysomal gradient peak patterns obtained with ⌬196. Typical results are shown in C and D. They reveal that eIF-4E is distributed throughout the gradient, whereas eIF-2␣ and ribosomal protein L15 show more localized distributions, acting as markers for 43 S preinitiation complexes and 60 S ribosomal subunits, respectively. Overexpression of ⌬196 resulted in a profile more typical of cells with a comparatively small amount of wild-type eIF-4E (compare E and A).
Evidently, more amino acids can be removed from the N terminus without inactivating the protein than from the C terminus. The deleted sequences belong to those regions of yeast eIF-4E showing no significant amino acid sequence identity to the equivalent regions of the higher eukaryotic eIF-4E proteins (Fig. 1). None of the deletions eliminated any of the eight tryptophan residues that are generally conserved in eIF-4E amino acid sequences from various sources and that are suspected to be significant for the recognition of the cap structure (40,41). Mutation of either the first (Trp-43) or the last (Trp-166) of these to phenylalanine was found previously to abolish cap binding activity (41). Given that the mutation of especially these two respectively N-terminal and C-terminal proximal tryptophans has the most drastic effect on cap binding (41), it may be possible to delete further sections of the interior region of the eIF-4E sequence without fully eliminating activity. The limits to this type of mutation will be at least partially dictated by the necessity of individual residues for the maintenance of the active conformation and/or stability of the protein. In another study, we have shown that the N-terminal sequence also contains two phosphorylation sites of yeast eIF-4E, both of which are recognized by casein kinase II (18). An important further issue to be considered here is that cap binding is evidently not the only property of eIF-4E that is relevant to cell viability. This is illustrated by comparison of, for example, ⌬30/206 and ⌬196. ⌬196 binds more weakly to the cap analogue column, yet allows growth at the low level of eIF-4E supported by the TRP1 promoter. In contrast, ⌬30/206 shows measurable cap binding activity, yet cannot support growth when expressed from the TRP1 promoter. One possible explanation is that the N-terminal deletion affects the ability of eIF-4E to interact with the ribosome and/or p20 or p150/p130.
In conclusion, we have obtained in vivo evidence that eIF-4E normally mediates the selectivity of yeast ribosomes for capped mRNAs. However, cells can survive with a greatly reduced preference for the capped state, accompanied by an overall reduction in the ability of eIF-4E to promote functional ribosomal binding, provided the selectivity does not drop to a point where uncapped mRNA becomes significantly competitive. The absolute priority for capped mRNA is clearly not essential for cell viability. Overexpression of CDC33 mutants apparently allows the cell to partially compensate for the defective ribosome-mRNA binding pathway. The simplest rationalization of these observations is that eIF-4E directly mediates 40 S ribosomal interactions with the mRNA, whereby the factor itself provides the specificity of cap-dependent binding. However, this does not rule out that eIF-4E can act indirectly, for example, by promoting the localization of (capped) mRNAs to a compartment where ribosomes can bind, by mediating the action of other proteins, or via another step preceding or subsequent to the actual ribosome-mRNA binding interaction.