Functional characterization of five eIF4E isoforms in Caenorhabditis elegans.

Recognition of the 5'-cap structure of mRNA by eIF4E is a critical step in the recruitment of most mRNAs to the ribosome. In Caenorhabditis elegans, approximately 70% of mRNAs contain an unusual 2,2,7-trimethylguanosine cap structure as a result of trans-splicing onto the 5' end of the pre-mRNA. The characterization of three eIF4E isoforms in C. elegans (IFE-1, IFE-2, and IFE-3) was reported previously. The present study describes two more eIF4E isoforms expressed in C. elegans, IFE-4 and IFE-5. We analyzed the requirement of each isoform for viability by RNA interference. IFE-3, the most closely related to mammalian eIF4E-1, binds only 7-methylguanosine caps and is essential for viability. In contrast, three closely related isoforms (IFE-1, IFE-2, and IFE-5) bind 2,2, 7-trimethylguanosine caps and are partially redundant, but at least one functional isoform is required for viability. IFE-4, which binds only 7-methylguanosine caps, is most closely related to an unusual eIF4E isoform found in plants (nCBP) and mammals (4E-HP) and is not essential for viability in any combination of IFE knockout. ife-2, ife-3, ife-4, and ife-5 mRNAs are themselves trans-spliced to SL1 spliced leaders. ife-1 mRNA is trans-spliced to an SL2 leader, indicating that its gene resides in a downstream position of an operon.

Eukaryotic mRNAs and small nuclear RNAs synthesized by RNA polymerase II are posttranscriptionally modified to form a 5Ј-5Ј GpppN linkage (1). The 5Ј-terminal G is methylated at N7 while still in the nucleus to yield an MMG 1 cap. The cap of small nuclear RNAs is then further methylated at N2 in the cytoplasm to yield a TMG cap (2). Methylation of small nuclear RNAs is dependent upon the binding of Sm proteins to form small nuclear ribonucleoproteins. Formation of the TMG cap is the targeting signal for import of small nuclear ribonucleoproteins back into the nucleus to take part in pre-mRNA splicing (3,4). mRNAs, on the other hand, which possess only the MMG cap, remain in the cytoplasm.
In some primative eukaryotes, including Caenorhabditis elegans, mRNAs acquire a TMG cap through the process of trans-splicing (5). Primary transcripts from approximately 70% of protein-coding genes are trans-spliced to 22-nt SL sequences, such that the original MMG caps are replaced with the TMG caps from the SL small nuclear RNAs (6,7). Also common in C. elegans is the organization of genes into operons that are transcribed from a single promotor into a polycistronic RNA (8). trans-Splicing results in the processing of these primary transcripts into monocistronic mRNAs. Generally, the mRNA from the first cistron is trans-spliced to SL1,whereas mRNAs from downstream cistrons are trans-spliced to SL2 or SL2 variants (8). mRNAs that are not trans-spliced retain the original MMG cap. Thus, both MMG-and TMG-capped mRNAs are found in the cytoplasm of C. elegans. Both types of mRNA enter polyribosomes and are translated, indicating that they are competent to interact with the translational machinery (9).
The recruitment of mRNAs to ribosomes is catalyzed by the eIF4 group of translation initiation factors (reviewed in Refs. 10 and 11). The mRNA cap is specifically recognized by eIF4E. At least two isoforms of eIF4E exist in humans, eIF4E-1 (12) and 4E-HP (13), which are quite divergent in primary sequence. In plants, three isoforms have been described, eIF4E (14,15), eIF(iso)4E (16), and nCBP (17). All eIF4E proteins characterized from higher eukaryotes are highly selective for MMG caps. Cap analogs containing TMG are 17-fold less inhibitory than the corresponding MMG-containing cap analogs in a rabbit reticulocyte cell-free translation system (18). Substitution of a TMG cap for the MMG cap on ␤-globin mRNA reduces its translational efficiency by 75% in the same system (19).
Because naturally occurring TMG-capped mRNAs are abundant in C. elegans, the organism must possess a mechanism to recognize these mRNAs and initiate their translation. We previously reported the characterization of MMG-and TMG-binding proteins from C. elegans and the cloning of three cDNAs (ife-1, ife-2 and ife-3) encoding eIF4E isoforms (20). Among these isoforms, IFE-3 has a cap specificity similar to mammalian eIF4E-1, binding only MMG. IFE-1 and IFE-2, on the other hand, bind either MMG or TMG, although apparently with different affinities. All three isoforms can be purified from worm extracts. We now report the cloning of two additional eIF4E cDNAs from C. elegans, ife-4 and ife-5, as well as the characterization of their encoded proteins and the requirement for each of the five proteins for viability.
Sequence Analysis of C. elegans eIF4E cDNAs-The predicted protein sequences of IFE-4 and IFE-5 were identified on cosmid C05D9 and yeast artificial chromosome Y57A10, respectively, from the genomic sequences generated by the C. elegans Genome Sequencing Consortium (21). The TBLASTN algorithm (22), run on the Washington University Genome Sequencing Center server, was used to identify sequences that encode proteins with homology to human eIF4E-1 and 4E-HP. C. elegans EST sequences were obtained through the EST data base BLAST server at the DNA Data Bank of Japan. Amino acid sequences of IFE-1 through IFE-5 were aligned using the PILEUP algorithm in the Wisconsin Software Package (Genetics Computer Group, Madison, WI). Multiple sequence alignments were used to calculate protein distance values with PHYLIP software on the Institut Pasteur or the National University of Singapore server. These values were used to generate a relational tree with TreeView software (version 1.5.2) from the University of Glasgow server (23). The percentage of identity was determined from pairwise alignments using the GAP algorithm in the Wisconsin Software Package. DNA was sequenced by the DNA Sequencing Facility at Iowa State University. cDNA sequences were submitted electronically to GenBank TM .
Cloning of C. elegans ife-4 and ife-5 cDNAs by RT-PCR-Total RNA was isolated from wild-type (N2 strain) C. elegans by either the LiCl precipitation method (24) or the guanidinium thiocyanate method (25). The protein-coding sequences of ife-4 and ife-5 cDNAs were amplified from 1 g of N2 total RNA by RT-PCR using primers 4.1 and 4.2 for ife-4 and primers 5.1 and 5.2 for ife-5 (Table I). RNA was reverse transcribed with rTth polymerase (Perkin-Elmer) at 61°C for 15 min. DNA was amplified with the same polymerase by 45 cycles of the following regimen: 95°C for 15 s, 61°C for 30 s, 72°C for 1 min, and a final extension at 72°C for 7 min. The resulting products were digested with NdeI and XhoI and subcloned into pET21b (Novagen) to yield pETife4 and pE-Tife5. Inserts were removed from these plasmids with XbaI and XhoI and inserted between the same sites in pBluescript SK (Stratagene) to form pSKife4 and pSKife5. pSKife1, pSKife2, and pSKife3 were excised as phagemids from EST clones yk364a1, yk452e8, and yk81f11 in Zap (generously supplied by Yuji Kohara, National Institute of Genetics, Mishima, Japan) using the Rapid Excision Kit (Stratagene). cDNAs for ife-1, ife-2, and ife-3 were amplified from pSKife1, pSKife2, and pSKife3 with primers 1.1, 2.1, and 3.1, respectively, together with primer T7. The PCR products for ife-1 and ife-2 were digested with NdeI and XhoI and subcloned into plasmid pET21a (Novagen) to form pETife1 and pETife2, respectively. The product for ife-3 was digested with NcoI and XhoI and subcloned into pET21d (Novagen) to form pETife3. pET constructs were used to express IFE proteins (except IFE-2) in Escherichia coli, and pSK constructs were used for in vitro transcription (see below).
Cap Binding Specificity Assay-Purified recombinant IFE-4 and IFE-5 (25 g) were diluted into 0.1 ml of buffer A (50 mM HEPES-KOH, pH 7.6, 50 mM KCl, 2 mM EDTA, 1 mM dithiothreitol, 5% (v/v) glycerol) and applied to a 0.1-ml m 7 GTP-Sepharose or a m 3 2,2,7 GTP-Sepharose column. The columns were washed with 0.2 ml of buffer A, and the proteins were eluted with 0.1 ml of 100 M m 7 GTP or m 3 2,2,7 GTP in buffer A. Aliquots (1/25 th volume) of the starting dilution and subsequent fractions were analyzed by SDS-PAGE.
Immunological Procedures-Peptides of the sequences CLRDQSS-YRHTTKNI and CALKYKFSLKSIV were synthesized by Biosynthesis (Dallas, TX) and used to generate isoform-specific antibodies for IFE-4 and IFE-5, respectively. The NH 2 -terminal Cys residue was not present in the sequence of IFE-5. Preparation of anti-peptide antibodies and immunoblotting were performed as described previously (27). Antibodies against IFE-4 were purified on columns of Affi-Gel 501 (Bio-Rad) to which the immunogenic peptide was linked via the Cys residue (28). Serum against IFE-5 was used directly at a dilution of 1:500. Affinitypurified antibodies against IFE-1, IFE-2 and IFE-3 were previously described (20).
Spliced Leader Assay-Poly(A) ϩ RNA was purified from C. elegans total RNA by a single round of oligo(dT)-cellulose chromatography (29). N2 poly(A) ϩ RNA (40 ng) was reverse transcribed with rTth polymerase (Perkin-Elmer) at 60°C for 20 min with specific ife primers (primers 1.2-5.2 for ife-1 through ife-5, respectively). Either primer SL1 or SL2, corresponding to the spliced leaders SL1 or SL2 (30), was added, and DNA was amplified for 35 cycles with the same polymerase according to the following regimen: 95°C for 1 min, 59°C for 1 min, 72°C for 1 min, and a final extension of 72°C for 7 min. Products from the primary RT-PCR reaction were purified by phenol extraction and ethanol precipitation and then resuspended in 15 l of buffer TE (10 mM Tris-HCl, pH 8.0, 1 mM EDTA). Secondary PCR was performed for 20 cycles with 0.5 l of the purified products using primers SL1 or SL2 and the same (ife-2, primer 2.2) or a nested (ife-5, primer 5.3) ife-specific primer and Taq polymerase (31). Products obtained from the primary RT-PCR reactions for ife-1 and ife-4 mRNAs, as well as the secondary PCR reaction for ife-5 mRNA, were subcloned into pGEM-T (Promega) to

Five eIF4E Isoforms in C. elegans
produce plasmids p2a (SL2-ife-1), p41a (SL1-ife-4), p51xc (SL1-ife-5) and p52xc (SL2-ife-5). The 5Ј end of each insert was sequenced (Ͼ200 bp). RNAi Assays-General procedures for handling and maintaining the C. elegans N2 strain have been described (32). Double-stranded RNAs were prepared by transcribing individual ife cDNAs in plasmids pSKife1 through pSKife5 with T3 and T7 RNA polymerases (Epicenter Technologies, Madison, WI). The sense and antisense strands were mixed together at 100 ng/l each and injected into either the intestine or gonads of young adult hermaphrodites (33,34). 6 -8 h after injection, each animal was transferred onto a fresh plate. Eggs were counted at 24-h intervals for 3-4 days and monitored for embryonic lethality. For every set of RNAi injections, a 200-ng/l solution of double-stranded RNA for the 64-kDa subunit of Cleavage Stimulatory Factor, which produces 100% embryonic lethality (Williams and Blumenthal, unpublished), was injected as a positive control. The negative control was buffer TE.

RESULTS
Cloning of ife-4 and ife-5 cDNAs-Homology searches using the human eIF4E-1 (12) and 4E-HP (13) sequences revealed five genes encoding C. elegans eIF4E isoforms. These were termed ife-1, ife-2, etc., and their encoded proteins, IFE-1, IFE-2, etc. Characterization of ife-1, ife-2, and ife-3 and identification of corresponding ESTs were reported earlier (20). No ESTs corresponding to ife-4 or ife-5 have yet been identified in the C. elegans data base of the DNA Data Bank of Japan. cDNAs containing the complete open reading frame sequences of the ife-4 and ife-5 genes were amplified by RT-PCR from C. elegans RNA. Analysis of trans-splicing (see below) verified that the predicted initiation codons for ife-4 and ife-5 were the 5Ј-most ATG codons in each mRNA.
The cDNA sequences matched the ife-4 and ife-5 genes absent the intron sequences (data not shown). 5Ј and 3Ј splice sites (including trans-splicing acceptor sites; see below) in both genes conformed closely to the consensus sequences for C. elegans (5). The ife-4 gene contains three introns of 292, 483, and 419 bp following, respectively, codons 21 (Met), 81 (Thr), and 161 (Arg). This gene structure is unique among the ife genes. ife-5 contains a single intron of 54 nt after codon 152 (Gly). This Gly residue is conserved in IFE-1, IFE-2, IFE-3, and IFE-5 (see below), but an intron is found at this position only in the genes encoding IFE-1, IFE-2, and IFE-5 (data not shown).
Alignment of all five IFE amino acid sequences shows that they are more similar throughout the central core and more divergent in the NH 2 -and COOH-terminal regions (Fig. 1A). The sequence of ife-4 encodes a putative protein of 212 amino acids with a molecular mass of 24,583 Da. Surprisingly, IFE-4 is more similar to human 4E-HP (48% identity) than to the other C. elegans eIF4E isoforms (27-35% identity; Fig. 1B). Despite its sequence divergence from the other IFEs, however, IFE-4 contains the five Trp residues in its central core that are characteristic of eIF4E proteins (12,35).
The sequence of ife-5 encodes a putative protein of 201 amino FIG. 1. Homology among the five eIF4E isoforms from C. elegans. A, amino acid sequences were deduced from the cloned cDNAs. Alignment was performed using PILEUP from the Wisconsin Software Package (Genetics Computer Group) with a gap penalty of 3.0 and a gap extension penalty of 0.1. Alignment was not significantly changed by gap penalties from 2.0 to 3.5. Sequences are ordered according to relatedness (e.g. IFE-1 is most closely related to IFE-5). Residues identical in three of the five sequence are shaded. B, the percentage of amino acid identity was calculated from pairwise alignments with the GAP algorithm of the Wisconsin Software Package using a gap penalty of 2.0 and a gap extension penalty of 0.05. Similar alignments to human eIF4E-1 (GenBank TM accession number M15353) and human 4E-HP (AF047695) were performed. The ability of each protein to bind caps containing MMG and TMG is shown by ϩ or Ϫ. C, a multiple protein sequence alignment was performed as in A, but also included were human eIF4E-1(hu eIF4E-1), human 4E-HP (hu 4E-HP), Arabidopsis nCBP (ar nCBP; AF028809), Arabidopsis eIF(iso)4E (Ar p26; Y10547), and Arabidopsis eIF4E (Ar p28; Y10548). The alignment was used to generate a phlyogenetic tree to display the degree of relatedness between two protein sequences (see "Experimental Procedures"). Distances along a branch are proportional to the extent of sequence divergence between two proteins.

Five eIF4E Isoforms in C. elegans
acids with a molecular mass of 23,277 Da, the smallest of the five IFE proteins. IFE-5 is highly homologous to IFE-1, sharing 80% amino acid identity (Fig. 1B) distributed over the entire length of both proteins (Fig. 1A). IFE-5 is next most closely related to IFE-2 (57% identity). Despite its similarity to IFE-1, IFE-5 is predicted to be considerably more basic (pI 7.7) than IFE-1 (pI 5.5) or the other IFE proteins (pI 6.1-6.6).
Sequence comparisons among all of the IFE proteins, the two human eIF4E isoforms, and three known plant (Arabidopsis thaliana) eIF4E isoforms can be represented in a phylogenetic tree (Fig. 1C). The proteins fall into four major branches. The first branch includes human eIF4E-1 and C. elegans IFE-3. IFE-1, IFE-2, and IFE-5 cluster tightly on a second branch. The unusual eIF4E proteins human 4E-HP, plant nCBP and C. elegans IFE-4 are found on the third branch. The fourth branch includes the plant proteins, p28 and p26, also known as eIF4E and eIF(iso)4E.
rIFE-4 and rIFE-5 were produced in E. coli, purified on m 7 GTP-Sepharose, and specifically eluted with m 7 GTP, indicating that the cDNAs generated by RT-PCR encode authentic cap-binding proteins (Fig. 2A). The molecular masses estimated by electrophoretic migration of rIFE-4 (ϳ25 kDa) and rIFE-5 (ϳ23 kDa) were consistent with the masses predicted from the cDNA sequences (24,583 and 23,277 Da, respectively).
IFE-4 and IFE-5 Are Expressed in C. elegans-The amplification of cDNAs corresponding to fully spliced ife-4 and ife-5 mRNAs from C. elegans RNA by RT-PCR indicated that these genes are expressed, at least at the transcriptional level. To verify the presence of IFE-4 and IFE-5 protein in worms, antisera were raised against unique peptides derived from COOHterminal amino acid sequences of each predicted protein. The specificity of the antipeptide antibodies was verified by Western blotting against purified rIFE-1, rTS-IFE-2, rIFE-3, rIFE-4, and rIFE-5. Antisera against IFE-4 ( Fig. 2F) and IFE-5 (Fig. 2G) gave robust signals for their cognate recombinant proteins but showed no cross-reactivity with the other IFE proteins. m 7 GTP-binding proteins were prepared from C. elegans strain N2 lysate (1.36 mg of total protein) and subjected to immunoblotting with the antisera against IFE-4 ( Fig. 2B) and IFE-5 (Fig. 2, C and D). Each antiserum recognized a single ϳ25-kDa protein, verifying that the ife-4 and ife-5 genes are expressed in vivo.
Cap Binding Specificities of IFE-4 and IFE-5-The binding of rIFE-4 and rIFE-5 to MMG and TMG caps was assayed. Because both recombinant and natural forms of the two proteins were purified on m 7 GTP-Sepharose (Fig. 2, A-D), they recognize MMG caps, as do IFE-1, IFE-2, and IFE-3 (20). Purified rIFE-4 and rIFE-5 were then applied to a m 3 2,2,7 GTP-Sepharose column (TMG cap analog resin). rIFE-4 failed to bind and was observed in the column flow-through (Fig. 3A, lanes F) rather than in the m 3 2,2,7 GTP elution (lane E). rIFE-5, on the other hand, was retained on the affinity resin and specifically eluted with m 3 2,2,7 GTP (Fig. 3B, lane E), indicating that it is a TMG-binding isoform.
Targeted Inactivation of ife Genes-We interfered with expression of each of the ife genes using RNAi (36) to determine which of the IFEs had essential functions in C. elegans. In this technique, expression of a specific gene is inhibited by the injection of the corresponding double-stranded RNA into young adult worms, and then phenotypes are assayed in their progeny. Phenotypes obtained through the RNAi technique have been shown to be gene-specific and are essentially equivalent to phenotypes observed for null mutations (34,37).
Various ife genes were inactivated individually and in combination by RNAi (Table II). Inactivation of ife-3, which is the most similar to the human eIF4E-1 gene, resulted in 100% embryonic lethality (Set 1), whereas inactivation of ife-1, ife-2, ife-4, or ife-5 alone showed no embryonic lethality whatsoever. RNAi against the combination of ife-1 and ife-2 was lethal to 75% of the embryos (Set 2). Similarly, inactivation of ife-2 and ife-5 caused 89% embryonic lethality. Simultaneous inactivation of ife-1, ife-2, and ife-5 resulted in 99% embryonic lethality. On the other hand, none of the combinations in which ife-4 was inactivated was deleterious to the worms, including simultaneous inactivation of ife-4, ife-1, and ife-5 (Set 3). GTP (E). Equivalent volume aliquots of each fraction were resolved and visualized as in Fig. 2A. A, analysis of rIFE-4. B, analysis of rIFE-5.

Five eIF4E Isoforms in C. elegans
trans-Splicing of ife mRNAs-The fact that some eIF4E isoforms recognize TMG caps suggests that they are required to translate trans-spliced mRNAs. We next determined whether the mRNAs encoding the IFEs were themselves trans-spliced. Antisense oligonucleotides to each of the ife cDNAs were used in conjunction with sense primers to either SL1 or SL2 to amplify products by RT-PCR from poly(A) ϩ RNA. At least one trans-spliced mRNA was detected for each ife gene.
A single amplified product for ife-1 mRNA indicated that it is trans-spliced to SL2 (Fig. 4A). The presence of the SL2 leader suggests that the ife-1 gene is downstream in a cluster of genes that form an operon (8). Consistent with that prediction, ife-1 is the third of five genes clustered within a 14-kilobase region of C. elegans chromosome III (Fig. 4C) (38). Intergenic distances in this putative operon range from 122 to 449 bp. The sequence of the SL2-ife-1 PCR product indicated that the spliced leader was added 2 nt upstream of the initiation codon of the open reading frame reported for ife-1 (20), similar to the spacing observed for most trans-spliced mRNAs (5). Three of the 11 known ife-1 ESTs contain more sequence upstream of this ATG (yk364a1, 11 bp; yk385e3, 31 bp; yk275b11; 92 bp). They are colinear and correspond to part of a putative exon 161 nt upstream of (and spliced to) the same 3Ј splice site used by SL2. None of the longer mRNAs was detected by RT-PCR, suggesting that they are not trans-spliced. Only clone yk275b11 contains another ATG upstream, and it is in frame with the reported ife-1 open reading frame (20).
Three products were observed with SL1 and primer 2.2 (Fig.  4A); however, sequencing of the two larger products indicated that they were not related to ife-2 (data not shown). The smallest SL1-ife-2 product was further amplified by a second round of PCR using the same primers (Fig. 4B). The length of the ife-2-derived product was consistent with existing ESTs for ife-2, one of which contains part of the SL1 sequence spliced 6 nt upstream of the ATG (yk469e7).
Major RT-PCR products corresponding to ife-3 and ife-4 mRNA were produced using primer SL1 but not primer SL2 (Fig. 4A). The length of the ife-3-derived product was consistent with existing ESTs for ife-3. EST clone yk81f11 encodes part of the SL1 sequence 4 nt upstream of the initiation codon of ife-3. The sequence of the SL1-ife-4 PCR product indicated that the mRNA was trans-spliced 2 nt upstream of the ATG that begins the predicted open reading frame of the ife-4 gene. The SL1-ife-4 PCR product therefore represents an mRNA that encodes the complete IFE-4 protein.
No discrete products of the expected size were observed for ife-5 mRNA with either the SL1 or the SL2 primer in the primary RT-PCR reaction (Fig. 4A), indicating either that it is not transspliced or that the abundance of ife-5 mRNA is low. Secondary PCR using a nested ife-5 primer, primer 5.3, produced both SL1and SL2-containing products of different sizes (Fig. 4B). The sequence of the SL1-ife-5 secondary PCR product indicated an mRNA that is trans-spliced 3 nt upstream of the putative initiation codon. The sequence of the SL2-ife-5 secondary PCR product indicated that the corresponding mRNA contained an additional exon of 120 nt that was likewise spliced (but in cis) 3 nt upstream of the predicted ATG (data not shown). This putative exon matched sequences 2.5 kilobases upstream of the ife-5 gene on the C. elegans genomic clone, Y57A10. It contained a 93-nt open reading frame followed by a termination codon and then by two out-of-frame ATGs. Although this product may correspond to an authentic bicistronic mRNA, it more likely represents an incompletely spliced intermediate or an aberrant splicing product. Overall, these data suggest that ife-1 mRNAs are transspliced to SL2 and that ife-2, ife-3, ife-4, and probably ife-5 mRNAs are trans-spliced to SL1. DISCUSSION Previously we identified three eIF4E isoforms in C. elegans, IFE-1, IFE-2, and IFE-3 (20). The present study identifies two more isoforms, IFE-4 and IFE-5. Because the sequencing of the 14 0 a Results are organized according to the group of IFEs (A, B, or C; see "Discussion") targeted for inactivation.
b Sequence to which double-stranded RNA was produced for injection. c Young adult hermaphrodites were injected, and F1 progeny were monitored for 4 days.
d Calculated as the number of dead embryos produced divided by the total number of embryos produced ϫ 100%.  (Table I). A, primary RT-PCR products were resolved by electrophoresis in a 2% agarose gel and stained with ethidium bromide. 5Ј primers (SL) correspond to SL1 or SL2. 3Ј primers (ife) were primer 1.2 (lanes 1), primer 2.2 (lanes 2), primer 3.2 (lanes 3), primer 4.2 (lanes 4), and primer 5.2 (lanes 5). B, approximately 20 ng of purified primary RT-PCR product was subjected to secondary PCR (20 cycles) using ife-2-or ife-5-specific primers (primers 2.2 and 5.3, respectively) in conjunction with SL1 or SL2 primers, and the products were analyzed as in A. C, organization of the ife-1 operon. Five predicted genes are encoded in the operon: (a) a homolog of the mouse nud C gene, (b) a putative CCAAT-binding transcription factor gene, (c) translation initiation factor gene, ife-1, (d) a putative acetyl-CoA thiolase gene, and (e) a homolog of the mouse myotubularin/ mtm1 gene. All five genes are colinearly transcribed from a single upstream promotor (arrowhead) and trans-spliced to monocistronic mRNAs.
Five eIF4E Isoforms in C. elegans C. elegans genome is essentially complete (21) and because no other obvious candidate genes are found by BLAST sequence searches (22), these five proteins are likely to represent the complete set. Their characterization with respect to primary structure, gene structure, cap binding specificity, and requirement for embryonic viability allows us to group them into three classes.
Class A contains IFE-3, the most similar of the five IFE proteins to human eIF4E-1 (47% identity). Mammalian eIF4E-1 is by far the best characterized translational capbinding protein (10,39). It is found free in the cytoplasm, in a complex with PHAS-I (4E-BP), and in initiation complexes bound to eIF4G, where it brings together the 5Ј-and 3Ј-termini of the mRNA, the 40 S ribosomal subunit, and the RNA helicase eIF4A. The rate of protein synthesis and regulation of the cell cycle are highly dependent on intracellular levels of eIF4E-1 (10, 40) and eIF4G-1 (41,42). Depletion of eIF4E-1 levels by antisense RNA (43) or binding to PHAS-I (44) results in reduced protein synthesis and slow cell growth; elevation of eIF4E-1 levels by ectopic vector expression or microinjection causes accelerated cell growth (45), malignant transformation (46), and protection against apoptosis (47). eIF4E-1 discriminates against TMG caps in favor of MMG caps (18,19), a characteristic shared by IFE-3 (20). Of all the IFE proteins, only IFE-3 is absolutely required for embryonic viability (Table  II, Set 1). Surprisingly, the other IFE proteins failed to substitute for IFE-3, although all of them bind to MMG caps (Ref. 20 and Fig. 2A). This is not due to level of expression, because IFE-3 and IFE-1 are approximately equally abundant in preparations of cap-binding proteins from C. elegans (20). Therefore, IFE-3 plays some unique role in protein synthesis (e.g. specific interactions with other initiation factors, binding to non-trans-spliced mRNA, tissue distribution, temporal expression, etc.) that makes it essential to the nematode.
Class B includes IFE-1, IFE-2, and IFE-5. These isoforms are the most closely related in primary structure of the five proteins (57-80% identity; Fig. 1B), and they form a cluster on the phylogenetic tree (Fig. 1C). Their genes contain an identically placed intron that is absent in ife-3 and ife-4. Finally, all three bind TMG as well as MMG caps, whereas IFE-3 and IFE-4 bind only MMG caps. These similarities in structure and function suggest that Class B ife genes are descended from a common ancestral gene.
The structural basis for the difference in cap-binding specificities of the IFE proteins is not known. Two possibilities are: (a) the proteins that bind MMG caps exclusively (IFE-3 and IFE-4) are unable to bind TMG caps because of steric hindrance with the additional methyl groups on N2, but this steric hindrance is absent in the TMG-binding proteins (IFE-1, IFE-2, and IFE-5) and (b) the TMG-binding proteins contain residues that form Van der Waals' contacts with the N2 methyl groups, but these residues are missing in the MMG-binding proteins. Thus, it is instructive to examine amino acid sequence motifs common to all three TMG-binding proteins that are absent in both MMG-binding proteins. This comparison can be further limited by the three-dimensional structure of mouse eIF4E-1 (48), which indicates that the residues closest to the N2 position of m 7 GDP are in the "S1-S2 loop" (Lys-49 to Asn-59) and the "S3- RNA interference experiments also supported the similarity of the TMG-binding isoforms (Class B). Unlike Class A, the knockout of any one Class B member did not affect viability (Table II, Set 2), suggesting an overlap in their functions. However, knockout of all combinations of Class B members except ife-1 plus ife-5 produced embryonic lethality, the strongest effect (99% lethality) being the knock-out of all three Class B members. This suggests that the presence of TMG-binding isoforms is necessary for the viability of the developing embryo. TMG-capped mRNAs have been observed to efficiently associate with polyribosomes in vivo (9), and it is presumed that these proteins promote the translation of TMG-containing mRNAs. However, the three Class B members are not equivalent. IFE-2 was sufficient for viability in the absence of both IFE-1 and IFE-5. Neither IFE-1 nor IFE-5 alone was sufficient in the absence of the other Class B proteins. Only IFE-1 plus IFE-5 were sufficient to compensate for the loss of IFE-2. Paradoxically, IFE-2 binds only weakly to TMG caps (20), yet is more important for viability than the strong TMGbinding isoforms. Class B proteins may be unique to organisms such as nematodes and trypanosomes, which possess TMGcapped mRNAs (8).
Recognition of TMG-capped mRNAs by Class B IFEs may also have an autoregulatory function. Each of the five isoforms is encoded by at least one trans-spliced mRNA. trans-Splicing of ife mRNAs also means that they contain TMG caps, yet only Class B IFE proteins are able to bind such mRNAs. It is possible that the translational efficiency of TMG-capped mRNAs, including all five ife mRNAs, is regulated by the level of Class B proteins. The fact that the ife-1 gene resides in an operon suggests that it may also be coordinately regulated with the other four genes in that operon at the level of transcription. However, the identity of the genes clustered with ife-1 (Fig. 4C) do not immediately suggest a rationale for such coordinate regulation.
Class C consists of IFE-4, which is the most unusual of the C. elegans eIF4E isoforms. It is more similar to Arabidopsis nCBP (17) and mammalian 4E-HP (13) than to any other IFE protein (Fig. 1C). Also, its gene has an intron/exon structure that is unique among the ife genes. RNAi experiments also suggest that the requirement for Class C proteins also differs from that for Class A and Class B proteins. IFE-4 was completely dispensible in the worms; no lethality was produced by knock-out of ife-4 either singly or in combination with other ifes (Table II,  Set 3). Either IFE-4 is functionally redundant in the worm or its function is restricted to a small population of cells that have no apparent role in animal development. It is curious that a nonessential eIF4E isoform would be conserved in widely separated species across two eukaryotic kingdoms.