Two zebrafish eIF4E family members are differentially expressed and functionally divergent.

Eukaryotic translation initiation factor 4E (eIF4E) is an essential component of the translational machinery that binds m(7)GTP and mediates the recruitment of capped mRNAs by the small ribosomal subunit. Recently, a number of proteins with homology to eIF4E have been reported in plants, invertebrates, and mammals. Together with the prototypical translation factor, these constitute a new family of structurally related proteins. To distinguish the prototypical translation factor eIF4E from other family members, it has been termed eIF4E-1 (Keiper, B. D., Lamphear, B. J., Deshpande, A. M., Jankowska-Anyszka, M., Aamodt, E. J., Blumenthal, T., and Rhoads, R. E. (2000) J. Biol. Chem. 275, 10590-10596). We describe the characterization of two eIF4E family members in the zebrafish Danio rerio. Based on their relative identities with human eIF4E-1, these zebrafish proteins are termed eIF4E-1A (82%) and eIF4E-1B (66%). eIF4E-1B, originally termed eIF4E(L), has been reported previously as the zebrafish eIF4E-1 counterpart (Fahrenkrug, S. C., Dahlquist, M. O., Clark, K., and Hackett, P. B. (1999) Differentiation 65, 191-201; Fahrenkrug, S. C., Joshi, B., Hackett, P. B., and Jagus, R. (2000) Differentiation 66, 15-22). Sequence comparisons suggest that the two genes probably evolved from a duplication event that occurred during vertebrate evolution. eIF4E-1A is expressed ubiquitously in zebrafish, whereas expression of eIF4E-1B is restricted to early embryonic development and to gonads and muscle of the tissues investigated. The ability of these two zebrafish proteins to bind m(7)GTP, eIF4G, and 4E-BP, as well as to complement yeast conditionally deficient in functional eIF4E, show that eIF4E-1A is a functional equivalent of human eIF4E-1. Surprisingly, although eIF4E-1B possesses all known residues thought to be required for interaction with the cap structure, eIF4G, and 4E-BPs, it fails to interact with any of these components, suggesting that this protein serves a role other than that assigned to eIF4E.

Eukaryotic mRNAs are modified post-transcriptionally by the addition of a 7-methylguanosine linked by a 5Ј-5Ј-triphosphate bridge to the first transcribed residue (reviewed in Refs. 4 and 5). This modification, known as a "cap" structure, is required for recruitment of mRNA by the translational machinery and is recognized by eukaryotic initiation factor 4E (eIF4E) 1 (reviewed in Refs. 6 and 7). Recruitment of mRNAs to the ribosome by eIF4E is accomplished through its interaction with the factor eIF4G (reviewed in Ref. 8). eIF4G interacts with eIF4E, eIF4A (an ATP-dependent RNA-helicase (9, 10)), and ribosome-bound eIF3 to place the small ribosomal subunit at the 5Ј end of mRNA in close proximity to the cap (11)(12)(13)(14). eIF4E has been characterized not only as a general translation factor but as a protein that can modulate both the overall rates of translation and the mRNA selectivity of the translational apparatus (6,7). The importance of the function of eIF4E is illustrated by the lethality of eif4e gene disruption in Saccharomyces cerevisiae (15). Both the activity and the structure of eIF4E are highly conserved throughout evolution; the human and yeast homologues are 32% identical at the amino acid level, and the mammalian factor can rescue eif4e gene disruption in S. cerevisiae (15).
Sequence comparisons and mutational analyses of eIF4E have revealed a phylogenetically conserved ϳ170-amino acid core with eight conserved tryptophans, which appears to be sufficient for cap binding (16 -18). The three-dimensional structure of the mouse and yeast eIF4E core sequences bound to cap analogues have been solved by x-ray crystallography and NMR, respectively (16,19). The core protein shows a cuppedhand shape, consisting of a curved, eight-stranded, antiparallel ␤-sheet backed by three ␣-helices (16,19). The cap analogue binds in a narrow slot within the concave surface (16,19). The binding of the nucleotide residue of the cap structure results from the stacking of the alkylated base between two conserved Trp residues and the formation of hydrogen bonds between the nucleotide base, the polypeptide backbone, and a conserved Glu residue (Trp-56, Trp-102, and Glu-103 for mouse eIF4E) (16,18,20,21). The presence of the N 7 -methyl moiety of the cap is detected by another Trp residue (Trp-166 in mouse eIF4E) (16,19). Crystallographic studies have also revealed that residues His-37, Pro-38, Val-69, Trp-73, Leu-131, Glu-132, and Leu-135 of mouse eIF4E make contacts with the eIF4E-binding domains of eIF4Gs and 4E-BPs (22), although mutational analyses have so far only identified Trp-73 (or its equivalent) as being re-quired for these interactions (19,22). In contrast to the conserved cap binding core, the N and C termini of eIF4Es vary in length, showing little or no conservation across the phyla. Most of the residues in these regions have been shown to be dispensable for cap binding and complementation in yeast (17).
Both the level and activity of eIF4E are regulated (reviewed in Refs. 23 and 24). Higher levels of eIF4E generally correlate with higher rates of protein synthesis and cell growth (reviewed in Ref. 8). eIF4E undergoes regulated phosphorylation by Mnk1 and -2 (25,26) at the residue equivalent to Ser-209 in human eIF4E-1 (27)(28)(29) (reviewed in Refs. 24 and 30). The relationship between eIF4E phosphorylation and eIF4E activity is not fully understood, although phosphorylation has been shown to be important for normal growth (31). The positive correlation between phosphorylation and overall protein synthetic rate initially led to the supposition that phosphorylation increases the ability of eIF4E to bind cap structures. However, subsequent kinetic analysis has revealed that phosphorylation diminishes the ability of eIF4E to bind cap by increasing the off-rate of eIF4E, suggesting that phosphorylation may serve to facilitate release of eIF4E from the 5Ј cap structure during the initiation cycle (32). Recruitment of mRNA by eIF4E can also be regulated by its association with a family of small eIF4Ebinding proteins (4E-BPs), the members of which share homology with eIF4G and act as competitive inhibitors of the eIF4E/ eIF4G interaction (22,24,33,34).
Recently, a number of proteins with homology to eIF4E have been reported in plants, invertebrates, and mammals. Together with the translation factor, these constitute a new family of structurally related proteins. This multiplicity of eIF4Es could reflect simple functional redundancy or that the eIF4E-related proteins support alternate roles. These issues have been addressed to a limited extent in Caenorhabditis elegans, in which the existence of several eIF4E-like factors can be partially explained by the necessity of recruiting mRNAs with both 7-methyl-and 2,2,7-methylguanosine caps to the initiation complex. In most systems, however, the biological significance of multiple eIF4E-like genes remains to be explored.
We describe here the characterization of two eIF4E family members in the zebrafish Danio rerio. Based on their relative identity with human eIF4E-1, these zebrafish proteins are termed eIF4E-1A and eIF4E-1B. The similarity of structure in the core regions of eIF4E-1A and eIF4E-1B to human eIF4E-1 suggests that both proteins could function as bona fide translation factors. However, we demonstrate that only eIF4E-1A possesses eIF4E-like activities and that this factor, unlike eIF4E-1B, is able to complement eIF4E deficiency in S. cerevisiae. Thus, by the analyses described, zebrafish eIF4E-1A is a functional equivalent of human eIF4E-1, whereas the role of eIF4E-1B remains to be determined.

EXPERIMENTAL PROCEDURES
Rearing and Spawning Zebrafish-Adult Fish were purchased from Aquarium Center (Baltimore, MD) and kept at 28.5°C in a constant flow-through system. Embryos were obtained by spontaneous spawning, maintained at 28.5°C, and staged as described (35). Staged embryos were immediately snap-frozen and stored at Ϫ80°C for future use.
Culture of Zebrafish Cell Lines, zfl-Cells were grown at 28°C in L-15 medium supplemented with 10% fetal calf serum but without sodium bicarbonate.
Reverse Transcription and PCR Amplification Reactions-One volume of fresh tissue, embryos, or harvested cells was homogenized in 8 volumes of Trizol (Invitrogen), essentially as described (36). For PCR, 1 g of total RNA was reverse-transcribed using Superscript II reverse transcriptase (Invitrogen) and oligo(dT) 18 in accordance with the manufacturer's instructions. Amplification reactions (primers as indicated) were performed under standard conditions for either Pfu (Stratagene) or Taq (Fisher) DNA polymerases with 5 cycles of annealing at 58°C, followed by 35 cycles of annealing at 68°C. Extension for all cycles was allowed for 1 min at 72°C. Whenever purified PCR products were transferred into plasmids, insertions were sequenced in both orientations to ensure that no errors had been introduced due to amplification.
For expression studies, DNA-contamination controls included reactions treated identically but lacking reverse transcriptase, as well as reactions lacking template. The products were resolved by TAE-agarose electrophoresis and recorded in a fluorimager (Amersham Biosciences).
Generation of cDNAs Encoding Zebrafish eIF4E Family Members eIF4E-1A and eIF4E-1B-A partial cDNA (EST AA542678) cloned into the vector pSPORT1 (Invitrogen) was purchased from Genome Systems Inc. and found to contain residues equivalent to all but the initiation codon and 5Ј-UTR of eIF4E-1A mRNA. From this partial cDNA and overlapping sequence information from the EST data base, a cDNA fragment encoding the complete eIF4E-1A protein was generated by PCR amplification using Pfu polymerase, the sense primer z4E1ANcoI5-23(F), named for the corresponding nucleotides following the ATG of eIF4E-1A, and the antisense primer z4E1ABamHI927-906(R), named by the same convention. Similarly, a cDNA fragment encoding eIF4E-1B was generated from the plasmid pBKSII(Ϫ)-zeIF4E-L (2), using the sense primer z4E1BNcoI5-42(F) and the antisense primer z4E1BBamHI653-632(R). In either case, the purified PCR product was digested with NcoI and BamHI and transferred into the vector pCITE4a(ϩ) (Novagen).
Generation of cDNA Encoding Zebrafish 4E-BP-A DNA fragment encoding zebrafish 4E-BP was isolated following amplification by RT-PCR using Taq DNA polymerase. Primers for amplification were based on the nucleotide alignment of multiple partial sequences encoding zebrafish 4E-BP in the EST data bank, forward primer z4EBPNdeI3-24(F) and reverse primer z4EBPBamHI339 -319(R). PCR products were transferred into the vector pGEM-T-easy (Promega) generating pGEMTEz4EBP. Based upon consensus sequences from alignment of multiple ESTs, no errors had been introduced during amplification. The insert was transferred into pCITE4a(ϩ) following digestion with NdeI and BamHI generating pCITE4az4EBP.
Generation of cDNA Encoding the eIF4E-binding Domain of Human eIF4GI-The 1366-bp ApaI/BamHI fragment from plasmid pBSKII(Ϫ)HFC1 (a gift from the laboratory of Dr. R. E. Rhoads) was modified by addition of the sequence CATATGTCCATGG, 5Ј of the intact ApaI site such that the ATG of both the NdeI and NcoI sites introduced were in-frame with the eIF4GI coding sequence. Following the modification, the NdeI/BamHI fragment was transferred into the NdeI and BamHI sites within pCITE4a(ϩ), generating pCITE4ah4GI4EBDHis. This encodes a protein containing amino acids 159-614 of eIF4GI fused to a C-terminal His tag and containing the eIF4E-binding domain.
Use of eIF4E-deficient Yeast Strain (JOS003)-Generation of a G418resistant S. cerevisiae strain, JOS003, lacking an endogenous functional yeast eif4e gene is described elsewhere (37). In essence, the endogenous yeast eif4e gene was replaced by homologous recombination with a linear DNA fragment containing the G418 resistance cassette KanMX4 (37,38). This strain is able to survive due to the expression of Leu-selectable, extrachromosomal, human eIF4E-1, driven by the glucose-repressible, galactose-dependent GAL1 promoter. Medium lacking galactose and containing glucose suppresses growth of this strain, allowing complementation tests to be performed. All yeast cells were grown at 30°C in either YP medium (1% yeast extract, 2% bactopeptone) or synthetic complete (SC) medium lacking only leucine or both leucine and uracil, as necessary. Medium was supplemented with either 2% D-galactose or 2% D-glucose (dextrose) as carbon source and 200 g/ml G418 for continuous selection of cells lacking the yeast eif4e gene.
eIF4E Complementation Assay in S. cerevisiae Conditionally Lacking Functional eIF4E-Initially, NcoI-BamHI fragments from pCITE4aeIF4E constructs were transferred into the Escherichia coli expression vector pET11d. XbaI-BamHI fragments derived from the resulting pET11deIF4E constructs were subsequently transferred into the Ura-selectable yeast expression vector pRS416GPD (39), which allows expression under the control of the constitutive yeast glyceraldehyde-3-phosphate dehydrogenase promoter. JOS003 transfected with pRS416GPDeIF4E constructs were selected on SCGal medium containing G418 and lacking Leu and Ura. Complementation tests were performed either by streaking or plating ϳ1,000 cells from a log-phase culture on YP-medium containing G418 and either galactose or dextrose. Plates were incubated at 30°C for 3 days, and growth was assessed visually by colony formation. Growth on plates containing dextrose indicates the ability of an ectopic eif4e gene to complement eIF4E deficiency.
Preparation of Recombinant eIF4E and 4E-BP from E. coli-The relevant pET11deIF4E constructs were transfected into BL21(DE3)-pLysS cells and expressed as described (40). 500-ml cultures, derived from the relevant cells grown in LB medium, were induced at 0.5-0.6 A 600 with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 30 min, followed by further 1.5 h of induction in the presence of 0.2 g/ml rifampicin. The recombinant proteins were recovered from inclusion bodies, essentially by the method of Rudolph and co-workers (41, 42) using solubilization in guanidine-HCl, followed by staged dialyses. His-tagged 4E-BP was purified from E. coli cell extracts by affinity chromatography using HisBind (Novagen).
In Vitro Expression and Protein/Protein Interaction Assays-For production of 35 S-radiolabeled proteins translated in vitro, the relevant pCITE4a constructs were used as templates in the rabbit reticulocytecoupled transcription-translation system, containing [ 35 S]Met (0.25 mCi/ml), essentially as described previously (43). For interaction assays, the fragment of human eIF4GI containing the eIF4E-binding domain from pCITE4ah4GI4EBDHis or zebrafish 4E-BP was co-translated with either S-tagged eIF4E-1A or eIF4E-1B in 50-l reactions for 60 min at 30°C. Reactions were diluted with 10 volumes of S-binding/ washing buffer (Novagen) and incubated with 15 l of S-proteinagarose (Novagen) for 45 min at 20°C. S-protein-agarose beads were recovered by centrifugation and washed 10 times with buffer (1 ml each), prior to elution with SDS-PAGE sample buffer. Samples of fractions, equivalent to 2 l of the initial translation reactions, were analyzed by SDS-PAGE and autoradiography.
For the interaction of zebrafish ovary eIF4Es with His-tagged 4E-BP, anti-His/protein G-Sepharose beads were prepared by binding anti-His 5 antibody (Qiagen) to protein G-Sepharose beads in binding buffer containing 300 mM NaCl, 25 mM KCl, 20 mM HEPES-KOH, pH 7.2, 2 mM EGTA, 10% glycerol, 1 mg/ml soybean trypsin inhibitor, 200 g/ml bovine serum albumin, at 4°C, with shaking, overnight. The beads were washed 6 times in binding buffer and incubated with His-tagged 4E-BP for 2 h at 4°C. Binding of His-tagged 4E-BP was confirmed by Western analysis of bound material, using INDIA-HisProbe-HRP (Pierce) and ECL for visualization of bead-bound His-tagged protein.
Recombinant eIF4E-1A and eIF4E-1B or ovary extracts (pooled from 10 fish) were incubated with His-tagged human 4E-BP bound to anti-His/ protein G-Sepharose beads for 2 h at 4°C. After 3 washes in binding buffer and 3 washes in low salt binding buffer containing 25 mM KCl, 20 mM HEPES-KOH, pH 7.2, 2 mM EGTA, 10% glycerol, 200 g/ml bovine serum albumin, bound proteins were eluted in 100 mM glycine, pH 2.5, and analyzed for the presence of eIF4E-1A and -1B by high-Tris SDS-PAGE and immunoblotting. m 7 GTP-affinity Chromatography-Ovary extracts (from one fish) or in vitro translation reactions (50 l) were bound to 50 l of m 7 GTP-Sepharose equilibrated with column buffer containing 25 mM HEPES-KOH, pH 7.2, 10% glycerol, 100 mM KCl, 0.1 mM EDTA, and 1 mM dithiothreitol. The resin was washed extensively with 50 volumes of the same buffer containing 200 M GTP, prior to elution with buffer containing 200 M m 7 GTP. Equivalent volumes of fractions representing either 10 l of ovary extract or 2 l of translation reaction were used for further analyses. For chromatography of proteins produced from E. coli, 50 ml of buffer containing 1.6 g/ml each of recombinant eIF4E-1A and eIF4E-1B and 100 g/ml total E. coli proteins was applied to 0.5 ml of m 7 GTP-Sepharose resin. The resin was washed with 100 volumes of buffer prior to the collection of 0.5-ml fractions from progressive elutions with 5 ml of the buffers as indicated. Samples from fractions, equivalent to 3 l of the applied material, were used for SDS-PAGE and immunoblotting analyses.
SDS-PAGE-For separation of different eIF4Es, samples were electrophoresed on 17.5% polyacrylamide/high-Tris for 1600 V-h for increased resolution. All buffers and solutions were essentially as described (44).
VSIEF-The isoelectric focusing method of O'Farrell (45), modified for use in vertical slab gels, was used essentially as described (46) but with modifications to the Pharmalyte composition and buffers used to give optimal resolution of eIF4E-1A and eIF4E-1B. For eIF4E-1A, samples were focused through pH gradients of 5 to 8.5, using Pharmalytes 3-10 in the presence of 9 M urea, 5% v/v ␤-mercaptoethanol, 2% w/v CHAPS (Fluka) with 25 mM NaOH at the anode and 25 mM aspartic acid at the cathode. For eIF4E-1B, samples were focused through pH gradients of 8 -10, using a mixture of Pharmalytes 3-10 and 9-11 (4:1), with 25 mM NaOH at the anode and 25 mM glutamic acid at the cathode. Samples were electrophoresed at 2 mA/gel for 20 h prior to immunoblotting. pH strips were used to verify the pH gradient of the gels after each run.
Immunoblotting-Proteins were electrotransferred from gels to PVDF membranes and probed with rabbit anti-TATKSGSTTKNRFVV (gift from Dr. Simon Morley) at 60 V for 1.25 h for SDS-PAGE and 1 h for VSIEF gels, prior to visualization by enhanced chemiluminescence, as described (47).

Cloning and Sequence Characterization of a cDNA Encoding a New eIF4E-related Protein from Zebrafish
An EST clone (GenBank TM accession number AA542678), generated by the Washington University zebrafish genome project, was identified by BLAST search (48) for clones encoding proteins similar to human eIF4E-1. This clone was purchased from Genome Systems Inc. and represents a cDNA with coding potential for a protein 82% identical to human eIF4E-1 and 84% identical to Xenopus eIF4E-1. From this partial cDNA and overlapping sequence information from the NCBI data bases, a cDNA fragment containing a complete eIF4E-1-related protein coding sequence was generated by PCR amplification. 2 Sequence comparison of this cDNA (GenBank TM accession number AF257519) with that of the cDNA encoding eIF4E(L) (Gen-Bank TM accession number AF176317) (3), showed that the two are distinct, having considerable differences in the length and sequence of their 3Ј-UTR (data not shown). The close sequence relationship between these cDNAs suggests that the two genes encoding these zebrafish eIF4E homologues share a common origin. Based on the closer sequence similarity of AF257519 to the human and Xenopus eIF4E-1 cDNAs, we termed the predicted protein encoded by AF257519 as eIF4E-1A, and we renamed the less similar predicted protein, formerly termed eIF4E(L), eIF4E-1B.
The deduced amino acid sequence of zebrafish eIF4E-1A is aligned with those of zebrafish eIF4E-1B, human eIF4E-1, and Xenopus eIF4E-1 in Fig. 1, with predicted biochemical characteristics given in Table I. Zebrafish eIF4E-1A and eIF4E-1B are Ͼ80% identical to each other in their core sequences (Fig. 1) and share extensive homology over the core sequences with human and Xenopus eIF4E-1. Over the entire protein, eIF4E-1A shows 82% identity and 91% similarity to human eIF4E-1, whereas eIF4E-1B shows 66% identity and 77% similarity. Higher sequence identities are seen in the conserved core (equivalent to residues 36 -200 in human eIF4E-1), with greater sequence diversity in the N termini. Over the conserved core, eIF4E-1A shows 89% identity and 96% similarity to human eIF4E-1, whereas eIF4E-1B shows 75% identity and 87% similarity. Both proteins contain the eight tryptophan residues conserved in all eIF4E-1 sequences known. In addition, all the residues recognized as essential for cap binding activity (16,19), those equivalent to Trp-56, Trp-102, and Trp-166 of human eIF4E-1 and the conserved Glu at a position equivalent to Glu-103, are strictly conserved in both zebrafish eIF4Es (highlighted in red in Fig. 1), suggesting that both are good candidates to function as eIF4E. The tryptophan residue equivalent to Trp-73 in human eIF4E-1, which is required for interaction with eIF4G and 4E-BP (22), is found at the corresponding positions in both zebrafish eIF4Es within the conserved region involved in this interaction (highlighted in dark purple in Fig.  1). Similarly, all residues known to make contacts with the eIF4E-binding domains of eIF4Gs and 4E-BPs, those equivalent to His-37, Pro-38, Val-69, Trp-73, Leu-131, Glu-132, and Leu-135 in mouse eIF4E-1 (22), are present in both proteins. Like all other eIF4E sequences characterized so far, the N termini of the zebrafish eIF4Es are quite divergent, both from each other and from eIF4Es from other species. Like prototypical eIF4E-1s, zebrafish eIF4E-1A has an acidic N-terminal region, contributing to its predicted acidic isoelectric point (5.5) ( Table I). The molecular weight of eIF4E-1A (24.8) is also in good agreement with those of human and Xenopus eIF4Es (50 -52). The predicted molecular weight eIF4E-1B is 24.7. However, eIF4E-1B is unusual in having a lysine-rich region (PKKKVEKKK) N-terminal to the conserved core, which contributes to its overall basic predicted isoelectric point (9.39). Short sequences rich in basic amino acids are characteristic of nuclear localization sequences, such as that for SV40 T-antigen, although specific consensus sequences have not been formulated (53,54). This opens up the possibility that the Nterminal region of zebrafish eIF4E-1B could function in this capacity.

Analysis of the Expression Patterns of eIF4E-1A and eIF4E-1B
Expression in Different Tissues-To assess whether eIF4E-1A and eIF4E-1B represent eIF4E family members that are actually expressed in zebrafish, we looked for the transcripts of each in a variety of tissues using RT-PCR (Fig. 2). A representative experiment in which equal amounts of RNA from each tissue were subjected to oligo(dT)-primed reverse transcription, followed by PCR with gene-specific primers, is shown in Fig. 2A. mRNA for eIF4E-1A was detected in all tissues examined, whereas eIF4E-1B mRNA was detectable only in ovary and muscle.
The suggestion of cell type-specific expression of eIF4E-1A and -1B observed by RT-PCR analysis was re-evaluated in a variety of tissues at the protein level. Although resolution of purified eIF4E-1A and eIF4E-1B could be achieved by high-Tris SDS-PAGE, for the analysis of crude extracts we found superior separation of the two forms by taking advantage of their differing isoelectric points. Following vertical slab isoelectric focusing (VSIEF), proteins from adult tissues were electrotransferred to a membrane and subjected to immunoblotting by using an antibody raised against a peptide corresponding to the C-terminal sequence of human eIF4E-1, TATKSGSTTKNR-FVV (Fig. 2B). The C terminus of eIF4E-1A differs by one residue from this sequence, whereas eIF4E-1B differs at three positions from the human-derived peptide. Nevertheless, both  1. Multiple alignment of predicted amino acid sequences of zebrafish eIF4Es, human eIF4E-1, and Xenopus eIF4E-1. The predicted amino acid sequences of eIF4E-1A and eIF4E-1B were aligned with those of human eIF4E-1 and X. laevis eIF4E-1. The period indicates identity at that position, relative to human eIF4E-1. A dash represents a gap in the sequence. The residues within the core region are indicated in black. Variations relative to human eIF4E-1 are identified by the symbol of the respective residue at each position. Residues in red are essential for cap binding activity. Residues in blue represent the region N-terminal to the conserved core. Residues in purple represent the conserved region surrounding the tryptophan (Trp-73 in human, highlighted in dark purple) that is essential for interaction with eIF4G and 4E-BPs (22). Residues in green define the sequence of the peptide used to raise the anti-eIF4E antibody used in this investigation. Highlighted in dark green is Ser-209 from human eIF4E-1, the site of phosphorylation by cellular kinases (27). recombinant eIF4E-1A and eIF4E-1B interact with the antibody. As expected, recombinant eIF4E-1A focused at a pH between 5 and 7, whereas recombinant eIF4E-1B focused between pH 9 and pH 10. eIF4E-1A was detected in all tissues examined. eIF4E-1B on the other hand, was detected only in ovary, testes, and muscle, consistent with the RNA expression data. Ovary cell fractionation studies indicate that both eIF4E-1A and -1B are found in both nucleus and cytoplasm (results not shown).
Developmental Expression-We further analyzed the expression characteristics of both zebrafish eIF4Es, by assaying RNA from embryos at different developmental stages (Fig. 3). RT-PCR was performed under conditions similar to those used for the tissue distribution assay. eIF4E-1A mRNA was detected at all stages of zebrafish development examined, whereas eIF4E-1B mRNA was detected only in stages that precede Prim6 (24 h post-fertilization). The analyses of the expression patterns of the eIF4E family members suggest that eIF4E-1A is ubiquitously expressed, as expected for an essential translation factor. However, eIF4E-1B appears non-essential for cell viability, given that neither its mRNA nor its protein accumulate in many cell types, including an actively replicating zebrafish liver cell line, zfl (see Fig. 2 2. Expression of zebrafish eIF4E-1A and eIF4E-1B in different tissues. A, expression of RNA. Total RNA (1 g) from samples of the indicated tissue was reverse-transcribed and used as template for PCR, with primers specific for either eIF4E-1A or eIF4E-1B cDNAs. Amplicons were resolved in TAE-agarose gels and stained with ethidium bromide. The size of the expected amplicons obtained by using eIF4E-1A or eIF4E-1B cDNAs as templates is indicated to the right of each panel. The target of each pair of primers used for the reactions is indicated to the left of each panel. B, analysis of eIF4E-1A and eIF4E-1B expression in different cell types by isoelectric focusing/ immunoblotting analysis. Recombinant eIF4E-1A and eIF4E-1B, and extracts from adult zebrafish tissues and zfl cells, were subjected to vertical slab gel isoelectric focusing and immunoblotting, essentially as described (46). Samples were focused in the presence of 9 M urea, 5% v/v ␤-mercaptoethanol, 2% w/v CHAPS (Fluka). In the top panel, a pH range of 5-8.5, using ampholytes 3-10 with 50 mM NaOH at the anode and 20 mM aspartic acid at the cathode, was used. In the bottom panel, samples were focused through a pH range of 8 -10, using a mixture of ampholytes 3-10 and 9 -11 (4:1), with 25 mM NaOH at the anode and 25 mM glutamic acid at the cathode. Proteins were electrophoresed at 2 mA/gel for 20 h. Proteins were transferred to PVDF membranes and probed with anti-TATKSGSTTKNRFVV. Recombinant human eIF4E-1 was included as a marker of pI 6.1 (only seen in top panel). tity), the mammalian factor can sustain growth of yeast deficient in eIF4E (15). The yeast strain, JOS003 (37), was used to assess the functionality of eIF4E-1A and -1B. JOS003 cells lack an endogenous yeast eif4e gene and express human eIF4E-1 under the control of the galactose-dependent and glucose-repressible GAL1 promoter (37). As a consequence, strain JOS003 is able to survive in medium containing galactose as carbon source but is not viable in medium containing glucose due to depletion of human eIF4E-1. Growth of JOS003 in glucose can be mediated by ectopic expression of a functional eIF4E, the regulation of which is under the control of a promoter active in the presence of glucose. cDNAs encoding the zebrafish eIF4E homologues were cloned into vectors with a URA selection marker, allowing expression from the constitutive glyceraldehyde-3-phosphate dehydrogenase (GPD) promoter. Following transfection and selection on media lacking uracil, cells containing control vector, or vectors for the expression of eIF4E-1A or eIF4E-1B, were streaked on plates of medium containing either galactose or glucose as carbon source (Fig. 4). It is evident that eIF4E-1A, but not eIF4E-1B, is able to rescue the JOS003 strain under conditions in which human eIF4E-1 is depleted. It is also apparent that growth of JOS003 cells in galactose medium is independent of the expression of either eIF4E-1A or eIF4E-1B. This genetic evidence suggests that zebrafish eIF4E-1A is functionally equivalent to prototypical eIF4E from other organisms.
Assessment of the Cap-binding Activity of eIF4E-1A and eIF4E-1B-The lack of complementation by eIF4E-1B suggests that this eIF4E homologue is unable to interact efficiently with the yeast translational machinery. One possibility is that the protein binds inefficiently to mRNAs. eIF4E has an intrinsic high affinity for the m 7 GTP cap of eukaryotic mRNAs, and this activity can be assayed by the ability to bind m 7 GTP-Sepharose. We asked whether the affinities of eIF4E-1A and eIF4E-1B for the cap structure are distinguishable from each other (Fig. 5). Ovary extract was applied to an m 7 GTP-Sepharose column, followed by extensive washing and elution with excess m 7 GTP. Eluted proteins were resolved by high-Tris SDS-PAGE (44), prior to electrotransfer and visualization of eIF4E-1A and eIF4E-1B using anti-TATKSGSTTKNRFVV (Fig. 5A). Although the ovary extract contains both eIF4E-1A and eIF4E-1B (lane 3), only eIF4E-1A bound to the m 7 GTPmatrix and was specifically eluted with m 7 GTP (lanes 4 -6). Equivalent results were obtained using recombinant zebrafish eIF4Es. When mixtures of recombinant eIF4E-1A and eIF4E-1B (Fig. 5B, lane 1) were subjected to m 7 GTP-Sepharose affinity chromatography, eIF4E-1A, but not eIF4E-1B, bound   FIG. 4. Ability of zebrafish eIF4Es to rescue the growth of S. cerevisiae, JOS003. The S. cerevisiae strain, JOS003, was transformed with the Ura-selectable vector, pRS416GPD, containing cDNAs encoding one of the following products: eIF4E-1A, the core of eIF4E-1A, the core of eIF4E-1A fused to the N terminus of eIF4E-1B, eIF4E-1B, the core of eIF4E-1B, or the core of eIF4E-1B fused to the N terminus of eIF4E-1A, as indicated. Following selection on SC medium with galactose lacking uracil and leucine, yeast from the resulting single colonies were transferred to YP-agar media containing G418 and either glucose (right) or galactose (left) as carbon source.  A, lanes 1 and 2, and in B, lanes 9 and 10. to the column and was specifically eluted with m 7 GTP (lanes 4 -6). Analyses using in vitro transcribed and translated 35 S-eIF4E-1A and 35 S-eIF4E1B gave similar results (data not shown). These data demonstrate that eIF4E-1A is a bona fide cap-binding protein and further suggest that eIF4E-1B is not functionally equivalent to human eIF4E-1.
eIF4E-1A, but Not eIF4E-1B, Interacts in Vitro with Human eIF4GI-The finding that eIF4E-1B is deficient in cap binding activity prompted us to consider that a possible function of eIF4E-1B is as a tissue-specific translational repressor. We hypothesized that eIF4E-1B could compete in vivo with eIF4E-1A for binding to eIF4GI. An interaction between eIF4E-1B and eIF4GI would preclude eIF4E-1A from incorporating into functional eIF4F complex, thus inhibiting cap-dependent translation. A mechanism analogous to this has been characterized in mammals; the 4E-BP family of translational repressors compete with cellular eIF4G for a common binding site on eIF4E-1 (22,24,33,34). eIF4E-1B could therefore be involved in a novel mechanism for control of translation by mimicking the eIF4G binding activity of eIF4E. We tested this hypothesis by asking if eIF4E-1B could interact with the eIF4E-1-binding domain of a vertebrate eIF4G homologue.
A polypeptide corresponding to residues 159 -614 of human eIF4GI (molecular mass ϳ45 kDa), which contained the eIF4E-1 interaction domain, was co-translated with S-tagged variants of either eIF4E-1A or eIF4E-1B in a reticulocyte cellfree translation system in the presence of [ 35 S]Met (Fig. 6,  lanes 1 and 5). After synthesis, reactions were incubated with S-protein-agarose. Following extensive washing, proteins bound to the matrix were eluted with SDS-PAGE sample buffer (lanes 3 and 4 and lanes 7 and 8). Fractions were resolved by high-Tris SDS-PAGE and analyzed by autoradiography. Whereas the eIF4GI polypeptide co-purified with eIF4E-1A, eIF4E-1B failed to interact with the same polypeptide at detectable levels.
These data, coupled with the yeast complementation data, demonstrate that like human eIF4E-1, zebrafish eIF4E-1A is able to interact with human eIF4GI in vitro and yeast eIF4G in vivo. In contrast, eIF4E-1B has a low affinity for eIF4GI and thus is unlikely to function as an efficient in vivo competitor of eIF4E-1A. This was corroborated by the finding that neither the addition of eIF4E-1A nor of eIF4E-1B at concentrations up to 5-fold that of the endogenous rabbit eIF4E-1 protein produced any significant effect on the kinetics of protein synthesis in the reticulocyte translation system (data not shown).
eIF4E-1A, but Not eIF4E-1B, Interacts with Zebrafish 4E-BP-Since eIF4Gs and 4E-BPs bind to eIF4E through common motifs, we anticipated that zebrafish eIF4E-1A, but not eIF4E-1B, would be targeted by the 4E-BP repressors. However, there remained the possibility that a homologue of eIF4E that is deficient in both cap binding activity and eIF4G interaction could potentially bind to 4E-BPs and work as a translational de-repressor.
By using RT-PCR, programmed with primers based on zebrafish ESTs encoding a 4E-BP, we amplified and cloned a zebrafish cDNA encoding a protein of 117 amino acids. An alignment of the predicted protein encoded by the zebrafish cDNA with human 4E-BP1-3, is presented in Fig. 7A. The close sequence similarity and conservation of the eIF4E-binding site strongly suggests that this cDNA (GenBank TM accession number AF332983) encodes a member of the 4E-BP family that is 68% identical to human 4E-BP3 (55) To assess the functionality of the cloned zebrafish 4E-BP, and to explore the possibility that eIF4E-1B is a de-repressor of 4E-BP-mediated inhibition of translation, we performed in vitro interaction assays with S-tagged variants of eIF4E-1A and eIF4E-1B in a manner analogous to the eIF4E/eIF4G interaction assay described above. After synthesis, reactions were incubated with S-protein-agarose. Following extensive washing, proteins bound to the matrix were eluted with SDS-PAGE sample buffer (Fig. 7B, lanes 4 and 5 and lanes 9 and 10). The data showed that zebrafish 4E-BP was enriched in the fraction of bound proteins in the presence of eIF4E-1A but not eIF4E-1B. These data demonstrate that the 4E-BP-mediated translational repression pathway is conserved in zebrafish. However, eIF4E-1B does not bind to zebrafish 4E-BP with an affinity that would be consistent with a role as a de-repressor of 4E-BP-mediated inhibition of translation.
The failure of eIF4E-1B to interact with eIF4G or 4E-BP could reflect the fact that the recombinant protein is simply misfolded and therefore cannot interact with eIF4G or 4E-BP. To address this possibility, a co-purification approach using His-tagged 4E-BP was implemented to look at the interaction of endogenous eIF4E-1B from zebrafish ovary with 4E-BP. A mixture of recombinant eIF4E-1A and -1B or zebrafish ovary extracts were incubated with His-tagged 4E-BP bound to anti-His/protein G-Sepharose beads (Fig. 8). The ovary extracts contained similar levels of eIF4E-1A and -1B (Fig. 8, lanes  4 -6). eIF4E-1A associated with the beads in a 4E-BP-dependent manner (Fig. 8, lanes 7-14). In contrast to eIF4E-1A, neither recombinant eIF4E-1B nor eIF4E-1B from zebrafish ovary bound to His-tagged 4E-BP (Fig. 8, lanes 11-14), consistent with the inability of in vitro translated eIF4E-1B to interact with 4E-BP (Fig. 7). To ensure that eIF4E-1A was binding to 4E-BP and not to the His tag in this assay, recombinant Histagged growth hormone was used as a substitute for His-4E-BP1 in control experiments. As expected, eIF4E-1A did not bind to this His-tagged control protein in place of His-4E-BP (data not shown). The higher molecular weight band visible in the lanes containing bead-bound proteins represents the light chain of the anti-His antibody that interacts with the secondary antibody of the immunoblotting procedure.

The Functional Divergence of eIF4E-1B Maps to a Region
Equivalent to the Core of eIF4E-1A The failure of eIF4E-1B to perform as a functional equivalent of eIF4E-1 in a variety of assays is surprising since the protein contains all the residues thought to be important for cap binding activity and eIF4G and/or 4E-BP interaction. By far the largest difference between the primary structures of functional eIF4Es (including zebrafish eIF4E-1A) and eIF4E-1B maps to the region N-terminal to the conserved core. As seen in Fig. 1, regions N-terminal to the conserved core of functional eIF4E proteins contain acidic residues that contribute to an overall acidic isoelectric point. However, the equivalent region of eIF4E-1B is extremely basic. Several lines of evidence suggest that this N-terminal region is not required for translation initiation; truncated mouse eIF4E-1, lacking sequence equivalent to this region, was used in structural studies to elucidate the molecular mechanisms of the interaction of the protein with the cap structure, eIF4G, and 4E-BP1 (16,19). Furthermore, a minimal yeast eIF4E, lacking the 30 N-terminal residues outside of the conserved core, is able to sustain cell viability (17). However, we hypothesized that the divergent, highly basic N terminus of eIF4E-1B might preclude the protein from functioning as a bona fide eIF4E. Furthermore, Fahrenkrug et al. (3) have shown that an alternative splicing variant of eIF4E-1B mRNA occurs in vivo that potentially encodes the conserved core region of eIF4E-1B, which may indeed be functional. To test this hypothesis, we engineered yeast expression constructs encoding the variants eIF4E-1A-(⌬1-33) and eIF4E-1B-(⌬1-32), equivalent to eIF4E-1A and eIF4E-1B, lacking the region N-terminal to the conserved core. We also engineered constructs encoding the hybrid molecules, eIF4E-1B-(⌬33-214)-eIF4E-1A-(⌬1-33) and eIF4E-1A(-⌬34 -215)-eIF4E-1B-(⌬1-32), in which the conserved core of eIF4E-1A or eIF4E-1B was fused to the N-terminal region of eIF4E-1B or eIF4E-1A, respectively. The constructs were then transferred into the yeast strain JOS003 and assessed for their ability to function as eIF4E in the yeast complementation assay (Fig. 4). to complement the lack of functional eIF4E, suggesting that substitutions of residues within the conserved core region of eIF4E-1B relative to that of eIF4E-1A are responsible for the failure of eIF4E-1B to function. The ability of eIF4E-1A-(⌬1-33) to support the growth of yeast lacking a functional eif4e gene demonstrates that 33 amino acids at the N terminus, representing the entire region N-terminal to the conserved core, are not essential for function. This is the most extensive N-terminal truncation product of eIF4E-1 to date that shows biological activity. DISCUSSION Our studies have shown that zebrafish eIF4E-1A appears to represent the functional equivalent of human eIF4E-1. In contrast to this, a related form, eIF4E-1B, constitutes a previously unknown type of vertebrate eIF4E-1 with unexpected properties and a restricted tissue distribution. The translation factor, eIF4E, functions to recruit mRNA to the ribosome by binding simultaneously to the 5Ј cap structure and to the scaffold translation factor eIF4G. Prototypical eIF4E is an essential translation factor; thus it is expressed in every cell type and is required for cell viability and cell cycle progression. With this in mind, we propose that zebrafish eIF4E-1A is a suitable candidate for eIF4E in this organism. Such an assignment is consistent with the expression data, the functional assays, and the ability of eIF4E-1A to support growth of the yeast strain, JOS003, conditionally deficient in eIF4E, in medium containing glucose. In contrast to eIF4E-1A, the data presented here show that neither recombinant nor endogenous eIF4E-1B exhibits any detectable activity in standard assays of eIF4E function. This is despite the fact that eIF4E-1B contains extensive similarity to known vertebrate eIF4Es and contains all amino acid residues shown to be required for binding the mRNA-cap structure, eIF4G, and 4E-BPs. Furthermore, data from expression analyses in zebrafish tissues, cultured cells, and at different stages of development suggest that the presence of eIF4E-1B is not essential for survival.
The position and phase of the intron/exon junctions in eIF4E-1B closely resemble those described for human eIF4E-1 (3,56,57). This, along with the high identity of the cDNA sequence of eIF4E-1B to eIF4E-1A, suggests that eIF4E-1B has arisen via a gene duplication event that occurred during vertebrate evolution. Interestingly, the 5Ј end of zebrafish eIF4E-1B and mammalian eIF4E-1 transcripts are quite distinct, and comparison of the mammalian eIF4E-1 promoter sequences with zebrafish eIF4E-1B promoter reveals no obvious conservation (2). It is possible that the gene duplication event that gave rise to eIF4E-1B excluded the 5Ј region of the ancestral eif4e gene or that selective pressure is responsible for the divergence in this region.
Although we have shown here that eIF4E-1B lacks three of the most prominent activities of eIF4E, cap binding, eIF4G binding, and 4E-BP binding, it remains to be addressed whether eIF4E-1B can interact with other eIF4E-binding proteins (e.g. 4E-T, maskin) (58,59). However, since all known proteins that associate with eIF4E seem to share a common eIF4E-binding domain, it seems unlikely that eIF4E-1B will interact with them. The novel N terminus of eIF4E-1B, with its potential nuclear localization sequence, may be indicative of its ability to undertake an alternate function such as RNA/ribonucleoprotein transport, RNA splicing, or localization of mRNAs during specific developmental stages or differentiated states. We have observed, however, that in normal fish ovary eIF4E-1B seems to be associated with both cytoplasmic and nuclear compartments, as assayed by cell fractionation (data not shown). Future studies will be required to clarify the intracellular distribution of eIF4E-1A and eIF4E-1B in intact cells. The expression patterns of eIF4E1-B in fish embryos suggest it may fulfill a specialized function in zebrafish germ cells or during development/differentiation.
Recently, cDNAs or genes encoding proteins that exhibit different degrees of identity to known prototypical eIF4E counterparts have been identified from a number of organisms (1,60,62,63). 3,4 Many possible examples of eIF4E-related proteins are still represented only by cDNA sequences from ESTs, or as genomic sequences, and their possible functions have not been reported or investigated. Several of these eIF4E-related factors have been functionally characterized, including the Schizosaccharomyces pombe, eIF4E2 (65), the eIF4E family members in C. elegans that interact with trimethylated cap structures (1), and the eIF4Elike protein 4EHP/nCBP/IFE-4 (1,60,61). Some of these studies demonstrate that the multiplicity of proteins related to prototypical eIF4E cannot simply be explained by functional redundancy. Although S. pombe eIF4E1 and eIF4E2 show almost identical cap binding characteristics, eIF4E2 alone cannot support viability (65). Likewise, in C. elegans, RNA interference experiments have shown that depletion of IFE-3 alone is lethal, suggesting that the other four forms cannot fulfill the same function or do not exhibit the same expression pattern (1). Other studies have shown that eIF4E-related proteins can fulfill specialized functions. Recognition of trimethylated cap structures by IFE-1, -2 and -5 in C. elegans is to date the best example of specialized function for eIF4E-related proteins (1,62,64). This particular function is likely to reflect the specialized nature of nematode mRNAs, the majority of which have trimethylated cap structures (65). A role for vertebrate eIF4Es (including eIF4E-1B) in the recruitment of mRNAs with trimethylated cap structures seems unlikely, since vertebrate mRNAs are not known to possess such structures. This possibility, however, has not been formally excluded.
Even though it is possible that the presence of multiple proteins related to eIF4E is the result of unproductive gene duplications giving rise to non-functional genes, the apparent conservation of at least some of these eIF4E-related factors suggests they serve important functions. One such example is 4EHP, conserved over vast evolutionary stretches from C. elegans, Drosophila, plants, and mammals and represented in EST clones across the eukaryotic phyla. Likewise, ESTs encoding eIF4E-1B-like proteins have been identified in Oncorhynchus mykiss, Xenopus laevis, Silurana tropicalis, and Sus scrofa (GenBank TM accession numbers BX303305, BQ398016, AL867687, and BF080360, respectively), and evidence for its presence in humans can be found in the human genome sequence. These observations, together with the data presented here in zebrafish, suggest that eIF4E-1B may have a role in all vertebrates.
The recognition of eIF4E-1A as the prototypical eIF4E in zebrafish will allow studies on the role of this regulatory translation factor during vertebrate development more easily than can be achieved in mouse. Similarly, the recognition of eIF4E-1B as a form not functional in cap binding or eIF4G/ 4E-BP binding will provide unique opportunities for investigating the relationship between structure and function in eIF4E family members. Furthermore, further studies of eIF4E-1B should allow the uncovering of additional roles for eIF4E family members.