Characterization of Translation Initiation Factor 5 (eIF5) from Saccharomyces cerevisiae FUNCTIONAL HOMOLOGY WITH MAMMALIAN eIF5 AND THE EFFECT OF DEPLETION OF eIF5 ON PROTEIN SYNTHESIS IN VIVO AND IN VITRO*

Eukaryotic translation initiation factor 5 (eIF5) interacts in vitro with the 40 S initiation complex (40 SzAUGzMet-tRNAfzeIF2zGTP) to mediate the hydrolysis of ribosome-bound GTP. In Saccharomyces cerevisiae, eIF5 is encoded by a single copy essential gene, TIF5, that encodes a protein of 45,346 daltons. To understand the function of eIF5 in vivo, we constructed a conditional mutant yeast strain in which a functional but a rapidly degradable form of eIF5 fusion protein was synthesized from the repressible GAL promoter. Depletion of eIF5 from this mutant yeast strain resulted in inhibition of both cell growth and the rate of in vivo protein synthesis. Analysis of the polysome profiles of eIF5-depleted cells showed greatly diminished polysomes with simultaneous increase in free ribosomes. Furthermore, lysates of cells depleted of eIF5 were dependent on exogenously added yeast eIF5 for efficient translation of mRNAs in vitro. This is the first demonstration that the TIF5 gene encodes a protein involved in initiation of translation in eukaryotic cells. Additionally, we show that rat eIF5 can functionally substitute yeast eIF5 in translation of mRNAs in vitro as well as in complementing in vivo a genetic disruption in the chromosomal copy of TIF5.


Eukaryotic translation initiation factor 5 (eIF5) interacts in vitro with the 40 S initiation complex (40
S⅐AUG⅐Met-tRNA f ⅐eIF2⅐GTP) to mediate the hydrolysis of ribosome-bound GTP. In Saccharomyces cerevisiae, eIF5 is encoded by a single copy essential gene, TIF5, that encodes a protein of 45,346 daltons. To understand the function of eIF5 in vivo, we constructed a conditional mutant yeast strain in which a functional but a rapidly degradable form of eIF5 fusion protein was synthesized from the repressible GAL promoter. Depletion of eIF5 from this mutant yeast strain resulted in inhibition of both cell growth and the rate of in vivo protein synthesis. Analysis of the polysome profiles of eIF5-depleted cells showed greatly diminished polysomes with simultaneous increase in free ribosomes. Furthermore, lysates of cells depleted of eIF5 were dependent on exogenously added yeast eIF5 for efficient translation of mRNAs in vitro. This is the first demonstration that the TIF5 gene encodes a protein involved in initiation of translation in eukaryotic cells. Additionally, we show that rat eIF5 can functionally substitute yeast eIF5 in translation of mRNAs in vitro as well as in complementing in vivo a genetic disruption in the chromosomal copy of TIF5.
Eukaryotic translation initiation factor 5 (eIF5), 1 in conjunction with GTP and other initiation factors, plays an essential role in initiation of protein synthesis (for reviews, see Refs. [1][2][3][4][5]. In vitro studies using purified initiation factors have shown that the overall initiation reaction proceeds with the initial binding of the initiator Met-tRNA f as the Met-tRNA f ⅐eIF2⅐GTP ternary complex to a 40 S ribosomal subunit followed by the positioning of the 40 S preinitiation complex (40 S⅐eIF3⅐Met-tRNA f ⅐eIF2⅐GTP) at the initiation AUG codon of the mRNA to form the 40 S initiation complex (40 S⅐eIF3⅐mRNA⅐Met-tRNA f ⅐eIF2⅐GTP). The initiation factor eIF5 then interacts with the 40 S initiation complex to mediate the hydrolysis of ribosome-bound GTP. Hydrolysis of GTP causes the release of eIF2⅐GDP (and P i ) from the 40 S ribosomal initiation complex which is essential for the subsequent joining of the 60 S ribosomal subunit to the 40 S complex to form a functional 80 S initiation complex (80 S⅐mRNA⅐Met-tRNA f ) that is active in peptidyl transfer (6 -9). eIF5 was purified originally from rabbit reticulocyte lysates on the basis of its ability to stimulate in vitro translation of globin mRNA in a partially reconstituted system (10 -12). The purified protein was shown to be essential for the joining of 60 S ribosomal subunits to the 40 S initiation complex to form the 80 S initiation complex (7,13,14). In these initial studies the apparent molecular weight of the purified protein varied between 125,000 (10) and 150,000 -168,000 (11,12). In our laboratory, however, using an assay that measured directly the formation of an 80 S initiation complex by the joining of 60 S ribosomal subunits to a preformed AUG-dependent 40 S initiation complex, we purified eIF5 both from mammalian cells (15)(16)(17)(18) as well as from the yeast Saccharomyces cerevisiae (19). The purified mammalian and yeast proteins were shown to be monomers that migrated on SDS gels with an apparent molecular weight of about 58,000 and 54,000, respectively (15)(16)(17)(18)(19). More recently, we have cloned and expressed the mammalian cDNA encoding functional eIF5 of calculated M r ϭ 48,926 (20 -22). The S. cerevisiae gene encoding eIF5, designated TIF5, has also been characterized and shown to be a single copy essential gene that encodes a protein of 45,346 daltons (23). The purified mammalian protein has also been used to characterize in vitro the eIF5-mediated subunit joining reaction using AUG-dependent 40 S initiation complex as a substrate (8,9,17,18,24). However, the involvement of either the 49-kDa mammalian eIF5 or the 45-kDa yeast protein in translation of natural mRNAs in vivo or in vitro has not yet been defined. This is relevant in view of previous observations (1-5) that proteins, e.g. eIF5A (25), isolated on the basis of their ability to stimulate partial reactions in the initiation pathway, were later shown not to be involved in translation of mRNAs in vivo.
In this paper, we have used the S. cerevisiae system to further characterize the function of eIF5 in protein synthesis in vivo. We have constructed a conditional eIF5 expression system in which functional eIF5 was expressed from a transcription unit consisting of a ubiquitin gene cassette which acts as a protein destabilizing genetic element fused to the NH 2 terminus of the open reading frame of the yeast TIF5 gene under the transcriptional control of GAL10 promoter. Similar expression systems were previously used by Park et al. (26) and Kang and Hershey (25) as an effective way to rapidly deplete yeast cells of protein of interest. The effect of depletion of eIF5 on cell growth as well as protein synthesis in vivo and in vitro were then analyzed. The results of these experiments show that eIF5 plays an essential role in translation of mRNAs in vivo and in vitro. Additionally, the rat cDNA encoding functional eIF5 was expressed in yeast cells lacking TIF5. We demonstrate that the rat protein can functionally substitute for the homologous yeast protein in vivo.

EXPERIMENTAL PROCEDURES
Media, Growth Conditions, and DNA Manipulations-Yeast strains were grown at 30°C in either YPD medium (1% (w/v) yeast extract, 2% (w/v) Bacto-peptone, and 2% (w/v) dextrose) or in YPGal medium where 2% (w/v) galactose (glucose-free) replaced dextrose as the carbon source. Where indicated, haploid yeast cells were also grown in either synthetic complete medium containing galactose as the sole carbon source, designated SGal media (2% galactose, 0.67% Bacto-yeast nitrogen base without amino acids, 0.2% amino acid mixture) or in SD medium where 2% dextrose replaced galactose as the carbon source. For in vivo [ 35 S]methionine incorporation, the 0.2% amino acid mixture in either the SGal or the SD media did not contain methionine. These methionine-lacking media were designated SGal-Met and SD-Met, respectively. Yeast transformations and tetrad analysis were performed as described by Rose et al. (27). Methods for plasmid and genomic DNA preparations, restriction enzyme digestion, DNA ligation, cloning, and bacterial transformation were according to standard protocols (28).
Construction of Plasmids for Conditional Expression of Yeast eIF5 or Mammalian eIF5-A 2.8-kb yeast genomic fragment containing the entire eIF5 gene (TIF5) (23) was separately cloned into the XhoI/ HindIII sites of centromeric plasmids pRS316 (URA3-based) (29) and pRS315 (LEU2-based) (29) to yield pRS316-TIF5 and pRS315-TIF5, respectively. In these newly constructed plasmids the expression of yeast eIF5 is under the transcriptional control of its endogenous promoter present in the inserted 2.8-kb fragment. For expression of yeast and mammalian eIF5 under the control of galactose-inducible GAL1 promoter, we first isolated a 0.6-kb BamHI/EcoRI fragment containing GAL1-GAL10 promoter from plasmid pBM272 (30), and cloned it at the same sites of pRS315 to generate a new recombinant vector, pTM100. For cloning of the yeast eIF5 open reading frame into pTM100 vector, two primers were synthesized bearing the NH 2 -terminal and COOHterminal ends of yeast eIF5 coding sequences (23) that were flanked by BamHI and EcoRI restriction sites: NH 2 terminus, 5Ј-dGCGCGCG-GATCCATGTCTATTAATATTTGTAG-3Ј; COOH terminus, 5Ј-dATC-CGGAATTCCTATTCGTCGTCTTCTTC-3Ј. These primers were used in high fidelity-polymerase chain reaction using Pyrococcus DNA polymerase I (Stratagene) and the polymerase chain reaction product was digested with BamHI and EcoRI and cloned into the same sites of pGEM7Zf(ϩ) vector to yield pGEM-TIF5. DNA sequencing of the eIF5 coding region in the recombinant plasmid indicated error-free DNA synthesis. The 1.218-kb TIF5 coding region was then excised from pGEM-TIF5 by digestion with BamHI and XbaI and cloned into the same sites of pTM100 vector to generate the new recombinant plasmid pTM100-TIF5. For expression of mammalian eIF5 under GAL1 promoter, the recombinant rat eIF5 expression plasmid, pETIF5 (20,22), was digested with EcoRI, blunt-ended by treatment with klenow DNA polymerase in the presence of 4 dNTPs, and then cut with XbaI to generate the 1.34-kb fragment that contained the entire coding sequence of rat eIF5 cDNA, starting 42 base pairs upstream of the initiating ATG codon and 8 base pairs after the termination codon. This fragment was then cloned into the NotI (blunt-ended)/XbaI sites of pTM100 vector to yield the new recombinant plasmid pTM100-EIF5. We also constructed a plasmid for conditional expression of a rapidlydegradable form of yeast eIF5 fusion protein (see Fig. 1). For this purpose, the coding region of TIF5 was first fused in-frame to a lacI-flu segment present in the plasmid pGEM-flu (26) as follows. The vector pGEM-TIF5 was digested with BamHI and SphI and the 1.235-kb BamHI (blunt-ended)/SphI fragment containing the entire TIF5 coding region was cloned into the PstI (blunt-ended)/SphI site of the vector pGEM-flu to yield pGEM-flu-TIF5. This new vector was then cut with BamHI and SphI and the resulting 1.34-kb lacI-flu-TIF5 fragment was fused in-frame to the ubiquitin gene, UBI4 (26) under the control of the GAL10 promoter as follows. The plasmid pUB23 which contains the UBI4 gene was cut with EcoRI and BamHI and the resulting EcoRI/ BamHI fragment that encodes URA3 upstream activation sequence of the GAL promoter (UAS Gal )-ubiquitin (Ub)-X (where X is the trinucleotide codon for arginine) was ligated with the BamHI/SphI fragment that encodes the lacI-flu-TIF5 gene derived from the pGEM-flu vector. The resulting EcoRI/SphI fragment was then cloned into the EcoRI/ SphI site of the centromeric HIS3 vector pSE362 (26) to generate pUB-TIF5R (Fig. 1). Expression of yeast eIF5 from the GAL10 promoter in this plasmid initially generates a NH 2 -terminal ubiquitinylated eIF5 fusion protein, which is rapidly processed in yeast cells by a deubiquitination enzyme to yield free ubiquitin plus an eIF5 fusion protein having arginine (R) as the NH 2 -terminal amino acid (see Fig. 1).

Construction of Haploid Yeast Strains for Expression of Yeast and Mammalian eIF5
Proteins-The haploid yeast strain TMY101 (see Table I), containing the chromosomal copy of the TIF5 gene disrupted with TRP1 marker gene and harboring the plasmid pRS316-TIF5 that expresses yeast eIF5 from its own promoter, was constructed from the diploid strain W303(a/␣) following the one-step gene disruption method of Rothstein (31) as described for a similar disruption procedure with URA3 gene by Chakravarti and Maitra (23). The newly generated diploid strain (tif5::TRP1/TIF5) was then transformed with plasmid pRS316-TIF5 and Ura ϩ transformants were selected. One of the transformants was then sporulated at 30°C, and the resulting tetrads were dissected and spores were germinated on YPD plates followed by selection on SD-Trp-Ura plates. Only spores containing the tif5::TRP1 gene and harboring the plasmid, pRS316-TIF5 will germinate to yield Trp ϩ Ura ϩ colonies. This haploid strain, named TMY101 (Table I), was the host strain used for all plasmid shuffling experiments described in this study. The genotype of the TIF5 loci on the chromosome and on the plasmid of TMY101 was confirmed by Southern blot analysis (23) using relevant DNA restriction fragments as probes. For construction of haploid yeast strains for conditional expression of ubiguitinated eIF5 fusion protein, the strain TMY101 was transformed with LEU2-plasmid pRS315-TIF5, and Trp ϩ Ura ϩ Leu ϩ transformants were selected on SD plates and replica-plated onto another SD plate which contained uracil and 5-FOA to promote the loss of the URA3 plasmid, pRS316-TIF5. The yeast strain recovered from the 5-FOA plate, designated TMY203, has the disrupted chromosomal copy of TIF5 but now harboring pRS315-TIF5 as the complementing plasmid. This strain was then transformed with pUB-TIF5R that contains the TIF5 gene fused to the ubiquitin gene cassette (see Fig. 1). The resulting Leu ϩ His ϩ Ura ϩ transformants were then selected and grown in SGal-His-Ura media to promote the loss of the LEU2-plasmid, pRS315-TIF5 by segregation. The newly generated yeast strain, which has the disrupted chromosomal copy of the TIF5 gene, but harboring only the plasmid, pUB-TIF5R, was selected on appropriate SGal media. This strain, designated TMY201R, expresses R-eIF5 fusion protein (arginine at the NH 2 terminus of the mature protein ( Fig. 1).
Polysome Profile Analysis-The haploid yeast strains W303␣ or TMY201R were grown in YP medium containing 2% galactose (YPGal) to early logarithmic phase. The cells were harvested and 3 A 600 units of cells were resuspended in 100 ml of YPD medium supplemented with 0.4 mg/ml adenine sulfate and grown for about 10 h at 30°C. Following addition of cycloheximide (50 g/ml), cultures were rapidly chilled in an ice-water bath and centrifuged. The cells were washed twice with LHB buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 30 mM MgCl 2 , and 50 g/ml cycloheximide). The washed cells were then suspended in 1.2 ml of the LHB buffer and lysed by vortexing with an equal volume of glass beads. After adding an additional 0.5 ml of the same buffer, the cell lysates were clarified by centrifugation first at 12,000 ϫ g for 10 min, followed by recentrifugation at 17,600 ϫ g for another 10 min. Equivalent amounts of A 260 absorbing material (approximately 10 A 260 units in 200 l) were layered on 11 ml of 7-47% (w/v) sucrose gradients in either a low salt buffer (10 mM Tris acetate, pH 7.0, 12 mM MgCl 2 , 50 mM NH 4 Cl) or in a high salt buffer (10 mM Tris-HCl, pH 7.5, 15 mM MgCl 2 , 800 mM KCl (32)) and centrifuged at 40,000 rpm for 2.5 h at 4°C in a Beckman SW41 rotor. The gradients were fractionated in an ISCO density gradient fractionator and the absorbance profile at 254 nm was analyzed in an ISCO UA-5 absorbance monitor.
Cell-free Translation-The strain TMY201R was grown in YPGal supplemented with adenine sulfate (0.4 mg/ml) overnight at 30°C (A 600 Ϸ 1.0). Cells were harvested and suspended in YPD medium containing adenine sulfate (0.4 mg/ml) such that the initial absorbance at 600 nm was about 0.03 and grown at 30°C for about 18 h (about 2 generations in YPD medium) when growth was nearly completely arrested. Cells were then harvested and cell-free translation extracts were prepared and mRNA-dependent cell-free translation was performed using [ 35 S]methionine as the labeled amino acid as described by Hussain and Leibowitz (33).
Other Materials and Methods-Total RNA of S. cerevisiae was isolated as described by Rose et al. (27) while poly(A) ϩ RNA was isolated from total yeast RNA by chromatography on oligo(dT)-cellulose columns (Pharmacia Biotech Inc.) using the protocol supplied by the manufacturer. Luciferase mRNA was purchased from Promega and capped at the 5Ј end with vaccinia guanylylmethyltransferase as described (34). The vaccinia enzyme was a kind gift of Dr. Stewart Shuman of Sloan-Kettering Cancer Institute, New York. Purified eIF5 from rabbit reticulocyte lysates as well as recombinant rat eIF5 were isolated as described (18,22). Yeast eIF5 was purified from S. cerevisiae strain BJ926, as described by Chakravarti et al. (19). The purified yeast eIF5 preparation exhibited two polypeptide bands of 54 and 56 kDa upon SDS-polyacrylamide gel electrophoresis. Each of the two polypeptides was found to contain eIF5 activity and they were immunologically related to each other (19). The preparation of specific rabbit antisera against purified rabbit and yeast eIF5 and their affinity purification were as described (17,19). The procedure used for preparation of yeast cell lysates for immunoblot analysis was adapted from that described by Sachs and Davis (35).

Construction and Expression of a Conditional Mutant
Form of eIF5 in S. cerevisiae-To understand the function of eIF5 in the yeast S. cerevisiae, we depleted yeast cells of endogenous eIF5 and measured the effects of such depletion on cell growth as well as protein synthesis in vivo and in vitro. We initially constructed a haploid yeast strain in which the chromosomal copy of the TIF5 gene was inactivated by disruption with TRP1 marker gene and the essential eIF5 function was provided by maintenance of a centromeric plasmid that contained the open reading frame of the TIF5 gene under the transcriptional control of the GAL1 promoter. This yeast strain, designated TMY201 (Table I), as expected, grew on YPGal plates but did not form detectable colonies on YPD plates (data not shown). To deplete yeast cells of eIF5, TMY201 cells were grown in galactose-containing medium to early exponential growth phase and then transferred to glucose medium to turn off transcription of the TIF5 gene from the GAL1 promoter and thus, in essence turning off synthesis of new eIF5 protein in cells. Surprisingly, we observed that under these conditions, cell growth continued normally for five to six generations before declining. Immunoblot analysis of cell lysates showed that while the level of eIF5 progressively decreased with time following transfer from galactose to glucose medium, a significant level of eIF5 was still detectable in cell lysates even after six generations of growth (data not shown). Presumably, the rate of degradation of preexisting eIF5 protein was not rapid enough to deplete yeast cells of detectable levels of eIF5 within one or two generations. In view of the observation (9, 17) that eIF5 acts catalytically in initiation of protein synthesis in vitro, the presence of very low levels of the protein in yeast cells may be sufficient to sustain slow cell growth for many generations even in the absence of new eIF5 synthesis. Furthermore, because depletion of eIF5 in TMY201 cells required several generations of growth under repressing conditions, we were concerned of the possibility that many secondary effects might be induced under these conditions that could complicate interpretation of results obtained with such eIF5-depleted cells.
To further lower the cellular level of eIF5 and to ensure more rapid depletion of eIF5 from yeast cells following transfer to repressing conditions, we constructed a haploid yeast strain TMY201R as described under "Experimental Procedures" in which the chromosomal copy of the TIF5 gene was inactivated with TRP1 marker gene and the strain was kept viable by maintenance of the URA3-centromeric plasmid that harbors a conditional eIF5 expression system in which eIF5 was expressed from a genetic cassette designed to destabilize the eIF5 protein. In such a system, constructed by utilizing a strategy first used by Park et al. (26) and later by Kang and Hershey (25), eIF5 was expressed from a transcription unit consisting of a ubiquitin gene cassette (Ub-X-lacI-TIF5) which acts as a protein destabilizing element fused to the NH 2 terminus of the open reading frame of the TIF5 gene under the transcriptional control of GAL10 promoter (Fig. 1). The eIF5 fusion protein synthesized from this construct contains ubiquitin at the NH 2 terminus followed by a 31-amino acid segment of the lacI repressor which acts as a recognition element for ubiquitin-dependent protein degradation by NH 2 -terminal recognition pathway (36,37). Deubiquitination of this eIF5 fusion protein exposes arginine (R) as the NH 2 terminus amino acid (see Fig.  1) and should lead to rapid degradation of the protein by the N-end rule (36,37). As expected the strain TMY201R grew well on plates containing galactose as the carbon source while it did not form detectable colonies on plates containing glucose as the sole source of carbon (data not shown). When growth of the strain TMY201R was monitored in liquid cultures containing galactose as the carbon source, it grew with doubling time of 3.4 h comparable with 3.2 h for the wild-type W303␣ strain (Fig. 2, panel A). However, when the exponentially growing cultures of these strains were shifted from galactose to glucosecontaining medium to repress transcription of the TIF5 gene from the GAL10 promoter, the strain W303␣, in which eIF5 expression occurred from its own promoter, as expected, grew well with a doubling time of about 2 h (Fig. 2, panel B). In contrast, the growth rate of TMY201R, in which expression of eIF5 is under the control of GAL10 promoter, began to decrease after about one generation and the growth rate was drastically reduced in about two generations (17 h in glucose medium) (Fig. 2, panel B).
The level of eIF5 protein in cell lysates following the shift from galactose to glucose medium was also analyzed using affinity-purified anti-yeast eIF5 antibodies as probes. Fig. 2  (lower panel) shows that the level of eIF5 in cell lysates prepared from non-repressed W303␣ cells remained constant during the growth period. In contrast, there was a rapid disappearance of R-eIF5 fusion protein in strain TMY201R. After about 17 h there was no detectable level of R-eIF5 fusion protein (Fig. 2, lower panel). These results demonstrate that in strain TMY201R, which synthesizes eIF5 fusion protein having Arg as the NH 2 terminus, a more rapid depletion of eIF5 can be achieved following repression of transcription of the TIF5 gene cassette from the GAL10 promoter, and this depletion of eIF5 This work level in cells correlated well with the decrease in the rate of cell growth.
Analysis of Protein Synthesis in Vivo upon Depletion of eIF5-We used strain TMY201R to investigate the effect of depletion of eIF5 on protein synthesis in yeast cells. For this purpose, an exponentially growing culture of TMY201R cells was transferred from galactose to glucose-containing medium and the rate of protein synthesis was then monitored by measuring the incorporation of [ 35 S]methionine into protein with 5-min pulses of the labeled amino acid into 1 A 600 unit of cells over the growth period. Fig. 3 shows that there was a clear correlation between the rate of protein synthesis in vivo with the growth rate (shown in Fig. 2) of TMY201R cells. During the first 7.5 h of growth following the shift from galactose to glucose medium, as TMY201R cells continued to grow exponentially due to the presence of pre-existing eIF5 in cells, the rate of protein synthesis remained fairly constant. However, between 7.5 and 10 h, there was a marked decrease in the rate of protein synthesis. After about 17-18 h (ϳ2 generations in glucose medium), the protein synthesis rate was only about 20% of the initial rate. This is also the time when yeast cells were nearly completely depleted of eIF5 as judged by immunoblot analysis (data not shown here, see Fig. 2). In contrast, when TMY201R cells were allowed to grow in galactose medium, or wild-type W303␣ cells were grown in glucose (SD) medium, the rate of protein synthesis remained fairly constant during the entire exponential growth period of about 18 h (Fig.  3). These results show that eIF5 is required for translation in yeast cells.
For a more detailed analysis of protein synthesis inhibition in eIF5-depleted cells, the polyribosome content of exponen-tially growing cultures of yeast cells containing eIF5 (wild-type W303␣ cells or TMY201R cells growing in galactose medium) and growth-arrested TMY201R cells in YPD medium that are depleted of eIF5 was analyzed by sucrose gradient centrifugation (Fig. 4). The gradient profile from eIF5-depleted TMY201R cells growing in glucose (YPD) medium showed a marked relative reduction in the amount of large polyribosomes with a proportionate increase in the amount of monoribosomes (Fig. 4, panel E) when compared with wild-type W303␣ cells growing in YPD medium (Fig. 4, panel B) or TMY201R cells growing in galactose medium (Fig. 4, panel D). When equivalent amounts of cell extracts were also analyzed in sucrose gradients containing 0.8 M KCl, the majority of the accumulated monoribosomes in eIF5-depleted TMY201R cells were found to be dissociated into subunits (Fig. 4, panel F), indicating that they are neither bound to mRNA nor arise due to nonspecific RNase activity in extracts. The observed decrease in polyribosome content with an increase in the amount of free 80 S ribosomes and free 60 S and 40 S ribosomal subunits when eIF5 was depleted, indicates a block in translation initiation, and not in the elongation or termination of protein synthesis.
eIF-5-dependent Cell-free Yeast Translation System-We used eIF5-depleted TMY201R cells to prepare a yeast cell-free translation system to demonstrate the requirement of eIF5 for translation of mRNAs in vitro (Fig. 5). Cell-free translation extracts prepared (33) from exponentially growing cultures of wild-type W303␣ cells in glucose medium was highly active in translation of total yeast poly(A) ϩ RNA without any exogenously added protein factors (Fig. 5, panel A). In contrast, when similar cell-free extracts were prepared from TMY201R cells that were initially grown in galactose-containing medium and then shifted to glucose medium and grown until the growth of cells virtually stopped (about 18 -20 h) due to depletion of eIF5, they were rather inactive in translation of both total yeast poly(A) ϩ RNA (Fig. 5, panel B) as well as luciferase mRNA (Fig. 6). Translation could be restored in these lysates  (panel B). Cell growth was monitored by measuring absorbance at 600 nm. Lower panel, at the indicated times following the shift from SGal to SD medium, cell lysates were prepared from strains W303␣ and TMY201R as described under "Experimental Procedures." Approximately 10 g of protein from each cell lysate were electrophoresed through SDS-10% polyacrylamide gel electrophoresis followed by electrophoretic transfer to a polyvinylidene difluoride membrane. The blot was then probed with affinity-purified rabbit anti-yeast eIF5 antibodies as described under "Experimental Procedures." Goat anti-rabbit IgG coupled to alkaline phosphatase was used to detect the binding of the primary antibody to eIF5 in the blots. It should be noted that the wild-type eIF5 migrated as a doublet with an apparent M r of 54,000 and 56,000 (19) while the R-eIF5 fusion protein migrated with an apparent M r of 62,000. The lower molecular weight immunoreactive polypeptides are presumably degradation products of R-eIF5. by the addition of purified yeast eIF5 (Figs. 5B and 6). In the absence of mRNA, addition of eIF5 had virtually no effect ( Fig.  5B and Fig. 6). Recombinant rat eIF5 was also able to substitute for yeast eIF5 in restoring translation in such eIF5-depleted cell-free extracts (Fig. 5B). However, the rat protein was about 50% as effective as an equimolar amount of yeast eIF5 in restoring translation in eIF5-depleted translation extracts (Fig. 5, panel B, compare stimulation of translation by 10 ng of each protein factor). In contrast to strong dependence of eIF5 for translation of mRNAs in these cell extracts, extracts prepared from wild-type yeast cells which were highly active in translation of poly(A) ϩ RNA, was not stimulated by exogenously added eIF5 (Fig. 5A), indicating that wild-type yeast extracts contained saturating levels of eIF5.
Mammalian eIF5 Can Functionally Substitute for the Homologous Yeast Protein in the Yeast S. cerevisiae-Results presented above showed that recombinant rat eIF5 can substitute for yeast eIF5 in supporting protein synthesis in in vitro yeast translation system. We employed the plasmid shuffling technique to determine if the wild-type mammalian (rat) eIF5 protein can functionally substitute for the corresponding yeast protein in vivo. For this purpose, we constructed the haploid S. cerevisiae strain TMY101 as described under "Experimental Procedures" and Table I. The strain TMY101 carries inactive TIF5 disrupted with the TRP1 marker gene and is kept viable by maintenance of a centromeric URA3 plasmid, pRS316-TIF5 that contains the yeast wild-type TIF5 gene under the transcriptional control of its own promoter. For expression of rat eIF5 in TMY101, we cloned the open reading frame of rat eIF5 cDNA under the control of yeast galactose-inducible GAL1 promoter in the vector, pTM100 that contains LEU2 as a selectable marker to yield the plasmid, pTM100-EIF5 (see "Experimental Procedures"). As controls, we constructed two additional plasmids, pTM100-TIF5 and pRS315-TIF5, in which the expression of yeast eIF5 is under control of the GAL1 promoter and endogenous yeast eIF5 promoter, respectively ("Experimental Procedures"). The host strain TMY101 was transformed individually with each of the three recombinant eIF5 expression plasmids and the parental vector, pRS315. Trp ϩ Ura ϩ Leu ϩ transformants were selected on SGal plates (Fig. 7A, left panel) and replica-plated onto another similar plate which also contained uracil and 5-fluoroorotic acid (5-FOA) to promote the loss of the original URA3 plasmid, pRS316-TIF5. Fig. 7 (panel A) shows that cells transformed with control plasmids, pRS315-TIF5 and pTM100-TIF5 (that express yeast eIF5 from endogenous eIF5 promoter and GAL1 promoter, respectively), grew on 5-FOA plates, as expected. Cells transformed with pTM100-EIF5 that express rat eIF5 from GAL1 promoter were also able to lose the URA3 plasmid, pRS316-

FIG. 3. Inhibition of translation in eIF5-depleted yeast cells.
Exponentially growing cultures of W303␣ and TMY201R in galactosecontaining methionine labeling medium (SGal-Met) were harvested and an equal number of cells of each cell type (approximately 3 A 600 units) was then inoculated into 100 ml of glucose-containing methionine labeling medium (SD-Met). A third SGal-Met medium was also inoculated with TMY201R cells as a control. Each culture was grown with vigorous shaking at 30°C. At the indicated times, 1 A 600 unit of cells from each culture was harvested, and suspended separately either in 300 l of SD-Met medium containing 100 Ci of [ 35 S]methionine (1175 Ci/mmol) or in 300 l of SGal-Met medium containing 100 Ci of [ 35 S]methionine. Cells were shaken for 5 min at 30°C for incorporation of [ 35 S]methionine into cellular proteins followed by the addition of 1 ml of a "stop buffer" containing 1.2 mg/ml non-radioactive methionine and 0.5 mg/ml cycloheximide to terminate in vivo incorporation of 35 S radioactivity. The amount of [ 35 S]methionine incorporated into cellular proteins was then measured by an adaptation of the method of Kang and Hershey (25) as follows. Yeast cells in frozen incubation mixtures were lysed in 1.85 N NaOH containing 0.2 M 2-mercaptoethanol and proteins precipitated by the addition of hot trichloroacetic acid to 10% final concentration. After centrifugation, the precipitate was washed twice in acetone, dissolved in 200 l of 1% SDS solution and heated at 95°C for 5 min. An aliquot of the SDS extract was counted for 35 S radioactivity in a liquid scintillation spectrometer to determine the amount of [ 35 S]methionine incorporated into proteins. The concentration of protein in the SDS extract was determined by the micro-BCA protein assay method using reagents obtained from Pierce Chemical Co. The rate of protein synthesis at each time point was calculated as counts/min of 35 S radioactivity incorporated per 1 g of protein per min.

FIG. 4. Analysis of the polyribosome profile of wild-type and eIF5-depleted yeast cells.
Exponentially growing cultures of wildtype W303␣ and TMY201R in YPGal media containing adenine sulfate (0.4 mg/ml) were harvested and suspended in glucose (YPD) media such that initial absorbance of each culture at 600 nm was about 0.03. At about 10 h after the shift, the cells from each culture were harvested, lysed, and about 10 A 260 units of each lysate were subjected to 7-47% (w/v) sucrose gradient either in low salt, no KCl buffer (panels B and E) or in high salt, 0.8 M KCl buffer (panels C and F). An equivalent amount of cell lysates of each strain prepared from cells prior to transfer from YPGal to YPD media were also analyzed (panels A and D). Each gradient was fractionated in an ISCO gradient fractionator and the absorbance profile at 254 nm was analyzed in an ISCO UA-5 absorbance monitor. It should be noted that for analysis of the gradient fractions in the left panels (panels A, B, and C), the UV absorbance monitor was set at twice the "sensitivity" as that used for analysis of the right panel gradients.
TIF5, and could thus grow on 5-FOA plates. In contrast, cells transformed with the vector pRS315 failed to grow, as expected. These results show that mammalian eIF5 when expressed from the GAL1 promoter in an extrachromosomally replicating plasmid can substitute for yeast eIF5 function in yeast cells that do not express yeast eIF5.
Further confirmation of the above conclusion came from the following observations. First, all three yeast strains recovered individually from 5-FOA plates and harboring plasmids pTM100-TIF5 (strain TMY201), pTM100-EIF5 (strain TMY202), and pRS315-TIF5 (strain TMY203) required uracil for growth, thereby confirming that these strains have lost the original URA3 plasmid, pRS316-TIF5 (data not shown). Sec-ond, all strains except the one (TMY203) expressing yeast eIF5 from its own promoter in the plasmid pRS315-TIF5 failed to grow on plates containing glucose as the sole carbon source (Fig. 7, panel B). Final confirmation that the growth of the yeast strain harboring the plasmid pTM100-EIF5 (strain TMY202) is due solely to expression of rat eIF5 that came from immunoblot analysis using monospecific anti-rabbit and antiyeast eIF5 antibodies as probes (Fig. 7, panel C). These antibodies are highly specific for their own antigen and do not cross-react either with heterologous eIF5 or any other yeast protein (Fig. 7C, lanes a and b, upper and lower panels). When lysates of strain TMY202 which express rat eIF5 from the GAL1 promoter in plasmid pTM100-EIF5 were subjected to immunoblot analysis, a strong immunoreactive polypeptide band (apparent M r ϭ 58,000) was observed only with antirabbit eIF5 antibodies (Fig. 7C, upper panel, lane f). Anti-yeast eIF5-antibodies did not recognize any polypeptide in this cell lysate (Fig. 7C, lower panel, lane f). As expected, lysates of strain TMY201 harboring plasmid pTM100-TIF5 and wild-type strain W303␣ reacted strongly only with anti-yeast eIF5 antibodies yielding an immunoreactive polypeptide band corresponding to the molecular mass of yeast eIF5 (56 and 54 kDa doublets) (lanes e and c of Fig. 7C, lower panel). Anti-mammalian eIF5 antibodies did not recognize any polypeptide in these cell lysates (lanes e and c of Fig. 7C, upper panel). However, a yeast strain carrying null mutations in the chromosomal copy of TIF5 and harboring both plasmids, pRS316-TIF5 and pTM100-EIF5, expressed both mammalian and yeast eIF5 (lane d of Fig. 7C, upper and lower panels). These results firmly establish that yeast cells without intact endogenous eIF5 loci but expressing mammalian eIF5 can maintain cell growth and viability. Thus, mammalian eIF5 expressed from an extrachromosomally replicating plasmid can functionally substitute for the homologous yeast protein in vivo although such cells did not grow as rapidly as either the wild-type W303␣ cells or the strain TMY201 overexpressing yeast eIF5 from the GAL1 promoter (Fig. 7, panel D) under "Cell-free Translation" except that where indicated, 5 g of capped luciferase mRNA was used as the template RNA and 100 ng of purified yeast eIF5 were also added. Following incubation at 25°C for 30 min, an aliquot (5 l) of each reaction mixture was assayed for luciferase activity as described by Russell et al. (39).
showed that while the molar concentration of eIF5 in strain TMY201 in which yeast eIF5 was expressed from the GAL1 promoter was nearly 4-fold higher relative to wild-type W303␣ cells, the concentration of rat eIF5 expressed from the GAL1 promoter in strain TMY202 was similar to that of the W303␣ strain (data not shown). DISCUSSION eIF5 was isolated in our laboratory both from mammalian cell extracts (15)(16)(17)(18) as well as from the yeast S. cerevisiae (19) based on an in vitro assay that measured the ability of the factor to mediate the joining of 60 S ribosomal subunits to a preformed AUG-dependent 40 S initiation complex (40 S⅐AUG⅐Met-tRNA f ⅐eIF2⅐GTP) to from the functional 80 S initiation complex. Studies with the purified factor as well as subsequent cloning and characterization of the mammalian cDNA and yeast gene encoding eIF5 (20 -23) have led to the demonstration that both mammalian and yeast proteins are monomers of calculated M r 48,926 and 45,346, respectively. For reasons summarized below, it is, however, important to show that the 45-49-kDa eIF5, purified on the basis of in vitro partial initiation reaction assay, is indeed involved in the translation of natural mRNAs in vivo and in vitro. First, prior to our work, eIF5 was isolated from mammalian cells in several laboratories (11)(12)(13)(14) based on its ability to stimulate globin mRNA-directed protein synthesis in a partially reconstituted translation system. The purified factor was reported to be a monomer of about 150,000 -160,000 kDa (10 -14) and shown to be specifically required for the joining of the 60 S ribosomal subunits to the 40 S initiation complex formed with natural mRNA to form the 80 S initiation complex. Second, there are reports in literature of several protein factors that were isolated from rabbit reticulocyte lysates based on the in vitro activity of each protein in stimulating a partial reaction in the initiation pathway (for review, see Refs. [1][2][3][4][5]. Although these protein factors were called translation initiation factors, the role of many of these proteins, e.g. eIF5A (formerly called eIF4D), eIF6, CoeIF2A, CoeIF2B, in initiation of protein synthesis in vivo is far from clear. For example, initiation factor eIF5A, a 16 -18-kDa highly conserved protein in eukaryotes (38,40,41), was isolated from ribosomal salt-wash proteins of rabbit reticulocytes based on its activity to stimulate the formation of methionyl puromycin by interaction of Met-tRNA f bound to the 80 S initiation complex with puromycin (12,41,42). Based on the activity of this protein in the above in vitro assay, eIF5A was thought to be a translation initiation factor required for the synthesis of the first peptide bond following 80 S initiation complex formation. However, subsequent cloning of the S. cerevisiae genes encoding eIF5A (43,44) and studies on protein synthesis in vivo in eIF5A-depleted yeast cells (25) showed that while eIF5A is an important protein required for yeast cell growth and viability, it may not be a translation factor required for global protein synthesis in yeast cells. These observations lend strong support to the view that the requirement of a protein isolated on the basis of an in vitro partial reaction assay must be demonstrated for protein synthesis in vivo before it can be regarded as a translation factor.
In this paper, we have used the S. cerevisiae system to investigate the function of the 45-kDa yeast eIF5 in protein synthesis in vivo and in vitro. We have used a conditional eIF5 expression system to rapidly deplete eIF5 from yeast cells. Data presented in this paper clearly show that as yeast cells were depleted of eIF5, the rate of protein synthesis in these cells was inhibited in parallel. Analysis of the polysome profiles of eIF5-depleted cells showed greatly diminished polysomes with the simultaneous increase in 80 S ribosomes (Fig. 4), which suggests that eIF5 plays an essential role in the initiation phase of protein synthesis. Furthermore, lysates of cells depleted of eIF5 were inactive in translation of mRNAs in vitro. Addition of 45-kDa purified yeast eIF5 was able to restore translation in these cell lysates. These results show that the TIF5 gene product, a protein of 45,346 daltons, is a translation factor involved in initiation of protein synthesis. Furthermore, in agreement with the 39% identity and 60% homology in amino acid sequence between mammalian and yeast eIF5 (21), we observed that rat eIF5 can functionally substitute for yeast eIF5 in restoring in vitro translation of mRNAs in eIF5-depleted yeast cell-free translation extracts (Fig. 5). In accord- FIG. 7. Growth of haploid yeast transformants expressing yeast or rat eIF5 from the GAL1 promoter. A, haploid yeast strain TMY101 harboring the URA3 plasmid pRS316-TIF5 was transformed separately with different recombinant eIF5 expression plasmids as indicated. Transformants were patched onto SGal plates lacking tryptophan, leucine, and uracil (SGal-Trp-Leu-Ura) and then replica-plated onto both SGal-Trp-Leu-Ura plates (left panel) and SGal plates containing 5-FOA and uracil (SGal-Trp-LeuϩUraϩ5-FOA) (right panel). Cells were allowed to grow on these plates for 3 and 5 days, respectively. B, haploid yeast cells recovered from 5-FOA plates each containing different eIF5 expression plasmids, pTM100-TIF5 (strain TMY201), pTM100-EIF5 (strain TMY202), and pRS315-TIF5 (strain TMY203) were grown on a SGal-Trp-Leu plate (left panel), or on a SD-Trp-Leu plate (right panel) at 30°C for 3 days. C, immunoblot analysis of eIF5 in lysates of yeast cells expressing either yeast or rat eIF5 or both from the recombinant plasmids. Yeast cells harboring different recombinant eIF5 expression plasmids were grown to mid-logarithmic phase in synthetic medium containing 2% galactose as the sole source of carbon. Cell lysates were prepared as described under "Experimental Procedures" and analyzed by Western blotting using either anti-rabbit eIF5 antibodies (upper panel) or anti-yeast eIF5 antibodies (bottom panel). Lanes a and b, purified recombinant rat eIF5, and purified yeast eIF5, respectively; lane c, extracts from W303␣ cells; lanes d-f, extracts from tif5::TRP1 yeast cells harboring eIF5 expression plasmids as follows; lane d, pRS316-TIF5 and pTM100-EIF5; lane e, pTM100-TIF5; lane f, pTM100-EIF5. D, growth curves of yeast strains expressing either yeast eIF5 or mammalian eIF5 from GAL1 promoter. The indicated yeast strains were grown in synthetic media containing 2% galactose as the only carbon source and the growth was monitored by measuring absorbance at 600 nm in a Beckman spectrophotometer. The growth curve of wild-type parental yeast strain W303␣ is also shown. ance with these results, we also observed that in ⌬TIF5 yeast cells in which no functional eIF5 was synthesized due to disruption of the essential chromosomal copy of the TIF5 gene, expression of either rat or S. cerevisiae eIF5 from centromeric plasmids under the transcriptional control of the GAL1 promoter can support growth and viability of yeast cells. It is to be noted, however, that the growth rate of ⌬TIF5 yeast cells expressing mammalian eIF5 from the GAL1 promoter is about 40 -50% of that exhibited by either the wild-type W303␣ strain or by a yeast strain expressing yeast eIF5 from the GAL1 promoter. This difference in growth rate between the two eIF5 expression systems presumably reflects subtle evolutionary differences between the components of the translation machinery of the unicellular yeast and the multicellular mammalian systems. We should also point out that among many translation initiation factors that have been tested for the ability of their cDNAs to functionally substitute for the corresponding yeast gene in yeast cells, only eIF4A has not been found to complement (for review, see Ref. 5). These observations lend strong support to the view that the basic mechanism of initiation of translation is highly conserved from the unicellular yeast to mammals. Thus, the information gathered from molecular genetic studies of translation in the yeast S. cerevisiae will complement in vitro biochemical studies in the mammalian system leading to a better understanding of the process of translation and its regulation in eukaryotic cells.
Finally, availability of both a cell-free yeast translation system dependent on exogenous eIF5 for efficient in vitro translation of mRNAs and a ⌬TIF5 yeast strain in which functional eIF5 is produced in vivo from the TIF5 gene under the control of GAL promoter should be powerful tools to analyze structurefunction studies of eIF5 in vivo and in vitro.