Sequence Analysis of the Translational Elongation Factor 3 from Saccharomyces cerevisiae*

The gene YEF-3 encoding the elongation factor for protein synthesis in Saccharomyces cerevisiae is an essential gene as shown by one-step gene disruption and is located on chromosome XII as determined by orthogonal field alternation gel electrophoresis. The nucleotide sequence of the gene was determined from a sequential series of subclones generated from the YEF-3 gene cloned into bacteriophage M13. The HOMOLl sequence and the RPG box, which are con- sidered to be enhancer elements involved in coordinate regulation of transcription of the genes coding for yeast ribosomal proteins and protein synthesis factors, are found in the 6’-flanking region of the gene. A dyad symmetry that enables hairpin loop formation in the DNA molecule is found in the 3’-terminal at the termination site of transcription. An open reading frame of 3132 nucleotides codes for a deduced protein of 115,860 Da. A striking feature of the elongation factor 3 deduced polypeptide is the internal repeat of a region with approximately 200 amino acids which includes an ATP-binding site and shares similarity with some transport and drug-resistant proteins. Another char- acteristic is the presence of a highly

is the presence of a highly charged C-terminal region composed of three basic polylysine blocks, suggesting interaction with RNA. The sequence supports the hypothesis that YEF-3 encodes a protein synthesis factor and suggests that its main role may be to transduce nucleoside triphosphate energy into mechanical energy for translocation during translation.
The elongation factor 3 (EF-3)' was discovered by collaborators (1976, 1977) to be required for in vitro protein synthesis by yeast ribosomes in addition to elongation factors 1 and 2 which are interchangeable with the equivalent factors from other eukaryotes. EF-3 has been purified from * This work was supported by the California College of Medicine Support Foundation, Grace Garvey Trust; by Natural Science Foundation of China Grant 3870123; by a grant to support young outstanding teachers from State Commission of Education, China; by Grant BC 87-68 from Third World Academv of Sciences: and bv a uostdoctoral training grant from Conselho" National de Desenvoivmento Cientifico, CNPq, Brazil. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) 505197.
several fungi species and found to consist of a single polypeptide chain with a molecular weight of about 125,000 (Dasmahapatra and Chakraburtty, 1981;Uritani and Miyazaki, 1988a). The evidence that EF-3 is a protein synthesis factor is compelling.
In vitro protein synthesis systems directed by poly(U) and natural mRNA exhibits a 30-fold dependence on elongation factor 3 (Skogerson and Wakatama, 1976;Skogerson and Engelhardt, 1977;Herrera et al., 1984). Immuno-inactivation experiments with monoclonal antibodies specific for EF-3 demonstrated that EF-3 was required for protein synthesis (Hutchison et al., 1984). A temperature-sensitive yeast mutant blocked in the elongation phase of protein synthesis was shown to have a thermolabile EF-3 (Herrera et al., 1984). The gene encoding EF-3 has been isolated and shown to complement the temperature-sensitive defect in the yeast mutant with the thermolabile EF-3 (Qin et al., 1987).
In some respects EF-3 is the best characterized elongation factor. However, its mechanism of action remains unclear despite a number of studies (Skogerson and Wakatama, 1976;Skogerson and Engelhardt, 1977;Dasmahapatra and Chakraburtty, 1981;Hutchison et al., 1984;Herrera et al., 1984;Miyazaki et al., 1988;Uritani and Miyasaki, 1988a, 198813). The basic problem is that although the overall elongation reaction with both natural and synthetic mRNAs strongly depends on EF-3, it has not been possible to assign EF-3 to a specific step in the elongation process. We have undertaken to sequence the gene encoding EF-3 to determine if the sequence can illuminate the mechanism of action of EF-3. Our results indicate that the YEF-3 is an essential gene located at chromosome XII. The presence of the consensus sequences HOMOLl and RPG box as well as thymidine-rich sequences 3' of RPG box include the gene in the group of coordinate regulated genes of the translational machinery. The unique open reading frame of this gene codes for a predicted peptide of 1044 amino acids with a molecular mass of 115.86 kDa. The polypeptide seems to be a soluble protein both by amino acid composition and hydropathy profile. The amino acid sequence indicates the presence of internal repeats bearing ATP-binding sites that share significant similarity with a group of proteins that comprise a family of structurally and functionally related subunits which share a common evolutionary origin, bind ATP, and probably serve to couple ATP hydrolysis to a different biological process (Higgins et al., 1986).

EXPERIMENTAL PROCEDURES
Gene Disruption-A one-step gene disruption experiment was carried out as described bv Rothstein (1983). A 3.5-kb XbaI fraement containing almost the complete YEF-3 gene was cloned into pcC19. A 3.0-kb fragment containing the selectable marker LEU2 was then inserted into the BglII site within the YEF-3. The disrupted YEF-3 by LEUP was linearized with XbaI and used to transfer a diploid strain SSUlO leu2 (Fig. 1). leu' transformants were allowed to sporulate for tetrad analysis. Factor 3  1904  I kb   sa   PII  PI Xh/Xb  5  PII Ts  Xb  Xb  I  I  I  I  I  I  II  I  I  . . , >

Sequence of Elongation
The vertical line is the insert from pEF-3.
The hollow orrow is the transcript unit of YEF-3 (Qin et nl., 1987). The bottom line is the fragment that was used to construct Ml3 subclones for dideoxy nucleotide sequence analysis. The arrows indicate the direction and extent of nucleotide sequences obtained from subclones. The restriction enzyme sites are shown as follows: R, BglII; X, XbaI; Xh, XhoI; A, EcoRI.

Orthogonal
Field Alternation Gel Electrophoresis (OFAGE) and Southern Blot Hybridization-Sample preparations from yeast strain DC04 and OFAGE were performed as published (Schwartz and Cantor, 1984;Olsen, 1984, 1985). After staining with ethidium bromide and destaining, the gel was soaked in 0.25 N HC1 for 20 min and then transferred to Zetabind membrane with 0.4 N NaOH for 3 h. The blot was cut into three strips and hybridized with chromosome probes and the YEF-3 DNA, respectively. Cloning of Fragments-3.5-kb XbaI fragment containing the major part of the transcription unit of YEF-3 and 1.5-kb EcoRI-XbaI fragment containing the upstream sequence were prepared from plasmid pEF-3 (Qin et al., 1987;Fig. 2). The 3.5-kb fragment was cloned into M13mp18, and two clones with inserts in opposite orientation were used to derive a series of overlapping subclones. The 1.5-kb fragment was cloned into M13mp18 as well as M13mp19. A 2.5.kb XhoI-BglII fragment was also cloned into M13mp18 to be used for determining the junction sequence of the 1.5-and 3.5-kb fragments (Fig. 2) The nucleotide and the deduced amino acid sequences of the YEF-3 gene are shown. In the 5'-flanking region, box 1 is HOMOLl; box 2 is RPG box, and TATA boxes are underlined.
In the coding region, the sequence motifs of regions A and B are underlined and double underlined, respectively. In the 3'-flanking region, the polyadenylation signal sequence is double underlined.
The signal for transcriptional termination proposed by Henikoff et al. (1983) is underlined, and the transcription terminal signal proposed by Zaret and Sherman (1982) is ouerlined.
to transform a leu-diploid yeast strain SSUlO. Three stable leu+ transformants were obtained and one of them was allowed to sporulate for tetrad analysis. Fig. 3A shows 12 tetrads. Only two of the four spores of each tetrad were viable for 10 out of the 12 tetrads. All viable spores are leu-. This segregation pattern shows that the disruption of the gene encoding EF-3 is lethal in haploids.
To confirm the site of integration of the transforming DNA in the genome of SSUlO, the DNA from the parental diploid leu-, the leu+ transformants, and a pair of viable spores was prepared and digested with PuuI or EcoRI, followed by blotting and hybridization with YEF-3 and LEU2 DNA frag-ments, respectively. Since YEF-3 is a single copy gene and there are no PuuI sites within YEF-3 and LEU2, only one band can be seen in the parental diploid (SSUlO) and the pair of viable spores after hybridization with the YEF-3 probe (Fig. 3B, lanes 1, 3, and 4), while from the leu+ transformants two bands are found (Fig. 3B, lane 2), which was expected if the linear YEF-3::LEU2 fragment had been integrated into one of the two haploid genomes. In Fig. 3C, the EcoRI-digested DNAs were hybridized with a LEU2 probe. Because there is an EcoRI site within LEUS, two bands can be seen in the parental diploid and the pair of viable spores (lanes 1, 3, and 4), while four bands are seen from the leu+ transformant (lane Sequence of Elongation Factor 3 2). The results described above indicate that the lethal mutation is genetically linked to the disrupted YEF-3 gene. Therefore, YEF-3 is an essential gene.
YEF-3 Gene Is Localized in Chromosome XII-In order to determine in which chromosome the YEF-3 gene is localized, a number of OFAGE studies were carried out using the yeast strain DC04 (Carle and Olsen, 1985) at different gel concen- Isa trations followed by blotting and probing with the YEF-3 fragment and different chromosomal probes. We ruled out that the YEF-3 was in chromosome IV, VII, or XV (data not shown) and found that the band probed with YEF-3 always corresponded to the band probed with rDNA, the probe for chromosome XII. This chromosome behaves anomalously and irreproducibly under the OFAGE conditions. However, occasionally it migrates as a well defined band. One such gel is shown in Fig. 4 and demonstrates that YEF-3 is on chromosome XII.
YEF-3 Gene Has Only One Open Reading Frame-The complete nucleotide sequence of the 5034-base pair fragment of the YEF-3 gene and the deduced amino acid sequences are shown in Fig. 5. Only one open reading frame with 3132 amino acids was found with no splicing sequences in the cloned fragment, which is consistent with the results of Sl mapping previously reported (&in et al., 1987). The molecular mass of the predicted EF-3 protein is 115,860 daltons, which is somewhat less than the 125,000 Da determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Dasmahapatra and Chakraburtty, 1981). Both amino acid composition (42.3% nonpolar amino acids) (Table I) and the hydropathy profile ( Fig. 6) based on hydropathy values of Kyte and Doolittle (1982) indicate that EF-3 should be a soluble protein which is consistent with the purification procedures previously published (Skogerson and Wakatama, 1976;Dasmahapatra and Chakraburtty, 1981;Qin et al., 1987;Uritani and Miyasaki, 1988a). The amino acid composition indicates that elongation factor 3 is an acidic protein with an isoelectric point of 5.7. The codon usage in the YEF-3 coding region is presented in Table II. A codon bias of 0.87 obtained for this gene reflects the preferential codon usage of yeast genes expressed at high levels such as glucose-3-phosphate dehydrogenase and ADHl (Bennetzen and Hall, 1982  gion.s of the Gene-The initiation codon AUG is preceded by 1552 nucleotides that presents stop codons in all reading frames, indicating that this region is untranslated. Observations in Saccharomyces cerevisiae show that the genes encoding the components of translation and transcription apparatus, such as r-proteins, elongation factors, and RNA polymerase subunits, contain a number of conserved elements in their 5'-flanking regions (Teem et al., 1984;Leer et al., 1985;Nagashima et al., 1986;Mann et al., 1987). The most striking common elements are HOMOLl and RPG box occurring 200-500 nucleotides from the translation starts. They are considered as enhancer-like elements involved in coordinate regulation of transcription of these genes and recognized by a special binding protein, TUF (Huet et al., 1985;Vignais et al., 1987). Such elements can be found in the 5' region of the YEF-3 gene at positions -286 and -485 (Fig. 5). , ' ,' ,d,' egion downstream of the RPG box, which have been found in genes involved in the process of translation and correlated with highly expressed genes for glycolytic enzymes in yeast (Dobson et al., 1982), can be seen at positions -60 to -75 and -200 to -227, but they are not followed by the CAAG motif as in other yeast genes. Perhaps most striking is the presence of a dyad symmetry in this region with the sequence GAAAAAGAAA/CTTTTTTTTT at positions 5' (-248) and 3' (-213), respectively. The role if any of this dyad symmetry remains to be determined. Nevertheless, the presence of these three characteristic sequences (HOMOLl, RPG box, and T-rich blocks) in 18 of the cloned ribosomal protein genes (Leer et al., 1985;Rotenberg and Woolford, 1985;Teem et al., 1984), as well as in the elongation factor 1 LY genes TEFl and TEF2 (Huet et al., 1985), suggests that the transcription of YEF-3 gene may be coregulated with other genes coding for the components of translational apparatus in yeast. Four potential TATA boxes were found. Two of them are located around positions -124 and -165, respectively, and the other two are far from the translation initiation site (-584 and -640). According to the transcription unit map of YEF-3, the transcription initiation may start in a purine-rich site between -49 and -42 nucleotides, which resembles the transcription initiation regions of other yeast genes, uiz. the sequence PyAAPu (Burke et al., 1983). The environment of the initiation codon AUG in the YEF-3 gene with an A at position -3 is associated with the most efficient translation in eukaryotes (Kozak, 1981;Bairn and Sherman, 1988) and fits well with the consensus sequence of the initiator region 5'-A/YAA/UAAUGUCU-3' found in most of the yeast genes (Cigan and Donahue, 1987).
The coding region ends with a TAA terminator codon followed by a segment of about 350 nucleotides of the 3'untranslated region. Several sequence elements have been proposed to be involved in transcription termination and polyadenylation in yeast. A putative polyadenylation signal AATAA was found 50 nucleotides downstream of the stop  'Higgins et al., 1982. d Gros et al., 1986. e Dolittle et al., 1986. 'Walker et al., 1982. B Serrano, 1988 h Cotrelle et al., 1985. ' Laursen et al., 1981 ' Sacerdot et al., 1984. ' Powers et al., 1984 codon. Sequences similar to the consensus TAG...TATG-T...TTT proposed by Zaret and Sherman (1982) as transcriptional termination signals were found in the YEF-3 gene, as well as the sequence TTTTTATA 37 base pairs downstream of the stop codon (Fig. 5), which as stated by Henikoff et al. (1983) is for transcription termination.
Based on the transcription unit map published previously (&in et al., 1987), the termination site of transcription is located about 110 bp downstream of this consensus sequence. Interestingly, at this site we found a dyad symmetry that is exactly the same as we found in the upstream region, but in the opposite orientation (AAAAGAAAAAG/TTTTTTTTTTC) at positions +3283 and +3323. Moreover, at the end of the coding region of the gene, in sites that correspond to three clusters of lysine in the C-terminal region of the protein, other dyad symmetries were found. Because of the repetition in tandem of the AAG codons in these clusters and the presence of pyrimidine residues (T/ C) between them, hairpin loops could be formed in the mRNA molecule.
Internal Repeats in the Amino Acid Sequence Share an ATP-binding Domain, and the C-terminal of the Polypeptide Presents a Highly Charged Region Composed of Three Basic Polylysine Block-Dot matrix analysis shows the presence of internal repeats in the EF-3 predicted polypeptide (Fig. 7A). The alignment of EF-3 with itself demonstrates the existence of two large repeated sequences encompassing residues 435 to 608 and 673 to 961 (Fig. 8). When the codon preference of these two repeats was examined the same codon bias was found, suggesting that a gene duplication probably occurred during the evolution of EF-3. Each repeat consists of two regions of high similarity (Fig. 8). Regions A of both repeats exhibit 26.9% identity in amino acid sequence and regions B 33% identity.
If conservative replacements are considered, similarities of 76 and 78% are obtained. These regions present two short amino acid sequence motifs, GXXGXGKS/T and (np)4 D, which seem to occur in many different proteins that bind nucleotides (Walker et al., 1982;Brands et al., 1986), suggesting that these repeats have a nucleotide-binding domain. Table III lists some of the proteins that share these sequence motifs in regions A and/or B. The presence of these conserved sequences in such diverse examples as ATPases, bacterial permeases, kinases, GTP-binding proteins, proteins involved in DNA repair, drug resistance, Rhizobium nodulation, membrane transport, and the yeast elongation factor 3 among others suggests that this model of nucleotide binding has been very successful in the course of evolution.
The consensus sequence NKXD preceded by a group of hydrophobic amino acids, which is often found in elongation and initiation factors of protein synthesis and seems to be involved in the binding of the guanine moiety of the nucleotide (Brands        This structural feature together with the biochemical studies (Skogerson and Wakatama, 1976;Skogerson and Engelhardt, 1977;Miyazaki, 1988a, 1988b) suggest that the nucleotide domain of the repeats is an ATP-binding domain. In the archetype adenylate kinase mode of ATP binding, the lysine in region A binds to the cu-P of ATP and the aspartate at sequence B binds to the fi-and y-P of ATP through the chelated magnesium. However, other modes of ATP binding exist in nature like the ATP-binding site of phosphofructokinase, phosphoglycerate kinase, hexokinase, pyruvate kinase, and the proton ATPases with phosphorylated intermediate (Serrano, 1988). The proton ATPases that operate via a phosphoenzyme intermediate belong to the family of P-ATPases and are inhibited by vanadate (Pedersen and Carafoli, 1987;Nelson and Taiz, 1989). Miyazaki (1988a, 1988b) demonstrated that the ATPase activity of EF-3 but not the GTPase activity of EF-1 and EF-2 is inhibited by vanadate, and they suggest that like the cation-pumping P-ATPases a phosphoprotein inter-mediate might be involved. The presence of an aspartate at positions 568 and 921 preceded by hydrophobic amino acids in regions B of the two repeats is consistent with this suggestion. Yet the structure of EF-3 differs significantly from the cation pumps (Serrano, 1988;Nelson and Taiz, 1989). Region A of EF-3 is not present in the P-ATPases, and in the P-ATPases the aspartate residue is always followed by the same amino acids KTGT that are not present in EF-3 (Table III). The lack of similarity to the structure of the ATP-driven cation pumps suggests that one must seek direct evidence of EF-3 phosphorylation.
The EF-3 polypeptide also contains a consensus target sequence for potential modification by CAMP-dependent protein kinases (Cohen, 1985) at position 263-266 in the Nterminal region of the protein. Five potential sites for Nlinked glycosylation (NXS/T) were found throughout the protein.
The C-terminal region of the protein is highly hydrophilic and particularly rich in basic amino acids with three clusters of lysine residues, which may interact directly with nucleic 1910 Sequence of Elongation Factor 3 acid (Clarck and Felsenfeld, 1971). Similar domains were found in the yeast and human eukaryotic initiation factor 2 in the N-terminal of the polypeptide (Donahue et al., 1988;Pathak et al., 1988).

EF-3 Is a Member of a Family of Closely Related ATP-
binding Proteins-Searches of the NBRF-PIR data bank using the FASTA program show that although EF-3 is unique it is similar to some proteins that belong to a family of ATPbinding proteins with 14 members already identified (Higgins et al., 1988). This similarity extends over an entire internal repeat of about 200 amino acids. The proteins with the highest scores of similarity to EF-3 include Escherichia coli phosphate transport subunit, pstB (Amemura et al., 1985); E. coli ribose transport subunit, rbsA ); E. coli maltose transport subunit, malK (Gilson et al., 1982); Salmonella typhimurium polypeptide permease subunits, oppD (Higgins, et al., 1985) and oppF (Hiles et al., 1987); S. typhimurium histidine transport subunit, hisP (Higgins et al., 1982); Rhizobium leguminosarum nodulation protein, nod1 ); E. coli hemolysin secretion protein, hlyB (Femlee et al., 1985); liverwort chloroplast predicted protein mbpX (Oyama et al., 1986); E. coli chlorate-resistant protein, chlD (Johann and Hinton, 1987); and human and mouse drugresistant P-glycoproteins, mdr (Chen et al., 1986;Gros et al., 1986;Gerlach et al., 1986). pstB, oppD, oppF, malK, hisP, and rbsA are known components of bacterial active transport transmembrane complexes (Ames, 1986). The other proteins are associated with nodulation in Rhizobium, export of hemolysin in E. coli, drug resistance in E. coli and tumor cells, and one with an unknown function encoded by the mbpX gene of liverwort chloroplast. Fig. 9 shows the alignment of domain I (435 to 608 amino acid residues) and domain II (673 to 961 amino acid residues) of the predicted EF-3 polypeptide with some of the proteins with the best scores of similarity and that are involved in different cellular processes. The statistical significance of similarities was examined to 100 shuffled comparisons with the window for local shuffling set to 10 amino acids. The initial score obtained by the alignment of the EF-3 polypeptide domains with each other was 17.35 SD. above the mean, while the initial scores obtained by the alignment of domain I with pstB, mbpX, hisP, malK, nodI, mdrl, chlD, and hlyB were 21.5, 22.0, 19.2, 18.55, 17.2, 12.53, 11.27, and 7.9 S.D. above the mean, respectively. The initial scores obtained by the alignment of those proteins with domain II were 13. 52, 16.53, 10.07, 14.43, 14.47, 6.93, 5.82, and 5.93 S.D. above the mean, respectively. The similarities, considering identities and conservative replacements (BESTFIT), using the algorithm of Smith and Waterman (1981), of domain I with each one of the above proteins is 54.3,50.9,51.1,50.0,51.2,55.2,50.9,and 54.4%,respectively,and of domain II is 50.2,42.8,53.6,43.3,45.8,51.6,48.6,and 43.6%. When these proteins were analyzed for local similarity with EF-3, the highest level of similarity was observed in the regions that bear the nucleotide-binding site. Nevertheless, the similarity is not restricted to these sites. The dot plots in Fig. 7 showing the internal repeats of EF-3 (Fig. 7A) and the alignment of each one of the related proteins with the two domains of EF-3 (Fig. 7, B-F) clearly demonstrate the extensive similarity among them. Moreover, it is important to emphasize that no similarity could be detected between EF-3 and any other nucleotidebinding proteins, in spite of the preservation or the consensus sequences (Table III), indicating that additional functional similarities aside from potential nucleotide-binding properties could be detected between EF-3 and this group of proteins. Fig. 9 shows that the 174 amino acid residues of EF-3 domain I match very well with this group of proteins, being almost restricted to the regions of highest similarity, as can be seen by the consensus sequence. On the other hand, the 289 amino acid residues of the domain II bear a region of about 45 amino acid residues after the G-G-GKS/T motif with very few matches. Nevertheless, domain II extends through all the length of the alignment and shares identities with the different sequences of the group as though it was the sum of these sequences.
The proteins of this family are associated with a variety of different biological processes including membrane transport, cell division, protein export, DNA repair, and multidrug resistance. The inclusion of the eukaryotic elongation factor 3 in this family brings a new function to these ATPases and expands the discussion about the mechanisms of action of these proteins.
Binding and hydrolysis of ATP have been demonstrated until now for only one single member of this family, the UvrA protein (Seeberg and Steinum, 1982). ATP hydrolysis has not yet been demonstrated by any of the transport components, although the binding of ATP has unambiguously been shown to oppD, hisP, and malK (Higgins et al., 1985;Hobson et al., 1984). This fact has raised the question whether ATP-binding sites play a role other than coupling ATP hydrolysis to this biological process, for instance a regulatory (Chen et al., 1986;Ames et al., 1986) or a structural role (Cross and Nalin, 1982). The fact that ATP hydrolysis is clearly associated with the function of EF-3 and that EF-3 shares an extensive similarity with this family of proteins is consistent with the suggestion that ATP hydrolysis supplies the energy necessary for the specific biological processes with which the other components of this family are associated (Hiles et al., 1987;Higgins et al., 1988).
Most of the ATP-binding proteins are associated with membrane events. The members of the bacterial transport protein family interact with periplasmatic substrate-binding proteins as well as with hydrophobic integral membranes proteins, are relatively hydrophilic, and seem likely that they are peripheral membrane proteins (Ames, 1986). The other proteins HlyB, P-glycoproteins, and Nod1 are also thought to be membrane-associated. However, the UvrA protein which is a member of this family is not a membrane-associated protein since it binds to both single-and double-stranded DNA and, when complexed with UvrB and UvrC, cuts damaged DNA at two sites separated by 12 or 13 nucleotides (Sancar and Rupp, 1983). Likewise, the EF-3 protein is not a membrane protein because it purifies as a soluble protein and must interact with other elongation factors and the ribosome itself to accomplish its function.
The internal repeats in the EF-3 polypeptide expand the suggestion of Hiles et al. (1987) that the proteins of the bacterial transport family function as dimers, since the E. coli oligopeptide transport system requires two ATP-binding subunits, oppD and oppF. Moreover, the RbsA subunit of the E. coli ribose transport contains two ATP-binding domains in tandem , as well as the UvrA protein. In eukaryotes, the mammalian multidrug-resistant membrane glycoproteins (Chen et al., 1986;Gros et al., 1986;Gerlach et al., 1986) and the white protein in the fruit fly Drosophila melanogaster (Dreesen et al., 1988), members of this family of ATP-binding proteins, also provide a corresponding pair of ATP-binding domains required for the decrease of drug accumulation in multidrug-resistant cells and for the deposition of pteridine pigments in the compound eyes and other tissues, respectively. Moreover, all the prokaryote members of this family belong to a multi-component system of proteins that could be energized by ATP hydrolysis. In eukaryotes, the organization of the single protein is similar to the multicomponent system of prokaryotes (Higgins et al., 1988).
These extensive sequence similarities suggest that the basic role of EF-3 is to transduce chemical energy stored in nucleoside triphosphate into mechanical energy involved in translocating the various components of the protein synthesis system druing the elongation process. The fact that EF-3 interacts with the ribosome is demonstrated by the ribosome dependence of its ATPase activity. The C-terminal region of the protein with its polylysine blocks could interact with RNA. EF-3 is required in every cycle of chain elongation (Hutchison et al., 1984) but merely stimulates the partial reactions carried out by EF-1, EF-2, and peptidyltransferase (Skogerson and Engelhardt, 1977;Dasmahapatra and Chakraburtty, 1981;Uritani and Miyazaki, 1988a). It may be that EF-3 carries out an as yet undefined step in the elongation process, but an alternative interpretation is that it acts to speed up the whole process and removes a rate-limiting step, probably involving the dissociation of the factors or tRNA molecules from the ribosomes, not assayed by the partial reactions. Although yeast has 80 S ribosomes with a molecular mass around 4.5 x lo6 Da like other eukaryotes they achieve a rate of protein synthesis only a little slower than bacteria with their 70 S, 2.5 x lo6 Da ribosomes. E. coli with 70 S ribosomes can achieve a rate of amino acid polymerization of 17 amino acids/s/ribosome (Dennis and Bremer, 1974), three times the rate of protein synthesis in yeast which is 5.5 amino acids/s/ribosome (Waldron and Lacroute, 1975). The rate of protein synthesis of Chinese hamster ovary cell in culture is 0.60 amino acids/s/ribosome, calculated from data presented by Fischer and Moldave (1981) and Harlow and Lane (1988). Mammalian ribosomes are eight times slower than yeast in spite of the similarities of the ribosome particles in both eukaryotes.
The energy for EF-3 role in protein synthesis almost certainly comes from ATP. EF-3 affinity for ATP is two to three times its affinity for GTP (Uritani and Miyazaki, 1988a). The concentration of ATP in yeast is five times higher than the concentration of GTP (Swedes et al., 1979). This lo-15-fold preference for ATP over GTP is somewhat surprising. GTP is the favored energy source for protein synthesis in all organisms examined. If the role of EF-3 is to accelerate the elongation process on a eukaryote 80 S ribosome to approach the speed of a 70 S ribosome, the use of ATP as an energy source may allow fungi to directly regulate the rate of protein elongation in response to energy limitation. Studies on the regulation of protein synthesis in yeast during energy limitation have demonstrated that yeast inhibits both initiation and elongation in response to energy limitations. The effect on initiation is most likely due to the impact of changes in the GTP:GDP ratio on eukaryotic initiation factor 2 which is inhibited by GDP (Walton and Gill 1975a, 1975b, 1976. The effect on elongation is not explained by the known affinities of EF-1 and EF-2 for GTP or GDP and may well involve the regulation of EF-3 activity. We and others (Skogerson and Wakatama, 1976;Dasmahapatra and Chakraburtty, 1981;Herrera et al., 1984;Hutchison et al., 1984;Uritani and Miyazaki, 1988a) have been able to find elongation factor 3 in several fungal species. Bacterial, plant, and mammalian sources have been examined without success for enzymatic activity that would complement a yeast extract depleted in EF-3 and supernatant proteins that would cross-react with monoclonal and polyclonal antibodies against yeast EF-3. Uritani and Miyazaki (1988a) have noticed that polyclonal antibodies raised against yeast EF-3 cross-react with ribosomal proteins from Z'etruhymenu, brine shrimp, and rat liver. We have extended their observation to include cross-reaction with ribosomal proteins from mouse ascites cells, soybean, and corn.2 The meaning of the cross-reaction with ribosomal proteins awaits further study, but it is clear that there is no evidence for a supernatant factor such as EF-3 in mammalian or plant ribosomes. Negative results are intrinsically unsatisfying, but sufficient effort has been expended to suggest that EF-3 is characteristic of fungal protein synthesis and is not present in higher plants and animal. It may mark a major step in the evolutionary history of protein synthesis. The question whether EF-3 activity is only required for fungal ribosomes or whether it assumes a different form, presumably as a ribosomal protein(s), in mammalian and plant ribosomes awaits further study as does the broader evolutionary question of whether EF-3 appears outside the fungal kingdom.