Site-directed Mutagenesis of Yeast eEF1A

Site-directed mutants of eEF1A (formerly eEF-1α) were generated using a modification of a highly versatile yeast shuttle vector (Cavallius, J., Popkie, A. P., and Merrick, W. C. (1997) Biochim. Biophys. Acta 1350, 345–358). The nucleotide specificity sequence NKMD (residues number 153–156) was targeted for mutagenesis, and the following mutants were obtained: N153D (DKMD), N153T (TKMD), D156N (NKMN), D156W (NKMW), and the double mutant N153T,D156E (TKNE). All of the yeast strains containing the mutant eEF1As as the sole source of eEF1A were viable except for the N153D mutant. Most of the purified mutant eEF1As had specific activities in the poly(U)-directed synthesis of polyphenylalanine similar to wild type, although with a K m for GTP increased by 1–2 orders of magnitude. The mutants showed a reduced rate of GTP hydrolysis, and most displayed misincorporation rates greater than wild type. The mutant NKMW eEF1A showed unusual properties. The yeast strain was temperature sensitive for growth, although the purified protein was not. Second, this form of eEF1A was 10-fold more accurate in protein synthesis, and its rate of GTP hydrolysis was about 20% of wild type. In total, the wild-type protein contains the most optimal nucleotide specificity sequence, NKMD, and even subtle changes in this sequence have drastic consequences on eEF1A function in vitro or yeast viability.

Eukaryotic elongation factor 1A (eEF1A) 1 binds aminoacyl-tRNAs in a GTP-dependent manner and positions the bound aminoacyl-tRNA in the A site of the ribosome. After or concomitant with the proper recognition of codon and anticodon, GTP is hydrolyzed, and eEF1A⅐GDP is released from the ribosome allowing for peptide bond formation with the peptidyl-tRNA in the ribosomal P site.
Besides being involved in the synthesis of every peptide bond, eEF1A is an exceptionally abundant protein comprising 1-3% of the soluble protein in most eukaryotic cells. Elongation factor 1A amino acid sequences have been inferred from more than 100 different organisms including bacteria, archaebacteria, plants, and animals. The relatively slow rate of change of the sequence of EF1A as it evolved into eEF1A has made it an excellent sequence for determining phylogenetic trees. In large measure, it would seem that the overall tertiary structure has probably been maintained given the 33% identity and 56% similarity of Escherichia coli EF1A (formerly EF-Tu) with human eEF1A, and a discussion of the evolution of the EF1A into the eEF1A sequence has been published (1).
Beyond its role in protein synthesis, eEF1A has also been of interest as a member of the G protein family, as the only crystal structures known are for EF1A (2), EF2 (formerly EF-G) (3), Ras (4), and transducin (5). Crystals, but no structure, have been reported for archaebacterial EF1A (6). The very high homology between EF1A and eEF1A has allowed us to use the available crystal structures for EF1A to model the changes we have made in the yeast eEF1A mutants. Many attempts have been made to mutate GTP utilizing proteins to make them use XTP (7)(8)(9)(10), including the prokaryotic counterpart to eEF1A, EF1A (11,12).
To study the nucleotide specificity of eEF1A, site-directed mutagenesis has been used to alter the binding pocket for the nucleotide in domain I of eEF1A. To do this, a highly efficient chromogenic selection system on a shuttle vector was used (13). We have made five mutants of yeast eEF1A at asparagine 153 and aspartic acid 156, the conserved amino acids in the nucleotide specificity sequence, NKMD. Several yeast strains grow exclusively on the mutant form of eEF1A, and the NKMN mutant eEF1A, when purified, uses XTP just as well as the wild type uses GTP. Curiously, the yeast strain for the NKMW mutant was temperature-sensitive, whereas the protein itself was not.
Oligonucleotides-Oligonucleotides were synthesized by the Molecular Biology Core Laboratory at Case Western Reserve University on an Applied Biosystems model DNA synthesizer, deprotected, and purified by the OPC method as described by the manufacturer (Applied Biosystems). The following oligonucleotides were made for site-directed mutagenesis. The oligonucleotides N153D, N153T, D156N, D156W, and D156E (using the N153T mutant as starting sequence) were all designed to make the respective indicated amino acid changes to the nucleotide binding sequence. N153D oligonucleotide, 5ЈCTCAGGTA-GAACAGCTGTCGTTGTTAG3Ј; N153T oligonucleotide, 5ЈTAAACT-GCCTCAGGTAGAACCACTGTCGTTGTTAGTTAACAG3Ј; D156N oligonucleotide, 5ЈAAGCAGGGTAAACTGCCTCAAGTAGAACAACTGT-CGTTGTTA3Ј; D156W oligonucleotide, 5ЈTAAGCAGGGTAAACTGCC-TGGTGTAGAACAACTGTCGTTGTTA3Ј; and D156E oligonucleotide (using the N153T mutant as starting sequence), 5ЈGGGTAAACTGC-CTAAGGTAGAACCACT3Ј. All the site-directed mutation oligonucleotides were constructed so that they would introduce a change in the pattern of restriction sites and yet not introduce a rare codon as defined * This work was supported in part by National Institutes of Health Training Grant in Metabolism DK 07319 (to J. C.) and Grant GM 26796 (to W. C. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In Vitro Mutagenesis-The entire mutagenesis procedure has been published (13). In short, the yeast shuttle vector PRS314-JCTEF2 (1) was the starting material. The PRS314-JCTEF2 vectors all contain the wild-type eEF1A promoter in the wild-type settings in front of the coding region. All the cloning took place in XL1-Blue (Stratagene) cells. The single-stranded DNA is recovered from the PRS314-JCX phagemid (JCX indicates a generic mutant X; in a specific mutant, X will be substituted with a number), which can be secreted as single-stranded DNA in the presence of M13 helper phage VCSM13 (Stratagene). The Sculptor IVM from Amersham Pharmacia Biotech was used for mutagenesis.
XL1-Blue cells were eletroporated according to Bio-Rad's Gene Pulser apparatus protocol. The mutants were identified by the desired color change of the colonies, according to the usage of the second oligonucleotide in the mutagenesis reaction (13). Plasmid preparations were made using the Promega Wizard Minipreps DNA purification system. Restriction digests were made as a first confirmation of the mutants. For the N153D mutant, digestion with the endonuclease SalI lacked a band compared with the wild-type digest when the digested plasmids were run on a standard agarose gel. For the mutants N153T and all three mutants at position 156 (D156N, D156E, and D156W), the identifications were performed using the restriction enzymes HincII and Tth111 I, respectively. Final sequencing of the mutants was performed using the DNA Sequenase kit version 2.0 (United States Biochemical) with [ 35 S]dATP (Amersham). Sequencing gels were made with premixed Long Ranger (AT Biochem).
Transformation of Saccharomyces cerevisiae-LiCH 3 CO 2 -washed M213 yeast were prepared as described before (1). Aliquots of 200 l were stored at the vapor temperature of liquid nitrogen. Just before use, cells was thawed on ice and mixed with up to 5 g of DNA and 200 g of salmon sperm carrier DNA (Sigma) in a maximum volume of 20 l. The mixture was incubated for 30 min at 30°C with agitation. Then, 1.2 ml of 40% PEG-4000, 0.1 M LiCH 3 CO 2 , 10 mM Tris-HCl, pH 7.5, and 1 mM EDTA was added and incubated for 30 min at 30°C with agitation. Heat shock for 15 min was performed at 42°C in a water bath. The cells were centrifuged for 5 s and washed twice with 0.5 ml of TE, pH 7.5. Finally, the cells were resuspended in 300 l of TE, pH 7.5, and 100 l was plated on selective supplemented minimal medium (SMM) plates (the rest was saved at 4°C). Normally, it took 1 1 ⁄2 to 2 days of incubation at 30°C until transformants appeared.
In Vivo Analysis of Mutationally Altered eEF1A Genes-The strategy for in vivo analysis of altered eEF1A genes (TEF1 and TEF2) in yeast involves the use of plasmid shuffle. The strain M213 (genotype: Mata leu2-3, 112 Dtef1::LEU2 tef2-D2 lys2-20 his4 -713 met2-1 ura3-52 trp1-D1 (YCpMS29 (TEF2, URA3))) served as the recipient for transformation (13,14). This strain contains nonfunctional chromosomal copies of both TEF1 and TEF2. This genotype is normally lethal due to the absence of TEF function, but viability is maintained by the presence of a yeast centromeric plasmid carrying a wild-type TEF2 gene. This strain has a growth rate identical to yeast strains with chromosomal copies of TEF2. The plasmid carrying the TEF2 gene is never lost because it is required for viability. Due to the CEN (yeast centromere sequence) and ARS (autonomously replicating sequences) elements in the PRS314 vectors, they are mitotically stable (15). The selectable markers TRP1 and URA3 in PRS314 and YCpMS29, respectively, have the following functions. TRP1 is a tryptophan auxotrophic marker, and URA3 is a 5-fluoro-orotic acid (5-FOA) marker used to select against URA3containing plasmids in the final test of the mutagenesis products in vivo.
The plasmid shuffle was carried out as described earlier (13). In short, yeast strain M213 can be transformed with plasmid PRS314-JCTEF2 to a Trpϩ phenotype. The resulting transformant will harbor two plasmids, one (YCpMS29) carrying wild-type TEF2 and URA3 and the other, PRS314-JCX, carrying mutagenized TEF2 and TRP1. In the plasmid shuffle, colonies are first screened for the Trpϩ phenotype, indicating the presence of the plasmid carrying the mutagenized TEF2. If the TEF2 gene product were dominant lethal (negative), no Trpϩ phenotypes would be observed. This did not occur with the mutants tested. The ability of the mutagenized TEF2 gene to support viability was then tested on 5-FOA plates. In the presence of the URA3 gene product, 5-FOA turns into a potent toxin for the cell (13). The M213 strain used as host in these studies harbors the plasmid YCpMS29 carrying TEF2 and URA3. By selecting on 5-FOA medium, only colonies that have lost the YCpMS29 plasmid, and thereby lost the URA3 gene, will survive. At the same time, these colonies will only survive if they have gained a viable copy of the TEF2 gene via the transfected PRS314-JCX plasmid. This procedure allows the determination of whether the mutagenized gene can support growth of yeast in the absence of wildtype eEF1A. If no Trpϩ colonies are observed (which was only the case for the N153D mutant, JC2), this means that the mutagenized TEF2 gene cannot support growth and substitute for wild-type eEF1A.
Purification of eEF1A from Yeast-Purification of eEF1A was performed according to a scheme modified from Carvalho et al. (16). After several passages on 5-FOA plates (except for the N153D, JC2, mutant), the mutant yeast strains were grown in standard YPD media to an A 600 of 2.5, concentrated by centrifugation (15 min at 4,000 ϫ g), and stored as a cell pellet at liquid nitrogen vapor temperature until purification was started. The JC2, N153D mutant did not support growth alone, so it was passaged on SMM (17) plates and grown in SMM medium at 30°C. All operations were carried out at 4°C as described earlier (13) unless stated otherwise. In short, cell-free extracts were prepared in a glass bead blender (Biospec Products, Inc., Bartlesville, OK) with start buffer (60 mM Tris-HCl, pH 7.5, 50 mM NH 4 Cl, 5 mM MgCl 2 , 10% glycerol, 1 mM dithiothreitol (DTT), 0.1 mM EDTA, pH 8.0, 1 l/ml aprotinin, and 0.2 mM phenylmethylsulfonyl fluoride). Cell debris was removed by centrifugation at 12,000 ϫ g for 20 min. The supernatant was added to DEAE-cellulose (DE52, Whatman) pre-equilibrated with buffer 1 (20 mM Tris-HCl, pH 7.5, 25% glycerol, 1 mM DTT, 0.1 mM EDTA, pH 8.0, 1 l/ml aprotinin, and 0.2 mM phenylmethylsulfonyl fluoride) with 100 mM KCl. The unbound material and the wash were added to phosphocellulose (P-11, Whatman) pre-equilibrated with buffer 1 with 100 mM KCl. The phosphocellulose slurry was washed with buffer 1 with 100 mM KCl and then made 0.5 M KCl by adding solid KCl. The released material and wash was dialyzed overnight to yield a dialyzed solution that was 50 mM with respect to KCl. The supernatant was applied to a CM-cellulose column (CM52, carboxymethyl cellulose, Whatman) pre-equilibrated with buffer 1 with 50 mM KCl. The column was eluted with a linear salt gradient (total of 10 ϫ column volume) from 50 to 300 mM KCl in buffer 1. Purity, as checked by SDS-polyacrylamide gel electrophoresis, was found to be approximately 98%.
The DKMD mutant (JC2), purified as described above, was dialyzed overnight in buffer 1 with 100 mM KCl and 1 mM Mg(CH 3 CO 2 ) 2 . The dialyzed solution was mixed for 1 h at 4°C with GTP-agarose preequilibrated with buffer 1 with 100 mM KCl and 1 mM Mg(CH 3 CO 2 ) 2 . The slurry was packed into a 0.9 ϫ 14-cm column. The run-through was collected and reapplied to the column at a flow rate of 5 ml/h. The column was washed with buffer 1 with 100 mM KCl and 1 mM Mg(CH 3 CO 2 ) 2 . The wash-through, containing the mutant JC2, DKMD, was concentrated on a phosphocellulose column pre-equilibrated with buffer 1 with 100 mM KCl. Protein was eluted from the phosphocellulose in buffer 1 with 1 M KCl and dialyzed in buffer 1 to give a final salt concentration of 100 mM KCl. The wild-type eEF1A was eluted off the GTP-agarose column with buffer 1 with 1 M KCl and 1 mM Mg(CH 3 CO 2 ) 2 and likewise dialyzed.
Binding Assay for [ 14 C]Phe-tRNA with either GTP or GDPNP-Binding assays either with GTP or GDPNP (a nonhydrolyzable analogue of GTP) were performed under conditions similar to the poly(U) assay described above but with the following alterations. GTP or GDPNP was added at a concentration of 1 mM. Incubation was 10 min at 37°C. Quantitation of [ 14 C]Phe-tRNA bound to ribosomes was determined as retention on nitrocellulose filters (Millipore filters, type HA) (18). Use of GDPNP in this assay allowed for the determination of the number of active eEF1A molecules, as the eEF1A acts stoichiometrically, not catalytically, under these conditions. The samples were transferred to nitrocellulose and washed with wash buffer (20 mM Tris-HCl (pH 7.5), 10 mM Mg(CH 3 CO 2 ) 2 , 100 mM KCl). The amount of [ 14 C]Phe-tRNA retained on the filters was determined by liquid scintillation spectrometry.
GTPase Activity and Stimulation of GTPase Activity by Aminoacyl-tRNA and Ribosomes-GTPase activity was measured as described by Merrick (19)  After these additions, the free [␥-32 P] was extracted with 2 ml of isobutanol:benzene (1:1), and the liquids were mixed by vortexing for 30 s. After centrifugation for 3 min at 1500 rpm in a Beckman model TJ-6 centrifuge at 4°C, a 1-ml aliquot of the organic (upper) phase was removed and mixed with CytoScint from ICN. The amount of hydrolyzed [␥-32 P]GTP was determined by liquid scintillation spectrometry.

RESULTS
The target for site-directed mutagenesis was the NKMD sequence of eEF1A, known as the nucleotide specificity sequence. Fig. 1A shows how in EF1A the hydrogen bonding pattern can account for the observed specificity of EF1A for GTP (2). Given the 50% sequence identity of eEF1A to EF1A in the GTP binding domain, domain 1, we have assumed that a similar hydrogen bonding pattern exists in eEF1A. As with the binding of ATP, there are other hydrogen bonds or salt bridges with the ribose and phosphate moieties, but these are not shown as they do not contribute to nucleotide specificity. The numerous hydrogen bond possibilities likely account for the fact that the K d for GTP in G proteins (GTP-binding proteins associated with signal transduction) is often about 1 M, whereas that for ATP-utilizing proteins is in the 50 -500 M range. A proof of the hydrogen bonding pattern was tested by Hwang and Miller (11) who changed the NKXD sequence to NKXN. As illustrated in Fig. 1B, this allows for the same network of hydrogen bonds but with the nucleotide XTP, not GTP. In their studies, the EF1A NKXN mutant had the same affinity for XTP as the wild-type protein did for GTP. However, the NKXN mutant of EF1A was unable to support the growth of E. coli as the sole source of the elongation factor.
We have decided to try a similar experiment in eukaryotic cells to see if it was possible to isolate and purify mutant eEF1A proteins. As a number of investigators have been unable to express eEF1A in soluble form in E. coli, we chose to use the yeast protein and purify the expressed proteins from yeast. In addition, we have attempted to alter the nucleotide specificity of the eEF1A based upon the observation of some differences in the NKXD nucleotide specificity sequence that have been reported for a number of proteins that utilize GTP (24). These changes are indicated in Table I. The wild-type NKXD and the mutant NKXN sequences were discussed above. The NKXW sequence is found in two enzymes, phosphoenolpyruvate carboxykinase and GTP:AMP phosphotransferase. Both of these proteins will use ITP as well as GTP. The TKXE sequence is found in the enzyme guanyltransferase. Here, one could imagine that because the glutamic acid residue is one methylene group longer than aspartic acid, a compensatory shortening would have to occur at the position normally occupied by asparagine. In this instance, the threonine residue provides the hydroxyl group, and this group is one methylene group closer to the main chain. Thus, one would expect to maintain the GTP specificity, although the possible loss of one or more of the hydrogen bonds labeled G, F, or H in Fig. 1A might result in a slight increase in the K d or K m for GTP. Finally, it was thought that the substitution of an aspartic acid in place of the asparagine would weaken the binding for GTP and perhaps even allow a hydrogen bond between the aspartic acid and the 6amino group of ATP. Thus, these substitutions were made in yeast eEF1A, and the yeast strains were analyzed for growth on nonselective media, where both the wild-type and mutant eEF1A would be expressed, and on selective media (5-FOA), where the mutant eEF1A would be the sole source of the eEF1A. Table II presents the characteristics of the yeast strains with mutant eEF1A species. All of the yeast strains that contained both host and mutant eEF1A were viable, indicating that none Note that for this mutant eEF1A, the binding of GTP would occur with 1 less hydrogen bond and possibly some steric hindrance between the amide group of asparagine and the amino group of GTP. Further, note that many additional hydrogen bonds to the ribose or phosphates would be expected to be the same for the mutants and wild-type eEF1A. a This is the wild-type eEF1A sequence (25). b By analogy to the crystal structure and experiments involving EF1A (11,12).
of the mutant eEF1As were behaving as dominant negative mutants. When grown on selective media, only the DKMD mutant, JC2, failed to grow, although the NKMW mutant, JC33, did display a temperature-sensitive phenotype with wild-type growth rates at or below 18°C and no growth at elevated temperatures. Within experimental error, the yeast strains containing just the mutant form of eEF1A grew in liquid media at wild-type growth rates. Similarly, growth on alternate carbon sources (glycerol, lactose, or sucrose) did not show any difference between yeast with the wild-type or mutant eEF1As (data not shown).
To examine the nucleotide specificity of the eEF1A species, the proteins were purified as described previously except for the DKMD mutant eEF1A. As yeast containing only this form of eEF1A were not viable, the yeast containing both the mutant and wild-type eEF1A were grown. The eEF1A was purified, and the wild-type protein was separated from the mutant by GTP-agarose chromatography. The first test of the mutant eEF1As was their ability to direct the synthesis of polyphenylalanine on poly(U)-programmed ribosomes. An Eadie-Hofstee plot for the wild-type eEF1A is shown in Fig. 2A. The apparent scatter in the data likely reflects that for polyphenylalanine synthesis, there are two GTP-requiring enzymes, eEF1A and eEF2. eEF2 promotes the translocation of the mRNA and the anticodon ends of the tRNAs during polypeptide synthesis (for an overview, see Ref. 20). In contrast, plots of the enzymatic activity of the mutant eEF1As yielded linear plots with little error (Fig. 2B and Table III). In large measure, this reflects the fact that the K m values for the mutant eEF1As were well above the wild-type K m (estimated to be about 0.14 M) 2 ranging from 2.7 to 13.1 m.
Of the mutant eEF1As examined, most of the mutant eEF1As had a specific activity equivalent to wild type (i.e., as seen as V max ). The TKMD mutant eEF1A was about 50% as active as wild type. However, if additional protein was added, it was possible to achieve the same high rate of protein synthesis possible with wild-type protein. In contrast, the NKMW mutant eEF1A was quite different. With the highest concentration of the NKMW mutant eEF1A tested, only about 35% of the maximal rate of polyphenylalanine synthesis was achieved. Possible causes for the reduced rates of protein synthesis are examined below.
An examination of the ability of the various mutant forms of eEF1A to use alternate nucleotides indicated that the NKMN mutant utilized XTP with the same affinity as the wild-type protein had for GTP (Table IV), as was also observed with the equivalent substitution into EF1A (19). However, the NKMN protein was also capable of utilizing GTP, but with a much higher K m value. This ability to use GTP explains why this mutant could serve as the sole source of eEF1A in the absence of an XTP pool in yeast. With the normal intracellular concentration of GTP in cells in the hundreds of micromolar range, the protein would be well saturated with GTP. At the same time, 2 Under the conditions of the assay, the concentration of eEF1A is approximately 0.3 M. As most preparations are 20 -50% active, as measured by the GDPNP-dependent binding of Phe-tRNA to poly(U)programmed ribosomes, the concentration of active eEF1A is in the range of 0.06 to 0.15 M. As this concentration is close to the estimated K m value, our determination of the K m for the wild-type eEF1A must be considered an apparent K m . The true K m is likely to be similar to or smaller than the apparent K m .  the ability of the NKMN mutant to use XTP allowed for the determination of the K m for GTP for the second enzyme required for polyphenylalanine synthesis, eEF2. In the presence of 2 M XTP, the K m for eEF2 for GTP was determined to be 1.1 M (Fig. 3).
Besides the NKMN mutant eEF1A, all of the other forms of eEF1A were also tested for their ability to use XTP, ITP, and ATP. The results are presented in Table V. Only the TKME mutant was capable of utilizing a nucleotide other than GTP where both ITP and XTP were used about one-fifth as well as GTP. The difference in all of the alternate nucleotides used (for both the NKMN and the TKME mutant eEF1As) was only in K m ; a similar maximal velocity was achieved. It should be noted that some activity was observed with ATP for all of the eEF1A proteins. However, as the K m for ATP was several hundred micromolar and the commercial ATP was only 95% pure, it is likely that the observed activity reflects GTP contamination in the ATP.
As indicated above, most of the mutant forms of eEF1A were capable of sustaining wild-type growth rates and could achieve the same maximal velocity in polyphenylalanine synthesis as the wild-type protein. The single exception is the NKMW mutant of eEF1A, which yielded a yeast strain that was temperature sensitive and a purified protein that was able to achieve only about 13% of the wild-type rate of polyphenylalanine synthesis (Table III). As this form of eEF1A came from a temperature-sensitive yeast strain, it was possible that this form of eEF1A was less stable to purification; however, efforts to improve the recovery of activity by increasing the glycerol concentration during purification were unsuccessful. Therefore, a direct test of the thermotolerance of the protein was made. As is evident from Fig. 4, the NKMW mutant eEF1A was actually more thermostable than the wild-type protein.
As an independent check of the activity of the various forms of eEF1A, a nucleotide-dependent binding of Phe-tRNA to ribosomes assay was performed with GTP and with GDPNP. The assay with GDPNP should determine the number of active molecules, as there is a one-to-one stoichiometry when GTP hydrolysis is blocked. In this assay, the eEF1A proteins were routinely 20 -50% active, and there seemed to be only slight differences between the wild-type and mutant proteins except for the NKMW mutant, which displayed only about 25% of the wild-type activity (data not shown). When GTP was used as the nucleotide substrate, the binding increased by a factor of two or three, indicating that the eEF1A species were catalyzing more than a single binding event. However, the NKMW mutant of eEF1A showed no stimulation of binding under these conditions (data not shown). To more directly address the deficiency in the NKMW mutant eEF1A, all of the eEF1A species were assayed for their ability to catalyze GTP hydrolysis in the presence of activators (aminoacyl-tRNA, ribosomes, and poly(U)). The results in Table VI indicate that all of the mutant forms of eEF1A show a reduced rate of GTP hydrolysis. Although most of the mutants seemed to be about half as active, the NKMW mutant eEF1A was only about 20% as active. This very low level of GTP hydrolysis is consistent with the observation that the NKMW mutant eEF1A showed no stimulation in the Phe-tRNA binding assay when GTP was substituted for GDPNP. and V max for wild-type and five mutant eEF1As K m and V max values were derived from poly(U) assays using 10 pmol of eEF1A (except for mutants NKMN and TKME, where 15 pmol were used) and calculated in an Eadie-Hofstee plot (see Fig. 1 4. Thermotolerance of wild-type and NKMW mutant eEF1A in the polyphenylalanine synthesis assay. The poly(U)directed polyphenylalanine synthesis assay was performed using eEF1As that have been preincubated at the given temperatures for 10 min prior to assay (10 pmol of eEF1A per assay). The activities are shown as percent of the activity observed after preincubation at 4°C. The wild type was shown as filled circles and NKMW, eEF1A activity as open triangles. 100% activity was 0.63 and 0.04 mol of Phe/mol of eEF1A for wild-type and NKMW eEF1A, respectively.
A final test of the function of the eEF1A proteins was that of misincorporation. In this assay, equal molar amounts of [ 3 H]Leu-tRNA and [ 14 C]Phe-tRNA are added, and the ratio of [ 3 H]Leu/[ 14 C]Phe-labeled protein is determined. The wild-type protein had a misincorporation rate of about 3%, a value similar to that reported previously in this assay using bacterial components (21). Most of the mutant proteins displayed misincorporation rates greater than wild type, although the maximal difference was less than 2.5-fold (Table VII). Surprisingly, the relatively inactive NKMW mutant eEF1A showed a misincorporation rate that was roughly 10% that of wild type, indicating that this form of eEF1A was about 10-fold more accurate than the wild-type protein.
In an attempt to correlate nucleotide affinity with error rate, the two eEF1A mutants that would use more than a single nucleotide were tested for their error rates. The idea here was that each of the eEF1A proteins might have a slightly different conformation due to the single or double amino acid substitutions, and thus differences in misincorporation could reflect differences in nucleotide affinity or conformation. By testing the same protein with different nucleotides, it should be possible to minimize conformation differences. When this was done (see Table VIII), it was observed that the nucleotide with the lowest K m value led to the least misincorporation. However, given the almost 2 order of magnitude difference in K m values of the NKMN mutant eEF1A and only a 30% increase in misincorporation, it is clear that a direct linear correlation of nucleotide affinity and misincorporation rate does not exist. DISCUSSION Although the attempts at site-directed change in nucleotide specificity could have failed, as there is no endogenous pool of either ITP or XTP in yeast, the in vitro analysis of the purified proteins provided an answer as to why these efforts were successful. All of the mutant proteins were able to use GTP as the nucleotide triphosphate for the synthesis of polyphenylalanine. As expected from the studies with EF1A, the NKMN mutant eEF1A was able to use XTP as the nucleotide triphosphate and did so with the same apparent K m for XTP as the wild-type protein exhibited for GTP. Additionally, this same protein was able to use GTP, although with reduced affinity. However, as the cellular concentration of GTP is in the hundreds of micromolar range, this mutant would have been fully saturated with GTP, as its K m for GTP was 13 M, about 2 orders of magnitude greater than the K m for XTP and 1 order of magnitude less than the physiologic concentration of GTP. All of the mutants preferred GTP over other nucleotide triphosphates except for the NKMN mutant, which preferred XTP. Based upon studies where overexpression of eEF1A obviated the need for the recycling protein eEF1B (22), it is anticipated that all of the mutants will allow for growth in the absence of the recycling protein (this assumes that the K d for GTP and GDP is the same as is observed with the wild-type protein). This hypothesis will be tested in the near future.
Another in vitro characteristic of the mutant eEF1A proteins is that they all seemed to have similar specific activities (as judged in the poly(U) assay or the nitrocellulose filter assay) except for the NKMW mutant. As this eEF1A was derived from a temperature-sensitive strain of yeast, it was anticipated that the lower specific activity reflected a thermal denaturation of the protein during the purification process. However, a direct test of the thermolability of this eEF1A indicated that, if anything, this mutant eEF1A was somewhat more temperature stable than wild type (Fig. 4). Thus, at the present time we have no simple explanation for the temperature sensitivity of this yeast strain (JC33, D156W) nor the low specific activity of the NKMW mutant protein, although these effects may possibly be explained by the relatively low level of GTPase activity observed with this eEF1A (see Table VI).
As might be expected, it would seem that nature has selected the ideal eEF1A nucleotide specificity sequence. As this sequence is anticipated to be in a loop (based upon analogy with EF1A), it was thought that the site-directed mutants would not alter the tertiary structure of eEF1A to any large degree. Despite the anticipated subtle changes, all of the mutant eEF1As exhibited a considerable change in K m for GTP. This change was partially anticipated based upon the disruption of the hydrogen bonding network that has been proposed for EF1A based on the crystal structure for either EF1A⅐GDP or EF1A⅐GDPNP (see Fig. 1). Further examination revealed that all of the mutant proteins also showed a reduced rate of GTP hydrolysis. The fact that only the NKMW mutant eEF1A displayed a reduced V max in polyphenylalanine synthesis (i.e., the synthesis rate could not be increased with increased protein Values are shown as mol of free phosphate released from GTP (background corrected) per mol of eEF1A in the assay (10 pmol of eEF1A for all except from mutants JC6 and JC32, where 15 pmol were used). The stimulation of the GTPase activity by the addition of aa-tRNA, ribosomes, and poly(U) is also shown. The free phosphate generation in the assay mixes without eEF1A were 2, 2, 2, and 31 pmol, with no additions, with aa-tRNA, with both aa-tRNA and ribosomes, and with aa-tRNA, ribosomes and poly(U), respectively.  NKMN and TKME mutant eEF1As The misincorporation rates were derived from poly(U) assays (using 15 pmol of eEF1A) using equal amounts of [ 3 H]Leu-tRNA and [ 14 C]Phe-tRNA. Asterisk indicates that the observed value was not reliable as several hundred micromolar of ITP was needed to drive the reaction, and the commercial ITP was only 95% pure (possible GTP contamination, see "Results"). 100% corresponds to 3.0 leucines incorporated into a polypeptide chain for every 100 phenylalanines (see the legend to added) suggests that in most instances the rate-limiting step in the elongation cycle is likely to be the GTP-dependent translocation step catalyzed by eEF2. In this same vein, the wild-type eEF1A seemed to be the most accurate form of eEF1A except for the NKMW mutant eEF1A, which showed a 10-fold greater accuracy than the wild-type protein. The fact that the yeast strain containing this accuracy mutant displayed a temperature-sensitive phenotype is reminiscent of E. coli mutants that "require" the antibiotic streptomycin for optimal growth (23). These bacteria grow slowly normally, but their rate of growth is increased in the presence of streptomycin, as is the translational error rate in these bacteria. The conclusion reached in the bacterial studies was that nature had selected for a certain degree of error and rapidity in translation. To decrease the error rate would decrease the translation rate and visa versa. In our studies, the penalty for increased accuracy with the NKMW mutant eEF1A is a dramatically reduced translation rate in vitro and no growth at temperatures above 18°C. Finally, it is noted that the designed, site-directed mutants failed to provide in vitro the nucleotide specificity anticipated except for the NKMN mutant eEF1A, which preferred XTP over GTP by about 70-fold (Table IV). This failure most likely reflects two things. First, there is no crystal structure for eEF1A, and it is possible that there are slightly different constraints on the binding of GTP relative to EF1A. These subtle differences might be anticipated as EF1A has a 100-fold preference for GDP over GTP, whereas eEF1A binds both nucleotides with about the same affinity. Second, the nucleotide specificity sequences chosen as alternates were based upon enzymes that use GTP. As these enzymes nominally have K m values for GTP in the 50 -200 M range, it is possible that the specificity for GTP does not involve the extensive hydrogen bonding network evidenced in EF1A. Another view would be that in the attempt to identify the nucleotide specificity sequence based upon consensus elements in the linear amino acid sequence (24), the incorrect elements had been chosen. Future studies that provide the crystal structure for each of the model enzymes and the definition of the elements for GTP specificity will provide the answer to this dilemma.