INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

A special methionine tRNA is used for the initiation of protein synthesis in all organisms that have been studied. Of the two classes of methionine tRNAs found universally, the initiator is used exclusively for initiation and the elongator is used for insertion of methionine into internal peptidic linkages (33, 48). Eubacteria, mitochondria, and chloroplasts use formylmethionine-tRNA (41) for initiation (10, 33), whereas the cytoplasmic protein synthesis system of eukaryotes uses methionyl-tRNA (Met-tRNA) without formylation (25, 59).

Because of their unique function, initiator tRNAs from eubacteria and eukaryotes possess a number of special properties distinct from those of elongator tRNAs. For eukaryotic cytoplasmic initiator tRNAs, these properties are (i) the formation of a specific complex among the initiator Met-tRNA, eukaryotic initiation factor 2 (eIF2), and GTP (42); (ii) the binding of the initiator Met-tRNA to the ribosomal P site; and (iii) the exclusion of the initiator Met-tRNA from the ribosomal A site. In contrast, elongator aminoacyl-tRNAs form a ternary complex with the eukaryotic elongation factor 1 (eEF1) and GTP and bind to the ribosomal A site.

Along with their special properties, eukaryotic initiator tRNAs also possess a number of unique sequence and structural features not found in elongator tRNAs (49, 57). These include (i) an A1:U72 base pair at the end of the acceptor stem, (ii) a sequence of three consecutive G:C base pairs (G29G30G31:C39C40C41) in the anticodon stem, and (iii) A54 and A60 in the TC loop instead of T54 and pyrimidine 60 found in virtually all elongator tRNAs. Interestingly, the first two of these unique features are also found in archaebacterial initiator tRNAs (60). In previous work, we have used in vitro functional studies on mutant human initiator tRNAs expressed in mammalian CV-1 cells or in yeasts to study the role of the above features in the activities of the mutant tRNAs in initiation of protein synthesis and in binding to eIF2 (13, 17). Here, we describe functional studies, in vivo and in vitro, to identify the determinants on the human initiator tRNA which prevent it from acting as an elongator in mammalian cells.

Functional studies in vitro have shown that the primary determinant which blocks the participation of yeast and wheat germ initiator tRNAs in the elongation step of protein synthesis is a bulky 2'-O-phosphoribosyl modification of ribose at position 64 (11, 19, 20, 57). This is most likely due to protrusion of the bulky group into the minor groove of the TC stem helix, leading to a steric block in binding of these tRNAs to eEF1 (4, 19). Further support comes from isolation of mutants of the yeast Saccharomyces cerevisiae that are defective in modification of the ribose 64 in the initiator tRNA and from the demonstration that in this mutant strain, the initiator methionine tRNA can act as an elongator in vivo (1).

In contrast to fungi and plants (20, 48, 57), vertebrate initiator tRNAs do not have a 2'-O-phosphoribosyl modification of the ribose at position 64 (21, 22, 46, 58, 60). This raises the question of how the vertebrate initiator tRNAs are prevented from acting as elongators in the corresponding protein synthesis systems. In this paper, using functional studies of mutant tRNAs in vivo in mammalian COS1 cells and in vitro in rabbit reticulocyte cell extracts, we show that the primary determinant preventing the human initiator tRNA from acting as an elongator is the nature of the base pairs 50:64 and 51:63 in the TC stem. A secondary determinant is the A1:U72 base pair at the end of the acceptor stem. Mutation of A1:U72 alone to G1:C72 endows the mutant tRNA with partial activity in elongation, but less so than mutations in the TC stem. A mutant human initiator tRNA carrying both the G1:C72 mutation and mutations in base pairs 50:64 and 51:63 in the TC stem is almost as active in elongation in vitro as the elongator methionine tRNA.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Plasmids, tRNA, and reporter genes. The wild-type human initiator methionine tRNA gene carried on a 140-bp BamHI fragment and the G1:C72, A29U31:U39U41, and U54U60 mutants have been described previously (13, 16). The additional mutant tRNAs described here were made on M13 viral DNA templates by oligonucleotide-directed mutagenesis with either the Amersham Sculptor Mutagenesis kit or by the method described by Kunkel (34). The mutant tRNA genes were then cloned into pSVBpUC (14) for expression in mammalian cells. Some mutant tRNA genes were made directly in pSVBpUC by the Stratagen Quickchange method.

Transfections. COS1 cells in 3.5-cm dishes were transfected essentially as described by Kluxen and Lubbert (32). Cells at approximately 80% confluence were rinsed twice with Dulbecco's modified Eagle's medium (DMEM). DNA was then added to the culture in 1 ml of 400 µg of DEAE dextran per ml-100 µM chloroquine-10% Nuserum (Collaborative)-25 mM Tris-HCl (pH 7.4). A 2.5-µg amount of pRSVCAT reporter and 1 µg each of pSVBpUC carrying the tRNA genes and pCDNA1 carrying the E. coli metRS or glnRS gene were used per dish. The cells were incubated for 5 h at 37°C, rinsed twice with DMEM, and treated for 2 min with 10% dimethyl sulfoxide in calcium- and magnesium-free phosphate-buffered saline. The cells were rinsed again with DMEM and incubated for 60 h with DMEM supplemented with 10% calf serum. Preparation of cell extracts from transfected cells and assays for CAT activity were as described previously (6, 14).

Preparation of RNA and Northern analysis. Total RNA was prepared from 3.5-cm dishes of COS1 cells transfected as described above by the guanidine thiocyanate-phenol-chloroform method (8) (Tri-Reagent; Molecular Research Center, Inc.). The final RNA pellet was suspended in 50 µl of 5 mM sodium acetate (pH 5.0). Northern analysis on denaturing 7 M urea-polyacrylamide gels and acid urea-polyacrylamide gels (24, 61) was performed as described previously (14) with 5'-³²P-labeled oligonucleotides complementary to the anticodon stem and loop of the U35A36 or C35 mutant initiator tRNAs as probes.

Generation of SV40 virus stocks. pSVBpUC DNA carrying the initiator tRNA genes were digested with BamHI and religated to remove pUC sequences. The recombinant simian virus 40 (SV40) DNA carrying the tRNA genes was used to transfect CV-1 cells with either DEAE dextran (14) or Superfect (Qiagen). The cells underwent complete lysis usually in about 10 to 12 days. Secondary stocks were made by infecting a 15-cm dish of CV-1 cells with 2 ml of the primary virus stock and allowing the cells to undergo complete lysis (for about 5 to 6 days). This yielded about 30 ml of secondary virus stock.

Large-scale isolation of tRNA for use in in vitro protein synthesis. A total of 5 to 10 dishes (15 cm in diameter) of CV-1 cells were infected with 2 ml each of the secondary virus stocks, and the dishes were incubated for 60 h. Isolation of tRNAs from large-scale infections of CV-1 cells with recombinant SV40 and gel purification of mutant tRNAs on 15% native or semidenaturing polyacrylamide (acrylamide to bisacrylamide, 19:1) gels were done as described before (13). Typically, the tRNA expressed from the tRNA gene carried on the SV40 vector represented 80 to 90% of the total methionine-accepting tRNA.

Aminoacylation of tRNA. Aminoacylation of the initiator tRNA with [³⁵S]methionine with purified E. coli MetRS was essentially as described before (15). The elongator methionine tRNA was aminoacylated with [³⁵S]methionine with a crude rabbit liver extract also as described before.

In vitro protein synthesis assays. Reactions were performed in a 25-µl mixture containing 12.5 µl of untreated reticulocyte lysate (Promega), 20 mM hemin, 75 mM potassium acetate, 0.5 mM magnesium acetate, 40 mM HEPES-KOH (pH 7.4), 8 mM creatine phosphate, 0.2-mg/ml creatine kinase, and 0.04 mM amino acids (except methionine, which was at a concentration of 3 mM). [³⁵S]Met-tRNAs were added (approximately 20,000 to 60,000 cpm), and the reaction mixtures were incubated at 30°C. Aliquots were removed at various times and processed essentially as described previously (13) to measure the percentage of [³⁵S]methionine transferred from the tRNA to proteins.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

In vitro studies with rabbit reticulocyte cell extracts. (i) Effect of mutations in sequences conserved in eukaryotic initiator tRNAs on activity of mutant human initiator tRNAs in elongation. Figure 1 indicates the conserved sequences in the three different regions, the acceptor stem, the anticodon stem, and the TC loop, which were mutated to produce the corresponding mutant tRNAs. The mutant tRNA genes were expressed in CV-1 cells by using SV40 vectors, and the mutant tRNAs, which were overproduced, were purified by polyacrylamide gel electrophoresis (13). The mutant tRNAs were then aminoacylated with [³⁵S]methionine by using purified E. coli MetRS. Under the conditions used for aminoacylation, this enzyme aminoacylates only the initiator species of human methionine tRNA and not the elongator species (15). The [³⁵S]Met-tRNAs thus obtained were added to a rabbit reticulocyte protein synthesizing system programmed with endogenous globin mRNA, and transfer of [³⁵S]methionine to protein was monitored. In this assay system, the [³⁵S]methionine used to initiate - and -globin chains is rapidly removed by methionine aminopeptidase and is unstable, whereas [³⁵S]methionine incorporated internally by a tRNA with elongator activity is stable (13, 25). Therefore, the stable transfer of [³⁵S]methionine to - and -globins is a measure of the elongator activity of the mutant tRNAs. The results are shown on Fig. 2. The wild type, anticodon stem, and TC loop mutants are all inactive in elongation. However, the G1:C72 mutant is clearly active, although its activity is less than that of the elongator methionine tRNA.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Cloverleaf structure of the vertebrate initiator tRNA. The unique features of eukaryotic initiator tRNAs are boxed. Arrows indicate the mutations introduced. The posttranscriptional base modification at the sites of mutations are not indicated in this figure. U54 is modified to T, and U55 is modified to . In addition, U31:U39 is modified to 31:39.

View larger version (20K): [in this window] [in a new window]   FIG. 2.   Activities of mutant [35S]Met-tRNAs in elongation in a reticulocyte cell-free system programmed with globin mRNAs. The elongation activities of the wild-type (Wt) initiator tRNA and elongator methionine are also shown. The percentage of transfer of [35S]methionine to proteins is a measure of elongation activity (see text).

(ii) Effect of mutations in base pairs 50:64 and 51:63 in the TC stem on the activity of the mutant tRNAs in elongation. The partial activity of the G1:C72 mutant initiator tRNA in elongation suggested that the conserved A1:U72 base pair in human initiator tRNA is most likely not the primary determinant blocking its activity in elongation. Therefore, we considered the possibility that the sequence of the TC stem of the human initiator tRNA itself causes a perturbation of the RNA helix in the TC stem and thereby prevents the tRNA from binding to eEF1. According to this hypothesis, the mechanism used to block the activity of human initiator tRNA in elongation is the same as that in the yeast and wheat germ initiator tRNAs, except for a sequence-dependent perturbation of the RNA helix in the TC stem instead of a bulky modification of the ribose at position 64 (1, 19, 30). Consequently, we mutated base pairs 50:64 and 51:63 from A:U to U:A and U:A to G:C, respectively. Both base pairs 50:64 and 51:63 were mutated, since we found (data not shown) that mutation of just the 50:64 base pair resulted in no accumulation of the mutant tRNA in COS1 cells presumably because of alternate possible structures or because the tRNA was unstable. In view of the partial activity of the G1:C72 mutant in elongation, the TC stem mutations were also combined with the G1:C72 mutation.

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Activities of [35S]Met-tRNAs carrying G1:C72 and/or the U50C54:G63A64 mutation in the tRNA. The elongation activities of the wild-type (Wt) initiator tRNA and elongator methionine are also shown. The percentage of transfer of [35S]methionine to proteins is a measure of elongation activity (see text).

In vivo studies with COS1 cells. (i) Mutant initiator tRNAs and design of reporter systems. Mutant initiator tRNAs carrying changes in the anticodon sequence from CAU to CCU (the C35 mutant) or to CUA (the U35A36 mutant) were used for functional studies of the mutant tRNAs in elongation in COS1 cells (Fig. 4, top). A mutant CAT gene, CATM173R, was used to monitor the activity of the mutant initiator tRNAs carrying the C35 anticodon sequence change. The rationale for the design of CATM173R (Fig. 4, bottom right), which has a methionine-to-arginine change at position 173 of the CAT protein, is as follows. The C35 mutant of human initiator tRNA is most likely aminoacylated with methionine, similar to the corresponding mutants of E. coli and yeast initiator tRNAs (12, 53). This mutant has the anticodon CCU, which is complementary to the arginine codon AGG. Therefore, if mutants derived from this tRNA are active in elongation, they will read the arginine codon AGG but insert methionine. In other words, they would act as methionine-inserting missense suppressors (5, 7, 43) and correct the functional defect of a methionine-to-AGG arginine codon mutation in a reporter gene. We searched for reporter genes which contain a highly conserved methionine residue, which is critical for function, that could be mutated to arginine to produce a nonfunctional protein. The commonly used reporter gene CAT was found to have such a potentially critical methionine residue at position 173. In the crystal structure of the protein, methionine 173 lies between amino acid residues in the chloramphenicol binding pocket and the acetyl coenzyme A-binding pocket (38). The AUG codon for methionine 173 of the CAT gene was, therefore, mutated to AGG to generate the CATM173R mutant. This is the only AGG codon in the CATM173R gene. The mutant protein was found to be essentially inactive in the acetylation of chloramphenicol, since E. coli cells carrying the CATM173R gene produced CAT protein (as detected by immunoblot analysis) but were chloramphenicol sensitive (data not shown). To ensure that the C35 mutant initiator tRNAs were maximally aminoacylated with methionine in vivo, COS1 cells cotransfected with plasmids carrying the mutant tRNA and the CAT M173R reporter gene were also cotransfected with a plasmid carrying the E. coli metRS gene.

View larger version (12K): [in this window] [in a new window]   FIG. 4.   Cloverleaf structure of vertebrate initiator tRNA. The mutant tRNA anticodon sequences and the reporter genes designed to measure activity of these tRNAs in elongation in vivo are shown. Mutations in the tRNA acceptor and T stems are also indicated.

(ii) Effect of mutations in the TC stem and in the acceptor stem on the activity of the mutant tRNAs in elongation in CV-1 cells. To study the effect of mutations in the TC stem and in the acceptor stem, mutations at these sites were coupled to either of the mutations in the anticodon sequence (37, 62). The mutant initiator tRNA genes were cloned into an expression vector, and COS1 cells were cotransfected in duplicate with the three plasmids shown in Fig. 5. pSVBpUC contains the mutant initiator tRNA gene to be tested for elongation activity, pRSV CAT contains either the CATM173R or the CATam27 gene (see above), and pCDNA1 contains either the E. coli metRS or the E. coli glnRS gene. Extracts were made from cells approximately 65 h after transfection and were assayed for CAT activity. Figure 6 shows the results of a representative thin-layer chromatographic assay for CAT activity in extracts from cells transfected in duplicate with various mutant initiator tRNA genes carrying the C35 anticodon sequence mutation. The C35 mutant initiator tRNA is essentially inactive as an elongator (Fig. 6, lanes 1 and 2). The G1:C72/C35 mutant has some activity as an elongator (lanes 3 and 4) and the C35/U50G51:C63A64 mutant has substantial activity as an elongator (lanes 5 and 6), whereas the G1:C72/C35/U50G51:C63A64 mutant has the highest activity as an elongator (lanes 7 and 8). Table 1 provides a quantitative measure of CAT activities in the various extracts and confirms (i) that initiator tRNA carrying the TC stem mutation is more active than the one carrying the G1:C72 mutation in the acceptor stem and (ii) that the initiator tRNA carrying both the TC stem and the G1:C72 mutations is more active in elongation that the one carrying the TC stem mutation alone. These results obtained in vivo are in complete agreement with the results for the corresponding mutants (without the anticodon sequence change) obtained in vitro (Fig. 3).

View larger version (14K): [in this window] [in a new window]   FIG. 5.   Plasmids used for cotransfection of COS1 cells. pSVBpUC, pRSVCAT, and pCDNA1 were used for the expression of the tRNA gene, the CAT reporter gene, and the E. coli glnRS and metRS genes, respectively. The double arrows indicate the sites of insertion of the tRNA genes into the BglII site in pSVBpUC. RSV, Rous sarcoma virus.

View larger version (62K): [in this window] [in a new window]   FIG. 6.   Thin-layer chromatographic analysis of CAT activities in extracts (30 µg) of COS1 cells cotransfected with SV40-based recombinant plasmids carrying the indicated initiator tRNA genes, pRSVCATM173R, and pCDMetRS. Assays from duplicate transfections are shown. The positions of acetyl-chloramphenicol (Ac-CAM) and unreacted chloramphenicol (CAM) are indicated.

Aminoacylation levels of mutant tRNAs in COS1 cells. Figure 7 shows a Northern blot analysis of tRNAs isolated from cells transfected with the U35A36, U35A36/U50G51:C63A64, and G1:C72/U35A36/U50G51:C63A64 mutant initiator tRNA genes. The COS1 cells were also cotransfected with either pCDNA1 or pCDNA1 carrying the E. coli glnRS gene. tRNAs were isolated under acidic conditions, and tRNA and aminoacyl-tRNA species were separated by acid urea-polyacrylamide gel electrophoresis (24, 61). The mutant human initiator tRNAs were detected with a probe complementary to the anticodon stem and loop sequence of the U35A36 mutant tRNA. The results show (i) that aminoacylation of the mutant initiator tRNAs in vivo is essentially totally dependent on coexpression of E. coli GlnRS (in Fig. 7, compare lanes 1 to 4 to lanes 5 to 8, respectively) and (ii) that introduction of the G1:C72 mutation leads to a reduction in steady-state levels of aminoacyl-tRNA (compare lanes 7 and 8 to lanes 5 and 6). PhosphorImager analysis of the blot showed that while 60% of the tRNA in the case of U35A36 and the U35A36/U50G51:C63A64 mutants was aminoacylated, only 40% of the tRNA was aminoacylated in the case of the G1:C72/U35A36/U50G51:C63A64 mutant tRNA.

View larger version (49K): [in this window] [in a new window]   FIG. 7.   RNA blot hybridization of tRNAs isolated from COS1 cells cotransfected with SV40-based plasmids carrying the indicated initiator tRNA genes and either pCDNA1 (lanes 1 to 4) or pCDGlnRS (lanes 5 to 8). The tRNA and aminoacyl-tRNA species were separated on an acid urea-polyacrylamide gel.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

A clear result of this work is that the primary determinant blocking the activity of the human initiator tRNA in elongation is the nature of the base pairs 50:64 and 51:63 in the TC stem. Mutation of base pair A50:U64 to U50:A64 and base pair U51:A63 to G51:C63 allows the mutant initiator tRNA to act as an elongator in vitro and in vivo. This is unlikely to be due to the fact that base pairs U50:A64 and G51:C63 are directly recognized by eEF1, since different elongator tRNAs have different base pairs at these positions (60) and elongation factors are thought not to recognize specific sequences in elongator tRNAs. In the three-dimensional structure of the Thermus thermophilus elongation factor EF-Tu·GDPNP·aminoacyl-tRNA ternary complex (44, 45), with the exception of the 3'-terminal A residue, all of the contacts of EF-Tu with the tRNA are with the sugar phosphate backbone (see below). Therefore, it is more likely that the human initiator tRNA is inactive as an elongator because of the sequence-dependent perturbation of the RNA helix in the TC stem of the initiator tRNA that blocks its binding to the elongation factor. Because the sequences of all vertebrate initiator tRNAs are identical (21, 22, 46, 58, 60), our conclusion is likely to be valid for all vertebrate initiator tRNAs and vertebrate protein synthesis systems.

The structural perturbation could be due to weakening of the RNA helix because of the consecutive U:A and A:U base pairs at positions 50:64 and 51:63, respectively, or due to local change in the helical structure which alters the sugar phosphate backbone conformation and/or position from that of a regular RNA helix. Although unlikely in this case, a stiffening of the RNA helix could also have a similar effect on EF binding. In the crystal structure of the EF-Tu·GDPNP·aminoacyl-tRNA ternary complex, the acceptor stem-TC stem extended helix is bent at the junction and twisted in such a way that the position of the 3'-terminal A is shifted by as much as 16 Å (44). A stiffening of the acceptor stem-TC stem junction could affect binding of the aminoacyl-tRNA to the EF by preventing this change.

In fungal and plant initiator tRNAs, the presence of a bulky 2'-O-phosphoribosyl modification at ribose 64 (11, 20, 57) prevents these tRNAs from acting in elongation (1, 19, 30). This bulky modification protrudes into the minor groove of the TC stem helix (4) and most likely acts as a steric block for the binding of eEF1. Therefore, the basic principle used to block the binding of the vertebrate, plant, and yeast initiator tRNAs to eEF1 is similar, sequence-dependent perturbation of the RNA helix in the TC stem in the case of vertebrates and use of a bulky modification in the cases of fungi and plants.

Besides initiator tRNAs, there is another class of tRNAs, selenocysteine tRNA (tRNA^Sec), which also does not bind to the normal EF, i.e., EF-Tu in E. coli and eEF1 in eukaryotes (3). A special EF, the product of the selB gene in E. coli, is used to transport E. coli tRNA^Sec to the ribosome (18). In tRNA^Sec, the nature of the base pairs corresponding to base pairs 49:65, 50:64, and 51:63 of normal tRNAs is important for blocking its binding to EF-Tu (52). These base pairs correspond to the eight, ninth, and tenth base pairs in the acceptor stem-TC stem extended helix in the three-dimensional structure of tRNA (31, 50), and two of these base pairs (50:64 and 51:63) overlap with the sites of the base pair mutations described in the present work. Movement of these base pairs closer to the CCA end by deleting one base pair from the acceptor stem is enough to allow tRNA^Sec to bind to EF-Tu (52), suggesting that EF-Tu is quite sensitive to the structure at or near the junction of the acceptor stem-TC stem extended helix. Therefore, while the rest of tRNA^Sec probably plays a role in preventing tRNA^Sec from binding to EF-Tu (39), an unusual conformation of the tRNA backbone at the junction of the acceptor stem-TC stem extended helix also plays an important role. A similar conformational perturbation could account for the exclusion of eukaryotic tRNA^Sec from binding to eEF1 (52, 60). Therefore, the principle of sequence-dependent perturbation of the RNA helix appears to be used quite widely to exclude specific tRNAs and possibly other RNAs from binding to certain proteins. The effect of the tRNA sequence on conformation at the end of the acceptor stem and on its interaction with proteins has been described before (35, 36, 63).

A secondary determinant which prevents the human initiator tRNA from acting in elongation is the highly conserved A1:U72 base pair at the end of the acceptor stem. Mutation of A1:U72 to G1:C72 allows the human initiator tRNA to clearly act as an elongator in vitro (Fig. 3) and in vivo (Table 1), although the activity of this mutant initiator tRNA is significantly less than those of the TC stem mutants. The corresponding mutant tRNA in the yeast S. cerevisiae also acts as an elongator (2). The role of the A1:U72 base pair as a secondary determinant preventing the activity of eukaryotic initiator tRNAs in elongation is somewhat analogous to the role of bases 1 and 72 in the E. coli initiator tRNA. There, a mismatch between bases 1 and 72 (C1xA72 mismatch in most eubacterial initiator tRNAs) is the primary determinant that prevents the initiator Met-tRNA from binding to EF-Tu. Mutation of C1xA72 to U1:A72 or to C1:G72 allows the mutant tRNAs to bind to EF-Tu and act in elongation in vivo and in vitro (55, 56). It is quite likely that the E. coli initiator tRNA with a C1xA72 mismatch adopts a structure at the end of the acceptor stem that is distinct from that of normal elongator tRNAs (63, 65). Whether the presence of base pair A1:U72 in eukaryotic initiator tRNAs, which is a weaker base pair than the more common base pair G1:C72, allows the initiator Met-tRNAs to adopt a different structure (4, 17) is not known. Interestingly, the only exception to the presence of A1:U72 in eukaryotic initiator tRNAs is in Schizosaccharomyces pombe, in which it is 1:A72 (29), which is also a weak base pair.

Previously, we showed that base pair A1:U72 in human initiator tRNA was important for its activity in initiation in vitro and that it was important specifically for binding to eIF2 (17). This base pair is also important for initiation in vivo in the yeast S. cerevisiae (64). Thus, the A1:U72 base pair, which is conserved in eukaryotic initiator tRNAs, has at least two functions, i.e., (i) in initiation and (ii) in preventing the tRNA from acting in elongation. This dual role of the A1:U72 base pair in eukaryotic initiator tRNAs is reminiscent of the situation in eubacteria such as E. coli, in which the highly conserved mismatch between bases 1 and 72 of the initiator tRNA also has multiple roles (47). The A1:U72 base pair is also conserved in all archaebacterial initiator tRNAs (60). It will be interesting to see if it plays a role similar to that of the eukaryotic initiator tRNAs in archaebacterial protein synthesis systems.

The possible importance of tRNA conformation at the end of the acceptor stem and near the junction of the acceptor stem-TC stem extended helix for binding to the EF can be understood from the crystal structure of the T. thermophilus EF-Tu·GDPNP·aminoacyl-tRNA ternary complex (44, 45) and previous work on the region of the tRNA involved in binding to EF-Tu and eEF1 (27, 28, 51). With the exception of the 3'-terminal A and the aminoacyl-ester group, in the crystal structure, EF-Tu makes no base-specific contacts with the tRNA. All contacts are with the sugar phosphate backbone. The crystal structure shows intimate contact of the protein with the sugar phosphate backbone of the aminoacyl-tRNA at the end of the acceptor stem and in the minor groove of the TC stem region, particularly around the sugar phosphate backbones of nucleotides 50 through 54 and 64 through 67. The work of Haenni, Joshi, and their coworkers with tRNA-like structures in plant viral RNAs also suggests contacts between the TC stem region and EF-Tu and the eEF1 (27, 28). These findings, combined with our results with a vertebrate initiator tRNA and those of Byström, Sprinzl and coworkers with yeast and plant initiator tRNAs (1, 19, 30), suggest that the requirements for eEF1 are quite similar to those for EF-Tu and suggest a common mode of binding of the eubacterial and eukaryotic EFs to the corresponding aminoacyl-tRNAs.

The assay for the elongation activity of mutant initiator tRNAs carrying the U35A36 anticodon mutation relied on their abilities to suppress the amber codon in the CATam27 mRNA (6). Because the U35A36 mutant tRNA is not aminoacylated to any appreciable extent by the endogenous COS1 cell aminoacyl-tRNA synthetases (Fig. 7), the activity of the U35A36/U50G51:C63A41 mutant tRNA as an amber suppressor is dependent on coexpression of E. coli GlnRS. Therefore, similarly to our previous work using an amber suppressor derived from E. coli tRNA^Gln in COS1 cells (14), the present work provides another example of the use of an exogenous aminoacyl-tRNA synthetase for suppression of an amber codon in mammalian cells. As pointed out previously (9), systems such as this may be useful for the isolation of mutants in aminoacyl-tRNA synthetases which activate noncognate amino acids or amino acid analogs and incorporate them into amber suppressor tRNAs (40).

Finally, combination of mutant initiator tRNAs (with the CCU anticodon sequence) and the CATM173R reporter gene used here provides the first documentation of missense suppression in mammalian cells or in any organism except in eubacteria and in yeasts (5, 7, 43).