Suppression of amber codons in vivo as evidence that mutants derived from Escherichia coli initiator tRNA can act at the step of elongation in protein synthesis.

The absence of a Watson-Crick base pair at the end of the amino acid acceptor stem is one of the features which distinguishes prokaryotic initiator tRNAs as a class from all other tRNAs. We show that this structural feature prevents Escherichia coli initiator tRNA from acting as an elongator in protein synthesis in vivo. We generated a mutant of E. coli initiator tRNA in which the anticodon sequence is changed from CAU to CUA (the T35A36 mutant). This mutant tRNA has the potential to read the amber termination codon UAG. We then coupled this mutation to others which change the C1.A72 mismatch at the end of the acceptor stem to either a U1:A72 base pair (T1 mutant) or a C1:G72 base pair (G72 mutant). Transformation of E. coli CA274 (HfrC Su- lacZ125am trpEam) with multicopy plasmids carrying the mutant initiator tRNA genes show that mutant tRNAs carrying changes in both the anticodon sequence and the acceptor stem suppress amber codons in vivo, whereas mutant tRNA with changes in the anticodon sequence alone does not. Mutant tRNAs with the above anticodon sequence change are aminoacylated with glutamine in vitro. Measurement of kinetic parameters for aminoacylation by E. coli glutaminyl-tRNA synthetase show that both the nature of the base pair at the end of the acceptor stem and the presence or absence of a base pair at this position can affect aminoacylation kinetics. We discuss the implications of this result on recognition of tRNAs by E. coli glutaminyl-tRNA synthetase.

Every prokaryotic initiator tRNA sequenced has a 5'-terminal nucleotide which is not base-paired to the corresponding nucleotide on the 3' side of the amino acid acceptor stem (1). In Escherichia coli initiator tRNA, the 5'-terminal nucleotide is C (Cl), and the corresponding 3' nucleotide is A (A72). Recently, we showed that a single base mutation which generated either a U1:A72 base pair (T1 mutant) or a C1:G72 base pair (G72 mutant) enabled these tRNAs to act as elongators, i n vitro, whereas the double mutant (ClA72 + UlG72) did not (2). The G72 mutant was more active than the T1 mutant, and this was ascribed to a tighter binding of the G72 mutant to elongation factor EF-Tu' compared to the T1 * This work was supported by National Institutes of Health Grant GM17151 and American Cancer Society Grant NP114. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The most direct approach to demonstrate i n vivo activity of mutant initiator methionine tRNAs in elongation is to isolate E. coli strains in which elongator tRNAMet synthesis or function is conditionally defective and to show that this defect could be complemented by mutants derived from initiator tRNAs. Isolation of such strains is, however, difficult since there are two genes for elongator tRNAMet, and both of these genes are part of a multiple tRNA gene operon in which one of these elongator tRNAMet genes is the promoter proximal gene for the entire operon (3, 4). Thus, conditional inactivation of the promoter or disruption of tRNAMet genes would affect expression or processing of transcripts of other tRNA genes which are downstream and some of which are essential for growth (3, 4). An alternative approach is to generate a nonsense suppressor derived from the initiator tRNA. Nonsense suppression, by definition, is a property of elongator tRNAs since suppressors insert amino acids internal to the polypeptide chain. In this work, we have changed the anticodon sequence of E. coli initiator tRNA from CAU into CUA, combined the anticodon mutation with mutations which introduce a base pair at the end of the acceptor stem, and examined whether these mutants suppress amber codons i n vivo. Since mutants carrying the anticodon sequence change from CAU to CUA are aminoacylated with glutamine (5), we have also analyzed the kinetic parameters for aminoacylation of the partially purified mutant E. coli initiator tRNAs (6) using purified E . coli GlnR synthetase. We show that both the presence or absence of a base pair at the end of the acceptor stem and the nature of this base pair can affect kinetic parameters of aminoacylation, and we discuss the implications of this result concerning recognition of tRNAs by E. coli GlnR synthetase.

MATERIALS AND METHODS
Isolation of Mutant tRNAs-Mutagenesis of E. coli initiator t R N A F gene and isolation of mutant tRNAs were as described (2, 6).
In Viuo Actiuity of Mutant tRNAs in Suppression-The mutant tRNA genes were subcloned into the PstI-EcoRI site of pBR322 and used to transform E. coli CA274 (HfrC Su-IacZ125am trpEam) (7).
Transformants were scored for suppression of amber codons using the following criteria: formation of purple colonies on MacConkey/ lactose plates, formation of blue colonies on IPTG/X-Gal plates, and ability of transformants to grow on minimal plates lacking tryptophan and with lactose as the sole carbon source (8).
Assay for &Galactosidase Activity in E. coli CA274 Transformants-&Galactosidase was measured according to Miller (9). Briefly, E. coli cells were grown in 2YT media containing tetracycline and 1 mM IPTG to Am of 0.6 to 0.8. After centrifugation, cell pellets were resuspended into one-tenth volume of lacZ buffer and chilled on ice. An aliquot of cell suspension was diluted into 1 ml of lacZ buffer, and 1 drop of CHCb and 1 drop of 0.1% SDS were added and vortexed 6504 briefly for lysis. The amount of cell suspension used varied for the different transformants, more for cells producing low levels of @galactosidase and less for those producing high levels of @-galactosidase. After 5 min at 28 "C, 0.2 ml of 0-nitrophenyl-@-galactopyranoside solution was added, and the mixture was incubated at the same temperature, following which 0.5 ml of Na2C03 was added. The sample was then chilled on ice and centrifuged. Supernatant was taken out for measurement of absorbance at 420 nm. The @-galactosidase activity units given are in hours.

Amimacylation of Mutant tRNAs by E. coli GlnR Synthetase-
Assay conditions for aminoacylation by GlnR synthetase were essentially as described (5). The incubation mixture (23 pl) contained 100 mM sodium cacodylate, pH 7.1, 10 mM magnesium acetate, 2 mM ATP, 310 PM [3H]glutamine, and varying amounts of either E. coli tRNAf'" or tRNA"" mutants which were isolated by gel electrophoresis. Incubation was for 2.5 min at 37 "C in the presence of appropriate dilutions of purified E. coli GlnR synthetase (gift from John Perona, Yale University). An aliquot (18 p1) from each reaction was used for measurement of acid-insoluble radioactivity.

Activity of Mutant tRNA in Vivo in Suppression of Amber
Codons-Analysis of CA274 (HfrC Su-lacZam trpEam) transformants carrying the mutant tRNA genes in multicopy plasmids showed that mutants with changes in either the anticodon sequence alone (T35A36 mutant) or in the acceptor stem alone (G72G73 mutant) were inactive in suppression, whereas those with changes in both regions (the Tl/T35A36, T35A36/ G72, and T35A36/G72G73 mutants) were all active in suppressing the amber mutant alleles in lacZ125am and in trpEam (Table I and Fig. 2). The intensity of colony color on MacConkey/lactose and IPTG/X-Gal plates indicated that, of the three mutants active in suppression in vivo, the T35A36/G72G73 mutant was the strongest suppressor. This was also indicated by the relative growth rate of the various transformants on minimal lactose plates lacking tryptophan. Following incubation at 37 "C for 24 h, growth was visible only in case of CA274 transformants carrying the T35A36/ G72G73 mutant tRNA gene or the E. coli Su+2 tRNAG1" gene. A further 24-36 h of incubation was necessary for visible growth of CA274 transformants carrying the Tl/T35A36 or the T35A36/G72 mutant tRNA genes (Fig. 2). Further evidence for a high level of amber suppression in CA274 transformants carrying the T35A36/G72G73 mutant initiator tRNA gene also comes from the finding (data not shown) that this transformant is able to sustain growth of the recombinant virus M13mp8, which has amber mutations in two of its essential genes.
The efficiency of suppression of the lacZ125am allele in E. coli CA274 by the various mutant initiator tRNA9 was quantitated by measuring @-galactosidase activity in the various transformants (Table 11). This assay (9) also indicated that the T35A36/G72G73 mutant was the strongest suppressor, whereas the Tl/T35A36 and the T35A36/G72 mutants were relatively weak suppressors. In contrast, the T35A36 and the G72G73 mutants were essentially inactive. The efficiency of suppression with the T35A36/G72G73 mutant was approximately half that of the E. coli Su+2 suppressor, when the latter tRNA gene was part of a multicopy plasmid vector, which carried the ColEl replication origin and in which the Su+2 tRNA gene was under the control of the lacUV5 promoter (10).
Amirwacylation of Mutant Initiator tRNA with Glutamine-Schulman and Pelka (5) showed previously that an E. coli initiator tRNAMet mutant in which the anticodon sequence was changed from CAU to CUA was aminoacylated with glutamine in vitro. We have confirmed this with each of the mutant tRNAs derived from t R N A p which carries the anticodon sequence change (data not shown). In addition, with the T35A36/G72G73 mutant which is the strongest of the amber suppressors (Table I, Fig. 2), we have examined whether this tRNA inserts glutamine in vivo in response to amber codons as analyzed by its ability to suppress the lacZam mutation in E. coli BT235 (11). This E. coli strain carries the lacZlOOOarn allele which requires insertion of glutamine at the site of amber mutation for enzyme activity (7, 12). Based on the ability of E. coli BT235 transformants to grow on minimal lactose plates and to form blue colonies on IPTG/ X-Gal plates (data not shown, experiment kindly performed by Dr. John Rogers of Yale University), the T35A36/G72G73 mutant E. coli tRNA"" most likely inserts glutamine in vivo in response to amber codons.
Kinetic Parameters in Aminoacylation of Mutant Initiator tRNAs by E. coli GlnR Synthetase- Table I11 compares the kinetic parameters in aminoacylation of the various mutant initiator tRNAs to those for E. coli tRNA$'". Of the mutants used in this study, the best substrate for E. coli GlnR synthetase is the T35A36 mutant with the anticodon sequence changed from CAU to CUA. The Vmex/KIP for this mutant is about 200-fold lower than that for E. coli tRNA:' " (Table 11).
This value agrees reasonably well with an approximately 400fold difference previously reported by Schulman and Pelka (5) between the corresponding tRNArmet. mutant and tRNA?'". Interestingly, the effects of introducing changes in the acceptor stem on the T35A36 mutant initiator tRNA are quite variable. Introduction of a Ul:A72 base pair at the end

DISCUSSION
We have shown that the absence of a base pair at the end of the acceptor stem, which is a hallmark of all prokaryotic initiator tRNAs, prevents these tRNAs from acting in elongation in vivo. We changed the anticodon sequence of E. coli initiator tRNA such that the tRNA has the potential capability of reading the amber termination codon UAG and combined this with mutations which generate either a U:A or a C:G base pair at the end of the acceptor stem. The mutant tRNA with the anticodon sequence change alone is inactive in amber suppression in uivo, whereas mutants carrying both the anticodon sequence change and the acceptor stem sequence change are active.
The inactivity of the T35A36 mutant tRNAmet as an amber suppressor differs from previous results demonstrating in vivo suppression by mutants derived from E. coli Su+3 tRNA which lack a Watson-Crick base pair at the end of the acceptor stem (7). There are several possible explanations for this discrepancy. 1) The T35A36 mutant tRNAmet is prevented from acting as a suppressor because of formylation of the T35A36 Gln-tRNAMet in vivo to fGln-tRNAmet. This is unlikely since the T1 mutant tRNAmet, which is almost as good a substrate for Met-tRNA formylase in vitro as the wild-type tRNA"": is active as an amber suppressor (Table I and Fig. 2).
2) The T35A36 mutant tRNA""' is an extremely weak suppressor, and the criteria we have used for amber suppression (Table   I) E. coli GlnR synthetase V, . and K,,,"PP were determined from Lineweaver-Burke plots. Relative Vm.,/KmaPP is the ratio of Vmru/KmaPp of each tRNA to the VmaJKmaPp of the T35A36/G72 mutant.

Mutants
Relative V-/KmW tRNA:'" 6325 T35A36 Tl/T35A36 9 T35A36/G72 1 T35A36IG72G73 5 -~ ~~ provided as aminoacyl-tRNAs, and assays are carried out in the absence of formyltetrahydrofolate. Thus, the activity of a mutant tRNA in elongation in vitro is simply a function of its affinity for EF-Tu and for the ribosomal A site. Second, while the mutants used for studying in vitro elongation activity contained only changes in the acceptor stem and are aminoacylated with methionine, mutants used for assay of in uiuo elongation activity (i.e. suppression) contained changes in both the anticodon sequence and the acceptor stem and are aminoacylated not with methionine but with glutamine (5). Furthermore, the kinetic parameters for aminoacylation with glutamine are different for the various mutants (Table   111). Thus, in uiuo levels of Gln-tRNA"et for the various mutants could vary widely. This possibility could explain the fact that the efficiency of amber suppression in cells carrying the Tl/T35A36 and the T35A36/G72 mutant tRNAmet genes are about equal. The T35A36/G72 mutant tRNA may have a higher affinity for EF-Tu than the Tl/T35A36 tRNA (2), but this increased affinity for EF-Tu could be compensated for by a lower concentration of Gln-tRNAMet corresponding to the T35A36/G72 mutant compared to the Tl/T35A36 mutant. Similarly, an explanation for the higher efficiency of suppression with the T35A36/G72G73 mutant tRNA over the T35A36/G72 mutant tRNA is that the former is a better substrate for E. coli GlnR synthetase. Consequently, the steady state concentration in uiuo of Gln-tRNA"" corresponding to the T35A36/G72G73 mutant is likely to be higher than that of the T35A36/G72 mutant.
Results on kinetic parameters for aminoacylation of the mutant tRNA9 with glutamine provide insights into how E. coli GlnR synthetase might interact with its tRNA substrate. The most surprising result is that a mutation which generates a Ul:A72 base pair conserved in all eubacterial, chloroplast, and mitochondrial glutamine tRNA5 does not result in an increase in the VmaX/K,,,. The T35A36 mutant tRNA with a C1. A72 mismatch at the end of the acceptor stem is as good, if not a better substrate, as the Tl/T35A36 mutant with a Ul:A72 base pair. The T35A36/G72 mutant with a Cl:G72 base pair is, on the other hand, a much poorer substrate. Thus, the Ul:A72 base pair conserved in glutamine tRNA5 may be there not because GlnR synthetase contacts specific functional groups in a U:A base pair, but is there to provide structural flexibility at the end of the acceptor stem. The structural flexibility provided by a weak base pair such as Ul:A72 or a mismatch may be needed for GlnR synthetase to interact with G73 or for the tRNA -CCA end to tit into the catalytic pocket of GlnR synthetase or both. The occurrence of Ul:A72 base pair in eubacterial, mitochondrial, and chloroplast tRNAG'" instead of no base pair, as in E. coli tRNAmet, may be viewed as a compromise between the need for a base pair at this position for binding to EF-Tu and requirements for E. coli GlnR synthetase that this base pair be relatively weak compared to other base pairs.
The hypothesis that structural flexibility at the end of the acceptor stem may be important in aminoacylation of tRNA by E. coli GlnR synthetase has been proposed before (7, 16), although in the absence of data on kinetic parameters for aminoacylation, and is now supported by analysis of mutants derived from several different E. coli tRNA9 which can be aminoacylated by E. coli GlnR synthetase. 1) Three of the four mutants derived from E. coli Su+3 tRNATy' which insert glutamine in uiuo, have a disruption in either the first or the second base pair in the acceptor stem (7). 2) The Su+7 amber suppressor derived from tRNAm by an anticodon sequence change from CCA to CUA inserts glutamine in uiuo. This tRNA has Al:U72, a "weak" base pair, and not G:C or C:G base pair at this position (13). 3) Mutants derived from E. coli Su+l tRNA"' which insert glutamine in uiuo have recently been generated by site-specific mutagenesis (17). One of the changes replaced the Gl:C72 base pair by a Ul:A72 base pair. 4) Finally, a mutant tRNA derived from E. coli tRNAGly which inserts glutamine in vivo has also been generated recently (18). This tRNA differs from other tRNAs aminoacylated with glutamine in that it has a Gl:C72 base pair at the end of the acceptor stem. Interestingly, the mutation that resulted in its aminoacylation with glutamine in vivo was the change of G3:C70 in the acceptor stem to G3:U70. Presumably, the G:U base pair destabilizes the acceptor stem sufficiently to provide the structural flexibility needed for aminoacylation by E. coli GlnR synthetase. These results taken together support strongly the hypothesis that for aminoacylation of a tRNA by E. coli GlnR synthetase, the exact nature of the base pair at the end of the acceptor stem is not as crucial as the fact that the base pair be such as to provide structural flexibility in this region of the molecule.
The possible importance of structural flexibility in the acceptor stem in aminoacylation of tRNA by E. coli GlnR synthetase provides a second example in which structural flexibility could play an important role in interactions between tRNA and proteins. Studies on formylation of E. coli tRNAmet mutants ( 19)2 have shown that tRNA5 with C1. A72 or U1. G72 mismatches, or a Ul:A72 base pair at the end of the acceptor stem, are all almost equally good substrates for Met-tRNA formylase, whereas tRNA with a Cl:G72 base pair is a very poor substrate. It would, therefore, not be surprising if further work reveals other examples in which structural flexibility in the region proximal to the -CCA end of tRNA is important in interaction between tRNA and aminoacyl-tRNA synthetases or other proteins.