Structural and sequence elements important for recognition of Escherichia coli formylmethionine tRNA by methionyl-tRNA transformylase are clustered in the acceptor stem.

We show that the structure and/or sequence of the first three base pairs at the end of the amino acid acceptor stem of Escherichia coli initiator tRNA and the discriminator base 73 are important for its formylation by E. coli methionyl-tRNA transformylase. This conclusion is based on mutagenesis of the E. coli initiator tRNA gene followed by measurement of kinetic parameters for formylation of the mutant tRNAs in vitro and function in protein synthesis in vivo. The first base pair found at the end of the amino acid acceptor stem in all other tRNAs is replaced by a C.A. "mismatch" in E. coli initiator tRNA. Mutation of this C.A. to U:A, a weak base pair, or U.G., a mismatch, has little effect on formylation, whereas mutation to C:G, a strong base pair, has a dramatic effect lowering Vmax/Kappm by 495-fold. Mutation of the second basepair G2:C71 to U2:A71 lowers Vmax/Kappm by 236-fold. Replacement of the third base-pair C3:G70 by U3:A70, A3:U70, or G3:C70 lowers Vmax/Kappm by about 67-, 27-, and 30-fold, respectively. Changes in the rest of the acceptor stem, dihydrouridine stem, anticodon stem, anticodon sequence, and T psi C stem have little or no effect on formylation.

Of the two classes of methionine tRNAs present in all organisms, the initiator is used exclusively for initiation of protein synthesis, whereas the elongator is used for inserting methionine into internal peptide linkages (1). In eubacteria, mitochondria, and chloroplasts the initiator is used as formylmethionyl-tRNA (fMet-tRNA).' Following aminoacylation of the initiator tRNA (tRNAfMet), the methionyl-tRNA (Met-tRNA""') is formylated to fMet-tRNAfMet by Met-tRNA transformylase (2). The recognition of Met-tRNAfMet by the Met-tRNA transformylase is highly specific, the enzyme formylates only Met-tRNA""' but not Met-tRNAMet, the elongator species, or any other aminoacyl-tRNA (3). The sequence and/or structural features important for this specific recognition are not known.
In previous work, we described mutagenesis of the E. coli tRNA"" gene followed by functional studies of the mutant tRNAs in protein synthesis in vitro and in vivo. We showed (i) that the GGG:CCC sequence conserved in the anticodon stem of all initiator tRNAs was important in targeting the * This work was supported by Grants GM17151 from the National Institutes of Health and NP114 from the American Cancer Society.
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.
tRNA to the ribosomal P site (4) and (ii) that the Cl.A72 mismatch at the end of the acceptor stem was important in preventing the tRNA from functioning in the elongation step of protein synthesis ( 5 , 6 ) . Here we describe the generation of several more mutants and measurement of kinetic parameters for formylation of the mutant Met-tRNAs by Met-tRNA transformylase. Our results show that the sequence and/or structural elements important for recognition of tRNA by Met-tRNA transformylase are localized mostly around the end of the acceptor stem. These include the discriminator base A73, either a weak or a disrupted base pair in the first position, a G2:C71 base pair in the second position and a C3:G70 base pair in the third position.

MATERIALS AND METHODS
Oligonucleotide-directed Site-specific Mutagenesis-Mutagenic primers were synthesized on an Applied Biosystems 380A DNA synthesizer and purified as described (7). Mutagenesis of single-stranded DNA in M13 vectors used the gapped duplex approach (8) or the phosphorothioate-based approach (9) using the kit from Amersham. The mutant tRNA genes were characterized by sequencing of the entire tRNA gene. Most of the mutant tRNAs were also characterized by fingerprint analysis of nuclease digests (10) and by modified base composition analysis (data not shown).
Identification, Isolation, and Purification of Mutant tRNA-Total tRNAs were isolated (4) and fractionated by gel electrophoresis on either 20 or 12% nondenaturing polyacrylamide gels. Most of the mutant tRNAs could be overproduced. tRNA5 were visualized by ethidium bromide staining for analytical gels or by UV shadowing for preparative gels. The location of mutant tRNAs which were not overproduced and which migrated differently from wild type tRNA were determined by Northern hybridization or hybrid selection as described below.
Localization of Mutant tRNAs on Gels by Northern Hybridization-Northern hybridization followed a procedure modified from that described before (11). One AZM) unit of total tRNA was electrophoresed on a 12% native polyacrylamide gel. tRNA bands were visualized by ethidium bromide staining and transferred to a nytran membrane (Schleicher & Schuell) using an electro-transfer apparatus (Hoefer Scientific Instruments). The membrane was baked a t 75 "C for 1 h and prehybridized at 42 "C for 3 h in a solution containing 50% formamide, 5 X Denhardt's reagent, 0.1% SDS, 5 X SSC, and 150 pg/ ml salmon sperm DNA. Hybridization was at 42 "C overnight in the same solution with a 5'-32P-labeled oligonucleotide probe. The probe used was 24 nucleotides long and was complementary to nucleotides 9-31 of the mutant tRNA in the D-stem and loop regions. The membrane was washed a t room temperature for 30 min in 6 X SSC before autoradiography.
Hybrid Selection of "P-Labeled Mutant tRNAs-Hybrid selection used a procedure slightly modified from that described previously (12). Single-stranded M13 DNA containing the sequence complementary to the desired tRNA was fixed on a piece of nitrocellulose membrane paper (10 pg DNA per 0.5 x 0.5 cm). The membrane was prehybridized a t 42 "C for 1 h with a solution (50 pl) containing 50% formamide, 5 X SSC, and 250 pg/ml E. coli ribosomal RNA in a microcentrifuge tube. "'P-labeled total tRNA (12) was added and hybridization was carried out at 42 "C overnight. The membrane was washed three times with 0.2 X SSC at 42 "C for 60 min each. Hybrid selected tRNAs were eluted by boiling the nitrocellulose paper in 250 pl of water for 5 min, recovered by precipitation with ethanol and by gel electrophoresis on a 12% denaturing polyacrylamide gel. The '"Plabeled hybrid selected tRNAs were renatured and then used as markers to locate the position of the nonradioactive mutant tRNAs on a preparative gel.
Synthesis and Purification of ['"C]Formyl Tetrahydrofolate-["C] Formyl tetrahydrofolate (["C]fTHF) was prepared from tetrahydrofolate and "C-labeled sodium formate using the enzyme formyl tetrahydrofolate synthetase from Clostridium acidi-uria (13), which was a generous gift from Dr. Jesse Rabinowitz (University of California at Berkeley). The ["CJfTHF was purified by chromatography on a column of Whatman cellulose powder (14), lyophilized to dryness, dissolved in 0.1 N HCI, and aliquots (100 pl) were stored at -70 "C. Specific activity of the ['"CJfTHF was 40 pCi/pmol. Under acidic conditions, fTHF is converted to a stable form, N",N"-methenyltetrahydrofolic acid. Before use in formylation, an aliquot was neutralized to pH 7.5 with 1 N NaOH and incubated a t room temperature for 30 min to convert the N",N'"-methenyl-THF into N"'-formyl tetrahydrofolate. N'"-fTHF, the active form for the formylation reaction, is unstable a t neutral pH and was, therefore, used immediately.
Measurement of Kinetic Parameters in Formylation of tRNAs-The assay for formylation used a two-step reaction. tRNA was first quantitatively aminoacylated with cold methionine using purified E. coli MetRS. It was then formylated with ['"CJfTHF using an E. coli extract (S-100) freed of nucleic acids as the source of Met-tRNA transformylase (15).
Incubation was a t 37 "C for 5 min. An aliquot (16 pl) was applied on to Whatman 3MM paper discs, which were then washed for 30 min with cold 10% trichloroacetic acid containing 0.1% folic acid followed by three washes for 10 min each with cold 5% trichloroacetic acid and once for 5 min with ethanol. After drying acid insoluble radioactivity was measured. Activity of Mutant tRNAs in Protein Synthesis in Vivo-Assays for activity of mutant tRNAs in initiation and elongation steps of protein synthesis were as described in the following paper (16).

Mutants of E. coli Initiator tRNA
The mutants used in this work contain changes in the acceptor stem, dihydrouridine stem, anticodon stem, anticodon sequence, and the T$C stem (Fig. 1). For in vivo functional studies changes in the dihydrouridine stem, anticodon stem, or the TJ.C stem were coupled to changes in the anticodon sequence. Mutation in the second base pair of the acceptor stem alone produced very little mutant tRNA presumably because mutant tRNA precursors were either not processed in vivo or were unstable. Therefore, mutations in the second and third base pairs were in most cases coupled to either the T1 or the G72 mutations to generate a base pair between nucleotides 1 and 72.

Isolation and Purification of Mutant tRNAs
Mutant tRNAs with changes in the acceptor stem or anticodon stem migrated to the same position on a native polyacrylamide gel as wild type tRNAgMet. These mutant tRNAs were purified using a single step of gel electrophoresis as described before (4).
Mutant tRNAs with changes in the dihydrouridine stem were expressed to a lower extent and migrated differently from wild type tRNAfMet and tRNAcMet. Their position on 12% native polyacrylamide gels were determined by hybridization. Fig. 2 3 and 4 ) . The mutant tRNA is clearly separated from the elongator tRNAMet and both of the initiator species and can, therefore, be isolated free of these tRNAs.
Mutant tRNAs with changes in the T$C stem also migrated differently from wild type tRNAiMet. The positions of these tRNAs on native polyacrylamide gels were determined by using gel purified, hybrid selected 32P-labeled mutant tRNAs (Fig. 3A) as markers. Fig. 3B shows (i) that each of the hybrid selected tRNAs consists of single tRNA species (lanes 2-4) and (ii) that the C52:G62 and the A49G50A51C52: G62T63C64T65 mutant tRNAs (lanes 4 and 3, respectively) are clearly separated from the wild type E. coli elongator and both of the initiator methionine tRNAs.
Prior to their use as substrate for Met-tRNA transformylase, the methionine acceptance activity of the mutant tRNA species listed in Table I was determined. The mutant tRNAs were found to be essentially pure with the exception of those carrying changes in the D stem or the T$C stem which migrated differently from wild type tRNA on native polyacrylamide gels. The purity of these tRNAs, based on methionine acceptance, were estimated to be 50% for the Cll:G24 mutant, 20% for the CllC12A13:A22G23G24 mutant, 28% for the C52:G62 mutant, and 30% for the A49G50A51C52: G62T63C64T65 mutant.
Kinetic Parameters for Aminoacylation of Mutant tRNAs with E. coli Met-tRNA Synthetase Kinetic parameters for aminoacylation of the mutant tRNAs listed in Table I were determined using purified E.
coli MetRS.' Our results agree with the conclusions of Schulman and co-workers that the anticodon sequence is the major site on E. coli methionine tRNAs for MetRS recognition (17).
Mutation of A73 to G73 and C3:G70 to G3:C70 had small effects in aminoacylation by MetRS (6-7-fold decrease in Vmax/PzP). The rest of the mutants had basically the same K,,, and V, , , parameters as wild type tRNAfMet.
Although the G72 mutant is a poor substrate for Met-tRNA transformylase, it can be quantitatively formylated with an excess of enzyme (20). However, the G72G73 mutant is essentially not formylated. This suggests that the "discriminator" base A73 is also important for formylation.
Mutation of G4:C69 to C4:G69 has a small effect on Vm,J Kzp in formylation (5-fold). Mutations in the rest of the acceptor stem (C5:G68, C6:G67, and C7:G66 mutants) have little or no effect on formylation. These results show that sequences clustered around the end of acceptor stem in E. coli tRNAfMet are important in recognition of tRNAs by E. coli Met-tRNA transformylase. Dihydrouridine Stem Mutants-One of the unique sequence features found in all prokaryotic initiator tRNAs is the presence of a purine 1l:pyrimidine 24 base pair instead of pyrimidine 1l:purine 24 base pair. Mutation of All:U24 in the E. coli tRNA to Cll:G24 lowers Vmax/F$P by &fold. This effect is primarily due to an increase in KZP. However, a mutation which changes 3 of the 4 base pairs in the dihydrouridine stem including the All:U24 base pair (CllC12A13: A22G23G24 mutant, which has the same sequence in the D stem as found in E. coli tRNAG'" and in yeast tRNATy') has no effect on formylation.
Anticodon pairs unique to all initiator tRNAs to T29C30A31:T39G40A41 or of all 5 base pairs in the anticodon stem to the sequence found in t R N A p (C27A28T29C30A31:T39G4OA4lT42G43 mutant) had virtually no effect on formylation. Thus, although the 3 G:C base pairs in the anticodon stem of tRNA""' are important for binding of the tRNA to the ribosomal P site during initiation of protein synthesis (4), they have no direct role in recognition of tRNA by E. coli Met-tRNA transformylase. Anticodon Sequence Mutant-The T35A36 mutant in which the anticodon sequence has been changed from CAU + CUA is now aminoacylated with glutamine (6, 21). The mutant tRNA aminoacylated with glutamine is still a good substrate for Met-tRNA transformylase, (a 2.5-fold difference in V,,,/F~p). This result indicates that the anticodon sequence is not important in recognition of tRNAs by Met-tRNA transformylase.
T$C Stem Mutants-The G52:C62 base pair is conserved in all initiator tRNAs, prokaryotic or eukaryotic, which are substrates for E. coli Met-tRNA transformylase. Mutation of G52:C62 to C52:G62 has no effect on formylation. Mutation which changes 4 of the 5 base pairs in the T$C stem (A49G50A51C52:G62T63C64T65 mutant) including the G52:C62 base pair also has little effect on formylation, suggesting that sequences in the T$C stem are not directly involved in Met-tRNA transformylase recognition of tRNAs.
Properties of the Mutant tRNAs in Initiation of Protein Synthesis in Vivo-The above results suggest that major sequence alterations in the D stem, anticodon stem, anticodon sequence, and T$C stem have no effect on kinetic parameters for formylation of the mutant tRNAs. Therefore, mutant tRNAs carrying these alterations would be expected to function in initiation of protein synthesis unless the mutations affect a step in initiation subsequent to formylation of the initiator tRNA. To test for this, we coupled the multiple base pair mutations in the dihydrouridine stem, anticodon stem or the T$C stem, to a change in the anticodon sequence from CAU to CUA. The latter mutation enables the mutant tRNAs to potentially initiate protein synthesis from UAG in a mutant chloramphenicol acetyl transferase (CAT) gene whose initiator codon has been changed from AUG to UAG (22). Mutant tRNAs which are active in initiation in vivo should produce CAT protein in E. coli transformed with plasmids carrying the various mutant tRNA genes and the mutant CAT gene. Results of such an experiment are shown in Table 11. Mutant tRNAs with alterations in 3 of the 4 base pairs in the D stem and 4 of the 5 base pairs in the T$C stem are quite active in initiation in vivo -50%. This confirms that the sequences are, most likely, not directly important for Met-tRNA transformylase recognition of tRNA and also suggests that they play no direct role in activity of tRNA in initiation.
The mutant tRNAs with changes in the 3 G:C base pairs in the anticodon stem are inactive (Table 11). This result is expected and supports our previous conclusion, based on in vitro studies, that the 3 G:C base pairs in the anticodon stem are important for binding of the initiator tRNA to the ribososmal P site (4). Table I1 also shows that the Tl/T29C30A31: T39G40A41/T35A36 mutant tRNA, which is inactive in initiation is, nevertheless, active in suppression of amber codons at the ribosomal A site (rightpanel). Therefore, lack of activity of this mutant in initiation in vivo is not due to defects in synthesis or in aminoacylation of the tRNA.

DISCUSSION
We have shown that the sequence and/or structural features important for the highly specific recognition of E. coli tRNAfMe' by E. coli Met-tRNA transformylase are, virtually all, clustered around the end of the acceptor stem. These include a weak or disrupted base pair between positions 1 and 72 at the end of the acceptor stem, a G2:C71 base pair, a C3:G70 base pair and A73 at the discriminator position (Fig.  1). The G4:C69 base pair and the All:U24 base pair could also play some role in the recognition process. These features are conserved in all eubacterial, mitochondrial, and chloroplast tRNAs (23). Alterations in the rest of the tRNA in the D stem, anticodon stem, anticodon sequence, T\CC stem, and in base pairs 5-7 in the acceptor stem of E. coli tRNAiMet have no effect on the kinetic parameters for formylation. Base Pair 1:72"The constellation of nucleotides in the acceptor stem important for specific interaction of tRNAtMet with Met-tRNA transformylase appears, at first glance, to be similar to that of E. coli tRNAG1" with GlnRS (24). Thus, like GlnRS, Met-tRNA transformylase requires a weak or a disrupted base pair between nucleotides 1 and 72. In the crystal structure of GlnRS-tRNAG'" complex, there are no sequence specific contacts between GlnRS and nucleotides 1 or 72 of tRNAG1" implying that the nature of nucleotides 1 and 72 i s less important than the fact that the base pair between them be disrupted (6,24). Whether this is also the case for Met-tRNA transformylase interaction with tRNA is not known. It is possible that besides requiring that the 1:72 base pair be disrupted, the Met-tRNA transformylase also makes direct contact with one or both of the nucleotides. Some indication for this comes from the fact that while most eukaryotic cytoplasmic initiator tRNAs (yeast, human etc.) can be quantitatively formylated by the E. coli enzyme ( E ) , the V,,,/ KZp for these tRNAs is 50-100-fold lower than for E. coli  (Table I). Thus, as in the case of E. coli GlnRS-tRNAG1" interaction, this base pair is important for recognition of tRNAfMet by Met-tRNA transformylase. The importance of this base pair in formylation is also underscored by the fact that although the G72 mutant with a "strong" Cl:G72 base pair at the end of the acceptor stem is a very poor substrate, it can be quantitatively formylated in vitro with an excess of enzyme (20). However, coupling of the T2:A71 mutation with the G72 mutation essentially abolishes formylation. The importance of G2:C71 in formylation may also explain why plant cytoplasmic initiator tRNAs are not formylated by the E. coli . Of all the eukaryotic cytoplasmic initiator tRNAs, the plant tRNAs are the only ones which contain a U2:A71 base pair instead of a G2:C71 base pair in others. Base Pair 3:70"Mutation of C3:G70 to any one of the other 3 base pairs reduces V,,,,,/P~p by a factor of 27-to 67-fold ( Table I). Assuming that Met-tRNA transformylase makes sequence-specific contact with this base pair, the fact that mutation of C3:G70 to any of the other 3 base pairs has more or less equally detrimental effects would seem to disagree with the notion that proteins contact base-paired regions of RNAs primarily in the minor groove. While the results obtained with the T3:A70 and A3:T70 mutants would be understand- able, that with the G3:C70 mutant would be less so, since G:C and C:G base pairs are thought to have basically the same functional groups at approximately similar positions in the minor groove (26-28). However, it is still possible that Met-tRNA transformylase interacts directly with the 3:70 base pair. The lower Vmax/PzP for the G3:C70 mutant compared to the wild type tRNA is more likely due to a change in the structure of the acceptor stem in the mutant than due to loss of a sequence specific contact in the minor groove (see following paper (16) for detailed discussion).
Outside of the acceptor stem, the only mutant with an alteration in kinetic parameters for formylation is the All:U24 to Cll:G24 mutant. The finding that mutation of All:U24 to Cll:G24 increases Kzp for formylation is consistent with the expectation that the Pull:Pyr24 base pair, unique to eubacterial, mitochondrial, and chloroplast initiator tRNAs, would play some role in the overall process of protein synthesis initiation. However, a mutant with changes in 3 of the 4 base pairs in the D stem including the same All:U24 to Cll:G24 change behaves normally in formylation. In addition, when coupled to an anticodon sequence change from CAU + CUA, the mutant with 3 of the 4 base pair changes in the D stem appears quite active in initiation in uiuo (Table 11). Thus the importance of the All:U24 base pair for formylation of tRNAgMet depends upon the sequence context in and around the D stem; it is, therefore, unlikely that Met-tRNA transformylase interacts directly with the All:U24 base pair of tRNAtMet. One possibility is that All:U24 to Cll:G24 change alters local structure of the tRNA and that the increase in P , P P in formylation for the Cll:G24 mutant is due to subtle effects of an altered tRNA structure. It is interesting to note in this regard that Smith and Yarus (29) have proposed subtle structural changes involving alternative tertiary base pairing between the nucleotide at position 9 and base pairs 12:23 or 11:24, depending upon the nature of the 11:24 base pair, as being responsible for certain types of misreading of noncognate codons on the ribosome.
Finally, are sequence and/or structural features around the end of the acceptor stem, in the context of a tRNA structure, sufficient for recognition of a tRNA by Met-tRNA transformylase or do other regions of the tRNA also contribute towards the overall specificity of the recognition process? To answer this question, we have introduced the above features in the acceptor stem of E. coli initiator tRNA to E. coli glutamine and methionine elongator tRNA species. Although kinetic parameters for formylation have not been determined, glutamine tRNA carrying multiple changes including those at positions 1, the 3:70 base pair and 73 among others is a substrate for Met-tRNA transformylase. Similar studies are in progress on mutants of E. coli elongator methionine tRNA.