Effect of the amino acid attached to Escherichia coli initiator tRNA on its affinity for the initiation factor IF2 and on the IF2 dependence of its binding to the ribosome.

We show that the nature of the amino acid in the formylaminoacyl-tRNA influences initiation factor (IF) 2 dependence of its ribosome binding and that this IF2 dependence reflects the relative affinity of the formylaminoacyl-tRNA for the initiation factor IF2. We compared the template-dependent ribosome binding activities, in the presence of initiation factors, of wild type and anticodon sequence mutants of Escherichia coli initiator tRNAs that carry formylmethionine (fMet), formylglutamine (fGln), or formylvaline (fVal). The fGln-tRNA bound less well than fMet-tRNA whereas the fVal-tRNA bound as well as fMet-tRNA. The rate and extent of binding of fGln-tRNA to the ribosome was significantly increased by further addition of purified initiation factor IF2. In contrast, the binding of fVal-tRNA or fMet-tRNA was not affected much by the addition of IF2. Using gel mobility shift assay, we have measured the apparent Kd values of the IF2·formylaminoacyl-tRNA binary complexes. These are 1.8, 3.5, and 10.5 μM for fMet-tRNA, fVal-tRNA, and fGln-tRNA, respectively.

Two classes of methionine tRNAs are present in all organisms (reviewed in Refs. [1][2][3][4]. One of these, the initiator, is used for initiation of protein synthesis whereas the other, the elongator, is used for insertion of methionine internally. In eubacteria and in eukaryotic organelles such as mitochondria and chloroplasts, the initiator tRNA is used as formylmethionyl-tRNA (fMet-tRNA). The fMet-tRNA binds to the P site on the ribosome, whereas the elongator tRNAs bind to the A site. Binding of the fMet-tRNA to the P site is facilitated by the initiation factors IF1, 1 IF2, and IF3 in Escherichia coli (5). IF2 is thought to help the ribosome select fMet-tRNA over other tRNAs by virtue of the fact that, outside of the peptidyl-tRNAs on the ribosome, the fMet-tRNA is the only tRNA in the cell that carries an N-acyl-amino acid (6). However, the mechanism by which IF2 helps the ribosome achieve this selection is not clear.
Results of in vitro studies originally led to the suggestion that IF2 acts as a carrier of fMet-tRNA to the ribosome in much the same way as the elongation factor EF-Tu (or eEF-1 in the case of eukaryotes) does for aminoacyl-tRNAs and the eukaryotic initiation factor elF-2 does for the corresponding initiator methionyl-tRNA (reviewed in Ref. 1). However, although IF2 and fMet-tRNA can form a specific binary complex, this com-plex is weak and dissociates readily in the presence of magnesium ions (7)(8)(9). Based on the estimated concentrations of IF2 in cells and the affinity of IF2 for the 30 S ribosome in the presence of the other initiation factors, it has been suggested that all of the 30 S ribosomal subunits are saturated with IF2 in vivo (5,10). This and other in vitro results indicating that IF2 stimulated the binding also of elongator aminoacyl-tRNAs to the 30 S ribosome without any evidence for the formation of a binary complex between aminoacyl-tRNA and IF2 has led to the proposal that IF2 works at the ribosomal level by binding first to the 30 S subunit and stimulating the binding of fMet-tRNA to the ribosome rather than binding to fMet-tRNA and carrying it to the ribosome (3,6,11).
For structure-function relationship studies of E. coli initiator tRNA in vivo, we previously developed a strategy based on use of anticodon sequence mutants of E. coli initiator tRNA (12,13). Two of these mutant tRNAs, U35A36 and G34C36, have the anticodons CUA and GAC and are aminoacylated, respectively, with glutamine and valine (14,15). These mutant tRNAs initiate protein synthesis in vivo with fGln or fVal in the presence of mutant chloramphenicol acetyltransferase genes which carry UAG or GUC, respectively, as the initiation codons (12,16). A surprising result was that overproduction of IF2 led to a dramatic increase in the amount of chloramphenicol acetyltransferase protein initiated with fGln (9). Overproduction of IF2 also increased the activity in initiation of a tRNA derived from E. coli elongator glutamine tRNA and which most likely carried fGln (9,17). In contrast, the amount of chloramphenicol acetyltransferase protein initiated with fVal was not affected by overproduction of IF2 (16). These results raised the possibility that fGln-tRNA binds less well to IF2 and is, therefore, less active in initiation compared to fMet-tRNA or fVal-tRNA. Overproduction of IF2 would compensate for the poor binding and lead to increased utilization of fGln-tRNA in initiation and, thereby, to increased synthesis of the chloramphenicol acetyltransferase protein. This paper reports the results of in vitro work designed to test this possibility. We have compared the codon-and initiation factor -dependent binding of fMet-tRNA, fGln-tRNA, and fVal-tRNA to the ribosome and the effect of further additions of IF2 on the binding. We have also measured directly the relative affinities of IF2 for the mutant initiator tRNAs carrying the different formyl amino acids. The results of these in vitro experiments support previous interpretations of the in vivo results (18) and suggest that IF2 acts as a carrier of the initiator tRNA to the ribosome. coli CA274 overexpressing the ValRS gene from a plasmid (16) and was freed of tRNAs by DEAE-cellulose chromatography (21).
Purification of Wild Type and Mutant Initiator tRNAs-The wild type and U35A36 mutant initiator tRNAs were expressed in E. coli B105 (22). The G34C36 mutant tRNA was expressed in E. coli CA274. The desired tRNA was purified by electrophoresis of total tRNA on a 15% native polyacrylamide gel (23). This system separates the wild type and mutant initiator tRNA 2 fMet species from all other tRNAs (19,23). 5Ј Labeling of Wild Type and Mutant Initiator tRNAs-tRNA was dephosphorylated by treatment with calf intestinal phosphatase and then labeled at the 5Ј-end with 32 P using T4 polynucleotide kinase (24). The 5Ј-labeled tRNA, purified by electrophoresis on a 10% polyacrylamide, 8 M urea gel (40 ϫ 20 ϫ .04 cm), was eluted from the gel slice by shaking overnight in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA containing 10% phenol. After extraction with phenol/chloroform, the 5Ј-32 P-labeled tRNA was precipitated with ethanol in the presence of 5 g of the corresponding nonradioactive tRNA. This 32 P-labeled tRNA was mixed with the corresponding unlabeled purified tRNAs for the preparative aminoacylation and formylation reactions described below.
Preparative Aminoacylation and Formylation of Wild Type and Mutant Initiator tRNAs-Two types of formylaminoacyl-tRNAs (fAA-tRNAs) were prepared. One contained a 5Ј-32 P label in the tRNA in which the fAA moiety is unlabeled. The other does not have a 5Ј-32 P label but carries a label in the amino acid of the fAA moiety. The labels used were 35  After incubation, the fAA-tRNAs were extracted with phenol/chloroform and precipitated with 0.3 M NaOAc (pH 6.0) and 2.5 volume of ethanol. The tRNA pellet was dissolved in 100 l of 5 mM NaOAc, pH 6.0, and passed through a Sephadex G-50 column. The tRNA was precipitated again with ethanol, and finally redissolved in 100 l of 5 mM NaOAc, pH 6.0, and stored at Ϫ20°C. The extent of aminoacylation and formylation of the tRNAs were determined using acid urea gel electrophoresis (25). The 5Ј-32 P-labeled fAA-tRNAs were used for the ribosome binding experiments shown in Fig. 3 and for the gel mobility shift assays.
Purification of Trinucleotides-Trinucleotides (GUC and UAG, each about 15 A 260 units) deprotected in 1 M tetra-n-butylammonium hydroxide were synthesized by the MIT Bio-Polymer Lab. To remove salt, 75 l of the oligonucleotide solution was mixed with 75 l of water and applied as an ϳ8-cm band onto a sheet of Whatman 3MM paper (22 ϫ 60 cm). The chromatogram was developed with 1-propanol/concentrated NH 4 OH/H 2 O (55:10:35, v/v/v) for approximately 12 h (26). The position of the oligonucleotide was localized under UV light and the oligonucleotide was eluted from the paper with water. AUG was purchased from Miles Laboratories.
Gel Mobility Shift Analysis of IF2⅐fAA-tRNA Complexes-Gel mobility shift assay (reviewed in Ref. 29) was modified from a protocol provided by Dr. C. Forster. 2 24 pmol of IF2 (in 500 mM NaCl, 20 mM Tris-HCl, pH 8.0, 10 mM dithiothreitol, and 50% glycerol) from B. stearothermophilus was mixed with 2 l of 5 ϫ buffer (250 mM MOPS/ NaOH, pH 7.5, 50 mM MgCl 2 , 100 mM KCl, and 5 mM dithiothreitol) and 4 g of total E. coli tRNAs in a total volume of 5 l. After leaving in ice for 5 min, the solution was mixed with 5 l of the appropriately diluted 5Ј-32 P-labeled fAA-tRNA for another 10 min at room temperature. 2 l of 50% glycerol was added to this mixture and the mixture was subjected to electrophoresis on a 4% polyacrylamide gel (acrylamide/bisacrylamide ratio, 40:1; 10 ϫ 10 ϫ 0.1 cm). The gel was run in 20 mM MOPS/NaOH, pH 7.5, at 100 V and room temperature for about 2 h. The gel was fixed in 20% ethanol, 10% acetate, dried, and exposed. The amounts of the free fAA-tRNA and the bound fAA-tRNA were determined using a Molecular Dynamics PhosphorImager with Image Quant Software. Fig. 1 shows the clover leaf structure of the wild type and mutant initiator tRNAs. The tRNAs were purified by polyacrylamide gel electrophoresis, aminoacylated in vitro with either methionine (for the wild type tRNA), glutamine (for the U35A36 mutant), or valine (for the G34C36 mutant), and then formylated. Two types of formylaminoacyl-tRNAs (fAA-tRNAs) were used for ribosome binding studies. One carried the radioactive label in the amino acid ([ 35 S]methionine, [ 3 H]glutamine, and [ 3 H]valine), the other carried the radioactive label at the 5Ј-terminal phosphate. Fig. 2 shows an acid urea-polyacrylamide gel electrophoretic analysis of the 5Ј-32 P-labeled fAA-tRNAs used and demon-2 C. Forster, personal communication. strates that the tRNAs are fully aminoacylated and formylated (25). In each case, there is a single band corresponding to fAA-tRNA (lanes 2, 4, and 6), which migrates differently from the corresponding tRNAs (lanes 1, 3, and 5, respectively) or aminoacyl-tRNAs.

Preparation of Formylaminoacyl-tRNAs Used in this Study-
Binding of fMet-, fVal-, and fGln-tRNAs to the Ribosome- Table I shows the binding of the fAA-tRNAs to NH 4 Cl washed ribosomes in the presence of the corresponding codons and mixture of initiation factors present in a ribosomal salt wash. Binding is stimulated significantly in the presence of the trinucleotide complementary to the tRNA anticodon and the initiation factors. The ribosome binding reaction is complete in about 10 min at 25°C and reaches a plateau at a trinucleotide concentration of 40 M except for fGln-tRNA (data not shown). The binding of [ 3 H]fGln-tRNA is, however, only about 45% of [ 3 H]fVal-tRNA which had essentially the same specific activity, even though a higher concentration (80 M) of the trinucleotide was used. Fig. 3 shows the effect of addition of IF2 from B. stearothermophilus on ribosome binding of the various fAA-tRNAs. Addition of IF2 has essentially no effect on the binding of fMet-tRNA or fVal-tRNA to the ribosome. In contrast, at all of the concentrations of the trinucleotides used, the rate and extent of binding of the fGln-tRNA to the ribosome is stimulated by IF2. At low concentrations (10 M) of the trinucleotide UAG and in the absence of any additional amounts of IF2, the binding of fGln-tRNA to the ribosome is lower by at least a factor of 6 compared to that of fMet-tRNA. There is a gradual increase in binding up to the highest concentration (30 M) of the UAG used. The amount of IF2 used (1 M) was saturating for stimulating the binding of fGln-tRNA to the ribosome at an UAG concentration of 40 M.

Effect of Additional Amounts of IF2 on the Binding of fMet-, fVal-, and fGln-tRNAs to the Ribosome-
Gel Retardation Analysis of the Binding of IF2 to fMet-, fVal-, and fGln-tRNAs-The stimulatory effect of IF2 on the ribosome binding, specifically, of fGln-tRNA suggested that IF2 might have a reduced affinity for fGln-tRNA (18) compared to fMet-tRNA or fVal-tRNA. Because the binding of IF2 to fMet-tRNA is known to be relatively weak, we used gel retardation analyses to determine the relative affinity of IF2 for the various fAA-tRNAs. Fig. 4 shows the results of such an analysis carried out in the presence of a fixed concentration of IF2 (2.4 M) and increasing concentrations (0.3-1.5 M) of the fAA-tRNAs. Under the conditions of gel electrophoresis, all three fAA-tRNAs showed a major band and a minor band corresponding to the free fAA-tRNA. The fAA-tRNAs in both bands bind to IF2 and form complexes. The regions of radioactivity corresponding to the bound and free fAA-tRNAs from this gel and another using higher concentrations (2-10 M) of fAA-tRNAs (data not shown) were quantified using a PhosphorImager. A double reciprocal plot of 1/r, where r is the fraction of IF2 bound with fAA-tRNA against [1/fAA-tRNA free] is shown on Fig. 5. The apparent K d values of the IF2⅐fAA-tRNA binary complexes were estimated from the slope of each curve using the equation, 1/r ϭ K d (1/fAA-tRNA free) ϩ 1 (30). The apparent K d values for the IF2⅐fMet-tRNA, IF2⅐fVal-tRNA, and IF2⅐fGln-tRNA complexes were found to be 1.8, 3.5, and 10.4 M respectively. The effect of the fAA moiety attached to the initiator tRNA on its affinity for IF2 is reminiscent of the effect of the amino acid moiety on interaction of aminoacyl-tRNAs with other proteins such as the E. coli methionyl-tRNA transformylase (31)(32)(33), the E. coli elongation factor EF-Tu (34 -36), the E. coli selenocysteinyl-tRNA specific SelB protein (37), and the eukaryotic translation initiation factor eIF2 (30). The apparent K d of 1.8   2. Acid-urea gel electrophoretic analysis of 5-32 P-labeled wild type, G34C36 and U35A36 mutant tRNAs following their aminoacylation and formylation. Lanes 1, 3, and 5 contained 5Ј-32 P-labeled tRNAs that were untreated, and lanes 2, 4, and 6 contained tRNAs that were treated with the corresponding aminoacyl-tRNA synthetase and the formylating enzyme.
M for the B. stearothermophilus IF2⅐fMet-tRNA complex is similar to the previously estimated K d of 1 M for the E. coli IF2⅐fMet-tRNA complex based on protection of fMet-tRNA against chemical deacylation by IF2 (38). DISCUSSION We have compared the codon-dependent ribosome binding of wild type and mutant E. coli initiator tRNAs which carry different amino acids and the effect of varying the levels of IF2 on the ribosome binding. Binding of fGln-tRNA requires a higher concentration of IF2 than that of fMet-tRNA or fVal-tRNA. Results of gel mobility shift analysis of IF2⅐fAA-tRNAs indicate that this is most likely due to lower affinity of IF2 for fGln-tRNA compared to fMet-tRNA or fVal-tRNA (Fig. 5). These in vitro results are in excellent correlation with the effect of overproduction of IF2 on activities of fGln-tRNA and fVal-tRNA in initiation in vivo (9,16). Overproduction of IF2 greatly stimulated the activity of fGln-tRNA whereas it had no effect on activity of fVal-tRNA. The strong binding to ribosomes of initiator fVal-tRNA carrying the wild type anticodon sequence has been described before (39).
Previous studies, based on overproduction of MetRS in vivo, have shown that the U35A36 mutant initiator tRNA carrying fMet is a better initiator than the one carrying fGln (40). Thus some component of the translational machinery which interacts with the initiator tRNA subsequent to its formylation prefers fMet over fGln. The results reported in this paper are consistent with our previous suggestion that this component is most likely IF2 (9).
The relationship between the effect of IF2 on the UAG dependent binding of fGln-tRNA to the ribosome and the previously reported "high" IF2 dependence for use of AUU as initiation codon in an E. coli cell-free translation system is not clear (41). In contrast to the UAG-dependent binding of fGln-tRNA to the ribosome, translation initiation with AUU utilizes fMet-tRNA. It is possible that IF2 stabilizes the otherwise "weak" pairing between AUU initiation codon and the CAU anticodon in the fMet-tRNA, which involves a U:C base pair in the wobble position.
The only differences between the mutant initiator tRNAs used in these studies is the amino acid attached to the tRNA and the anticodon sequences. Since IF2 interacts with the acceptor stem of the tRNA and requires a formylated or Nacylated amino acid for binding (8,28,42,43), the reduced affinity of IF2 for fGln-tRNA is most likely due to the amino acid attached to the tRNA. Also, IF2 must make close contact with the amino acid attached to the tRNA since it protects the ester linkage in fMet-tRNA from deacylation (38). Recent results from cross-linking of IF2 to fMet-tRNA indicate, however, a possible interaction between IF2 and the anticodon stem and loop of the tRNA (44). Therefore, the possibility that the anticodon sequence change from CAU to CUA in the fGln-tRNA also contributes toward a lower affinity of IF2 for this tRNA must be left open.
The mechanism by which IF2 functions in initiation in E. coli is not established. The gene for IF2 is essential in E. coli and IF2 is required for the binding of fMet-tRNA to the P site on the ribosome (5,45). Earlier studies in vitro suggested a role for IF2 as a carrier of fMet-tRNA to the P site on the ribosome in much the same way as EF-Tu as a carrier of aminoacyl-tRNAs to the A site. However, recent studies, also in vitro, have led to the notion that all of the 30 S ribosomes in vivo are saturated with IF2 and that IF2 selects the fMet-tRNA from the pool of cellular tRNAs only after binding to the 30 S ribosomal subunit (5,6,10,11). Our results on the effect of overproduction of IF2 on the increased activity of fGln-tRNA in initiation in vivo are not easily explained if ribosomes are already saturated with IF2 and if IF2 binds first to 30 S ribosomes and then to the fAA-tRNA. They are, intuitively, more easy to understand if IF2 acts as a carrier of fGln-tRNA to the ribosome. According to this hypothesis (18), the fGln-tRNA is less active than fMet-tRNA because IF2 has a lower affinity for fGln-tRNA. In cells overproducing IF2, more of the IF2 would bind to the fGln-tRNA and carry it to the ribosome. This hypothesis would also answer the otherwise puzzling question of how, under normal circumstances, the fMet-tRNA gets to the ribosome during initiation. It is worth noting that every other tRNA including aminoacyl-tRNAs, the eukaryotic initiator tRNAs, and even the special selenocysteine inserting tRNA is carried to the ribosome by a protein carrier, the elongation factor EF-Tu or EF1, eIF2, and the SelB protein, respectively (5,37,46).
The suggestion that IF2 acts as a carrier of fMet-tRNA to the ribosome differs from the conclusions of Gualerzi and Pon (3), based on an extensive series of careful experiments, that IF2 functions as a component of the initiating 30 S ribosome and kinetically helps select the fMet-tRNA over other tRNAs. It is difficult to reconcile these differences except to note that one is based on results of experiments done in vitro with N-acetyl Phe-tRNA as an initiator, whereas the other is based on a combination of in vivo and in vitro experiments with anticodon mutants of initiator tRNA. Further work, including detailed kinetic analysis, is necessary to determine whether our suggestions, based on in vivo work using anticodon mutants of initi- ator tRNAs which are overproduced and using cells overproducing IF2 (9), are valid under normal conditions of protein synthesis initiation. It would also be desirable to extend this study to mutant initiator tRNAs carrying other formylaminoacids to determine whether in these cases also there is a correlation between the effect of overproduction of IF2 on activities of the tRNAs in vivo and their affinities for IF2 in vitro.
Finally, our in vitro work on effect of increasing IF2 levels on ribosome binding of fAA-tRNAs and measurement of K d values of the IF2⅐fMet-tRNA complex has been carried out using IF2 from B. stearothermophilus instead of E. coli. It is unlikely that our results are influenced to any significant extent by the use of the hetorologous IF2 protein. Proteins from B. stearothermophilus, including IF2, have been studied extensively (47) and are known to behave similarly to those from E. coli. In addition, the apparent K d of 1.8 M for the B. stearothermophilus IF2⅐E. coli fMet-tRNA complex that we have measured is similar to the previously estimated value of 1 M for the E. coli IF2⅐fMet-tRNA complex.