Perturbation of the tRNA Tertiary Core Differentially Affects Specific Steps of the Elongation Cycle*

The tRNA tertiary core region is important for both tRNA stability and activity in the translation elongation cycle. Here we report the effects of mutating each of two highly conserved base pairs in the tertiary core of Phe-tRNAPhe, 18-55 and 19-56, on rate and equilibrium constants for specific steps of this cycle, beginning with formation of aminoacyl-tRNA·EF-Tu·GTP ternary complexs and culminating with translocation of A-site-bound peptidyl-tRNA into the P-site. We find that codon-dependent binding of aminoacyl-tRNA to the A/T-site and proofreading of near-cognate tRNA are sensitive to perturbation of either base pair; formation of the ternary complex and accommodation from the A/T to the A-site are sensitive to 18-55 perturbation only, and translocation of peptidyl-tRNA from the A- to P-site is insensitive to perturbation of either. These results underline the importance of the core region in promoting the efficiency and accuracy of translation, and they likely reflect different requirements for structural integrity of the core during specific steps of the elongation cycle.

The L-shaped conformation of tRNA is maintained by interactions between the D-and T-loops within the "tertiary core" region of the molecule. These interactions include two universally conserved base pairs, G18-⌿55 (⌿ is a modification of U) and G19-C56, and a stacking interaction of three purines at the 18, 57, and 58 positions (Fig. 1A). The L-shaped conformation allows ribosome-bound tRNA to contact simultaneously the 30 S subunit decoding center, via its anticodon loop, and the 50 S subunit peptidyltransferase center, via its CCA end. Ribosome-bound tRNA, however, assumes conformations distinct from the classical L-shaped structure. Indeed, tRNA has been likened to a flexible spring that reversibly changes conformation during the process leading to recognition of cognate aminoacyl-tRNA at the A-site and peptide bond formation (1). Departures from the classical tRNA structure have also been seen for tRNA bound at the P-and E-sites (2)(3)(4).
Recent molecular dynamics studies have suggested that maintaining a balance between tRNA flexibility and rigidity is essential for optimizing the rate and accuracy of cognate tRNA recognition (5,6), and it is reasonable to suppose that this balance might be important for other steps of the elongation cycle as well, such as translocation. Mutations in the tertiary core and D-arm that could affect this balance are known to affect tRNA function during elongation. Thus, maintenance of the geometries of one or both of the 18 -55 and 19 -56 bp is essential for optimal aminoacyl-tRNA⅐EF-Tu⅐GTP ternary complex (TC) 4 formation (7,8) and for rapid translocation of P-site tRNA (9). Related studies on bacterial cells have also demonstrated the importance of tertiary core interactions for the overall rate of translation (8,10). Similarly, elevated error rates have been observed for D-arm variants (11,12, including the Hirsh mutation (G24A) (13), which, although little affecting TC binding to a cognate codon, allows decoding of a near-cognate codon with relatively high efficiency (14). Despite these earlier examples, relatively little is known (15) about the role of the tertiary core in tRNA function at the A-site.
Here, using single and multiple turnover kinetics and binding experiments, we examine the effects of mutations in the G18-U55 and G19-C56 bp on each of four major steps comprising the elongation cycle (Fig. 1B) as follows: (i) TC formation; (ii) pre-accommodation (PRE-AC) complex formation which proceeds in three resolvable steps, an initial 2nd order binding step of the TC in which there is no codon-anticodon recognition, followed by two first-order steps in which cognate codon-anticodon recognition is rapidly followed by GTP hydrolysis (16); (iii) pretranslocation complex (PRE-TR) formation, in which peptidyl-tRNA and discharged tRNA are located in the A-and P-sites, respectively. PRE-TR formation from the PRE-AC complex proceeds via P i release from EF-Tu followed by a rapid conformational change in EF-Tu that results in dissociation of aminoacyl-tRNA from EF-Tu (17) and either accommodation of aminoacyl-tRNA into the 50 S A-site followed by rapid peptide bond formation (18) or dissociation of aminoacyl-tRNA from the ribosome prior to peptide bond formation, the latter constituting a form of proofreading; and (iv) EF-G⅐GTP-dependent translocation, in which discharged tRNA and peptidyl-tRNA move from the P-to E-site and A-to P-site, respectively, completing the elongation cycle. Translocation proceeds via a hybrid state intermediate, with the 3Ј ends of tRNAs moving toward the P-and E-sites on the 50 S subunit, whereas the anti-codons of tRNAs remain in A-and P-sites on the 30 S subunit (19).

EXPERIMENTAL PROCEDURES
Ribosomes and Proteins-Tight-coupled ribosomes were prepared from Escherichia coli MRE600 cells as described (9). Clones of E. coli His-tagged proteins EF-G, EF-Tu, IF1, IF2, and IF3 were obtained as described (9), and the proteins were purified on nickel-nitrilotriacetic acid (Qiagen) columns, except that EF-G went though an additional fast protein liquid chromatography Mono Q column with a gradient of 50 -350 mM KCl in buffer containing 50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 0.5 mM EDTA, and 6 mM 2-mercaptoethanol. EF-Ts was a gift from Dr. Yale E. Goldman (University of Pennsylvania).
mRNAs-mRNA022UUU was prepared from JM109 cells transformed with pTZ18 plasmid containing the 022 sequence under a T7 promoter, which was provided by Dr. C. Gualerzi (University of Camerino). The plasmid was extracted, restricted, and then transcribed using the EPICENTRE Ampliscribe T7 flash transcription kit. mRNA was purified by precipitation with 2.5 M LiCl on ice (30 min). mRNA022CUC was made by mutation of UUU to CUC, using the QuickChange protocol (Qiagen) followed by a preparation procedure essentially identical to that described for making mRNA022UUU.
tRNAs-Native E. coli tRNA fMet and tRNA Phe were acquired from Chemical Block (Moscow, Russia). Mutations were introduced into plasmid pTFMa (a derivative of pUC18) containing E. coli WT-tRNA Phe sequence using the QuickChange protocol (Qiagen) and confirmed by DNA sequencing. Plasmids harboring the WT-tRNA Phe sequence and its variants were purified, restricted with BstNI, and transcribed by T7 RNA polymerase.
His-tagged E. coli PheRS, purified on a nickel-nitrilotriacetic acid (Qiagen) column was used to charge both native tRNA Phe and transcribed tRNA Phe (WT or variants). Charging was carried out by incubating 20 M tRNA Phe , 80 M [ 3 H]Phe, and 30 g/ml PheRS in Buffer C (100 mM Tris-HCl (pH 7.8), 4 mM ATP, 20 mM MgCl 2 , 1 mM EDTA, and 7 mM 2-mercaptoethanol) at 37°C for 10 min. Purification of the charged tRNAs was carried out on fast protein liquid chromatography using a Mono Q column with a gradient of 0 -1 M NaCl in 50 mM sodium acetate (pH 5.0). tRNA fMet labeled with proflavin (prf), denoted tRNA fMet -(prf), was prepared by replacing the dihydrouridine residue at position 20 with proflavin, as described (9, 19 -21). 40 l of NaBH 4 (100 mg/ml in 0.01 M KOH) was added to tRNA fMet (1 mg in 400 l of 40 mM Tris-HCl (pH 7.5)), followed by incubation on ice for 60 min. The reaction was stopped by adding 20 l of 6 M acetic acid and ethanol precipitation. The resulting pellet was dissolved in 50 l of water, which was then mixed with 1 ml of 2 mM proflavin (in 0.1 M sodium formate (pH 3.0)) and incubated at 37°C for 45 min. To remove unreacted dye, the solution was adjusted to pH 7.5 with Tris, extracted with buffer-saturated phenol (three times), and precipitated with ethanol (twice).
Unlabeled and prf-labeled tRNA fMet were charged and formylated with partially purified E. coli tRNA synthetase containing MetRS and formyltransferase by incubating (22). Buffer TC contains all components of Buffer A, 100 M GTP, 10 g/ml pyruvate kinase, and 2 mM phosphoenolpyruvate. EF-Tu was incubated in buffer TC at 37°C for 5 min to exchange any GDP bound to EF-Tu for GTP. Such exchange occurs because of contamination of EF-Tu with small amounts of EF-Ts (23). Samples of EF-Tu⅐GTP solution (10 nM to 16 M, 25 l) were mixed with 25 l of 50 nM [ 3 H]Phe-tRNA Phe in buffer TC on ice, incubated at 37°C for 5 min, and placed on ice for 30 min. 5 l of 0.05 mg/ml bovine pancreatic RNase A (Sigma) was then added, and incubation was continued for 15 s on ice, followed by quenching with 200 l of 10% cold trichloroacetic acid. EF-Tu protected [ 3 H]Phe-tRNA Phe was collected on nitrocellulose filters after three washes of 1 ml of 5% cold trichloroacetic acid. The filters were dried, dissolved in 1 ml of ethyl acetate, and counted in a scintillation counter, providing a measure of TC concentration. K d values for various Phe-tRNA Phe were calculated by fitting results determined as a function of EF-Tu to Equation 1, using Igor-Pro (Wavemetrics).
Determination of Rate Constants and Functional Homogeneity-With one clear exception, the U55A variant, each Phe-tRNA Phe preparation (WT or variant) gave rates and stoichiometries of ribosome-dependent reactions consistent with complete or near-complete functional homogeneity. Thus, rates of fMet-Phe-tRNA dissociation, GTP hydrolysis, fMet-Phe formation, and translocation were well fit to single exponential equations, giving apparent first-order rate constants, with stoichiometries per TC formed similar to those obtained with native or WT Phe-tRNA Phe . Fitting the rates of accommodation of tRNA Phe transcripts required a double exponential equation. However, for all of the transcripts except U55A, 75-90% of the total fluorescence change occurs in the rapid first phase of reaction, so even with this assay the extent of any putative functional heterogeneity would be quite limited. In contrast, the U55A variant preparation could have considerable functional heterogeneity, because its ribosome-dependent reactions proceed with low stoichiometry, and the fluorescent change accompanying accommodation does not show a dominant first phase. Fits to exponential equations were carried out with Igor-Pro (Wavemetrics). k cat and K m values were obtained by fitting the curves of kЈ GTP as a function of 70SIC concentration to the Michaelis-Menten equation.

RESULTS
tRNA Variant Preparation-tRNA Phe was transcribed and contained the wild-type sequence (WT-tRNA Phe ) or variations in the 18, 19, 55, and/or 56 positions (Fig. 1A). WT transcript differs from the native tRNA Phe , isolated from E. coli cells, in having no post-transcriptional modification, and it is used for direct comparison with transcribed variants.
Initial Screening of Variants-Four assays (Table 1) were used to test each of the 12 tRNA Phe variants for functional Phe-tRNA Phe interaction with the A-site, following addition of the corresponding [ 3 H]Phe-tRNA Phe ⅐EF-Tu⅐GTP TC to a 70 S initiation complex (70SIC) containing fMet-tRNA fMet in the P-site and an empty A-site programmed with an UUU codon as follows: (i) the initial stoichiometry of Phe-tRNA Phe binding to the A-site via the TC complex; (ii and iii) the stoichiometry and rate (kЈ dp ) of dipeptide formation; and (iv) the rate of peptidyl-tRNA dissociation (kЈ d ) from the A-site following dipeptide formation. Here an increased value indicates destabilization of peptidyl-tRNA binding to the A-site.
Measured at an EF-Tu concentration of 1 M, mutation effects in reducing initial Phe-tRNA Phe binding are, for the most part, modest ( Table 1). The strongest effects on TC binding are found for U55A (4-fold decrease) and C56G and G18U/ U55A (ϳ2-fold decreases). The latter two decreases largely disappear when the measurement is repeated at a higher EF-Tu concentration (5 M), indicating that they are due more to a defect in TC formation (see below) than in TC binding to the ribosome. However, raising EF-Tu concentration does not significantly increase the stoichiometry of Phe-tRNA Phe binding for U55A, suggesting that this mutation directly perturbs Phe-tRNA Phe interaction with the A-site. Additional evidence for such a perturbation is provided by the larger negative effect that U55A mutation has on the stoichiometry of dipeptide formation than on the stoichiometry of initial Phe-tRNA Phe binding. This contrasts with the largely similar effects that almost all of the other mutants (G19C excepted) have on both stoichiometric assays.
More pronounced effects are seen on apparent rate constants (Table 1). For kЈ dp , the largest decreases are for U55A, G18U/ U55A, and G19C (8 -10-fold) and G18A, C56A, and C56G (3-4-fold), whereas for kЈ d , the largest increases (5-9-fold) are found for U55G, G18U/U55A, G19U, and C56G. The dissociation rate for U55A was not determined accurately because of its low binding stoichiometry but is likely to be relatively rapid, given the evidence for perturbed Phe-tRNA Phe binding to the A-site. In both assays, WT function is fully or partly restored in tRNA Tertiary Core Affects the Elongation Cycle 19 -56 double variants that preserve Watson-Crick base pairing.
Selection of Variants for Further Study-Five variants (U55A, G18U/U55A, G19U, C56A, and G19U/C56A) were selected for particularly close scrutiny, to determine which step or steps in the overall process of PRE-TR complex formation were responsible for the effects observed. The first four variants were selected because, with the exception of U55A, they have relatively large effects on the rate and smaller effects on the stoichiometry of dipeptide formation (U55A strongly affects both), allowing us to concentrate our analysis on the rate constants. The G19U/C56A variant was selected to more fully examine the rescuing effect of a second mutation that restores a 19 -56 Watson-Crick base pair.
Binding of Phe-tRNA Phe to the ribosomal A-site, leading to PRE-TR complex formation, includes three readily distinguishable steps (Fig. 1B) as follows: (a) formation of the TC, (b) for-mation of the PRE-AC complex, and (c) formation of the PRE-TR complex. Below we determine the effects of the selected mutations on each of these three steps.
TC Formation-The binding affinities to EF-Tu⅐GTP of native Phe-tRNA Phe , its WT transcript, and the five selected variants were determined by the ribonuclease protection assay (22), giving the dissociation constants displayed in Table 2. The largest effects are seen for U55A and G18U/U55A, which have 15-20-fold lower affinities than WT-Phe-tRNA Phe for EF-Tu⅐GTP. G19U and C56A are less affected, with 6-and 1.7fold lower affinities, respectively, although for G19U/C56A binding affinity is restored to a level close to that of the WT transcript. Structural studies of the ternary complex (26) show a contact between EF-Tu and the backbone of nucleotide 54. Thus, although EF-Tu does not directly interact with the 18 -55 bp, the comparatively large effects of 18 -55 mutations on TC formation may reflect a perturbation of the stacking interaction that the 18 -55 bp makes with the intraloop 54 -58 bp (Fig. 1A). The smaller effects of mutations at 19 -56 are consistent with their being even further from EF-Tu within the TC, and with earlier results showing that the G19C and C56G mutations have only minor effects on TC formation (15). The similarity in the values for native and WT-Phe-tRNA Phe binding to EF-Tu is also in accord with earlier results (27).
PRE-AC Complex Formation-Rates of PRE-AC complex formation for the selected variants were determined by quenched flow rapid kinetic studies of EF-Tu-dependent GTP hydrolysis ( Fig. 2A and Table 2). The stoichiometries of GTP hydrolysis per TC complex bound to the ribosome were similar for all of the transcribed tRNAs and for native tRNA, except for the markedly lower stoichiometry of the U55A variant. Measured at a single 70SIC concentration ( Fig. 2A), the apparent net rate constant for GTP hydrolysis (kЈ GTP , see Table 2), relative to that for WT transcript, is 2-fold larger for native tRNA, 10-fold smaller for the U55A and G18U/U55A variants, and 5-fold smaller for the G19U and C56A variants, the latter effects being fully rescued by the G19U/C56A double mutation. Measuring kЈ GTP as a function of increasing 70SIC concentration for WT transcript, G18U/U55A, and C56A allowed effects on the net rate constant k cat,GTP to be distinguished from effects on K m values (defined as the concentration of 70SIC giving half-maximal velocity) (Fig. 2B). For G18U/U55A k cat,GTP is decreased 12-fold with little change in K m , but for C56A both k cat (3-fold decrease) and K m values (3-fold increase) are affected.
PRE-TR Complex Formation-Rates of PRE-TR complex formation were determined by measurement of the rate of accommodation. In accord with earlier results (16,18), this rate is indistinguishable from that of dipeptide formation (data not shown). Accommodation was measured by the change in the fluorescence of P-site-bound fMet-tRNA fMet (prf) for a constant amount of TC added (Fig. 3). Previously we showed that this change occurs at the same rate as the fluorescent change of Phe-tRNA Phe (prf), a proflavin derivative of Phe-tRNA Phe (see supplemental Fig. 1C in Ref. 19) that signals accommodation into the A-site (16). For the TC formed with the WT transcript, the change in fMet-tRNA fMet (prf) fluorescence proceeds in two phases as follows: a large, fast decrease followed by a slow, small decrease, and it is the former that corresponds to accommodation. The two phases are well separated for all the variants but U55A, permitting reliable estimation of the stoichiometries of accommodated Phe-tRNA Phe from the magnitude of the first phase decrease in fluorescence, as well as of the apparent net rate constants for accommodation (kЈ acc ) ( Table 2). kЈ acc is reduced 15-fold in G18U/U55A and 2-4-fold in the 19 -56 single variants, which are once again rescued by the G19U/C56A double mutation. For U55A, the two phases of fluorescence decrease are not well separated, allowing only upper limit estimates of the stoichiometry of accommodation and kЈ acc , the latter being Ն18-fold less than for the WT transcript.
The differences in the apparent net rate constants for GTP hydrolysis (kЈ GTP ) and accommodation allow calculation of  The results emphasize the dramatic difference between the marked effects of the 18 -55 variants (U55A and G18U/U55A) on k* acc (Ն20-fold decreases) versus the total lack of effect of the 19 -56 variants.
Misreading-To investigate the effect of mutations in the core region of tRNA Phe on misreading, we formed 70 S initiation complexes containing a near-cognate CUC codon (coding for Leu) at the A-site in place of the original cognate UUU codon and determined, following individual addition of selected Phe-tRNA Phe s, the following: (i) the rates of single turnover GTPase; (ii) the rates of single and multiple turnover dipeptide formation; and (iii) the proofreading ratios P i formed/dipeptide formed. The results of these studies are presented in Fig. 4 and Table 3.
Replacing cognate UUU with CUC decreases kЈ GTPase for each of the Phe-tRNA Phe s tested (Fig. 4A). Although the reductions are larger for native and WT Phe-tRNA Phe (6 -8-fold), than for the variants G18U/U55A and C56A (2-4-fold), the order of reactivity with the CUC codon is unchanged from that for the UUU codon, native Ͼ WT Ͼ G18U/U55A, C56A.
More dramatic effects are seen on dipeptide formation (Fig.  4, B and C), which proceeds much more slowly and for which the order of reactivity for the CUC codon, G18U/U55A, C56A Ͼ WT Ͼ native is inverted with respect to that found for the UUU codon (Fig. 5), indicating that mutations in the tRNA tertiary core promote misreading.
The biphasic character of near-cognate dipeptide formation, evident in Fig. 4, B and C, is accounted for quantitatively by scheme 1 in Fig. 4D. Here r, which is the ratio of the rate constant for accommodation and peptide bond formation to that for aminoacyl-tRNA release, provides a measure of proofreading. The very slow rate of second phase reaction in Fig. 4B is determined by the rate of EF-Tu⅐GDP recycling to EF-Tu⅐GTP and the value of r Ϫ1 . For experiments performed in Buffer A, estimates of r were determined either by fitting the results in Fig. 4, A and B, to scheme 1 in Fig. 4D or by directly determining R, the measured ratio of P i formed/dipeptide formed, and setting r equal to R Ϫ 1. Both methods yielded values that agree within experimental error (Table 3). Very similar values of r were obtained in a "high fidelity" polyamine-containing buffer (25).
It is the comparatively poor proofreading by the G18U/U55 and C56A variants (r ϭ 3-4) that accounts for their higher rates of dipeptide synthesis as compared with WT-tRNA (r ϭ 7) and native (r ϭ 28) (Fig. 5). That proofreading is considerably more stringent for native versus WT-tRNA agrees with earlier results (27). Finally, using the overall rate of dipeptide formation with the CUC codon as a criterion, it is clear that proofreading is also decreased in the G18A and G19U variants, but it is maintained in the G19U/C56A double variant that restores the 19 -56 Watson-Crick base pair (Fig. 5).
Translocation-Apparent rate constants of translocation (kЈ tr ) using cognate mRNA were determined by quenched flow  With one exception (U55A) curves are fit to a double exponential equation. For WT Phe-tRNA Phe and most variants, the large decrease in the first reaction phase corresponds to accommodation, and it is followed by a small, slow decrease in the second phase of unclear origin, possibly because of a minor amount of misfolded tRNA. For U55A, which binds ternary complex slowly and with low stoichiometry, the first phase is not clearly distinguishable from the second, and results are fit with a single exponential. Relative amplitudes are observed changes for 0.2 M ternary complex added and are thus corrected for incomplete TC formation, using the K d values in Table 2. Monitoring of P-site tRNA fMet does not permit observation of initial binding of Phe-tRNA Phe to the A/T-site.

tRNA Tertiary Core Affects the Elongation Cycle
kinetic measurements of fMet-Phe-puromycin formation following rapid mixing of the PRE-TR complex with EF-G⅐GTP in the presence of puromycin (Fig. 6). Although the WT transcript fMet-Phe-tRNA Phe is translocated 3-fold more slowly than native, none of the 12 mutations studied further decreased the rate by more than 2-fold, and some showed slightly increased rates over WT transcript. This result is in marked contrast to the large decreases in translocation rate that we previously observed with the corresponding tRNA fMet variants bound in the P-site (9) (Fig. 6), especially at the 18 and 55 positions.

Effects of 18 -55 and 19 -56 Mutations on tRNA Function
during Elongation-Here we demonstrate that mutations of the strictly conserved 18 -55 and 19 -56 bp in the tRNA tertiary core differentially affect specific steps of the elongation cycle. The present results, taken together with related earlier results describing mutational effects on translocation from the P-site (9), may be summarized as follows (Fig. 1B). (i) The overall rate of PRE-AC complex formation, as measured by kЈ GTP , the stability of peptidyl-tRNA bound to the A-site, as measured by kЈ d , and the rate of dipeptide formation in response to a near-cog-  (Table 3). Virtually identical fits were obtained if r values were set equal to R Ϫ 1 (Table 3). D, scheme 1, which accounts quantitatively for the results presented in B. k 3 , determined by fitting, is equal to 0.005 s Ϫ1 and reflects the amount of EF-Ts in the reaction mixture as a contaminant of EF-Tu (21). Adding EF-Ts (1 M) to the reaction mixture (light blue trace in C) strongly accelerates dipeptide formation. k 2 is not evaluated but is Ͼ Ͼk 3 . No reaction is observed when EF-Tu is omitted (C), showing that dipeptide formation does not proceed from nonenzymatic binding of Phe-tRNA. a r, the ratio of the rate constant for accommodation and peptide bond formation to that for aa-tRNA release, is determined by fitting results in Fig. 4, A and B, to scheme 1 in Fig. 4D. b r ϭ R Ϫ 1, where R is equal to P i formed/dipeptide formed. R was determined on a reaction mixture incubated for 50 min that was identical to that used for misreading assays (Fig. 4) except that 20 M GTP was used instead of 200 M GTP to reduce background GTP hydrolysis contributing to P i formation. nate codon are each sensitive to perturbation of either base pair.
(ii) The formation of TC, accommodation of Phe-tRNA Phe into the A-site (as measured by k* acc ), and translocation of tRNA fMet from the P-site to the E-site are each sensitive to perturbation of the 18 -55 bp only. (iii) The translocation of Phe-tRNA Phe from the A-site to the P-site is insensitive to perturbation of either base pair. Overall, these results demonstrate the importance of the 18 -55 and 19 -56 bp for functional tRNA interaction with the ribosome. However, it must be acknowledged that the validity of our results for what might be observed in vivo, along with the results of others who also study tRNA variants produced by in vitro transcription (7,15,22), is subject to the caveat that the quantitative effects of mutation could be different in the context of native tRNA, which contains modified bases.
The changes in function accompanying mutation of any of the 4 bases in the 18 -55 or 19 -56 bp appear to result mainly from perturbations in local tRNA structure, rather than from disruptions of specific interactions of these bases with either EF-Tu or the ribosome. Thus, the deleterious effects of single mutations in the G19-C56 Watson-Crick base pair at the periphery of the tertiary core structure (28,29) (Fig. 1A) are largely or fully rescued by double mutations that restore Watson-Crick pairing. As well, all four Watson-Crick pairs at the 19 -56 position yield similar values for the stoichiometry of TC binding and for the stoichiometry and apparent rate constant of dipeptide formation (Table 1), and the double variant G19U/C56A has very similar values to WT-tRNA for TC formation, kЈ GTP and proofreading (Table 2 and Fig. 5). The situation is less straightforward for the G18-U55 base pair because these nucleotides form a bifurcated, three center interaction between the 2 carbonyl of U55 and N 1 and the 2-amino group of G18 (30), rather than a Watson-Crick base pair. Moreover, G18 makes a stacking interaction with bases G57 and A58, which form a so-called "purine trap" (8). As a result, the G18U/U55A double variant does not restore the disruption in the tRNA structure that results from U55A substitution, accounting for its general failure to reverse the effects of U55A mutation on TC formation, kЈ d , kЈ GTP , kЈ dp , and k* acc (Tables 1 and 2). On the other hand, we find high relative activities for the G18A/U55G variant, in accord with modeling studies showing that the geometry of the A-G base pair is similar to that of the G-U base pair within the tRNA tertiary core (8) and with several functional studies demonstrating the near interchangeability of the A-G and G-U (i.e. WT) base at the 18 -55 position pairs (7)(8)(9)31). An analogous explanation may account for the higher activity of the U55G variant than the U55A variant with respect to both Phe-tRNA Phe interaction with the A-site ( Table 1) and translocation of P-site-bound tRNA fMet (9) (Fig. 6), because in vivo selection of effective variant suppressor tRNAs yields some G-G pairs but no G-A pairs in positions equivalent to 18 -55 (8, 32).
Comparing the Present Results with Those of Nazarenko et al. (15)-The importance of the 18 -55 and 19 -56 bp for tRNA function in elongation was less evident in an earlier study (15). This latter work, which carried out two measurements (the stoichiometries of GTP hydrolysis and dipeptide formation) on a limited set of variants (U55C, G19C, and G19C/C56G), concluded that the conserved nucleotides 18, 19, 55, and 56 of tRNA Phe are not essential for functional Phe-tRNA Phe interaction with the ribosome. Our more extensive study of these base pairs, including determination of the effects of mutation on reaction rates as well as stoichiometries, clearly show that some disruptions of 18 -55 and 19 -56 bp have major impacts on Phe-tRNA Phe interaction with the ribosome. That U55A substitution, studied here, has a more deleterious effect on function than the U55C substitution studied by Nazarenko et al. (15) is consistent with modeling studies (31) showing that A-G, G-U, and G-C are the only three combinations of 18 -55 bases that form hydrogen bonds within the constraints of the elbow structure of tRNA Phe .
Despite the different conclusions reached in the two studies, there is only one apparent disagreement between the present results and those of Nazarenko et al. (15). This concerns the effect of G19C mutation on dipeptide stoichiometry, with  . Apparent rate constants for translocation with tRNA variants in the A-or P-sites. Comparison of the relative apparent rate constants for translocation of PRE-TR complexes containing either A-site (white bars) or P-site (black bars) variants. For A-site variants, translocation is measured by fMet-Phe-puromycin formation. Results for P-site variants are taken from Pan et al. (9), in which translocation was measured by the fluorescence change in fMet-Phe-tRNA Phe (prf). Both assays of translocation give essentially the same results.
tRNA Tertiary Core Affects the Elongation Cycle Nazarenko et al. (15) reporting no effect, although we find a 2.5-fold decrease (Table 1). This relatively minor difference is likely attributable to different experimental protocols; Nazarenko et al. (15) used poly(U)-programmed ribosomes, N-acetyl-Phe-tRNA Phe as peptide donor, and yeast Phe-tRNA Phe as peptide acceptor, whereas we employ 0.22mRNA-programmed ribosomes containing fMet-tRNA fMet as peptide donor and E. coli Phe-tRNA Phe as peptide acceptor. In addition, Nazarenko et al. (15) performed their assays at relatively high concentrations of ribosomes (up to 2 M) and Mg 2ϩ (10 mM), conditions that would be expected to minimize decreases observed in the stoichiometry of dipeptide formation arising from weaker TC affinity for the ribosome. By comparison, the experiments reported in Table 1 utilized 0.4 M ribosomes and 7 mM Mg 2ϩ .
Effects of 18 -55 and 19 -56 Mutations on Misreading-Fidelity in decoding mRNA is accomplished via two processes, an initial selection of cognate tRNA via an induced-fit, leading to EF-Tu-dependent GTP hydrolysis and PRE-AC complex formation, and a kinetic proofreading step that involves partitioning of ribosome-bound aminoacyl-tRNA that has been released from EF-Tu between accommodation followed by dipeptide formation and dissociation from the ribosome (14, 33) (scheme 1 in Fig. 4D). The rate of EF-Tu-dependent GTPase provides one measure of the induced-fit process, whereas the overall rate of dipeptide formation provides a measure of proofreading. Our results show that when a near-cognate codon is used, mutations disrupting tertiary base pairing, while little affecting relative rates of GTP hydrolysis (variant versus WT-Phe-tRNA Phe ) ( Fig. 2A and Fig. 4A), result in enhanced relative rates of dipeptide synthesis (Fig. 5). Thus, the rate of dipeptide formation for the G18U/U55A variant relative to WT Phe-tRNA Phe increases some 40-fold, from 0.06 to 2.5, when the cognate UUU codon is replaced by near-cognate CUC, whereas the corresponding increase for the C56A variant is 10-fold. These results provide the first indication of the importance of tRNA tertiary core structure in maintaining the fidelity of protein synthesis.
The induced-fit process is thought to proceed via large scale conformational changes of the 30 S subunit in response to cognate codon:anticodon recognition that trigger specific interactions between the TC and the 50 S subunit, resulting in GTP hydrolysis (1,33,34). The antibiotic kirromycin, while allowing formation of the PRE-AC complex and P i release, prevents both the conformational change in EF-Tu⅐GDP that normally follows P i release as well as EF-Tu⅐GDP dissociation from the ribosome (17). Cryo-EM studies of the kirromycin-stalled cognate complex reveals a distortion in the tRNA bound as part of the TC, relative to the crystal structure of tRNA alone, stemming from a kink near bases 26 and 44/45 (35). This has led to the hypothesis that the bending of aminoacyl-tRNA, which allows it to acquire the orientation required for codon recognition, is also a part of the communication between the decoding site and the GTPase activation center (1,35). The elevated error rates that have been observed for D-arm variants (11)(12)(13)(14), noted above, lend support to this hypothesis, if it is assumed that such variants adopt a kinked conformation even when binding to a near-cognate codon.
Portions of the GTPase-activation region of the 50 S subunit are in contact with the tertiary core of tRNA within the kirromycin-stalled complex (1,35). Such contacts have led to the suggestion (6) that mutations in the tertiary core region could lead to reduced GTPase activity by distorting the three-way interaction between the switch region in EF-Tu, the acceptor arm of tRNA, and the sarcin-ricin loop. Our results demonstrating the importance of maintaining both the 19 -56 and 18 -55 bp in the transition state for GTP hydrolysis (Table 2) support this suggestion. Indeed, as judged by effects on the value of kЈ GTPase , perturbing these base pairs by mutation has as strong an effect on the induced-fit process as does replacement of the cognate UUU codon by the near-cognate CUC codon (Fig. 3E).
A recent molecular dynamics simulation emphasizes the sensitivity of the rate of the accommodation process to the precise alignment of the acceptor arm relative to the 50 S subunit (5,6). This is because of the narrow accommodation corridor through which the acceptor arm must navigate as it moves from the A/T-site to the A-site (34). As a result, although the tertiary structure of tRNA is required to flex during accommodation, it also must be sufficiently stiff to avoid significant steric clashes. Our results, showing the accommodation rate constant k* acc for a cognate codon-tRNA interaction to be sensitive to mutation in the 18 -55 bp but not in the 19 -56 bp ( Table 2), suggest that the proper balance is achieved by maintaining the 18 -55 bp, which is embedded within the tRNA tertiary core (Fig. 1A), and allowing the 19 -56 bp at the periphery of the tRNA elbow to come apart, at least at the beginning of the accommodation process. This suggestion is consistent with results indicating that the 19 -56 Watson-Crick base pair is not maintained in the kirromycin-stalled complex (36). On the other hand, the balance between accommodation and tRNA dissociation for a near-cognate codon-tRNA interaction is clearly altered by perturbation of both the 18 -55 and 19 -56 bp, with mutation leading to increased misreading. This suggests that optimal sensing of a noncognate codon-anticodon interaction, leading to tRNA dissociation, requires strict maintenance of the tertiary core base pair interactions.
EF-G-catalyzed translocation of tRNAs proceeds with GTP hydrolysis and involves large scale movements of each tRNA, which, given the distorted tRNA structures that have been found at the P-and E-sites (2)(3)(4), are likely to require some flexing of tRNA. This would be consistent with our results showing that, as is the case for accommodation, maintenance of the 19 -56 bp is not required for the efficient translocation of either tRNA. On the other hand, translocation of A-site tRNA differs from translocation of P-site tRNA in also being insensitive to maintenance of the 18 -55 bp (Fig. 6). This could be interpreted as indicating that translocation of A-site tRNA to the P-site requires more flexibility within the tertiary core than translocation of P-site tRNA to the E-site. It would be of interest to determine whether molecular dynamics simulations of translocation (37,38) support this suggestion.