Stringent Nucleotide Recognition by the Ribosome at the Middle Codon Position

Accurate translation of the genetic code depends on mRNA:tRNA codon:anticodon base pairing. Here we exploit an emissive, isosteric adenosine surrogate that allows direct measurement of the kinetics of codon:anticodon base formation during protein synthesis. Our results suggest that codon:anticodon base pairing is subject to tighter constraints at the middle position than at the 5′- and 3′-positions, and further suggest a sequential mechanism of formation of the three base pairs in the codon:anticodon helix.

. (A) Isosteric guanosine and adenosine surrogates th G and th A; (B) mRNAs site-specifically labeled with th A (in red).

th A Fluorescence Intensity Changes Accompanying PRE and POST Complex Formation
Ribosomes programmed with mRNAs th A1, th A4, th A5, and th A6 displayed substantial functionality in forming 70SIC (Figure 2A), PRE complex ( Figure 2B), and POST complex ( Figure 2C). Stoichiometries of complex formation, measured at a single standard condition, were generally ≥50% relative to ribosomes programmed with control mRNA lacking th A, denoted fMKFR, although POST complex formation with th A5 (30%) was clearly lower.
For 70SICs programmed with th A4, th A5 or th A6, PRE complex formation on addition of cognate EF-Tu.GTP.Lys-tRNA Lys (Lys-TC) leads, in each case, to significant decreases in fluorescence intensity ( Figure 3A-C). These changes arise as a consequence of codon:anticodon interaction, since almost no changes are seen with th A1 ( Figure 3D) or on replacing Lys-TC with non-cognate Arg-TC ( Figure S2A-C). Translocation of PRE complexes to POST complexes on addition of EF-G.GTP proceeds with a large decrease in the fluorescence intensity of th A1 as tRNA fMet dissociates from the ribosome ( Figure 3D), but with moderate to negligible increases in the fluorescence intensities of th A4, th A5 and th A6 ( Figure  3A-C). These intensity changes, summarized in Figure 3E, provide spectroscopic signatures for monitoring the kinetics of PRE and POST complex formation during the polypeptide elongation cycle, as described below. Since th A fluorescence responds to alterations in several factors, including solvation, base-pairing, stacking and interactions with proximal chromophores, it is not currently possible to ascribe fluorescence intensity changes to specific molecular perturbations in the th A microenvironment.

th A Fluorescence Intensity Changes Accompanying PRE and POST Complex Formation
Ribosomes programmed with mRNAs th A1, th A4, th A5, and th A6 displayed substantial functionality in forming 70SIC (Figure 2A), PRE complex ( Figure 2B), and POST complex ( Figure 2C). Stoichiometries of complex formation, measured at a single standard condition, were generally ≥50% relative to ribosomes programmed with control mRNA lacking th A, denoted fMKFR, although POST complex formation with th A5 (30%) was clearly lower.
For 70SICs programmed with th A4, th A5 or th A6, PRE complex formation on addition of cognate EF-Tu.GTP.Lys-tRNA Lys (Lys-TC) leads, in each case, to significant decreases in fluorescence intensity ( Figure 3A-C). These changes arise as a consequence of codon:anticodon interaction, since almost no changes are seen with th A1 ( Figure 3D) or on replacing Lys-TC with non-cognate Arg-TC ( Figure S2A-C). Translocation of PRE complexes to POST complexes on addition of EF-G.GTP proceeds with a large decrease in the fluorescence intensity of th A1 as tRNA fMet dissociates from the ribosome ( Figure 3D), but with moderate to negligible increases in the fluorescence intensities of th A4, th A5 and th A6 ( Figure 3A-C). These intensity changes, summarized in Figure 3E, provide spectroscopic signatures for monitoring the kinetics of PRE and POST complex formation during the polypeptide elongation cycle, as described below. Since th A fluorescence responds to alterations in several factors, including solvation, base-pairing, stacking and interactions with proximal chromophores, it is not currently possible to ascribe fluorescence intensity changes to specific molecular perturbations in the th A microenvironment. Control experiments were performed, as indicated, with either no added mRNA, or with added unmodified mRNA. Values reported in (B) were normalized to the value obtained for unmodified fMKFR mRNA after subtraction of background obtained with no mRNA. Values reported in (C) were obtained following the previously published procedure for th G5 [28] and normalized to the value obtained for unmodified fMKFR mRNA. Traces are shown for purified 70SIC (black); PRE complex (red); POST complex (green); (E) Fluorescence intensity changes were obtained by subtracting the spectrum of a purified complex (70SIC or PRE) from the spectrum of a complex prepared in situ (PRE or POST, respectively) and then normalizing the intensity differences to the peak intensity value of purified complex (70SIC or PRE, respectively).

Kinetics of PRE Complex Formation
We and others previously have used changes in the proflavin (prf) fluorescence intensities of aa-tRNA aa (prf) (as part of a TC) and fMet-tRNA fMet (prf) to monitor, respectively, the initial rates of TC binding to the 70SIC [12] and of aa-tRNA accommodation into the A-site prior to dipeptide formation [31]. In this work we compare the rates of prf fluorescence change of Lys-tRNA Lys (prf) and fMet-tRNA fMet (prf) with those obtained with th A present in th A4-mRNA ( Figure 4A), th A5-mRNA ( Figure 4B), or th A6-mRNA ( Figure 4C) to determine the rates of different substeps within overall PRE complex formation. Reaction mixtures contained either a single fluorescent component, th A alone, or two fluorescent components, th A plus either Lys-tRNA Lys (prf) or fMet-tRNA fMet (prf). In the latter cases, changes in both th A and prf fluorescence were monitored simultaneously.   Traces are shown for purified 70SIC (black); PRE complex (red); POST complex (green); (E) Fluorescence intensity changes were obtained by subtracting the spectrum of a purified complex (70SIC or PRE) from the spectrum of a complex prepared in situ (PRE or POST, respectively) and then normalizing the intensity differences to the peak intensity value of purified complex (70SIC or PRE, respectively).

Kinetics of PRE Complex Formation
We and others previously have used changes in the proflavin (prf) fluorescence intensities of aa-tRNA aa (prf) (as part of a TC) and fMet-tRNA fMet (prf) to monitor, respectively, the initial rates of TC binding to the 70SIC [12] and of aa-tRNA accommodation into the A-site prior to dipeptide formation [31]. In this work we compare the rates of prf fluorescence change of Lys-tRNA Lys (prf) and fMet-tRNA fMet (prf) with those obtained with th A present in th A4-mRNA ( Figure 4A), th A5-mRNA ( Figure 4B), or th A6-mRNA ( Figure 4C) to determine the rates of different substeps within overall PRE complex formation. Reaction mixtures contained either a single fluorescent component, th A alone, or two fluorescent components, th A plus either Lys-tRNA Lys (prf) or fMet-tRNA fMet (prf). In the latter cases, changes in both th A and prf fluorescence were monitored simultaneously. Relative PRE formation -mRNA fMFKR th A1 th A4 th A5 th A6 fMFKR th A1 th A4 th A5 th A6 fMFKR th A1 th A4 th A5 th A6 Figure 3. Fluorescence changes of th A-containing mRNAs between ribosomal complex conversion (A-D); Traces are shown for purified 70SIC (black); PRE complex (red); POST complex (green); (E) Fluorescence intensity changes were obtained by subtracting the spectrum of a purified complex (70SIC or PRE) from the spectrum of a complex prepared in situ (PRE or POST, respectively) and then normalizing the intensity differences to the peak intensity value of purified complex (70SIC or PRE, respectively).

Kinetics of PRE Complex Formation
We and others previously have used changes in the proflavin (prf) fluorescence intensities of aa-tRNA aa (prf) (as part of a TC) and fMet-tRNA fMet (prf) to monitor, respectively, the initial rates of TC binding to the 70SIC [12] and of aa-tRNA accommodation into the A-site prior to dipeptide formation [31]. In this work we compare the rates of prf fluorescence change of Lys-tRNA Lys (prf) and fMet-tRNA fMet (prf) with those obtained with th A present in th A4-mRNA ( Figure 4A), th A5-mRNA ( Figure 4B), or th A6-mRNA ( Figure 4C) to determine the rates of different substeps within overall PRE complex formation. Reaction mixtures contained either a single fluorescent component, th A alone, or two fluorescent components, th A plus either Lys-tRNA Lys (prf) or fMet-tRNA fMet (prf). In the latter cases, changes in both th A and prf fluorescence were monitored simultaneously.
Apparent rate constants, collected in Table 1, were evaluated using Equation (1) or Equation (2) (Section 4). In all three cases, the Lys-TC(prf) fluorescence change was the most rapid. The apparent first order constants for 70SICs programmed with th A4-mRNA ( Figure 4A), th A6-mRNA ( Figure 4C), and fMKFR-mRNA, 44-62 s −1 (Table 1), correspond closely to previously published values [12], whereas the value with th A5-mRNA is~3-fold slower (18 s −1 ) ( Table 1). For both th A4-mRNA and th A6-mRNA, rates of fMet-tRNA fMet (prf) fluorescence change, which also agree with previously published values [31], are somewhat slower than those for th A fluorescence change. These results lead to the conclusion that, for these labeled mRNAs, the change in th A fluorescence that depends on codon:anticodon interaction follows TC binding and precedes Lys-tRNA Lys accommodation, a kinetic mechanism that is consistent with the accepted view of the sequence of events involved in PRE complex formation [5,9]. Apparent rate constants, collected in Table 1, were evaluated using Equation (1) or Equation (2) (Section 4). In all three cases, the Lys-TC(prf) fluorescence change was the most rapid. The apparent first order constants for 70SICs programmed with th A4-mRNA ( Figure 4A), th A6-mRNA ( Figure 4C), and fMKFR-mRNA, 44-62 s −1 (Table 1), correspond closely to previously published values [12], whereas the value with th A5-mRNA is ~3-fold slower (18 s −1 ) ( Table 1). For both th A4-mRNA and th A6-mRNA, rates of fMet-tRNA fMet (prf) fluorescence change, which also agree with previously published values [31], are somewhat slower than those for th A fluorescence change. These results lead to the conclusion that, for these labeled mRNAs, the change in th A fluorescence that depends on codon:anticodon interaction follows TC binding and precedes Lys-tRNA Lys accommodation, a kinetic mechanism that is consistent with the accepted view of the sequence of events involved in PRE complex formation [5,9]. Rates of th A fluorescence change were virtually identical using either unlabeled fMet-tRNA fMet and unlabeled aa-tRNA (green), unlabeled fMet-tRNA fMet and Lys-tRNA Lys (prf) (red), or fMet-tRNA fMet (prf) and unlabeled Lys-tRNA Lys (purple). Also shown are rates of fluorescence changes of Lys-tRNA Lys (prf) (dark blue) and of fMet-tRNA fMet (prf) (cyan). The inset in (C) shows an expanded time scale, making clear biphasic nature of th A5 fluorescence change. Solid black lines are fitted to Scheme 1, using rate constants provided in Table 2. This order of reaction is captured by globally fitting the results for th A4-mRNA and th A6-mRNA ( Figure 4A,B) to Scheme 1 for PRE complex formation (more detailed schemes are presented in references [9] and [32]), in which steps 1, 2 and 3 correspond to initial binding of TC to 70SIC (Complex C1), formation of the codon:anticodon helix (Complex C2) and Lys-tRNA Lys accommodation into the A-site followed by rapid peptide bond formation [5,12,32,33] to complete PRE complex formation, respectively. The fitted microscopic rate constants and relative fluorescence intensities for this Scheme are presented in Table 2. The situation is more complicated for th A5-mRNA ( Figure 4C), for which the th A fluorescence change has clear biphasic character, with a minor, fluorescence decrease that occurs more rapidly than the monophasic change seen with th A4-mRNA or th A6-mRNA and a major fluorescence change that occurs much less rapidly, at a rate that is quite similar to that observed for fMet-tRNA fMet (prf) fluorescence change (Table 1). These differences suggest that, although we are able to globally fit the results obtained with th A5-mRNA to Scheme 1, giving the values for k1, k2, and k3 displayed in Table 2, the identity of C2 differs from that assigned in the case of th A4-mRNA or th A6-mRNA (see Section 3).   Table 2.
This order of reaction is captured by globally fitting the results for th A4-mRNA and th A6-mRNA ( Figure 4A,B) to Scheme 1 for PRE complex formation (more detailed schemes are presented in references [9] and [32]), in which steps 1, 2 and 3 correspond to initial binding of TC to 70SIC (Complex C 1 ), formation of the codon:anticodon helix (Complex C 2 ) and Lys-tRNA Lys accommodation into the A-site followed by rapid peptide bond formation [5,12,32,33] to complete PRE complex formation, respectively. The fitted microscopic rate constants and relative fluorescence intensities for this Scheme are presented in Table 2. Apparent rate constants, collected in Table 1, were evaluated using Equation (1) or Equation (2) (Section 4). In all three cases, the Lys-TC(prf) fluorescence change was the most rapid. The apparent first order constants for 70SICs programmed with th A4-mRNA ( Figure 4A), th A6-mRNA ( Figure 4C), and fMKFR-mRNA, 44-62 s −1 (Table 1), correspond closely to previously published values [12], whereas the value with th A5-mRNA is ~3-fold slower (18 s −1 ) ( Table 1). For both th A4-mRNA and th A6-mRNA, rates of fMet-tRNA fMet (prf) fluorescence change, which also agree with previously published values [31], are somewhat slower than those for th A fluorescence change. These results lead to the conclusion that, for these labeled mRNAs, the change in th A fluorescence that depends on codon:anticodon interaction follows TC binding and precedes Lys-tRNA Lys accommodation, a kinetic mechanism that is consistent with the accepted view of the sequence of events involved in PRE complex formation [5,9].  Table 2.
This order of reaction is captured by globally fitting the results for th A4-mRNA and th A6-mRNA ( Figure 4A,B) to Scheme 1 for PRE complex formation (more detailed schemes are presented in references [9] and [32]), in which steps 1, 2 and 3 correspond to initial binding of TC to 70SIC (Complex C1), formation of the codon:anticodon helix (Complex C2) and Lys-tRNA Lys accommodation into the A-site followed by rapid peptide bond formation [5,12,32,33] to complete PRE complex formation, respectively. The fitted microscopic rate constants and relative fluorescence intensities for this Scheme are presented in Table 2. The situation is more complicated for th A5-mRNA ( Figure 4C), for which the th A fluorescence change has clear biphasic character, with a minor, fluorescence decrease that occurs more rapidly than the monophasic change seen with th A4-mRNA or th A6-mRNA and a major fluorescence change that occurs much less rapidly, at a rate that is quite similar to that observed for fMet-tRNA fMet (prf) fluorescence change (Table 1). These differences suggest that, although we are able to globally fit the results obtained with th A5-mRNA to Scheme 1, giving the values for k1, k2, and k3 displayed in Table 2, the identity of C2 differs from that assigned in the case of th A4-mRNA or th A6-mRNA (see Section 3). The situation is more complicated for th A5-mRNA ( Figure 4C), for which the th A fluorescence change has clear biphasic character, with a minor, fluorescence decrease that occurs more rapidly than the monophasic change seen with th A4-mRNA or th A6-mRNA and a major fluorescence change that occurs much less rapidly, at a rate that is quite similar to that observed for fMet-tRNA fMet (prf) fluorescence change (Table 1). These differences suggest that, although we are able to globally fit the results obtained with th A5-mRNA to Scheme 1, giving the values for k 1 , k 2 , and k 3 displayed in Table 2, the identity of C 2 differs from that assigned in the case of th A4-mRNA or th A6-mRNA (see Section 3).  1 Ranges are average ± average deviation for 2 independent determinations. All fits were to a single exponential except for k thA of th A5, which was fit to a double exponential. In the latter case, the major change in fluorescence intensity was 2.2 fold greater than that of the minor change.

Kinetics of POST Complex Formation
Apparent rate constants for EF-G.GTP-dependent translocation of PRE complexes programmed with th A-containing mRNAs to form POST complexes were determined from the time-dependent changes in fluorescence intensity of either th A (k A ) or fMet-aa-tRNA aa (prf) (k prf ) [17,18] (Figure 5, Table 3; the lack of a spectral change precluded k A determination for PRE complexes programmed with th A6).   1 Ranges are average ± average deviation for 2 independent determinations. All fits were to a single exponential except for kthA of th A5, which was fit to a double exponential. In the latter case, the major change in fluorescence intensity was 2.2 fold greater than that of the minor change. 2

Kinetics of POST Complex Formation
Apparent rate constants for EF-G.GTP-dependent translocation of PRE complexes programmed with th A-containing mRNAs to form POST complexes were determined from the time-dependent changes in fluorescence intensity of either th A (kA) or fMet-aa-tRNA aa (prf) (kprf) [17,18] (Figure 5, Table 3; the lack of a spectral change precluded kA determination for PRE complexes programmed with th A6). The relative values of kA vs. kprf were different for PRE complexes programmed with th A4, th A1, or th A5. For th A4-programmed ribosomes kA is indistinguishable from kprf, indicating that the movements of tRNA and mRNA are strongly coupled. In contrast, kprf for th A1-programmed PRE complexes is 3-fold larger than kA. This most likely reflects a more dramatic change in th A1 fluorescence on tRNA fMet dissociation from the ribosomal E-site, a step following translocation. Movements of tRNA and The relative values of k A vs. k prf were different for PRE complexes programmed with th A4, th A1, or th A5. For th A4-programmed ribosomes k A is indistinguishable from k prf , indicating that the movements of tRNA and mRNA are strongly coupled. In contrast, k prf for th A1-programmed PRE complexes is 3-fold larger than k A . This most likely reflects a more dramatic change in th A1 fluorescence on tRNA fMet dissociation from the ribosomal E-site, a step following translocation. Movements of tRNA and mRNA appear quite decoupled for th A5-programmed PRE complexes, with the apparent rate constants for the major and minor changes in prf fluorescence intensity, k prf1 and k prf2 , being 6-fold larger and 2-fold smaller, respectively, than k A . The rate constant k prf1 for th A5 is also considerably larger (13 s −1 ) than the values of k prf for th A1, th A4, and th A6 (2.6-4.4 s −1 ), an acceleration that might arise from weaker codon:anticodon interaction (see Section 3). The similarity in the values of k A for th A1 and of k prf2 for th A5 (Table 2) suggests that the latter may also correspond to tRNA fMet dissociation.  1 Ranges are ±average deviation from 2 independent determinations. 2 not determined. 3 Fluorescence change is biphasic (top, k prf1 , bottom, k prf2 ). The major change in fluorescence intensity was 1.7 fold greater than that of the minor change.

Discussion
The results presented here show that th A, like th G [24], can be incorporated into functional mRNAs, giving rise to fluorescence intensity changes on conversion of 70SICs to PRE complexes and PRE complexes to POST complexes. These changes make possible the monitoring of microenvironment changes of specific codon nucleotides during a cycle of polypeptide elongation, allowing measurement of the kinetics of both PRE complex formation on cognate TC binding to the ribosome and of mRNA translocation catalyzed by EF-G.GTP. Our most interesting results were obtained for PRE complex formation on the first AAA elongation codon, and it is on these results that we will focus in this section. The rate constants presented in Table 2 demonstrate that whereas th A substitution for adenosine at the first and third positions ( th A4 and th A6) show little or no interference with normal mRNA function both in TC binding (k 1 , 120-170 µM −1 ·s −1 ) and in the overall rate constant for aa-tRNA accommodation following TC binding [k 2 k 3 /(k 2 + k 3 ), 6 s −1 ], major perturbations in mRNA function are observed for th A substitution in the middle position ( th A5), including a 3-4-fold decrease in the rate of TC binding (40 µM −1 ·s −1 ) and a much larger 15-fold decrease in the overall rate of aa-tRNA accommodation (0.4 s −1 ).
We attribute these th A5-induced perturbations to the greater sensitivity on the middle codon position to structural changes arising from substitution of the bulkier, less polar, thiophene C-S-C grouping for the N 7 -C 8 -N 9 purine grouping. Although we posit that TC binding (step 1, Scheme 1) does not involve codon:anticodon base pairing, the more extensive backbone-rRNA H-bonding network at the middle codon position [1,34] could make it more sensitive to structural change than at the 5 and 3 codon positions. This interpretation is consistent with results measuring the effects of disruption of the H-bonding network on TC binding, which are more pronounced at the 5-position than at the 4-or 6-positions [35]. The larger effect on the overall rate constant for aa-tRNA accommodation is also reflected in the large differences in the time courses of th A fluorescence change that directly monitors codon:anticodon interaction, which is biphasic for th A5 with rate constants of 40 s −1 (minor phase) and 0.4 s −1 (major phase) but monophasic for th A4 and th A6, with rate constants of~10-11 s −1 . These differences raise two major questions. First, why does the major fluorescence change with th A5 proceed so much more slowly than the monophasic changes observed with th A4 and th A6? Second, why is the th A5 fluorescence change biphasic?
In addressing the first question we note that codon:anticodon base pairing at the middle position is subject tighter structural constraints within the ribosomal decoding center than at the 5 and 3 codon positions [34,36,37], in accord with the observation that misreading by the ribosome occurs least frequently at the middle codon nucleotide [38,39]. Specifically, A5:U is stacked between two adjacent base pairs, which form a tight cavern. In contrast, neither 5 -A4:U nor 3 -A6:U, which each have only one stacking interaction, are faced with such a cavern. The "solvent volume" of a base-pair, defined as the volume within a cutoff radius (5 Å) from the base-pair that can be occupied by a solvent molecule [40], provides a quantitative measure of the steric constraint imposed by this cavern and is considerably reduced for A5:U as compared with either A4:U or A6:U (Table 4). Higher constraint at the middle base-pair position provides at least a partial rationale, based on steric crowding, for the dramatically reduced rate of base-pairing. Other factors may also be in play. Thus, PRE complex formation using mRNAs containing the guanosine surrogate th G ( Figure 1A), showed a much less marked reduction in the rate of th G fluorescence change when th G is placed at the middle base pair as compared with the 5 and 3 base-pairs [28], even though solvent volumes calculated for G:C base pairs also show a much reduced value for the middle base pair (Table 4). A clear rationale for the apparent context dependence of the rate reduction phenomenon is currently lacking, but might be related to the much more stable stacking and hydrogen bonding interactions generally seen for G-C as opposed to A-U base pairs [41,42]. With regard to the biphasic nature of th A5 fluorescence change, we speculate that the answer lies in the detailed kinetic mechanism by which the three base pairs in the codon:anticodon helix are formed, whether simultaneously or sequentially. Although such helix formation in solution proceeds with a single apparent rate constant, it has been assumed that multiple base pairings occur in a series of rapid, unresolved steps [43], in accord with simulation studies of RNA helix formation [44,45]. The same scenario would explain why virtually all of the fluorescence change on conversion of C 1 to PRE complex with either th A4-mRNA or th A6-mRNA occurs in Step 2 (Table 2). However, the putative slow formation of the middle base pair with th A5-mRNA affords partial kinetic resolution of the base pairing steps. Accordingly, we interpret the rapid, minor phase for th A5 as corresponding to the formation a first base pair to either A4 or A6, and the very slow, major phase to formation of the second base pair to th A5, which is followed by much the more rapid formation of the third base pair. Interestingly, the 3.7 ± 1.5-fold higher k 2 value for th A5-mRNA vs. th A4-or th A6-mRNA (Table 2), which we attribute to formation of the first base pair, is the approximate value one would expect if individual base pairs to th A4-or th A6-mRNA are formed sequentially with each having a first-order rate constant equal to the k 2 value found for th A5-mRNA. Very slow formation of the base pair with th A5 would make that step rate-determining for all subsequent steps leading to PRE complex formation, accounting for the similarity in the apparent rates of the major decreases in the fluorescence intensities of both th A and tRNA fMet (prf) ( Table 1).
In summary, our results indicate that emissive isosteric purine nucleoside mimetics can be used not only to measure rates of specific steps in ribosome-catalyzed polypeptide elongation, but also to probe position-dependent constraints that modulate the rate of codon:anticodon helix formation, particularly at A-U base pairs. In this connection, it would be interesting to extend these studies to the recently synthesized isothiazolo-based nucleosides, in which the CH group of th A and th G is replaced by N, in a position corresponding to the purines' N7 [46]. Extension of this approach to examination of the helix formation within other RNA structures, such as ribozymes, IRESs and riboswitches, should be relatively straightforward.

Synthesis of Modified mRNAs
mRNAs were prepared using standard solid-phase phosphoramidite synthesis on an Expedite 8909 synthesizer on a 500 Å CPG (1 µmol scale) column. Commercially available reagents and phosphoramidites (Glen Research, Sterling, VA, USA), and a th A-phosphoramidite, prepared as described [25], were employed. Cleavage from the solid support and deprotection were accomplished with 50:50 mixture of MeNH 2 in water (40 wt %) and in ethanol (33 wt %) at 35 • C for 6 h. The 2 -TOM group was removed by TEA·3HF at 65 • C for 3 h and the residue was desalted by precipitation (Glen Report 19-20, Glen Research). All oligonucleotides were purified by preparative polyacrylamide gel electrophoresis (PAGE) using the crush and soak method; the desired band was cut out, pulverized, extracted with 50 mM TEAA (pH 7.0) for a minimum of 12 h (while shaking) and decanted. The buffer containing the purified oligonucleotide was lyophilized and the residue was taken up in 0.2 M TEAB (pH 7.0) buffer and desalted on a Sep-pak C-18 (Waters, Milford, MA, USA). The oligonucleotides were eluted with 40% acetonitrile in water. All oligonucleotides were >98% pure as determined by analytical high resolution PAGE ( Figure S3)

Complex Preparation and Quantification
70SIC was prepared, purified and quantified following a previously published procedure [28]. Modified mRNA was used as half of 70S concentration for experiments presented in Figure 2 or 2-fold excess over 70S for all other experiments. All experiments reported in this paper used purified 70SIC except for those shown in Figure 2B, for which the ultracentrifugation step described in [28] was omitted. TC was formed as described [28]. PRE complex was formed, purified and quantified by the amount of co-sedimenting [3H]-Lys [28]. All experiments reported in this paper used purified PRE complexes. POST complexes were prepared and quantified as described [28].

Fluorescence Experiments
Fluorescence experiments were carried out in Buffer A. Steady-state fluorescence was measured on a Fluorolog-3 spectrofluorometer (Horiba Jobin Yvon, Edison, NJ, USA). Emission and excitation spectra for th A-containing mRNAs were determined using excitation at 360 nm and monitoring emission at 410 nm, respectively. PRE complex traces shown in Figure 3 are the average of two independently determined traces: (i) PRE complex prepared in situ by addition of TC to 70SIC; (ii) purified PRE complex used in the preparation of POST complex. To account for small concentration differences arising from the two preparation methods, the traces for purified PRE complexes were normalized to those for in situ generated PRE complexes, using an 8 nm window covering the fluorescence maximum to generate a normalization factor. The resulting two very similar traces were then averaged. POST complexes were prepared in situ from purified PRE complexes by incubation with EF-G.GTP (2.5 µM for 20 s), and traces were normalized using the same normalization factors as above. For all traces background fluorescence of the corresponding complex made with unlabeled mRNA was subtracted from the observed fluorescence of each labeled ribosome complex prior to normalization to the peak intensity value of 70SIC for each mRNA. For Figure 3E, multiple determinations of single experiments indicate that fluorescence intensity differences of less than 10% are unreliable because of small errors associated with both pipetting and background fluorescence subtraction.
Kinetic experiments were carried out at 25 • C on a KinTek SF-300X stopped-flow spectrofluorometer (KinTek, Snow Shoe, PA, USA). The excitation wavelengths and band-pass filters used were: 360 nm and 410 ± 10-nm for th A-containing mRNAs; 462 nm and 515 ± 15-nm for proflavin. In each independent kinetic experiment, traces of fluorescence intensity changes were obtained as an average of at least 4 shots. In stopped-flow experiments containing both th A and a prf-labeled tRNA, th A and prf signals were collected on successive pulses from the same syringes. Each experiment was performed at least twice.

Solvent Volume Calculations
Structural models of codon-anticodon-ribosome interactions were based on structures of Yusupova and co-workers (protein databank accession code 3TVF [34]). Bases in the codon and anticodon were changed via molecular replacement using the swapna routine of chimera, which replaces base moieties. Structures were subsequently checked for clashes [47]. For each individual codon-anticodon base pair considered, an environment consisting of residues of the tRNA, mRNA and ribosome within 5.5 Å was used. To extract the solvent volume of these environments, a two-probe volume voxelator method was used, with a minor probe radius of 1 Å, major probe radius of 10 Å, and high grid resolution [40]. Chimera was used to obtain the subset of these solvent volumes within distance r bp of the codon-anticodon base pair of interest, with r bp = 5 Å. The volumes (in Å 3 ) and areas (in Å 2 ) of these solvent volumes were calculated with Chimera.

Apparent Rate Constants
Apparent rate constants (k app s) were obtained by either single (Equation (1)) or double (Equation (2)) exponential fitting of each independent experiment using Origin (OriginLab), respectively.

Conflicts of Interest:
The authors declare no competing financial interests.