1. Biochemistry and Chemical Biology
Download icon

Kinetics of initiating polypeptide elongation in an IRES-dependent system

  1. Haibo Zhang
  2. Martin Y Ng
  3. Yuanwei Chen
  4. Barry S Cooperman  Is a corresponding author
  1. University of Pennsylvania, United States
Research Article
  • Cited 12
  • Views 935
  • Annotations
Cite this article as: eLife 2016;5:e13429 doi: 10.7554/eLife.13429

Abstract

The intergenic IRES of Cricket Paralysis Virus (CrPV-IRES) forms a tight complex with 80S ribosomes capable of initiating the cell-free synthesis of complete proteins in the absence of initiation factors. Such synthesis raises the question of what effect the necessary IRES dissociation from the tRNA binding sites, and ultimately from all of the ribosome, has on the rates of initial peptide elongation steps as nascent peptide is formed. Here we report the first results measuring rates of reaction for the initial cycles of IRES-dependent elongation. Our results demonstrate that 1) the first two cycles of elongation proceed much more slowly than subsequent cycles, 2) these reduced rates arise from slow pseudo-translocation and translocation steps, and 3) the retarding effect of ribosome-bound IRES on protein synthesis is largely overcome following translocation of tripeptidyl-tRNA. Our results also provide a straightforward approach to detailed mechanistic characterization of many aspects of eukaryotic polypeptide elongation.

https://doi.org/10.7554/eLife.13429.001

eLife digest

Inside cells, machines called ribosomes make proteins using instructions carried by molecules of messenger RNA (or mRNA). The ribosomes bind to the mRNA and then move along it to assemble the proteins in a process called translation. The first step of translation – when the ribosome binds to the mRNA – is known as initiation. In human and other eukaryotic cells, initiation mainly occurs through a mechanism that requires many proteins called initiation factors to recruit the ribosome to a cap structure formed at one end of the mRNA.

When viruses infect cells, they hijack the ribosomes of the host cell to produce large quantities of viral proteins. However, unlike their host cells, many viruses use a different pathway to initiate translation of their mRNAs. The mRNAs of these viruses have regions known an internal ribosome entry sites (IRESs) that host cell ribosomes can bind to instead.

After initiation, the ribosome progressively assembles the building blocks of proteins (amino acids) into a chain until the new protein is complete. Molecules called transfer RNAs bind to individual amino acids and bring them to the ribosome. Previous research has shown that, prior to initiation, IRESs on Cricket Paralysis Virus mRNAs bind to the ribosome and occupy sites where transfer RNAs would normally bind. However, it was not clear how this affects the elongation process. Zhang et al. now address this question using a cell-free system that allowed them to recreate and observe translation outside of the normal cell environment.

Zhang et al. found that the binding of an IRES to a ribosome slows down the early steps of elongation. A likely explanation for this is that the IRES elements have to be displaced from the ribosome before the incoming transfer RNAs can occupy the three tRNA sites. However, as elongation progresses, the effects of the IRES elements are overcome and the pace of elongation increases significantly. Zhang et al.’s findings provide a convenient approach that could be used for future studies of elongation. This approach could also help researchers find out how abnormalities in translation contribute to human diseases, including muscle-wasting disorders.

https://doi.org/10.7554/eLife.13429.002

Introduction

Initiation of protein synthesis in eukaryotic cells proceeds via two well-established pathways. The cap-dependent pathway involves recognition of 7-methyl-guanosine at the 5’-terminus of mRNA by a preinitiation complex of 40S ribosomal subunit and a host of initiation factors prior to a scanning step that results in initiator aminoacyl-tRNA(aa-tRNA) pairing with a cognate start codon, followed by 60S binding to form the 80S initiator complex (Jackson et al., 2010; Aitken and Lorsch, 2012). The second pathway involves binding of the ribosome to an internal ribosome entry site (IRES), a structure that is present in many virus-encoded mRNAs, as well as in some cellular mRNAs (Fitzgerald and Semmler, 2009). Initiation of protein synthesis from an 80S·IRES complex can take place in the absence of some or even all of the initiation factors required in the cap-dependent pathway (Filbin and Kieft, 2009), depending on the IRES source. The intergenic IRES of Cricket Paralysis Virus (CrPV-IRES) forms a complex with 80S ribosomes that is capable of initiating the synthesis of complete proteins in cell-free assays completely lacking initiation factors (Jan et al. 2003; Pestova and Hillen, 2003). More recently, high resolution structural studieshave shown that, prior to polypeptide chain initiation, the closely related Dicistroviridae IRES structures from CrPV (Fernandez et al., 2014; Muhs et al., 2015) and Taura syndrome virus (Koh et al., 2014) occupy all three tRNA binding sites (E, P, and A) on the ribosome, with the protein coding region beginning immediately downstream from IRES segment occupying the A-site (Figure 1).

Figure 1 with 1 supplement see all
Structure of CrPV-IRES bound to the 80S ribosome superposed on A, P, and E tRNA binding sites.

The position of the first codon is indicated. Adapted from Fernandez et al. (2014).

https://doi.org/10.7554/eLife.13429.003

CrPV-IRES binds with high affinity (Kd ~ 10 nM) to the 80S ribosome (Jang and Jan, 2010), raising the question of what effect the necessary IRES dissociation from the tRNA binding sites, and ultimately from all of the ribosome as well, has on the rates of initial peptide elongation steps as nascent peptide is formed (Muhs et al., 2015). Since prior to the work reported in this paper nothing had been published concerning the rate of initial oligopeptide synthesis by an 80S·CrPV-IRES complex, it has been unclear whether there is a retarding effect due to the presence of IRES on the ribosome, and, if so, how many cycles of peptide elongation are required before the ribosome begins to form peptide bonds at a higher rate. In considering this question, we make use of the simplified 12-step scheme of initial tetrapeptide synthesis shown in Figure 2, which provides a useful framework for presenting the results described in this paper. In this scheme Steps 1–3 show the reactions required for initial binding of the first tRNA to the A site followed by translocation to the P-site, and reactions 4–6, 7–9, and 10–12 represent three elongation cycles, ending with P-site bound tetrapeptidyl-tRNA, completing the third cycle of polypeptide synthesis. This model makes the reasonable assumption that binding of successive aminoacyl-tRNAs (aa-tRNAs) cognate to the mRNA requires the progressive removal of IRES structures from each of the tRNA binding sites, such that translocation of dipeptidyl tRNA to the P-site (structure 7) requires removal of the IRES from the last of the three tRNA binding sites. In the work reported below, we demonstrate first, that the initial elongation steps are indeed quite slow and are limited by the translocation step of the elongation cycle, and second, that the rate of elongation accelerates following translocation of tripeptidyl-tRNA to the P-site.

Proposed scheme for initial tetrapeptide synthesis on CrPV IRES-programmed ribosomes.

This simplified scheme neglects the several substeps, including GTP hydrolysis, Pi release, and elongation factor release, that accompany both productive binding of ternary complex to the ribosome (Steps 2, 4, 7, 10) and tRNA translocation (Steps 3, 6, 9, 12).

https://doi.org/10.7554/eLife.13429.005

Results

In our experiments, eukaryotic ribosomes are prepared from shrimp cysts (Iwasaki and Kaziro, 1979), elongation factors are prepared from yeast, and charged tRNAs are prepared from yeast and E. coli. In addition, the peptide coding sequence attached to the 3’-end of the CrPV-IRES (Figure 1) has been mutated for ease in detection of peptide synthesis via 35S-Met incorporation. In all such mutants the initial wt-codon triplet GCU encoding Ala has been replaced by UUC, encoding Phe, a change that has little effect on the expression of active luciferase in a cell-free protein synthesis assay (Figure 1—figure supplement 1). The initial coding sequences of the mutants used in this work are presented in Supplementary file 1. Collectively, they allow monitoring of the rates of PheMet, PheLysMet, PheValLysMet and PheLyValArgGlnTrpLeuMet synthesis. In presenting the results below, Steps 1–12 and structures 113 are as described in the scheme for initial tetrapeptide synthesis proposed in Figure 2. Values of t1/2 for Steps 1–12, determined as described, are summarized in Table 2.

Rates of Phe-TC binding to the 80S·IRES complex: Steps 1–3, structures 1–4

We previously have utilized two assays to measure binding of the ternary complex Phe-tRNAPhe·eEF1A·GTP (Phe-TC) to the 80S·CrPV-IRES (80S·IRES) complex (Ruehle et al., 2015). The increase in proflavin-labeled Phe-tRNAPhe fluorescence anisotropy measures binding to either the A- or P-site (structures 3 and 4, respectively, Figure 2). [3H]-Phe-tRNAPhe cosedimentation with the 80S·IRES complex measures accumulation of 4 only, since A-site binding is too labile to survive the ultracentrifugation step (Yamamoto et al., 2007).

In Figure 3 we present time-resolved application of the anisotropy assay that allows us to measure the rates of Phe-TC binding to form Structure 3 from 1. These resultswere fit to the scheme shown in Figure 2, giving values for k1, k-1, and k2 in both the presence and absence of eEF2·GTP that are summarized in Table 1. In the absence of eEF2 (blue trace), the equilibrium position of Step 1, a so-called pseudo-translocation step (Muhs et al., 2015) in which the IRES vacates the A-site, favors Structure 1 over Structure 2 by approximately 20-fold, consistent with recent structural studies (Fernandez et al., 2014; Koh et al., 2014; Muhs et al., 2015). Phe-TC binds to Structure 2 yielding Structure 3, in a process where the rate-limiting step is the conversion of Structure 1 to Structure 2. Preincubation of 80S·IRES complex with 1 µM or 3 µM eEF2·GTP leads to clear biphasic binding of Phe-TC, with the more rapid and slower phases each accounting for ~50% of binding, respectively (red and black traces). These results indicate that, consistent with recent results of Petrov et al. (2016), the equilibrium between Structures 1 and 2 is shifted in the presence of eEF2·GTP, such that approximately half of 80S·IRES is present as 2.Phe-TC binding to 2, resulting in the formation of Structure 3, accounts for the rapid phase in the red and black traces. Further formation of 3 is limited by the slower rate of 1 to 2 conversion. Although added eEF2·GTP decreases all three apparent rate constants, the effect is much greater on k-1 (~50-fold reduction) than on either k1 (~twofold reduction) or k2 (~fourfold reduction). The near identity of the red and black traces, performed at different eEF2·GTP concentrations, suggests that this factor interacts with both 1 and 2, with a dissociation constant significantly less than 1 µM. The large inhibitory effect of eEF2·GTP on k-1 is consistent with its role as a translocase, and with recent results demonstrating that a principal role of EF-G, the prokaryotic equivalent of eEF2, is to inhibit back-translocation (Adio et al., 2015). eEF2·GTP inhibition of k2 may be due, at least in part, to a requirement for eEF2·GDP dissociation prior to Phe-TC binding.

Figure 3 with 2 supplements see all
Rates of initial Phe-tRNAPhe binding measured by fluorescence anisotropy or Phe-tRNAPhe cosedimentation.

Fluorescence anisotropy changes were monitored after rapid mixing of Phe-tRNAPhe (Prf) ternary complex (0.1 µM final concentration, containing 1 mM GTP)with 80S·FVKM-IRES complex (0.1 µM final concentration) either in the absence of eEF2 (blue line) or with 80S·FVKM-IRES complex that was pre-incubated with either 3 µM (black line) or 1 µM eEF2·GTP (red line) for 1–2 hr. These long times ensured full equilibration prior to TC addition. In the latter cases, eEF2 concentration was kept constant by including 3 µM or 1 µM eEF2, respectively, in the TC solution. eEF2 displays virtually no GTPase activity when it is not bound to the ribosome (Nygård and Nilsson, 1989). Rates of Phe-tRNAPhe binding to the P site, as determined by cosedimentation, were measured by rapidly mixing Phe-TC (1.6 µM final concentration) with 80S·FVKM-IRES complex (0.8 µM final concentration) pre-incubated for 5’ – 60’ in the presence (1 µM) (□) or absence of eEF2·GTP (○). In both cases, eEF2 final concentration after mixing was adjusted to 1 µM, by including 1 µM or 2 µM eEF2·GTP, respectively, in the TC solution. After quenching with 0.5 M MES buffer (pH 6.0), ribosome bound Phe-tRNAPhe was measured by cosedimentation. In the preincubation experiment, three-fold increases of both eEF2·GTP and Phe-TC concentrations, or of just eEF2·GTP concentration, had little effect on the cosedimentation results. Results in this Figure are corrected for IRES-independent changes in fluorescence anisotropy or Phe-tRNAPhe cosedimentation (Figure 3—figure supplements 1,2). All three solid green lines are best fits of the results obtained to the scheme in Figure 2, using the numerical integration program Scientist.

https://doi.org/10.7554/eLife.13429.006
Table 1

Apparent rate constants for Steps 1 and 2.

https://doi.org/10.7554/eLife.13429.009
Apparent rate constants (s-1)-eEF2+eEF2
k10.0071 ± 0.00330.0033 ± 0.0001
k-10.15 ± 0.040.0034 ± 0.0001
k2 ([Phe-TC] = 0.1 µM)0.11 ± 0.040.0256 ± 0.0002

Formation of Structure 4 from Structure 1, as measured by the co-sedimentation assay, requires the presence of eEF2·GTP and proceeds at a considerably slower rate than formation of Structure 3 from Structure 1 (Figure 3), allowing estimation of a t1/2 for Step 3, a second pseudo-translocation step involving conversion of 3 to 4, of 210 ± 10 s. It is this further slow step that accounts for the lack of significant effect of preincubation with eEF2·GTP (5’ or 60’) on the rate of formation of 4 from 1 (Figure 3).

Rates of oligopeptide formation and Met-tRNAMet cosedimentation

Using ribosomes programmed with the appropriate coding sequence mutants (Supplementary file 1) and [35S]-Met-TC, we employ a rapid mixing and quench assay to measure rates of PheMet, PheLysMet, and PheLysValMet synthesis, with detection and quantification of product by thin layer electrophoresis (TLE) (Figure 4A and Figure 4—figure supplement 1). For PheMet synthesis (Figure 4B) we preform Structure 4 and measure its conversion to Structure 6. We measurePheLysMet synthesis, Structure 9, starting from either Structure 4 or Structure 7 (Figure 4C) and PheValLysMet synthesis, Structure 12, starting from either Structure 7 or Structure 10 (Figure 4D). In all three cases, reactions involving only TC binding and a single peptide bond formation (4 to 6; 7 to 9; 10 to 12) proceed in remarkably similar fashion, each showing biphasic behavior with a rapid phase accounting for 65 ± 10% of reaction proceeding with a t1/2 of ~6–9 s and a slower, minor phase proceeding much more slowly (t1/2 ~220–240 s), possibly corresponding to defective ribosomes. Reactions involving formation of two peptide bonds, as in the conversion of 4 to 9 or 7 to 12 are well approximated as single phase reactions with t1/2 values of 90–110 s. Conversion of 4 to 9 proceeds via Steps 4 – 8, allowing the t1/2 value for the translocation Step 6 to be estimated as 84 s, from the difference between the t1/2 value for the 4 to 9 reaction and the sum of the t1/2 values for the 4 to 6 and 7 to 9 reactions (major phases). Similarly, the t1/2 value for the translocation Step 9 can be estimated as 110 s from the difference between the t1/2 value for the 7 to 12 reaction and the sum of the t1/2 values for the 7 to 9 and 10 to 12 reactions. Since the di-, tri- and tetrapeptides synthesized in the results reported in Figure 4 use different coding sequence mutants, these estimates of translocation t1/2 values depend on the not unreasonable assumption that the identities of the tRNAs undergoing translocation do not have a major influence on the translocation rate. With this caveat, the results presented in Figure 4 lead to the clear conclusion that translocation is the rate limiting step in each of the first two cycles of polypeptide elongation, proceeding from 4 to 10.

Figure 4 with 2 supplements see all
Kinetics of peptide synthesis and Met-tRNAMet cosedimenting with ribosomes.

Reaction mixtures were quenched at various times after mixing. Peptide synthesis aliquots were quenched with 0.8 M KOH, and the released [35S]-containing peptide was resolved and quantified by TLE and autoradiography (Materials and methods). Cosedimentation assay aliquots were quenched with with 0.5 M MES buffer (pH 6.0) and [35S] cosedimenting with ribosomes was determined. For all the reactions shown, final concentrations of reactants after mixing were: 80S·IRES complexes (0.8 μM); all added TCs (1.6 µM); eEF2·GTP (1 µM). The numbers in blue in parts (BD) refer to the Structures in Figure 2 whose rates of conversion are measured. For example, the peptide synthesis result in part (B) labeled 46 measures conversion of Structure 4 to Structure 6. (A) Time course for formation of PheValLysMet tetrapeptide as determined by TLE. 80S·FVKM-IRES complex was mixed with Phe-TC, Val-TC, Lys-TC and [35S]-Met-TC. The migration positions of [35S]-Met and [35S]-PheValLysMet (*) are indicated. (B) 80S·FM-IRES complexes with Phe-tRNAPhe at the P site were mixed with [35S]-Met-TC. Dipeptide synthesis (□); cosedimentation assay (■). (C) Tripeptide synthesis: 80S·FKM-IRES complexes with either Phe-tRNAPhe (O) in the P site (Structure 4) or PheLys-tRNALys (Δ) in the P site (Structure 7) were mixed with either Lys-TC and [35S]-Met-TC or with just [35S]-Met-TC, respectively. Cosedimentation assay: 80S·FKM-IRES complex with PheLys-tRNALys in the P site was mixed with [35S]-Met-TC (■). (D) Tetrapeptide synthesis: 80S·FVKM-IRES complexes with either PheVal-tRNAVal (O) in the P site (Structure 7) or PheValLys-tRNALys (Δ) in the P site (Structure 10) were mixed with either Lys-TC and [35S]-Met-TC or with just [35S]-Met-TC, respectively. Cosedimentation assay: 80S·FKM-IRES complex with PheValLys-tRNALys in the P site was mixed with [35S]-Met-TC (■). Solid lines are best fits using single (B, 4–7; C, 4–9; D, 7–12) or double (B, 4–6; C, 7–8, 7–9; D, 10–11, 10–12) exponentials.

https://doi.org/10.7554/eLife.13429.010
Table 2

t1/2 values*.

https://doi.org/10.7554/eLife.13429.013
Step (s)t1/2 (s)
1
1 (+eEF2)
230 ± 5
237 ± 5
2
2 (+eEF2)
15 ± 9
30 ± 5
3210 ± 10
4 + 58 ± 2
4-898 ± 15
6 = (4-8) – (4+5) – (7+8)§84 ± 16
73 ± 1
8 = (7+8) – 7§4 ± 2
7 + 86 ± 2
7-11128 ± 26
9 = (7-11) – (7+8) – (10+11)§110 ± 30
102 ± 1
11 = (10 + 11) – 10§7 ± 3
10 + 119 ± 2
12<10
  1. * Error ranges shown are based on the variances of fits to single or double exponentials of the results presented in Figure 4, unless otherwise noted.

  2. Calculated as 0.69 (k-1 + k2)/k1k2 (see Table 1).

  3. Calculated as 0.69 (k-1 + k2)/k22 (see Table 1).

  4. § Error ranges for these steps, which are not observed directly, are based on the error ranges of the directly observed steps.

In an attempt to resolve the TC binding step (reactions 4, 7, and 10) from the peptide formation step (reactions 5, 8, and 11) we also employed a rapid mixing and quench assay to determine the rates with which [35S]-Met-tRNAMet is able to cosediment with the ribosome following mixing of [35S]-Met-TC with structures 4, 7, or 10. This strategy was successful for [35S]-Met–TC reaction with structure 7 (containing P-site bound PheLys-tRNALys, Figure 4C) or structure 10 (containing P-site bound PheValLys-tRNALys Figure 4D), in which the [35S]-Met-TC cosedimentation rates outpace the rates of peptide bond formation with Met-TC. These rate differentials permit estimates to be made for the t1/2 values of TC binding (Step 7, 3 s; Step 10, 2 s) and peptide bond formation (Step 8, 4 s; Step 11, 7 s). They also provide a clear indication that, within Structures 8, 9, 11 and 12, Met-tRNAMet, PheLysMet-tRNAMet, and PheValLysMet- tRNAMet, whenbound to the A-site, efficiently cosediment with ribosomes, which is typical for A-site bound tRNAs in conventional (non-IRES) elongation complexes (Warner and Rich, 1964; Nwagwu, 1975).

However, for [35S]-Met–TC reaction with structure 4 (containing P-site bound Phe-tRNAPhe), the [35S]-Met-TC cosedimentation rate is much slower than the dipeptide formation rate (Figure 4B). This indicates that PheMet-tRNAMet, and possibly Met-tRNAMet as well, are not bound stably to the ribosome in Structures 5 and 6, and that only PheMet-tRNAMet bound to the P-site (Structure 7) is fully recovered by cosedimentation. As a result, the cosedimentation assay does not permit estimation of the t1/2 values for Steps 4 and 5. It is possible that the lability of the A-site tRNAs in structures 5 and 6 is due to IRES binding to the E-site, which is absent in structures 8, 9 and 11, 12, and may reflect an allosteric A-site: E-site interaction. Evidence for allosteric A-site/E-site interactions has been presented for both bacterial and eukaryotic ribosomes (Nierhaus 1990; Chen et al., 2011; Ferguson et al., 2015), although the general validity of this interaction has been questioned (Semenkov et al., 1996; Petropoulos and Green, 2012).

Translocation of tetrapeptidyl-tRNA (Step 12) is much more rapid than of tripeptidyl-tRNA (Step 9)

The results presented in Figure 4 show that translocation proceeds slowly through the first two elongation cycles of nascent protein synthesis, raising the question of how far nascent protein synthesis has to proceed to overcome the retarding effect of ribosome-bound IRES. In Figure 5 we present the results of two experimental approaches demonstrating that translocation of tetrapeptidyl-tRNA proceeds much more rapidly than translocation of tripeptidyl-tRNA.

Figure 5 with 1 supplement see all
Tetrapeptide translocation (Step 12) is faster than tripeptide translocation (Step 9).

(A) Puromycin reaction with PheValLys-tRNALys bound either at the A site (D) or at the P-site (O) of the 80S·FVKM-IRES complex or being translocated from the A site to the P site (□). (B) Puromycin reaction with PheValLysMet-tRNAMet either bound at the P-site (O) of the 80S·FVKM-IRES complex or being translocated from the A site to the P site (□). Lines in A. and B. Are fits to single exponentials. (C) Time dependence of PheLysValArgGlnTrpLeuMet octapeptide synthesis from the 80S·FKVRQWLM-IRES complex containing various peptidyl-tRNAs pre-bound at the P site, as indicated. The pre-bound peptidyl tRNAs were prepared using the standard procedure (see Complex Preparations in Materials and methods) by incubating the 80S-IRES complex with the relevant TCs for 15 min. The remaining TCs needed for octapeptide synthesis, including [35S]-Met-TC, were then added, each at a concentration of 1.6 µM, for the indicated times prior to quenching. PheLysValArgGlnTrpLeuMet octapeptide synthesis was measured by [35S]-Met cosedimenting with 80S ribosome.

https://doi.org/10.7554/eLife.13429.014

The first approach makes use of the fact that formation of peptidyl-puromycin proceeds more rapidly with peptidyl-tRNA bound to the P-site than to the A-site, permitting puromycin reactivity to distinguish A-site from P-site peptidyl-tRNA. As shown in Figure 5A, puromycin (1 mM) reacts with A-site bound PheValLys-tRNALys, Structure 9, about 20times more slowly (t1/2 1400 ± 300 s) than it reacts with P-site bound PheValLys-tRNALys(t1/2 76 ± 16 s). The corresponding t1/2 value for puromycin reaction with PheValLys-tRNALys undergoing translocation from the A- to P-site is 170 ± 30 s. This increase of approximately 100 s for translocating PheValLys-tRNALysvs. translocated PheValLys-tRNALys closely matches the t1/2 value of 110 ± 30 s estimated above for the translocation of tripeptidyl-tRNA (Table 2) and can be assigned to the translocating step. In contrast, the rates of puromycin reaction with translocating and translocated PheValLysMet-tRNAMet (Structure 13)are indistinguishable from one another (t1/2 values of 37 ± 4 s and 46 ± 7 s, respectively, Figure 5B), a clear demonstration that translocation of PheValLysMet-tRNAMet proceeds rapidly with respect to puromycin reaction. Our results allow us to estimate an upper limit value of t1/2 for the translocation Step 12 of ≤10 s.

Puromycin reacts at similar rates with translocated PheValLys-tRNALys (Structure 10, t1/2 76 ± 16 s) and PheValLysMet-tRNAMet (Structure 13, t1/2 46 ± 7 s). These rates, while consistent with those reported by others for puromycin reaction with eukaryotic P-site bound Met-tRNAMet (Lorsch and Herschlag, 1999), N-AcPhe-tRNAPhe (Ioannou et al., 1997), and Cy3-Met-tRNAMet (Ferguson et al., 2015), are several hundred-fold slower than those measured for puromycin reaction with prokaryotic P-site bound peptidyl- or fMet-tRNA. This largely explains why the rate reduction for puromycin reaction with A-site vs. P-site bound peptidyl-tRNA is so much more modest for eukaryotic ribosomes (~20-fold, Figure 5A) than for prokaryotic ribosomes (103–104-fold, Pan et al., 2007; Semenkov et al., 1992 ; Sharma et al., 2004; Peske et al., 2004 ).

Above we have demonstrated that, under our conditions, aa-tRNA binding and peptide bond formation proceed with an overall t1/2of 6 – 9 s for each of the three elongation steps we have studied. This relative constancy, coupled with the much slower translocation of tripeptidyl-tRNA (Step 9) vs. tetrapeptidyl-tRNA (Step 12), leads to the prediction that synthesis of a longer peptide that required the tripeptidyl-tRNA translocation step (Step 9) would proceed significantly more slowly than synthesis not requiring this step.

In the second approach we verified this prediction by demonstrating that octapeptide FKVRQWLM formation, as measured by the cosedimentation assay, is much slower when synthesis is initiated with P-site bound PheLys-tRNALys (Structure 7) vs. P-site bound PheLysVal-tRNAVal (Structure 10) (Figure 5C). Indeed, the rates of FKVRQWLM synthesis are only marginally increased when reaction is initiated with P-site bound tetrapeptidyl-tRNA or pentapeptidyl-tRNA as compared with tripeptidyl-tRNA, reinforcing the notion that the retarding effect of ribosome-bound IRES on protein synthesis is largely overcome following translocation of tripeptidyl-tRNA.

Discussion

The results presented in this paper constitute the first time that rates of reaction have been determined for the initial cycles of IRES-dependent elongation. They demonstrate quite clearly that the first two cycles of elongation proceed much more slowly than subsequent steps, and that these reduced rates arise from slow, rate-determining, pseudo-translocation and translocation steps. Translocation during the first elongation cycle (Step 6) clearly requires displacement of the IRES from the E-site, so it is not unexpected that it would be slow. Less predictable is the slow translocation in the second elongation cycle, (Step 9) after the IRES structure has, presumably, already left the E-site (Figure 1). The slow rate of Step 9 might be due to a full dissociation of IRES from the ribosome during this step, a suggestion that could be tested by appropriately designed structural studies. In any case, our results do clearly demonstrate that, following translocation of tripeptidyl-tRNA from the A- to P-site, the pace of nascent peptide chain elongation picks up dramatically. Further work, comparing quantitatively the rates of successive cycles of nascent peptide elongation following tetrapeptide formation (i.e, cycles 4, 5, 6, 7, etc.) will be required to determine how many cycles are required before any retarding influence of bound CrPV-IRES is completely eliminated.

Our results also clarify an aspect of the initial binding of the first aa-tRNA to the 80S·CrPV-IRES complex. Prior results have shown that initial aa-tRNA binding, in the form of a ternary complex, to an 80S·IRES complex, as measured either by cosedimentation (Fernandez et al., 2014), or by filter binding and toeprinting (Yamamoto et al., 2007), requires eEF2·GTP, leading to the conclusion that initial aa-tRNA binding can only bind to the 80S·IRES complex after an eEF2-dependent translocation event (Fernandez et al., 2014). While we agree with the experimental results, and have in fact reproduced the cosedimentation result in our own work, we disagree with the conclusion. This is because these earlier experiments only measured stable aa-tRNA binding, corresponding to formation of Structure 4 in which aa-tRNA binds to the P-site. However, it is clear from the anisotropy experiment conducted in the absence of added eEF2·GTP (Figure 3, blue trace) that ternary complex binding measured in situ, which can monitor labile binding to the A-site (Structure 3)does not require eEF2·GTP. This is easily understood as an example of Le Chaltelier’s principle, in which the equilibrium between Structure 1 (closed A- site)and Structure 2 (open A- site), which strongly favors Structure 1, is pulled to the right by aa-tRNA binding. Preincubation with eEF2 also shifts the equilibrium to the right, leading to an initial rapid phase of reaction with Phe-TC (Figure 3).

This latter shift, for which the results presented in Figure 3 provide strong inferential evidence, appears to be at odds with earlier toeprinting results showing no shift in IRES position within the 80S·CrPV-IRES complex on addition of eEF2 alone (Pestova et al., 2003; Jan et al., 2003). In agreement with the suggestion of Muhs et al. (2015), we believe it likely that this apparent inconsistency arises from eEF2 dissociation from the ribosome during the toeprinting assay (Pestova et al., 1996), with the consequent favoring of Structure 1. This is because GTP is required for tight binding of eEF2 to the ribosome (Nygård and Nilsson, 1984), but the toeprinting assay is carried out for an extended period of time (45 min) under non-denaturing conditions in the absence of added GTP, conditions that would eventually deplete GTP due to ribosome-dependent eEF2·GTP hydrolysis (Nygård and Nilsson, 1989). In addition, the toeprinting assay is performed at a Mg2+ concentration of 10.5 mM, considerably higher than the 5 mM used in our kinetic studies, which could also affect the 1 to 2 equilibrium position.

How relevant are the present results for in vivo initiation of IRES-dependent protein synthesis? We note three potential concerns. First, our in vitro systemis quite heterogeneous, with ribosomes derived from shrimp cysts, yeast elongation factors, and yeast and E. coli charged tRNAs. However, as reviewed in Koh et al. (2014), IRESs can initiate translation on ribosomes from many eukaryotic organisms, including shrimp (Cevallos and Sarnow, 2005), indicating that the molecular mechanism is not species-specific. CrPV IRESs in particular can initiate translation on ribosomes from yeast (Thompson et al., 2001) to human (Spahn et al., 2004). Furthermore, eukaryotic elongation factors have structures that are very strongly conserved (Soares et al., 2009; Jørgensen et al., 2006), and there is strong evidence that charged tRNAs from one species form functional complexes with both eEF1A and ribosomes from a different species (Jackson et al., 2001; Ferguson et al., 2015). Second, the coding sequences employed in this work are different from that immediately downstream of wt-CrPV-IRES (Supplementary file 1). This is also unlikely to pose a major difficulty, given the strong evidence that mutations in the downstream sequence are, in general, tolerated without substantial effect on initiation of translation (Tsukiyama-Kohara et al., 1992; Wang et al., 1993; Hellen and Sarnow, 2001Rijnbrand et al., 2001), although mutations of some downstream sequences do give rise to relatively minor changes in IRES activity (Kim et al., 2003; Shibuya et al., 2003; Wang et al., 2013). Third, the elongation rate of even the later cycles of IRES-dependent elongation (Figure 5C) is quite slow (~0.1 s-1). Although this rate is essentially identical to that reported for tripeptide synthesis in a cap-dependent yeast-based in vitro translation system which requires five initiation factors and eEF3 in addition to eEF1A and eEF2 (Acker et al., 2007; Eyler and Green, 2011; Gutierrez et al., 2013), it is 1.5–2 orders of magnitude slowerthan rates of peptide elongation that have been estimated for intact eukaryotic cells at 37°C (3–10 s-1) (Boehlke and Friesen, 1975; Hershey, 1991). There is evidence that, in many eukaryotic cells, the protein synthesis machinery is highly organized, containing several components, including ribosomes, a multi-aminoacyl-tRNA synthetase complex, eEF-1A, and several auxiliary proteins (Negrutskii et al., 1994; Negrutskii and El’skaya, 1998; David et al., 2011). It has been suggested that this organized structure optimizes translation rate by coordinating synthetase activities to facilitate channeling of aa-tRNAs to the elongating ribosomes. Thus, protein synthesis in a permeabilized mammalian cell, in which this structure is likely to be preserved, proceeds 40-fold faster than what is obtained in a cell-free system prepared from the same cells which presumably lacks this structure (Negrutskii et al., 1994). The slow rates measured for both the IRES-dependent and cap-dependent in vitro systems could be due, at least in part, to their lack of aa-tRNA channeling. Such channeling would be unlikely to accelerate the very slow translocation rates in the initial peptide elongation cycles reported in this work, although we cannot exclude the possibility that other proteins present in vivo might have such effects. Future efforts will address this issue. Here, incorporation of some of the features of a recently introduced in vitro protein synthesissystem in which initiation is carried out using the IRES from hepatitis C virus could be useful (Machida et al., 2014).

Detailed mechanistic characterization of many aspects of eukaryotic polypeptide elongation has been held back by the lack of a convenient system for its study. The very simple in vitro IRES-dependent elongation system described here should be useful in overcoming this limitation. As one example, it is generally assumed, based on extensive structural similarities (Jørgensen et al., 2006), that eEF2 functions in catalyzing eukaryotic elongation in much the same way that EF-G catalyzes prokaryotic elongation, but this assumption does not take into account some important structural differences, including the fact that eEF2 is subject to post-translational modifications not found in EF-G, with clear consequences for activity but, as yet, little understanding of mechanism (Dever and Green, 2012; Mittal et al., 2013 ; Greganova et al., 2011; Liu et al., 2012). The CrPV-IRES based system should permit detailed rate and structural dynamic studies of eEF2 catalytic function, of the kind that have proved so useful in elucidating EF-G function in bacterial protein synthesis (Pan et al., 2007; Chen et al., 2011; Chen et al., 2013; Holtkamp et al., 2014; Salsi et al., 2015).

Materials and methods

Plasmid construction and cloning

The wt CrPV Phe-IRES vector, as well as several variants in which the first Ala codon is replaced by a Phe codon, were the kind gifts of Dr. Eric Jan.This replacement, which has little effect on the initiation of translation (see Discussion and Figure 1—figure supplement 1), was made as a matter of convenience, since tRNAPhe was available to us and the appropriate tRNAAla acceptor was not. The vectors encoding the PheMet, PheValMet, PheValLysMet, and PheLysValArgGlnTrpLeuMet were generated by PCR insertion of corresponding sequences (Supplementary file 1) into the CrPV Phe-IRES vector. All cloned sequences were verified by standard sequencing methods using appropriate primers.

In vitro transcription

For in vitro transcription of full-length mRNA for the Luciferase assay, the WT and mutated Phe-IRES plasmids were linearized with XbaI, which cleaves the plasmids after the firefly luciferase coding region. mRNA was transcribed in vitro using the AmpliScribe T7 transcription kit (EPICENTRE) according to the manufacturer. For in vitro transcription of short-length mRNAs, the mutated IRES plasmids were linearized with NarI, which cleaves 33 nt downstream of the ATG start codon of the luciferase coding region.

Luciferase assay

In vitro translation of firefly luciferase with WT and mutated F-IRES mRNA (1 μg in 50 μL of reaction mixture) was performed using the Flexi Rabbit Reticulocyte Lysate System (Promega) according to the manufacturer. IRES mRNA was omitted in the control reaction. Fluc activities (Figure 1—figure supplement 1) were determined using a plate reader (Envision 2103, Perkin-Elmer) to detect the luminescence signal.

Ribosomes, elongation factors and tRNAs

Shrimp (A. salina) 80S ribosomes were prepared from dried, frozen cysts as previously described (Iwasaki and Kaziro, 1979) with some modifications. After the shrimp cysts were ground open, debris was removed by centrifugation at 30,000xg for 15 min and crude 80S ribosomes were precipitated from the supernatant by addition of 4.5% (w/v) PEG 20K (Ben-Shem et al., 2011). 40S and 60S subunits were resolved on 10–30% sucrose gradients after puromycin treatment. E. coli 30S subunits were prepared as described (Grigoriadou et al., 2007). eEF1A was purified from yeast according to published methods (Thiele et al., 1985). His6-eEF2 was isolated from an overexpressing yeast strain (TKY675) generously provided by Dr. Terri Kinzy, and purified as described (Jørgensen et al., 2002). Proflavin-labeled Phe-tRNAPhe, denoted Phe-tRNAPhe(prf), was prepared as previously described (Wintermeyer and Zachau 1974, Betteridge et al., 2007). Yeast tRNAPhe was purchased from Sigma. Other isoacceptor tRNAs were prepared from bulk tRNA (Roche) from either E. coli (tRNAGln, tRNALys, tRNAMet) or yeast (tRNAArg, tRNALeu, tRNATrp, tRNAVal) by hybridization to immobilized complementary oligoDNAs, as described (Barhoom et al., 2013 ; Liu et al., 2014). E. coli and yeast tRNAs were charged with their cognate amino acids as described (Pan et al., 2006, 2009).

Complex preparations. TCs and various 80S·IRES complexes

All complexes were prepared in buffer 4 (40 mM Tris-HCl pH 7.5, 80 mM NH4Cl, 5 mM MgOAc2, 100 mM KOAc, 3 mM 2-mercaptoethanol) at 37°C. For the preparation of ternary complexes (TC, aa-tRNA·eEF1A·GTP) and 80S·IRES complexes containing either Phe-tRNAPhe or peptidyl-tRNA bound in the P-site, buffer 4 was supplemented with 1 mM GTP and 1 mM ATP. All TC complexes were prepared by incubating the relevant charged tRNA (1.6 μM, based on amino acid stoichiometry) with eEF1A (8 μM) for 5 min. 80S·IRES complexes were formed by incubation of shrimp 40S (0.8 µM) and 60S (1.6 µM) subunits with the appropriate IRES (2.4 µM) for 5 min. 80S·IRES complexes containing Phe-tRNAPhe or peptidyl-tRNA bound in the P-site were formed by mixing 80S·IRES complexes (0.8 µM) with 1 μM eEF2 and the appropriate TCs (1.6 µM for each) for 15–40 min. To determine radioactively labeled aa-tRNA binding stoichiometries, 40 µL samples were subjected to ultracentrifugation at 4°C (540,000xg) for 40 min through a 1.1 M sucrose cushion. Excess bacterial 30S bacterial ribosome subunits (600 pmol/15 ± 5 µL) were added as carrier to enhance pelleting and allow facile calculation of complex recovery. Control experiments carried out in the absence of IRES or of both IRES and 80S ribosomes demonstrated that only negligible amounts of labeled peptidyl-tRNA cosedimented due to binding to 30S subunits (Figure 4—figure supplement 2). The pellets were gently washed twice with buffer 4 and dissolved in 100 μL of buffer 4 for A260 determination. Ribosome recoveries typically varied between 60 and 80%.

Kinetic measurements

Unless otherwise noted, all reactions were performed at 37°C in buffer 4 supplemented with 1 mM GTP. All kinetic results reported are the averages of 2–4 independent determinations, performed on different days. No systematic effort was made to carry out duplicate experiments using independently made stock reagent solutions, although this was sometimes done. Error bars in figures are shown as average deviations.

Rates of Phe-TC binding by fluorescence anisotropy change (Figure 3)

Phe-tRNAPhe(prf)·eEF1A·GTP ternary complex was rapidly mixed with 80S·FVKM-IRES complex in the presence or absence of eEF2·GTP using a KinTek stopped-flow spectrofluorometer model SF-300X. Proflavin labeled Phe-tRNAPhe was excited at 462 nm and monitored using a pair of 495 nm long-pass filters. A T-shape configuration was utilized such that instrument-specific polarizers were attached to both the excitation and the two emission light paths. In each independent measurement, 15–20 shots (rapid mixing of samples) were averaged to provide the time course of anisotropy change. The g-factor and anisotropy value were calculated using the instrument software as described (Lakowicz 1999, Ameloot et al., 2013). Experimental data were processed and analyzed by Felix software (from PTI).

Rates of [3H]-Phe-TC or [35S]-Met-TC binding by cosedimentation

80S·IRES complex (0.8 μM) with no tRNA bound (Figure 2) was rapidly mixed with [3H]-Phe-TC in the presence of eEF2·GTP in a KinTek Corporation RQF-3 Rapid Quench-Flow Instrument the reaction mixture was quenched at various times with 0.5 M MES buffer (pH 6.0), and the stoichiometry of ribosome-bound [3H]-Phe-TC was determined by ultracentrifugation as described above for complex characterization. Similar procedures were used to determine the kinetics of [35S]-Met-TC binding to 80S·IRES complexes containing Phe-tRNAPhe or peptidyl-tRNA in the P-site (Figure 4).

Rates of peptide synthesis

Di-, Tri- and Tetrapeptide

80S·IRES complexes containing either Phe-tRNAPhe or the appropriate peptidyl-tRNA in the P-site, prepared using the standard procedure (see Complex preparations above), were rapidly mixed with [35S]-Met-TC (1.6 µM) and additional TCs as required (all 1.6 µM) in a KinTek Corporation RQF-3 Rapid Quench-Flow Instrument, and the reaction mixture was quenched at various times with 0.8 M KOH. [35S]-Met-containing peptide was released from tRNAMet by further incubation at 37°C for 3 hr. The pH of the samples were adjusted with acetic acid to pH 2.8, lyophilized, suspended in water, and centrifuged to remove particulates, which contained no 35S. The supernatant was analyzed by thin layer electrophoresis as previously described (Youngman et al., 2004), using the same running buffer, and the labeled peptide was located by autoradiography. The identities of PheMet, PheLysMet, and PheValLysMet, (Figures 4A, Figure 4-figure supplement 1, Figure 5C) were confirmed by their comigrations with authentic samples obtained from GenScript (Piscataway, NJ). A further demonstration of tetrapeptide identity was provided by matrix-assisted laser desorption/ionization (MALDI) mass spectrometric analysis (Ultraflex III TOF/TOF, Bruker: Phe-Val-Lys-Met(Na+), calculated, 546.7; found 546.6. In addition, the 80S·IRES complex containing P-site bound Phe-Val-Lys-Met-tRNA was reacted for 40 min with 10 mM puromycin (37°C, buffer 4 plus 1 mM GTP). The resulting puromycin adduct, Phe-Val-Lys-Met-puro(H+), released into solution, was also identified by MALDI: calculated, 978.7; found, 978.9.

Octapeptide

80S·FKVRQWLM-IRES complexes containing the appropriate peptidyl-tRNA in the P-site, prepared using the standard procedure (see Complex preparations above)were mixed with [35S]-Met-TC (1.6 µM) and additional TCs as required (all 1.6 µM), for various times followed by quenching with 0.5 M MES (pH 6.0) buffer. PheLysValArgGlnTrpLeuMet octapeptide synthesis was measured by [35S]-Met cosedimentation with 80S·FKVRQWLM-IRES complexes. Over the time scale of these measurements (60 –600 s, Figure 5C), all [35S]-Met-tRNAMet stably bound to the ribosome undergoes a peptide transfer reaction (see Table 2). PheLysValArgGlnTrpLeuMet peptide was released from tRNAMet using the base treatment described above for other peptidyl tRNAs and its identity was confirmed by its comigration during thin layer electrophoresis with an authentic sample obtained from GenScript (Piscataway, NJ) (Figure 5C—figure supplement 1).

Rates of puromycin adduct formation

Rates of puromycin adduct formation were measured for FVK-tRNALys bound in the A-site, and for both FVK-tRNALys and FVKM-tRNAMet either pre-bound in the P-site or undergoing translocation from the A-site to the P-site. In all cases, reaction mixtures were quenched with 0.5 M MES (pH 6.0) buffer. The quenched samples were next ultracentrifuged and the radioactivity co-sedimenting with the ribosome, which decreases as more puromycin adduct is formed, was determined. All incubations and reactions were carried out at 37°C. All complexes were reacted with puromycin (1 mM, final concentration) for various times before quenching. No decreases in radioactivity co-sedimenting with the ribosome were observed in the absence of added puromycin.

FVK-tRNALys bound in the A-site (Structure 9, PRE-3)

80S·FVKM-IRES complex with FVK-tRNA pre-bound at A-site was formed by incubating the 80S·FVKM-IRES complex containing FV-tRNA at the P-site (Structure 7, POST-2, 0.4 µM), purified by sedimentation through a sucrose cushion, with 0.8 μM [3H]-Lys-TC for one minute. The resulting complex was then reacted with puromycin.

FVK-tRNALys bound in the P-site (Structure 10, POST-3)

Two procedures were employed, which yielded equivalent results. Procedure 1: 80S·FVKM-IRES complex containing FV-tRNA at the P-site (Structure 7, POST-2, 0.4 µM), purified as described above, was incubated with 0.8 μM [3H]-Lys-TC and eEF2.GTP (1.0 µM) for 15 min at 37°C. The resulting complex was then reacted with puromycin. Procedure 2: 80S·FVKM-IRES complex (0.8 µM) was preincubated for 15 min with eEF2·GTP (1 µM) and Phe-TC, Val-TC, and [3H]-Lys-TC (all TCs present at 1.6 µM). The resulting complex was then reacted with puromycin.

FVKM-tRNAMet bound in the P-site (Structure 13, POST-4)

80S·FVKM-IRES complex (0.8 µM) was preincubated for 15 min with eEF2·GTP (1 µM) and Phe-TC, Val-TC, Lys-TC and [35S]-Met-TC (all TCs present at 1.6 µM). The resulting complex was then reacted with puromycin.

FVK-tRNALys undergoing translocation (Structure 9 becoming Structure 10)

80S·FVKM-IRES complex (0.8 µM) was preincubated for 15 min with eEF2·GTP (1 µM) and Phe-TC and Val-TC, both present at 1.6 µM, yielding PheVal-tRNAVal bound in the P-site. This complex was then mixed for 1 min with ([3H]-Lys-TC (1.6 µM) to form PheValLys-tRNALys bound in the A-site, which was then reacted with puromycin in the presence of additional added eEF2·GTP (final concentration 1 µM).

FVKM-tRNAMet undergoing translocation (Structure 12 becoming Structure 13)

80S·FVKM-IRES complex (0.8 µM) was preincubated for 15 min with eEF2·GTP (1 µM) and Phe-TC, Val-TC, and Lys-TC (all TCs present at 1.6 µM), yielding PheValLys-tRNALys bound in the P-site. This complex was then mixed for 1 min with ([35S]-Met-TC (1.6 µM) to form PheValLysMet-tRNAMet bound in the A-site, which was then reacted with puromycin in the presence of additional added eEF2·GTP (final concentration 1 µM).

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
    Cellular content of ribonucleic acid and protein in Saccharomyces cerevisiae as a function of exponential growth rate: calculation of the apparent peptide chain elongation rate
    1. KW Boehlke
    2. JD Friesen
    (1975)
    Journal of Bacteriology 121:429–433.
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
    Principles of Fluorescence Spectroscopy
    1. JR Lakowicz
    (1999)
    New York: Kluwe Academic/ Plenum.
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
    Elongation factor 1 alpha from Saccharomyces cerevisiae. Rapid large-scale purification and molecular characterization
    1. D Thiele
    2. P Cottrelle
    3. F Iborra
    4. JM Buhler
    5. A Sentenac
    6. P Fromageot
    (1985)
    Journal of Biological Chemistry 260:3084–3089.
  65. 65
  66. 66
    Internal ribosome entry site within hepatitis C virus RNA
    1. K Tsukiyama-Kohara
    2. N Iizuka
    3. M Kohara
    4. A Nomoto
    (1992)
    Journal of Virology 66:1476–1483.
  67. 67
    Translation of human hepatitis C virus RNA in cultured cells is mediated by an internal ribosome-binding mechanism
    1. C Wang
    2. P Sarnow
    3. A Siddiqui
    (1993)
    Journal of Virology 67:3338–3344.
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72

Decision letter

  1. Alan G Hinnebusch
    Reviewing Editor; National Institute of Child Health and Human Development, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "One, two, three, go – initiating polypeptide elongation in an IRES-dependent system" for consideration by eLife. Your article has been favorably evaluated by John Kuriyan (Senior editor) and three reviewers, one of whom, Alan Hinnebusch, is a member of our Board of Reviewing Editors.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary of work:

This paper conducts in vitro experiments using a purified system to study the rates of initial oligopeptide synthesis by an 80S·CrPV-IRES complex. The results suggest that the first two elongation cycles involved in producing a di- or tripeptide are slow and are limited by the translocation step, and that the rate of elongation accelerates following translocation of tripeptidyl-tRNA to the P-site (the second true translocation event). The IRES was suitably modified to replace the 1st Ala codon with a Phe codon, allowing fluorescence anisotropy measurements of the rate of proflavin-modified Phe-tRNAPhe binding to either the A- or P-site, and to allow incorporation of labeled Met-tRNAMet to monitor production of FM, FKM, FVKM, and FKVRQWLM synthesis. Sedimentation of complexes through sucrose gradients was used to assay stable binding of Met-tRNA or peptidyl-tRNAs containing labeled Met to the ribosome. The anisotropy results in Figure 3 suggest that EF2 shifts the equilibrium of the first pseudotranslocation step in which the IRES is moved in the decoding center to vacate the A-site, and thereby increases the rate of Phe-tRNAPhe binding to the A site in a biphasic reaction. The sedimentation assay indicates that the second pseudotranslocation step, in which Phe-tRNA is translocated from the A to the P site is much slower than the combination of the first pseudotranslocation plus Phe-tRNAPhe binding to the A site. They go on to use a rapid mixing and quench assay to measure rates of FM, FKM, and FKVM synthesis, with detection and quantification of product by thin layer electrophoresis (TLE), using the appropriate pre-formed substrates for addition of labeled Met to Phe-tRNAPhe or the relevant peptidyl tRNA. Results in Figure 4 indicate that the rates of aa-tRNA binding to the A site and formation of a peptide bond are very similar for all 3 substrates that produce a di-, tri-, or tetrapeptide, and the calculated rates of translocation in these reactions indicate that the rates of translocation for the dipeptidyl- and tripeptidyl-tRNAs are slow and rate-limiting in forming the tri- and tetrapeptides, respectively. Using the sedimentation assay, they could also measure the very rapid rates of ternary complex binding to the A site in the 2nd and 3rd elongation cycles producing the tri- and tetrapeptides, but found that A-site bound TC or FM-tRNA was too unstable to capture for the intermediates 5-6 containing the IRES in the E site, which they speculate might involve an allosteric effect of the IRES in destabilizing A-site binding. By then measuring rates of puromycin reaction with either the tri- or tetrapeptidyl tRNAs either under conditions where prior translocation from the P-site to the A-site is required or where the peptidy-tRNAs are prebound to the A site, they deduce that the rate of the 3rd translocation of tetrapeptidyl-tRNA from the P to A site is very rapid compared to the 2nd or 3rd translocation events.

Essential revisions:

The following issues raised in the referees' comments shown below need to be addressed in the revised manuscript.

Ref. #1: All of the requests for clarification of reaction conditions or interpretations of results should be addressed with appropriate revisions of text.

Ref.#2: The request for complete details about fitting the kinetic data in comment #1 should be honored.

As described in comment #2, the need for modifying the IRES sequence should be more explicitly stated in Materials and methods, as it would have been technically feasible to carry out all experiments with the WT sequence. In addition, it is important to compare the rates of IRES-mediated translation given by the native sequence versus your modified IRES sequence either in a cell-free extract or in your reconstituted system.

To respond to points 3. & 4., it would be sufficient to revise the text to remove statements about allostery; for point 5, to explain why you monitored the puromycin reaction as you did in Figure 5A-B. For points 6-10, clarify the relevant issues with the appropriate revisions of text, and provide the requested control to rule out Met-tRNA binding to 30S subunits.

Ref. #3: Points 1.-3 concern your conclusion that eEF2 does not stimulate the rate of TC binding in step 2 and rather only shifts the equilibrium in step 1 towards structure 2, which is at odds with other published work. As this is a hypothetical interpretation of the biphasic kinetics in Figure 3, it is important that you attempt to provide evidence for this deduced effect of eEF2 on the equilibrium of step 1 using toe-printing analysis of the complexes. It is also necessary to justify the statement that k-1 is much smaller than k2, which is not self-evident, and explain how it could be that eEF2 would shift the equilibrium to structure 2 without altering the t1/2 of step 1. You must also discuss the discrepancy with the previous results, particularly those of Ruehle et al.

You should clarify the issues raised in points 4-6 with the appropriate revisions of text.

Original reviews:

Reviewer #1:

The results are significant in providing strong evidence that the two pseudotranslocation steps (steps 1 and 3 in Figure 2) as well as the two conventional translocation steps (steps 6 and 9) involved in producing a tripeptide and translocating it to the P site occur slowly and are rate-limiting for the first two elongation cycles of CrPV IRES-directed translation, and that the next translocation step occurs much more rapidly and does not appear to be rate-limiting for production of longer polypeptides. The slow rate of the 2nd translocation step is surprising because the IRES should no longer occupy the decoding sites, so they may be detecting a novel intermediate with the IRES still bound to the ribosome outside of the decoding center and affecting a ribosome conformational change involved in translocation. The results are also significant in providing evidence against the previous conclusion by others that EF2 is required for the first pseudotranslocation that moves the IRES out of the A site (step 1) to permit recruiting the first TC to the A-site, as here they found no effect of EF2 on the rate of this reaction when it is monitored directly by fluorescence anisotropy measurements as opposed to stable TC binding to the P site assayed by sedimentation, which occurs only after the second pseudotranslocation reaction (step 3). Finally, the results using the CrPV IRES are valuable in laying the groundwork for studying eukaryotic elongation in a simplified purified system without the need for initiation factors to form the first peptide bound. There are a number of instances where the reaction conditions or interpretations of results need to be more carefully described, as follows:

1) In the second paragraph of the subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”: This section is very confusing as the composition of the aa-tRNA or peptidyl-tRNA species in question, and whether it is bound to the A or P sites, is generally unclear. Particularly, in the clause "…These rate differentials provide a clear indication that peptidyl- Met-tRNAMet and Met-tRNAMet bound to the A-site (Structures 8, 9 and 11, 12), as well as peptidyl-Met-tRNAMet binding to the P-site (Structures 10 and 13), efficiently…", the clarity could be improved by mentioning the specific peptidyl-tRNAs by their amino acid compositions for each of the structures, as "peptidyl-Met-tRNAMet" is imprecise nomenclature.

2) In the subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”: Also, it's not perfectly clear whether the instability of Met-tRNA binding to the A-site observed for structure 3 and 5, and FM-tRNA to structure 6 using the co-sedimentation assay is unusual; whereas stable binding of both Met-tRNA and peptidyl-tRNA to the A-site in structures 8, 9, 11-12 is typical of A-site-bound ligands in conventional (non-IRES) elongation complexes; and this distinction needs to be carefully spelled out.

3) In the second paragraph of the subsection “Translocation of tetrapeptidyl-tRNA (Step 12) is much more rapid than of tripeptidyl-tRNA (Step 9)”: requires citation of a figure for puromycin synthesis rates.

4) In the second paragraph of the subsection “Translocation of tetrapeptidyl-tRNA (Step 12) is much more rapid than of tripeptidyl-tRNA (Step 9)” and Figure 5A-B: more detail is required about how the experiment in Figures 5A-B was done in terms of starting reactants.

5) In the last paragraph of Results: more detail is required to understand how binding of peptidyl tRNA harboring labeled Met is distinguished from Met-tRNA binding to the A-site in these assays.

6) Error estimates have not been defined in any of the figures or tables and this needs to be rectified.

Reviewer #2:

The paper by Cooperman et al. provides a first kinetic description of the initial steps of translation by eukaryotic ribosomes after the initiation at an IRES. There are very few papers on mechanistic aspects of eukaryotic translation in general and the present work provides very interesting, significant insights. It is a real challenge to obtain clean kinetic data for the eukaryotic system and this paper provides valuable information on the kinetic properties of eukaryotic ribosomes. The technical quality of the data is high (I particularly like the quality of the stopped-flow traces, they are really excellent) and the conclusions are warranted. My only reservation is that many technical aspects are not clearly described. Given that this paper delineates new kinetic approaches to study eukaryotic translation, it is particularly important to provide a very detailed description of methods. This really has to be carefully revised. I personally do not like the title, because it is not informative at all; a more rigorous title would be better.

Detailed comments:

1) Fitting of the kinetic data in this paper is mostly non-exponential and would require fitting to a model by numerical integration. This part is completely missing in Materials and methods. This must be described in all details required to understand and reproduce the calculations.

2) Among all potential criticism of the experimental system (Discussion, third paragraph), I am particularly worried about the altered coding sequence. It is not clear why the native Ala codon was replaced by Phe, as unlabeled amino acids are used anyway and the tRNAs are prepared by a method which should readily yield tRNA(Ala), i.e. one could label tRNA(Ala) with Prf (Kothe and Rodnina, 2007). This means that in principle one could use the native sequence. Furthermore, the measurements and rate determinations would be even more straightforward if [14C] and [3H]-labeled amino acids were used instead of [35S]Met; in this case one could use the native sequence. The need for modifying the sequence should be more explicitly stated in Materials and methods. Clearly, the whole set of experiments cannot be repeated with Ala as a 1st amino acid; however, it would be highly desirable to compare IRES-mediated translation with a native sequence and a modified sequence in the cell-free translation system (commercial or reconstituted from components).

3) The "instability" of aa-tRNA binding in A site is not likely to be caused by the dissociation of aa-tRNA from the A site prior to peptide bond formation. As authors show, peptide bond formation is rapid (Table 1); so as soon as aa-tRNA binds, it will be rapidly incorporated into peptide. Instead, peptidyl-tRNA tends to drop-off easily (see Semenkov 2000 and Konevega 2004). Although this dissociation is relatively slow, it normally explains the drop-off during centrifugation (which takes minutes to hours). In any case, there is no evidence for the allosteric interactions between the A and E sites. The statements on the allosteric interplay should be removed, as they only weaken the paper (subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”, third paragraph) (By the way, the so-called "allosteric effect" is simply caused by the presence of deacylated tRNA in aa-tRNA preps, which chases the labeled tRNA from the E site. There are several other groups, in addition to Rachel Green, who provided strong evidence against the model).

4) The instability issues addressed in point 3 could be overcome by using nitrocellulose filtration assays, which are really very commonly used to study ribosome studies. Are there any technical issues which preclude the use of nitrocellulose filter in this case? If yes, this should be described in Materials and methods.

5) The authors use puromycin reaction to identify the P-site position of the peptidyl-tRNA, which is a very reliable and well-established method. It is therefore surprising that the authors did not use the time-resolved puromycin assays (which they have established for the prokaryotic ribosomes) to determine the rates of translocation in a more direct fashion than described in Table 1? If there is some technical issue specific for eukaryotic ribosomes, it would be important to indicate this in Materials and methods.

6) Discussion: "Further work will be required to determine how many cycles of elongation are required before any retarding influence of bound IRES is completely eliminated": I am confused here. Table 1 shows that k12 is very fast, i.e. the 5th round is already rapid. This should be clarified.

7) In the subsection “Complex preparations. TCs and various 80S·IRES complexes”. 30S subunits were added to the centrifugation assays as a carrier. However, 30S subunits can bind [35S]Met (at least to some extent) even in the absence of the mRNA. Are there controls for that?

8) Figure 1 legend. "in the both cases" – it is not clear which two cases are meant. It is also not clear why the incubation with EF2 is so long (1-2 hours) – is it really necessary? I guess the authors just wanted to be on the safe side, but this has to be clearly stated in Materials and methods.

9) Figure 4 legend; "or just [35S]Met-TC”. My expectation is that in the absence of the preceding Lys-TC Met should not be incorporated at all. This has to be clarified.

10) In the subsection “Kinetic measurements”, "averages of 2-4 independent determinations" – do the authors mean technical replicates or independent experiments (i.e. biological replicated)?

Reviewer #2 (Additional data files and statistical comments):

The statistical information is appropriate except for the clarification needed in the subsection “Kinetic measurements” (point 10 in the review).

Reviewer #3:

Zhang and colleagues have investigated the effect of the CrPV IRES RNA on the first rounds of translation elongation including the initial pseudo-translocation steps. They have performed for the first time experiments to determine the rates of the various main steps of the elongation cycles. They show that the IRES initial significantly slows down the translocation steps are down until the tetrapeptide stage is reached. The CrPV is an important minimal model system for translation initiation in eukaryotes The results are novel and interesting. However, the taking into account the following points is recommended before the paper can be published.

1) According to Figure 3, blue curve, the binding of tRNA to the A-site (structure 2 in the nomenclature of the authors) reaches nearly the same level with or without eEF2. However, in previous experiments binding of tRNA was dependent on eEF2 or at least significantly increased (Yamamoto et al., 2007; Fernandez et al., 2014; Ruehle et al., 2015). The experiments by Ruehle et al. have been done by colocalization using fluorescence. The discrepancy should be discussed. Furthermore, the authors should perform experiments to establish the nature of the complex in a more direct manner, e.g. by toeprinting of the complex.

2) The authors state that k-1 is much smaller than k2 and was assumed to be negligible. How has this been determined? In fact, a toeprint signal indicative of translocation cannot be obtained by incubation of binary 80S-IRES complexes with eEF2 alone, but requires an A-site ligand, too (Jan et al., 2003; Muhs et al., 2015). This suggest that the translocated complex (structure 2) is kinetically labile and contradicts the present statement that the back-translocation is negligible. Again, the authors should do experiments, e.g. toeprinting, to provide additional evidence that pre-incubation with eEF2-GTP shifts the equilibrium between structures 1 and 2 from 95:5 to 50:50 as stated.

3) The explanation about the effect of eEF2 on the equilibrium between structure 1 and 2 (Discussion) is curious. They propose that eEF2 shifts the equilibrium towards structure 2 at the same time do not measure a significant effect on the t1/2 of step 1. How this is possible? Do the authors propose that k-1 is becoming smaller, i.e. that eEF2 slows down back-translocation without accelerating forward translocation?

4) Can the authors rule out that back-translocation or peptidyl-tRNA drop off influences the following steps towards tetrapeptide synthesis?

Additional points:

5) The first sentence of the paper "Initiation of protein synthesis in eukaryotic cells proceeds via two well-established pathways" is to some extent misleading. As the authors acknowledge later factor requirement of different IRES RNAs is diverse and the structure and mechanism of different IRES RNAs is also different. So there are several pathways leading to internal initiation.

6) The peptide coding sequence is not really part of the IRES. Therefore, it is misleading to talk about mutant IRES, when the changes are in the ORF.

7) At the beginning of the Results part the authors should specify in which system they are working.

8) The green line is not mentioned in the legend to Figure 3.

9) A-site/E-site allostery has been recently reported by Ferguson et al., 2015, Mol Cell for human 80S ribosomes. This is relevant for the respective discussion in the third paragraph of the subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Kinetics of initiating polypeptide elongation in an IRES-dependent system" for further consideration at eLife. Your revised article has been favorably evaluated by John Kuriyan (Senior editor), a Reviewing editor, and two reviewers.

The manuscript has been improved but there are some remaining issues raised by Reviewer #3 that need to be addressed before acceptance. In considering these new comments, Reviewer #2 also asks that you carefully review the Materials and methods to insure that they contain all of the critical information that would be required to duplicate the work.

Reviewer #2:

The authors have significantly improved the manuscript by satisfactorily addressing the comments and concerns of the reviewers. The story they prevent extends our knowledge on the function IRES-dependent systems and can be published in eLife.

Reviewer #3:

In the revised version Zhang and colleagues have addressed many points raised by the reviewers and have improved their paper. However, there are still some points where discussion should be extended.

1) The authors now confirm that the effect of eEF2 is to inhibit k-1 and write that this is "consistent with its role as a translocase". However, the accepted role of EF-G/eEF2 is to accelerate translocation, not to inhibit back-translocation. This warrants additional discussion.

2) The authors have pre-incubated 80S-FVKM-IRES complexes with eEF2-GTP for 1 – 2 hr. How they can rule out that GTP consumption has an impact on the experiment? Is it possible that accumulation of eEF2-GDP during pre-incubation results in a significant fraction of 80S FVKM-IRES-eEF2 complexes (between Structures 1 and 2) to slow down k2.

3) The authors write that the anisotropy assay in Figure 3 reports on formation of Structure 3 from 1. However, as the co-sedimentation assay reports, during the measurement time there should be also formation of Structure 4 at later time.

4) The authors state that A-site-bound Phe-tRNAPhe is labile. Can they estimate the off-rate k-2. Is it valid to neglect k-2 for numerical integration?

https://doi.org/10.7554/eLife.13429.017

Author response

Essential revisions:

The following issues raised in the referees' comments shown below need to be addressed in the revised manuscript.

Ref. #1: All of the requests for clarification of reaction conditions or interpretations of results should be addressed with appropriate revisions of text.

This has been addressed.

Ref.#2: The request for complete details about fitting the kinetic data in comment #1 should be honored.

This has been addressed.

As described in comment #2, the need for modifying the IRES sequence should be more explicitly stated in Materials and methods, as it would have been technically feasible to carry out all experiments with the WT sequence.

This was simply a matter of convenience, as now stated explicitly in the first paragraph of the Materials and methods. When we started this work tRNAPhe was available to us and the tRNAAla acceptor was not, and Eric Jan told us that the Phe codon UUC could be substituted for the Ala GCU codon with little or no functional consequence.

In addition, it is important to compare the rates of IRES-mediated translation given by the native sequence versus your modified IRES sequence either in a cell-free extract or in your reconstituted system.

This has been addressed by addition of Figure 1—figure supplement 1.

To respond to points 3. & 4., it would be sufficient to revise the text to remove statements about allostery; for point 5, to explain why you monitored the puromycin reaction as you did in Figure 5A-B. For points 6-10, clarify the relevant issues with the appropriate revisions of text, and provide the requested control to rule out Met-tRNA binding to 30S subunits.

These points have been addressed, but, as explained in response to Reviewer #2, point #3, we would prefer to leave the allosteric discussion in the text.

Ref. #3: Points 1.-3 concern your conclusion that eEF2 does not stimulate the rate of TC binding in step 2 and rather only shifts the equilibrium in step 1 towards structure 2, which is at odds with other published work. As this is a hypothetical interpretation of the biphasic kinetics in Figure 3, it is important that you attempt to provide evidence for this deduced effect of eEF2 on the equilibrium of step 1 using toe-printing analysis of the complexes. It is also necessary to justify the statement that k-1 is much smaller than k2, which is not self-evident, and explain how it could be that eEF2 would shift the equilibrium to structure 2 without altering the t1/2 of step 1. You must also discuss the discrepancy with the previous results, particularly those of Ruehle et al.

We disagree with the request for a toeprinting experiment, for reasons elaborated in response to Reviewer 3’s comments #s 1 – 3. However, we have carried out a more complete analysis of our kinetic data, resulting in the addition of a new table (revised Table 1) and a more cogent presentation of t1/2 values for Steps 1 and 2 in revised Table 2 (formerly Table 1). We are indebted to Reviewer 3 for raising this point. We also rebut the statement that there is a discrepancy between our present results and those of Ruehle et al.

You should clarify the issues raised in points 4-6 with the appropriate revisions of text.

We have addressed points 4 and 6 in our response. We disagree with point 5, as stated in our detailed response.

Original reviews:

Reviewer #1:

The results are significant in providing strong evidence that the two pseudotranslocation steps (steps 1 and 3 in Figure 2) as well as the two conventional translocation steps (steps 6 and 9) involved in producing a tripeptide and translocating it to the P site occur slowly and are rate-limiting for the first two elongation cycles of CrPV IRES-directed translation, and that the next translocation step occurs much more rapidly and does not appear to be rate-limiting for production of longer polypeptides. The slow rate of the 2nd translocation step is surprising because the IRES should no longer occupy the decoding sites, so they may be detecting a novel intermediate with the IRES still bound to the ribosome outside of the decoding center and affecting a ribosome conformational change involved in translocation. The results are also significant in providing evidence against the previous conclusion by others that EF2 is required for the first pseudotranslocation that moves the IRES out of the A site (step 1) to permit recruiting the first TC to the A-site, as here they found no effect of EF2 on the rate of this reaction when it is monitored directly by fluorescence anisotropy measurements as opposed to stable TC binding to the P site assayed by sedimentation, which occurs only after the second pseudotranslocation reaction (step 3). Finally, the results using the CrPV IRES are valuable in laying the groundwork for studying eukaryotic elongation in a simplified purified system without the need for initiation factors to form the first peptide bound. There are a number of instances where the reaction conditions or interpretations of results need to be more carefully described, as follows:

1) In the second paragraph of the subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”: This section is very confusing as the composition of the aa-tRNA or peptidyl-tRNA species in question, and whether it is bound to the A or P sites, is generally unclear. Particularly, in the clause "…These rate differentials provide a clear indication that peptidyl- Met-tRNAMet and Met-tRNAMet bound to the A-site (Structures 8, 9 and 11, 12), as well as peptidyl-Met-tRNAMet binding to the P-site (Structures 10 and 13), efficiently…", the clarity could be improved by mentioning the specific peptidyl-tRNAs by their amino acid compositions for each of the structures, as "peptidyl-Met-tRNAMet" is imprecise nomenclature.

The requested clarifications have been made, both in the Results and Experimental Sections.

2) In the subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”: Also, it's not perfectly clear whether the instability of Met-tRNA binding to the A-site observed for structure 3 and 5, and FM-tRNA to structure 6 using the co-sedimentation assay is unusual; whereas stable binding of both Met-tRNA and peptidyl-tRNA to the A-site in structures 8, 9, 11-12 is typical of A-site-bound ligands in conventional (non-IRES) elongation complexes; and this distinction needs to be carefully spelled out.

The requested distinction has been made. See subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”, end of second paragraph and start of the third paragraph.

3) In the second paragraph of the subsection “Translocation of tetrapeptidyl-tRNA (Step 12) is much more rapid than of tripeptidyl-tRNA (Step 9)”: requires citation of a figure for puromycin synthesis rates.

The references for the reactivity of P-site bound aminoacyl-tRNA with puromycin in eukaryotic ribosomes were originally provided (Lorsch and Herschlag, 1999 and Ioannu et al., 1997). In writing the revised version we became aware of a paper published in late 2015 by Ferguson et al., which also provided a pertinent result, and this reference was also added. In addition, in considering this comment we realized that there were no reliable data extant for the reactivity of A-site bound aminoacyl-tRNA with puromycin in eukaryotic ribosomes. Accordingly, we performed the relevant experiment for A-site bound PheValLys-tRNALys, demonstrating that A-site reactivity is approximately 20-fold less than P-site reactivity. This result, which has been added to Figure 5A with a description in the third paragraph of the subsection “Translocation of tetrapeptidyl-tRNA (Step 12) is much more rapid than of 182 tripeptidyl-tRNA (Step 9)”, fully supports our original conclusion, based on results presented in our initial submission, that translocation of PheValLys-tRNALys is much slower than translocation of PheValLysMet-tRNAMet. We also added two sentences (in the aforementioned paragraph) explicitly comparing A-site vs. P-site reactivity with puromycin in prokaryotic and eukaryotic ribosomes.

4) In the second paragraph of the subsection “Translocation of tetrapeptidyl-tRNA (Step 12) is much more rapid than of tripeptidyl-tRNA (Step 9)” and Figure 5A-B: more detail is required about how the experiment in Figure 5A-B was done in terms of starting reactants.

More detail was added to the description of the puromycin reaction in the subsection entitled “Rates of puromycin adduct formation” as requested. In addition, the reference to Figure 4 in the original version of this section was corrected to Figure 5.

5) In the last paragraph of Results: more detail is required to understand how binding of peptidyl tRNA harboring labeled Met is distinguished from Met-tRNA binding to the A-site in these assays.

A sentence justifying the use of the cosedimentation assay to measure octapeptide formation involving peptide transfer to ribosome-bound Met-tRNAMet has been added to the subsection entitled “Octapeptide” in the Experimental section. We also added Figure 5C—figure supplement 1 that confirms octapeptide synthesis.

6) Error estimates have not been defined in any of the figures or tables and this needs to be rectified.

In the subsection labeled “Kinetic measurements” the last sentence has been amended to read “Error bars in figures are shown as average deviations.” In addition, several additions were made to revised Table 2 (formerly Table 1), including a full description of the error ranges, added as footnotes a. and d., as well as the measured t1/2 values for the combined steps 4-8 and 7-11.

Reviewer #2:

The paper by Cooperman et al. provides a first kinetic description of the initial steps of translation by eukaryotic ribosomes after the initiation at an IRES. There are very few papers on mechanistic aspects of eukaryotic translation in general and the present work provides very interesting, significant insights. It is a real challenge to obtain clean kinetic data for the eukaryotic system and this paper provides valuable information on the kinetic properties of eukaryotic ribosomes. The technical quality of the data is high (I particularly like the quality of the stopped-flow traces, they are really excellent) and the conclusions are warranted. My only reservation is that many technical aspects are not clearly described. Given that this paper delineates new kinetic approaches to study eukaryotic translation, it is particularly important to provide a very detailed description of methods. This really has to be carefully revised. I personally do not like the title, because it is not informative at all; a more rigorous title would be better.

The title has been changed as requested.

Detailed comments:

1) Fitting of the kinetic data in this paper is mostly non-exponential and would require fitting to a model by numerical integration. This part is completely missing in Materials and methods. This must be described in all details required to understand and reproduce the calculations.

Only the results presented in Figure 3 and Figure 3—figure supplement 2 were fit using the numerical integration program Scientist. Although Scientist was mentioned in the Figure 3 legend of the original version, this point is now emphasized in the revised version by adding the phrase “the numerical integration program” to the last line of the figure legend.

2) Among all potential criticism of the experimental system (Discussion, third paragraph), I am particularly worried about the altered coding sequence. It is not clear why the native Ala codon was replaced by Phe, as unlabeled amino acids are used anyway and the tRNAs are prepared by a method which should readily yield tRNA(Ala), i.e. one could label tRNA(Ala) with Prf (Kothe and Rodnina, 2007). This means that in principle one could use the native sequence. Furthermore, the measurements and rate determinations would be even more straightforward if [14C] and [3H]-labeled amino acids were used instead of [35S]Met; in this case one could use the native sequence. The need for modifying the sequence should be more explicitly stated in Materials and methods. Clearly, the whole set of experiments cannot be repeated with Ala as a 1st amino acid; however, it would be highly desirable to compare IRES-mediated translation with a native sequence and a modified sequence in the cell-free translation system (commercial or reconstituted from components).

An experiment demonstrating that replacement of the initial Ala codon GCU by UUC encoding Phe has little effect on active luciferase expression has been added to the text as Figure 1—figure supplement 1 and referenced in the text (1st para in Results), with experimental details provided (subsections “In Vitro Transcription” and “Luciferase assay”).

3) The "instability" of aa-tRNA binding in A site is not likely to be caused by the dissociation of aa-tRNA from the A site prior to peptide bond formation. As authors show, peptide bond formation is rapid (Table 1); so as soon as aa-tRNA binds, it will be rapidly incorporated into peptide. Instead, peptidyl-tRNA tends to drop-off easily (see Semenkov 2000 and Konevega 2004). Although this dissociation is relatively slow, it normally explains the drop-off during centrifugation (which takes minutes to hours). In any case, there is no evidence for the allosteric interactions between the A and E sites. The statements on the allosteric interplay should be removed, as they only weaken the paper (subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”, third paragraph) (By the way, the so-called "allosteric effect" is simply caused by the presence of deacylated tRNA in aa-tRNA preps, which chases the labeled tRNA from the E site. There are several other groups, in addition to Rachel Green, who provided strong evidence against the model).

There are two points raised here. We and the reviewer are in agreement that PheMet-tRNAMet is not bound stably to the A-site. However, we also think it possible that Met-tRNAMet bound to the A-site may also be bound unstably. This is based on our observation of a large difference in the relative stoichiometries measured by the co-sedimentation and peptide synthesis assays at 3 s and 10 s after Met-TC addition (Figure 4B). Based on our estimated t1/2 values for the peptide formation steps 8 and 11 (4 s and 7s, respectively), we would expect both PheMet-tRNAMet and Met-tRNAMet to be bound to the ribosome within 10 s after Met-TC addition, and so cannot rule out that Met-tRNAMet is bound unstably to the A-site following step 4. To reflect this uncertainty, we have changed the wording of the relevant sentence in the text to read:

“This indicates that PheMet-tRNAMet, and possibly Met-tRNAMet as well, are not bound stably to the ribosome in Structures 5 and6”.

As to the second point, there are indeed other groups than Green et al. who have argued against an allosteric A-site/E-site effect, in particular Wintermeyer, Rodnina and their co-workers, and an appropriate reference (Semenkov et al., 1996) has been added to the text. However, there remains a need to explain the very evident differences we observe between the relative stoichiometries measured by the co-sedimentation and peptide synthesis assays in the first (Figure 4B), second (Figure 4C), and third (Figure 4D) peptide elongation steps. As stated in the revised text we think that:

“It is possible that the lability of the A-site tRNAs in structures 5and6is due to IRES binding to the E-site, which is absent in structures 8, 9and 11, 12.

So then the question becomes why would a putative IRES binding to the E-site cause this difference in behavior. Above we have mentioned studies arguing against A-site:E-site allosteric interaction, but there are also studies that support the notion, most notably by Nierhaus (Nierhaus 1990), but also by ourselves (Chen et al., 2011), and most recently, by Ferguson et al. (2015), the latter notably based on work with eukaryotic ribosomes. Given this background, we think it is permissible for us to speculate that the lability of A-site tRNAs in structures 5and6:

“may reflect an allosteric A-site: E-site interaction”.

(see also Reviewer #3, point #9). This said, we are willing to withdraw the latter phrase and the references to the A-site:E-site controversy if the editors object to their inclusion.

4) The instability issues addressed in point 3 could be overcome by using nitrocellulose filtration assays, which are really very commonly used to study ribosome studies. Are there any technical issues which preclude the use of nitrocellulose filter in this case? If yes, this should be described in Materials and methods.

We did not use nitrocellulose filter assays, and do not think that such assays are required for this paper.

5) The authors use puromycin reaction to identify the P-site position of the peptidyl-tRNA, which is a very reliable and well-established method. It is therefore surprising that the authors did not use the time-resolved puromycin assays (which they have established for the prokaryotic ribosomes) to determine the rates of translocation in a more direct fashion than described in Table 1? If there is some technical issue specific for eukaryotic ribosomes, it would be important to indicate this in Materials and methods.

There are technical issues. The rate of puromycin reaction with peptidyl-tRNA bound in the P-site of the eukaryotic ribosome is much slower (several hundred-fold) than the corresponding rate in the prokaryotic ribosome, so that only if the translocation rate is even slower (as in the case of the tripeptidyl-tRNA) is it possible to use the time-resolved puromycin assay to get information about the translocation rate. Second, we were not confident of the stability of A-site bound peptidyl-tRNA toward co-sedimentation and so decided to generate A-site bound peptidyl-tRNA in situ prior to puromycin addition. More detail about the puromycin reaction has been added to both the Results (subsection “Translocation of tetrapeptidyl-tRNA (Step 12) is much more rapid than of tripeptidyl-tRNA (Step 9)”) and Experimental (subsection “Kinetic measurements”) sections of the revised manuscript (see also response to Reviewer 1’s comments #s 3 and 4).

6) Discussion: "Further work will be required to determine how many cycles of elongation are required before any retarding influence of bound IRES is completely eliminated": I am confused here. Table 1 shows that k12 is very fast, i.e. the 5th round is already rapid. This should be clarified.

A phrase has been added to the relevant sentence to make this point clear.

7) In the subsection “Complex preparations. TCs and various 80S·IRES complexes”. 30S subunits were added to the centrifugation assays as a carrier. However, 30S subunits can bind [35S]Met (at least to some extent) even in the absence of the mRNA. Are there controls for that?

Yes. Co-sedimentation due to binding to 30S subunits is negligible. This point has been addressed by adding both an appropriate sentence to the subsection in Materials and methods entitled “Complex Preparations”and Figure 4C—figure supplement 1 which provides illustrative data.

8) Figure 1 legend. "in the both cases" – it is not clear which two cases are meant. It is also not clear why the incubation with EF2 is so long (1-2 hours) – is it really necessary? I guess the authors just wanted to be on the safe side, but this has to be clearly stated in Materials and methods.

The reviewer is referring to the Figure 3 legend and is correct about the reasoning. A sentence has been added to the legend make this point explicit.

9) Figure 4 legend; "or just [35S]Met-TC”. My expectation is that in the absence of the preceding Lys-TC Met should not be incorporated at all. This has to be clarified.

Clarifying text has been added. The basic point is that, in 4C, the kinetics of tripeptide synthesis was followed for either 5 steps (Structure 4 converted to Structure 9) or two steps (Structure 7 converted to Structure 9). Similarly, in 4D the kinetics of tetrapeptide synthesis was followed for either 5 steps (Structure 7 converted to Structure 12) or two steps (Structure 10 converted to Structure 12).

10) In the subsection “Kinetic measurements”, "averages of 2-4 independent determinations" – do the authors mean technical replicates or independent experiments (i.e. biological replicated)?

Clarifying language on this point has been added to the subsection “Kinetic measurements”.

Reviewer #3:

Zhang and colleagues have investigated the effect of the CrPV IRES RNA on the first rounds of translation elongation including the initial pseudo-translocation steps. They have performed for the first time experiments to determine the rates of the various main steps of the elongation cycles. They show that the IRES initial significantly slows down the translocation steps are down until the tetrapeptide stage is reached. The CrPV is an important minimal model system for translation initiation in eukaryotes The results are novel and interesting. However, the taking into account the following points is recommended before the paper can be published.

1) According to Figure 3, blue curve, the binding of tRNA to the A-site (structure 2 in the nomenclature of the authors) reaches nearly the same level with or without eEF2. However, in previous experiments binding of tRNA was dependent on eEF2 or at least significantly increased (Yamamoto et al., 2007; Fernandez et al., 2014; Ruehle et al., 2015). The experiments by Ruehle et al. have been done by colocalization using fluorescence. The discrepancy should be discussed. Furthermore, the authors should perform experiments to establish the nature of the complex in a more direct manner, e.g. by toeprinting of the complex.

The reviewer makes two suggestions for revising the manuscript in this point, and we disagree with both.

There is no conflict with the Ruehle et al. paper, on which we are co-authors. The colocalization result is reported in Figure 6 of Ruehle et al. and utilizes single molecule observation. The relevant experimental section describing this experiment in the Ruehle et al. paper reads as follows:

“The 80S-IRES ribosome complex was incubated in the flowcell for 5 min and components that remained untethered to the surface of the microfluidic flowcell at the conclusion of the 5 min were washed out of the flowcell using an imaging buffer…”.

This washing procedure has the effect of removing the labile A-site bound Phe-tRNAPhe, accounting for the much lower stoichiometry observed in the absence of added eEF2. This point is addressed directly in Ruehle et al. in the section entitled “Loop 3 facilitates eEF2’s ability to translocate ac-tRNA on IGR IRES-80S ribosome complexes”:

“To examine eEF1A-dependent ac-tRNA delivery, we assembled TC with Phe353

tRNAPhe(Cy5)+eEF1A+GTP and delivered this to the immobilized IRES-80S complexes without eEF2. Compared to the reactions lacking eEF1A, both IRESs show increased and similar ac-tRNA occupancies (WT: 17.9 ± 4.8%, Δ3: 20.8 ± 5.4%). These data initially seem at odds with the anisotropy data in which eEF2-independent ac-tRNA association with 80S-WT IRES ribosome complexes is much greater than complexes with Δ3. This apparent discrepancy is likely due to the fact that anisotropy data are obtained under equilibrium conditions where transient interactions are observed, whereas the single-molecule fluorescence data are collected after the flowcell is flushed and thus only show stable long-lived association.”

The second suggestion requests that we clarify “the nature of the complex” through use of a toeprinting experiment. We are not completely clear to which complex the Reviewer is referring: Structure 2 with the A-site empty in either the absence or presence of eEF2; or Structure 3, with aa-tRNA bound in a labile manner in the A-site. Of these three possibilities, we consider two to be noncontroversial. In the absence of eEF2 and Phe-TC, our results suggest that the equilibrium between Structures 1 and 2 strongly favors 1, in agreementwith the results of others. With respect to Structure 3, the results presented in the Ruehle et al. paper show the tRNA binding to be enhanced for cognate vs. non-cognate tRNA, supporting the structure shown.

Where our results may appear to be at odds with published results, likely prompting the Reviewer’s suggestion, are for Structure 2 in the presence of added eEF2·GTP. We address this point directly in the revised text (Discussion, third paragraph). The basic conclusion is that the reaction conditions employed in the toeprinting assay make it an unreliable method for measuring eEF2·GTP effects on the 1 to 2equilibrium position, in agreement with an earlier suggestion of Muhs et al. (2015) (subsection “Di-, Tri- and Tetrapeptide.”), which has recently been supported by results of Petrov et al. (2016). Because of this, we do not agree with the reviewer that a toeprinting experiment is likely “to establish the nature of the complex in a more direct manner”.

2) The authors state that k-1 is much smaller than k2 and was assumed to be negligible. How has this been determined? In fact, a toeprint signal indicative of translocation cannot be obtained by incubation of binary 80S-IRES complexes with eEF2 alone, but requires an A-site ligand, too (Jan et al., 2003; Muhs et al., 2015). This suggest that the translocated complex (structure 2) is kinetically labile and contradicts the present statement that the back-translocation is negligible. Again, the authors should do experiments, e.g. toeprinting, to provide additional evidence that pre-incubation with eEF2-GTP shifts the equilibrium between structures 1 and 2 from 95:5 to 50:50 as stated.

3) The explanation about the effect of eEF2 on the equilibrium between structure 1 and 2 (Discussion) is curious. They propose that eEF2 shifts the equilibrium towards structure 2 at the same time do not measure a significant effect on the t1/2 of step 1. How this is possible? Do the authors propose that k-1 is becoming smaller, i.e. that eEF2 slows down back-translocation without accelerating forward translocation?

These two valid points are linked, and we thank the reviewer for raising them. To address them we fit the anisotropy results presented in Figure 3 to the Scheme presented in Figure 2 for Steps 1 and 2, yielding the apparent kinetic constants collected in new Table 1. The text has been modified (subsection “Rates of Phe-TC binding to the 80S·IRES complex: Steps 1-3, structures 1 – 4”, second paragraph) to include these results. The values in Table 1 show that, as expected, in the absence of eEF2·GTP, k-1 >> k1, whereas in the presence of eEF2·GTP, k-1 ~ k1. This change results from the effects of eEF2·GTP in strongly decreasing k-1 (~ 50-fold) while only modestly decreasing k1 (~ 2-fold). The rate constants in Table 1 are then used to calculate t1/2 values for Steps 1 and 2 in Table 2, as described in footnotes b and c.

4) Can the authors rule out that back-translocation or peptidyl-tRNA drop off influences the following steps towards tetrapeptide synthesis?

We do not think that peptidyl-tRNA drop off is a major issue, because the yield of peptide per ribosome is constant for di-, tri- and tetrapeptide. However, we acknowledge that many of the details of the process, including possible back translocation during oligopeptide synthesis, remain to be elucidated.

Additional points:

5) The first sentence of the paper "Initiation of protein synthesis in eukaryotic cells proceeds via two well-established pathways" is to some extent misleading. As the authors acknowledge later factor requirement of different IRES RNAs is diverse and the structure and mechanism of different IRES RNAs is also different. So there are several pathways leading to internal initiation.

We disagree with the reviewer. Our division of initiation of protein synthesis via two pathways is a common way of introducing the subject and follows the presentation in a review article we cite (Jackson et al., 2010). The qualification that IRES-initiated synthesis can take place in diverse ways comes three lines later in the opening paragraph, so we are hardly neglecting this point.

6) The peptide coding sequence is not really part of the IRES. Therefore, it is misleading to talk about mutant IRES, when the changes are in the ORF.

We accept this criticism and have made the appropriate changes in the Results section and in Supplementary file 1.

7) At the beginning of the Results part the authors should specify in which system they are working.

The requested change has been made in revising the first sentence of Results.

8) The green line is not mentioned in the legend to Figure 3.

This criticism is addressed in the revised last sentence of the Figure 3 legend.

9) A-site/E-site allostery has been recently reported by Ferguson et al., 2015, Mol Cell for human 80S ribosomes. This is relevant for the respective discussion in the third paragraph of the subsection “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”.

We agree. We only became aware of the Ferguson et al. article (published November 2015) after we had submitted our initial MS to eLife. It is relevant to our manuscript with respect both to A-site:E-site allostery (see the response to Reviewer #2, point 3) and puromycin reactivity. The manuscript has been altered to include references to Ferguson et al. (2015) in the subsections: “Rates of oligopeptide formation and Met-tRNAMet cosedimentation”, last paragraph; “Translocation of tetrapeptidyl-tRNA (Step 12) is much more rapid than of tripeptidyl-tRNA (Step 9), third paragraph and Discussion, fourth paragraph.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues raised by Reviewer #3 that need to be addressed before acceptance. In considering these new comments, Reviewer #2 also asks that you carefully review the Materials and methods to insure that they contain all of the critical information that would be required to duplicate the work.

Reviewer 2 raises no specific objections to the previous submission (Revision 1) but we note that some experimental detail has been added to the Figure 3 legend in response to Reviewer 3 that might address the concern expressed in your overview relevant to “critical information.” Our point-by-point response to Reviewer 3 is given below.

Reviewer #3:

In the revised version Zhang and colleagues have addressed many points raised by the reviewers and have improved their paper. However, there are still some points where discussion should be extended.

1) The authors now confirm that the effect of eEF2 is to inhibit k-1 and write that this is "consistent with its role as a translocase". However, the accepted role of EF-G/eEF2 is to accelerate translocation, not to inhibit back-translocation. This warrants additional discussion.

We strongly disagree with the reviewer on this point. In fact, a principal role of EF-G, the prokaryotic equivalent of eEF2, has long been considered to be to inhibit back-translocation (Savelsbergh et al., Mol Cell. 2003 11:1517-23; Ratje et al., Nature. 2010 468:713-6), a view backed up by recent experiments. A reference to a particularly compelling set of results (Adio et al., 2015) has been added to the text to make this point explicit (subsection “Rates of Phe-TC binding to the 80S·IRES complex: Steps 1-3, structures 1 – 4”, end of second paragraph).

2) The authors have pre-incubated 80S-FVKM-IRES complexes with eEF2-GTP for 1 – 2 hr. How they can rule out that GTP consumption has an impact on the experiment? Is it possible that accumulation of eEF2-GDP during pre-incubation results in a significant fraction of 80S FVKM-IRES-eEF2 complexes (between Structures 1 and 2) to slow down k2.

We understand the reviewer’s concern, and have added additional text to make it clearer that we have adequately addressed this concern. As stated in the experimental section (subsection “Kinetic Experiments”) “Unless otherwise noted, all reactions were performed at 37 oC in buffer 4 supplemented with 1 mM GTP”. To eliminate any ambiguity, we have made two additions to the Figure 3 legend. First, we added the phrase “containing 1 mM GTP”. Second, we added the sentence “eEF2 displays virtually no GTPase activity when it is not bound to the ribosome (Nygård and Nilsson, 1989)”. This makes it clear to the reader that the eEF2·GTP added as part of the Phe-TC solution will not undergo hydrolysis prior to its rapid mixing with 80S·IRES. Lastly, as mentioned in the text (Discussion, third paragraph), GTP is required for tight binding of eEF2 to the ribosome, so that eEF2·GTP should easily outcompete eEF2·GDP generated during the preincubation step for binding to 80S·IRES.

3) The authors write that the anisotropy assay in Figure 3 reports on formation of Structure 3 from 1. However, as the co-sedimentation assay reports, during the measurement time there should be also formation of Structure 4 at later time.

The reviewer is correct, but as Phe-TC is bound to the 80S·IRES complex in both Structures 3 and 4, any anisotropy change accompanying Structure 4 formation from Structure 3 is expected to be minor.

4) The authors state that A-site-bound Phe-tRNAPhe is labile. Can they estimate the off-rate k-2. Is it valid to neglect k-2 for numerical integration?

Productive TC binding is accompanied by GTP hydrolysis, an essentially irreversible step. Thus, the lability of bound Phe-tRNAPhe would not be measured by k-2, but by a separate reaction following dissociation of eEF1A·GDP from the ribosome, the kinetics of which we have not measured. However, this question by the reviewer prompted us to add a second sentence to the Figure 2 legend, explaining that Scheme 1 is simplified and neglects many substeps in the elongation process. As to the validity of neglecting the actual k-2, which describes the reversible binding of ternary complex, the answer is that it is always valid to use the minimum number of parameters to fit the available data. This results in obtaining apparent kinetic constants, as listed in Table 1. Different experimental results, which are outside the scope of this manuscript, could allow k-2 to be estimated, with, perhaps, altered apparent values of k1, k-1, and k2.

https://doi.org/10.7554/eLife.13429.018

Article and author information

Author details

  1. Haibo Zhang

    Department of Chemistry, University of Pennsylvania, Philadelphia, United States
    Present address
    Spark Therapeutics, Philadelphia, United States
    Contribution
    HZ, Designed experiments and acquired, Analyzed and interpreted data
    Competing interests
    The authors declare that no competing interests exist.
  2. Martin Y Ng

    Department of Chemistry, University of Pennsylvania, Philadelphia, United States
    Contribution
    MYN, Acquired, analyzed and interpreted data
    Competing interests
    The authors declare that no competing interests exist.
  3. Yuanwei Chen

    Department of Chemistry, University of Pennsylvania, Philadelphia, United States
    Present address
    Hacker Way, Menlo Park, United Stetas
    Contribution
    YC, Acquired, analyzed and interpreted data
    Competing interests
    The authors declare that no competing interests exist.
  4. Barry S Cooperman

    Department of Chemistry, University of Pennsylvania, Philadelphia, United States
    Contribution
    BSC, Conceived and designed the experiments, Interpreted results, Wrote the manuscript
    For correspondence
    cooprman@pobox.upenn.edu
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6989-9788

Funding

National Institutes of Health (RO1GM 080376)

  • Haibo Zhang
  • Martin Y Ng
  • Yuanwei Chen
  • Barry S Cooperman

The funder had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Eric Jan for gifts of wt and variant CrPV Phe-IRES vectors and for helpful discussions.

Reviewing Editor

  1. Alan G Hinnebusch, National Institute of Child Health and Human Development, United States

Publication history

  1. Received: December 1, 2015
  2. Accepted: June 1, 2016
  3. Accepted Manuscript published: June 2, 2016 (version 1)
  4. Version of Record published: July 27, 2016 (version 2)

Copyright

© 2016, Zhang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 935
    Page views
  • 338
    Downloads
  • 12
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Biochemistry and Chemical Biology
    2. Structural Biology and Molecular Biophysics
    Nicholas Chim et al.
    Short Report
    1. Biochemistry and Chemical Biology
    2. Structural Biology and Molecular Biophysics
    Erica CIrri et al.
    Research Article