Author response: Distinct stages of the translation elongation cycle revealed by sequencing ribosome-protected mRNA fragmentsArticle Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract During translation elongation, the ribosome ratchets along its mRNA template, incorporating each new amino acid and translocating from one codon to the next. The elongation cycle requires dramatic structural rearrangements of the ribosome. We show here that deep sequencing of ribosome-protected mRNA fragments reveals not only the position of each ribosome but also, unexpectedly, its particular stage of the elongation cycle. Sequencing reveals two distinct populations of ribosome footprints, 28–30 nucleotides and 20–22 nucleotides long, representing translating ribosomes in distinct states, differentially stabilized by specific elongation inhibitors. We find that the balance of small and large footprints varies by codon and is correlated with translation speed. The ability to visualize conformational changes in the ribosome during elongation, at single-codon resolution, provides a new way to study the detailed kinetics of translation and a new probe with which to identify the factors that affect each step in the elongation cycle. https://doi.org/10.7554/eLife.01257.001 eLife digest To make a protein from a gene, the gene is first transcribed to produce a molecule of messenger RNA (mRNA), which then passes through a molecular machine called a ribosome. The ribosome reads the genetic code in the mRNA in groups of three letters at a time, and each triplet of letters (or codon) represents an amino acid. The ribosome then joins the relevant amino acids together to build a protein. The ribosome processes about six amino acids per second, on average, but the mRNA is not fed through at a constant rate. Instead, the ribosome changes its shape to ratchet along the mRNA from one codon to the next: it then reads the new codon and adds another amino acid to the protein. However, many of the details of this ratcheting process are not fully understood. In this study, Lareau, Hite et al. have used a technique called 'ribosome profiling' to explore the movement of ribosomes along mRNA molecules. First, all of the pieces of mRNA molecules that are not protected inside a ribosome were chemically destroyed. The sequences of the protected fragments were then read and matched to the full-length gene sequences. The protected fragments came in two different sizes: some were about 28–30 letters long, and others were about 20–22 letters long. Lareau, Hite et al. suggest that these different fragment sizes occur because the ribosome switches between two shapes at each codon as it ratchets along the mRNA, and so it protects different lengths of mRNA. In previous ribosome-profiling experiments, the fragments had all been about 28 letters long; but these experiments had used a chemical to halt the progress of the ribosomes along the mRNAs before measuring the length of the fragments. Lareau, Hite et al. argue that this chemical locks the ribosome in the same shape when it brings the ribosome to a halt, and so the protected fragments always have the same length. Further, other chemicals that halt ribosomes appear to lock this molecular machine in the other shape, and so it can only protect the shorter fragments. The findings of Lareau, Hite et al. show that ribosomal profiling experiments can reveal much more than simply where a ribosome is on an mRNA molecule. Further study into the different stages of the ribosome ratcheting process will help uncover how the speed that a ribosome translates an mRNA into a protein can be encoded in the mRNA sequence itself. https://doi.org/10.7554/eLife.01257.002 Introduction To accomplish the huge task of translation elongation—in each cycle, accurately incorporating a new amino acid into a nascent peptide every 1/6th of a second, then moving precisely three nucleotides along the mRNA template—the ribosome undergoes a series of major structural rearrangements (Figure 1) (reviewed in Chen et al., 2012 and Noeske and Cate, 2012). During the initial decoding step of elongation, aminoacylated tRNAs are delivered to the decoding site (A site) as part of a ternary complex with EF-Tu (in prokaryotes) or the orthologous eEF1A (in eukaryotes). When the anticodon of one of these aminoacylated tRNAs is able to base-pair stably with the specific mRNA codon in the decoding site (A site), a new peptide bond is formed between the nascent polypeptide and the specified amino acid. The ribosome then undergoes a massive rearrangement in which the ribosomal subunits rotate relative to each other (Frank and Agrawal, 2000; Zhang et al., 2009). Along with this rotation, the A and P site tRNAs move from 'classic' to 'hybrid' states: the anticodon ends stay in their original A and P sites and the acceptor ends move to the P and E sites (Moazed and Noller, 1989; Munro et al., 2007). This rotated state of the ribosome undergoes additional conformational changes in preparation for translocation (Zhang et al., 2009; Fu et al., 2011). The ribosome can fluctuate between rotated and non-rotated states until EF-G (eEF2 in eukaryotes) binds and stabilizes the rotated ribosome (Agirrezabala et al., 2008). GTP hydrolysis by EF-G then promotes translocation of the mRNA along the ribosome, coupled to a large intra-subunit rotation of the 30S head (Ratje et al., 2010), after which the ribosome subunits rotate back to a closed formation for the next cycle (Gao et al., 2009). Structural and biochemical studies have revealed many of the atomic-level changes that allow this complicated process to occur (Pulk and Cate, 2013; Tourigny et al., 2013; Zhou et al., 2013), and new details continue to emerge, reshaping models, raising new questions, and leaving other questions still unanswered. Figure 1 Download asset Open asset Schematic representation of the eukaryotic elongation cycle. Blue overlay denotes stages at which the ribosome has undergone a large inter-subunit rotation. Ribosome shapes are for illustration only, not a literal representation of the structure or degree of rotation. https://doi.org/10.7554/eLife.01257.003 Recently, 'ribosome profiling' by high-throughput sequencing of ribosome-protected fragments has provided a powerful tool for identifying the position of ribosomes on mRNAs across the entire transcriptome (Ingolia et al., 2009). Cell lysates are treated with nuclease to degrade all mRNA not physically protected by ribosomes, and the ribosome-protected fragments are extracted, sequenced, and mapped back to the genome to show ribosome positions, revealing the overall translation level of each gene as well as the distribution of ribosomes along the mRNA. Nucleotide-level precision of ribosome positions is possible because of the very consistent size of ribosome footprints in the conditions assayed. The authors of the method used a nuclease protection assay to establish that, in yeast treated with the elongation inhibitor cycloheximide, each ribosome protects a footprint of 28 nucleotides (nt), confirming earlier reports (Steitz, 1969; Wolin and Walter, 1988). While performing ribosome-profiling experiments in Saccharomyces cerevisiae, we serendipitously noticed a population of smaller ribosome-protected fragments. To better capture these fragments and to investigate their origins, we revised the ribosome-profiling protocol originally established by Ingolia et al. Our experiments revealed that, in the absence of cycloheximide, the small ribosome-protected fragments were abundant, consistent with an early observation of short ribosome footprints in the absence of cycloheximide (Wolin and Walter, 1988). We show here that the small fragments originate from ribosomes in a conformation distinct from that previously observed in the presence of cycloheximide. The ability to discern distinct ribosomal structural states by ribosome profiling has given us insight into how codon, tRNA, and amino acid identity and translational speed relate to ribosome structure. This additional dimension of ribosome-profiling data will provide a valuable new layer of molecular and mechanistic information, at codon resolution, for future studies of translation. Results Ribosomes can protect two distinct mRNA fragment sizes We began our investigation of ribosome footprint size by isolating ribosome-protected mRNA fragments from yeast using a modified ribosome-profiling procedure. The standard ribosome-profiling protocol includes a size selection for RNA fragments of around 28 nt. To eliminate the bias against smaller fragments, we broadened the initial size range and selected RNA fragments between 18 and 32 nt after RNase I digestion. By selecting fragments in this broader size range, and, importantly, by carrying out the entire procedure in the absence of cycloheximide or other inhibitors, we observed two clearly distinct, abundant populations of ribosome-protected mRNA fragments ('footprints'), 28–30 nt and 20–22 nt long. We visualized fragment lengths and positions with a three-dimensional 'metagene' representation: sequence reads representing the ribosome-protected fragments from all expressed genes were aligned relative to the start codon of the corresponding gene and tallied by fragment length and position to show the average pattern of translation along all annotated coding regions (Figure 2A–C, Figure 2—figure supplement 1). Figure 2 with 1 supplement see all Download asset Open asset Ribosome-protected fragment positions and size distributions from yeast not treated with elongation inhibitors. (A) The position of each fragment was calculated relative to the start codon of its gene. The 5′ end positions (x axis) and lengths of all fragments (y axis) were tallied across all genes with a coding region of at least 300 nt. Higher color intensity reflects more fragments. RNA fragments between 18 and 32 nucleotides were selected after gel electrophoresis; shorter and longer fragments are not entirely excluded but their read counts are presumed to be unrepresentative of their true abundance. (B) Profiles of the 5′ end positions of all 20 nt and 28 nt fragments relative to the start codon of their genes, as in (A). (C) Total counts of mapped fragment lengths. (D) Distribution of 21 nt and 28 nt fragments in coding regions and untranslated regions of mRNAs. (E) Positions of 21 nt and 28 nt fragments relative to the reading frame. (F) Interpretation of fragment positions on an arbitrary gene fragment. Arrowheads show hypothetical nuclease cleavage sites relative to a ribosome in a non-rotated or rotated conformation (shape is for illustration only). The resulting fragments are shown with the inferred decoding site (A site), and their positions in a grid as in Figure 2A are shown with corresponding colors. https://doi.org/10.7554/eLife.01257.004 We found overwhelming evidence that both populations of fragments came from translating ribosomes. The 21 and 28 nt fragments were both found almost entirely within annotated coding regions (CDS) and not in 5′ or 3′ UTRs; 98.3–99.7% of mappable 21 nt fragments, and 96.5–99.6% of mappable 28 nt fragments, mapped within the annotated CDS in three replicates (Figure 2D). Both populations also showed the 3-nucleotide periodicity expected of fragments originating from elongating ribosomes (Figure 2E). We conclude that fragments of both sizes are footprints of translating ribosomes. The 5′-most peaks in the metagene represent ribosomes with the start codon in the P site and the second codon in the A site (Kapp and Lorsch, 2004; Ingolia et al., 2009). Using this as a reference for phasing all the footprints, we inferred that for ribosomes with a given codon in the A site, small and large footprints generally had the same 5′ ends positioned 15–16 nt upstream of the A-site codon, and differed at their 3′ ends: extending 2–3 nt beyond the A-site codon in the small footprints and 10 nt beyond the A-site codon in the large footprints, respectively (Figure 2F). Different elongation inhibitors stabilize distinct conformations and bias the footprint size distribution During elongation, at each codon, the ribosome cycles through a stereotyped sequence of steps as it incorporates the specified amino acid and translocates to the next codon. These steps are accompanied by major rearrangements of the ribosome structure, including a rotation of the large subunit relative to the small subunit upon peptide bond formation. We hypothesized that the non-rotated, pre-peptide-bond ribosomes and rotated, post-peptide-bond ribosomes might protect different lengths of mRNA, and that the two resulting footprint sizes might, therefore, represent these two conformations. To determine what footprint sizes were protected by ribosomes in distinct stages of elongation, we performed ribosome profiling on yeast treated with inhibitors that block different steps of the cycle. Cycloheximide is an elongation inhibitor that binds to the E site of ribosomes, preventing the E site tRNA from leaving the ribosome. When cycloheximide was added to the yeast immediately before harvest and was present throughout lysis and RNase I treatment, the most prevalent footprints were 28–30 nt long and were distributed along the coding sequence with a 3-nt periodicity (Figure 3A–C, Figure 3—figure supplement 1). Apart from a distinct peak at the start codon, there were very few 20–22 nt footprints. Figure 3 with 1 supplement see all Download asset Open asset Ribosome-protected fragment positions and size distributions from yeast treated with elongation inhibitors. (A and B) As in Figure 2A,B, fragment position and size distribution for yeast treated with cycloheximide. (C) Distribution of mapped fragment lengths for yeast treated with cycloheximide. (D and E) Fragment position and size distribution for yeast treated with anisomycin. (F) Distribution of mapped fragment lengths for yeast treated with anisomycin. https://doi.org/10.7554/eLife.01257.006 Our data confirmed previous evidence that the ribosome predominantly protects a 28 nt footprint in the presence of cycloheximide, and suggest that cycloheximide stabilizes one stage of the elongation cycle. Previous work shows that cycloheximide bound alongside a tRNA in the E site prevents either the incorporation of the next aminoacylated tRNA in the A site or peptide bond formation (Schneider-Poetsch et al., 2010). In either case, it is expected to trap the ribosome in a non-rotated conformation, suggesting that the non-rotated conformation protects 28–30 nt of mRNA. We next conducted ribosome-profiling experiments using yeast treated with anisomycin, an elongation inhibitor that binds to the peptidyl transferase center (Grollman, 1967; Hansen et al., 2003). We observed almost exclusively small footprints in yeast treated with anisomycin (Figure 3D–F, Figure 3—figure supplement 1). By comparison to the effects of cycloheximide treatment, we inferred that anisomycin stabilizes a distinct conformation of the ribosome that protects 20–22 nt of mRNA. Although anisomycin's precise mechanism is not characterized, it has higher affinity for post-translocation ribosomes than for pre-translocation, cycloheximide-treated ribosomes, suggesting that it preferentially binds a ribosome conformation distinct from that stabilized by cycloheximide (Barbacid and Vazquez, 1974, 1975). Lincomycin and other antibiotics that bind the peptidyl transferase center induce translocation, and lincomycin-treated ribosomes prefer a rotated conformation in in vitro FRET experiments (Fredrick and Noller, 2003; Ermolenko et al., 2013). It is possible that anisomycin acts similarly to stabilize a rotated conformation. We have thus demonstrated that two distinct ribosome conformations can be stabilized using elongation inhibitors. Stabilization of distinct conformations by two drugs resulted in a nearly complete reciprocal bias in the size of ribosome footprints, providing evidence that large and small footprints originate from distinct ribosomal conformations. We hypothesize that each ribosome cycles through both conformations, protecting first a large footprint and then a small footprint at each codon. The footprints identified by high-throughput sequencing in a ribosome-profiling experiment represent a deep sampling of ribosomes in different states, and thus the ratio of large to small footprints in untreated cells could show, at single-codon resolution, how many ribosomes are in each stage of elongation. Increased decoding time produces more large footprints To enrich for ribosomes in a single, defined stage of the elongation cycle, we induced conditions expected to result in the depletion of a specific aminoacyl-tRNA and thus to increase the decoding time when the cognate codon is in the A site. We treated yeast with 3-amino-1,2,4-triazole (3-AT), an inhibitor of histidine biosynthesis, to create a specific shortage of His-acylated tRNA and cause ribosomes to pause on histidine codons (Figure 4A). We would therefore expect ribosomes to accumulate at histidine codons in a pre-peptide-bond conformation. Estimating codon-specific occupancy as described in more detail below, we found that the shortage of His-tRNA dramatically increased the relative abundance of large footprints from ribosomes with His codons in the A site, with minimal effect on the abundance of small footprints (Figure 4B,C, Figure 4—figure supplement 1). During the decoding phase of elongation, before peptide bond formation, the ribosome is in a non-rotated conformation (Frank and Agrawal, 2000; Gao et al., 2009); these results therefore strongly suggest that the decoding phase of elongation (the non-rotated conformation) is represented by large footprints. Figure 4 with 1 supplement see all Download asset Open asset Effect of 3-amino 1,4 triazole on translation of histidine codons. (A) Schematic representation of the hypothesized effect of 3-AT. 3-AT reduces intracellular concentrations of histidyl-tRNA and thus is expected to increase time spent decoding histidine codons (i.e., in the decoding phase of the cycle, with a His codon in the A-site). (B) All 61 sense codons are plotted by the log2 of the relative abundance of large footprints with the specified codon in the A-site for untreated cells (x axis) against the log2 relative abundance of large footprints for yeast treated with 3-AT (y axis). Values shown are the average of three untreated replicates and two 3-AT treatments (10 min and 60 min). Histidine codons are denoted in red (CAT) and cyan (CAC). (C) As in (B), showing the relative abundance of small footprints. https://doi.org/10.7554/eLife.01257.008 The footprint size distribution varies by codon Recently, ribosome profiling has revealed that translation speed varies systematically by codon (Tuller et al., 2010; Stadler and Fire, 2011; Dana and Tuller, 2012); we hypothesized that there might be distinct codon-specific effects on the rate of the two distinct phases of elongation represented by small and large footprints. Using data from untreated cells, we calculated the number of large and small footprints corresponding to ribosomes with a given codon in the A site, for each codon position in the yeast transcriptome. Large footprints were defined as 28 or 29 nt and small footprints were defined as 20, 21, or 22 nt with 5′ ends positioned relative to the inferred A site as depicted in Figure 2F. We found substantial variation in the characteristic length distribution between codons: small footprints ranged from 38 ± 12% (UUU) to 87 ± 9% (CGG) of the total footprints for a given codon identity, averaged across three replicates. To explore this codon effect, we computed the relative occupancy of each of the 61 sense codons in the A site. We started by considering an individual gene and calculated the over- or underrepresentation of footprints at each codon position compared to the average for all codon positions in that gene, including both small and large footprints (an example from a highly expressed gene is shown in Figure 5A). After performing this computation for every gene, we averaged these multipliers across all occurrences of a given codon in the genome to provide the 'relative occupancy' for that codon, representing, on a relative scale, how frequently we observed ribosomes with that codon positioned at the A site. The relative occupancies varied over a fivefold range, from 0.48 ± 0.04 (GGU) to 2.6 ± 0.67 (CCG) (unitless, average of three replicates) and were highly correlated between independent replicates (Figure 5B). As a control, we also analyzed the occupancy based on the codon one position 3′ of the A site, which has not yet entered the decoding site. We found that the range of occupancies relative to the codon in the A site was much broader than the range of occupancies relative to the next codon, suggesting that the A-site occupancies reflect an aspect of translation, not merely confounding factors such as biases in fragment capture (Figure 5—figure supplement 1). Figure 5 with 1 supplement see all Download asset Open asset Codon-specific variation in large and small footprint abundance. (A) Distribution of ribosome footprint counts on the highly expressed gene an arbitrary codons Ribosome footprint counts per position were consistent between replicates and varied between of the same codon in this occupancy was based on the codon in the inferred A site. Total footprint at each codon of a gene was computed relative to the average for that gene, then averaged by codon across all genes to provide relative abundance of small or large footprints was computed the of small or large footprints at each codon of a gene against the average for that gene, then averaged by codon across all of small and large footprint abundance at two specific codons in are (B) occupancies of all 61 codons compared between two with of codons and the first codons of each gene were excluded from small footprint abundance (C) and large footprint abundance (D) compared between replicates. Codon-specific in ribosome occupancy could have been by variation in small footprint variation in large footprint or revealing the of each stage of elongation. We inferred the relative abundance of ribosomes in each state at each codon using a to the one we used to overall relative but considering counts of either small or large footprints (Figure 5A). As with overall the relative of small footprints and the relative abundance of large footprints were both highly correlated between replicates (Figure This that codon identity both the pre-peptide-bond and post-peptide-bond stages of elongation. However, the effect of codon identity on the inferred of these two phases of the elongation cycle was the codon-specific relative of small and large footprints were almost average of three This us to for of the codon-specific occupancy is to amino acid and We found that a major and of the abundance of footprints from each conformation was the identity of the amino acid encoded by the A-site codon. We found a much of small footprints at codons amino acids than at codons amino The relative abundance of small footprints at codons a given amino acid was correlated with of of the cognate amino such as the of of from to when by amino by codon, Figure the relative abundance of large footprints showed to amino acid by amino by codon) 2007). These data strongly suggest that the chemical of the amino acid specified by the codon in the A site affect the of the rotated, post-peptide-bond conformation of the ribosome. We hypothesize that between the ribosome and amino acids to the A-site tRNA can translocation Figure Download asset Open asset of footprint abundance. (A) footprint averaged for all codons the same amino acid plotted against of of from to as a of with from the average of three (B) occupancy of codons relative occupancy of codons that the same tRNA with Values are the average of three replicates. shows the expected occupancy were by tRNA and As in (B), showing small and large footprint abundance. factors have been to affect translation speed at a given codon, tRNA abundance. In the number of genes a specific tRNA has been shown to be highly correlated with both codon and tRNA concentrations et al., A of codon is the tRNA which to codons in translational by for tRNA and codon et al., We found that the relative occupancy per codon was only correlated with and with tRNA number and average of three replicates) and that the was not correlated with the relative abundance of either small footprints or large footprints and average of three unexpectedly, codon as represented by the not appear to be a major of relative ribosome occupancy the conditions The 3-AT data show that in an case, a of the tRNA cognate to the A-site codon translation during the In our overall results in untreated yeast suggest that the in abundance tRNAs in cells have only a effect on relative ribosome occupancy of the cognate codons We also the between relative and the of large and small footprints. at the A site has been with elongation in and and Fire, 2011). We compared codons with the codons that with the same tRNA et al., 2008). While we found consistent increased occupancy at we observed higher occupancy on a of codons and (Figure these three codon we see a dramatic increase in short footprints, representing stages of translation (Figure The codon is to be a inhibitor of translation in and its effect is more to decoding than tRNA abundance and after the initial decoding et al., 2010). Our data that is one of the most and its relative occupancy is to increased abundance of small footprints, suggesting that its elongation is to a the abundance of footprints from each step of elongation was clearly by distinct codon-specific with and Discussion A ribosome cycle through a series of with mRNA to the one codon at a The of the one to precisely where ribosomes on to the codon isolating and sequencing ribosome-protected mRNA fragments. We were to that the ribosome protects two different footprint sizes nt and 20–22 as the original ribosome-profiling experiments and nuclease protection only the longer footprints (Ingolia et al., 2009). The is by the the small footprints were revealed only after we out cycloheximide, a translation inhibitor used to stabilize ribosomes on mRNA for ribosome early study of ribosome found that when cycloheximide was nt footprints in to the footprints from cycloheximide-treated ribosomes (Wolin and Walter, 1988). As in our experiments, the small and large footprints observed had the same 5′ and differed at the 3′ We that the two footprints sizes originate from two ribosome conformations corresponding to different stages of large footprints from