Spontaneous Intersubunit Rotation in Single Ribosomes

Transcription

1 Article Spontaneous Intersubunit Rotation in Single Ribosomes Peter V. Cornish, 1,4 Dmitri N. Ermolenko, 3,4 Harry F. Noller, 3, * and Taekjip Ha 1,2, * 1 Department of Physics 2 Howard Hughes Medical Institute University of Illinois, 1110 West Green Street, Urbana, IL 61801, USA 3 Department of Molecular, Cell, and Developmental Biology and Center for Molecular Biology of RNA, University of California, Santa Cruz, Santa Cruz, CA 95064, USA 4 These authors contributed equally to this work *Correspondence: harry@nuvolari.ucsc.edu (H.F.N.), tjha@uiuc.edu (T.H.) DOI /j.molcel SUMMARY During the elongation cycle, trna and mrna undergo coupled translocation through the ribosome catalyzed by elongation factor G (EF-G). Cryo-EM reconstructions of certain EF-G-containing complexes led to the proposal that the mechanism of translocation involves rotational movement between the two ribosomal subunits. Here, using single-molecule FRET, we observe that pretranslocation ribosomes undergo spontaneous intersubunit rotational movement in the absence of EF-G, fluctuating between two conformations corresponding to the classical and hybrid states of the translocational cycle. In contrast, posttranslocation ribosomes are fixed predominantly in the classical, nonrotated state. Movement of the acceptor stem of deacylated trna into the 50S E site and EF-G binding to the ribosome both contribute to stabilization of the rotated, hybrid state. Furthermore, the acylation state of P site trna has a dramatic effect on the frequency of intersubunit rotation. Our results provide direct evidence that the intersubunit rotation that underlies ribosomal translocation is thermally driven. INTRODUCTION Protein synthesis is a dynamic process carried out by the ribosome, an RNA-based molecular machine. During protein synthesis, trna and mrna are translocated through the ribosome in a series of complex, large-scale molecular movements catalyzed by elongation factor G (EF-G) and GTP. However, translocation can occur, albeit very slowly, in the absence of EF-G and GTP (Cukras et al., 2003; Fredrick and Noller, 2003; Gavrilova et al., 1976; Gavrilova and Spirin, 1971; Pestka, 1969). Thus, translocation is a property of the ribosome itself, rather than of EF-G, and is thermodynamically favored even in the absence of GTP hydrolysis. Chemical probing studies provided the first direct evidence that translocation takes place in two steps involving an intermediate hybrid state (Moazed and Noller, 1989b). In the first step, the acceptor ends of the trnas move relative to the 50S subunit, from their classical A/A- and P/P-binding states into hybrid A/P and P/E states (in which the peptidyl-trna is bound in the 30S A site and the 50S P site and the deacylated trna is bound in the 30S P site and the 50S E site; Figure 1A). The specific affinity of the acceptor end of deacylated trna for the 50S E site (Lill et al., 1986) helps to account for the thermodynamic driving force for spontaneous formation of the hybrid state. In the second step, which strongly depends on participation of EF-G, their anticodon ends move on the 30S subunit, coupled with mrna movement, into the posttranslocational P/P and E/E states. Cryo-EM studies have identified a conformation of the ribosome in which the 30S subunit is rotated by about 3 10 counterclockwise relative to the 50S subunit in complexes containing EF-G$GDPNP (a nonhydrolyzable analog of GTP) or EF-G$GDP$fusidic acid (Frank and Agrawal, 2000; Gao et al., 2003; Valle et al., 2003). This finding led to the proposal of a ratchet-like mechanism, in which translocation of trna and mrna is linked to intersubunit rotational movement (Frank and Agrawal, 2000; Frank et al., 2007; Tama et al., 2003; Valle et al., 2003). Recently, this model has been directly tested by formation of a disulfide bridge between ribosomal proteins S6 and L2 designed to restrict intersubunit movement, resulting in a specific block in translocation (Horan and Noller, 2007). The hybrid-state and ratchet models have now converged. Recent bulk FRET measurements combined with chemical probing experiments show that the EF-G-induced rotation of the 30S subunit observed in cryo-em reconstructions corresponds to formation of the hybrid state characterized by chemical probing studies (Ermolenko et al., 2007a, 2007b). Although EF-G binding was found to stabilize the rotated, hybrid state (Spiegel et al., 2007), rotation of the 30S subunit was also observed in the absence of EF-G under conditions favoring the hybrid state (Ermolenko et al., 2007a), consistent with previous biochemical experiments with pretranslocation complexes (Sharma et al., 2004). Furthermore, spontaneous movement of two fluorescently labeled trnas relative to each other, interpreted as movement of the trnas between the classical and hybrid states, was observed in individual pretranslocation ribosomes using singlemolecule FRET (smfret) (Blanchard et al., 2004b; Kim et al., 2007; Munro et al., 2007). 578 Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc.

2 Figure 1. Experimental Design (A) Cartoon showing the movement of deacylated trna fmet (initially in the P site) and peptidyl-trna analog N-Ac-Phe-tRNA Phe (initially in A site) during translocation between classical pretranslocation, hybrid pretranslocation, and classical posttranslocation states. (B) Positions of fluorescent dyes (orange spheres) coupled to proteins L9, S6, and S11 in the 70S ribosome, viewed from the E site interface side of the crystal structure (Korostelev et al., 2006). The 50S subunit is on the left (23S and 5S rrnas are in gray, proteins in magenta), and the 30S subunit is on the right (16S rrna in cyan, proteins in blue). The E site trna (red) can be seen spanning the interface. The red arrows indicate the direction of intersubunit rotation accompanying hybrid-state formation. (C) Ribosomes were immobilized by hybridization of the 3 0 tail of the mrna to a biotin-derivatized DNA strand that was bound via neutravidin to a quartz coverslip. Although the above-mentioned evidence points to the role of ribosome structural dynamics in translocation, the underlying molecular mechanism of this process remains elusive. Intersubunit movements inferred from cryo-em and static bulk FRET experiments have been performed at equilibrium and on the ensemble level and have yet to be observed in real time; moreover, there is so far no thermodynamic and kinetic description of ribosomal intersubunit movement. Finally, the proposal, based on cryo-em (Frank and Agrawal, 2000; Gao et al., 2004) and FRET studies (Munro et al., 2007; Pan et al., 2007), that ribosomal subunits may occupy more than one intermediate conformational state has yet to be established. Here, we address these questions directly using smfret (Ha et al., 1996) and total internal reflection microscopy (Zhuang et al., 2000). This method has been used previously to probe trna dynamics during and after trna accommodation (Blanchard et al., 2004a, 2004b; Gonzalez et al., 2007; Kim et al., 2007; Lee et al., 2007; Munro et al., 2007) and EF-G dynamics (Wang et al., 2007) on the ribosome. In our experiments, using fluorescently labeled ribosomal subunits, we use this approach to directly monitor the dynamics of the ribosome itself. We observe the hypothesized ratchet-like motions of individual ribosomes and characterize the determining factors of their dynamics. The ability of ribosomes to undergo spontaneous intersubunit rotation in the absence of EF-G or GTP has strong implications for the molecular mechanism of translocation. RESULTS Intersubunit Movement in Individual Pretranslocation Ribosomes We conjugated fluorescent labels to specific cysteine residues introduced by directed mutagenesis into ribosomal proteins S6 (D41C), S11 (E75C), and L9 (N11C) (Figure 1B) (Ermolenko et al., 2007a, 2007b). Proteins S6 or S11 labeled with acceptor (Cy5) dye and protein L9 labeled with donor (Cy3) dye were incorporated into 30S and 50S subunits, respectively, by in vitro reconstitution as described previously (Ermolenko et al., 2007a). The labeled subunits were then associated to create two kinds of doubly labeled 70S ribosomes, 70S:S6(Cy5)/L9(Cy3) and 70S:S11(Cy5)/L9(Cy3). In vitro assays showed that at least 50% 60% of purified reconstituted, labeled ribosomes were active in in vitro translocation (Ermolenko et al., 2007a) and 80% 100% active in in vitro translation of a defined mrna (L. Lancaster, personal communication). In order to study the intrinsic structural dynamics of the pretranslocation ribosome, the peptidyl-trna analog N-Ac-Phe-tRNA Phe was bound to the A site of ribosomes containing deacylated trna fmet bound to the P site in the presence of a defined mrna. Pretranslocation ribosome complexes were then immobilized in polymer-passivated microscope slide/coverslip chambers via a biotin-derivatized DNA oligonucleotide annealed to the mrna (Figure 1C) (Blanchard et al., 2004b) and were visualized using total internal reflection microscopy (Zhuang et al., 2000). This immobilization approach preserved the ribosome s translocation activity (see below). Time traces of individual S6-Cy5/L9-Cy3 pretranslocation complexes showed spontaneous, time-dependent anticorrelated changes in donor (Cy3) and acceptor (Cy5) fluorescence intensities (Figure 2A). Calculation of apparent FRET efficiency (FRET = I Cy5 /[I Cy5 +I Cy3 ]) from donor (I Cy3 ) and acceptor (I Cy5 ) intensities revealed that pretranslocation ribosomes fluctuate between high (0.56) and low (0.40) FRET states. smfret measurements performed with the Cy3 and Cy5 dyes reversed gave similar results (data not shown). Time traces recorded for S11-Cy5/L9-Cy3 ribosomes show a similar pattern of spontaneous fluctuations but inverted from that of the S6/L9 construct (data not shown), because S11 moves closer to L9 in the hybrid state, whereas S6 moves away from L9 (Ermolenko et al., 2007a). Below, we present only data from the S6/L9 construct, because of the previously demonstrated strong anticorrelation between the FRET changes for the S6/L9 and S11/L9 dye pairs (Ermolenko et al., 2007a). The high (0.56) FRET state for the S6/L9 pair corresponds to the nonrotated conformation of the ribosome, in which the trnas are bound in the classical state, whereas the low (0.4) FRET state corresponds to the conformation in which the small ribosomal subunit is rotated and the trnas are bound in Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc. 579

3 Figure 2. HMM Analysis of FRET Data Obtained from S6-Cy5/L9-Cy3 Pretranslocation Ribosomes (A) Representative trace showing fluorescence intensities observed for the Cy3 donor (green) attached to L9 and a Cy5 acceptor attached to S6 (red) in ribosomes containing trna fmet in the P site and N-Ac-Phe-tRNA Phe in the A site. (B) Schematic showing the observed FRET values for the two states and the forward and reverse transition frequencies (k 1 and k 1 ). (C) Transition density plot (TDP) for the pretranslocation complex. The TDP is constructed by plotting values for each transition based upon the FRET value from which the transition originated (x axis) and to which FRET value the transition ends (y axis). The transition paths are indicated by the broken red arrows. the A/P and P/E hybrid states (Ermolenko et al., 2007a, 2007b). The latter conformation is equivalent to the ratcheted state observed in cryo-em studies of complexes of the ribosome bound with EF-G (Frank and Agrawal, 2000; Valle et al., 2003). Thus, our single-ribosome traces show that the pretranslocation ribosome, in the absence of EF-G or GTP, fluctuates spontaneously between the rotated and nonrotated conformations, corresponding to the hybrid and classical states, respectively. To ask whether the ribosomal subunits also move through any additional, previously unobserved transient rotational states, the presence of which could be masked by noise in our FRET traces, we subjected our data to a hidden Markov modeling (HMM) algorithm (McKinney et al., 2006). This approach allows for determination of the number of states present in the system and the rates of exchange between them. HMM analysis of the S6-Cy5/L9-Cy3 pre-translocation complex (612 total molecules showing on average 30 transitions per molecule; Table 1) assuming three states (Figures 2C and 3) showed that the pretranslocation complex fluctuates between just two distinct states, nonrotated and rotated, with a forward rotation (nonrotated to rotated) rate of 0.27 ± 0.08 s 1 and a reverse rotation (rotated to nonrotated) rate of 0.19 ± 0.02 s 1 at 25 C under our in vitro conditions (Figures 2B and 2C; Table 1). The same analysis performed on ribosomes containing only a P site trna (trna fmet or trna Met ) also resulted in just two observed FRET states (Table 1 and see Figure S1 available with this article online). In addition, dwell-time analysis for all three complexes fit to a single exponential decay, consistent with two states (Figure S2). Effects of the Acylation State of the P Site trna and EF-G on Intersubunit Rotation We next asked how the acylation state of P site trna and EF-G binding affects the nonrotated/rotated states equilibrium. The equilibrium constant, K eq = ½%rotated Š =½%non rotatedš,is determined from double Gaussian fits to smfret histograms (Dahan et al., 1999) built from several hundred molecules for each construct (Figure 4 and Figure S3; Table 2). A majority (75%) of posttranslocation ribosomes, containing the peptidyltrna analog N-Ac-Phe-tRNA Phe bound in the P site with a vacant A site, exhibited a high FRET value (Figure 4A), corresponding to the classical, nonrotated conformation, in agreement with chemical probing (Moazed and Noller, 1989b) and bulk FRET (Ermolenko et al., 2007a) experiments. In a complex containing a different peptidyl trna, fmet-trna fmet, bound to the P site, 66% of ribosomes were also in the nonrotated conformation (Figure 4B). Likewise, 52% and 79% of authentic posttranslocation complexes obtained by incubation of the pretranslocation complex (the former containing the N-Ac-Phe-tRNA Phe bound to the A site and deacylated trna fmet bound to the P site and the latter containing N-Ac-Phe-tRNA Phe bound to the A site and deacylated trna Tyr bound to the P site) with EF-G$GTP were found in the nonrotated conformation (Figures 4C and 4D). Sixty-one percent of vacant ribosomes (i.e., ones lacking trna) also exhibited a high FRET value (Figure 4K). In contrast to posttranslocation complexes, ribosomes containing deacylated trna fmet in the P site without an A site trna showed a majority of ribosomes (77%) in the low-fret rotated state (Figure 4F). An even more pronounced effect was observed in the case of deacylated trna Phe, which shifted 85% of the ribosome population to the rotated state (Figure 4E), consistent with the higher propensity of trna Phe to occupy the hybrid P/E state, compared with trna fmet (Dorner et al., 2006; Spiegel et al., 2007). A similar difference was also observed between trna fmet and trna Met (Studer et al., 2003). In contrast, only 29% of ribosomes containing only a trna anticodon stem loop (ASL) bound to the P site were found in the hybrid, rotated state (Figure 4L), 580 Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc.

4 Table 1. Kinetic Rates Measured between 0.56 and 0.40 FRET States P Site trna/a site trna Forward Transitions (k 1 ) k 1 (s 1 ) Reverse Transitions (k -1 ) k -1 (s 1 ) TransitIons per Trace trna fmet /Vacant ± ± trna fmet /N-Ac-Phe-tRNA Phe ± ± trna Met /Vacant ± ± Results of fitting FRET time trajectories with the HMM algorithm. Each data set was divided into three and analyzed separately. The reported number is an average from each of the three data sets with the standard deviation. indicating that interactions between the elbow and/or acceptor end of a deacylated trna with the 50S E site promote stabilization of the P/E hybrid state. When EF-G$GDPNP was bound to the complex containing deacylated trna fmet, 94% of the ribosomes were observed in the low-fret, rotated state (Figure 4J; compare with Figure 4F), demonstrating that EF-G$GDPNP converts almost the entire ribosome population to the rotated, hybrid-state conformation. The translocation inhibitor viomycin also converted 90% of trna fmet -containing ribosomes into the rotated state (Figure 4N) (Ermolenko et al., 2007b; Spiegel et al., 2007). In contrast, a significantly lower number of ribosomes containing only an ASL in the P site (29%) or vacant ribosomes (51%) was observed in the rotated, hybrid-state conformation in the presence of EF-G$GDPNP (Figures 4O and 4P). These results show that a deacylated trna in the P site, with or without EF-G bound to the ribosome, contributes to stabilization of the rotated, hybrid state and that EF-G alone is insufficient to convert all ribosomes to the hybrid state (Figure 4Q). Deacylation of P Site trna Has a Dramatic Effect on Ratcheting Kinetics A significant portion of vacant ribosomes and ones containing an ASL or peptidyl-trna exist in the rotated conformation (Figures 4A, 4B, 4K, and 4L), yet only 5% or less of the time traces exhibit any transition between FRET states (Table 2). For example, only 2% of the traces for complexes containing fmet-trna fmet show any transitions, as compared to 71% of traces for deacylated trna fmet. To estimate the interconversion rate at which transitions are very rare, for which HMM analysis is not applicable, we divided the total number of observed transitions by the total observation time from hundreds of molecules and obtained an estimate for the forward rotation (nonrotated to rotated) and reverse rotation (rotated to nonrotated) rates in the range of s 1 (Table 2 and Experimental Procedures). These calculated rates potentially represent an upper limit due to the paucity of observed transitions and may be further complicated by the presence of inactive ribosomes. However, these estimates are sufficient to clearly demonstrate that while there is at most a 5-fold difference in the equilibrium constants between these complexes and their deacyl counterparts, there is as much as a 100-fold difference in kinetic rates (Table 2). These differences are solely due to the presence of a deacylated trna in the P site of the ribosome, which significantly affects both the forward and reverse rotation rates of the ribosome (Figure 4R). Addition of N-Ac-Phe-tRNA Phe to the A site of ribosomes that contained a single deacylated trna in the P site had only a modest effect on the equilibrium and kinetic rates of intersubunit rotation (Table 2 and Figures 4F and 4M compared to Figures 4G and 4H, respectively). In the case of trna fmet, for example, there was only a 1.5-fold increase in the equilibrium constant upon adding N-Ac-Phe-tRNA Phe to the A site and a 1.9-fold increase in the forward rotation rate with no change in the reverse rotation rate. Overall, a total of six complexes were tested that contained a deacylated trna in the P site. For the complexes that contained trna fmet in the P site, 71% and 59% (the latter with N-Ac- Phe-tRNA Phe in the A site) of traces showed FRET transitions between states. The fraction of traces showing FRET transitions was lower for other P site trnas, with 51%, 45%, 12%, and 14% for trna Met,tRNA Phe,tRNA Tyr, and trna Tyr with N-Ac-PhetRNA Phe in the A site, respectively (Table 2). This indicates that the identity of the P site trna can also influence ribosome dynamics. The different propensities of the various trnas to favor intersubunit rotation is consistent with previous results (Dorner et al., 2006; Spiegel et al., 2007). EF-G Facilitates Ratcheting Mildly and Stabilizes the Rotated State The kinetic rates for movement from nonrotated to rotated conformations of the ribosome, for complexes containing deacylated trna bound to the P site, range from 0.27 to 2.53 s 1 (Tables 1and2). As these rates are reduced relative to known in vitro translocation rates measured at room temperature of between 1 and 10 s 1 (Dorner et al., 2006; Pan et al., 2007; Studer et al., 2003), we asked whether the additional rotational rate enhancement is due to EF-G. As mentioned above, various complexes were prepared with EF-G$GDPNP or viomycin present in solution providing near-complete conversion of ribosomes containing deacyl-trna in the P site to the rotated conformation (Figure 4 and Figure S3). In all of these cases, less than 8% of the time traces exhibited transitions between FRET states, and the equilibrium constants are 9 and higher (Table 2). The reduced occurrence of transitions between the two states can be primarily attributed to a significant reduction in the reverse rotation rate by stabilization of the rotated state with EF-G$GDPNP. To test whether EF-G has any effect on the forward rotation rate, we calculated a dwell-time histogram of the nonrotated state from the time traces that showed one or more transient excursions to the nonrotated state (Figures S4 and S5). The dwell-time histogram of the nonrotated state for ribosomes containing trna fmet in the P site with EF-G$GDPNP were fit to a single exponential resulting in a kinetic rate of 1.2 s 1. This corresponds to a rate enhancement of only 2-fold compared to the forward rotation rate for the same complex in the absence of EF-G$GDPNP calculated by HMM (0.5 s 1 ) or dwell-time analysis (0.6 s 1 )(Table 1; Figure S2B). An accurate determination of the forward rotation rates for other samples is not possible for the current data set, as the dwell times are consistently shorter than 1 s and transitions are infrequent (Table 2). For vacant ribosomes Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc. 581

5 Figure 3. Representative Time Traces from HMM Analysis of S6-Cy5/L9-Cy3 Ribosomes Containing Deacylated trna fmet in the P Site and N-Ac-Phe-tRNA Phe in the A Site Cy3 and Cy5 intensities are shown as green and red traces, respectively. The calculated FRET curve is shown in blue with the HMM-determined fit overlaid in black. or for ribosomes with an ASL in the P site, EF-G$GDPNP had minimal effect on ribosome dynamics (Table 2). Real-Time Observation of Intersubunit Movement Triggered by Deacylation, EF-G Binding, and Translocation smfret measurements also allowed us to visualize the real-time dynamics of hybrid-state formation caused by in situ deacylation of peptidyl-trna or binding of EF-G. When puromycin was flowed into a sample cell containing immobilized ribosomes occupied with fmet-trna fmet, a decrease in FRET was observed within 3 s (Figure 5A), indicating that deacylation of fmet-trna fmet triggered movement into the rotated, hybrid-state conformation. When EF-G$GDPNP was added to ribosomes containing deacylated trna fmet bound to the P site, spontaneous fluctuations ceased after 15 s on average, and the ribosomes became stabilized in the rotated, hybrid-state conformation (Figure 5B). Translocation induced by adding EF-G$GTP to a pretranslocation complex in which trna fmet was bound to the P site and N-Ac-Phe-tRNA Phe to the A site converted ribosomes into the nonrotated, classical state, accompanied by disappearance of fluctuations after 22 s on average (Figure 5C). These single-molecule data further support our previous assignment of conformational states of the ribosome and demonstrate that our immobilization scheme preserves ribosome function. As such, the results presented here serve as a basis for future single-molecule measurements of the ribosome s conformational dynamics during protein synthesis. DISCUSSION The Intersubunit Rotational States Are in Dynamic Equilibrium Our smfret experiments have allowed direct observation of the long-hypothesized ribosomal intersubunit movements in real time. Remarkably, spontaneous forward and reverse rotation be- tween the nonrotated and rotated states was observed in the absence of EF-G and GTP. Consistent with earlier reports (Ermolenko et al., 2007a, 2007b; Spiegel et al., 2007; Valle et al., 2003), the forward rotation to the rotated state appeared to be coupled to the transition of deacylated trna from the P/P state into the hybrid P/E state. Indeed, significant fluctuations between the two major conformational states were observed only when a deacylated trna was bound to the P site, regardless of A site occupancy. Conversely, ribosomes containing either peptidyl-trna or an ASL in the P site were fixed predominantly in the classical, nonrotated state (Figures 4A, 4B, and 4L). Since stable interaction between trna and the 50S E site has a stringent requirement for a deacylated acceptor end (Lill et al., 1986), the presence of an amino acid or an elongating peptide on the P site trna is expected to block its movement into the P/E hybrid state (Korostelev et al., 2006; Selmer et al., 2006). Unexpectedly, a significant fraction of posttranslocation ribosomes occupied the rotated state (Figures 4A and 4B), although intersubunit fluctuations were nearly absent and the majority of ribosomes were in the nonrotated state. Possible explanations are (1) spontaneous deacylation of trna, leading to formation of the P/ E state; (2) incomplete saturation of the P site by peptidyl-trna; (3) reverse translocation, as recently reported (Konevega et al., 2007; Shoji et al., 2006), leading to formation of the A/P hybrid state; (4) the presence of inactive ribosomes unable to bind trna but adopting the rotated conformation; and (5) an unprecedented movement of a fraction of peptidyl-trna into the P/E hybrid state. Our previous probing, filter-binding, and toe-printing experiments (Spiegel et al., 2007) indicate that significant spontaneous deacylation of peptidyl-trna is unlikely under our experimental conditions. Moreover, this explanation is inconsistent with the lack of rotational fluctuations observed for posttranslocation ribosomes (Table 2). Increasing the concentration of peptidyl-trna in the imaging buffer by two orders of magnitude did not markedly shift the distribution between classical and hybrid states (Figure S6), ruling out partial occupancy of the P site. Finally, the results of biochemical experiments do not support reverse translocation under these conditions (data not shown). Future smfret experiments using fluorescently labeled peptidyl-trna should help to test the remaining possibilities, that a fraction of ribosomes are unable to bind trna and so are able to occupy the rotated conformation or whether our results are, in fact, indicative of peptidyl-trna bound in the hybrid P/E state. 582 Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc.

6 Comparing Ribosome Dynamics and trna Dynamics at the Single-Molecule Level Under our conditions, even in the absence of EF-G, the hybrid state appears to be thermodynamically favored over the classical state by about kcal/mol for pretranslocation ribosomes (Figure 2 and Table 1). The frequencies of oscillation of fluorescently labeled trna molecules between states that are believed to correspond to the classical and hybrid states (1 5 s 1 ) measured in recent smfret experiments (Kim et al., 2007; Munro et al., 2007) are about an order of magnitude higher than the frequencies of intersubunit motion observed here with trna fmet in the P site ( s 1, Table 2). In addition, the trna dynamics data showed that the classical state is favored over the hybrid state under most conditions. These differences are most likely due to the different experimental conditions and constructs used in the two studies. The reconstituted ribosomes used in our experiments, fluorescent labeling of trnas in the studies by Kim et al. (2007) and Munro et al. (2007), and the different ionic conditions may all affect the observed frequencies. For example, differences in magnesium ion concentrations can increase the propensity of the ribosome for the classical state by up to 5-fold (Kim et al., 2007). However, because no one has measured the single-molecule dynamics of the ribosome and the trnas simultaneously, we can not presently rule out the possibility that the trnas may undergo movement independently of intersubunit rotation, at a faster rate. Rates for the overall step of translocation in the presence of EF-G$GTP have been reported as between 1 and 10 s 1 at 25 C, depending on experimental conditions (Dorner et al., 2006; Pan et al., 2007; Studer et al., 2003). The differences between previously reported rates of translocation and the rates of intersubunit rotation determined here may be at least partially accounted for by an acceleration of the rate of hybrid-state formation by EF-G (Spiegel et al., 2007), as suggested by experiments performed in the presence of EF-G$GDPNP (Figure 4, Figure S3, and Table 2). Recently, the existence of an additional hybrid-state intermediate, containing trnas bound in the A/A-P/E state, was proposed based on smfret (Munro et al., 2007) and stopped-flow (Pan et al., 2007) experiments using fluorescently labeled trnas. This proposal implies independent movement of A and P site trnas into the A/P and P/E hybrid states. Moreover, different degrees of small ribosomal subunit rotation (3 10 ) have been observed in different cryo-em reconstructions (Frank and Agrawal, 2000; Gao et al., 2003, 2004; Valle et al., 2003) and bulk FRET experiments (Ermolenko et al., 2007a), also raising the possibility of intermediate rotational states. However, in our experiments only a single hybrid-state intermediate was found. This is consistent with a recent maximum likelihood classification of cryo-em data that was able to only identify two genuine classes of ribosomes in the data set (Scheres et al., 2007). Analysis of our single-molecule data shows that the different magnitudes of FRET changes observed in ensemble experiments (Ermolenko et al., 2007a) are caused by shifts in the population of two major conformations rather than by the existence of additional conformations with intermediate degrees of rotation. If any undetected rotational intermediates exist, their lifetimes must be lower than the time resolution of our experiments ( ms) or the magnitude of rotation accompanying formation of these intermediates must be too small for detection by FRET. Although we found no evidence for existence of an additional A/A-P/E hybrid-state intermediate, it is possible that movement of trna from the A/A to the A/P state would result in a change in intersubunit orientation below the limit of detection of our method. This would not be surprising, since, in contrast to the transition from the P/P to P/E state, which entails movement of the CCA-end of trna by about 50 Å, transition from the A/A to A/P states involves a much smaller movement (Korostelev et al., 2006; Noller et al., 2002; Valle et al., 2003). The P Site trna Significantly Impacts Ribosome Dynamics One of the most striking observations is that deacylation of the P site trna dramatically enhances both the forward and reverse rotation rates of intersubunit rotation, in addition to shifting the equilibrium toward the rotated state. A nearly 100-fold increase in the forward rotation rate was observed upon deacylation of the P site trna (fmet-trna fmet versus trna fmet in the P site; Table 2). This result supports previous proposals suggesting that deacylation of peptidyl-trna plays a decisive role in triggering both trna and intersubunit movement (Ermolenko et al., 2007a; Moazed and Noller, 1989a; Valle et al., 2003; Zavialov and Ehrenberg, 2003). It is remarkable that a single formyl-methionyl or acetyl-phenylalanyl acyl group can critically affect the dynamic behavior of a 2.4 MD macromolecular complex. The identity of the deacylated trna in the P site conferred modest but measurable differences in intersubunit rotational rates for the four different deacylated trnas used. The largest differences were observed between the trna fmet and trna Tyr complexes where the forward rotation rate for trna Tyr was 5-fold faster than for trna fmet (Figures 4F and 4M). A similar effect was observed for these two trnas when N-Ac-Phe-tRNA Phe was present in the A site (Figures 4C and 4D). Because the elbow and/or acceptor stem moieties facilitate trna movement into the hybrid state (compare Figures 4F and 4L; Table 2), these differences may be influenced by the differential affinities of these features of the respective trnas for the P and E sites. Addition of N-Ac-Phe-tRNA Phe to the A site had less than a 2-fold effect on the forward rotation rate while leaving the reverse rotation rate unchanged. Other constituents on the A site trna were not investigated in this study but have been investigated biochemically and at the single-molecule level in other studies (Kim et al., 2007; Munro et al., 2007; Sharma et al., 2004; Shoji et al., 2006). These reports suggest that the identity of the A site trna constituent can influence the dynamics of ribosomal movement. Nevertheless, more systematic studies will be needed to fully understand the contributions of the A site trna to intersubunit dynamics. Intersubunit Movement and the Mechanism of Translocation Thermal Brownian motion has been shown to play a key role in the mechanical movement of macromolecular machines (Astumian, 1997; Cordova et al., 1992) and may also underlie the mechanism of ribosomal translocation (Spirin, 2004). Although the timeresolved smfret trajectories (Figures 2, 3, and 5 and Figure S4) show relatively long (1 10 s) dwell times, the actual transit times between the hybrid, rotated state and the classical, nonrotated Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc. 583

7 Figure 4. Histograms Compiled from Several Hundred Time Traces Showing Distributions of FRET Values for Different Ribosome Complexes S6-Cy5/L9-Cy3 ribosomes were assembled with mrna and (A) N-Ac-Phe-tRNA Phe in the P site, (B) fmet-trna fmet in the P site, (C) trna fmet in the P site and N-Ac- Phe-tRNA Phe in the A site translocated with EF-G$GTP, (D) trna Tyr in the P site and N-Ac-Phe-tRNA Phe in the A site translocated with EF-G$GTP, (E) trna Phe in 584 Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc.

8 Table 2. Statistical Data for All Tested Complexes Figure Percent NR Percent R K eq k 1 (s 1 ) k 1 (s 1 ) Method Percent Trans Vacant and ASL in P Site Vacant 4K 61 a 39 a 0.63 b c c 1, 1 d 4 e Vacant with EF-G$GDPNP 4O , 1 5 ASL fmet 4L , 1 3 ASL fmet with EF-G$GDPNP 4P , 1 2 Peptidyl-tRNA in P Site N-Ac-Phe-tRNA Phe 4A , 1 4 fmet-trna fmet 4B , 1 2 Pretranslocation Complexes trna fmet /N-Ac-Phe-tRNA Phe 4G ± ± , 2 59 trna Tyr /N-Ac-Phe-tRNA Phe 4H ± , 1 14 Posttranslocation Complexes trna fmet /N-Ac-Phe-tRNA Phe with 4C , 1 8 EF-G$GTP trna Tyr /N-Ac-Phe-tRNA Phe with EF-G$GTP 4D , 1 9 Deacyl-tRNA in P Site trna fmet 4F ± ± , 2 71 trna fmet with EF-G$GDPNP 4J ± , 1 7 trna fmet viomycin 4N ND ND 3 trna Phe 4E ± ± , 3 45 trna Tyr 4M ± , 1 12 trna Tyr with EF-G$GDPNP S3B ND ND 1 trna Tyr with viomycin S3D ND ND 0 trna Met 4I ± ± , 2 51 trna Met with EF-G$GDPNP S3A ND ND 3 trna Met with viomycin S3C ND ND 0 ND, insufficient data to calculate rate information. For complexes stabilized in the rotated conformation (e.g., complexes containing EF-G or viomycin), equilibrium constants calculated from the relative populations of nonrotated and rotated ribosomes deviate significantly from constants calculated from the ratios of rates for forward and reverse rotation. This discrepancy is likely due to the presence of a fraction of inactive ribosomes that skews the value of the equilibrium constant calculated from the distribution. a Values in this column are derived from fitting the histograms providing the percent of the molecules in the nonrotated (NR) or high FRET state and the rotated (R) or low FRET state. b Equilibrium constants in this column are calculated from the relative populations of nonrotated and rotated ribosomes (see text). c The rates in this column were calculated as described in the text and in the Experimental Procedures. d In this column, the method used to calculate the rates in which 1 is the number of transitions/dwell time, 2 is HMM analysis, and 3 is dwell-time analysis. A more detailed description is in the Experimental Procedures. e The percentages in this column represent the total number of traces that contain at least one unambiguous FRET transition between the fitted high and low FRET states divided by the total number of traces. states are faster than the time resolution ( ms) of our measurements. The rapid transit times, which are consistent with a correlation time for rotational Brownian diffusion of the 30S ribosomal subunit of 1 2 ms (Amand et al., 1977), and the fact that intersubunit movement can occur spontaneously in the absence of EF-G (Figures 2 and 3) provide evidence that thermal energy is sufficient to drive the ribosomal movements associated with translocation, obviating the need for a power stroke from GTP hydrolysis for the first step of translocation. Although EF-G binding contributes to stabilization of the hybrid the P site due to deacylation of the complex in (A) with puromycin, (F) trna fmet in the P site, (G) trna fmet in the P site and N-Ac-Phe-tRNA Phe in the A site, (H) trna Tyr in the P site and N-Ac-Phe-tRNA Phe in the A site, (I) trna Met in the P site, (J) trna fmet with EF-G$GDPNP, (K) no trna, (L) anticodon stem loop (ASL) from trna fmet, (M) trna Tyr in the P site, (N) trna fmet with viomycin, (O) no trna with EF-G$GDPNP, and (P) anticodon stem loop (ASL) from trna fmet with EF-G$GDPNP. The FRET data was smoothed with a five-point window and can be fitted to two Gaussians. (Q) A graphical depiction of the percent of each complex in the nonrotated state (Table 2). It is clear that the nonrotated state is weakly populated if the P site trna is deacylated in comparison to other constructs. (R) A scatter plot showing the forward versus reverse rotation rates for various ribosomal complexes (Table 2). The ribosomes with a deacylated trna have much higher rates of forward and reverse rotation. The data with EF-G are not included in the plot. Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc. 585

9 Figure 5. Single-Molecule Time Traces of S6-Cy5/L9-Cy3 Ribosome Complexes Showing Intersubunit Movement Induced by Reaction with Puromycin, EF-G Binding, or Translocation (A) Ribosomes containing fmet-trna fmet bound to the P site were reacted with puromycin (1 mm) in buffer B (see the Experimental Procedures) flowed into the microscope slide 20 s after beginning the fluorescence recording (arrow). Intersubunit movement is seen as a sharp decrease in the FRET signal (indicated by the vertical dashed line). Thirteen of twenty-four molecules that did not photobleach before addition of puromycin showed similar behavior with an average deacylation time of 20 s. (B) EF-G$GDPNP (300 nm, 250 mm) in buffer B (see the Experimental Procedures) was introduced at 40 s (arrow) into complexes containing deacylated trna fmet bound to the P site and a vacant A site. EF-G-induced intersubunit rotation is observed as stabilization of the low FRET state (vertical dashed line). Forty-two of fifty-one molecules showed similar behavior with an average time between addition of EFG$GDPNP and the last FRET fluctuation of 15 s. (C) EF-G$GTP (300 nm, 250 mm) in buffer B (see the Experimental Procedures) was introduced at 20 s (arrow) to pretranslocation complexes containing deacylated trna fmet bound to the P site and N-Ac-Phe-tRNA Phe bound to the A site. Translocation is observed as the transition to the high FRET state (vertical dashed line). Forty of seventy-six molecules showed similar behavior with an average translocation time of 22 s after addition of EF-G$GTP. state (Figure 4 and Figure S3), a more critical role for EF-G may be to uncouple movement of the small subunit from that of the mrna-trna complex to promote the second step of translocation. In this study, we investigated 20 different ribosomal complexes, only one of which has previously been studied, using an alternative approach of labeling two trna molecules (Blanchard et al., 2004b; Kim et al., 2007; Munro et al., 2007). This comprehensive analysis of intersubunit dynamics in pre- and posttranslocation ribosomes represents a first step toward a description of the internal movements of the ribosome during the complete process of protein synthesis at the single-molecule level. EXPERIMENTAL PROCEDURES Buffers Buffer A consists of 20 mm HEPESKOH (ph 7.5), 6 mm MgCl 2, 150 mm NH 4 Cl, 6 mm b-mercaptoethanol, 2 mm spermidine, and 0.1 mm spermine. Buffer B consists of 20 mm HEPESKOH (ph 7.5), 6 mm MgCl 2, 150 mm NH 4 Cl, 6 mm b-mercaptoethanol, 2 mm spermidine, 0.1 mm spermine, 0.8 mg/ml glucose oxidase, 0.625% glucose, 1.5 mm 6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic (Trolox), and 0.03 mg/ml catalase. Materials and Sample Preparations trna fmet was purchased from MP Biomedicals; GTP, GDPNP, puromycin, trna Phe, trna Tyr, and trna Met were purchased from Sigma. Defined mrna m291 and m301 were transcribed in vitro and further purified as described previously (Fredrick and Noller, 2002). The biotin-labeled DNA primer (5 0 biotin, 586 Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc.

10 CTTTATCTTCAGAAGAAAAACC-3 0 ) was synthesized by Integrated DNA Technologies. NeutrAvidin used for sample immobilization at a final concentration of 0.2 mg/ml was purchased from Pierce. fmet-trna fmet, N-Ac-Phe-tRNA Phe, and EF-G with a 6-histidine tag were prepared and purified as previously described (Dorner et al., 2006; Moazed and Noller, 1989b; Wilson and Noller, 1998). The components of the oxygen scavenging system (glucose oxidase from Aspergillus Niger, glucose, and 6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid) were purchased from Sigma. Catalase from beef liver was from Roche. Ribosomal subunits were prepared from E. coli strains MRE600 (wild-type) and CSH142/DL9K (containing a chromosomal deletion of the gene encoding protein L9) as described (Hickerson et al., 2005). Mutant variants of ribosomal proteins S6 (D41C), S11 (E75C), and L9 (N11C) were created by site-directed mutagenesis expressed, purified, and labeled separately with maleimide derivatives of Cy3 (donor) or Cy5 (acceptor) dyes (Amersham Biosciences) as described previously (Hickerson et al., 2005; Lieberman et al., 2000). Labeled proteins S6 or S11 were incorporated into 30S subunits by in vitro reconstitution from purified 16S rrna and the other 19 individually purified ribosomal proteins according to published procedures (Culver and Noller, 1999; Hickerson et al., 2005). Labeled protein L9 was incorporated into 50S subunits by partial reconstitution from 50S subunits carrying an L9 deletion, and doubly labeled 70S ribosomes were isolated using previously described procedures (Ermolenko et al., 2007a). Assembly and Immobilization of Ribosomal Complexes Ribosomal complexes were constructed in buffer A. Ribosome$P sitetrna$mrna complexes were constructed by incubation of 70S ribosomes (0.3 mm) with mrna m291 or m301 (0.6 mm) preannealed to biotin-labeled primer (0.8 mm) and trna (trna fmet, trna Met, trna Tyr, N-Ac-Phe-tRNA Phe,or fmet-trna fmet ), depending on the experiment, (0.7 mm) for 20 min at 37 C. Pretranslocation complexes were made by binding N-Ac-Phe-tRNA Phe (0.6 mm) to a P site complex containing deacylated trna fmet and m291 or trnatyr and m301 for 30 min at 37 C. All complexes were subsequently diluted with buffer A to a final concentration of 2 nm and were immobilized on the quartz surface of microscope slides prepared for these experiments using an established protocol (Zhuang et al., 2000). To prevent photobleaching during data acquisition the sample buffer was exchanged for imaging buffer B, containing an oxygen scavenging system (Rasnik et al., 2006). For the flow experiments (Figure 5), the average event time is an estimated difference between when the solution was added to the sample cell and the appearance of the expected event. smfret Data Acquisition and Analysis For single-molecule measurements, an Olympus IX-70 or IX-71 microscope with a UPlanApo 603/1.20w objective lens was used. A 532 nm laser (Crysta- Laser or Meshtel) was used for excitation of Cy3. Total internal reflection was obtained by using prism-type TIR (Cornish and Ha, 2007). The fluorescence emission was split in two (Cy3, Cy5 emission) with a 630 nm dichroic mirror (Chroma). For visualizing the fluorescence signal, an Andor ixon or ixon+ EMCCD camera was used. Various optical components were purchased from Thorlabs, Newport, or Edmund Optics. TIRF movies were obtained using in-house software. For all constructs, the time binning of the data was set at 100 ms with some of the data acquired for histogram analysis at 200 ms. The resultant image files were processed using IDL software and analyzed using Matlab. Time traces were selected from the data set by choosing only traces that contained single photobleaching steps for Cy3 and Cy5. Thus, data that contain multiple dyes or only a single donor were not included in subsequent analysis. Furthermore, fluorescence blinking events, although extremely rare, were removed by truncating the traces prior to the blinking event. Histograms were created from the selected traces and smoothed with a five-point window. Histograms were fit to Gaussian distributions using Origin. The peak position was left unrestrained. Minor deviations in peak position were observed for complexes that existed predominately in one state and therefore represent an inability to accurately fit this small portion of the distribution as opposed to an actual change in the FRET value. For Figure 4 and Figure S3, the data were best fit to two Gaussians in which the resultant widths were self-consistent and the residuals were random except for Figures S3B and S3C, which were best fit to a single Gaussian. HaMMy was used for HMM analysis of the FRET data (McKinney et al., 2006). A total of 572, 612, and 385 molecules were analyzed for S6-Cy5/L9-Cy3 ribosomes containing trna fmet, trna fmet /N-Ac-Phe-tRNA Phe, and trna Met, respectively. Fitting to three FRET states resulted in just two observable FRET states with a total of 19,808, 18,451, and 6,498 transitions, respectively. Kinetic rates were determined by transition density plot analysis (McKinney et al., 2006). The data were fit to three FRET states, which is more than the expected FRET states, to determine if hidden intermediate states existed. The data were also fit to two and five different states, each resulting in the same two states as observed in a transition density plot (data not shown). Markov analysis was not performed on the remainder of the data, since the time traces failed to contain at least five transitions in 50% of the total data set. To estimate the kinetic rates for all of other complexes, one of the following two methods was performed. In the first, the kinetic rates were calculated by dividing the total number of transitions from one state to another by the sum of the dwell times from which the transitions originated (k = #transitions 1/2 = P t 1, where t is the dwell time in state 1). This analysis was performed on the vacant, ASL, and acyl-trna-containing ribosomes including these constructs in the presence of EF-G$GDPNP. Also, this analysis was performed on the rotated to nonrotated state transition for all constructs that had less than 15% of the data in the nonrotated state (Table 2). To calculate the nonrotated to rotated state transitions for these complexes, a second method was employed. For this method, the dwell times in the nonrotated state were tabulated, plotted as a histogram, and fit to an exponential decay (Figure S5). This dwell-time analysis was also performed on the constructs analyzed by HMM analysis (Figure S3). For several constructs, kinetic rates could not be determined, since it was determined that the rate was on the order of or faster than the time resolution of the experiments. SUPPLEMENTAL DATA Supplemental Data include six figures and can be found with this article online at ACKNOWLEDGMENTS These studies were supported by grant number GM from the National Institutes of Health (NIH) to H.F.N., grant number GM from the NIH and a National Science Foundation (NSF) CAREER award to T.H., postdoctoral fellowship PF GMC from the American Cancer Society to P.V.C., and a NATO-NSF postdoctoral fellowship to D.N.E. The authors thank Robyn Hickerson for small ribosomal subunit proteins used in reconstitution; Robyn Hickerson, Zigurts Majumdar, and Michelle Nahas for their early contributions to the project; and Andrei Korostelev and Laura Lancaster for discussions. T.H. is an Investigator of the Howard Hughes Medical Institute. Received: March 11, 2008 Revised: April 29, 2008 Accepted: May 7, 2008 Published: June 5, 2008 REFERENCES Amand, B.,Pochon,F., andlavalette, D. (1977). Rotationaldiffusion ofescherichia coli ribosomes. I. Free 70 S, 50 S and 30 S particles. Biochimie 59, Astumian, R.D. (1997). Thermodynamics and kinetics of a Brownian motor. Science 276, Blanchard, S.C., Gonzalez, R.L., Kim, H.D., Chu, S., and Puglisi, J.D. (2004a). trna selection and kinetic proofreading in translation. Nat. Struct. Mol. Biol. 11, Blanchard, S.C., Kim, H.D., Gonzalez, R.L., Jr., Puglisi, J.D., and Chu, S. (2004b). trna dynamics on the ribosome during translation. Proc. Natl. Acad. Sci. USA 101, Cordova, N.J., Ermentrout, B., and Oster, G.F. (1992). Dynamics of single-motor molecules: the thermal ratchet model. Proc. Natl. Acad. Sci. USA 89, Cornish, P.V., and Ha, T. (2007). A survey of single-molecule techniques in chemical biology. ACS Chem. Biol. 2, Molecular Cell 30, , June 6, 2008 ª2008 Elsevier Inc. 587